Vegetation Dynamics in Australia's Wet Tropics

Two systems or one? Vegetation dynamics in Australia’s Wet Tropics

Photos: L.Warman

Laura Warman

Thesis submitted for the degree of Doctor of Philosophy

Evolution and Ecology Research Centre School of Biological, Earth and Environmental Sciences University of New South Wales

March 2011

PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: Warman

First name: Laura Other name/s:

Abbreviation for degree as given in the University calendar:

School: School of Biological, Earth and Environmental Sciences Faculty: Science Evolution and Ecology Research Centre

Title: Two systems or one? Vegetation dynamics in Australia’s Wet Tropics

Abstract 350 words maximum: (PLEASE TYPE) Around the world tropical rainforests intergrade with open, fire-dependent (pyrophytic) vegetation forming landscape- scale mosaics. However, rainforests and open vegetation are mostly studied independently from each other. This has been the case in the Australian Wet Tropics. This thesis begins by arguing that the vegetation of north-eastern Queensland can be considered as a complex system where rainforests and pyrophytic vegetation represent alternative stable states. This is important because it creates a new context for understanding the region’s biological dynamics. It allows us to move forward from an understanding where only abiotic factors determine the distribution of vegetation, to one in which the vegetation itself plays an important and active role at several scales. Globally, studies of abrupt boundaries between closed forests and pyrophytic vegetation focus on the contrast between abiotic factors on either side of the boundaries. I used a novel approach considering controls in non-boundary vegetation and comparing them to parameters across boundaries at a regional scale. My results show that contrasts in soil chemistry across boundaries are not greater than the variance found within either vegetation type, and that the vegetation itself can change soil chemistry remarkably quickly. This highlights the active, rather than passive, role of vegetation in the region. In contrast to other systems presenting alternative stable states (such as coral reefs, savannas or kelp forests), I found herbivory in the Wet Tropics was similar across vegetation types. I also show for the first time that the advantages of compound vs. simple leaves do not lie in leaf construction or herbivore avoidance, addressing a long debated topic to which there is still no satisfying answer. Lastly, a computer model compares the effects on landscape-scale vegetation patterns of changing species interactions under varying environmental conditions. The model evaluates the outcome of obligate rainforest and pyrophytic “species” competing with pioneers and ‘traitors’ and proposes long-term hypotheses that may be tested in the field. In conclusion, using ideas about alternative stable states as a starting point has allowed me to make contributions towards understanding the interaction between different factors involved in large-scale vegetation dynamics in the Australian Wet Tropics.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).

………………………………………………………… ……………………………………..…………… ……….……………………...… Signature Witness Date

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award:

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed ……………………………………………......

Date …………………………………….………......

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed ……………………………………………......

Date …………………………………….………......

ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ……………………………………………......

Date …………………………………….………......

ii Statement of contributions of co-authors and declarations of permission to publish

All publishers and co-authors have granted permission for the following publications to be submitted and published as a thesis. No other authors will be submitting this work as part of their thesis submissions. The contribution of each author to the respective publications is stated at the start of each chapter. All photographs were taken by Laura Warman.

Warman, L. and Moles, A.T. 2009. Alternative stable states in Australia's wet tropics: a theoretical framework for the field data and a field case for the theory. Landscape Ecology 24: 1-13. DOI: 10.1007/s10980-008-9285-9

Reproduced with kind permission from Springer Science+Business Media: Landscape Ecology, Alternative stable states in Australia's wet tropics: a theoretical framework for the field data and a field case for the theory, 24, 2009, 1-13, Warman, L. & Moles A.T, © Springer Science+Business Media B.V. 2008.

Warman L., Moles A.T and Edwards W. 2011. Not so simple after all: Searching for the ecological advantages of compound leaves. Oikos 120:813-821. DOI: 10.1111/j.1600-0706.2010.19344.x

Reproduced with kind permission from Oikos and Wiley-Blackwell. Oikos grants authors the right to reproduce their article in a new publication of which they are the author, editor or co-editor, including a thesis or dissertation.

Warman L., Bradford M.G. and Moles A.T. A broad approach to abrupt boundaries: Looking beyond the boundary at soil attributes within and across tropical vegetation. In review at Journal of Biogeography

Warman L., Fletcher C.S. and Moles A.T. Traitors on the landscape? A model of interactions between rainforests and fire-prone forests. In preparation for Journal of Ecology

iii

Acknowledgements

The first acknowledgement must go to Professor Des Cooper; if not for him and the MConBio program I might not have returned to science. With humour and encouragement, Des let me convince myself that I wanted to pursue a PhD (and tried to pass the blame later). For all of this, the sav blanc and proper introduction to Melbourne Cup Day: Thank you again Des!

If Des opened the door to the PhD, then Angela Moles lay down the welcome mat and invited me in. For all her good intentions, she has since been saddled with nagging questions, endless discussions on soil, many e-mails sent at the wee hours of dawn and never-ending first chapter drafts. Angela: my most heartfelt thanks. I seriously don’t believe I could have started, let alone finished a PhD without your constant support and faith in me (both personally and as a scientist). Admittedly, the hard-lines probably helped as well. Thank you so much for all you’ve taught me. Thank you for all the carrots and the sticks, the slaps-upside the head, the encouragement, the hugs, cups of tea, and especially the ever-important smiley-faces on manuscript drafts. I’m looking forward to climbing trees with you again!

Will Edwards literally opened the door to me and, in return, promptly had his car filled with mud, leeches and leaves (more than once). Thank you for giving me an anchorage in the Far North and the foothold I needed to get going. Thank you as well for all those hours in the forest tagging leaves, all the advice, comments and all those discussions late into the tropical night.

Special thanks to Paul Adam for asking me hard questions in kind ways and answering all my questions. Thank you as well for all of the relevant reading material that appeared magically on my desk throughout my candidature. A bit belatedly, but thank you for the dry wit and lectures during MConBio which planted a seed of fascination and curiosity about the “double-decker forests” growing up north.

Thanks to David Hilbert for listening to the jumbled thoughts in my head, for suggesting I go talk to a certain modeller that happened to be on site, and for all the discussions under the porch or in the banana jungle. iv For help in finding and accessing field sites in Far North Queensland, thanks to Dan Metcalfe as well as Mick Blackman and John Kanowski from the Australian Wildlife Conservancy (especially for access to Mt. Lewis). Thanks to Matt Bradford for showing me lovely wet sclerophyll forest, listening to my ideas and inviting me to brew.

Thanks to all my friends and fellow PhD students at Victoria University in Wellington and UNSW in Sydney. Special thanks go to everyone in the Big Ecology lab (and associated fauna). Heidy, you are definitely the most beautiful girl in the room. Jo, we need to reminisce about when we first reminisced (and eat almond croissants). Ray, this isn’t part of the thesis but it’s still very important (as is saying thanks and yes, you are awesome). Margo, so long and thanks for all the sushi (and for holding my hand in hospital when I needed a hand to hold so badly). Fiona, say hi to Paula Thomson for me. Emily, fancy a game of Buckhunter? Rhiannon, thanks for all the smiles and good cheer. Habacuc, thank you for the clover that gave me luck the last few weeks! Brett and Jemaine, thanks for all the hilarity during the bleak moments. Stephen, thanks for all your help (especially in obtaining funding) and for letting me parasitize your home and garage. Thanks to Frank Hemmings for putting up with all sorts of odd questions and teaching me so much, but especially for being a friend.

Dr. Fletcher, thank you for the huge effort and helping me to translate my metaphors into scores, percentages and species response curves. Cameron, thank you... For all your support, in so very many ways. For trying to stay up to keep me company while I scanned leaves late into the night. For helping me see beyond the immediate situation and making me use the closet and drawers for the past two and a half weeks. For opening the door to an Australia I could not have accessed otherwise. Also, for helping me laugh at myself and for reminding me that if it were all certain, neat and tidy it wouldn’t be science. You’ve kept me sane(ish) and I am forever grateful.

Mom, Dad... I don’t even know how to say thanks. For all the encouragement, love and support. For leaving the door open to departures and arrivals. For the support from the Warman Foundation. For believing in me and trying to convince me when I didn’t believe you. For letting me bounce ideas off you at bleak hours of dawn. For coming to see Queensland. For reminding me about context and home when I’ve needed it most...Thank you so much...Besototototes

Cameron: there are just no words, and yet too many already....

With all my heart, this is dedicated to you.

....With fire, woodland thrives. With water, rainforest spreads. Two systems, or one?

vi Abstract

Around the world tropical rainforests intergrade with open, fire-dependent (pyrophytic) vegetation forming landscape-scale mosaics. However, rainforests and open vegetation are mostly studied independently from each other. This has been the case in the

Australian Wet Tropics. This thesis begins by arguing that the vegetation of north- eastern Queensland can be considered as a complex system where rainforests and pyrophytic vegetation represent alternative stable states. This is important because it creates a new context for understanding the region’s biological dynamics. It allows us to move forward from an understanding where only abiotic factors determine the distribution of vegetation, to one in which the vegetation itself plays an important and active role at several scales.

Globally, studies of abrupt boundaries between closed forests and pyrophytic vegetation focus on the contrast between abiotic factors on either side of the boundaries. I used a novel approach considering controls in non-boundary vegetation and comparing them to parameters across boundaries at a regional scale. My results show that contrasts in soil chemistry across boundaries are not greater than the variance found within either vegetation type, and that the vegetation itself can change soil chemistry remarkably quickly. This highlights the active, rather than passive, role of vegetation in the region.

In contrast to other systems presenting alternative stable states (such as coral reefs, savannas or kelp forests), I found herbivory in the Wet Tropics was similar across vegetation types. I also show for the first time that the advantages of compound vs. simple leaves do not lie in leaf construction or herbivore avoidance, addressing a long debated topic to which there is still no satisfying answer. Lastly, a computer model

vii

compares the effects on landscape-scale vegetation patterns of changing species interactions under varying environmental conditions. The model evaluates the outcome of obligate rainforest and pyrophytic “species” competing with pioneers and ‘traitors’ and proposes long-term hypotheses that may be tested in the field.

In conclusion, using ideas about alternative stable states as a starting point has allowed me to make contributions towards understanding the interaction between different factors involved in large-scale vegetation dynamics in the Australian Wet Tropics.

viii Table of Contents

Thesis/Dissertation sheet Copyright and Authenticity statements ...... i Originality statement ...... ii Statement of contributions and declarations of permissions to publish ...... iii Acknowledgements ...... iv Abstract ...... vii

Chapter One: Introduction ...... 1 The Australian Wet Tropics ...... 5 Thesis structure ...... 8 References ...... 10

Chapter Two: Alternative stable states in Australia's wet tropics: a theoretical framework for the field data and a field case for the theory Statement of co-author contributions ...... 17 Abstract ...... 18 How the Australian Wet Tropics are understood today: a complex mosaic ...... 20 The importance of fire and water ...... 21 An alternative paradigm: alternative stable states ...... 24 Is there evidence of alternative stable states in Far North Queensland? ...... 30 Implications: new insights into old problems— why ASS are a well-rounded paradigm for the Wet Tropics ...... 34 Conclusions ...... 41 Acknowledgements ...... 42 References ...... 42

Chapter Three: Not so simple after all: Searching for the ecological advantages of compound leaves Statement of co-author contributions ...... 51 Abstract ...... 52 Introduction ...... 53 Materials and methods ...... 58 Results ...... 64 ix

Discussion ...... 69 Conclusions ...... 73 Acknowledgements ...... 74 References ...... 75 Appendices ...... 81

Chapter Four: A broad approach to abrupt boundaries: Looking beyond the boundary at soil attributes within and across tropical vegetation. Statement of co-author contributions ...... 85 Abstract...... 86 Introduction ...... 87 Methods ...... 92 Results ...... 97 Discussion ...... 108 Conclusions ...... 118 Acknowledgements ...... 118 References ...... 119 Appendices ...... 126

Chapter Five: Traitors on the landscape? A model of interactions between rainforests and fire-prone forests Statement of co-author contributions ...... 135 Abstract...... 136 Introduction ...... 137 Methods ...... 141 Results ...... 147 Discussion ...... 157 Conclusions ...... 163 References ...... 163

Chapter Six: General discussion and conclusions. General discussion ...... 169 Conclusions ...... 181 References ...... 184 x

Chapter One

Introduction

Photo: L.Warman

1 General introduction

Australia is an island continent renowned for its aridity, yet scattered across the northern, eastern and south-eastern coasts, lay a string of rainforest “islands” (Adam

1992, Bowman 2000). The contrast and proximity between lush, dark tropical rainforest and stark, sun-baked open vegetation has long caught the attention of explorers, naturalists, biologists and doctoral students in ecology (see Unwin 1983, House 1986,

Duff 1987, Adam 1992, Bowman 2000). In his appropriately titled 1910 presentation to the Royal Society of Queensland (“Some problems of Queensland's

Botanogeography”), Karel Domin remarked that: “The line of demarcation between them [open forests and “vine scrubs” (rainforests)] is most distinct, a fascinating phenomenon unique in the whole world” (Domin 1911). He went on to add that “it is very hard to explain this most extraordinary contact between forest and scrub”.

Photos: L.Warman Woodland in Dinden National Park, Qld. Rainforest in Daintree National Park, Qld.

Almost exactly one century later, mosaics and boundaries between tropical rainforest and open, fire-dependent vegetation have been described around the world. We know of their existence in Brazil (Hoffmann et al. 2009), Ghana (Markham and Babbedge 1979,

Swaine 1992), the Central African Republic (Beauvais 2009), Ivory Coast (Goetze et al. 2006, Hennenberg et al. 2006), India (Mariotti and Peterschmitt 1994, Puyravaud et

2 General introduction

al. 1994), Madagascar (Virah-Sawmy 2009), New Caledonia (Perry and Enright 2002,

Stevenson and Hope 2005), etc. However, we still find it quite difficult to explain this

“extraordinary contact”. For the most part, scientists continue to study rainforests and open vegetation as two independent biological systems that happen to occur close together. Scientists also, for the most part, continue to focus their questions on differences in abiotic factors between the two environments. These approaches have greatly improved our understanding of both vegetation types, but much remains to be done in order to understand how these ‘opposites’ thrive in such close proximity.

Photos: L.Warman Characteristic contrast in canopy closure between rainforest and woodland.

The main aim of this thesis is to contribute towards a cohesive understanding of the dynamics between rainforest and fire-dependent vegetation, particularly in the

Australian Wet Tropics. In other words, my aim is to approach the vegetation of Far

North Queensland as a single dynamic, complex system, rather than as two contrasting

3 General introduction

biological systems which grow side by side. In part this is achieved by using ideas based on alternative stable states (ASS) theory.

Thinking in terms of alternative stable states is quite common in studies of aquatic and marine systems (Scheffer et al. 1993, Hughes et al. 2007, Richardson et al. 2009,

Baskett and Salomon 2010), arid and semi-arid systems (Heffernan 2008, Odion et al.

2009), as well as in discussions of management and restoration, particularly in agroecology (Westoby et al. 1989, van de Koppel et al. 1997, Anderies et al. 2002,

Suding et al. 2004). However, until quite recently, few people had applied the idea to rainforests. One of the main advantages of ASS is that it allows one to think in terms of interactions. Rather than expecting a single ecological community as a response to a combination of abiotic factors, with ASS we would expect to find more than one ecosystem or community state for any particular set of environmental conditions

(Scheffer and Carpenter 2003). Perhaps more importantly though, thinking in terms of

ASS incorporates feedbacks created by the vegetation and interactions between vegetation types (Odion et al. 2009) as well as slow or gradually changing conditions

(such as climate; Scheffer et al. 2001, Scheffer et al. 2005). In other words, while not disregarding the importance of abiotic factors, ASS allows us to consider a series of scenarios where feedbacks created by the vegetation are equally as important as the abiotic conditions.

‘Rain forests vs. flame forests’ rainforest vegetation in Australia

Debates over what defines Australian rainforest vegetation and what controls their distribution across the continent have long raged among Australian scientists. While the term ‘rainforest’ is evocative of a certain type of vegetation for scientists and non-

4 General introduction

scientists alike (Adam 1992), it has proven considerably difficult and contentious to pinpoint and strictly define what constitutes rainforest in Australia (Webb and Tracey

1981). Currently the term ‘rainforest’ encompasses vegetation that ranges from the ‘dry rainforests’, ‘monsoon rainforests’ and ‘tropical rainforests’ of Northern Australia, through to the subtropical rainforests of Queensland and New South Wales, to the temperate rainforests of Tasmania (Adam 1992, Bowman 2000). At varying times, different authors have defined Australian rainforest in terms of degrees of canopy height and/or closure (e.g. Wood and Williams 1960, Specht et al. 1981), canopy closure, species composition and/or vegetation structure (e.g. Beadle and Costin 1952, Baur

1957) leaf anatomy and/or physiology (e.g. McLuckie and Petrie 1927) or presumed biogeographic origin (e.g. Hooker 1860). Furthermore, rainforests have been further classified into subgroups based on presence of ‘diagnostic life forms’ (Webb 1959), classified biogeographically into ecofloristic provinces (Webb et al.1984) and in the context of regional vegetation (Bowman 2000). Debate over the nature of rainforest and factors that limit their extent and location have centred on a series of abiotic factors which include water availability (e.g. Webb and Tracey), soil nutrients, particularly phosphorus and fertility (e.g. Beadle 1962, Webb 1969), soil parent material and topography (e.g. Ash 1988, Bowman 2000) and soil physical properties (e.g. Tracey

1969). However, it is generally agreed that fire plays a particularly important role in defining both the nature and extent of rainforest in Australia (Jackson 1968, Ash, 1988,

Bowman 2000 and 2001, Lynch and Neldner 2000).

The Australian Wet Tropics

Rainforests (including monsoon forests, tropical, subtropical and temperate rainforests) have been estimated to cover altogether less than 0.3% of Australia (Winter et al. 1987).

5 General introduction

The Australian Wet Tropics bioregion contains the greatest extent of rainforest in

Australia, and the best examples of tropical rainforest on the continent. While the region boasts 783,000 ha of tropical rainforest, 60% of the region (out of 2,138,000 ha) is composed of various types of fire-dependent sclerophyll vegetation (Williams et al.

1996). Almost 900,000 ha of the Wet Tropics are contained within the Wet Tropics

World Heritage Area (Trott, 1996).

Location of the Wet Tropics (and World Heritage Area) within Australia

Considering its small area, the Wet Tropics is home to a disproportionate amount of biodiversity. In less than 0.2% of Australia, the region contains more than 3,500 species of plants (about 25% of all Australian species, including 65% of its ferns; Trott 1996,

Moritz 2005), 28% of all Australian vertebrates (including 83 regionally endemic species; Williams et al. 2008) and 62% of Australian butterfly species (Moritz 2005).

Much of the animal biodiversity in the region is concentrated in specific areas, and this has been attributed to both altitudinal and habitat preferences (Williams et al. 1996).

6 General introduction

Approaching the vegetation dynamics of a large dynamic system like the Wet Tropics of Australia presents a series of challenges for an ecologist trying to understand long- term processes within a limited amount of time. It is a large area with complex, often rugged topography; dense vegetation and great expanses which are very difficult to access. The region features both a wet season (which brings cyclones, floods and impassable roads) and a dry one (when fires can unexpectedly consume prospective field sites and populations of herbivorous insects dwindle). In many accessible areas, much of the natural vegetation has been supplanted by agriculture and (sub)urbanization

(Stork et al. 2008) which is encroaching much as it does elsewhere in the world.

Perhaps the most crucial hurdle to understanding long-term processes over vast areas, is that the life-spans of many of the organisms involved and the critical processes which one seeks to observe occur over far longer periods of time that the ones allocated to the research time allowed to a PhD thesis.

However, in this regard we have gained several important advantages that Dr. Domin could have only dreamed of a century ago. For example, in contrast to most tropical areas, good palaecological records now exist for the Wet Tropics. Fossil deposits in inland Queensland (such as Riversleigh; Moritz 2005) have given significant insights into the fauna of prehistoric Australia. More importantly, pollen cores from Lynch’s

Crater and Lake Euramoo on the Atherton Tablelands and offshore cores from the Great

Barrier Reef have provided relatively complete records of past vegetation extending back some 230,000 years (Haberle 2005, Kershaw et al. 2007). Indeed, by pooling palynological data from coastal Queensland and New Caledonia, we probably know more about the region’s palaeo-environment than we do for anywhere else in Australia, and other rainforests globally (Schneider et al. 1998, Stevenson and Hope 2005).

7 General introduction

Ecologists now possess powerful tools including ideas about complex systems that have been tested in a variety of environments, computer modelling, access to sophisticated chemical analysis and a better understanding of interactions between plants, other organisms and the ecosystems in which they occur. In this thesis I have used several of these tools to try and answer the problem of the “most interesting contact between open forests and vine scrubs” (Domin 1911). However, one of the greatest advantages of starting a project such as this at the start of the twenty first century is that one can follow in the footsteps of many others who have been equally fascinated by the region.

Thesis structure

Chapter Two presents a general explanation of alternative stable state theory and the reasoning under which the fire-dependent and rainforest vegetation of Australian Wet

Tropics can be considered as ASS. I relate how existing field evidence from the region relates to ‘lines of evidence’ that have been proposed for systems supporting ASS. More importantly, I propose different scales at which vegetation interacts with water and fire, and a new context for understanding wet sclerophyll forests.

Chapter Three quantifies herbivory in rainforest and adjacent wet sclerophyll forest and woodland. Studies around the world and in a variety of environments have shown that herbivores can play an important role in maintaining boundaries and alternative stable states. Furthermore, in this chapter I compare herbivory and chemical and physical traits in compound and simple leaves. I ask whether compound or simple leaves receive more herbivore damage and whether this relates to differences in leaf traits in either leaf type, as has been previously hypothesized by several studies.

8 General introduction

In Chapter Four, I compare soil chemistry directly across abrupt boundaries and with control sites in both woodland and rainforest well away from boundaries. Soil has long been thought to play an important role in defining the position of abrupt vegetation boundaries. However, the lack of non-boundary controls makes it difficult to interpret the significance of previous cross-boundary contrasts. Similarly, the majority of previous studies focus on a single boundary or a single location, which makes it hard to interpret results at a regional scale. I ask whether cross-boundary contrasts in the

Australian Wet Tropics are actually greater than natural variation within either rainforests or woodlands, and whether there is evidence of boundaries representing

‘rainforest soil’ abutting against ‘woodland soil’. I also test whether, as rainforest invades wet sclerophyll forest, the soil chemistry under the latter changes to resemble that of the former.

Chapter Five uses a computer model to evaluate the effects on landscape-scale vegetation patterns of changing species interactions and environmental conditions. In the model, obligate rainforest and pyrophytic ‘species’ interact with pioneers from each vegetation type. Pioneer species act either as pioneers in the classic sense or as

‘traitors’, taking advantage of the opposite vegetation type’s preferred conditions to facilitate conditions necessary for obligate species of their own type. The model takes into account changing environmental conditions as well as competition and facilitation occurring at several scales within the vegetation. This chapter proposes long-term hypotheses that may be tested in the field.

Chapters Two to Five in this thesis have been written for stand-alone publication in ecological journals. Some overlap has been unavoidable, especially in references and

9 General introduction

introductory sections. I have, however, kept references at the end of each chapter for ease of reference and to keep the original “flow”. I have also kept the use of “we” and

“our” throughout the chapters because this is how they have been submitted to and/or appeared in publication. Contributions by co-authors are stated at the start of each chapter.

Photo: L.Warman

References

Adam, P. 1992. Australian Rainforests. - Oxford University Press.

Anderies, J. M., Janssen, M. A. and Walker, B. H. 2002. Grazing Management,

Resilience, and the Dynamics of a Fire-driven Rangeland System. - Ecosystems 5:

23-44.

Ash, J. 1988. The Location and Stability of Rainforest Boundaries in North-Eastern

Queensland, Australia. - Journal of Biogeography 15: 619-630.

Baskett, M. L. and Salomon, A. K. 2010. Recruitment facilitation can drive alternative

states on temperate reefs. - Ecology 91: 1763-1773.

10 General introduction

Baur, G. N. 1957. Nature and distribution of rain-forests in New South Wales. -

Australian Journal of Botany 5: 190-233.

Beadle, N. C. W. 1962. Soil Phosphate and the Delimitation on Plant Communities in

Eastern Australia II. - Ecology 43: 281-288.

Beadle, N. C. W. and Costin, A. B. 1952. Ecological classification and nomenclature. -

Proceedings of the Linnean Society of New South Wales 77: 61-82.

Beauvais, A. 2009. Ferricrete biochemical degradation on the rainforest-savannas

boundary of Central African Republic. - Geoderma 150: 379-388.

Bowman, D. M. J. S. 2000. Australian Rainforests: Islands of green in a land of fire. -

Cambridge University Press.

Bowman, D. M. J. S. 2001. On the elusive definition of 'Australian rainforest': response

to Lynch and Neldner (2000). - Australian Journal of Botany 49: 785-787.

Domin, K. 1911. Queensland's Plant Associations (Some problems of Queensland's

Botanogeography). - Proceedings of the Royal Society of Queensland 23: 57-74.

Duff, G. A. 1987. Physiological ecology and vegetation dynamics of North Queensland

upland rainforest-open forest ecotones. PhD Thesis, Department of Botany. -

James Cook University.

Goetze, D., Hörsch, B. and Porembski, S. 2006. Dynamics of forest–savanna mosaics in

north-eastern Ivory Coast from 1954 to 2002 -Journal of Biogeography 33: 653-

664.

Haberle, S. G. 2005. A 23,000-yr pollen record from Lake Euramoo, Wet Tropics of NE

Queensland, Australia. - Quaternary Research 64: 343-356.

Heffernan, J. B. 2008. Wetlands as an alternative stable state in desert streams. -

Ecology 89: 1261-1271.

11 General introduction

Hennenberg, K., Fischer, F., Kouadio, K., Goetze, D., Orthmann, B., Linsenmair, K.,

Jeltsch, F. and Porembski, S. 2006. Phytomass and fire occurrence along forest-

savanna transects in the Comoé National Park, Ivory Coast. - Journal of Tropical

Ecology 22: 303-311.

Hoffmann, W. A., Adasme, R., Haridasan, M., T. de Carvalho, M., Geiger, E. L.,

Pereira, M. A. B., Gotsch, S. G. and Franco, A. C. 2009. Tree topkill, not

mortality, governs the dynamics of savanna-forest boundaries under frequent fire

in central Brazil. - Ecology 90: 1326-1337.

Hooker J.D. 1860. The botany of the Antarctic voyage of H.M. Discovery ships Erebus

and Terror in the years 1839-1843. Part III.Flora Tasmaniae. Vol I.

Dicotyledones. Reeve Brothers, London (available at www.archive.org)

House, A. P. N. 1986. Seed exchange and storage across rainforest- sclerophyll

woodland boundaries in north Queensland. - Australian National University.

Hughes, T. P., Rodrigues, M. J., Bellwood, D. R., Ceccarelli, D., Hoegh-Guldberg, O.,

McCook, L., Moltschaniwskyj, N., Pratchett, M. S., Steneck, R. S. and Willis, B.

2007. Phase Shifts, Herbivory, and the Resilience of Coral Reefs to Climate

Change. - Current Biology 17: 360-365.

Jackson, W. D. 1968. Fire, air, water and earth- an elemental ecology of Tasmania. -

Proceedings of the Ecological Society of Australia 3: 9-16.

Kershaw, A. P., Bretherton, S. C. and van der Kaars, S. 2007. A complete pollen record

of the last 230 ka from Lynch's Crater, north-eastern Australia. -

Palaeogeography, Palaeoclimatology, Palaeoecology 251: 23-45.

Lynch, A. J. J. and Neldner, V. J. 2000. Problems of placing boundaries on ecological

continua - options for a workable national rainforest definition in Australia. -

Australian Journal of Botany 48: 511-530.

12 General introduction

Mariotti, A. and Peterschmitt, E. 1994. Forest savanna ecotone dynamics in India as

revealed by carbon isotope ratios of soil organic matter. - Oecologia 97: 475-480.

Markham, R. H. and Babbedge, A. J. 1979. Soil and Vegetation Catenas on the Forest-

Savanna Boundary in Ghana. - Biotropica 11: 224-234.

McLuckie, J. and Petrie, A.H.K. 1927. An ecological study of the flora of Mt. Wilson.

Proceedings of the Linnean Society of New South Wales 52: 161-184

Moritz, C. 2005. Rainforest history and dynamics in the Australian Wet Tropics. - In:

Bermingham, E., Dick, C. W. and Moritz, C. (eds.), Tropical Rainforests: Past,

present and future. University of Chicago Press, pp. 313-321.

Odion, D. C., Moritz, M. A. and DellaSala, D. A. 2009. Alternative community states

maintained by fire in the Klamath Mountains, USA. - Journal of Ecology 98: 96-

105.

Perry, G. L. W. and Enright, N. J. 2002. Spatial modelling of landscape composition

and pattern in a maquis-forest complex, Mont Do, New Caledonia. - Ecological

Modelling 152: 279-302.

Puyravaud, J.-P., Pascal, J.-P. and Dufour, C. 1994. Ecotone Structure as an Indicator of

Changing Forest-Savanna Boundaries (Linganamakki Region, Southern India). -

Journal of Biogeography 21: 581-593.

Richardson, A. J., Bakun, A., Hays, G. C. and Gibbons, M. J. 2009. The jellyfish

joyride: causes, consequences and management responses to a more gelatinous

future. - Trends in Ecology & Evolution 24: 312-322.

Scheffer, M. and Carpenter, S. R. 2003. Catastrophic regime shifts in ecosystems:

linking theory to observation. - Trends in Ecology & Evolution 18: 648-656.

Scheffer, M., Hosper, S. H., Meijer, M. L., Moss, B. and Jeppesen, E. 1993. Alternative

equilibria in shallow lakes. - Trends in Ecology & Evolution 8: 275-279.

13 General introduction

Scheffer, M., Holmgren, M., Brovkin, V. and Claussen, M. 2005. Synergy between

small- and large-scale feedbacks of vegetation on the water cycle. - Global

Change Biology 11: 1003-1012.

Schneider, C. J., Cunningham, M. and Moritz, C. 1998. Comparative phylogeography

and the history of endemic vertebrates in the Wet Tropics rainforests of Australia.

- Molecular Ecology 7: 487-498.

Specht, R. L. and Morgan, D. G. 1981. The balance between the foliage projective

covers of overstorey and understorey strata in Australian vegetation. - Austral

Ecology 6: 193-202.

Stevenson, J. and Hope, G. 2005. A comparison of late Quaternary forest changes in

New Caledonia and northeastern Australia. - Quaternary Research 64: 372-383.

Stork, N. E., Goosem, S. and Turton, S. M. 2008. Australian Rainforests in a global

context. - In: Stork, N. E. and Turton, S. M. (eds.), Living in a Dynamic Tropical

Forest Landscape. Blackwell Publishing, Ltd, pp. 5-20.

Suding, K. N., Gross, K. L. and Houseman, G. R. 2004. Alternative states and positive

feedbacks in restoration ecology. - Trends in Ecology & Evolution 19: 46-53.

Tracey, J. G. 1969. Edaphic Differentiation of some Forest Types in Eastern Australia:

I. Soil Physical Factors. - Journal of Ecology 57: 805-816.

Trott, L. 1996. Wet Tropics in profile: A reference guide to the Wet Tropics of

Queensland World Heritage Area. - Wet Tropics Management Authority.

Unwin G.L. 1983. Dynamics of the rainforest - eucalypt forest boundary in the

Herberton Highland, North Queensland. MSc Thesis, James Cook University:

Townsville.

14 General introduction

van de Koppel, J., Rietkerk, M. and Weissing, F. J. 1997. Catastrophic vegetation shifts

and soil degradation in terrestrial grazing systems. - Trends in Ecology &

Evolution 12: 352 - 356.

Virah-Sawmy, M. 2009. Ecosystem management in Madagascar during global change. -

Conservation Letters 2: 163-170.

Webb, L. J. 1969. Edaphic Differentiation of Some Forest Types in Eastern Australia:

II. Soil Chemical Factors. - Journal of Ecology 57: 817-830.

Webb, L. J. and Tracey, J. G. 1981. Australian rainforests: patterns and change. - In:

Keast, A. (ed.) Ecological Biogeography of Australia. Dr. W. Junk, pp. 605-694.

Webb, L. J., Tracey, J. G. and Williams, W. T. 1984. A floristic framework of

Australian rainforests. - Austral Ecology 9: 169-198.

Westoby, M., Walker, B. and Noy-Meir, I. 1989. Opportunistic management for

rangelands not at equilibrium. -.

Williams, S. E., Isaac, J. L., Graham, C. and Moritz, C. 2008. Towards an

understanding of vertebrate biodiversity in the Australian Wet Tropics. - In: Stork,

N. E. and Turton, S. M. (eds.), Living in a Dynamic Tropical Forest Landscape.

Blackwell Publishing, Ltd, pp. 133-149.

Williams, S. E., Pearson, R. G. and Walsh, P. J. 1996. Distributions and biodiversity of

the terrestrial vertebrates of Australia's Wet Tropics: a review of current

knowledge. - Pacific Conservation Biology 2: 327-362.

Winter, J. F., Atherton, R. G., Bell, E. C. and Pahl, L. I. 1987. An introduction to

Australian rainforests. - Australian Heritage Commission.

Wood, J. G. and Williams, R. J. 1960. Vegetation. - In: Leeper, G. W. (ed.) The

Australian environment. CSIRO and Melbourne University Press, pp. 67-84.

Photo: L.Warman

Chapter Two

Alternative stable states in Australia's wet tropics: a theoretical framework for the field data and a field

case for the theory.

Laura Warman and Angela T. Moles

Landscape Ecology (2009) 24: 1-13 DOI: 10.1007/s10980-008-9285-9

This paper was conceived by LW, who carried out the literature review and wrote the manuscript. Extensive supervision, guidance and corrections were provided by AM.

This chapter has been modified from the original paper specifically for this thesis.

17 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Abstract:

The vegetation of the Wet Tropics bioregion of Far North Queensland is a

complex system whose components (mainly tropical rainforests and fire-prone

forests and woodlands) have mostly been studied independently from each other.

We suggest that many characteristics of the vegetation are consistent with those

of a complex, dynamic, spatially heterogeneous system which exhibits

alternative stable states. We propose these states are driven and maintained by

the interaction of vegetation-specific positive feedback loops with the regions’

environmental parameters (such as topography, steep humidity gradients and

seasonality) and result in the rainforest/fire-prone vegetation mosaic that

characterises the area. Given the regions’ magnitude, biodiversity and

complexity, we propose the Wet Tropics as an important new example and a

good testing ground for alternative stable state and resilience theories in large

heterogeneous natural systems. At the same time, thinking in terms of alternative

stable states and resilience creates a new context for understanding the regions’

biological dynamics.

18 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Introduction

There is growing interest in resilience theory and the existence of alternative stable states (hereafter ASS) in nature. Increasing evidence indicates that biological systems ranging from cells (Angeli et al. 2004) to species assemblages (Konar and Estes 2003; van Nes and Scheffer 2004) and ecosystems (see Folke et al. 2004) can be found in alternative stable states. In these systems more than one stable state or regime is possible for a given set of environmental conditions (see Scheffer et al. 2001; Schröder et al. 2005). Each state is controlled by a different set of processes and rapid shifts can occur between states (Scheffer et al.2001; Folke et al. 2004). Regime shifts have been documented in a wide range of ecosystems including freshwater lakes that are turbid or clear depending on nutrient-loading (Scheffer et al. 1993; Carpenter et al. 1999); grasslands and woodlands where tree or grass density and dominance can depend on herbivores and/or fire (Westoby et al. 1989; Dublin et al. 1990); arid and semi-arid zones where plant-soil and plant-water relations influence plant cover (Rietkerk and van de Koppel 1997; Scheffer et al. 2001, 2005); desert streams and wetlands that depend on the interaction of geomorphic and biotic factors coupled to disturbances (Heffernan

2008); marine systems such as reefs where algal or coral dominance is linked to overfishing or catastrophic events (McManus and Polsenberg 2004; Hughes et al. 2007) or kelp beds adjacent to urchin ‘barrens’ (Konar and Estes 2003).

In this paper we suggest that many of the characteristics of the vegetation in the Wet

Tropics of Australia are consistent with those of systems supporting alternative stable states. ASS theory provides an alternative, more congruent set of solutions for ongoing debates on regional vegetation dynamics, and a unifying theoretical framework for seemingly disparate evidence from the field. Furthermore we propose that Queensland’s

19 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Wet Tropics represent a good testing ground for ASS in complex heterogeneous systems because of particular vegetation characteristics, such as strongly contrasting positive feedback loops. Understanding the nature and interaction of these feedback loops in Queensland may prove a useful exercise for finding similar dynamics in other complex heterogeneous systems in Australia and around the world.

How the Australian Wet Tropics are understood today: a complex mosaic

The Wet Tropics of Far North Queensland, Australia contain many different vegetation types which create a complex, dynamic mosaic (Hopkins et al. 1993; Hilbert et al.

2001). The region, which covers approximately 1.8 million hectares (Williams 2006), is characterised by a fragmented band of humid tropical rainforests interspersed with fire- prone sclerophyll vegetation (including grasslands, open woodlands and eucalypt forests) (Unwin 1989; Hopkins et al. 1993; Hilbert et al. 2001). Throughout the Wet

Tropics, these two major vegetation types can be found side by side in patches of different sizes ranging from small grasslands and clumps of a few trees to rainforest massifs comprising thousands of hectares (Unwin 1989; Hopkins et al. 1993). Further inland, sclerophyll vegetation becomes wholly dominant, while to the north the region is bound by both sclerophyll vegetation and more seasonal rainforest vegetation (see

Russell-Smith et al. 2004a, b). The region is less strongly seasonal than the Wet/Dry (or monsoon) Tropics (for detailed floristic descriptions as well as distribution and location of vegetation types in the region see Ash 1988; Unwin 1989; Hopkins 1993; Harrington and Sanderson 1994; Harrington et al. 2005; Williams 2006; Hilbert et al. 2007).

Abrupt boundaries between rainforest and non-rainforest vegetation are a notable and puzzling feature of the Wet Tropics. These borders commonly have only a few metres,

20 Chapter Two- Alternative stable states in Australia’s Wet Tropics

if any, of ecotone and are relatively stable, at least in the short term (Ash 1988; Unwin

1989; Adam 1992; Bowman 2000a). Wide ecotonal boundaries are less common and less stable (Ash 1988; Turton and Sexton 1996). Light, temperature, water and fire regimes within rainforests differ markedly from those of abutting sclerophyll vegetation

(Adam 1992; Turton and Sexton 1996).

The location of boundaries, their relative stability and the reason(s) they shift through time have long puzzled researchers. To date, the distribution of vegetation types and boundaries have been generally attributed to a varying set of interactions between abiotic factors. These factors include topography, rainfall gradients, soil type and fertility, site history and seasonal water stress (see Ash 1988; Unwin 1989; Adam 1992;

Turton and Sexton 1996; Bowman 2000a). Fire has been shown to play a particularly significant role in defining vegetation patterns and demarcating borders between sclerophyll forests and rainforests throughout Australia (Adam 1992; Bowman 2000a;

Williams 2000).

The importance of fire and water

Australian vegetation is often grouped into two broad categories, namely ‘pyrophobic’ and ‘pyrophytic’, depending on its tolerance to fire (Ash 1988; Bowman 2000a).

Pyrophytic vegetation burns readily and regularly. It is dominated by fire-adapted and fire-dependent species whose reproduction is typically aided by fire. These species are relatively unable to maintain and/or regenerate under a closed, shady canopy (Ash

1988). Conversely, ‘pyrophobic’ vegetation does not burn as easily (Ash 1988; Adam

1992) and is dominated by species considered to be fairly ‘fire intolerant’ (Ash 1988;

21 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Williams 2000). These species are generally regarded as especially susceptible to frequent fires.

In the Wet Tropics, open sclerophyll vegetation is pyrophytic, while the rainforests with closed canopies and shady, humid understories are considered to be pyrophobic. In the absence of fire, rainforest vegetation is able to infiltrate the boundaries and become dominant in areas previously occupied by pyrophytic vegetation (Hopkins et al. 1993;

Harrington and Sanderson 1994; Bowman 2000a).

The effect of fire as an environmental trigger on pyrophytic and pyrophobic vegetation is non-linear. In other words, vegetation types as a whole respond abruptly to environmental conditions created by the presence or absence of fire (see Scheffer et al.

2001; Scheffer and Carpenter 2003) (Fig. 1b). However, the effects of fire are strongly dependent on frequency and intensity (Jackson 1968, Bowman 2000a). The consequences of burning also depend on what species are present, as individual species have a varying range of tolerance and response to fire (e.g. Bowman 2000a; Williams

2000; Marrinan et al. 2005).

Like fire, rainfall gradients and seasonality strongly affect the species and vegetation types present in an area (e.g. ‘dry rainforests’ or monsoon rainforests) and how these interact with neighbouring vegetation types (see Fensham et al. 2003; Russell-Smith et al. 2004a). Annual water availability and seasonality (including the intensity and duration of dry seasons) also act as qualifiers for the effects of fire (Bowman and

Panton 1993; Russell-Smith et al. 2004a). For example, extended droughts coupled with

22 Chapter Two- Alternative stable states in Australia’s Wet Tropics

seasonal dry seasons strongly increase the likelihood and consequences of fire in otherwise fire-resistant rainforest areas (Marrinan et al. 2005).

Fig. 1 Two interpretations of current approaches to understanding vegetation

dynamics in the Australian Wet Tropics (after Scheffer et al. 2001; Scheffer and

Carpenter 2003): a As a linear, monostable system. In this case, ecosystems

occur along a gradient of environmental parameters and only one ecosystem

state is possible for any given set of environmental parameters. b As a non-

linear, monostable system where fire acts as a trigger which encourages the

pyrophytic vegetation and constrains the pyrophobic (rainforest) vegetation.

While neither of these interpretations is entirely incorrect, both ignore the

interactions between vegetation types. The second approach only partly

recognizes the interaction between vegetation and the environment.

23 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Large continuous areas of rainforest predominate in areas like the Daintree, which have very wet, aseasonal conditions. Likewise, drier areas with more fire-prone conditions support extensive pyrophytic communities. However, external parameters alone are not able to explain all current vegetation patterns. For example, not all areas in the Wet

Tropics with the appropriate environmental conditions currently support rainforest vegetation (Adam 1992). Likewise, pyrophytic vegetation can be found in isolated pockets surrounded by rainforest (Ash 1988) and dry rainforests in inland Far North

Queensland are able to maintain themselves in relatively arid conditions (e.g. Fensham

1995; Hansman 2001). In these cases, we suggest that internal drivers in the form of positive feedback loops allow the vegetation to maintain its stability.

An alternative paradigm: alternative stable states

Alternative states are also known as basins or domains of attraction and have been likened to dips or valleys in a three-dimensional ‘stability landscape’ (see Beisner et al.

2003; Scheffer and Carpenter 2003) (Fig. 2). Both external drivers and internal feedbacks contribute to shaping each valley and maintaining or shifting the ecosystem configuration from one basin of attraction into another (Scheffer et al. 2001; Beisner et al. 2003).

A bistable system ‘‘toggles between two discrete, alternative stable steady states, in contrast to a monostable system, which slides along a continuum of steady states’’

(Angeli et al. 2004). To date, the vegetation of the Wet Tropics has been approached as a monostable system, where vegetation types are distributed along a gradient of abiotic factors, and where fire encourages pyrophytic vegetation while limiting rainforest expansion (Fig. 1).

24 Chapter Two- Alternative stable states in Australia’s Wet Tropics State variables State variables

Environmental parameters

Fig. 2 A ‘stability landscape’ (defined by state variables and environmental

parameters, as well as the interactions between them; after Beisner et al.

2003; Scheffer et al. 2001; Scheffer and Carpenter 2003), for the vegetation of

the Wet Tropics. a and c correspond to ecosystem configurations as rainforest

or pyrophytic vegetation. These alternative states represent resilient basins of

attraction, so it takes a strong change in environmental parameters (which

changes the “depth” of the basin) or positive feedbacks (which push the ball)

to shift the ecosystem configuration. b Is an unstable or moderately stable

state between the basins. It is stable under a very particular, narrow set of

conditions, but lacks resilience and will easily slide into either basin if these

conditions (and/or feedbacks) change. We suggest that in Far North

Queensland, this ridge is occupied by specific types of wet sclerophyll

vegetation.

We propose that what we see in Far North Queensland is a dynamic, complex, non- linear system where rainforest and pyrophytic vegetation occur as alternative stable

25 Chapter Two- Alternative stable states in Australia’s Wet Tropics

ecosystem states, rather than as independent systems (Fig. 3). Under this scenario, rather than fire acting as an independent environmental trigger or constraint, we propose both fire and water have a three-fold role and interact with vegetation at several scales:

(1) Fire and water are both important components of external, large-scale

environmental parameters such as climate and seasonality. As such, both fire

frequency and water availability act to define suitable areas for vegetation to

establish and maintain itself according to broad habitat requirements and preferences.

(2) Locally, both fire and water act as limited resources for plants. Fire can be

considered a limited resource in the case of obligate pyrophytic species that depend

on fire for reproduction and/or to maintain themselves. The equivalent occurs for

obligate rainforest species that have minimum water availability/constancy

requirements. This acts as a further filter defining what communities and species can

establish, maintain themselves and outcompete others in any given area.

(3) Both fire and water form part of biological feedback loops whereby the vegetation

locally influences the frequency and intensity of fire as well as the availability of

water. In other words, pyrophytic vegetation requires and encourages fire, which

discourages rainforest; meanwhile rainforests require and encourage conditions with

higher water availability, which make fire more unlikely. Rather than simply

reiterating that pyrophytic vegetation burns and rainforests are intolerant of fire, we

suggest that both vegetation types in the region actively modify environmental

parameters.

26 Chapter Two- Alternative stable states in Australia’s Wet Tropics

In short, both fire and water act as external parameters and as ‘patchy’ resources whose presence and changing distribution can affect and constrain vegetation. However, once established, the vegetation itself is able to promote and maintain more amenable environmental conditions. Both positive water-vegetation and positive fire-vegetation feedbacks have been detailed for vegetation elsewhere (e.g. Bond and Midgley 1995;

Laurance and Williamson 2001; Sternberg 2001; Scheffer et al. 2005; Borgogno et al.

2007), but what makes the Wet Tropics system interesting is that it presents contrasting feedback loops associated with particular vegetation types. So, not only does each vegetation actively enhance environmental conditions it requires, but these conditions simultaneously hinder or constrain the opposite vegetation type.

The processes implied in maintaining alternative stable states in natural systems occur at different scales of space, time and levels of ecological organization (see Levin 2000;

Scheffer et al. 2001; Folke et al. 2004; Petraitis and Latham 1999; van Nes and Scheffer

2004). There are also interactions between scales (such as the interactions of fast and slow variables, like aridification and pest outbreaks, with local and regional processes) and the cumulative effects of small or gradual changes (e.g. Paine et al. 1998; Scheffer et al. 2001; van Nes and Scheffer 2004).

Water and fire feedback loops in the Wet Tropics probably act at several interrelated scales. Small scale water-vegetation feedbacks may be able to interact synergistically with precipitation on a larger scale (Scheffer et al. 2001, 2005). For example, some of the effects of seasonality may be ameliorated in patches of rainforest vegetation through positive feedback processes similar to ‘amplification of local moisture recycling’ (see

Sternberg 2001; Scheffer et al. 2005).

27 Chapter Two- Alternative stable states in Australia’s Wet Tropics

28 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Fig. 3 Building an alternative state scenario for the Wet Tropics (after Scheffer et al. 2001; Scheffer and Carpenter 2003), where vegetation types correspond to alternative stable states along an axis of “effective wetness”. This takes into account combined environmental parameters (including rainfall, seasonality, topography, soil type and fertility, etc…) as well as the vegetations’ interaction with these parameters via species habitat preferences and positive feedbacks. a The scenario for rainforest vegetation. Under effectively wet conditions, the vegetation behaves as a monostable system where rainforest vegetation is dominant and resilient. As “effective aridity” increases, resilience diminishes and the state encounters a “pyrophobic threshold” (the point at which neither feedbacks nor environmental parameters protect the vegetation from fire) where rainforest suddenly gives way to pyrophytic vegetation. b The equivalent scenario for pyrophytic vegetation, which is dominant under more arid conditions. This state reaches an “ombrophobic threshold” when the vegetation can no longer compete with, and is excluded by, rainforest vegetation. c When these scenarios are overlapped, the result is a nonlinear, bistable system that presents a range of conditions where effective wetness is equally suitable for either vegetation type. In these areas vegetation types coexist, compete and dominance shifts between them. Under this scenario, wet sclerophyll forests become “momentarily” stable at a point where feedbacks and abiotic conditions balance each other out

29 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Isolated pockets of a given vegetation type may be maintained by a few species that act as environmental engineers via feedback loops (sensu Guevara et al. 1992). For example, a few eucalypts that burn much hotter than the surrounding rainforest vegetation might be able to maintain fire-related feedback loops in a pyrophytic pocket with few fires (see Ash 1988). If these engineers are removed, then the pyrophytic feedbacks become unable to maintain the pocket and the vegetation turns into rainforest.

Along these lines, Hopkins et al (1996) described rainforest excluding pyrophytic vegetation after a cyclone destroyed the eucalypt forest canopy.

Is there evidence of alternative stable states in Far North Queensland?

Many of the distinctive features of the vegetation in the Wet Tropics correspond with those proposed as ‘lines of evidence’ for systems supporting alternative stable states

(e.g. Scheffer and Carpenter 2003; Schröder et al. 2005). These include the sharp boundaries between vegetation types and the formation of landscape-scale mosaics; two of the most prominent yet puzzling characteristics of the vegetation in the Wet Tropics.

Sharp transitions between alternative regimes in space

Rainforest and pyrophytic vegetation occur side by side under the same general environmental parameters throughout Far North Queensland. The sharp boundaries between them are often attributed to abiotic environmental factors (including soil and topography), but these factors alone have limited predictive power over both vegetation type and boundary location (see Bowman 2000a). In bistable systems, sharp transitions often arise between alternative stable states (Scheffer and Carpenter 2003; Schröder et al. 2005) and these borders maintain themselves because of internal state drivers. In

30 Chapter Two- Alternative stable states in Australia’s Wet Tropics

other words the drivers make each state relatively uninvasible by the other, despite external environmental conditions (see Konar and Estes 2003).

Rapid shifts between alternative stable states in time

Sharp transitions in time sequences occur in systems with alternative stable states

(Konar and Estes 2003; Scheffer and Carpenter 2003; Schröder et al. 2005). There is congruent palaeoecological and present day evidence of rapidly shifting boundaries between rainforest and pyrophytic vegetation throughout northern Australia (e.g.

Hopkins et al. 1993; Harrington and Sanderson 1994; Hopkins et al. 1996; Russell-

Smith et al. 2004a). In some areas rainforest margins are currently advancing more than a metre per year (Unwin 1989; Adam 1992) and areas known to have maintained pyrophytic vegetation only 200 years ago are now entirely covered by large areas of rainforest (Hopkins et al. 1996). In particular, the pollen record from Lynch’s Crater in the Wet Tropics indicates three major and relatively rapid shifts between dominance of rainforest and sclerophyll vegetation approximately 9,000 years B.P., 38,000 years B.P. and 78,000 years B.P. (Kershaw 1994).

Response to slow variables

Rapid shifts between alternative stable regimes can result from response to external or slow-changing environmental variables that eventually reach a threshold (Scheffer et al.

2001; Scheffer and Carpenter 2003). In the case of the Wet Tropics, we propose these slow variables correspond to long-term climatic changes, particularly in environmental humidity and rainfall. Palaeoecological records from the Wet Tropics seem to be consistent with present day evidence from throughout northern Australia. These show that rapid rainforest expansion has been linked to ‘more amenable’ conditions such as

31 Chapter Two- Alternative stable states in Australia’s Wet Tropics

increasing rainfall and/or water availability coupled with, and ‘strongly mediated by’ release from frequent fires (Hopkins et al. 1993, 1996; Williams 2000; Russell-Smith et al. 2004a; b). This is also consistent with evidence from savannas around the world, which shows recent historical transitions from open vegetation towards more closed canopies (e.g. Puyravaud et al. 1994; Roques et al. 2001; Ngomanda et al. 2007).

Stability and maintenance

Rapid shifts in ecosystems can be associated with dynamic processes other than long- term bistability, such as population explosions and crashes due to environmental cues

(Scheffer and Carpenter 2003). The fundamental difference between these ‘speedy’ processes and stable state transitions, is that state transitions denote ‘substantial, long- term reorganizations of complex systems’ where the crucial feedbacks loops that maintain one state are replaced by the ones that maintain another one (Carpenter and

Brock 2006). Additionally, it’s expected that the new regime should maintain itself and persist ‘beyond the replacement of its initial resident individuals’ (Scheffer and

Carpenter 2003; Schröder et al. 2005). In other words, in an ASS context the difference between successional and transitional processes is that succession occurs within a basin of attraction while transitional processes denote a shift towards a new regime with different ecosystem drivers.

Much of the same evidence that denotes rapid shifts in time also shows alternating long- term dominance between rainforest and pyrophytic vegetation types (e.g. Hopkins et al.

1993, 1996; Russell-Smith et al. 2004a). In other words, there is evidence of long periods of stability, followed by very rapid switches between states. Congruent pollen and charcoal evidence show that, on a regional scale, sclerophyll vegetation seems to

32 Chapter Two- Alternative stable states in Australia’s Wet Tropics

have dominated the Wet Tropics throughout arid conditions of the last glacial period

(Hopkins et al. 1993, 1996; Kershaw 1994). Starting 8,000 years before present, as conditions became less arid, the rainforests rapidly expanded and colonized previously pyrophytic areas (Hopkins et al. 1993, 1996; Hilbert et al. 2007). This is consistent with present day studies in northern Australia, which track rainforest expansion as a result of less fire and more water (Banfai and Bowman 2006).

Resilience

Resilience can be defined as the amount or magnitude of change or perturbation a state can tolerate before it reaches a threshold that forces a shift into another regime (Scheffer et al. 2001; Folke et al. 2004). The more resilient a state is, the greater the basin of attraction it represents, and the harder it is for the ecosystem configuration to climb out of it (Scheffer et al. 2001; Beisner et al. 2003) (Fig. 2). The resilience of a state depends partly on general environmental parameters and partly on feedback loops (sensu

Scheffer and Carpenter 2003).

If a state is highly resilient, large perturbations may be unable to shift the system into another regime. For example, small or gradual changes are often to blame for

‘catastrophic regime shifts’ rather than large infrequent disturbances (LIDs) (Paine et al.

1998; Scheffer et al. 2001; van Nes and Scheffer 2004). In the Wet Tropics isolated events (such as a very wet season, a cyclone or a very fiery dry season) are usually incapable of causing an ecosystem-level state transition between vegetation types because of existing system-resilience. However, LIDs can act as catalysts for ecosystem state shifts when accompanied by significant changes in environmental parameters that alter or weaken the positive feedback loops. For example, in 2002 an El Niño event

33 Chapter Two- Alternative stable states in Australia’s Wet Tropics

following a long drought resulted in fires spreading through areas of rainforest that would not normally burn (Marrinan et al. 2005 see also Laurance and Williamson 2001;

Scheffer et al. 2001; Holmgren et al. 2006).

Regime shifts can also occur if environmental conditions are marginal- that is, near a threshold of habitat suitability. For example, if a cyclone damages rainforest vegetation in a wet area, the vegetation can revert to rainforest through coppicing and successional pathways like the filling in of light-gaps with light-demanding pioneer species (e.g.

Webb 1958). However if site conditions are near the thresholds required by either vegetation type, and the cyclone is followed by fuel accumulation and dry conditions conducive to fire, the area may shift to sclerophyll vegetation as described by Webb

(1958) and Unwin et al. (1988).

Implications: new insights into old problems—why ASS are a well-rounded paradigm for the Wet Tropics

Incorporating ideas about feedback loops and resilience creates a broader and more comprehensive conceptual framework for understanding the vegetation patterns of the

Wet Tropics. Rather than contradict the current paradigm, this framework builds upon it and integrates ideas and field evidence that can currently seem contradictory.

Furthermore, the ASS paradigm provides new insights and ways to examine longstanding quandaries about the vegetation in the Wet Tropics, such as the nature of wet sclerophyll forests and the role of edaphic compensation and rainforest refugia. It can also inform conservation and management.

Wet sclerophyll forests

34 Chapter Two- Alternative stable states in Australia’s Wet Tropics

In the Wet Tropics, wet sclerophyll forests (WSF, also known as tall open forests) are characterised by a sclerophyll overstorey with a variety of rainforest and pyrophytic species underneath (see Hopkins et al. 1993; Harrington and Sanderson 1994;

Harrington et al. 2005). Transition woodlands and mosaics also exist between fire-prone savannas and fire-sensitive forests in West Africa (Hopkins 1965; Goetze et al. 2006).

The dynamics of the ‘fire-zone’ forests described in Ghana (Swaine 1992) seem particularly relevant and interesting, especially because these forests appear to be more akin to the drier, more deciduous forests found elsewhere in northern Australia (where

WSF are absent). Another interesting contrast with Africa is that while savanna-forest ecotones there can be vast (Smith et al. 1997), the WSF band is rarely more than 4 km wide in the Wet Tropics (Harrington and Sanderson 1994).

Wet sclerophyll forests raise very interesting questions and represent an intriguing

(some might say vexing) problem for the definition of rainforest vegetation in Australia

(e.g. Adam 1992; Lynch and Neldner 2000; Bowman 2001). Because of the mix of plant species they contain, associated endemic fauna, and a tendency to shift rapidly toward rainforest when fire is excluded, there is ongoing debate over the nature and stability of WSF in the Wet Tropics. Some authors consider WSF to be ecotonal boundaries (Ash 1988). Others view them as a specific vegetation type instead of transitional or seral communities (see Turton and Sexton 1996).

Viewed within an ASS framework, the WSF problem becomes one of contrasting feedbacks, rather than the mixing of otherwise vegetation specific species. In other words, because both feedback loops are present and ‘‘pulling’’ the vegetation towards different attractors, WSF could represent a narrow ‘‘ridge’’ between the basins of

35 Chapter Two- Alternative stable states in Australia’s Wet Tropics

attraction for rainforest and pyrophytic vegetation (Fig. 2). As such they might represent a ‘moderately stable state’ under particular conditions, but not a third stable state.

In theory, when ASS occur over a range of environmental conditions, they are separated by an unstable state (Konar and Estes 2003; Schröder et al. 2005) (Figs. 2 and 3). This unstable state is expected to shift rapidly towards stability, however, in the real world this shift could correspond to a longer process; especially when it depends on long-lived organisms such as plants. If WSF represent an unstable state shifting towards stability, then different types of WSF (see Harrington et al. 2005) represent varying successional and transitional stages and could be identified according to their predominating feedback regime in addition to species composition.

Schröder et al. (2005) also raise the possibility of ‘‘transient systems’’ and ‘‘quasi- stable states’’ in which the system alternates between two basins of attraction but is never able to achieve stability. If WSF are transient systems that remain unstable in the long term, they might represent a stable community in time but not in space. In other words, they are a constant presence in the landscape, but they move as the boundary between vegetation types advances and retreats. Figuratively, WSF through time might behave similarly to surf breaking along a coastline. From any given point on the beach one sees individual waves arrive, break and disappear, but seen from a plane there is a constant line of breaking surf.

Edaphic compensation

Viewing positive feedback loops as internal drivers for alternative vegetation states strengthens and explains the otherwise confusing and sometimes contradicting role of

36 Chapter Two- Alternative stable states in Australia’s Wet Tropics

abiotic environmental factors as external drivers in the Wet Tropics. For example, interactions between soil fertility and topography have been linked to ‘edaphic compensation’ that allows rainforests to grow in drier areas that would otherwise be unsuitable (Ash 1988; Bowman 2000a). However to date it is not understood how edaphic compensation works, or why it is not a universal phenomenon (Bowman

2000a). In an ASS context, edaphic compensation can be interpreted as the result of water related positive feedbacks that allow these rainforests to maintain themselves.

The idea that rainforest vegetation presents positive feedbacks also provides a more robust explanation of how external and internal drivers work synergistically. When external ‘large scale’ environmental parameters (such as climate and seasonality) are unable to determine vegetation type alone, secondary factors gain more relative importance because they mitigate or intensify water availability and fire-propensity (see

Scheffer et al. 2005). In other words, soil fertility alone does not define rainforest boundaries because what matters is not only the fertility itself, but how it alters the critical or effective precipitation level needed by the vegetation (see Scheffer et al.

2005) (Fig. 3).

Refugia

A lot of research has been carried out in the Wet Tropics on rainforest refugia and the effects of historical rainforest contractions and expansions on regional biodiversity (e.g.

Williams and Pearson 1997; Schneider et al. 1998; Hilbert et al. 2007). Consistent evidence indicates some areas of rainforest maintained some degree of integrity throughout the more arid conditions of the last glacial period (see Hopkins et al. 1993;

Schneider and Moritz 1999; Hilbert et al. 2007). Considering the Wet Tropics as a

37 Chapter Two- Alternative stable states in Australia’s Wet Tropics

system supporting ASS could lead to a reinterpretation of the role of refugia through time and a better understanding of regional historical and biogeographical processes.

For example, different types of vegetation (i.e. isolated endemic populations, dry rainforest communities, or pyrophytic pockets) are currently perceived as remnants of once widespread vegetation types (Ash 1988; Prider and Christophel 2000; Hansman

2001). In a resilience context, rather than seeing refugia as pockets of vegetation that were able to survive, it becomes more interesting to interpret refugia as areas where conditions maintained some degree of stability through time (sensu Graham et al. 2006).

Recent modelling by Hilbert et al. (2007) would also seem to support (and quantify) this view.

Conservation and management

Aside from being an interesting theoretical model, ASS has practical implications in Far

North Queensland. Considering how exotic species alter feedback loops within alternative regimes of native vegetation may give us new insight into how these exotic species alter the resilience of the native vegetation and impact the systems they invade.

One reason exotic species successfully invade both vegetation types may be because they are not limited by, and are able to disturb local feedback loops. For example,

Unwin et al. (1985) describe how Lantana camara (Verbenaceae) burns hotter than native rainforest edge vegetation. This may undermine the rainforest’s resilience by adding pyrophytic feedbacks to the understory vegetation. Similarly, Fensham (1995) discusses exotic species altering native vegetation fuel-loads in dry rainforests in

Australia.

38 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Many authors have stressed the importance of understanding system drivers and what constitutes an efficient fire management regime for northern Australia (e.g. Harrington and Sanderson 1994; Hopkins et al. 1996; Harrington et al. 2005; Banfai and Bowman

2007). Treating a bistable system as a linear one can bring about unexpected outcomes in restoration and conservation efforts, and it can bring about strong system resilience to restoration (Suding et al. 2004). For example, there is ample evidence showing that simply removing grazing pressure does not always restore grasslands to their original condition (Westoby et al. 1989). When feedback loops are altered, restoring abiotic parameters may not be sufficient to restore a system to its original condition (Suding et al. 2004). Hence, understanding the role of feedback loops as system drivers in the Wet

Tropics may have important repercussions for restoration efforts in the region and throughout northern Australia.

Future directions

Existing information and observed phenomena build an interesting case for alternative stable ecosystem states existing in the Wet Tropics. However, empirical evidence is needed to rule out other causes of non-linear behaviour within the system (see Scheffer and Carpenter 2003). Because of the time-scales involved, as well as the interacting processes that occur at different scales it is difficult to prove unequivocally whether

ASS exist in the system. Reassessing and reinterpreting existing literature, particularly from restoration, may provide empirical data about the functioning of feedback loops.

Likewise, reinterpreting the extensive palaeoecological literature for the Wet Tropics may provide further indication of evidence for ASS in the region, and new directions for palaeoecological research.

39 Chapter Two- Alternative stable states in Australia’s Wet Tropics

The terms pyrophytic and pyrophobic imply that the system is driven by the positive and negative feedbacks created by fire and that while pyrophytic vegetation plays an active role, rainforest is passively limited by fire. However, if two positive feedback loops are recognized, we suggest that the term ‘ombrophytic’ (from the Greek ombros, meaning ‘rain shower’, and phuton, meaning ‘plant’) is more fitting than ‘pyrophobic’.

Rather than emphasizing the negative effects of fire, this term acknowledges the moisture-related feedback loops created by closed canopy vegetation as well as the increasing evidence that not all Australian rainforest vegetation is equally pyrophobic

(see Bowman 2000a; Williams 2000; Marrinan et al. 2005). The prefix ‘ombro’ directly links the term to water and thus captures the dichotomy of Australian “rain forests vs. flame forests” (Bowman 2000b) in ombrophytic and pyrophytic vegetation.

Furthermore, while ‘ombrophytic’ does not directly reference the closed-canopy nature of rainforests in its etymology, it does through common usage. A similar existing term,

‘ombrophilous’, is used to describe specific types of dense, closed-canopy vegetation in both South America and New Caledonia (Webb and Tracey 1981, Scudeller et al. 2001).

Despite some important differences in floristics, structure and seasonality, we’ve drawn upon information and examples from analogous vegetation types and processes (such as positive feedbacks in the vegetation) that exist throughout northern Australia (see Adam

1992; Bowman 2000a; Williams 2000; Fensham et al. 2003; Russell-Smith et al. 2004a, b; Banfai and Bowman 2006, 2007). Research into the functional commonalities between different rainforest and sclerophyll vegetation types and their boundaries throughout the region could all feed back into a resilience/ASS context if these vegetation types were considered as parts of equivalent systems. Furthermore, it then becomes interesting to explore whether the different varieties of pyrophytic and

40 Chapter Two- Alternative stable states in Australia’s Wet Tropics

rainforest vegetation types exist along a continuum of ‘minimal habitat conditions’ based on species’ requirements. In other words, whether there are linear (monostable) systems and dynamics nested within the larger bi-stable system. Further research into the role of seasonality in an ASS context could also be of particular interest, especially in ascertaining the role it may play in limiting the distribution of WSF.

Conclusions

Integrating existing knowledge of the Australian Wet Tropics into an ASS theoretical context allows for incorporation of local data into a broader theoretical framework that resolves apparent contradictions, acknowledges the complexity, adaptability and variability of the system and has strong implications for local conservation and management (see Beisner et al. 2003; Folke et al. 2004; Suding et al. 2004). It also opens the door to new research; from looking for historical/palaeoecological evidence, to testing alternative hypothesis and looking into the resilience of the Wet Tropics as a socio-ecological system.

On a larger scale, this reinterpretation of the Wet Tropics as a system with ASS complements existing studies on alternative regimes in more arid vegetation systems in

Australia (see Anderies et al. 2002; Anderies 2005; Walker and Salt 2006). It also fits into a growing body of research on resilience and ecosystems as complex dynamic biological systems where feedbacks and other linked processes occur at many scales

(see Folke et al. 2004; Gillson 2004; Peters et al. 1998, 2004).

41 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Acknowledgements

This project was begun at Victoria University of Wellington, New Zealand, and continued at The University of New South Wales, Australia. We thank both universities for scholarships awarded to LW. We also thank J. Warman and D. Hilbert for insightful discussion on the manuscript, as well as J. van de Koppel, F. J. Weissing and three anonymous reviewers for their comments and suggestions on earlier drafts.

References

Adam P (1992) Australian rainforests. Oxford University Press, Oxford

Anderies JM (2005) Minimal models and agroecological policy at the regional scale: an

application to salinity problems in southeastern Australia. Reg Environ Change

5:1–17.

Anderies JM, Janssen MA, Walker BH (2002) Grazing management, resilience, and the

dynamics of a fire driven rangeland system. Ecosystems (N Y, Print) 5:23–44.

Angeli D, Ferrell JE Jr, Sontag ED (2004) Detection of multistability, bifurcations, and

hysteresis in a large class of biological positive-feedback systems. Proc Natl Acad

Sci USA 101:1822–1827.

Ash J (1988) The location and stability of rainforest boundaries in North-Eastern

Queensland, Australia. J Biogeogr 15:619–630.

Banfai DS, Bowman DMJS (2006) Forty years of lowland monsoon rainforest

expansion in Kakadu National Park, Northern Australia. Biol Conserv 131:553-

565.

Banfai DS, Bowman DMJS (2007) Drivers of rain-forest boundary dynamics in Kakadu

National Park, northern Australia: a field assessment. J Trop Ecol 23:73–86.

42 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Beisner BE, Haydon DT, Cuddington K (2003) Alternative stable states in ecology.

Front Ecol Environ 1:376–382

Bond WJ, Midgley JJ (1995) Kill thy neighbour: an individualistic argument for the

evolution of flammability. Oikos 73:79–85.

Borgogno F, D’Odorico P, Laio F, Ridolfi L (2007) Effect of rainfall interannual

variability on the stability and resilience of dryland plant ecosystems. Water

Resour Res 43:1–9.

Bowman DMJS (2000a) Australian rainforests: islands of green in a land of fire.

Cambridge University Press, Cambridge, UK

Bowman DMJS (2000b) Rainforests and Flame Forests: the Great Australian Forest

Dichotomy. Aust Geo Stud 38: 327-331

Bowman DMJS (2001) On the elusive definition of ‘Australian rainforest’: response to

Lynch and Neldner (2000). Aust J Bot 49:785–787.

Bowman DMJS, Panton WJ (1993) Factors that control monsoon- rainforest seedling

establishment and growth in north Australian eucalyptus savanna. J Ecol 81:297–

304.

Carpenter SR, Brock WA (2006) Rising variance: a leading indicator of ecological

transition. Ecol Lett 9:311–318.

Carpenter SR, Ludwig D, Brock WA (1999) Management of eutrophication for lakes

subject to potentially irreversible change. Ecol Appl 9:751–771.

Dublin HT, Sinclair ARE, McGlade J (1990) Elephants and fire as causes of multiple

stable states in the Serengeti-Mara woodlands. J Anim Ecol 59:1147–1164.

Fensham RJ (1995) Floristics and environmental relations of inland dry rainforest in

North Queensland, Australia. J Biogeogr 22:1047–1063.

43 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Fensham RJ, Fairfax RJ, Butler DW, Bowman DMJS (2003) Effects of fire and drought

in a tropical eucalypt savanna colonized by rain forest. J Biogeogr 30:1405–1414.

Folke C, Carpenter S, Walker B, Scheffer M, Elmqvist T, Gunderson L, Holling CS

(2004) Regime shifts, resilience, and biodiversity in ecosystem management.

Annu Rev Ecol Evol Syst 35:557–581.

Gillson L (2004) Evidence of hierarchical patch dynamics in an East African savanna?

Landscape Ecol 19:883–894.

Goetze D, Hörsch B, Porembski S (2006) Dynamics of forest—savanna mosaics in

north-eastern Ivory Coast from 1954 to 2002. J Biogeogr 33:653–664.

Graham CH, Moritz C, Williams SE (2006) Habitat history improves prediction of

biodiversity in rainforest fauna. Proc Natl Acad Sci USA 103:632–636.

Guevara S, Meave J, Moreno-Casasola P, Laborde J (1992) Floristic composition and

structure of vegetation under isolated trees in neotropical pastures. J Veg Sci

3:655–664.

Hansman DJ (2001) Floral biology of dry rainforest in north Queensland and a

comparison with adjacent savanna woodland. Aust J Bot 49:137–153.

Harrington GN, Sanderson KD (1994) Recent contraction of wet sclerophyll forest in

the wet tropics of Queensland due to invasion by rainforest. Pac Conserv Biol

1:319–327

Harrington GN, Bradford MG, Sanderson KD (2005) The wet sclerophyll and adjacent

forests of North Queensland: a directory to vegetation and physical survey data.

Cooperative Research Centre for Tropical Rainforest Ecology and Management,

Rainforest CRC, Cairns

Heffernan JB (2008) Wetlands as an alternative stable state in desert streams. Ecology

89:1261–1271.

44 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Hilbert DW, Ostendorf B, Hopkins MS (2001) Sensitivity of tropical forests to climate

change in the humid tropics of north Queensland. Austral Ecol 26:590–603.

Hilbert DW, Graham A, Hopkins MS (2007) Glacial and interglacial refugia within a

long-term rainforest refugium: the Wet Tropics Bioregion of NE Queensland,

Australia. Palaeogeogr Palaeoclimatol Palaeoecol 251:104–118.

Holmgren M, Stapp P, Dickman CR, Gracia C, Graham S, Gutierrez JR, Hice C, Jaksic

F, Kelt DA, Letnic M, Lima M, Lopez BC, Meserve PL, Milstead WB, Polis GA,

Previtali MA, Richter M, Sabat S, Squeo FA (2006) Extreme climatic events

shape arid and semiarid ecosystems. Front Ecol Environ 4:87–95.

Hopkins B (1965) Forest and savanna: an introduction to tropical plant ecology with

special reference to West Africa. Heinemann Educational Books, Nigeria

Hopkins MS, Ash J, Graham AW, Head J, Hewett RK (1993) Charcoal evidence of the

spatial extent of the eucalyptus woodland expansions and rainforest contractions

in North Queensland during the Late Pleistocene. J Biogeogr 20:357–372.

Hopkins MS, Head J, Ash JE, Hewett RK, Graham AW (1996) Evidence of a Holocene

and continuing recent expansion of lowland rain forest in humid, tropical North

Queensland. J Biogeogr 23:737–745.

Hughes TP, Rodrigues MJ, Bellwood DR, Ceccarelli D, Hoegh-Guldberg O, McCook

L, Moltschaniwskyj N, Pratchett MS, Steneck RS, Willis B (2007) Phase shifts,

herbivory, and the resilience of coral reefs to climate change. Curr Biol 17:360-

365.

Jackson WD 1968. Fire, air, water and earth- an elemental ecology of Tasmania. - Proc

Ecol Soc Aus 3: 9-16.

Kershaw AP (1994) Pleistocene vegetation of the humid tropics of northeastern

Queensland, Australia. Palaeogeogr Palaeoclimatol Palaeoecol 109:399–412.

45 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Konar B, Estes JA (2003) The stability of boundary regions between kelp beds and

deforested areas. Ecology 84:174–185.

Laurance WF, Williamson GB (2001) Positive feedbacks among forest fragmentation,

drought, and climate change in the Amazon. Conserv Biol 15:1529–1535.

Levin SA (2000) Multiple scales and the maintenance of biodiversity. Ecosystems

3:498–506.

Lynch AJJ, Neldner VJ (2000) Problems of placing boundaries on ecological continua –

options for a workable national rainforest definition in Australia. Aust J Bot

48:511–530.

Marrinan MJ, Edwards W, Landsberg J (2005) Resprouting of saplings following a

tropical rainforest fire in north-east Queensland, Australia. Austral Ecol 30:817–

826.

McManus JW, Polsenberg JF (2004) Coral-algal phase shifts on coral reefs: ecological

and environmental aspects. Prog Oceanogr 60:263–279.

Ngomanda A, Jolly D, Bentaleb I, Chepstow-Lusty A, Makaya Mv, Maley J, Fontugne

M, Oslisly R, Rabenkogo N (2007) Lowland rainforest response to hydrological

changes during the last 1500 years in Gabon, Western Equatorial Africa. Quat Res

67:411–425.

Paine RT, Tegner MJ, Johnson EA (1998) Compounded perturbations yield ecological

surprises. Ecosystems 1:535–545.

Peters DPC, Pielke RA Sr, Bestelmeyer BT, Allen CD, Munson- McGee S, Havstad

KM (2004) Cross-scale interactions, nonlinearities, and forecasting catastrophic

events. Proc Natl Acad Sci USA 101:15130–15135.

Peterson G, Allen CR, Holling CS (1998) Ecological resilience, biodiversity, and scale.

Ecosystems 1:6–18.

46 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Petraitis PS, Latham RE (1999) The importance of scale in testing the origins of

alternative community states. Ecology 80:429–442

Prider JN, Christophel DC (2000) Distributional ecology of Gymnostoma australianum

(Casuarinaceae), a putative palaeoendemic of Australian wet tropic forests. Aust J

Bot 48:427–434.

Puyravaud J-P, Pascal J-P, Dufour C (1994) Ecotone structure as an indicator of

changing forest-savanna boundaries (Linganamakki Region, Southern India). J

Biogeogr 21:581–593.

Rietkerk M, van de Koppel J (1997) Alternate stable states and threshold effects in

semi-arid grazing systems. Oikos 79:69–76.

Roques KG, O’Connor TG, Watkinson AR (2001) Dynamics of shrub encroachment in

an African savanna: relative influences of fire, herbivory, rainfall and density

dependence. J Appl Ecol 38:268–280.

Russell-Smith J, Stanton PJ, Edwards AC, Whitehead PJ (2004a) Rain forest invasion

of eucalypt-dominated woodland savanna, Iron Range, north-eastern Australia: II.

Rates of landscape change. J Biogeogr 31:1305–1316.

Russell-Smith J, Stanton PJ, Whitehead PJ, Edwards A (2004b) Rain forest invasion of

eucalypt-dominated woodland savanna, Iron Range, north-eastern Australia: I.

Successional processes. J Biogeogr 31:1293–1303.

Scheffer M, Carpenter SR (2003) Catastrophic regime shifts in ecosystems: linking

theory to observation. Trends Ecol Evol 18:648–656.

Scheffer M, Hosper SH, Meijer M-L, Moss B, Jeppesen E (1993) Alternative equilibria

in shallow lakes. Trends Ecol Evol 8:275–279.

Scheffer M, Carpenter S, Foley JA, Folke C, Walker B (2001) Catastrophic shifts in

ecosystems. Nature 413:591–596.

47 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Scheffer M, Holmgren M, Brovkin V, Claussen M (2005) Synergy between small- and

large-scale feedbacks of vegetation on the water cycle. Glob Change Biol

11:1003–1012.

Schneider C, Moritz C (1999) Rainforest refugia and evolution in Australia’s Wet

Tropics. Proc R Soc B Biol Sci 266:191

Schneider CJ, Cunningham M, Moritz C (1998) Comparative phylogeography and the

history of endemic vertebrates in the Wet Tropics rainforests of Australia. Mol

Ecol 7:487–498.

Schröder A, Persson L, De Roos AM (2005) Direct experimental evidence for

alternative stable states: a review. Oikos 110:3–19.

Scudeller V, Martins F, Shepherd G. 2001. Distribution and abundance of arboreal

species in the atlantic ombrophilous dense forest in Southeastern Brazil. Plant

Ecol 152: 185-199.

Smith TB, Wayne RK, Girman DJ, Bruford MW (1997) A role for ecotones in

generating rainforest biodiversity. Science 276:1855–1857.

Sternberg LDSL (2001) Savanna-Forest hysteresis in the tropics. Glob Ecol Biogeogr

10:369–378.

Suding KN, Gross KL, Houseman GR (2004) Alternative states and positive feedbacks

in restoration ecology. Trends Ecol Evol 19:46–53.

Swaine MD (1992) Characteristics of dry forest in West Africa and the influence of fire.

J Veg Sci 3:365–374.

Turton SM, Sexton GJ (1996) Environmental gradients across four rainforest-open

forest boundaries in northeastern Queensland. Austral Ecol 21:245–254.

Unwin GL (1989) Structure and composition of the abrupt rainforest boundary in the

Herberton Highland, North Queensland. Aust J Bot 37:413–428.

48 Chapter Two- Alternative stable states in Australia’s Wet Tropics

Unwin GL, Stocker GC, Sanderson KD (1985) Fire and the forest ecotone in the

Herberton highland, north Queensland. Proc Ecol Soc Aust 13:215–224

Unwin GL, Applegate GB, Stocker GC, Nicholson DI (1988) Initial effects of Tropical

Cyclone ‘Winifred’ on forests in north Queensland. Proc Ecol Soc Aust 15:283–

296 van Nes EH, Scheffer M(2004) Large species shifts triggered by small forces. Am Nat

164:255–266.

Walker BH, Salt D (2006) Resilience thinking: sustaining ecosystems and people in a

changing world. Island Press,Washington, DC

Webb LJ (1958) Cyclones as an ecological factor in tropical lowland rain-forest, North

Queensland. Aust J Bot 6:220–228.

Webb LJ, Tracey JG. (1994) The rainforests of northern Australia. In: Groves RH (ed.)..

Australian Vegetation. Cambridge University Press.

Westoby M, Walker B, Noy-Meir I (1989) Opportunistic management for rangelands

not at equilibrium. J Range Manage 42:266–274.

Williams PR (2000) Fire-stimulated rainforest seedling recruitment and vegetative

regeneration in a densely grassed wet sclerophyll forest of north-eastern Australia.

Aust J Bot 48:651–658.

Williams SE (2006) Vertebrates of the Wet Tropics rainforests of Australia: species

distributions and biodiversity. Cooperative Research Centre for Tropical

Rainforest Ecology and Management, Rainforest CRC, Cairns, Australia

Williams SE, Pearson RG (1997) Historical rainforest contractions, localized

extinctions and patterns of vertebrate endemism in the rainforests of Australia’s

Wet Tropics. Proc Biol Sci 264:709–716.

Photos: L.Warman

Chapter Three

Not so simple after all: Searching for the ecological

advantages of compound leaves.

Laura Warman, Angela T. Moles and Will Edwards

In press: Oikos DOI: 10.1111/j.1600-0706.2010.19344.x

The study was conceived by LW and AM. LW carried out the fieldwork, all of the image and data analysis, and the majority of the writing. WE helped with the fieldwork and provided invaluable logistical support. Supervision and manuscript corrections were carried out by AM and WE.

This chapter has been modified from the original paper specifically for this thesis.

This article has been reproduced with kind permission from Oikos and Wiley-Blackwell. Oikos grants authors the right to reproduce their article in a new publication of which they are the author, editor or co-editor, including a thesis or dissertation.

51 Chapter Three- Leaf type and herbivory

Abstract

Leaves come in many sizes and shapes, and the relationships between leaf traits

and the environments they occur in are better understood every day. However

we still know very little about the ecological consequences of plants having

either compound or simple leaves. We attempted to address this knowledge gap

by comparing chemical and physical characteristics (leaf area, length:width

ratio, water content, leaf mass per area, ‘toughness’ and C:N ratio), as well as

rates of herbivory between compound and simple leaves across 34 species in

adjacent rainforest, open woodland and wet sclerophyll (tall open forest)

vegetation in northeastern Australia. We found C:N ratio to be lower in

compound leaves, but this was the only leaf trait that differed significantly

between leaf types and did not stand up under phylogenetic analysis. Overall, we

found no differences in herbivory between simple and compound leaves. While

it remains unclear what the advantages of having one leaf type over another

might be, the differences do not seem to lie in construction, or in vulnerability to

herbivores, at least in the Australian Wet Tropics. Furthermore, we found no

evidence of herbivores playing a role in maintaining rainforest and sclerophyll

vegetation as alternative stable states in the region.

52 Chapter Three- Leaf type and herbivory

Introduction

There is ever-increasing knowledge of leaf traits and their role in the interactions between plants and the environments in which they occur. Numerous studies have examined the ways leaf size, mass, shape and thickness relate to light regimes, leaf longevity, plant life form and soil nutrients (Givnish and Vermeij 1976, Givnish 1979,

1987, Niinemets and Kull 2003, Wright et al. 2004, Hoffman et al. 2005, Poorter and

Bongers 2006). There is also growing understanding of the distribution and significance of different types of leaf margins in relation to climate; as well as the role of structures like drip-tips, trends in leaf ‘hairiness’, and chemical composition in relation to altitude and latitude (Baker-Brosh and Peet 1997, Ivey and DaSilva 2001, Wegner et al. 2003,

Reich and Oleyksyn 2004, Lucking and Bernecker-Lucking 2005, Løe et al. 2007). In contrast, we still know surprisingly little about the ecological significance of one of the most fundamental aspects of leaf form: the dichotomy between simple and compound leaves.

Compound leaves have arisen independently many times during plant evolution; that is, not all compound leaves are homologous (Bharathan and Sinha 2001, Friedman et al.

2004). Differences in leaf type occur throughout families and even within genera

(Taylor and Hill 1996), and in the case of some aquatic species they can even occur on the same individual (Sinha 1999). Many studies have been aimed at teasing apart the evolutionary histories of simple and compound leaves (Friedman et al. 2004) as well as their genetic and developmental differences (Bharathan and Sinha 2001). However, a surprising knowledge gap still exists about basic ecological costs and advantages afforded to plants by these different leaf types. The majority of ecological studies do not explicitly seek to compare compound and simple leaf types, but rather consider the leaf

53 Chapter Three- Leaf type and herbivory

forms only in passing (usually just recording leaf type as one of many plant or leaf traits observed).

Compound leaves seem to be more common in warmer and arid or semi-arid environments, high-light environments, and in light demanding species (Givnish 1978a, b, 1987, Stowe and Brown 1981). Their shape, arrangement and construction are thought to offer advantages in capturing light while reducing water loss and maintaining lower leaf lamina temperatures (Givnish 1984, Popma et al. 1992, Niinemets 1998).

Accordingly, it has been proposed that compound leaves should be more common in pioneer species and early succesional communities; that is, environments with high light availability and where species need to gain height quickly (Givnish 1978b, 1979, 1987).

Givnish (1978a, b, 1979, 1984) put forward the idea that compound leaves in these situations function as disposable branches. The rachis represents a smaller investment for a plant than a woody lateral branch and it can be disposed of during adverse conditions such as seasonal drought (Givnish 1978a, b). Thus, having compound leaves should allow species to invest relatively more energy into gaining height (and thus be more competitive) than those species with heavy investments in actual branches

(Givnish 1978a, 1979). However, field evidence has proven equivocal. Recent studies have found that compound leaves are not necessarily more frequent among light demanding species in rainforest light gaps (Popma et al. 1992), that shade-tolerance does not vary significantly between leaf types (Niinemets 1998) and that energetic tradeoffs between leaf construction and light capture (particularly in regard to size) can be found in both leaf types (Niinemets 1998, Niinemets and Kull 1999, Niinemets et al.

2006). Furthermore, Malhado et al. (2010) found that while trees with compound leaves in Amazonia grow faster than those with simple leaves, this was not related to the

54 Chapter Three- Leaf type and herbivory

species’ status as a pioneer. If the benefits of compound leaves are not purely a matter of support, leaf organization, or succesional status then the ecological advantages of having either leaf type must lie elsewhere.

Selective pressures exerted by herbivores can play an important role in shaping leaf coloration, morphology, plant architecture and community composition (Ritchie et al.

1998, Burns and Dawson 2006, Fine et al. 2004, Rudgers and Whitney 2006, Campitelli et al. 2008). Previous studies have suggested that compound leaves may receive less herbivory, but the idea remains untested. Providing a test of this idea is the first aim of our study. Both Brown et al. (1991) and Niinemets et al. (2006) have suggested that having many small leaflets, rather than a single large lamina, could make grazing less efficient for herbivores; it could also affect how ‘apparent’ leaflets are to herbivores

(sensu Feeny 1976). Having compound leaves may also enable plants to isolate and/or limit damage to discrete areas. In other words, it may be advantageous for a plant to lose a whole leaflet to herbivores or pathogens rather than to lose part of a leaf and risk having damage spread throughout a simple leaf's entire lamina (Brown et al. 1991).

Interestingly, Gall (1987) found that some plants with compound leaves allocate chemical defences differentially amongst leaflets which occur at varying positions along the rachis. These defences directly affect herbivore feeding choice regarding the defended leaflets and may provide an indirect defence for neighbouring leaflets (Gall

1987). If this is the case, a plant with compound leaves might be able to defend itself from herbivores with smaller energetic investments than can a plant with simple leaves.

Defence from herbivory might also be derived secondarily from traits inherent to compound leaves. Testing this idea was the second aim of our study. Many of the foliar

55 Chapter Three- Leaf type and herbivory

traits that have been linked to growing in hot, dry and/or high light environments are also known to influence herbivory levels. Specifically, smaller, tougher leaves, which are common in high light or low water environments (Givnish 1987, Bragg and

Westoby 2002), have been shown to receive less herbivory (Coley 1983, Moles and

Westoby 2000). So it could be hypothesised that compound leaves might receive less damage because some of the same traits that make them more efficient in warmer or drier environments may make leaflets less apparent or palatable to herbivores.

For both compound and simple leaves, the costs and effects associated with mechanical support and construction are strongly tied to leaf size (Niinemets 1998, Niinemets and

Kull 1999, Niinemets et al. 2006). Thus, one could expect strong differences in leaf traits when comparing small leaflets to large simple leaves. Niinemets et al. (2006) suggest that leaflet size may also influence herbivory because larger leaves invest a higher proportion of biomass into internal support structures. However, if a simple leaf's lamina were the same size as a compound leaflet, it is not clear whether there would be a difference in toughness that might affect herbivory.

In this paper we search for evidence of ecological advantages of compound leaves in regards to interaction with herbivores. We compare leaf traits that have previously been linked to both herbivory and possible construction differences between leaf types. Then we test whether compound and simple leaves experience different levels of herbivory.

We expected compound leaves to exhibit less herbivore damage than do simple leaves and for this difference to be associated with differences in traits including smaller leaflet area, greater toughness and higher C:N ratios.

56 Chapter Three- Leaf type and herbivory

We sampled herbivory levels in different environments to test for differences in herbivory between rainforest, sclerophyll and wet sclerophyll vegetation. These vegetation types can exist adjacent to one another in north-eastern Australia, and present strong contrasts in soil nutrients, light incidence and relative humidity all of which have an influence on both leaf traits and herbivory levels (Huntly 1991). More importantly, strong differences in herbivory between these adjacent environments may indicate that herbivores are playing a role in controlling the location and extent of the vegetation.

Studies from a variety of environments around the world have shown that herbivores can play an important role in both creating and maintaining alternative stable states

(ASS) at scales ranging from communities to landscapes. For example Fine et al. (2004) found that distinct vegetation communities growing on sand and clay soils were not limited by the substrate per se, but rather by the effects of herbivores. Likewise, elephant browsing in addition to seasonal fires contribute to the maintenance of savannah-forest boundaries in Africa (Dublin et al. 1990).

The role of terrestrial herbivores in controlling the maintenance of alternative ecosystem states in Australia becomes particularly interesting, given that Australia has relatively few native species of large mammalian herbivores (Andersen and Lonsdale 1990). For example, Andersen and Lonsdale (1990) point out the striking contrast between the density and diversity of mammalian herbivores on African savannas (which includes a variety of gazelles, zebras, wildebeest and elephants) and in Australian savannas (where the largest extant herbivores are introduced species including horses, cattle and water buffalo). Given this absence of mammalian herbivores, the role of insect herbivores becomes more important and it becomes interesting to test whether they can maintain alternative stable states through herbivory. Existing studies comparing herbivory

57 Chapter Three- Leaf type and herbivory

between rainforest and sclerophyll vegetation in Australia have not been carried out in the tropics and offered ambiguous results (see Lowman 1995).

The main questions we addressed in this study were: 1) are traits that have been linked to herbivory consistently different in compound and simple leaves? 2) Do levels of herbivory differ between compound and simple leaves, and does this relate to specific traits? 3) Are herbivory levels different in rainforest, wet sclerophyll and wet sclerophyll vegetation, and is there any evidence that herbivores themselves are playing a role in maintaining these vegetation types as alternative stable states? 4) Are herbivory levels for the different leaf types maintained across contrasting vegetation types?

Material and methods

Site descriptions

The study took place in the Wet Tropics of northeastern Queensland, Australia. The region contains both rainforest (traditionally characterized by closed canopies and shady understories with high humidity levels, Adam 1992) and sclerophyll vegetation

(woodlands and forests which have sparse canopies and more open understories with high light penetration, Hopkins et al. 1993). Abrupt boundaries between these vegetation types occur throughout the area. A third, intermediate, vegetation type known as wet sclerophyll or tall open forest is also present. Wet sclerophyll is composed of a variety of sclerophyll and rainforest species – often in the form of a rainforest understory under a tall sclerophyll canopy (Hopkins et al. 1993, Warman and

Moles 2009). Environmental conditions in a wet sclerophyll forest understory can range from those expected in fire-prone woodlands, to the shady conditions found in rainforests.

58 Chapter Three- Leaf type and herbivory

Field sites in all three vegetation types were chosen at both Brooklyn Station on Mt.

Lewis and near the Clohesy River Road in Dinden National Park and State Forest (site descriptions in Table 1). At both locations the three vegetation types were within two kilometres of each other.

Species selection and leaf analysis

At each site we chose the most abundant species with compound and simple leaves.

Only dicot species were considered because monocots have been shown to have tougher leaves than dicots in lowland rainforests (Dominy et al. 2008). At each site we selected four species with compound leaves and four species with simple leaves except at the sclerophyll sites where there were not enough individuals of each species with compound leaves (at the Brooklyn site we included two species and at the Clohesy site we only considered simple-leaved species). In total we studied 15 compound-leaved species and 19 simple-leaved species, for a total of 34 species in 19 families (Appendix

1). For simplicity, we refer to both leaves and leaflets as leaves throughout the rest of the paper (following Givnish 1978b).

We collected 15 to 20 leaves from at least three individuals of each species (600 leaves in total) to measure a suite of traits. We chose physical and chemical traits which have been associated with varying rates of herbivory but that have also been implied in the putative construction differences between simple and compound leaves. We compared leaf area and width:length ratio as a proxy for apparency and because these traits have been linked to both light capture and water loss (Malhado et al. 2009). Leaf mass per area (LMA), water content, toughness and C:N ratio were chosen because they may reflect differences in leaf construction and biomass allocation, as well as being known

59 Chapter Three- Leaf type and herbivory

to influence herbivory (Coley 1983, Nabeshima et al. 2001, Pérez- Harguindeguy et al.

2003).

Table 1: Field site descriptions

Site Coordinates Altitude Description

Brooklyn 16°36.085’S, 883 m Sclerophyll woodland dominated sclerophyll 145°15.624’E by Allocasuarina torulosa, Banksia woodland aquilonia and Syncarpia glomulifera.Open, grassy understory.

Brooklyn wet 16°35.989’S, 926 m Tall open forest with Eucalyptus sclerophyll forest 145°15.832’E grandis, E. resinifera and Syncarpia glomulifera in the canopy. Relatively open understory, some rainforest plants.

Brooklyn rainforest 16°35.756’S, 842 m Lower montane rainforest with a 145°16.529’E very closed canopy and dense understory.

Clohesy sclerophyll 16°56.299’S, 474 m Woodland dominated by woodland 145°36.310’E Allocasuarina torulosa and Eucalyptus tereticornis. Very open understory.

Clohesy wet 16°55.950’S, 503 m Tall open forest dominated by sclerophyll forest 145°37.074’E Eucalyptus grandis and E. pellita. Relatively dense rainforest understory.

Clohesy rainforest 16°56.294’S, 379 m Secondary rainforest; 145°37.066’E Argyrodendron a prominent component of a thick canopy. Relatively open understory.

60 Chapter Three- Leaf type and herbivory

All fresh leaves were individually weighed, scanned (using an HP psc2510 flatbed scanner) then oven dried at 60°C for two days, then weighed again. Trait values were measured for each leaf and averaged for species. Only mature, fully expanded leaves were considered (Moles and Westoby 2000), and these were collected along the length of the branches/rachises to account for natural variation in leaf size. To minimize potential water loss and loss of mass due to respiration, leaves were kept in sealed plastic bags with moistened tissue, placed in refrigeration as soon as possible and measured within two days of collection.

Leaf images were analysed using GNU Image Manipulation Program (GIMP ver. 2.6.2) and ImageJ (ver. 1.37, Rasband 2008) to calculate leaf areas and length-width ratios.

We then calculated LMA (dry weight/leaf area [g cmí2]) for each leaf and determined an average LMA for each species. We did not take the petioles/petiolules of the leaves into account. The dry leaf samples were ground using a mill and carbon and nitrogen content of leaves were assayed at Southern Cross University's Environmental Analysis

Laboratory.

We determined relative leaf ‘toughness’ in the field using a Chatillon Type 516 push- pull gauge as a penetrometer. Leaves were placed between two perforated acrylic panels for support while the gauge was pushed through the leaf lamina, avoiding obvious venation. ‘Toughness’ values were converted to percentage toughness for each species, whereby the species with the highest mean toughness was assigned a value of 100%, and values for all other species were expressed relative to this. We used this measure because we were interested in comparing the relative toughness of leaves in our sample

61 Chapter Three- Leaf type and herbivory

(and how it might affect a herbivore's choice), rather than in an absolute measurement of toughness or punch strength.

Sampling of herbivory

At each site we chose three to five individuals of each species, making sure to avoid plants from which leaves had already been collected (to avoid the possibility of herbivory results being affected by our artificially inducing defences during leaf collection). On each plant we tagged three leaves on each of three different branches (or rachises) using cable ties around the stem/rachis. In total we marked nine leaves per individual and 33 to 45 leaves per species at each site. Only mature, fully expanded leaves with the lowest possible levels of existing damage were chosen (Moles and

Westoby 2000). Pre-existing damage on leaves was marked with xylene-free permanent

marker, and if damage was extensive (more than ׽3% of the leaf area), a digital photograph was taken of the leaf for later comparison. Tagged leaves were left in situ in the field, and then collected after an eight week observation period which ran from

September to November 2008. The leaves were scanned and total leaf areas and percentage of damaged area were calculated for each leaf (again using GIMP and

ImageJ). In total 1779 leaves were scanned

Statistical analysis

Data were analyzed using general linear models (GLM) in SPSS Statistics (ver. 17.0).

All leaf trait values and herbivory data were log10-transformed prior to analysis. In order to avoid zeroes when log10-transforming herbivory values, leaves with no detectable levels of herbivory were assigned the value 0.006; which was equal to half of the smallest detected value. There was evidence of losses of tagged leaves due to leaf

62 Chapter Three- Leaf type and herbivory

senescence in one out of the 34 species observed (consistently higher rates of missing leaves as well as senesced tagged leaves on the ground beneath Alstonia muelleriana), so to avoid confounding our herbivory results, the 33 missing leaves from this species were excluded from the analysis.

Mean values for both leaf traits and herbivory were calculated for each species, and analysis was carried out at the species (rather than individual, ‘branch’ or ‘leaf’) level.

We applied full factorial models and tested for interactions between traits and herbivory for both leaf type and vegetation type. We also used a GLM to compare herbivory in compound and simple leaves using leaf size as a covariate. For analysis of herbivory within and between vegetation types, data were nested (species within vegetation type within site) in the GLMs. Regressions were used to test whether there exists a relationship between herbivory and leaf size, C:N ratio and LMA.

During cross species analyses each species is considered as an independent replicate.

However the species sampled have varying degrees of phylogenetic relatedness and shared evolutionary history and thus they are not truly fully independent. To account for this phylogenetic non-independence, we carried out phylogenetic analyses for both herbivory and leaf traits. These analyses use phylogenetically independent contrasts (in other words, the differences between species) as replicates rather than the species themselves. Phylogenetically independent contrasts start by plotting species onto a known phylogenetic tree. Then, for each terminal species pair and for each ancestral node, trait divergences (and in this case herbivory differences) are calculated (for example, if species A has an area of 200mm2 and its congeneric, species B, has an area of 150mm2, then the contrast for that species pair is 50mm2). These values are then used

63 Chapter Three- Leaf type and herbivory

in a regression which must pass through the origin (because the contrasts may have positive or negative values, arbitrarily, Garland et al 1992). For this study, species were arranged on phylogenetic trees using PHYLOMATIC (with maximally resolved tree

R20091110 as a backbone, Webb et al. 2008) and then phylogenetically independent contrasts (based on branch topology and length) were carried out using the Analysis of

Traits module of PHYLOCOM (‘Software for the Analysis of Phylogenetic Community

Structure and Character Evolution’, Webb et al. 2008).

Results

Leaf traits in compound and simple leaves

Surprisingly, C:N ratio was the only trait that was significantly different between compound and simple leaves (GLM, p = 0.007; Fig. 1a). Twenty three percent of the variation in the dataset was attributed to the term for leaf type. Compound leaves had lower C:N ratios, that is, higher foliar nitrogen concentrations relative to foliar carbon than simple leaves. When we analyzed the elements separately we found that compound leaves had significantly higher foliar nitrogen (GLM, p = 0.004) than simple leaves, while foliar carbon did not vary significantly between leaf types (GLM, p = 0.53). Since leaves with higher nitrogen content have been shown to be preferred by herbivores over leaves which are high in carbon compounds (Basset 1991, Pérez-Harguindeguy et al.

2003), our results suggest that compound leaves could be more palatable for herbivores than are simple leaves.

Simple leaves are slightly, but not significantly tougher than compound leaves (mean toughness of 28.12% and 40.93% respectively; GLM p = 0.066 with only 10% of the variation in the toughness data attributable to leaf type; Fig.1b). While this difference in

64 Chapter Three- Leaf type and herbivory

a) b) c)

100 0.1 scale] scale] scale] 10 100 ] p=0.004 p=0.06 0.05 e

10 p=0.09 l ) [log

50 2

sca 0.02 50 10

og 0.01

[l 30

o 30 ti 0.005

ra 20 20 N : C 0.002

10 10 (g/cm area per Leaf mass 0.001 Compound Simple Relative toughness (%)[log Compound Simple Compound Simple d) e) f)

100 200 40 p=0.9 p=0.6 100 p=0.4

scale] scale] 20 scale] scale] 80 scale] 10

10 50 10 10

) [log 2 20 60 10 4 50 5 2 Leaf area (cm Length-width ratio [log Water content (%) [log Water content (%) 40 2 1 Compound Simple Compound Simple Compound Simple

Figure 1- Comparison of leaf traits between species with simple and

compound leaves. Boxes span from the 25th to the 75th percentile, and the

horizontal line represents the median. Whiskers span from the 10th to the 90th

percentile, and outlying species are shown as dots. P values correspond to

GLM results.

relative toughness could simply reflect the species selection, it is interesting to speculate whether the trend might reflect differences in the construction of the laminas. Leaf toughness is often closely associated with how thick or dense the leaf itself is. However, despite simple leaves having slightly higher LMA (0.008 vs 0.006 g cmí2), we found no significant difference in LMA between leaf types (GLM, p = 0.09; Fig. 1c). There was also no significant difference in foliar water content (GLM, p = 0.6), leaf area

65 Chapter Three- Leaf type and herbivory

(GLM, p = 0.93) or length-width ratios (GLM, p = 0.47) between simple and compound leaves (Fig. 1d–f). As there was no difference in leaf area or in length-width ratio between compound leaflets and simple leaves, apparency to herbivores would not appear to be different between leaf types.

Phylogenetic analyses showed no consistent differences in any of the traits across the independent contrasts between simple and compound leaved taxa (all p > 0.12). That is, the lack of significance in our cross species analyses is not an artefact of the phylogenetic composition of our species. However the significant difference in C:N ratios found in cross species analysis may be an artefact caused by one or a few changes deep within the phylogeny.

Overall herbivory in compound and simple leaves

We found no significant difference in herbivory between simple and compound leaves

(GLM, p = 0.44; Fig. 2). Furthermore, we found no relationship in our samples between amount of herbivory and leaf area (GLM, p = 0.78), C:N ratio (GLM, p = 0.51), water content (GLM, p = 0.43), toughness (GLM, p = 0.74), shape (GLM, p = 0.96), LMA

(GLM, p = 0.82) or leaf type using leaf size as a covariate (GLM, p>0.05). Herbivory levels were relatively low, with 49.6% of all leaves receiving less than 2% damage during the eight week sample period. There was no discernible herbivory damage in

23.7% of the leaves sampled while only 1.7% of the leaves were completely removed.

Phylogenetic analyses showed a marginally significant (p = 0.051) difference in herbivory across contrasts between taxa with compound versus simple leaves. The difference between the results of the cross species and phylogenetic analyses suggests

66 Chapter Three- Leaf type and herbivory

100 50 p=0.44 20 10 5 scale] 10 2 1 0.5

0.2

Herbivory (%) [log 0.1 0.05

0.02 0.01 Compound Simple

Figure 2- Comparison of herbivory damage (percentage of leaf area removed)

over an eight week period between species with simple and compound leaves.

Boxes span from the 25th to the 75th percentile, and the horizontal line

represents the median. Whiskers span from the 10th to the 90th percentile, and

outlying species are shown as dots.

that one or more divergences deep in the phylogeny might be having an undue influence on the cross species results. We investigated this possibility using the contribution index

(Moles et al. 2005), a metric that calculates the relative influence of different divergences in the phylogeny on the present day distribution of traits. The highest contribution (contribution index of 0.44) to this result came from the divergence between Cissus hypoglauca and its closest relatives (Appendix 2). Cissus hypoglauca not only received more damage than its nearest rosid relatives, but also had the highest level of herbivory (16.5% over eight weeks) of all the species analyzed. Given the types of damaged observed on the leaves of C. hypoglauca, it is also possible that this species received more damage by mammalian herbivores, rather than damage consistent with insect herbivores as was the norm for the majority of the species sampled.

67 Chapter Three- Leaf type and herbivory

Herbivory across vegetation types

There was no significant difference in herbivory between compound and simple leaves in either sclerophyll (GLM, p = 0.85) or wet sclerophyll vegetation (GLM, p = 0.31;

Fig. 3). Furthermore, herbivory was remarkably similar across these vegetation types

(3.2% vs 2.9% of leaf area lost over eight weeks). The picture becomes murkier when considering the rainforest data (Fig. 3). We found no difference in herbivory between leaf types at the Brooklyn (Mt Lewis) rainforest site (GLM, p = 0.92; 0.81% of leaf area removed). Compound leaves at the Clohesy rainforest site had a comparable degree of damage to leaves of both types at the Brooklyn rainforest site (2.42%) and leaves from sclerophyll and wet sclerophyll vegetation. However the simple leaves at Clohesy had significantly less damage (0.15%; GLM, p = 0.007) than did their compound counterparts.

To understand the differences in herbivory between leaf types at Clohesy, we tested whether these could be explained by differences in leaf traits between leaf types at this location. Water content was the only trait to vary significantly (GLM, p = 0.035, with

50% of the variation corresponding to leaf type) between leaf types at the Clohesy rainforest site. However, water content was not related to herbivory (GLM, p = 0.85).

Thus, although there was a difference in herbivory between simple and compound leaves at one of the six study sites, this difference did not seem to be attributable to a difference in leaf traits. At this level of analysis (leaf types within one vegetation type at one site), we are only comparing four simple-leaved species with four compound-leaved species. A likely explanation is that unmeasured chemical defenses might drive this result. This seems especially plausible since the simple leaved species in question belong to families known to contain many secondary chemical compounds (namely

68 Chapter Three- Leaf type and herbivory

Rubiaceae, Myrtaceae, Euphorbiaceae and Celastraceae; in comparison to the compound leaved species which belong to the Araliaceae, Malvaceae, Rutaceae and

Connaraceae).

Sclerophyll vegetation Wet sclerophyll vegetation Rainforest vegetation 100 100 100

p=0.85 p=0.31 p=0.007 p=0.92 scale]

10 10 10 10

1 1 1 Herrbivory (%) [log

0.1 0.1 0.1 Compound Simple Compound Simple Compound Simple Compound Simple Clohesy Brooklyn

Figure 3- Mean herbivory levels (percentage of leaf area removed) over an

eight week period for species with compound and simple leaves in all three

vegetation types sampled. Error bars denote standard error. Site data are

pooled for sclerophyll and wet sclerophyll vegetation, and shown separate for

rainforest vegetation

Discussion

Traits and their relation to herbivory

Our trait analyses indicated that simple and compound leaves are very similar in overall form and composition. That is not to say that the leaves sampled in our study were uniform, but rather that considerable variation was present in both simple leaves and compound leaflets. For example, leaf area across both leaf types spanned three orders of 69 Chapter Three- Leaf type and herbivory

magnitude, from 1.3 cm2 to 325.3 cm2. It is also worth noting that the species with the smallest average leaf area had simple leaves (Alyxia ruscifolia; 2.77cm2), while the species with the highest average leaf area had compound leaves (Polyscias australiana;

151.65cm2). It has been noted before that leaflets need not be smaller than simple leaves

(Givnish 1978, Stowe and Brown 1981), but our results emphasize that the stereotype of small leaflets versus large simple leaves should be discarded.

In the absence of structural differences or differences in apparency, one might think that compound leaves should be more attractive to herbivores because they are relatively richer in nitrogen (Basset 1991, Ritchie et al. 1998, Pérez-Harguindeguy et al. 2003,

Kurokawa et al. 2010). However, differential C:N ratios in our sample clearly did not lead to a difference in herbivory levels between leaf types. One possible answer is that not all leaf nitrogen is available to herbivores. Nitrogen can be tied up by secondary compounds such as tannins (DeGabriel et al. 2008). Hence while compound leaves are richer in total nitrogen, the quantities of nitrogen available to herbivores, particularly invertebrates, may not actually differ between leaf types.

Our results do not support the hypothesis that having compound leaves may be beneficial to plants in avoiding herbivory. However, they do lend some support to the idea of compound leaves functioning as disposable structures, and add to the body of evidence on the energetic costs of large versus small leaves.

Previous studies have shown that while compound leaved species have a higher biomass investment in the rachis and petiole-like structures, within-leaf support for large simple leaves can also represent a large carbon and energy investment (Niinemets 1998,

70 Chapter Three- Leaf type and herbivory

Niinemets et al. 2006). Given that both leaf types in our sample were so similarly constructed (aside from C:N ratios) and had similar leaf areas, length-width ratios and

LMA, one would expect similar loads and thus similar construction costs for both the internal and external foliar support structures in both leaf types.This would lend support to the idea that the construction tradeoffs are bigger across a spectrum of leaf sizes than across leaf types (Niinemets 1998, Niinemets and Kull 1999, Niinemets et al. 2003).

Interestingly, in previous studies high foliar nitrogen has been linked to leaves with short lifespans (Poorter and Bongers 2006), a trait that could be expected in pioneer species and those with deciduous leaves. In that regard, our results would be consistent with Givnish's original proposal of compound leaves as disposable structures. LMA is also strongly correlated with leaf life span (Wright et al. 2004, Poorter and Bongers

2006). Although we did not find significant differences in LMA (p = 0.09) or toughness

(GLM, p = 0.07) between leaf types, the directions of our non-significant results are consistent with Givnish's proposal. The higher C:N ratios we found in simple leaves may also be a reflection of higher vascular/support tissue in the simple leaves

(Niinemets 1998, Pérez-Harguindeguy et al. 2003) which may also influence the tendency towards simple leaves being slightly tougher.

Perhaps the difficulty in finding a strong dichotomy in traits or ecological differences between compound and simple leaves indicates that, as with other leaf traits and plant life-strategies, leaf type is best understood as a continuum rather than a dichotomy

(Poorter and Bongers 2006).

71 Chapter Three- Leaf type and herbivory

Herbivory differences between vegetation types

We expected to find different rates of herbivory across vegetation types. However, herbivory levels were surprisingly similar at five of the six field sites. We found no evidence of herbivores playing a role in maintaining rainforest and sclerophyll vegetation as alternative stable states. This was remarkable given the strongly contrasting conditions found in the three vegetation types, especially in regard to light penetration. Previous studies have shown that sun leaves suffer higher rates of damage than do shade leaves on individual plants, and that the leaves of light demanding species experience higher rates of herbivory than do shade-tolerant species (Coley 1987, Dudt and Shure 1994). Additionally, Lowman (1985) found differences in herbivory between rainforests and sclerophyll vegetation in temperate and subtropical Australia. Finding similar rates of herbivory across vegetation boundaries in the Wet Tropics is interesting given that there is strong regional evidence to suggest that different herbivore assemblages are present in the different vegetation types, despite their proximity to one another (Williams and Marsh 1998, van Ingen et al. 2008).

Seasonality

The study was carried out at the end of the dry season (observations were carried out

September through October) on fully mature leaves that were likely to have been present for at least one growing season. During this time of year regional insect numbers increase after a seasonal lull (Frith and Frith 1985, Grimbacher and Stork

2009) and it has been reported as the seasonal peak of activity for herbivorous beetles in the Australian Wet Tropics (Grimbacher and Stork 2009). Given the type and amount of damage on the leaves, as well as the average height of most of the plants sampled, it is likely that most of the damage was caused by insect, rather than mammalian herbivores.

72 Chapter Three- Leaf type and herbivory

The overall observed mean leaf loss was 2.6% over eight weeks (3.6% in compound leaves and 1.7% in simple leaves). Although this value is relatively low it is not dissimilar from other discrete and short term studies which commonly report losses of three to 10% (Lowman and Box 1983, Lowman 1985). Longer term observations usually report higher losses of 4.8% to 32.5% (Lowman and Box 1983, Lowman 1985).

Our study represents a ‘snapshot in time’ at six field sites, rather than a comprehensive long term assessment of herbivory at those locations. As such, our results are likely to underestimate total (or yearly) rates of herbivory (Lowman 1985). However, if herbivory were playing a defining role in the evolution of leaf shape, we would have expected to find significant differences in the levels of herbivory between the leaf types, which we did not.

Conclusions

Our study did not find the expected differences in traits between simple and compound leaves. Nor did we find significant differences in the level of herbivory between leaf types. Thus, the ecological advantages of simple versus compound leaves do not seem to lie in construction, or in vulnerability to herbivores, at least in the Australian Wet

Tropics. In particular, the recurring idea that having compound leaves offers benefits in regards to damage from herbivores, because of small folioles rather than a single large lamina, should be abandoned. Field evidence has still not fully confirmed ecological hypotheses for the advantage of having one leaf type over the other, and many long-held ideas in this field are still awaiting formal testing. It is surprising that given how much we know about leaves, we still don't really understand such a basic and evident

73 Chapter Three- Leaf type and herbivory

characteristic as leaf type. For the moment the ecological advantages of compound leaves remain unknown, but the search for them may yet prove enlightening.

Acknowledgements

UNSW for UIPA scholarship awarded to LW. Thanks to Mick Blackman and John

Kanowski at the Australian Wildlife Conservancy for access to Brooklyn Station on Mt

Lewis. Michelle Nissen and Tony Hess for kind assistance in obtaining QPWS/EPA permits (permit no. WITK05321808). Thanks to Petrina Duncan for kind assistance in the field and Cameron Fletcher for invaluable assistance with logistics.

Photo: L.Warman

Tagged leaflets of Polyscias australiana at the Clohesy wet sclerophyll forest site. The

characteristic bole of Eucalyptus grandis is visible in the background

74 Chapter Three- Leaf type and herbivory

References

Adam P. 1992. Australian rainforests. – Oxford Univ. Press.

Andersen, A. N. and Lonsdale, W. M. 1990. Herbivory by insects in Australian tropical

savannas: A review. J Biogeog 17: 433 - 444.

Baker-Brosh K. F. and Peet R. K. 1997. The ecological significance of lobed and

toothed leaves in temperature forest trees. Ecology 78: 1250–1255.

Basset Y. 1991. Influence of leaf traits on the spatial distribution of insect herbivores

associated with an overstorey rainforest tree. Oecologia 87: 388–393.

Bharathan G. and Sinha N. R. 2001. The regulation of compound leaf development.

Plant Physiol. 127: 1533–1538.

Burns K. C. and Dawson J. W. 2006. A morphological comparison of leaf heteroblasty

between New Caledonia and New Zealand. N. Z. J. Bot. 44: 387–396.

Bragg J. G. and Westoby M. 2002. Leaf size and foraging for light in a sclerophyll

woodland. Funct. Ecol. 16: 633–639.

Brown V. K. et al. 1991. Herbivory and the evolution of leaf size and shape [and

discussion]. Phil. Trans. R. Soc. B 333: 265–272.

Campitelli B. E. et al. 2008. Leaf variegation is associated with reduced herbivore

damage in Hydrophyllum virginianum. Botany 86: 306–313.

Coley P. D. 1983. Herbivory and defensive characteristics of tree species in a lowland

tropical forest. Ecol. Monogr. 53: 209–229.

Coley P. D. 1987. Interspecific variation in plant anti-herbivore properties: the role of

habitat quality and rate of disturbance. New Phytol. 106: 251–263.

DeGabriel J. L. et al. 2008. A simple, integrative assay to quantify nutritional quality of

browses for herbivores. Oecologia 156: 107–116.

75 Chapter Three- Leaf type and herbivory

Dominy N. J. et al. 2008. In tropical lowland rain forests monocots have tougher leaves

than dicots, and include a new kind of tough leaf. Ann. Bot. 101: 1363–1377.

Dudt J. F. and Shure D. J. 1994. The influence of light and nutrients on foliar phenolics

and insect herbivory. Ecology 75: 86–98.

Dublin, H. T., Sinclair et al. 1990. Elephants and Fire as Causes of Multiple Stable

States in the Serengeti-Mara Woodlands. J Animal Ecol. 59: 1147-1164.

Feeney P. 1976. Plant apparency and chemical defense. Rec. Adv. Phytochem. 10: 1-40.

Fine P. V. A. et al. 2004. Herbivores promote habitat specialization by trees in

Amazonian forests. Science 305: 663–665.

Friedman W. E. et al. 2004. The evolution of plant development. Am. J. Bot. 91: 1726–

1741.

Frith C. B. and Frith D. W. 1985. Seasonality of insect abundance in an Australian

upland tropical rainforest. Austral Ecol. 10: 237–248.

Gall L. F. 1987. Leaflet position influences caterpillar feeding and development. Oikos

49: 172–176.

Garland, T. et al. 1992. Procedures for the Analysis of Comparative Data Using

Phylogenetically Independent Contrasts. Syst. Biol. 41: 18-32.

Givnish T. J. 1978a. Ecological aspects of plant morphology: leaf form in relation to

environment. Acta Biotheor. 27: 83–142.

Givnish T. J. 1978b. On the adaptive significance of compound leaves, with particular

reference to tropical trees. – In: TomlinsonP. B. and ZimmermanM. H. (eds),

Tropical trees as living systems. Proc. 4th Cabot Symp. at Harvard Forest,

Petersham, Massachusetts, April 1976. Cambridge Univ. Press, pp. 351–380.

Givnish T. J. 1979. On the adaptive significance of leaf form. – In: SolbrigO. T. et al.

(eds), Topics in plant population biology. Columbia Univ. Press, pp. 375–407.

76 Chapter Three- Leaf type and herbivory

Givnish T. J. 1984. Leaf and canopy adaptations in tropical forests. – In: MedinaE. et al.

(eds), Physiological ecology of plants of the Wet Tropics. Proc. Int. Symp. in

Oxatepec and Los Tuxtlas, Mexico, June 1983. Dr. W. Junk Publishers, pp. 51-84.

Givnish T. J. 1987. Comparative studies of leaf form: assessing the relative roles of

selective pressures and phylogenetic constraints. New Phytol. 106: 131–160.

Givnish T. J. and Vermeij G. J. 1976. Sizes and shapes of liane leaves. Am. Nat. 110:

743–778.

Grimbacher P. S. and Stork N. E. 2009. Seasonality of a diverse beetle assemblage

inhabiting lowland tropical rain forest in Australia. Biotropica 41: 328–337.

Hoffmann W. A. et al. 2005. Specific leaf area explains differences in leaf traits

between congeneric savanna and forest trees. Funct. Ecol. 19: 932–940.

Hopkins M. S. et al. 1993. Charcoal evidence of the spatial extent of the Eucalyptus

woodland expansions and rainforest contractions in North Queensland during the

Late Pleistocene. J. Biogeogr. 20: 357–372.

Huntly N. 1991. Herbivores and the dynamics of communities and ecosystems. Annu.

Rev. Ecol. Syst. 22: 477–503.

Ivey C. T. and DeSilva N. 2001. A test of the function of drip tips. Biotropica 33: 188-

191.

Kurokawa H. et al. 2010. Plant traits, leaf palatability and litter decomposability for co-

occurring woody species differing in invasion status and nitrogen fixation ability.

Funct. Ecol. 24: 513–523.

Løe G. et al. 2007. Trichome production and spatiotemporal variation in herbivory in

the perennial herb Arabidopsis lyrata. Oikos 116: 134–142.

Lowman M. D. 1985. Temporal and spatial variability in insect grazing of the canopies

of five different Australian rainforest tree species. Aust. J. Ecol. 10: 7–24.

77 Chapter Three- Leaf type and herbivory

Lowman, M. D. 1995. Herbivory in Australian forests -a comparison of dry sclerophyll

and rain forest canopies. Proc Linn Soc NSW 115: 77-87

Lowman M. D. and Box J. D. 1983. Variation in leaf toughness and phenolic content

among five species of Australian rain forest trees. Aust. J. Ecol. 8: 17–25.

Lucking R. and Bernecker-Lucking A. 2005. Drip-tips do not impair the development of

epiphyllous rain-forest lichen communities. J. Trop. Ecol. 21: 171–177.

Malhado A. C. M. et al. 2009. Spatial distribution and functional significance of leaf

lamina shape in Amazonian forest trees. Biogeosciences 6: 1577–1590.

Malhado A. C. M. et al. 2010. Are compound leaves an adaptation to seasonal drought

or to rapid growth? Evidence from the Amazon rain forest. Global Ecol.

Biogeogr. doi:10.1111/j.1466-8238.2010.00567.x

Moles A. T. and Westoby M. 2000. Do small leaves expand faster than large leaves, and

do shorter expansion times reduce herbivore damage? Oikos 90: 517–524.

Moles A. T. et al. 2005. A brief history of seed size. Science 307: 576–580.

Nabeshima E. et al. 2001. Effects of herbivory and light conditions on induced defense

in Quercus crispula. J. Plant Res. 114: 403–409.

Niinemets Ü. 1998. Are compound-leaved woody species inherently shade-intolerant?

An analysis of species ecological requirements and foliar support costs. Plant

Ecol. 134: 1–11.

Niinemets Ü. and Kull O. 1999. Biomass investment in leaf lamina versus lamina

support in relation to growth irradiance and leaf size in temperate deciduous trees.

Tree Physiol. 19: 349–358.

Niinemets Ü. and Kull K. 2003. Leaf structure vs nutrient relationships vary with soil

conditions in temperate shrubs and trees. Acta Oecol. 24: 209–219.

78 Chapter Three- Leaf type and herbivory

Niinemets Ü. et al. 2006. Leaf size modifies support biomass distribution among stems,

petioles and mid-ribs in temperate plants. New Phytol. 171: 91–104.

Pérez-Harguindeguy N. et al. 2003. Leaf traits and herbivore selection in the field and in

cafeteria experiments. Austral Ecol. 28: 642–650.

Poorter L. and Bongers F. 2006. Leaf traits are good predictors of plant performance

across 53 rain forest species. Ecology 87: 1733–1743.

Popma J. et al. 1992. Gap-dependence and leaf characteristics of trees in a tropical

lowland rainforest in Mexico. Oikos 63: 207–214.

Rasband W. S. 19972009. ImageJ. – US Natl Inst. of Health

Reich P. B. and Oleksyn J. 2004. Global patterns of plant leaf N and P in relation to

temperature and latitude. Proc. Natl Acad. Sci. USA 101: 11001–11006.

Ritchie M. E. et al. 1998. Herbivore effects on plant and nitrogen dynamics in oak

savanna. Ecology 79: 165–177.

Rudgers J. A. and Whitney K. D. 2006. Interactions between insect herbivores and a

plant architectural dimorphism. J. Ecol. 94: 1249–1260.

Sinha N. 1999. Leaf development in angiosperms. Annu. Rev. Plant Physiol. Plant Mol.

Biol. 50: 419–446.

Stowe L. G. and Brown J. L. 1981. A geographic perspective on the ecology of

compound leaves. Evolution 35: 818–821.

Taylor F. and Hill R. S. 1996. A phylogenetic analysis of the Eucryphiaceae. Aust. Syst.

Bot. 9: 735–748.

Van Ingen L. T. et al. 2008. Ant community structure along an extended rain forest

savanna gradient in tropical Australia. J. Trop. Ecol. 24: 445–455.

79 Chapter Three- Leaf type and herbivory

Warman L. and Moles A. T. 2009. Alternative stable states in Australia's Wet Tropics: a

theoretical framework for the field data and a field-case for the theory. Landscape

Ecol. 24: 1–13.

Webb C. O. et al. 2008. Phylocom: software for the analysis of phylogenetic community

structure and character evolution. Bioinformatics 24: 2098–2100.

Wegner C. et al. 2003. Foliar C:N ratio of ferns along an Andean elevational gradient.

Biotropica 35: 486–490.

Williams S. E. and Marsh H. 1998. Changes in small mammal assemblage structure

across a rainforest/open forest ecotone. J. Trop. Ecol. 14: 187–198.

Wright I. J. et al. 2004. The worldwide leaf economics spectrum. Nature 428: 821–827.

80 Chapter Three- Leaf type and herbivory

Appendices

Appendix 1. Phylogenetic tree of species included in the study, with family and leaf type.

The arrow indicates the node with the highest contribution (0.44) to a marginally significant (p=0.051) difference in herbivory across contrasts between taxa with compound vs simple leaves.

81 Chapter Three- Leaf type and herbivory

Appendix 2. Scatter plots for relationships between herbivory over eight weeks and

area, leaf mass per unit area (LMA) and C:N ratio. Closed circles represent compound

leaves and open circles represent simple leaves.

100 100 100 p=0.77 p=0.82 p=0.51

10 10 10 scale] 10

1 1 1

0.1 0.1 0.1 Herbivory (%) [log Herbivory

0.01 0.01 0.01 1 10 100 0.001 0.01 0.1 10 20 50 100 2 LMA (g/cm2)[log scale] C:N ratio Area (cm )[log10 scale] 10

82 Chapter Three- Leaf type and herbivory

Photo: L.Warman

Chapter Four

A broad approach to abrupt boundaries: Looking beyond the boundary at soil attributes within and

across tropical vegetation.

Laura Warman, Matt G. Bradford and Angela T. Moles

In review at Ecography

This study was conceived by LW who also carried out all of the fieldwork, rock crushing, data analysis, and most of the writing. MGB provided the 1995 data set as well as advice in the field and manuscript revisions. Supervision, input on statistical methods and manuscript revisions by AM.

85 Chapter Four- Soil beyond the boundaries

Abstract

Most research on boundaries between vegetation types emphasizes the contrasts

and similarities between conditions on either side of a boundary, but does not

compare boundary to non-boundary vegetation. While valuable, this approach

may overlook underlying aspects of landscape variability at a regional scale and

underestimate the effects that the vegetation itself has on the soil. We compared

25 soil chemistry variables in rainforest, woodland and across rainforest-

woodland boundaries in north-eastern Queensland, Australia. Studies worldwide

have reported strong contrasts in soil chemistry across vegetation boundaries,

however we did not find greater variation in chemical parameters across

boundary transects than in transects set in either rainforest or woodland. We did

not find that the soil chemistry across boundaries is representative of “rainforest

soil” abutting against “woodland soil”, rather soil on both sides of the boundary

is more similar to rainforest soil than to woodland soil. Transects in wet

sclerophyll forests with increasing degrees of rainforest invasion showed that as

rainforest invades wet sclerophyll forest, the soil beneath wet sclerophyll forest

becomes increasingly similar to rainforest soil. Our results have implications for

understanding vegetation dynamics, and considering soil-vegetation feedbacks

and the differences between soil at boundaries and in non boundary sites may

hold clues to some of the processes that occur across and between vegetation

types in a wide range of ecosystems.

86 Chapter Four- Soil beyond the boundaries

Introduction

Boundaries between different vegetation types occur throughout the world in a variety of environments and range in scale from localized populations and communities to interfaces that span hemispheres (such as the 13,400 km boreal boundary between tundra and taiga; Callaghan et al. 2002, Read et al. 2006, Andersen et al. 2009). Abrupt boundaries between vegetation types are an especially dramatic landscape component where two distinct vegetation forms (e.g. tropical rainforest and fire-prone savannas) abut against each other rather than being separated by an ecotone. These abrupt boundaries have long intrigued researchers and the role of soil in delimiting vegetation and maintaining abrupt boundaries continues to be debated (see Bowman 2000, Durigan and Ratter 2006, Bond 2010, Furley 2010).

While research on abrupt boundaries continues to grow, most studies approach boundaries as the limits of two vegetation types. Many studies, especially those considering soil chemistry, tend to focus on differences found directly across boundaries, on single sites or single transects, on one of the vegetation types involved or on the characteristics and drivers of specific boundaries (Furley et al. 1992, Turton and

Sexton 1996, Bowman 2000, Little et al. 2003, Read et al. 2006, Dick and Gillam

2007). These approaches have undoubtedly increased our understanding of the factors involved in determining boundary locations and dynamics. However, focusing on the boundaries alone may overlook or underestimate underlying aspects of landscape variability and long-term history at a regional scale. This includes the effects of the vegetation itself on the parameters being measured and the effect of environmental processes (such as succession and decomposition) occurring in the boundary zone. In this study we approach abrupt boundaries between rainforest and woodland vegetation

87 Chapter Four- Soil beyond the boundaries

at a regional scale in the Australian Wet Tropics, specifically trying to incorporate local diversity of both soils and vegetation types. We address an existing knowledge gap by comparing soil chemistry under boundaries to ‘controls’ in non-boundary vegetation

(rainforest and woodland far from boundaries) and investigate the changes that occur in soil following the shift from wet sclerophyll forest to rainforest.

Abrupt boundaries between tropical rainforest and open, fire-prone vegetation can be found throughout the world, notably in South America between the cerrado and gallery forest vegetation (Cruz Ruggiero et al. 2002, Hoffmann et al. 2004, Silva et al. 2008); in equatorial Africa (Longman and Jeník 1992, Swaine 1992, Schwartz et al. 1996, Goetze et al. 2006), India (Puyravaud et al. 1994) and in both the wet and wet-dry tropics of northern Australia (Ash 1988, Bowman 1992, Bowman and Cook 2002). Fire has a critical role in boundary stability and maintenance in all of these systems but it is generally accepted that no single factor is solely responsible for these abrupt boundaries

(Unwin 1989, Puyravaud et al. 1994, Hoffmann et al. 2003). In the Wet Tropics of

Australia, it is considered that rainforests flourish on basaltic soils while sclerophyll woodland is more common on less fertile granitic soils (Ash 1988). However, there has been little evidence for edaphic control of boundaries (Duff 1987, Ash 1988, Unwin

1989) and it is increasingly recognized that the importance of parent material has probably been overemphasised, and that there is much variation within and between communities (Bowman 2000).

Globally, abrupt boundaries between rainforest and open vegetation are characterised by strong contrasts in canopy cover, tree density, plant biomass and species composition as well as soil moisture and other microsite conditions (including light penetration, wind,

88 Chapter Four- Soil beyond the boundaries

temperature maxima and minima and leaf litter accumulation; Unwin 1983, Duff 1987,

Furley et al. 1992, Turton and Duff 1992, Puyravaud et al. 1994, Hoffmann et al. 2003,

Silva et al. 2008, Rossatto et al. 2009). In contrast, edaphic characteristics including pH and soil chemistry remain equivocal even at local scales. Some studies report strong contrasts in soil chemistry across boundaries (Hoffmann et al. 2009), while others report weaker and/or contradictory trends (Plowman 1979, Turton and Sexton 1996, Cruz

Ruggiero et al. 2002). This inconsistency may be due to inherent differences between boundaries, but may also reflect a sampling artefact caused by not taking into account the variability at non-boundary sites. The first aim of our study is to determine whether the variability in soil chemistry across boundaries is greater than the natural variability within either rainforest or woodland vegetation at a regional scale. If the strong contrasts that have been reported across some boundaries are specific to the boundaries themselves, then we would expect to find higher variance across boundaries than in similar measures taken wholly within either rainforest or woodland.

Studies around the world have shown that both boundaries and vegetation types have moved back and forth across the landscape through time (Puyravaud et al. 1994,

Hopkins et al. 1996, Silva et al. 2008, Silva et al. 2010). In North Queensland it has been estimated that boundaries can shift horizontally more than a meter per year (Unwin

1989). When studies compare soil directly across boundaries, there is an implicit assumption that they are measuring two distinct entities; in this case “woodland soil” abutting against “rainforest soil”. However, because boundaries are constantly shifting, many studies may have just compared two types of “boundary soil” instead. No previous studies have compared soil directly across boundaries to controls in independent non-boundary vegetation to verify whether or not there is a difference

89 Chapter Four- Soil beyond the boundaries

between boundary and non-boundary soils at regional scales. Rather, studies have compared adjacent contrasting vegetation (e.g. Bowman 1992; Read et al. 2006), looked at variations in soil within a single type of vegetation (e.g. Amorim and Batalha 2007), compared fine-scale differences in either vegetation near the boundary (e.g. Dick and

Gilliam 2007) or compared soil under vegetation types at a regional scale without comparing boundaries (e.g. Webb 1969). The second aim of this paper is to compare soil chemistry across boundaries with the corresponding vegetation type from independent non-boundary sites to ascertain whether the soil on either side of the boundary is actually representative of “woodland soil” and “rainforest soil”. Finding this pattern (and particularly finding the strongest contrasts across the boundaries) would be expected under the current understanding of rainforest distribution in

Australia. However, if one considers rainforest and woodland vegetation to represent alternative stable states, then it would be expected to find much stronger overlap between rainforest, woodland, boundary and non boundary soils. In particular one would expect boundaries to show intermediate values between rainforest and woodland soils.

There is increasing interest in and discussion of vegetation-soil feedbacks at the boundary level, and of how vegetation itself affects soil parameters (Bowman 1992,

Puyravaud et al. 1994, Bowman 2000, Hoffmann et al. 2005, Silva et al. 2008, Hedin et al. 2009, Hoffmann et al. 2009). Numerous studies have shown the effects on soil chemistry of removing, altering or artificially replacing rainforest with other vegetation and of subsequent re-vegetation of cleared areas (Holt and Spain 1986, Schwartz et al.

1996, Herbohn and Congdon 1998, Bautista-Cruz and del Castillo 2005, Paul et al.

2010a, b). Less emphasis has been placed on tracking changes in the soil during

90 Chapter Four- Soil beyond the boundaries

‘natural’ shifts from open to closed forest vegetation. Numerous studies looking at charcoal, soil carbon content and radiocarbon dating (through accelerator mass spectrometry and stable carbon isotope analyses) have shown that both forests and grasslands can leave “footprints” when they replace each other on the landscape (e.g.

Hopkins et al. 1996, Schwartz et al. 1996, Bowman and Cook 2002, Silva et al. 2008), but the effects of vegetation change on other aspects of soil chemistry and fertility are not clear. The third aim of our study is to quantify the effects of the vegetation itself on the soil; specifically, to quantify the effects of rainforest invading wet sclerophyll vegetation in a regional context. If the differences in soil under rainforest and woodland are, at least in part, due to an effect of the vegetation types themselves, then as rainforest invades wet sclerophyll forest, the soil under the latter should progressively change to resemble the soil under rainforest. Furthermore, if the vegetation itself is influencing soil parameters at the boundary, we would expect to find strong differences between rainforest and woodland soil, and for the boundaries to show intermediate

“boundary soil” values. Thinking in terms of alternative stable states, it would be expected to find that the differences in soil are strongly influenced by the vegetation types themselves.

In summary, the hypotheses we test are:

1) The contrasts in soil chemistry across rainforest-woodland boundaries at a regional scale are higher than the variability found within either vegetation type,

2) The soil chemistry on either side of the boundary corresponds more closely to

“boundary zone” soil, rather than either rainforest or woodland soil;

91 Chapter Four- Soil beyond the boundaries

3) There is a measurable effect on the soil of natural changes in the vegetation over time, and as rainforest invades wet sclerophyll forest, the soil under the former will increasingly resemble the latter.

Methods

Study region

The study took place in the Wet Tropics bioregion of north-eastern Queensland,

Australia (Map, Fig. 1). The vegetation of the Australian Wet Tropics is characterized by a mosaic of rainforests and more open, fire -prone sclerophyll vegetation (which ranges from grassy woodlands to eucalypt forests), often with abrupt boundaries between them (Ash 1988, Unwin 1989). The region as a whole is characterized by high variability in soils, topography, rainfall and climate (Hilbert et al. 2007).

A notable feature in the vegetation of the Australian Wet Tropics is the presence of a narrow band of tall open wet sclerophyll forests that forms an ecotone between rainforest and woodland. These forests contain a mixture of species, conditions and processes from both rainforest and woodland vegetation (Warman and Moles 2009 and references within). Under the right conditions, and in the absence of fire, rainforest can rapidly invade, and eventually replace, wet sclerophyll forest vegetation (Harrington et al. 2005). These forests offer an interesting opportunity to test ideas about vegetation- soil feedbacks.

Site selection

Sites were chosen to reflect and incorporate the broad regional diversity across and within vegetation types (Appendix 1). Thus, sites range from rainforest near the coast in

92 Chapter Four- Soil beyond the boundaries

Figure 1 - Map of 2009 transect locations.

Daintree National Park up into the Atherton and Herberton Tablelands as well as woodland vegetation through the Lamb and Herberton ranges. Sites range in altitude from 280 to 847 masl (Fig. 1 and Appendix 2), and regionally rainfall varies from around 900 mm/pa in the Lamb Range to over 3000 mm/pa (near Innisfail). For descriptions of the region’s vegetation see Tracey 1982, Unwin 1983, Duff 1987 and

Harrington and Sanderson 1994.

We chose relatively undisturbed sites, avoiding intersecting waterways, riparian gallery forests and areas extensively covered by invasive species. We sampled relatively flat

93 Chapter Four- Soil beyond the boundaries

areas to avoid runoff or concentration of nutrients down slopes (hillside transects were set across the slope face). Boundary sites were chosen on the basis of strong differences in canopy closure and species composition, particularly reflected in the density of grass cover. Two boundary transects (Surprise Creek and Herberton) were intersected by a track which could ostensibly function as a firebreak.

Soil collection and analysis

Sampling was carried out during April of 2009, at start of the dry season. Eighteen transects were laid out; five each in rainforest, woodland and across abrupt boundaries straddling these two vegetation types. Three additional transects were set up in wet sclerophyll forest sites with varying degrees of invasion by rainforest (see below). All transects were 60 m long and pooled samples were collected at 12 m intervals along each transect, for a total of 107 pooled samples. Each pooled sample consisted of five smaller samples taken to a depth of 15cm within one square meter of the transect line.

Leaf litter and other detritus were brushed away prior to sampling. Parent material was assessed on site. The starting points for transects were randomly chosen and, in the case of boundary transects, three sampling points were placed on each side of the boundary.

Soil samples were dried at 70o C for at least 48 hours then passed through a sieve to remove roots, woody material and small rocks greater than approximately 15mm in diameter. Dried samples were crushed using a TEMA mill at UNSW and then a suite of

25 general soil chemistry and fertility parameters was measured at the Environmental

Analysis Laboratory at Southern Cross University. Measured parameters included organic matter content, macronutrients (carbon, nitrogen, phosphorus, calcium, magnesium, potassium and sulphur), pH, conductivity, and some micronutrients

94 Chapter Four- Soil beyond the boundaries

(Appendices 3 and 4). Phosphorus tests included available P Bray I, potentially available P Bray II and total P Colwell. This suite of parameters was chosen because together they offer a general picture of soil fertility, availability of nutrients to plants

(rather than just total concentration) and biological activity.

Wet sclerophyll forest chronosequence data

As a proxy for a chronosequence, three transects were set up in wet sclerophyll forest sites, each one with an increasing degree of invasion by rainforest. We chose sites where Eucalyptus grandis was present in the canopy because this species can survive under a rainforest, but requires fire and relatively open conditions to establish and regenerate (Harrington and Sanderson 1994). Hence, the presence of E. grandis as canopy species in a mature rainforest indicates a vegetation shift through time from open vegetation to closed forest (Unwin 1983).

The vegetation at the three wet sclerophyll sites corresponds to wet sclerophyll forest types I, II and III as described by Harrington and Sanderson (1994), with E. grandis present and rainforest invasion increasing from type I to III. At the type I site

(Herberton) a mature canopy of Eucalyptus grandis grows over a mixture of grasses, sedges and shrubs in the understory. Type II (at Clohesy River Rd), represents the next progressive stage, with a stand of mature E. grandis and E. grandis/E. pellita hybrids as emergents over a developing rainforest understory. At the third site (Smith’s track, type

III) a few relict emergent E. grandis survive in an established mature rainforest with a canopy around 35 m high. We could find no evidence of regenerating eucalypts at this site.

95 Chapter Four- Soil beyond the boundaries

We combined the data from these three transects with a dataset previously collected in the dry season of 1995. This dataset includes measurements of phosphorus (%), carbon

(%), nitrogen (%) and electro-conductivity (μS/cm) for 235 soil samples from wet sclerophyll forest types I, II and III collected from transects at Baldy Range, the Carbine

Tablelands, Herberton Range, Kirrima Range, Koombooloomba, Lamb Range, Paluma

Range, Wallaman Falls and the Windsor Tableland. The pooled 1995-2009 dataset thus includes 252 samples spanning nearly three degrees of latitude, approximately 350 km north to south and 60 km inland from the coast.

Statistical analysis

All soil parameter measurements, except calcium, magnesium and sodium, were log10 transformed prior to analysis. To calculate within transect variation, we calculated the variance for each parameter measured within each transect and then compared the variances using a general linear model (GLM) with fixed effect for vegetation type in

SPSS (v. 17.0).

Next, we compared the values of soil parameters across boundaries and between boundary and non boundary vegetation taking into account the variation between transects. We used a linear mixed effect model in R (lmer; R Development Core Team

2009) with a random effect for site to compare all parameters in rainforest, woodland, boundary-rainforest and boundary-woodland vegetation. A similar linear mixed effect model was used in R to compare the soil parameters in the samples from wet sclerophyll forest with increasing levels of rainforest invasion. In this case, the 1995 and 2009 datasets were collated according to forest type and the model included a random effect for site. We repeated the tests using only samples that had either granite or rhyolite as

96 Chapter Four- Soil beyond the boundaries

parent material (all samples from basalt were in type I forest) to ascertain whether our results were influenced by parent material, but results were qualitatively unchanged in these tests.

Results

Variance across transects

We began by testing whether variance in soil parameters was higher for boundary transects than for transects in either rainforest or woodland. In other words, whether the changes in soil parameters are greater across boundaries than within a given vegetation type. We did not find the variance in any of the 25 soil parameters to be significantly higher in boundary transects than in transects within either rainforest or woodland (see

Fig. 2 for examples). Our results show some abrupt changes in parameters across some boundaries (Fig. 2c), but these changes are neither consistent across all parameters or boundaries, nor great enough to result in significantly higher variance for boundary transects overall. This lack of consistency across boundaries highlights the importance of comparing multiple boundaries. Additionally, we found that the variance of Bray II phosphorus was significantly higher (p=0.045; Fig. 2a) in rainforest transects than in woodland or boundary transects. In other words, the only soil parameter to show significantly higher variance, described more variability within a single vegetation type than across boundaries between vegetation types. Our results indicate that the differences in soil parameters are not greater along boundaries in the Australian Wet tropics than what could be expected within either rainforest or woodland vegetation.

These results highlight the importance of comparing boundaries to controls in both vegetation types.

97 Chapter Four- Soil beyond the boundaries

Soil parameters across boundaries and vegetation types: general patterns

Next, we tested whether soil chemistry on either side of boundaries corresponds to rainforest and woodland soil or to “boundary zone soil”. For clarity, we divided our results into two sections: general patterns and parameter values (see below). We found no cases where soil parameters in rainforest and at the rainforest boundary were significantly different to woodland and the corresponding woodland boundary. In other words, we did not find that the soil on one side of boundaries is “rainforest soil” abutting against “woodland soil” on the other.

Figure 2- Measurements of soil variables along transects. Each line

represents one transect with six measurements along its length. a) Bray II

phosphorus measurements show significantly higher within-transect

variance in rainforest, than across boundaries or woodlands. b) Aluminium

shows high variance in all transects and vegetation types. The Clohesy

transect (open triangles) in c) shows the type of pattern expected across

boundaries, but overall within-transect variance was not higher across

boundaries than within single vegetation types. There was no significant

difference in organic matter across boundaries. In contrast, C:N ratios (d)

were significantly different across boundaries, but within transect

differences were not higher than in rainforest or woodland vegetation.

Values for both of these vegetation types were significantly different to the

boundary vegetation.

98 Chapter Four- Soil beyond the boundaries

Rainforest Boundary Woodland a) 1000 1000 scale]

10 100 100

10 10

1 1

P (Bray II, ppm) [log II, ppm) P (Bray 0 1224364860 -30 -18 -6 6 18 30 0 1224364860 b) 3000 3000 1000 1000

scale] 300 300 10 100 100

30 30 10 10

3 3

Al (Kg/Ha) [log (Kg/Ha) Al 1 1 0 1224364860 -30 -18 -6 6 18 30 01224364860 c) 100 100

scale] 50 50 10

20 20 10 10 5 5

2 2 1 1 0 1224364860 -30 -18 -6 6 18 30 0 1224364860 Organic matter (%) [log (%) matter Organic d) 100 100

scale] 50 50 10

20 20

10 10 C:N Ratios[log 5 5 0 1224364860 -30 -18 -6 6 18 30 0 1224364860

Distance along transect (m)

99 Chapter Four- Soil beyond the boundaries

However our results aren’t necessarily consistent with the idea of “boundary zone soil” either. Only eight of the 25 soil parameters varied significantly across boundaries

(including electrical conductivity, nitrogen and related measures of C:N ratio, ammonium and nitrate, but not organic matter, pH, carbon or phosphorus; Fig. 3,

Appendix 3). While we found differences between rainforest soils and woodland soils as well as across boundaries, the patterns across the boundaries were not always consistent with the differences between rainforests and woodlands. In fact, nearly a quarter of our results showed that soil on both sides of the boundary was more similar to rainforest soil than to woodland soil. The patterns and/or values of soil parameters across vegetation types may be indicative of environmental processes involving both soil and vegetation in both vegetation types and on the boundaries themselves (Figs. 3 and 4).

Four main patterns appear in our results and account for 20 out of 25 (80%) of the parameters measured. Over a third of the parameters (nine out of 25, including cation exchange capacity, calcium, magnesium and available Bray II phosphorus) show no difference across the four vegetation types (rainforest, woodland and both types of boundary vegetation; Fig. 3a). Five parameters (including soil organic matter, magnesium and carbon) show a difference between woodland and the rest (Fig. 3b).

Four parameters (nitrogen, ammonium, sulphate sulphur and electro-conductivity) show no difference between rainforest and rainforest boundaries, but rainforest and rainforest boundaries are different to both woodland and woodland boundaries (Figs. 3c, 3f, 4b).

Finally, both pH and aluminium were significantly different between boundary and non- boundary vegetation, but showed no difference across boundaries or between rainforests

100 Chapter Four- Soil beyond the boundaries

and woodlands (Figs. 3d and 4c). In no case did we find that the parameters in rainforest soil differed significantly to the rest.

Soil parameters across boundaries and vegetation types: parameter values

Carbon and nitrogen both decreased substantially from rainforest (means= 7.2% C and

0.55% N), to woodland (means= 3.5% C and 0.14% N). Accordingly, C:N ratios dropped sharply from woodland (mean= 29.4) to woodland boundary (mean= 20.8) to rainforest boundary (mean= 18.3) and again to rainforest (mean=13.6), with a significant difference between all vegetation types (p<0.005; Fig. 4). C:N ratio was the only parameter to show an incremental pattern with significant differences between all vegetation types. These results are probably related to time since fire and fire frequencies (Turner et al. 2008) and support global patterns of accumulation of nitrogen by rainforests (Hedin et al. 2009).

Total nitrogen, ammonium and nitrate (different forms of nitrogen in soil) all increase from woodland to rainforest but show different contrasts between the four vegetation types (Figs. 3c, 3f, 3g). These contrasts indicate differential rates of decomposition and mineralization of nitrogen in the different environments. Ammonium and total nitrogen are highest in rainforests and rainforest boundaries, then drop at the woodland boundaries and again in woodlands. Nitrate was the only parameter whose values showed no difference (p=0.12) between woodlands (1.6 ppm) and woodland boundaries

(2.3 ppm). These values were lower than those of boundary rainforest (mean 3.3ppm; p<0.02) and rainforest itself (7.1ppm; p<0.001).

101 Chapter Four- Soil beyond the boundaries

Figure 3- Soil parameters measured in woodland (W), rainforest (RF), wet

sclerophyll forest (WSF) and along transects that crossed boundaries

between woodland (BnW) and rainforest (BnRF). Box plots mark the median

and span from the 25th to the 75th percentiles, with error bars at the 10th and

90th percentiles. Statistical difference is indicated when the letters above the

bars differ (i.e. “a” is not statistically different to “a”, but is statistically

different to “b” or “c”). Wet sclerophyll forest values (2009 data) included as

a comparison. a)-d) Show the four main patterns in our data: a) no difference

in soil between vegetation types; b) a difference between woodland values

and those from the other vegetation types; c) no difference between

rainforest and rainforest boundaries, but these are different to both

woodland and the woodland boundaries; d) a difference between boundary

vegetation and non-boundary vegetation. a) And e) show different results for

soil phosphorus, while c) and g) show different forms of nitrogen (f).

Manganese in f) shows a pattern which could be strongly misinterpreted if

only boundary values had been measured. Values for both of these

vegetation types were significantly different to the boundary vegetation.

102

a) b) c) d) 100 100 200 a a a a a b b b a b c c a b b a 100 scale] 50 scale] 1000 scale] 10 10 50 10 10

20 scale]

10 100 10 20 10 1 5 5 10 2 Bray I P (ppm) [log Bray 2 [log (kg/ha) Al 0.1 1 1 Ammonium (ppm)Ammonium [log W BnW BnRF RF WSF (%) [log matter Organic W BnW BnRF RF WSF W BnW BnRF RF WSF W BnW BnRF RF WSF

e) f) g) h) 1000 1000 a b b c 2 a b c c 50 a a b c ab b ac c 1 scale]

scale] 20 scale]

scale] 100 10

10 0.5 100 10 10 10 0.2 5 10 0.1 2 10 0.05 1 1

0.5 [log (ppm) Mn 0.02 Nitrate (ppm)Nitrate [log N content (%) [log

Colwell P (ppm)Colwell [log 1 0.01 0.2 W BnW BnRF RF WSF W BnW BnRF RF WSF W BnW BnRF RF WSF W BnW BnRF RF WSF

103 Chapter Four- Soil beyond the boundaries

Results for aluminium and pH are interesting both because of the pattern they display and the values themselves. The boundary vegetation showed a significantly lower pH

(p=0.01) and significantly higher aluminium (p<0.001) than did woodlands and rainforest. Soils under woodlands displayed a remarkably wide range of pH values from highly alkaline (8.8 on Mt. Baldy) to highly acidic (3.8 at Surprise Creek), while rainforest soils showed a narrower and more acidic range (6.4 to 4.5).

Two of the phosphorus tests (Bray I and II) showed no difference between vegetation types. The third (Colwell) showed values in boundary transects (19.3 ppm) which were intermediate to those in woodland and rainforest (mean values of 12.05 ppm and 50.4 ppm respectively; p<0.04). Bray measurements are used in agriculture to assess ‘plant available’ phosphorus, while Colwell is quantitative measure of total phosphorus

(Moody 2007). The applicability of Bray assessments of phosphorus availability for natural plant populations is unclear (Bond 2010). Additionally, it has been suggested that Colwell tests may overestimate phosphorus in the soil (Kirchhof et al. 2008). Thus, it may be that levels of phosphorus in the soil increase from woodlands to rainforest (as shown by our Colwell test) but not all the phosphorus in rainforests is freely available to plants, especially if it is bound with iron which was also higher in rainforests (Kirchhof et al. 2008).

Surprisingly cation exchange capacity (CEC; the capacity of the soil to retain cations) was not significantly different across vegetation types (p=0.14), despite strongly diverging mean values for rainforests and woodlands (20.2 and 6.9 cmol+/kg). Our model only explained 2.2% of the variation in the data, with 86% of the difference between sites not accounted for. This would indicate that although rainforest soil can

104 Chapter Four- Soil beyond the boundaries

potentially retain more nutrients (Bailey et al. 2008), there is no difference in this aspect of soil fertility between rainforest, woodlands and the boundaries between them.

Vegetation-soil feedbacks through time

Lastly, we looked for evidence of changes in the soil following a transition from wet sclerophyll forest to rainforest vegetation. Out of the six parameters available in the combined 1995-2009 dataset, four (electrical conductivity, pH, phosphorus and C:N, but neither carbon nor nitrogen alone) showed significant differences across wet sclerophyll forest with progressive invasion by rainforest. These parameters, for the most part, followed similar trends to the results we observed from soil under rainforest and woodlands. In other words as the vegetation shifts progressively towards that of an established rainforest, the soil of the wet sclerophyll vegetation appears to become increasingly similar to that of a rainforest.

C:N ratios were lower in wet sclerophyll forest with advanced invasion by rainforest

(type III; mean= 18.8, p= 0.003) than in either of the two previous stages (which showed no difference between them; means= 28.03 and 25.9, p>0.2). Values for electrical conductivity (eC; a measure of ions in the soil and a proxy for salinity) in wet sclerophyll forests (mean= 57.5 μS/cm) were lower overall than in woodland or rainforest vegetation (means of 86.26 and 191.8 μS/cm respectively), but they increased with advancing rainforest invasion (45.32 μS/cm in Type I forest to 94.41 μS/cm in

Type III) much in the same way as they do from woodland to rainforest (Fig. 4,

Appendix 3).

105 Chapter Four- Soil beyond the boundaries

Figure 4- Soil parameters measured in woodland (W), rainforest (RF) and along

transects that crossed boundaries between woodland (BnW) and rainforest

(BnRF), compared to wet sclerophyll forest (WSF) with increasing degrees of

invasion by rainforest (WSF1, WSF2 and WSF3 respectively). Box plots mark

the median and span from the 25th to the 75th percentiles with error bars at

the 10th and 90th percentiles. Statistical difference is indicated when the letters

above the bars differ (i.e. “a” is not statistically different to “a”, but is

statistically different to “b” or “c”). a) C:N values decrease from woodlands to

rainforests, and with increasing degrees of rainforest invasion in wet

sclerophyll forest. A similar but increasing pattern can be seen in b), the

electrical conductivity results. In c), pH drops as rainforest invades wet

sclerophyll forest, even though there is no overall significant difference

between rainforest and woodlands.

106 Chapter Four- Soil beyond the boundaries

Vegetation types Wet Sclerophyll a) 100 a b c d AAB 70

50 scale]

10 35

20 C:N Ratio [log 10

W BnW BnRF RF WSF1 WSF2 WSF3 b) 1000 a b c c A B C scale] 500 10

200

100

Electrical Conductivity [log Electrical Conductivity W BnW BnRF RF WSF1 WSF2 WSF3

c) 10 a bb a A B C 9 8 7

scale] 6 10 5 pH [log 4

3 W BnW BnRF RF WSF1 WSF2 WSF3

107 Chapter Four- Soil beyond the boundaries

pH dropped significantly with increasing rainforest invasion (from mean= 4.9 in Type I to means of 4.8 and 4.5 respectively in types II and III; p<0.006). This is interesting considering that although the mean pH for rainforest (5.66) was lower than in woodland

(6.02), these values were not significantly different (p=0.43). Additionally, pH along the boundary transects was significantly lower (mean=5.1) and their value is closer to that of wet sclerophyll forest on average (4.8). Results for soil phosphorus are difficult to interpret as the element decreases (p<0.001) from early to mid-stage wet sclerophyll vegetation (Type I to Type II; means=0.031 and 0.022% respectively), but then shows no difference between mid and late stages (Type II and III; p=0.99) or early and late stages (Type I and III; p=0.49).

Overall, if we interpret the data from the three types of wet sclerophyll forest as a proxy for a chronosequence, then we can infer that as rainforest invades and overtakes wet sclerophyll vegetation, the vegetation itself changes the parameters of the soil. This, together with the broader picture presented by comparing the boundary transects to

‘non-boundary’ rainforest and woodlands allows us to make comparisons between the vegetation and with previous studies carried out both in Australia and elsewhere.

Discussion

Changes in soil parameters across boundaries vs. within vegetation type

Within-transect variance was not higher across boundaries than within either woodland or rainforest vegetation. Considering contrasts across boundaries in the context of non- boundary variability allows for a more meaningful interpretation of the patterns in values of soil parameters. Rather than just identifying differences and similarities across boundaries or between vegetation types, this approach gives us a framework to

108 Chapter Four- Soil beyond the boundaries

understand whether these differences matter on a larger scale. In this case, our results indicate that contrasts across boundaries are less informative than previously thought as non-boundary soils present similar levels of variability to those at the boundary. Using appropriate non-boundary controls could provide important context and insights in studies of edge-effects, treelines, and a wide variety of comparisons between adjacent contrasting communities and ecosystems.

Our results also create a context for understanding earlier studies restricted to boundaries. For example, previous studies in Australian vegetation have found seemingly contradicting results for differences in pH across boundaries. Plowman

(1979) found pH in wet sclerophyll forest to be lower than in rainforest, while Turton and Sexton (1996) found pH to be lower in rainforest than in woodland and attributed the apparent contradiction with Plowman’s (1979) results to differences in parent material. We found pH decreased from woodland (6.023) to rainforest (albeit non- significantly different at 5.66) and was lowest in wet sclerophyll forest (4.8). Thus, our results are consistent with both studies and suggest that the pH differences described can be attributed to contrasts between ‘wet’ and ‘dry’ sclerophyll vegetation. Using this approach on boundaries elsewhere (such as the cerrado-rainforest boundaries of Brazil) might allow for better insights into the relative importance of soil chemistry in different systems and for more meaningful comparisons between systems (see below).

Soil parameters within and across vegetation types

Our results are not entirely consistent with having measured either “boundary soil” or contrasting “rainforest soil vs. woodland soil” across boundaries. Rather we found both differences and similarities across boundaries and between boundary and non boundary vegetation. In general, open forest transects had lower levels of carbon, nitrogen, soil

109 Chapter Four- Soil beyond the boundaries

organic matter and the highest C:N ratios, all of which are consistent with higher fire frequencies and presence of established forests (Menaut et al. 1993, Ludwig et al. 1998,

Turner et al. 2008, Hedin et al. 2009). While we do not have fire histories for our field sites, woodlands have higher fire frequencies than rainforests, and rainforest near a boundary is exposed more often to fire than rainforest in a massif. Higher nitrogen levels, and lower C:N ratios in rainforest have also been found in previous comparisons of closed and open vegetation, and studies across boundaries in Australia, Africa, India and Brazil (Swaine 1992, Puyravaud et al. 1994, Turton and Sexton 1996, Silva et al.

2008, Hedin et al. 2009, Hoffmann et al. 2009).

Similarities and differences between nitrogen, ammonium and nitrate merit closer consideration as they are indicators of environmental processes other than just fire frequency. Whereas levels of ammonium reflect rates of accumulation of nitrogen through decomposition, nutrient cycling and accumulation of organic matter, lower nitrate levels indicate lower bacterial activity (Baldock and Skjemstad 1999). As nitrate is more readily available to vegetation than ammonium, and particularly important to rainforest pioneer species (Turnbull 1999), the relative differences between ammonium and nitrate across boundaries could play a role in what species and thus which succesional pathways are able to establish across boundaries. Mycorrhizae may also play an important role in plant establishment across vegetation types (and in the invasion of wet sclerophyll by rainforest), as it is thought they help plants exploit nitrogen in non-nitrate form (Bowman and Panton 1993, Turnbull 1999).

Finding significant differences in pH and aluminium between the boundaries and non- boundary vegetation was surprising. Aluminium was the only cation that mirrored pH

110 Chapter Four- Soil beyond the boundaries

levels, which may be an indication that soils at the boundaries are acid mineral soils.

That is, that exchangeable aluminium cations, rather than hydrogen cations are responsible for the low pH (Kamprath 1972). High aluminium and low pH (<5.5) also indicates potential aluminium toxicity for some species at the boundaries (Blamey

1999). It is not obvious how aluminium toxicity affects natural communities (in contrast to agricultural vegetation), but it may influence which species can survive in the boundary zone.

Regarding soil physical parameters, our results show rainforest predominantly on clayey soils and most sclerophyll vegetation on shallower gravelly or sandy soils (Appendix 2).

Although this fits the perceived pattern of rainforests growing on soils that are better able to retain moisture, Australian rainforests have not been shown to be restricted to soil with specific textural properties (Bowman 2000). Indeed, studies throughout the continent have consistently failed to find a relationship between the presence of rainforest and a variety of physical properties including clay or sand content, field capacity, saturation moisture content, aeration or texture (Bowman 2000 and references within). The substrate at the boundary sites sampled in this study ranged from rich basaltic soil to sandy substrates but none of the boundaries sampled coincided with a change in parent material or soil texture.

Our study shows interesting contrasts and similarities with savanna-forest boundaries in

Brazil and India. For example, Puyravaud et al. 1994, Silva et al. 2008, and Hoffmann et al. 2009 found differences across boundaries in parameters including carbon, magnesium and iron, where we did not. However at a larger scale, we did find differences in these elements between open forest and rainforest. These differences and

111 Chapter Four- Soil beyond the boundaries

similarities may be due to stronger chemical contrasts across boundaries outside the

Australian Wet Tropics, but the previous results might become more similar to ours if comparisons to non-boundary vegetation and of variance along boundaries were incorporated. It may also be that grassland-forest boundaries present stronger chemical contrasts than do woodland-closed forest boundaries because of more similar litter- cycling in the latter (see Little et al. 2003).

Changes in soil through time

The results from our first two questions become particularly interesting in light of the results from the third. The accumulation of certain nutrients (particularly nitrogen and carbon) by rainforests has been discussed previously both in Australian and global contexts (e.g. Bowman 1992, Hedin et al. 2009). Our data support the hypothesis that as rainforest invades wet sclerophyll vegetation, rainforest vegetation itself changes the soil. With increased rainforest invasion, over periods that probably span a few hundred years, the soil under wet sclerophyll vegetation becomes increasingly similar to rainforest soil. Taking vegetation-soil feedbacks into account allows for a different, more dynamic interpretation both of soils and vegetation at the boundaries. For example, soil-vegetation feedbacks may help explain why both sides of the boundaries tend to be more similar to rainforests, and why we found no cases where soil parameters in rainforest were different to the other vegetation types while this was often the case for woodlands. If the boundaries between rainforest and woodlands are relatively fluid and have continued to move back and forth across the landscape through time, then we should expect to find the “signature” of recent rainforest “inputs” into the soil near current boundaries.

112 Chapter Four- Soil beyond the boundaries

Implications for understanding the distribution of Australian rainforest

In Australia the debate over the causal factors involved in abrupt vegetation boundaries has been made more complicated by the ongoing debate over the nature and definition of Australian rainforest, as well as what limits its growth and distribution across the continent. Some authors have concentrated on degree of canopy closure as a defining factor for Australian rainforests (notably Specht and Morgan 1981). Others have focused instead on perceived biogeographical origins (i.e. Hooker 1860 and Cromer and

Pryor 1943) or sensitivity to repeated fire (Jackson 1968, Ash 1988). Disagreements over what exactly defines vegetation types have strongly influenced different researchers’ views on what factors determine the extent and distribution of vegetation.

For example, Beadle (1954, 1962) was a strong proponent of the idea that relatively low phosphorus content in the soil restricts the distribution of vegetation types and taxa in

Australia. When Coaldrake and Haydock (1958) obtained results in Queensland that contradicted Beadle’s studies in New South Wales, Beadle (1962) argued that these results were not contradictory based on a different interpretation of the vegetation types that were examined. Since then, it has been shown that rainforests occur on soils with a wide range of phosphorus concentrations, and that phosphorus interacts with other factors such as slope or fire frequency (Bowman 2000). However Adam (1989) found that rainforest genera are more numerous on soils with more than 200ppm of total phosphorus, while sclerophyll taxa are generally more diverse on soils with lower total phosphorus.

Our study fit the pattern of rainforests occurring on soils with higher total phosphorus

(Colwell P, ppm) than sclerophyll vegetation. However, all samples, including those from rainforest, showed levels of total soil phosphorus lower than 200ppm (Fig 5a and

113 Chapter Four- Soil beyond the boundaries

Appendix 3). This illustrates the high variability in total phosphorus content (and other soil nutrients in general) described by Bowman (2000) and the difficulty in trying to define general rules in such complex variable systems.

a) b)

Figure 5- a) Total phosphorus concentrations (ppm, log10 scale) from transects in

rainforest (Rf), woodland (W) and on cross-boundary transects that span from

rainforest (BnRf) to woodland (BnW). The lowest phosphorus values were found in

woodland and the highest in rainforest (lmer, p<0.04). b) The same data overlaid

with a classic hysteresis curve (see Chapter 2). Although the data and results are the

same, this representation is consistent with what would be expected under a system

where rainforest and sclerophyll vegetation function as alternative stable states.

That is, areas with low total phosphorus (<7ppm) support woodlands and sclerophyll

vegetation and areas with high total phosphorus (>55ppm) support rainforest. The

grey area represents intermediate phosphorus concentrations expected to be able

to support rainforest, woodland and a mosaic containing both types of boundary

vegetation as stable ecosystem states.

114 Chapter Four- Soil beyond the boundaries

Taking into account the historical view of Australian vegetation controlled by a series of abiotic parameters, one would expect that vegetation boundaries would represent

“rainforest soil” abutting “woodland soil”. In other words that the important differences in soil nutrient results would be found across boundaries (i.e. between ‘boundary rainforest soil’ and ‘boundary woodland soil’, but not expecting to find big differences between rainforest and boundary rainforest soil, or between woodlands and woodland boundaries (Fig 6). None of our results matched this pattern. Under a paradigm where the vegetation is controlled solely by a series of interacting abiotic factors (i.e. the classic understanding of Australian vegetation), one finds ‘rules of thumb’ that are linked to a long series of exceptions (such as the of 200pmm total P ‘limit’ for rainforest genera). However, if one approaches the vegetation of the Australian Wet Tropics as a system where rainforest and sclerophyll vegetation function as alternative stable states, then a different perspective is possible. Under this scenario one might expect rainforest occurring on the soils with the highest total soil phosphorus, and sclerophyll vegetation on soils with the lowest total phosphorus; but one would also expect a wide range of total phosphorus concentrations where both vegetation types and boundaries between them are possible. This is consistent with the pattern that can be observed in our data

(Fig 5).

As with our results regarding pH in wet sclerophyll vs. woodland vs. rainforest vegetation, considering these vegetation types as part of a system with alternative stable states allows us to put the seemingly contradictory data within a theoretical framework that makes sense of the contrasts. Importantly, thinking in terms of alternative stable states also provides a framework that allows thinking in terms of relative scales instead of absolute values. Under this scenario, the results found by Coaldrake and Haydock

115 Chapter Four- Soil beyond the boundaries

(1958) and Beadle (1962) become complimentary rather than contradictory. By the same token, the high variability in abiotic factors for a given vegetation type (or for different communities within a same region (e.g. Bowman 2000) becomes an informative set of data, rather than exceptions to a perceived general pattern. Soil parameter X

W BnW BnRf Rf

Figure 6- Result pattern expected for the majority of soil nutrients, if soil nutrients

were determining the location of vegetation types and boundaries. Not only was this

not the general pattern found, but none of our results followed this pattern. This

scenario is not consistent with alternative stable states, because it implies that, for

the most part, only one vegetation state can exist for any given soil parameter

conditions.

Insights from wet sclerophyll vegetation

An unexpected outcome of our study was finding that several important parameters in wet sclerophyll forest soils were, on average, more similar to boundary transects than to 116 Chapter Four- Soil beyond the boundaries

either rainforest or woodland soils. This raises interesting questions relating to differences between “transitional” soils in comparison to those of “established” vegetation. A series of environmental processes related to species changes and changing site conditions (shade, soil moisture maxima and minima, etc) could result in differences between soil deep within tracts of established vegetation and soils at the boundaries and in transitional zones like the wet sclerophyll forests. The role of species from different succesional stages feeding back different elements into the soil could be particularly interesting in this context. For example, high levels of aluminium could be linked to species that are hyper-accumulators (such as Ceratopetalum apetalum; Webb

1954). Species of Gossia and Macadamia, both rainforest genera found in far north

Queensland have recently been found to accumulate metals in their foliage (Fernando et al. 2009).

The wet sclerophyll forests of north-eastern Australia provide a window into the processes of transition from open to closed vegetation, and these forests seem to have similarities with the wet savannas and cerradão of South America (Furley et al. 1992).

Comparing boundary to non-boundary vegetation in these South American vegetation types, and elsewhere in the world may provide interesting insights and comparisons into the dynamics of boundaries and transitions between open and closed forests. Obtaining a better understanding of abrupt vegetation boundaries and the transitions between vegetation types could prove to be especially important in understanding large-scale and long-term vegetation dynamics, especially in the face of climate change and for restoration and conservation worldwide.

117 Chapter Four- Soil beyond the boundaries

Conclusions

Our first finding was that variance in soil parameters was not higher across boundaries than within vegetation types. Thus, to be able to interpret and understand processes occurring across boundaries, it is important to include controls in both vegetation types, away from the boundaries. This finding has strong implications for other boundary and edge effect studies in ecology, not just in tropical vegetation, but as far reaching as studies of boundaries caused by currents and eddies in marine systems (e.g. Baltar et al.

2010). Our second main finding was that soil parameters across boundaries did not correspond entirely either to “boundary soil” or contrasting “rainforest soil vs. woodland soil”. Rather, our results reflect a complex mix of feedbacks, vegetation

‘footprints’ and ongoing processes at a variety of scales. This reiterates the importance of comparing boundary to independent non-boundary sites. Our third main finding was that rainforest vegetation transforms “wet sclerophyll forest soil” into “rainforest soil”.

Taken together, our results present a comprehensive picture of plant-soil feedbacks in both boundary and non-boundary vegetation in Australia’s Wet Tropics. Our results have implications for studies comparing transitional and established vegetation, and possibly for restoration efforts as well. Perhaps more importantly, our study emphasizes the importance of leaving behind the pervasive view of rainforests as static, pyrophobic and monolithic and incorporating more dynamic long-term views of landscape processes and interactions between vegetation types and communities.

Acknowledgements: UNSW for Science Faculty grant awarded to A. Moles and L.

Warman and UIPA scholarship awarded to L.Warman. Thanks to Tony Hess for kind

118 Chapter Four- Soil beyond the boundaries

assistance in obtaining QPWS/EPA permits (# WITK05321808), and Will Edwards and

Cameron Fletcher for invaluable help with soil collection logistics.

References

Adam, P. et al. 1989. Species-richness and soil phosphorus in plant communities in

coastal New South Wales. - Aust J Ecol 14: 189-198.

Amorim, P. K. and Batalha, M. A. 2007. Soil-vegetation relationships in hyperseasonal

cerrado, seasonal cerrado, and wet grassland in Emas National Park (central

Brazil). - Acta Oecol 32: 319-327.

Andersen, K. M. et al. 2009. Soil-based habitat partitioning in understorey palms in

lower montane tropical forests. - J Biogeogr 37: 278-292.

Ash, J. 1988. The location and stability of rainforest boundaries in north-eastern

Queensland, Australia. - J Biogeogr 15: 619-630.

Bailey, J. S. et al. 2008. Relationships between important soil variables in moderately

acidic soils (pH 5.5) in the highlands of Papua New Guinea and management

implications for subsistence farmers. - Soil Use Manage 24: 281-291.

Baldock, J. A. and Skjemstad, J. O. 1999. Soil organic carbon/soil organic matter. - In:

Peverill, K. I., Reuter, D. J., Sparrow, L.A. (ed.) Soil analysis: an interpretation

manual. CSIRO Publishing, 369p.

Baltar et al. 2010. Mesoscale eddies: hotspots of prokaryotic activity and differential

community structure in the ocean. - ISME J 4: 975-988.

Bautista-Cruz, A. I. and Castillo, R. F. 2005. Soil changes during secondary succession

in a tropical montane cloud forest area. Soil Sci. Soc. Am. J. 69: 906-914.

Beadle, N. C. W. 1954. Soil phosphate and the delimitation of plant communities in

eastern Australia.- Ecol 35: 370-375.

119 Chapter Four- Soil beyond the boundaries

Beadle, N. C. W. 1962. Soil Phosphate and the Delimitation on Plant Communities in

Eastern Australia II. - Ecology 43: 281-288.

Blamey, F. P. C. 1999. Soil acidity and toxicities. - In: Atwell, B., Kriedemann, P. and

Turnbull, C. (eds.), Plants in Action: Adaptation in nature, performance in

cultivation. Macmillan Publishers Australia Pty. Ltd.

Bond, W. 2010. Do nutrient-poor soils inhibit development of forests? A nutrient stock

analysis. - Plant Soil 334: 47-60.

Bowman, D. M. J. S. 1992. Monsoon forests in north-western Australia. II. Forest-

savanna transitions. - Aust J Bot 40: 89-102.

Bowman, D. M. J. S. 2000. Australian Rainforests: Islands of green in a land of fire. -

Cambridge University Press.

Bowman, D. M. J. S. and Panton, W. J. 1993. Factors that Control Monsoon-Rainforest

Seedling Establishment and Growth in North Australian Eucalyptus Savanna. – J

Ecol 81: 297-304.

Bowman, D. M. J. S. and Cook, G. D. 2002. Can stable carbon isotopes in soil carbon

be used to describe the dynamics of Eucalyptus savanna-rainforest boundaries in

the Australian monsoon tropics? - Austral Ecol 27: 94-102.

Callaghan, T. V. et al. 2002. The Dynamics of the Tundra-Taiga Boundary: An

Overview and Suggested Coordinated and Integrated Approach to Research. -

Ambio: 3-5.

Coaldrake, J. E., and K. P. Haydock. 1958. Soil Phosphate and Vegetal Pattern in Some

Natural Communities of South-Eastern Queensland, Australia. Ecol 39:1–5.

Cromer D.A.N. and Pryor, L.D. 1943. A contribution to rain-forest ecology. Proc Linn

Soc NSW 67: 249-268.

120 Chapter Four- Soil beyond the boundaries

Cruz Ruggiero, P. G. et al. 2002. Soil-vegetation relationships in cerrado (Brazilian

savanna) and semideciduous forest, Southeastern Brazil. - Plant Ecol 160: 1-16.

Dick, D. and Gilliam, F. 2007. Spatial heterogeneity and dependence of soils and

herbaceous plant communities in adjacent seasonal wetland and pasture sites. -

Wetlands 27: 951-963.

Duff, G. A. 1987. Physiological ecology and vegetation dynamics of North Queensland

upland rainforest-open forest ecotones. PhD Thesis. James Cook University, Qld.

Durigan, G. and Ratter, J. A. 2006. Successional changes in cerrado and cerrado/forest

ecotonal vegetation in western São Paulo State, Brazil, 1962-2000. - Edinburgh J

Bot 63: 119-130.

Fernando, D. R. et al. 2009. Foliar Mn accumulation in eastern Australian herbarium

specimens: prospecting for ‘new’ Mn hyperaccumulators and potential

applications in taxonomy. - Ann Bot London 103: 931-939.

Furley, P. 2010. Tropical savannas: Biomass, plant ecology, and the role of fire and soil

on vegetation. - Prog Phys Geog 34: 563-585.

Furley, P. A. et al (eds.) 1992. Nature and Dynamics of Forest-Savanna Boundaries.

Chapman & Hall.

Goetze, D. et al. 2006. Dynamics of forest–savanna mosaics in north-eastern Ivory

Coast from 1954 to 2002 -J Biogeog 33: 653-664.

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. - Pac Cons

Bio 1: 319-327.

Harrington, G. N. et al. 2005. The Wet Sclerophyll and adjacent forests of north

Queensland: A directory to vegetation and physical survey data. - Cooperative

121 Chapter Four- Soil beyond the boundaries

Research Centre for Tropical Rainforest Ecology and Management. Rainforest

CRC.

Hedin, L. O. et al. 2009. The nitrogen paradox in tropical forests ecosystems. - Annu.

Rev. Ecol. Evol. Syst 40: 613-635.

Herbohn, J. L. and Congdon, R. A. 1998. Ecosystem dynamics of disturbed and

undisturbed sites in north Queensland wet tropical rain forest. III. Nutrient returns

to the forest floor through litterfall. – J Trop Ecol 14: 217-229.

Hilbert, D. W. et al. 2007. Glacial and interglacial refugia within a long-term rainforest

refugium: The Wet Tropics Bioregion of NE Queensland, Australia. - Palaeogeogr

Palaeoecol 251: 104 - 118.

Hoffmann, W. A. et al. 2003. Comparative fire ecology of tropical savanna and forest

trees. - Funct Ecol 17: 720-726.

Hoffmann, W. A. et al. 2004. Constraints to seedling success of savanna and forest trees

across the savanna-forest boundary. - Oecologia 140: 252-260.

Hoffmann, W. et al. 2005. Seasonal leaf dynamics across a tree density gradient in a

Brazilian savanna. - Oecologia 145: 306-315.

Hoffmann, W. et al. 2009. Tree topkill, not mortality, governs the dynamics of savanna-

forest boundaries under frequent fire in central Brazil. -Ecology 90: 1326-1337.

Holt, J. A. and Spain, A. V. 1986. Some biological and chemical changes in a North

Queensland soil following replacement of rainforest with Araucaria cunninghamii

(Coniferae: Araucariaceae). - J Appl Ecol 23: 227-237.

Hooker J.D. 1860. The botany of the Antarctic voyage of H.M. Discovery ships Erebus

and Terror in the years 1839-1843. Part III.Flora Tasmaniae. Vol I.

Dicotyledones. Reeve Brothers, London (available at www.archive.org)

122 Chapter Four- Soil beyond the boundaries

Hopkins, M. S. et al. 1996. Evidence of a Holocene and continuing recent expansion of

lowland rain forest in humid, tropical North Queensland. -J Biogeogr 23: 737-745.

Jackson, W. D. 1968. Fire, air, water and earth- an elemental ecology of Tasmania. -

Proc Ecol Soc Aust 3: 9-16.

Kamprath, E. J. 1972. Soil acidity and liming. - In: Soils of the humid tropics. National

Academy of Sciences.

Kirchhof, G. et al. 2008. An evaluation of Colwell-P as a measure of plant-available

phosphorus in soils of volcanic and non-volcanic origins in the highlands of Papua

New Guinea. - Soil Use Manage 24: 331-336.

Little, D. et al. 2003. Soil & regolith patterns and nutrient processes across an abrupt

woodland-grassland ecotone. - In: Roach, I. C. (ed.) Advances in Regolith. CRC

LEME, Australia.

Longman, K. A. and Jeník, J. 1992. Forest-savanna boundaries: general considerations.

- In: Furley, P. A., Proctor, J. and Ratter, J. A. (eds.), Nature and Dynamics of

Forest-Savanna Boundaries. Chapman & Hall.

Ludwig, B. et al. 1998. Modelling cation composition of soil extracts under ashbeds

following an intense slashfire in a eucalypt forest. - Forest Ecol Manag 103: 9-20.

Menaut, J.-C. et al. 1993. Nutrient and organic matter dynamics in tropical ecosystems.

- In: Crutzen, P. J. and Goldammer, J. G. (eds.), The Ecological, Atmospheric and

Climatic Importance of Vegetation Fires. John Wiley & Sons Ltd, pp. 215-231.

Moody, P. W. 2007. Interpretation of a single-point P buffering index for adjusting

critical levels of the Colwell soil P test. – Aust J Soil Res 45: 55-62.

Paul, M. 2010a. Recovery of soil properties and functions in different rainforest

restoration pathways. - Forest Ecol Manag 259: 2083-2092.

123 Chapter Four- Soil beyond the boundaries

Paul, M. et al. 2010b. Does soil variation between rainforest, pasture and different

reforestation pathways affect the early growth of rainforest pioneer species? -

Forest Ecol Manag 260: 370-377.

Plowman, K. P. 1979. Litter and soil fauna of two Australian subtropical forests. - Aust

Ecol 4: 87-104.

Puyravaud, J.-P. et al. 1994. Ecotone Structure as an Indicator of Changing Forest-

Savanna Boundaries (Linganamakki Region, Southern India). - J Biogeog 21:

581-593.

R Development Core Team. 2009. R: A language and environment for statistical

computing. R Foundation for statistical computing. Vienna, Austria.

Read, J. et al. 2006. Does soil determine the boundaries of monodominant rain forest

with adjacent mixed rain forest and maquis on ultramafic soils in New Caledonia?

– J Biogeogr 33: 1055-1065.

Rossatto, D. R. et al. 2009. Differences in growth patterns between co-occurring forest

and savanna trees affect the forest–savanna boundary. - Funct Ecol 23: 689-698.

Schwartz, D. et al. 1996. Present dynamics of the savanna-forest boundary in the

Congolese Mayombe: a pedological, botanical and isotopic (13C and 14C) study. -

Oecologia 106: 516-524.

Silva, L. C. R. et al. 2008. Expansion of gallery forests into central Brazilian savannas. -

Glob Change Biol 14: 2108-2118.

Silva, L. et al. 2010. Not all forests are expanding over central Brazilian savannas. -

Plant Soil 333: 431-442.

Specht, R. L. and Morgan, D. G. 1981. The balance between the foliage projective

covers of overstorey and understorey strata in Australian vegetation. - Austral

Ecol 6: 193-202

124 Chapter Four- Soil beyond the boundaries

Swaine, M. D. 1992. Characteristics of dry forest in West Africa and the influence of

fire. - J Veg Sci 3: 365-374.

Tracey, J. G. 1982. The vegetation of the humid tropical region of far North

Queensland. - CSIRO.

Turnbull, M. H. 1999. Rainforest succession and nitogen nutrition. - In: Atwell, B.,

Kriedemann, P. and Turnbull, C. (eds.), Plants in Action: Adaptation in nature,

performance in cultivation. Macmillan Publishers Australia Pty. Ltd.

Turner, J. et al. 2008. Long term accumulation of nitrogen in soils of dry mixed

eucalypt forest in the absence of fire. - Forest Ecol Manag: 256: 1133-1142.

Turton, S. M. and Duff, G. A. 1992. Light environments and floristic composition

across an open forest-rainforest boundary in northeastern Queensland. - Austl

Ecol 17: 415-423.

Turton, S. M. and Sexton, G. J. 1996. Environmental gradients across four rainforest-

open forest boundaries in northeastern Queensland. - Aust Ecol 21: 245-254.

Unwin, G. L. 1983. Dynamics of the rainforest - eucalypt forest boundary in the

Herberton Highland, North Queensland. MSc. Thesis. James Cook University,

Queensland, Australia.

Unwin, G. L. 1989. Structure and Composition of the Abrupt Rain-Forest Boundary in

the Herberton Highland, North Queensland. - Aust J Bot 37: 413-428.

Warman, L. and Moles, A. T. 2009. Alternative stable states in Australia’s Wet Tropics:

a theoretical framework for the field data and a field-case for the theory -

Landscape Ecol 24: 1-13.

Webb, L. J. 1954. Aluminium accumulation in the Australian–New Guinea flora. - Aust

J Bot 2: 176-196.

125 Chapter Four- Soil beyond the boundaries

Webb, L. J. 1969. Edaphic Differentiation of Some Forest Types in Eastern Australia:

II. Soil Chemical Factors. - J Ecol 57: 817-830.

Appendices

Appendix 1 - Site photographs (All photos by L. Warman)

1- Characteristic boundary between closed rainforest and woodland dominated by grasses, cycads and eucalypts.

2- Woodland near the Davies Creek Site.

126 Chapter Four- Soil beyond the boundaries

3- Complex rainforest in the Daintree.

4- Wet sclerophyll forest. Eucalyptus grandis over a grassy understorey.

127 Chapter Four- Soil beyond the boundaries

Appendix 2 - Site descriptions

Note coordinates are often approximate, taken from the nearest site where it was possible to obtain a GPS signal.

Site Altitude Location Description (m) Woodland with grass and sedge understory, occasional Surprise 16° 51.328'S mossy beds and terrestrial orchids near a treefall. Creek 383 145° 38.435'E Metamorphic, whitish soils. Woodland site, with dense understory of sedges and Davies 16° 59.885'S grasses and regenerating woody plants under sparse Creek 2 456 145°33.897'E Eucalyptus spp. Red clay soil. Woodland. Grassy (Themeda spp.) understory with Xanthorrhea sp. Acacia spp. and Allocasuarina torulosa Davies 17°00.714'S in the canopy. Greyish gravelly soil, bare granitic rock in Creek 1 623 145°34.960'E spots. Woodland with Allocasuarina torulosa, Eucalyptus spp., Robson 17° 9.356'S Acacia spp. and a Themeda triandra understory. Track 280 145°42.088'E Shallow sandy soils over granite. Woodland with open Themeda triandra understory and 17° 16.704'S some cycads. Eucalyptus and Allocasuarina torulosa Mt. Baldy 847 145° 27.765'E canopy. Shallow soils over rhyolite. Lowland rainforest within Daintree National Park. Complex mesophyll vine forest. Canopy to 30-40 m. 16° 14.413'S Trees with large buttresses, many vines and epiphytes. Jindalba 353 145° 25.956'E Thick reddish clay. Rainforest with open understory under a dense canopy. Plentiful Calamus spp. and Alyxia ruscifolia; mature Smith's 16°53.327'S Agathis robusta, Podocarpus sp. and Gossia spp. Track 449 145°38.813'E present. Soil with high clay content. Mature upland rainforest on basalt. Relatively dense 17° 17.216'S understory. Ficus spp., Argyrodendron spp. and Gossia Eacham 756 145° 37.776'E spp.common. Dense orange soil with high clay content. Rainforest with remarkably sparse and open understory 17° 25.805'S under a dense and tall canopy. Psychotria nematopoda, Hypipamee 789 145° 29.201'E Ficus spp., Calamus spp. present. Shallow gravelly soil. Mature lowland rainforest with a canopy to 30 m. Samples taken near buttresses of large Ficus spp. 17° 35.407'S Shallow gravelly soil with clay, numerous roots in the Nandroya 354 145° 45.270'E soil. Basaltic parent material.

128 Chapter Four- Soil beyond the boundaries

Site Altitude Location Description (m) Dense rainforest with many ferns in sparse understory. Ficus spp., Agathis robusta and Calamus spp. in canopy. Surprise 16° 51.567' S Sedges plentiful in woodland understorey, under Acacia (Boundary) 410 145° 39.018'E spp. and, Syncarpia glomulifera. Fine basaltic black soil. Rainforest canopy to 25-30 m, shrubby understory. Smith's Very deep leaf litter in woodland, abundant grasses. Track 16° 53.061' S Eucalyptus spp. and Allocasuarina torulosa with (Boundary) 382 145° 39.059'E strongly scorched trunks. Loose soil with gravel. Abundant grasses in woodland, under Corymbia torreliana and Eucalyptus tereticornis. Rainforest Clohesy 16° 55.630' S understory very open in comparison, but very strongly (Boundary) 407 145° 36.257'E shaded with no grass. Relatively sandy soil. Open forest with dense grassy understory under a few Herberton 17° 16.428'S eucalypts abutting dense rainforest with a relatively (Boundary) 760 145° 25.774'E open understory. Boundary between young rainforest with a canopy to Hypipamee 17° 25.691' S 20m and open forest with dense understory of sedges (Boundary) 789 145° 29.178'E and grasses. Shallow soils over granite.

129 Chapter Four- Soil beyond the boundaries

Appendix 3- Mean values (±Std Dev) for soil parameters measured in open forest (OF), rainforest (RF), open forest boundary vegetation (bnOF) and rainforest boundary vegetation (bnRF); as well as degrees of significance for comparisons (using a linear mixed effect model in R) between parameters at each of these vegetation types.

OF v OF v bnO RF RF v OF Open Boundaries Rain- bnO bnR v v bnR v Forest forest bnR bnO RF bn OF bn RF

Organic 6.18 11.66 13.73 12.55 0.00 <0.0 n. s. n. s. n. s. <0.0 matter ±4.74 ±4.9 ±6.03 ±5.91 1 01 01 pH 6.02 5.15 5.13 5.66 0.02 0.02 n. s. 0.01 0.00 n. s. (1:5 water) ±1.18 ±0.43 ±0.3 ±0.61 9 1 5 9 eC μS/cm 86.26 110.46 147.93 191.80 0.02 <0.0 0.00 0.00 0.08 <0.0 ±28.53 ±28.23 ±36.01 ±76.37 01 1 1 01 CEC 6.90 8.56 9.95 20.25 n. s. n. s. n. s. n. s. n. s. n. s. ±3.11 ±3.01 ±4.02 ±18.29 C % 3.53 6.66 7.84 7.17 0.00 <0.0 n. s. n. s. n. s. <0.0 ±2.71 ±2.8 ±3.45 ±3.38 2 01 01 N% 0.14 0.32 0.42 0.54 <0.0 <0.0 0.00 0.00 n. s. <0.0 ±0.12 ±0.14 ±0.18 ±0.28 01 01 5 7 01 C:N 29.44 20.77 18.28 13.60 0.00 <0.0 0.00 <0.0 <0.0 <0.0 ±12.38 ±3.78 ±2.18 ±1.97 5 01 1 01 01 01 Nitrate ppm 1.62 2.33 3.26 7.14 n. s. 0.00 0.01 <0.0 0.00 <0.0 ±1.07 ±1.21 ±1.53 ±5.2 1 9 01 5 01 Amm ppm 9.90 20.12 28.96 43.32 0.00 <0.0 0.00 0.00 n. s. <0.0 ±7.07 ±10.98 ±12.36 ±9.02 2 01 8 4 01 P (Bray I) 3.53 3.48 4.94 2.53 n. s. n. s. n. s. n. s. n. s. n. s. ±4.62 ±4.86 ±6.53 ±2.74 P (Bray II) 8.72 8.06 10.80 42.21 n. s. n. s. n. s. n. s. n. s. n. s. ±6.84 ±5.35 ±7.15 ±59.11 P (Colwell) 12.05 17.82 20.73 50.38 0.01 0.00 n. s. 0.00 0.04 <0.0 ±10.65 ±9.02 ±10.04 ±48.84 7 2 8 0 01 ......

130 Chapter Four- Soil beyond the boundaries

OF v OF v bnO RF RF v OF Open Boundaries Rain- bnO bnR v v bnR v Forest forest bnR bnO RF bn OF bn RF

Ca kg/ha 1566.18 1167.46 1562.47 6080.80 n. s. n. s. n. s. n. s. n. s. n. s. ±1322.8 ±922.68 ±811 ±6833 3 Mg kg/ha 370.37 467.97 692.74 1157.48 0.03 0.00 n. n. s. n. s. <0.0 ±153.41 ±145.54 ±276 ±984.25 7 1 s.. 01 Ca:Mg 2.63 1.49 1.42 2.30 n. s. n. s. n. s. n. s. n. s. n. s. ±1.92 ±0.93 ±0.63 ±1.63 K kg/ha 898.51 652.26 703.9 745.901 n. s. n. s. n. s. n. s. n. s. n. s. ±402.62 ±374.43 ±370.36 ±335.25 Al kg/ha 108.16 482.11 636.77 210.87 <0.0 <0.0 n. s. <0.0 <0.0 n. s. ±176.85 ±322.9 ±479.7 ±244.09 01 01 01 01 Mn ppm 20.19 14.83 46.92 75.03 n. s. n. s. <0.0 0.00 n. s. 0.01 ±14.47 ±12.4 ±41.06 ±53.31 01 5 1 Fe ppm 140.77 260.65 271.59 295.95 <0.0 <0.0 n. s. n. s. n. s. <0.0 ±116.59 ±96.22 ±80.25 ±144.3 01 01 01 Cu ppm 0.52 0.77 1.31 2.07 n. s. n. s. n. s. n. s. n. s. n. s. ±0.26 ±0.6 ±1.26 ±0.97 B ppm 0.40 0.68 0.95 0.95 <0.0 <0.0 0.00 n. s. n. s. <0.0 ±0.11 ±0.24 ±0.35 ±0.39 01 01 3 01 Sulphate S 12.77 29.32 38.95 41.32 <0.0 <0.0 0.00 0 0.42 <0.0 ppm ±13.43 ±10.19 ±12.36 ±9.02 01 01 1 01 Na kg/Ha 114.88 97.15 112.3 129.08 n. s. n. s. n. s. n. s. n. s. n. s. ±37.49 ±21.26 ±49.41 ±92.1 Si ppm 86.57 135.71 163.95 165.09 0.01 <0.0 n. s. n. s. n. s. <0.0 ±34.87 ±58.02 ±68.36 ±84.66 1 01 01 Zn ppm 1.39 1.26 1.69 4.46 n.s. n. s. n. s. n. s. n. s. n. s. ±1.1 ±0.73 ±1 ±5.13

131 Chapter Four- Soil beyond the boundaries

Appendix 4- Analytical methods used by the Environmental Analysis Laboratory (EAL) as part as the Albrecht/Reams soil testing package. Albrecht methods follow Rayment and

Higgins (1992) Australian Laboratory Handbook of Soil and Water Chemical Methods.

Soil parameter Analytical methods and reference

Organic matter Calculated as organic matter = Total Carbon (%C) x 1.75 (%) 6B3 From Rayment and Higgins 1992, Reams diluted Morgan extract pH 1:5 Soil: Water suspension 4A1 From Rayment and Higgins 1992, Reams diluted Morgan extract eC μS/cm 1:5 Soil: Water suspension 3A1 From Rayment and Higgins 1992, Reams diluted Morgan extract Cation Exchange Calculated as the sum of exchangeable Mg, Ca, Na, K, H and Al Capacity CEC 15G1 From Rayment and Higgins 1992 cmol+/Kg Total C (%) Measured in a LECO CNS Analyser Total N (%) C:N Ratio based on total C (%) and total N (%) Nitrate ppm Modified Morgan extract (Andersen, 2000) , Reams diluted Morgan Ammonium extract ppm P (Bray I) Ammonium fluoride and hydrochloric acid extracts, 9E2 From Rayment P (Bray II) and Higgins 1992 P (Colwell) Bicarbonate extraction, 9E1 From Rayment and Higgins 1992 Ca Kg/ha Ammonium acetate equivalent extract. Mg Kg/ha 15B1 From Rayment and Higgins 1992. Na Kg/Ha Area based metric conversions used by the EAL: 1 cmol+/Kg = 460 K Kg/ha Kg/Hectare Sodium; 780 Kg/Ha Potassium; 240 Kg/Ha Magnesium; 400 Al Kg/ha Kg/Ha Calcium. Ca:Mg Ratio based on percent base saturation totals Mn ppm Fe ppm DPTA extract. 12A1 From Rayment and Higgins 1992, Reams diluted Cu ppm Morgan extract Zn ppm B ppm Hot calcium chloride extract. 12C2 From Rayment and Higgins 1992 Sulphate S ppm Modified Morgan extract, Reams diluted Morgan extract Si ppm Calcium chloride extract

132 Chapter Four- Soil beyond the boundaries

133

Photo: L.Warman

134

Chapter Five

Traitors in the landscape? A model of interaction

between rainforests and fire-prone forests.

Laura Warman, Cameron S. Fletcher and Angela T. Moles

In preparation for Journal of Ecology

This study was conceived by LW and CSF based on ideas by LW and CSF. The model was coded by CSF. Supervision and manuscript revisions by AM

135 Chapter Five- Pioneers or traitors?

Abstract

When rainforests and fire-dependent vegetation occur as alternative stable states,

feedback loops particular to each vegetation type allow it to facilitate the

conditions it needs and to compete with other vegetation types. We propose that

in this scenario some pioneer species play an important role as ‘traitors’, using

feedbacks that correspond to the opposite vegetation to facilitate their own

vegetation type. We present a type of driven cellular-automata as a minimal

model of interaction between pioneer and obligate species for rainforest and fire-

prone vegetation, where pioneers exhibit increasing degrees of “traitorousness”

and interact with changing environmental conditions. Our study is based on the

vegetation of the Australian Wet Tropics, where rainforest and fire-dependent

sclerophyll and wet sclerophyll vegetation co-occur on the landscape. Our

results show the vegetation tracking changing environmental conditions and

becoming much more responsive to environmental change when pioneers are

traitorous. We found that minimal rules of competition, facilitation and habitat

preference were enough to make wet sclerophyll vegetation maintain itself on

the landscape, despite being an only marginally stable vegetation type. Our

results have important implications for understanding vegetation feedbacks,

vegetation functioning as alternative stable states, the vegetation dynamics of the

Australian Wet Tropics, and the way in which vegetation as a whole responds to

changing environmental parameters.

136 Chapter Five- Pioneers or traitors?

Introduction

“It is not sheer numbers that make the living world complex, but rather the

enormous variety of ways in which organisms interact. Ecosystems are the

products of interactions. They emerge out of interactions within species, between

species, between organisms and their environment and between parts of a

landscape.” Green and Sadedin 2005.

Mosaics and abrupt boundaries between closed forest and fire-prone vegetation occur around the world. There is considerable evidence that both the distribution and relative proportion of both vegetation types has changed significantly through time (Eden and

McGregor 1992, Puyravaud et al. 1994, Hopkins et al. 1996, Sanaiotti et al. 2002). This raises a series of compelling questions: What controls these profound changes in vegetation? Why do some boundaries appear to be stable while others shift so quickly?

How are both vegetation types maintained on the landscape, without one vegetation type

“taking over”? The answers to these questions have traditionally been sought by focusing on a series of abiotic parameters (including fire, water and soil) and dramatic changes in environmental conditions over time (such as ice ages). The role of vegetation interactions and the effects of the vegetation itself on local and regional conditions have received much less attention. However, as rainforest and fire-dependent vegetation are increasingly considered to behave as alternative stable states (Perry and Enright 2002,

Hoffman et al. 2009, Warman and Moles 2009), the effect of plant interactions as drivers of landscape-scale patterns becomes much more important. In this paper we address the role of plant interactions (both with other plants and the environment), in shaping landscape mosaics and determining boundaries between rainforest and fire- dependent woodland in the Australian Wet Tropics.

137 Chapter Five- Pioneers or traitors?

The interplay between interactions such as facilitation and competition occurs throughout the life histories of plants, and at scales that range from individuals to species to populations and larger communities. Even when acting at a very local level

(e.g. between individuals), these interactions can affect vegetation dynamics at ecosystem and landscape levels (Hunter and Aarssen 1988, Green and Sadedin 2005,

Brooker 2006). Measuring and quantifying these types of complex multi-scale interactions has long been beyond the reach of ecological studies, especially in systems with very long lived organisms. However, computer simulation models now allow for testing these ideas in regards to large spatial systems over long time periods (Green and

Sadedin 2005).

Existing models have examined the effect of different parameters such as the effect of climate and abiotic factors on vegetation (Green 1989), vegetation feedbacks on the local/regional environments (Sternberg 2001) and the effects of slowly changing conditions and species interactions (Buenau et al. 2007). However, in many models at least one of these ‘sets’ of parameters is assumed to be stable or ignored, which may disregard important interactions between these sets of parameters at different scales. In this paper we create a type of driven cellular automata as a minimal model of community establishment and response. The model captures environmental change and successional processes within three different vegetation communities. In turn, the model also captures how these vegetation communities impact competitiveness at species level.

Because these processes feed back on each other at several scales, we expect small changes in the underlying interactions to create significant changes in the dynamics and distribution of vegetation in the landscape. This model allows us perspective to generate novel, testable hypotheses about long term vegetation dynamics.

138 Chapter Five- Pioneers or traitors?

We base our model on the vegetation of the Australian Wet Tropics. In this region of north-eastern Australia, a fragmented band of closed-canopy rainforests grow side by side with open, fire-dependent (sclerophyll) vegetation. The region is characterized by abrupt and highly variable topography and strong rainfall gradients (Hilbert et al. 2007).

The areas with highest rainfall (>3000 mm yr-1) are covered by rainforest, while areas with less than ~600 mm yr-1 are covered by sclerophyll vegetation (Bowman 2000).

However, in areas where rainfall ranges around1300-2000 mm yr-1 rainforest and sclerophyll vegetation form a mosaic. While rainforest and open vegetation are often separated by very abrupt boundaries, they can also be separated by wet sclerophyll vegetation. Also referred to as ‘tall open forests’, wet sclerophyll vegetation contains a mixture of rainforest, sclerophyll and endemic species. In Tasmania and other temperate and subtropical areas, wet sclerophyll vegetation can cover large expanses (Ash 1988).

By contrast, in the Wet Tropics, wet sclerophyll vegetation occurs as a narrow band between rainforest and sclerophyll vegetation, and it is absent in the monsoon tropics of the Northern Territory. Warman and Moles (2009) suggested that because wet sclerophyll vegetation contains a mix of rainforest and fire-dependent species, it also contains a mixture of feedbacks that “pull” the vegetation towards both rainforest and sclerophyll vegetation. This means that wet sclerophyll forest can only be stable under certain conditions where these opposing feedbacks cancel each other out (Warman and

Moles 2009).

When vegetation types such as rainforest and fire-dependent vegetation occur as alternative stable states, they effectively compete for space on the landscape. Feedback loops specific to each vegetation type allow the vegetation to facilitate the conditions that it requires, while at the same time limiting the other vegetation type (see Wilson and

139 Chapter Five- Pioneers or traitors?

Agnew 1992, Warman and Moles 2009). In other words, water and shade based feedback loops give rainforest a competitive edge, while fire does the same for pyrophytic vegetation. This allows rainforest and open vegetation to compete for space and coexist. What happens then, if some rainforest species can thrive with fire and some fire-dependent species can survive under shady canopies? We propose that, in a system with two vegetation types as alternative stable states, some pioneer species act as

“traitors” which “hijack” the feedbacks required by the opposite vegetation type to compete against it and facilitate the establishment of their own vegetation type. For example, Alstonia muelleriana (Domin) is a rainforest species which responds to fire by resprouting vigorously and creating shade (Williams 2000). We propose this species is a good example of a traitor .In this scenario, pioneers have a dual role as instigators of both succession within vegetation types (for example, in regeneration after a disturbance) and transition between vegetation types (when they colonize an area formerly controlled by the other vegetation type).

The existence of traitors on the landscape could have important implications for our understanding of how species and vegetation types respond to changing environmental conditions. This may be especially important in the case of wet sclerophyll vegetation, as traitors may help to explain how a marginally stable or even unstable vegetation type is able to persist in the landscape. We use the model to compare scenarios where pioneer species behave as pioneers in the ‘classic’ sense, and where they increasingly behave as

‘traitors’ using the conditions required by the opposite vegetation type to expand their range and facilitate conditions for their own vegetation type. Specifically, our hypothesis is that the ‘traitorousness’ of pioneers will increase the ability of the vegetation to colonize and persist in their non-preferred communities. Thus, traitors will increase the

140 Chapter Five- Pioneers or traitors?

rate at which vegetation responds to changing abiotic parameters and change the pattern of vegetation in the landscape.

In short:

1) We create a minimal model based on species interactions and habitat preferences based on the vegetation of the Australian Wet Tropics. This model allows us to ask questions about long-term, landscape scale dynamics under changing abiotic conditions and propose testable hypothesis for the real vegetation of the Wet Tropics.

2) We test the hypothesis that wet sclerophyll vegetation is able to persist on the landscape as a marginally stable vegetation type, based on species interactions and abiotic conditions.

3) We propose the existence of traitors, and test whether increased ‘traitorousness’ of pioneers results in shorter response-time by the vegetation to changing environmental conditions.

4) We test whether traitors alter the stability of wet sclerophyll vegetation on the landscape.

Methods

The model

Our model depicts four species groups, representative of rainforest obligates, rainforest pioneers, sclerophyll obligates, and sclerophyll pioneers respectively. The combination of species groups creates three vegetation types: rainforest, wet sclerophyll, and sclerophyll across the model landscape.

141 Chapter Five- Pioneers or traitors?

Our model landscape incorporates variation in abiotic parameters which affect the mortality of the different species groups. For the purposes of this minimal model only two simple parameters are included, which we call ‘water’ and ‘soil’. Water is represented as a linear gradient that ranges from low to high (‘0’ to ‘1’) across the model landscape, representing the qualitative arrangement found in the Australian Wet Tropics from inland (left, West) to the coast (right, East). Soil is represented as a similar base gradient to water, but random ‘patchiness’ is evolved into the landscape to break symmetry and qualitatively reflect the patchiness of real soil patterns. The other important difference between ‘water’ and ‘soil’ in the model is that ‘water’ can change over the timescale of a model run, while ‘soil’ remains constant. Although referred to as

‘soil’, this parameter can be understood to reflect a series of conditions including soil fertility (which is not constant in time in the real world, but is considered constant for the time scales in the model-see chapter 4), fire refugia, topography and aspect. These factors have a patchy distribution in the Wet Tropics Bioregion and influence habitat suitability for different vegetation types there (Ash 1988, Metcalfe and Ford 2008).

Each cell on the model grid lists the proportion of the four plant species groups at that location and vegetation type is determined using a scoring system, based on the presence of each species group (Table 1). In turn, the type of vegetation in each cell affects the mortality and competitiveness of the plant species groups at that location, which incorporates feedbacks into the model. These settings reflect circumstances in the Wet

Tropics where areas with fire-dependent sclerophyll vegetation tend to support fire, giving sclerophyll plants a competitive edge (Duff 1987, Williams et al. 2006) while rainforests support light limited conditions which give rainforest plants the advantage

(Unwin 1983, Duff 1987). Neither fire nor light limitation is explicitly modelled, but

142 Chapter Five- Pioneers or traitors?

rather is implicit in mortality rates and competition within each vegetation type. For the purpose of creating the simplest minimal model, the parameters in Tables 1 and 2 are kept symmetric. In other words, the shape of habitat preference curves or mortality and fecundity rates for species groups is dependent on their status as pioneers or obligates, but independent of whether they are rainforest or sclerophyll species. The implications of this are considered in the discussion.

Table 1

Vegetation Types

Rainforest Wet Sclerophyll Sclerophyll Rainforest obligates 1.0 0.0 -1.0 Rainforest pioneers 0.6 1.0 -0.6

Species Sclerophyll pioneers -0.6 1.0 0.6 Sclerophyll obligates -1.0 0.0 1.0

Coefficients of vegetation scores used to assign a vegetation type to the mix of

species groups present at any given cell on the landscape. At each cell, the

total number of mature individuals of each species group is multiplied by the

corresponding coefficient in each column. Then the totals are summed for

each column and the cell is classified according to the vegetation type that

receives the highest score.

143 Chapter Five- Pioneers or traitors?

Table 2

Vegetation Types Rainforest Wet Sclerophyll Sclerophyll Rainforest obligates 1.0 1.0 0.0 Rainforest pioneers 0.8 + traitorousness 0.8 0.8 + traitorousness

Species Sclerophyll pioneers 0.8 – traitorousness 0.8 0.8 - traitorousness Sclerophyll obligates 0.0 1.0 1.0

Competitiveness coefficients assigned to species groups in each vegetation

type. The total number of seedlings that germinates and reaches reproductive

maturity on a cell, given the vegetation there, equals the total seed rain,

multiplied by this coefficient and scaled to the carrying capacity of the cell.

Pioneers (when traitorousness = 0) are equally competitive across all

vegetation types, but less so than the dominant obligate species in each

vegetation type (i.e. obligate sclerophyll species always outcompete

sclerophyll pioneers). Traitorousness increases the competitiveness of a

pioneer in the opposite vegetation type, at the cost of decreasing its

competitiveness within its own vegetation. Pioneers do not generally

disappear from the landscape because they can survive in all vegetation types.

It is worth emphasizing that, while pioneers are outcompeted by obligates in

wet sclerophyll vegetation, pioneers must be present in a cell for it to be

classified as wet sclerophyll (see Table 1).

144 Chapter Five- Pioneers or traitors?

The total number of individual mature plants at each cell in the landscape is limited by the cell’s given carrying capacity. Plants of all species groups contribute equally to this carrying capacity. At each time step, some adult plants die in each cell (mortality is based on habitat suitability at each location), freeing up resources for new seedlings to establish. Seedlings establish based on the relative proportion of seed rain of each species, and their competitiveness based on the vegetation present at that cell (Table 2).

Seeds are dispersed from plants within each cell and the eight immediate neighbouring cells. In this way the relative proportion of species groups (and consequently vegetation type) at a given cell changes through time based on the relative mortality of different species, their relative competitiveness as seedlings, and the dispersal from neighbouring cells.

Measuring model outcomes

Starting from an initial random distribution of plants, a steady-state distribution will evolve across a landscape when abiotic parameters are constant. However, if abiotic parameters change, species groups and vegetation communities will respond to those changes over time.

1) Changes in water-availability: We ran three different types of scenarios to test model outcomes under changing water-availability. In the first scenario (‘steady-state’) conditions remain constant, in the second (‘step-change’) we introduced a single abrupt change in water availability and in the third (‘cycle’), water availability increases and decreases cyclically (as a sinusoid). We ran 100 stochastic simulations for each scenario and each model run lasted 3000 – 10000 model years. Model years correspond roughly to real years, in the sense that in the model it is assumed that plants reproduce once a

145 Chapter Five- Pioneers or traitors?

year. To compare scenarios we used several metrics of ecosystem performance. We measured species group abundances and the proportion of cells on the model landscape occupied by each vegetation type (rainforest, sclerophyll or wet sclerophyll). We also measured the width of the wet sclerophyll vegetation corridor as a proxy for its stability.

2) Changes in species interactions: To detect the effect of increasing traitorous behaviour by pioneer species, we ran 100 simulations under the steady-state scenario and measured species abundance, proportions of vegetation types and width of wet sclerophyll vegetation. Then we measured the speed at which landscape metrics respond to changing water conditions with increasing traitorousness. Speed of response is calculated by allowing the model to run (1000 years) until it reaches a steady-state, then applying a step-change in water availability and allowing the landscape to come to a new equilibrium. Then we fit an exponential recovery curve to the change from one steady state to the next for each parameter. The ‘response time’ is the size of the exponent time parameter of the best fit (in other words, the amount of time it takes the species or vegetation to change 73.2% (1-e) of the way to its new value following the change in water availability). We repeated the runs with pioneers at different levels of traitorousness.

3) Changing interactions on a changing landscape: Lastly, we ran scenarios where changes in water availability with different cycle lengths were combined with increasing traitorousness to see how the interaction between traitorousness and changing abiotic drivers affects vegetation response. Water cycle lengths varied from short (500 years) to long (2500 years) and the response of species groups and vegetation types to increasing traitorousness was gauged by comparing their variation from the basic sinusoidal

146 Chapter Five- Pioneers or traitors?

response curve (seen in Figs 2i and 2j). That is, the more responsive the species and vegetation types become, the more they vary in response to the fixed amplitude water forcing, and the larger the range of variation becomes.

Results

Our minimal model reproduces basic features of vegetation dynamics in the Wet

Tropics. When the model runs and is allowed to reach equilibrium the vegetation sorts itself following the water gradient that is superimposed on the landscape. That is, continuous rainforest establishes in the areas of high rainfall where it outcompetes sclerophyll vegetation and continuous sclerophyll vegetation establishes in drier areas where it outcompetes rainforest (Fig 1). In between the areas of continuous rainforest and sclerophyll vegetation, the boundary zone contains patches of rainforest, sclerophyll, and wet sclerophyll vegetation. Some of these patches are determined by patchiness in ‘soil’ conditions which differ from surrounding cells.

Model response to changing environment

When the total amount of water available across the model landscape changes, this causes the established water gradient to shift uniformly upwards or downwards, but the slope of the water gradient remains constant (Fig 1). In other words, the ‘wettest’ and

‘driest’ areas remain as such even though overall water availability increases or decreases across the landscape. As the total amount of water in the landscape decreases, the landscape becomes dominated by sclerophyll vegetation (Fig 1a) and conversely by rainforest if water availability increases (Fig 1c). When the model is driven by dynamic forcing (i.e. when the total amount of water increases and decreases dynamically) the vegetation tracks these changes and ‘moves’ across the landscape. As this happens, the

147 Chapter Five- Pioneers or traitors?

relative proportions of each species group and vegetation type change (Figs 2e, 2f, 2i,

2j).

Several interesting patterns emerge when comparing the scenarios for steady-state, stepwise and cyclical changes in water availability (Fig 2). Once the vegetation achieves a steady-state (by 1000 years), the proportions of species groups and vegetation types remain constant unless water availability changes (Fig 2a and 2b). Obligate species groups become more numerous than pioneers, which is not surprising since the former outcompete the latter in both rainforest and sclerophyll vegetation. Similarly, it is not surprising (given the default water gradient settings from Fig 1e) that there are equal proportions of sclerophyll and rainforest vegetation, and that both of these vegetation types are more abundant on the landscape than is wet sclerophyll vegetation.

When changes in water availability are introduced, the total landscape-cover by rainforest and sclerophyll vegetation changes in equal proportions. In other words, if water decreases sclerophyll vegetation will gain the landscape that rainforest loses (Fig

2f, 2j). It is important to note that this pattern emerges in the model, rather than being set as a rule, and is directly related to the slope of the water gradient on the landscape. In contrast, both the amount of wet sclerophyll vegetation on the landscape and its width

(Figs 2j and 2k), remain fairly constant independent of changes in water availability.

This means that as rainforest and sclerophyll vegetation move across the model landscape, wet sclerophyll is maintained as a relatively steady boundary between the two. From any one given point in the model landscape these changes would be interpreted as a change in dominant vegetation type. In particular, at the boundary it would appear as if wet sclerophyll were being replaced by either rainforest or

148 Chapter Five- Pioneers or traitors?

Figure 1- The top row shows the distribution of sclerophyll (brown), wet sclerophyll (yellow) and rainforest (green) vegetation on the model landscape.

The bottom row shows the corresponding water gradient driving each distribution. The gradient may shift up or down according to water availability, but maintains the same slope. b) Shows the default model setting, where the gradient is centred on the landscape. In a), ‘drier’ conditions lead to more abundant sclerophyll vegetation and the boundary is shifted towards the East.

In c), when water becomes more abundant the boundary moves westwards and rainforest is more abundant. The patches of contrasting vegetation are due to soil patchiness. The soil suitability in these patches remains constant in all three scenarios, and vegetation changes within them reflect water availability.

149 Chapter Five- Pioneers or traitors?

Figure 2 Mean values of landscape metrics from 100 stochastic runs of the

model under three different water availability scenarios. Error bars denote

standard error. The first column (a-d) shows the steady-state scenario; the

middle column (e-h) shows the step-change scenario and the third column (i-l)

shows the cyclical water change scenario. Dotted lines represent the point

when water availability corresponds to the ‘default’ water setting. The top row

(a, e, i) shows percentage of species group abundance; the second row (b, f, j)

shows percentage of vegetation type abundance; and the third row (c, g, k)

shows the width of wet sclerophyll vegetation. (d, h, l) show the pattern of

water variation driving the response, as the water availability measured at the

mid-point of the landscape. 150 Chapter Five- Pioneers or traitors?

sclerophyll vegetation. However, from a whole-landscape perspective no vegetation type disappears as water availability changes, rather the location of the vegetation types shifts through time.

The correspondence between the patterns of cyclical water availability (Figs 2d, 2h, 2l) and the patterns in the vegetation response offer an intriguing prediction. The amounts of rainforest and sclerophyll vegetation (Fig 2j) oscillate at the same frequency as water availability (Fig 2l) albeit with a small time lag. In other words, rainforest abundance peaks shortly after water availability does. In contrast, the peaks in amount and width of wet sclerophyll vegetation (Figs 2j, 2k) correspond to the times when water availability passes through the default water scenario (with the gradient centred on the landscape).

This means that the abundance and width of wet sclerophyll vegetation varies at twice the frequency of water change, or rainforest abundance. This creates testable predictions for long-term data sets about the relative abundance and permanence of wet sclerophyll vegetation.

When changes in water availability are introduced into the model universe, contrasts quickly become apparent between the behaviour of species groups and vegetation types

(particularly in the cyclical scenario; Figs 2i and 2j). While changes in the proportion of vegetation types on the model landscape follow relatively simple (zero sum) patterns, the trajectories of species group abundances are much more complicated (Figs 2e, 2f, 2i,

2j). This suggests that piecing together a long-term, landscape-scale understanding of vegetation dynamics may not only be difficult, but also lead to mistaken interpretations, if based solely on individual species’ trajectories and behaviours. Given that the model is symmetrical and only takes into account the very simplest of life-history traits and

151 Chapter Five- Pioneers or traitors?

interactions, species trajectories in the real world could be expected be even more diverse and confusing.

Response to changing species interactions

Increasing the level of traitorousness exhibited by pioneer species groups resulted in more obligates and fewer pioneers (of both rainforest and sclerophyll species groups) on the model landscape (Fig 3a). Counter-intuitively, this did not strongly affect the net amount of rainforest, sclerophyll and wet sclerophyll vegetation on the landscape (Fig

3b). This is because, in addition to changing species composition, increasing traitorous behaviour changes the distribution of pioneers across the landscape (Figs 3d-f). When traitorousness is low, pioneer species are distributed broadly both in areas which are recognized as rainforest or sclerophyll vegetation by the model and at the boundary between vegetation types. As traitorousness increases, there are fewer pioneer species across the landscape and these are concentrated in the boundary region. In other words, at low traitorousness there are more pioneers embedded within rainforest and sclerophyll vegetation, and at high traitorousness the ‘range’ of pioneer species becomes restricted to the boundary zone between rainforest and sclerophyll vegetation. At the highest levels of traitorousness, this migration (which is based on how traitors compete on the landscape) causes the pioneers to eventually “switch sides” on the landscape (Fig 3f).

That is, rainforest pioneer species move closer to sclerophyll vegetation and vice versa.

While in the real world one would still expect to find pioneer species throughout a vegetation type, this result suggests where traitors would be expected to be found on real landscapes: at boundaries between vegetation types and edging further into the opposite vegetation.

152 Chapter Five- Pioneers or traitors?

Figure3 Mean values and standard errors of 100 stochastic model runs on a steady state landscape with increasing traitorousness. a) Percentage of species group abundance and b) percentage of vegetation type abundance. c) Shows the width of wet sclerophyll vegetation on the landscape, which does not change significantly as traitorousness increases. d) When traitorousness is low, pioneer species are distributed across the boundary region and with significant populations in rainforest and sclerophyll vegetation. e) As traitorousness increases, there are fewer pioneer species across the landscape but these are forced closer together into the boundary region. f) At high traitorousness, pioneers “switch” places on the landscape.

153 Chapter Five- Pioneers or traitors?

Figure4 Mean values of 100 stochastic runs of the time it takes the species

and/or vegetation to reach a new equilibrium after a step-change in water

availability. Error bars denote the standard error of the exponential fit. a)

Species group and b) vegetation type response to changing (drying) water

conditions with increasing levels of traitorousness. As traitorousness increases

past a certain threshold, response time shortens, especially in the case of wet

sclerophyll vegetation.

154 Chapter Five- Pioneers or traitors?

Our results suggest that the interactions between both species and vegetation types play a crucial role in maintaining wet sclerophyll vegetation on the landscape. Furthermore, that pioneers acting as traitors may be directly related to the coherent functioning of the boundary between rainforest and sclerophyll vegetation as a distinct community. We also found that the formation of wet sclerophyll vegetation in our model is sensitive to the value assigned to pioneer competitiveness (not illustrated). In other words, when pioneers’ competitiveness was reduced, wet sclerophyll vegetation failed to persist in a static environment. However, since its’ component species did persist, wet sclerophyll was re-formed at the boundary following a change in water availability. Its prevalence was still strongly related to traitorousness. The relationship between pioneer competitiveness and wet sclerophyll persistence remains to be explored, but was not further considered for this chapter.

Response to traitors on a dynamically changing landscape

When pioneer species act as traitors, this strongly reduces the time it takes for individual species to reach a new equilibrium following a step change in water availability (Fig 4).

At low traitorousness, rainforest and sclerophyll vegetation types respond and reach a new equilibrium much more quickly (~100yrs) than do wet sclerophyll vegetation or the species groups (~400-700 years). In other words, vegetation types can become stable even while the proportions of species that combine to create the vegetation type are still changing and stabilizing. It is worth noting that this emerging stability becomes apparent despite only two species groups (pioneers and obligates) being present for each vegetation type. As traitorousness increases, the response time of wet sclerophyll vegetation decreases markedly (from ~400 years to ~100 years), while the response of rainforest and sclerophyll vegetation only changes marginally (Fig 4b). If this occurs in

155 Chapter Five- Pioneers or traitors?

the real world, then we could expect communities driven by traitors to be much better suited to rapidly changing environments.

Figure5 Coefficient of variation as a measure of vegetation ‘responsiveness’ to

cyclical water changes of different length in time, with different degrees of

traitorousness. Increased traitorousness allows all vegetation types to respond

more to shorter forcing cycles. a and c) Rainforests and sclerophyll vegetation

respond more to longer forcing cycles. In contrast, d) responsiveness of wet

sclerophyll decreases as cycles lengthen, especially at high traitorousness.

Our model shows that when cyclical environmental changes occur over shorter periods of time (500 years), traitorousness increases the responsiveness of all vegetation types

(Fig 5). As the water cycles lengthen (to 1500 and 2500 years), the responsiveness of rainforest and sclerophyll vegetation increases (Figs 5a and 5c). In other words, if the water cycles that drive the vegetation changes occur over longer periods of time, rainforest and sclerophyll vegetation are better able to track these changes across the

156 Chapter Five- Pioneers or traitors?

landscape. As traitorousness increases, the responsiveness of both rainforest and sclerophyll vegetation increases but would appear to peak after a certain point, especially in scenarios with longer water cycles. In other words, traitors may allow the rainforest and sclerophyll vegetation to respond more to quickly changing environments.

But if changes in environmental drivers occur over longer periods of time, the effect of traitors in these two vegetation types diminishes. By contrast, the responsiveness of wet sclerophyll vegetation decreases as water cycles lengthen (Fig 5b) and the peak responsiveness at intermediate levels of traitorousness is much more marked.

In other words, in contrast to rainforest and sclerophyll vegetation, the effect of high traitorousness creates a greater difference in response to length of water cycles.

Discussion

Based on simple rules, our model shows vegetation tracking suitable conditions across the landscape. Switches in landscape dominance between rainforest and open vegetation in the model landscape are consistent with palaecological data from the real vegetation in the Australian Wet Tropics (Kershaw et al 2007, Hopkins et al. 1993, Hopkins et al.

1996, Haberle 2005, Hilbert et al. 2007). Indeed, palaeoecological data reported by

Haberle (2005) support lag in extent of rainforest coverage on the Atherton Tablelands in respect to climatic forcing cycles (in this case reduced seasonality linked to

Milankovitch cycles). Furthermore, Tng et al. 2010 have recently shown that the most likely predictor for rainforest invasion of open vegetation during the past century is proximity of the open vegetation to existing rainforest. This means that the way in which the vegetation moves across the model landscape also appears to be consistent with how rainforest expansion has occurred in the Wet Tropics.

157 Chapter Five- Pioneers or traitors?

Wet sclerophyll on the landscape

Importantly, our model indicates the constant presence of wet sclerophyll vegetation on the landscape. This is consistent with the predictions made by Warman and Moles

(2009) that wet sclerophyll forests are “a constant presence in the landscape, but move as the boundary between vegetation types advances and retreats”. When water availability is constant on the model landscape (and thus the boundary doesn’t move), the boundary between rainforest and sclerophyll vegetation remains as a band dominated by wet sclerophyll vegetation. However, individual cells “flicker” between rainforest, sclerophyll and wet sclerophyll. In other words, although the vegetation is stationary at a landscape scale, the dynamics of wet sclerophyll vegetation cause small-scale changes during periods with no changing environmental drivers.

The width of wet sclerophyll vegetation is determined by the slope of water availability across the landscape, but our results show that as long as this gradient remains constant, wet sclerophyll width is independent of general water availability. In the model this measurement is more representative of wet sclerophyll stability than actual geographic width in the Wet Tropics, and should be interpreted carefully. However, this metric does become an interesting prediction which merits further examination, especially in the light of the latitudinal differences in the gradient of water availability and wet sclerophyll vegetation width across Australia.

Traitors in the landscape

The most consequential effect of pioneer species behaving as traitors was the increased responsiveness of vegetation types to rapidly changing water availability. This result is important because in the real world, changes in water availability are highly variable.

158 Chapter Five- Pioneers or traitors?

Rather than behaving as a distinct step-change, water availability changes will incorporate both small and large year-to-year changes, as well as much slower centennial to millennial trends. Landscapes in which vegetation dynamics are facilitated by traitorous pioneers are likely to be more responsive to rapid changes (<500 years) in water availability than similar landscapes facilitated by non-traitorous pioneers.

Understanding to what extent Australian vegetation is facilitated by traitorous pioneers could have important implications in assessing the adaptability of wet sclerophyll ecosystems to projected rapidly changing future climates.

Our results show that the presence of traitors on the landscape would not change overall vegetation patterns, but would affect species composition and their distribution across the vegetation. Thus, although the overall amount of wet sclerophyll was not dependent on traitorousness, at high traitorousness the proportion of pioneer species in the wet sclerophyll was higher. This results in higher stability for wet sclerophyll vegetation.

Is there evidence for traitors in the Wet Tropics?

A good example of what we consider traitorous behaviour in the Australian Wet Tropics is exhibited by the rainforest pioneer Alstonia muelleriana Domin. (Apocynaceae).

While many rainforest species are able to survive and regenerate (and/or germinate from soil seedbanks) following fire, A. muelleriana responds by multiplying its’ stem density and forming thickets (Williams 2000, Williams et al. 2006). These thickets (which thin out following periods without fire; Williams 2000, Williams et al. 2006) then produce shade which favours shade-tolerant plants and hinders pyrophytic species like grasses.

159 Chapter Five- Pioneers or traitors?

In the case of fire-dependent vegetation, some eucalypts may be functioning as traitors.

Unlike other open forest plants, species such as Eucalyptus grandis Hill ex Maiden or

Corymbia torelliana (F.Muell.) K.D.Hill & L.A.S.Johnson (once established) are able to survive and maintain themselves within rainforest margins for very long periods of time.

Although their seedlings are unable to regenerate in the rainforest shade, seeds from these species (which have aerial seed banks and depend on fire) have been found in soil seed banks in rainforest (House 1986). Indeed, House (1986) found that the greater part of the soil seed bank on both sides of the rainforest-open forest boundary belonged to sclerophyll forest species (but see Williams 2000). Bark streamers and other debris from

Eucalyptus grandis forms deep litter, which may inhibit germination and/or seedling survival under its canopy (Barton 1993). Furthermore these eucalypts may contribute fuel that promotes and/or intensifies occasional fires. For example, the bark streamers from E. grandis have been reported to catch alight and float through the forest spreading fire (Russell-Smith and Stanton 2002). Although these fires won’t necessarily kill rainforest plants, they will force rainforest plants to resprout from the ground rather than epicormically, which gives pyrophytic plants (which do resprout epicormically) a competitive advantage in the canopy and in regards to escaping the “fire trap”

(Williams et al. 2006).

The contrast between the life histories of Alstonia and the eucalypt species highlights the fact that, unlike in our model, the processes involving pioneers (and possibly traitors) are not symmetric across rainforest and open forest in the real world. While Alstonia can actively colonize a patch of open vegetation, thrive with fire and create conditions for rainforest species relatively quickly, eucalypt species would be more likely to act as sleeper species (or sit-and-wait predators). In other words, even if eucalypts were to be

160 Chapter Five- Pioneers or traitors?

present in the soil seed bank for a given area, they would require further suitable conditions (such as fire) before driving a significant change in community composition.

Implications and future perspectives

Our model contrasts the behaviour of vegetation types to the behaviour of species groups following changes in water availability. Our results suggests that understanding the behaviour of coherent functional groups (such as rainforest or woodland) through time may be a much simpler task than trying to discern a pattern based on a particular species’ history or behaviour. This would seem to echo findings by Bowman et al. (2005 and references within) that Callitris intratropica, a fire-sensitive species, is severely declining with increased exposure to fire, even as the monsoonal rainforests it inhabits appear to be expanding. By grouping species into functional groups, our model may also be of use in re-examining palaeoecological data. Fossil pollen has provided clear records of what species have been present in the Australian Wet Tropics for the past 230,000 years (e.g. Kershaw et al. 2007). Understanding which species (or species groups) are likely to act as traitors may allow us to infer past patterns of interactions and look for similar interactions in other areas where abrupt boundaries are known to exist, such as

New Caledonia or Amazonia (e.g. Colinvaux et al 1996, Stevenson and Hope 2005).

The relationship between changing species interactions and changing environmental conditions in the model poses questions that have direct bearing on our understanding of ecosystem and landscape-scale vegetation patterns and functioning. For example, on the contrast between concepts of “ecological nomads” (in which ecosystems and/or communities are viewed as cohesive groups which track environmental conditions; e.g.

Smith and Guyer 1983) and perspectives of “landscape fluidity” (which emphasize

161 Chapter Five- Pioneers or traitors?

changing interactions, changing species and shifting landscapes as a result of changing environmental conditions; Manning et al. 2009). This type of question in turn has direct relevance to conservation and management strategies (Green and Sadedin 2005). For example, as to what makes an appropriate management regime for wet sclerophyll vegetation (sensu Tng et al. 2010) in light of changing climatic conditions and changing fire regimes.

The main strength of the model lies in that, as a graphical extension of a ‘thought experiment’, it permits us to follow through and better understand the implications of traitors being present on the landscape as well as the role of long term, slowly-changing variables. The model captures processes that occur at a variety of interacting scales, and unlike a simpler model, makes it possible to ‘tweak’ the parameters linked to these processes at any scale. This in turn allowed us to better understand which of our assumptions on the functioning of the system are valid (such as the landscape-level effect of facilitation between species), offered unexpected results (particularly regarding time lags and vegetation ‘responsiveness’) and allows us to advance hypotheses to be tested through empirical data and quantitative models. In many ways this model is a necessary first step which opens the way for ongoing work. For example, as described in this study, the model does not yet produce the classic hysteresis pattern expected with alternative stable states (see Chapter 2 and Chapter 4). However, the model does capture the interplay between abiotic parameters. For example, past a given water threshold, two areas with different ‘soil’ conditions may both be covered by rainforest, while during

‘drier’ conditions one is covered by rainforest and one by sclerophyll vegetation (see

Fig. 1). Up to this point fire has been included as an implicit rather than explicit factor in the model. This does not negate the importance of fire in a system like the Wet Tropics,

162 Chapter Five- Pioneers or traitors?

but rather reflects that most models to date have included fire and fire-related feedbacks instead of water and water-related feedbacks (see Perry and Enright 2002, 2006 or Perry and Millington 2008). It may eventually be necessary to make fire an explicit variable in order to describe hysteresis in the system and capture dynamics of vegetation persisting despite unfavourable conditions via feedbacks, such as those described by Virah-Sawmy

(2009) in Madagascar.

Conclusion

Our model uses basic rules of interactions between species and species habitat preferences to show how both species and vegetation types respond to a changing environmental driver on a heterogeneous landscape. The model reproduces some important characteristics of the vegetation of the Australian Wet Tropics and offers insights into interactions between closed forest and fire dependent vegetation. We show that wet sclerophyll vegetation can remain on the landscape while being a marginally stable vegetation type. We also show that, if traitors are present on the real landscape, they may play an important role in plant communities’ response to changing climatic conditions. A key point of the paper is that considering interactions between functional species groups in the Australian Wet Tropics may be crucial in understanding long term vegetation patterns. This understanding will be vitally important if we are to conserve and manage these systems successfully.

References

Barton, A. M. 1993. Factors Controlling Plant Distributions: Drought, Competition, and

Fire in Montane Pines in Arizona. - Ecological Monographs 63: 367-397.

163 Chapter Five- Pioneers or traitors?

Bowman, D. M. J. S. 2000. Australian Rainforests: Islands of green in a land of fire. -

Cambridge University Press.

Bowman, D., Murphy, B. and Banfai, D. 2010. Has global environmental change

caused monsoon rainforests to expand in the Australian monsoon tropics? -

Landscape Ecology 25: 1247-1260.

Brooker, R. W. 2006. Plant–plant interactions and environmental change. - New

Phytologist 171: 271-284.

Buenau, K. E., Rassweiler, A. and Nisbet, R. M. 2007. The effects of landscape

structure on space competition and alternative stable states. - Ecology 88: 3022-

3031.

Colinvaux, P. A., Oliveira, P. E. D., Moreno, J. E., Miller, M. C. and Bush, M. B. 1996.

A Long Pollen Record from Lowland Amazonia: Forest and Cooling in Glacial

Times. - Science 274: 85-88.

Duff, G. A. 1987. Physiological ecology and vegetation dynamics of North Queensland

upland rainforest-open forest ecotones. Department of Botany. - James Cook

University of North Queensland.

Eden, M. J. and McGregor, D. F. M. 1992. Dynamics of the forest-savanna boundary in

the Rio Branco-Rupununi region of northern Amazonia. - In: Furley, P. A.,

Proctor, J. and Ratter, J. A. (eds.), Nature and dynamics of forest-savanna

boundaries. Chapman Hall.

Green, D. G. 1989. Simulated effects of fire, dispersal and spatial pattern on

competition within forest mosaics. - Plant Ecology 82: 139-153.

Green, D. G. and Sadedin, S. 2005. Interactions matter--complexity in landscapes and

ecosystems. - Ecological Complexity 2: 117-130.

164 Chapter Five- Pioneers or traitors?

Haberle, S. G. 2005. A 23,000-yr pollen record from Lake Euramoo, Wet Tropics of NE

Queensland, Australia. - Quaternary Research 64: 343-356.

Hilbert, D. W., Graham, A. W. and Hopkins, M. S. 2007. Glacial and interglacial

refugia within a long-term rainforest refugium: The Wet Tropics Bioregion of NE

Queensland, Australia. - Palaeogeography, Palaeoclimatology, Palaeoecology

251: 104 - 118.

Hoffmann, W. A., Adasme, R., Haridasan, M., T. de Carvalho, M., Geiger, E. L.,

Pereira, M. A. B., Gotsch, S. G. and Franco, A. C. 2009. Tree topkill, not

mortality, governs the dynamics of savanna-forest boundaries under frequent fire

in central Brazil. - Ecology 90: 1326-1337.

Hopkins, M. S., Ash, J., Graham, A. W., Head, J. and Hewett, R. K. 1993. Charcoal

Evidence of the Spatial Extent of the Eucalyptus Woodland Expansions and

Rainforest Contractions in North Queensland During the Late Pleistocene. -

Journal of Biogeography 20: 357-372.

Hopkins, M. S., Head, J., Ash, J. E., Hewett, R. K. and Graham, A. W. 1996. Evidence

of a Holocene and Continuing Recent Expansion of Lowland Rain Forest in

Humid, Tropical North Queensland. - Journal of Biogeography 23: 737-745.

House, A. P. N. 1986. Seed exchange and storage across rainforest- sclerophyll

woodland boundaries in north Queensland. - Australian National University.

Hunter, A. F. and Aarssen, L. W. 1988. Plants Helping Plants. - BioScience 38: 34-40.

Kershaw, A. P., Bretherton, S. C. and van der Kaars, S. 2007. A complete pollen record

of the last 230 ka from Lynch's Crater, north-eastern Australia. -

Palaeogeography, Palaeoclimatology, Palaeoecology 251: 23-45.

165 Chapter Five- Pioneers or traitors?

Manning, A. D., Fischer, J., Felton, A., Newell, B., Steffen, W. and Lindenmayer, D. B.

2009. Landscape fluidity - a unifying perspective for understanding and adapting

to global change. - Journal of Biogeography 35: 193-199.

Metcalfe, D. J. and Ford, A. J. 2008. Floristics and Plant Biodiversity of the Rainforests

of the Wet Tropics. - In: Stork, N. E. and Turton, S. M. (eds.), Living in a

Dynamic Tropical Forest Landscape. Blackwell Publishing, Ltd, pp. 123-132.

Perry, G. L. W. and Enright, N. J. 2002. Spatial modelling of landscape composition

and pattern in a maquis-forest complex, Mont Do, New Caledonia. - Ecological

Modelling 152: 279-302.

Perry, G. L. W. and Enright, N. J. 2006. Spatial modelling of vegetation change in

dynamic landscapes: a review of methods and applications. - Progress in Physical

Geography 30: 47-72.

Perry, G. L. W. and Millington, J. D. A. 2008. Spatial modelling of succession-

disturbance dynamics in forest ecosystems: Concepts and examples. -

Perspectives in Plant Ecology, Evolution and Systematics 9: 191-210.

Puyravaud, J.-P., Pascal, J.-P. and Dufour, C. 1994. Ecotone Structure as an Indicator of

Changing Forest-Savanna Boundaries (Linganamakki Region, Southern India). -

Journal of Biogeography 21: 581-593.

Russell-Smith, J. and Stanton, P. 2002. Fire regimes and fire management of rainforest

communities across northern Australia. - In: Bradstock, R. A., Williams, J. E. and

Gill, A. M. (eds.), Flammable Australia: The Fire Regimes and Biodiversity of a

Continent. Cambridge University Press.

Sanaiotti, T. M., Martinelli, L. A., Victoria, R. L., Trumbore, S. E. and Camargo, P. B.

2002. Past Vegetation Changes in Amazon Savannas Determined Using Carbon

Isotopes of Soil Organic Matter. - Biotropica 34: 2-16.

166 Chapter Five- Pioneers or traitors?

Smith, J. M. B. and Guyer, I. J. 1983. Rainforest-eucalypt forest interactions and the

relevance of the biological nomad concept. - Austral Ecology 8: 55-60.

Sternberg, L. D. S. L. 2001. Savanna-Forest Hysteresis in the Tropics. - Global Ecology

and Biogeography 10: 369-378.

Stevenson, J. and Hope, G. 2005. A comparison of late Quaternary forest changes in

New Caledonia and northeastern Australia. - Quaternary Research 64: 372-383.

Tng, D. Y. P., Sanders, G., Murphy, B., Williamson, G. J., Kemp, J. and Bowman, D.

M. J. S. 2010. Rainforest expansion in Far North Queensland. A Preliminary

Analysis of the Windsor and Carbine Tablelands. - Marine and Tropical Sciences

Research Faility (MTSRF) Transition Project Final Report. Reef and Rainforest

Research Centre Limited, Cairns: 23pp.

Unwin, G. L. 1983. Dynamics of the rainforest - eucalypt forest boundary in the

Herberton Highland, North Queensland. – MSc. Thesis, James Cook University.

Virah-Sawmy, M. 2009. Ecosystem management in Madagascar during global change. -

Conservation Letters 2: 163-170.

Warman, L. and Moles, A. T. 2009. Alternative stable states in Australia’s Wet Tropics:

a theoretical framework for the field data and a field-case for the theory -

Landscape Ecology 24: 1-13.

Williams, P. R. 2000. Fire-stimulated rainforest seedling recruitment and vegetative

regeneration in a densely grassed wet sclerophyll forest of north-eastern Australia.

- Australian Journal of Botany 48: 651-658.

Williams, P., Parsons, M. and Devlin, T. 2006. Rainforest recruitment and mortality in

eucalypt forests of the wet tropics - refining the model for better management.

Life in a Fire-Prone Environment: Translating Science into Practice. - Griffith

University.

167 Chapter Five- Pioneers or traitors?

Wilson, J. B. and Agnew, A. D. Q. 1992. Positive-Feedback Switches in Plant-

Communities. - Advances in Ecological Research 23: 263-336

168

Chapter Six

General discussion and conclusions

169 Chapter Six- General discussion and conclusions

General discussion and conclusions

The overall aim of this thesis was to contribute towards a better understanding of vegetation dynamics of the Australian Wet Tropics. Rather than considering rainforest and woodland separately and concentrating on the differences between vegetation types,

I approached them both as part of a larger complex and dynamic system. Considering both vegetation types as part of a single system allowed me a different perspective. It also allowed me to ask questions that might not otherwise be relevant, such as whether pioneer species can “hijack” the feedbacks required by the other vegetation type, or comparing variance in soil parameters across vegetation boundaries to non-boundary sites.

The advantage of Alternative Stable States theory

Using ASS as a starting point to re-interpret the vegetation patterns of the Australian

Wet Tropics has several advantages. To begin with, the theory helps create a broad conceptual framework in which seemingly contradictory data from previous research becomes complimentary (for example, the stability of some boundaries and dynamism of others; Ash 1988, or the different resprouting ability of rainforest plants after fire,

Williams 2000, Marrinan et al. 2005, Williams et al. 2006). Having ASS as a theoretical backbone for understanding vegetation dynamics in the Wet Tropics also allows for deeper comparisons with other places where closed forest and pyrophytic vegetation interact and coexist (not just in Australia, but around the world). This means that rather than concentrating on the “singularities” of a local abrupt boundary, we can ask new questions and make meaningful comparisons based on global patterns of abrupt boundaries. However, the main strength of a conceptual framework that involves ASS is

170 Chapter Six- General discussion and conclusions

that it assumes abiotic parameters and feedbacks will occur and interact at a variety of spatial and temporal scales. This allows for changing biotic and abiotic conditions over time (including climatic shifts and/or the cumulative effects of feedback loops) to modify the resilience of a system (sensu Scheffer and Carpenter 2003). This means that the “magnitude of disturbance” required to shift the system from rainforest to woodland

(or vice versa) would be expected to change through time, according to general climatic conditions as well as current local parameters. So, for example, the fire-regime (both quantity and intensity of fires) required to transform a patch of rainforest into woodland or grassland (or to shift the boundary between vegetation types) would be expected to differ depending on large-scale gradually changing climatic conditions (such as glaciations), gradually changing abiotic parameters (such as soil chemistry and condition) and the species present at a given point in time (i.e. whether they are fire- promoters, fire-sensitive species or even traitors). Rather than contradict or simply re- word previous theories based on feedbacks and disturbance (e.g. Jackson 1968, Wilson and Agnew 1992), ASS theory expands upon them and provides a larger, multi-scaled context in which to fit these ideas. It also allows us to look past the individual species involved and to approach vegetation dynamics from a larger scale.

Since I began working on this thesis, viewing tropical rainforest and fire-dependent vegetation as ASS has become more widely discussed and accepted in the scientific literature (e.g. Archibald 2010 or Hoffmann et al .2009). The published version of

Chapter Two (Warman and Moles 2009) has contributed to the discussion on the origins, maintenance and conservation of fire-dependent vegetation and the mosaics formed by grassy biomes and closed forests (Virah-Sawmy 2009, Bond and Parr 2010,

Zaloumis and Bond in press) elsewhere in the world.

171 Chapter Six- General discussion and conclusions

Herbivory and ASS

There is a strong body of literature about the effects of herbivores in creating and/or maintaining ASS in a variety of environments. These range from the effects of large herbivorous fish on coral reefs (Hughes et al. 2007), sea-urchins in kelp beds and temperate reefs (Konar and Estes 2003, Baskett and Salomon 2010), grazers on rangelands (Anderies et al. 2002), elephants on savannahs (Dublin et al. 1990) and caribou/reindeer in tundra (Wal 2006). In contrast, herbivory rates in the Australian Wet

Tropics showed no difference between rainforest and fire-dependent vegetation during four months of monitoring (eight weeks of which were reported in Chapter Three;

Warman et al., in press). The greater part of the observed herbivory bore the hallmarks of being caused by invertebrate rather than mammal folivores (i.e. damage caused by leaf-miners or relatively small holes in leaves, rather than large areas missing from leaves). Folivorous mammals occur in both sclerophyll and rainforest ecosystems in Far

North Queensland; ranging from agile, swamp and rock-wallabies and larger kangaroos

(such as the eastern grey which grows to ~65kg) in open habitats to a variety of pademelons (small rainforest kangaroos, less than 8kg in weight), tree kangaroos (up to

14.5 kg in weight) and possums in rainforests (Williams 2006). However, unlike other tropical terrestrial systems, the Australian Wet Tropics lacks large native browsers or grazers such as tapirs, okapi or elephants (all at least ~200kg). It may be that large, selective mammalian herbivores of this type are required to push and maintain terrestrial systems from one state to another (barring pest outbreaks sensu Scheffer and Carpenter

2003). It may also be that herbivores don’t play as important a role in maintaining the contrasts between fire-prone woodlands and rainforests as they do in maintaining open savannas and grasslands in contrast to rainforests. Long term monitoring, in a variety of

172 Chapter Six- General discussion and conclusions

regional vegetation types may ultimately be necessary to be able to distinguish between these possibilities.

Whiptail or pretty-faced wallaby (Macropus Lumholtz’s Tree-kangaroo (Dendrolagus parryi) in woodland near Cooktown lumholtzi) in rainforest near Atherton

Future attempts to understand how herbivory interacts with both rainforest and open vegetation could test the similarity and/or complementarities of invertebrate herbivores in each vegetation type. The role of herbivores at boundaries may prove particularly enlightening. Palmer et al. (2003) found that particular plants attracted more herbivores, increasing damage to surrounding vegetation and linked this to small-scale vegetation boundaries. If different subsets of herbivores are present in rainforest and open vegetation (which seems likely), then herbivory might be higher at the boundary, where the insect faunas from both vegetation types overlap, than within either rainforest or woodland. If herbivory is higher at the boundaries, this may play an important role in determining what species are present there. Herbivory might also influence successional processes and thus play a role in determining boundary shape or location. To test this, in addition to invertebrate herbivore studies, it could prove useful to set up exclusion tests at the boundaries for vertebrate herbivores and determine whether herbivory interacts with seedling survival (sensu Allock and Hik 2004).

173 Chapter Six- General discussion and conclusions

Regarding leaf types

It was surprising to have found so little difference in either construction or herbivory between compound and simple leaves. However, it is a much greater surprise that we still know so little about the ecological advantages of either leaf type; especially given that being simple or compound is a fundamental characteristic of leaves, easy to measure and easily accessible in most vegetation types. While I found no difference in either leaf construction or herbivory, it might prove useful to test whether differences exist in expanding leaves, especially considering that up to 70% of herbivory occurs on young leaves (Coley and Aide 1991). It may also prove useful to further explore the relationship between plant growth rate and leaf type (sensu Malhado et al. 2010) and to relate this to prevalence of compound leaves in different in environments.

Looking past the boundaries

My results show that the contrasts that have been described across boundaries may be less meaningful than has been previously thought, at least in the Australian Wet

Tropics. Overwhelmingly, research considers boundaries as the limits of two systems.

By not comparing what occurs at the boundary to what occurs in non-boundary regions

(on both sides of the boundary), we may be ignoring a big part of the overall picture and misinterpreting the parts we do see. The importance of using non-boundary controls (or their equivalent) goes beyond this type of boundary study into research on edge effects and beyond. My soil results also emphasize how dynamic the interplay between vegetation types really is, and the importance of vegetation as an ecosystem engineer.

Having found some differences between “rainforest soil” and “woodland soil” but comparatively few differences in soil across boundaries, at a regional scale, may be indicative of processes and interactions which are common to all boundaries. This

174 Chapter Six- General discussion and conclusions

impression is emphasized by the differences I found between soil in “established” vegetation and in more transitional (or at least recent) vegetation at boundaries and in wet sclerophyll vegetation. The role of low pH and high aluminium at the boundaries also merits further investigation. If high concentrations of aluminium are due to species which act as aluminium accumulators, then these species may play important roles in influencing succesional processes and competition between species and vegetation types

(either by facilitation recruitment of similar species, or hindering sensitive species).

The results from herbivory (Chapter Three; Warman et al., in press) and soil (Chapter

Four; Warman et al., in review) both indicate that different scenarios may apply when comparing closed and open forests than when comparing closed forests to savanna or grassland. Differences in leaf litter (both quantity and quality) between grassland/savanna and open forests/woodlands may play an important role in creating these different scenarios. By treating all types of fire-dependent vegetation equally, we may be overlooking important differences in how two different types of boundaries function. In order to tease this apart, and make global comparisons of boundaries meaningful, studies comparing boundaries to non-boundary controls and studies comparing rainforest/open-forest and rainforest/savanna-grassland, should be carried out.

Cascading interactions

Considering a system with multiple contrasting (“duelling”) positive feedbacks, rather than just a spectrum of reactions to fire, opens the door to questions about species and community interactions and the stability of different vegetation types through time. In

Chapter Two (Warman and Moles 2009) I proposed that a mixture of abiotic

175 Chapter Six- General discussion and conclusions

parameters, the vegetations’ interactions with these parameters, species’ habitat preferences and vegetation specific feedback loops could be used to build an ASS scenario for the vegetation of the Wet Tropics. I also suggested that wet sclerophyll forests might present a mixture of feedbacks as well as just rainforest and sclerophyll species. These ideas are explored further in Chapter Five and the idea of traitors is introduced. This idea, which only makes sense in a system with more than one set of feedbacks, raises further interesting (and possibly vexing) questions and ideas. For example, in Australia plant species are classed as belonging to rainforest or pyrophytic vegetation mainly based on how sensitive they are to fire (see Williams et al. 2006,

Bowman et al. 2010). The idea of traitors raises the possibility of a case where two species have similar habitat preferences for establishment and respond similarly to fire but differ in the feedbacks they create (where one creates conditions that hinder fire and the other fosters conditions that promote fire). How would these species then be considered?

So, is it alternative stable states after all?

The argument over whether alternative stable states are only a nice idea or actually a good representation of what happens in the natural world is ongoing. The presence of more than one community or ecosystem for a given set of environmental conditions has been remarked upon at least since Humboldt (Scheffer et al. 2005) although ideas regarding ‘alternative equilibrium states’ were more formally introduced in the late

1960’s and 1970’s (e.g. Holling 1973). Serious criticisms of ASS theory (Connel and

Sousa 1983, Peterson 1984, Petraitis and Latham 1999) have concentrated on the difficulty in defining states and variables and the lack satisfactory experiments. One of the main problems in “proving” the existence of ASS in large, natural systems is that of

176 Chapter Six- General discussion and conclusions

the time-scales involved (Petraitis and Latham 1999, Odion et al. 2010), especially in systems like rainforests with very-long lived organisms. Time scales considered in this thesis span from seasonal/yearly (for herbivory or the effects of droughts and ENSO influenced wet/dry seasons; see Marrinan et al 2005), to centuries (as rainforest establishes and replaces wet sclerophyll vegetation; see Chapter 4) into tens and hundreds of thousands of years (when considering palaecological studies such as carbon dating, pollen data, and fossil findings that span from the relatively recent last glacial maxima to well over 200,000 years ago; see Hopkins et al 1993, Kershaw et al 2003).

Choosing an appropriate field experiment regime to capture ecological processes that span these different time scales, or processes that occur at a variety of time scales presents problems. For this thesis I chose instead to reconsider previous studies on current, recent, historical and palaeoecology of the Wet Tropics, counterpoint them with specific field biology and modelling exercises and place all this information in a theoretical framework to better understand the larger panorama created.

‘Proving’ or ‘disproving’ the existence of ASS at a regional scale in the Wet Tropics goes well beyond the scope of any three-year project. However, other authors (e.g.

Perry and Enright 2002) have remarked on the value of re-considering a system via

ASS, independently of trying to prove their existence. In that sense, the importance of considering the Wet Tropics as ASS is that it becomes a useful exercise which creates new questions, insights and testable predictions, independently of ASS ultimately being

‘proven’ or ‘disproven’.

A different critique of ASS theory has centred on mosaics of closed and open vegetation having an anthropogenic origin, and thus not constituting a ‘natural system’ (e.g.

177 Chapter Six- General discussion and conclusions

Bowman and Haberle 2010). For example, the occurrence of fire-dependent vegetation in New Zealand’s South Island has been linked to the arrival of Maori burning ca 2000 years ago (Bowman and Haberle 2010). In systems like the Wet Tropics of Far North

Queensland, African grasslands or Neotropical llanos and cerrados, where humans have been a known presence on the landscape for much longer, the relationship between people, fire and fire-dependent vegetation is much harder to tease apart. Human habitation of the Australian Wet Tropics has been estimated to go back at least 40,000 years (Pannell 2008) and more than 20 Aboriginal tribal groups are linked to the area.

Aboriginal oral histories for the region incorporate records of dramatic landscape changes over time. These include the volcanic eruptions that formed the crater lakes on the Atherton Tablelands (c.a. 10,000BP) as well as changes to the coastlines during the last glacial maxima (ca. 13-10,000BP; Pannell 2008). Past interpretations of fossil pollen and charcoal analyses have attributed vegetation changes from rainforest to sclerophyll vegetation, dated at 38-45,000 years BP, to Aboriginal burning practices

(Kershaw 1986, Turney et al. 2001) and an increase in ENSO activity (Kershaw et al.

2003). However fossil pollen cores from New Caledonia show Araucaria rainforests declining and following similar trends to the Australian rainforests, at a time when there were no people on New Caledonia to set fire to the landscape (Stevenson and Hope

2005). Furthermore, fire-dependent landscapes have been suggested to predate humans by millions of years (Bond and Keeley 2005, Crisp et al. 2011). Descriptions exist in the literature of lightning-strikes being able to maintain an established sclerophyll pocket within rainforest (Hopkins et al. 1993). It has also been argued that Aboriginal land management, by groups like the Kuku-Yalanji people of the Wet Tropics, sought to maintain both rainforest and open vegetation, rather than just replace rainforest (Hill et al. 2000).

178 Chapter Six- General discussion and conclusions

Wet sclerophyll vegetation

While some authors consider wet sclerophyll forest to be a rainforest formation (see

Adam 1992, Bowman 2000), others consider it to be a type of sclerophyll vegetation

(Plowman 1979, Harrington and Sanderson 1994). Likewise, some researchers consider it to be an ecotone while others consider it to be a vegetation type in its own right (Ash

1988). All of these views are backed by evidence to support them. As scientists, we have much work left to do towards a thorough understanding of wet sclerophyll vegetation. For this thesis I treat the wet sclerophyll vegetation as an ‘unstable state’; that is, a real community in time, but not in space. This reflects the evidence for the importance of the species, interactions and processes in the wet sclerophyll (such as the presence of endemic bettongs; Johnson and McIlwee 1997) as a valid community. It also reflects the tension between the rivalling feedbacks associated with this community.

Implications for conservation and management

One of the consequences of rainforest and sclerophyll vegetation being approached as independent systems is that their conservation is also considered in independent terms.

Often this translates into conserving one vegetation type at expense of the other. For example, an argument has been made for grasslands and other fire-dependent vegetation globally being seen as derived from anthropogenic activities and thus being considered of lower conservation value (e.g. Bond and Parr 2010). Conversely, in the Wet Tropics, rainforest invasion of wet sclerophyll forest is seen as driving losses of endemic wet sclerophyll biodiversity (Harrington and Sanderson 1994, Chapman and Harrington

1997). While up to 60% of the vegetation of the Australian Wet Tropics is comprised by sclerophyll vegetation (Williams et al. 1996), rainforests on fertile lands have faced

179 Chapter Six- General discussion and conclusions

higher rates of clearing for agriculture (Laurance and Goosem 2008, Metcalfe and Ford

2008) both at low altitudes and on the tablelands.

The fact that the vegetation has moved across the landscape in response to big changes in climate, such as glaciations (Hopkins et al. 1996) could be construed as a good portent regarding the level of adaptability to change and disturbance displayed by both vegetation types (if not all the species contained within them; e.g. Williams et al. 2003).

However, anthropogenic pressure on the Wet Tropics is increasing and areas that had already been cleared for agriculture (including vast cane fields in the lowlands) are also becoming increasingly urbanized (Stork et al. 2008). Naturally fragmented vegetation is faced with additional anthropogenic fragmentation, barriers such as roads and power- line clearings (Laurance and Goosem 2008) and invasive species such as goats, pigs and weeds (Stork et al. 2008). These artificial barriers may have unforeseen consequences on vegetation that depends on being able to move across the landscape over time, especially given the additional disruption to natural processes caused by invasive species.

Future directions

The striking contrast and abrupt boundaries between rainforest and woodland will doubtless continue to catch our attention and curiosity, as they have for at least a century. I suggest that considering both fire and water related feedbacks at varying spatial and temporal scales may provide new answers to current questions, such as the apparent paradox of vegetation thickening while fire regimes intensify. While some studies at the boundaries themselves (for example of herbivores) may be useful, my results show that we also need to focus on the context in which boundaries occur.

180 Chapter Six- General discussion and conclusions

Lastly, climatic events like El Niño/La Niña have been shown to have both short and long term effects on vegetation communities, and to present unique opportunities to carry out experiments regarding the functioning of systems affected by these changes

(Holmgren et al. 2001 and 2006). Cyclones in Far North Queensland present similar opportunities. For example, the massive defoliation created by recent Cyclone Yasi offered a rather unique opportunity to measure the effect of shade related feedbacks

(and lack thereof) in rainforests followed by high intensity rainfall.

Conclusions

I found that existing data for the Wet Tropics support the premise of a single complex system where rainforest and sclerophyll vegetation function as alternative stable states.

ASS theory provides a contextual framework with which to reconcile results from a wide range of regional studies (from floristics, to soil studies, to species-specific fire tolerances and resprouting ability) and opens up scope for a series of new questions, and global comparisons with similar systems. Rather than contradict current understanding of regional vegetation dynamics, ASS allows us to find consistent links between current regional knowledge and palaecological studies and to build links with similar systems elsewhere in Australia and around the world.

I tested long-standing hypotheses on the ecological advantage of compound and simple leaf types in regards to defence from herbivores. I did not find that having compound leaves helps plants avoid herbivory. In fact I found both leaf types had remarkably similar construction and received similar damage from herbivores. However, my findings of higher nitrogen content and trends for lower toughness and leaf mass per

181 Chapter Six- General discussion and conclusions

area (LMA) are consistent with the idea that compound leaves function as disposable structures, which in turn allows plants to grow taller in less time (Malhado et al. 2010).

Together with recent findings that Amazonian trees with compound leaves grow faster than do those with simple leaves, this suggests that the answer to the ecological advantage of compound leaves may lie in the competitive advantage resulting from faster growth.

My herbivory results show no clear difference between rainforest and sclerophyll vegetation. This was surprising given that herbivory often plays an important role in maintaining the coexistence between closed forest and open vegetation in other systems around the world. This absence of differences, together with some of the soil results raises interesting new questions about the differences between closed forest boundaries with grassland/savanna or woodland/open forest. Finding consistent differences between these two types of boundaries would help solidify a global context for understanding abrupt boundaries and coexistence of closed forest and fire-dependent vegetation.

I compared soil chemistry across rainforest-sclerophyll boundaries and compared it to non-boundary rainforest, sclerophyll and wet sclerophyll vegetation at a regional scale in the Wet Tropics. By doing this, I tested implicit assumptions made by previous studies of boundaries and edge effects around the world. I found that variance in soil chemistry across boundaries is not greater than the variance found in either rainforest or woodland. This emphasizes the need to use controls from non-boundary vegetation in studies of boundaries, both in Australia and globally. I also found that soil across both sides of rainforest-woodland boundaries tends to be more similar to rainforest soil

182 Chapter Six- General discussion and conclusions

overall. This probably reflects the fluidity of boundaries in the Wet Tropics and the

“vegetation footprint” from past changes in boundary location. However, as soil at the boundary did not reflect “rainforest soil” abutting against “woodland soil”, this result also implies previous studies may not have measured and compared what they thought they were measuring. Regionally, I found soil at the boundaries has lower pH and higher aluminium concentrations than either non-boundary rainforest or woodland soil, which may act as a filter determining which species can establish at the boundaries. My finding that as rainforest invades wet sclerophyll vegetation, the soil becomes more similar to rainforest soil highlights how quickly the vegetation can affect changes in its environment. Furthermore, soil from boundaries appears to be more similar to soil in wet sclerophyll vegetation, which may be indicative of differences between soils from

‘established’ and ‘transitional’ vegetation. Put together, my soil results present a regional picture of what happens across vegetation boundaries in the Australian Wet

Tropics. More importantly they present a way to gage how meaningful contrasts are across vegetation boundaries regionally, and thus a way to make relevant comparisons across boundaries at a global scale.

Lastly, I presented a model which represents landscape-scale vegetation dynamics based on species’ habitat preferences, interactions between species and vegetation types as well as changing environmental parameters. I use this model to propose the existence of ‘traitors’ in vegetation systems with alternative stable states. Traitors are species which thrive under the feedbacks of one vegetation type, yet create a different set of feedbacks which encourage the competing vegetation type. In this way they not only allow for successional processes within vegetation type, but also for transitions between alternative stable states. I found that the presence of traitors allows vegetation to “react”

183 Chapter Six- General discussion and conclusions

much faster to environmental changes. The model also addresses the stability of wet sclerophyll vegetation. I found that with or without traitors, wet sclerophyll vegetation remains stable in time, yet not in space as suggested in the second chapter.

Considered cohesively, my thesis offers a new look at vegetation dynamics of the

Australian Wet Tropics. I approached the vegetation at a variety of scales, from leaves, to interactions between taxa, to interactions between vegetation types, to interactions between the vegetation as a whole with the environment. My work shows that, overall, an intriguing picture emerges when we consider the vegetation of the region as one system instead of two. This paradigm creates the scope for asking new questions, taking a second look at previous answers and for greater and more comprehensive comparisons with the vegetation elsewhere in Australia and around the world.

References

Adam, P. 1992. Australian Rainforests. - Oxford University Press.

Allcock, K. G. and Hik, D. S. 2004. Survival, growth, and escape from herbivory are

determined by habitat and herbivore species for three Australian woodland

plants. - Oecologia 138: 231-241.

Anderies, J. M., Janssen, M. A. and Walker, B. H. 2002. Grazing Management,

Resilience, and the Dynamics of a Fire-driven Rangeland System. - Ecosystems

5: 23-44.

Archibald, S. 2010. Fire regimes in southern Africa- Determinants, drivers and

feedbacks. Animal, plant and environmental sciences. PhD Thesis- University of

the Witwatersand.

184 Chapter Six- General discussion and conclusions

Ash, J. 1988. The Location and Stability of Rainforest Boundaries in North-Eastern

Queensland, Australia. - Journal of Biogeography 15: 619-630.

Baskett, M. L. and Salomon, A. K. 2010. Recruitment facilitation can drive alternative

states on temperate reefs. - Ecology 91: 1763-1773.

Bond, W. J. and Parr, C. L. 2010. Beyond the forest edge: Ecology, diversity and

conservation of the grassy biomes. - Biological Conservation 143: 2395-2404.

Bowman, D. M. J. S. 2000. Australian Rainforests: Islands of green in a land of fire. -

Cambridge University Press.

Bowman, D. M. J. S. and Haberle, S. G. 2010. Paradise burnt: How colonizing humans

transform landscapes with fire. - Proceedings of the National Academy of

Sciences 107: 21234-21235.

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.

Coley, P. D. and Aide, T. M. 1991. Comparison of herbivory and plant defenses in

temperate and tropical broad-leaved forests. - In: Price, P. W., Lewinsohn, T.

M., Fernandes, G. W. and Benson, W. W. (eds.), Plant-Animal Interactions:

Evolutionary Ecology in Tropical and Temperate Regions. Wiley & Sons, NY,

pp. 25-49.

Connell, J. H. and Sousa, W. P. 1983. On the Evidence Needed to Judge Ecological

Stability or Persistence. - The American Naturalist 121: 789-824.

Crisp, M. D., Burrows, G. E., Cook, L. G., Thornhill, A. H. and Bowman, D. M. J. S.

2011. Flammable biomes dominated by eucalypts originated at the Cretaceous-

Palaeogene boundary. - Nat Commun 2: 193. doi:10.1038/ncomms1191

185 Chapter Six- General discussion and conclusions

Dublin, H. T., Sinclair, A. R. E. and McGlade, J. 1990. Elephants and Fire as Causes of

Multiple Stable States in the Serengeti-Mara Woodlands. - Journal of Animal

Ecology 59: 1147-1164.

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.

Hill, R., Griggs, P. and B. B. N. I. 2000. Rainforests, Agriculture and Aboriginal Fire-

Regimes in Wet Tropical Queensland, Australia. - Australian Geographical

Studies 38: 138-157.

Hoffmann, W. A., Adasme, R., Haridasan, M., T. de Carvalho, M., Geiger, E. L.,

Pereira, M. A. B., Gotsch, S. G. and Franco, A. C. 2009. Tree topkill, not

mortality, governs the dynamics of savanna-forest boundaries under frequent fire

in central Brazil. - Ecology 90: 1326-1337.

Holling, C. S. 1973. Resilience and Stability of Ecological Systems. - Annual Review of

Ecology and Systematics 4: 1-23.

Holmgren, M., Scheffer, M., Ezcurra, E., Gutiérrez, J. R. and Mohren, G. M. J. 2001. El

Niño effects on the dynamics of terrestrial ecosystems. - Trends in Ecology &

Evolution 16: 89-94.

Holmgren, M., Stapp, P., Dickman, C. R., Gracia, C., Graham, S., Gutierrez, J. R., Hice,

C., Jaksic, F., Kelt, D. A., Letnic, M., Lima, M., Lopez, B. C., Meserve, P. L.,

Milstead, W. B., Polis, G. A., Previtali, M. A., Richter, M., Sabat, S. and Squeo,

F. A. 2006. Extreme climatic events shape arid and semiarid ecosystems. -

Frontiers in Ecology and the Environment 4: 87-95.

Hopkins, M. S., Ash, J., Graham, A. W., Head, J. and Hewett, R. K. 1993. Charcoal

Evidence of the Spatial Extent of the Eucalyptus Woodland Expansions and

186 Chapter Six- General discussion and conclusions

Rainforest Contractions in North Queensland During the Late Pleistocene. -

Journal of Biogeography 20: 357-372.

Hughes, T. P., Rodrigues, M. J., Bellwood, D. R., Ceccarelli, D., Hoegh-Guldberg, O.,

McCook, L., Moltschaniwskyj, N., Pratchett, M. S., Steneck, R. S. and Willis, B.

2007. Phase Shifts, Herbivory, and the Resilience of Coral Reefs to Climate

Change. - Current Biology 17: 360-365.

Jackson W.D. 1968. Fire, air, water and earth- an elemental ecology of Tasmania. -

Proceedings of the Ecological Society of Australia 3: 9-16.

Johnson, C. N. and McIlwee, A. P. 1997. Ecology of the Northern Bettong, Bettongia

tropica, a tropical mycophagist. - Wildlife Research 24: 549-559.

Kershaw, A. P. 1986. Climatic change and Aboriginal burning in north-east Australia

during the last two glacial/interglacial cycles. - Nature 322: 47-49.

Kershaw, A. P., van der Kaars, S. and Moss, P. T. 2003. Late Quaternary Milankovitch-

scale climatic change and variability and its impact on monsoonal Australasia. -

Marine Geology 201: 81-95.

Konar, B. and Estes, J. A. 2003. The stability of boundary regions between kelp beds

and deforested areas. - Ecology 84: 174-185.

Laurance, W. F. and Goosem, M. 2008. Impacts of Habitat Fragmentation and Linear

Clearings on Australian Rainforest Biota. - In: Stork, N. E. and Turton, S. M.

(eds.), Living in a Dynamic Tropical Forest Landscape. Blackwell Publishing,

Ltd, pp. 295-306.

Malhado, A. C. M., Whittaker, R. J., Malhi, Y., Ladle, R. J., Steege, H. t., Phillips, O.,

Aragão, L. E. O. C., Baker, T. R., Arroyo, L., Almeida, S., Higuchi, N., Killeen,

T. J., Monteagudo, A., Pitman, N. C. A., Prieto, A., Salomão, R. P., Vásquez-

Martínez, R., Laurance, W. F. and Ramírez-Angulo, H. 2010. Are compound

187 Chapter Six- General discussion and conclusions

leaves an adaptation to seasonal drought or to rapid growth? Evidence from the

Amazon rain forest. - Global Ecology and Biogeography 19: 852-862.

Marrinan M.J., Edwards W. and Landsberg J. 2005. Resprouting of saplings following a

tropical rainforest fire in north-east Queensland, Australia. Austral Ecology 30:

817-826.

Metcalfe, D. J. and Ford, A. J. 2008. Floristics and Plant Biodiversity of the Rainforests

of the Wet Tropics. - In: Stork, N. E. and Turton, S. M. (eds.), Living in a

Dynamic Tropical Forest Landscape. Blackwell Publishing, Ltd, pp. 123-132.

Odion, D. C., Moritz, M. A. and DellaSala, D. A. 2009. Alternative community states

maintained by fire in the Klamath Mountains, USA. - Journal of Ecology 98: 96-

105.

Palmer, S. C. F., Alison, J. H., Elston, D. A., Gordon, I. J. and Hartley, S. E. 2003. The

Perils of Having Tasty Neighbors: Grazing Impacts of Large Herbivores at

Vegetation Boundaries. - Ecology 84: 2877-2890.

Pannell, S. 2008. Cultural Landscapes in the Wet Tropics. - In: Stork, N. E. and Turton,

S. M. (eds.), Living in a Dynamic Tropical Forest Landscape. Blackwell

Publishing, Ltd, pp. 373-386.

Perry, G. L. W. and Enright, N. J. 2002. Humans, fire and landscape pattern:

understanding a maquis-forest complex, Mont Do, New Caledonia, using a

spatial 'state-and-transition' model. - Journal of Biogeography 29: 1143-1158.

Peterson, C. H. 1984. Does a Rigorous Criterion for Environmental Identity Preclude

the Existence of Multiple Stable Points? - The American Naturalist 124: 127-

133.

Petraitis, P. S. and Latham, R. E. 1999. The Importance of Scale in Testing the Origins

of Alternative Community States. - Ecology 80: 429-442.

188 Chapter Six- General discussion and conclusions

Plowman, K. P. 1979. Litter and soil fauna of two Australian subtropical forests. -

Austral Ecology 4: 87-104.

Scheffer, M. and Carpenter, S. R. 2003. Catastrophic regime shifts in ecosystems:

linking theory to observation. - Trends in Ecology & Evolution 18: 648-656.

Scheffer, M., Holmgren, M., Brovkin, V. and Claussen, M. 2005. Synergy between

small- and large-scale feedbacks of vegetation on the water cycle. - Global

Change Biology 11: 1003-1012.

Stork, N. E., Goosem, S. and Turton, S. M. 2008. Australian Rainforests in a global

context. - In: Stork, N. E. and Turton, S. M. (eds.), Living in a Dynamic Tropical

Forest Landscape. Blackwell Publishing, Ltd, pp. 5-20.

Stevenson, J. and Hope, G. 2005. A comparison of late Quaternary forest changes in

New Caledonia and northeastern Australia. - Quaternary Research 64: 372-383.

Turney, C. S. M., Kershaw, A. P., Moss, P., Bird, M. I., Fifield, L. K., Cresswell, R. G.,

Santos, G. M., Di Tada, M. L., Hausladen, P. A. and Zhou, Y. 2001. Redating

the onset of burning at Lynch's Crater (North Queensland): implications for

human settlement in Australia. - Journal of Quaternary Science 16: 767-771.

Virah-Sawmy, M. 2009. Ecosystem management in Madagascar during global change. -

Conservation Letters 2: 163-170.

Zaloumis, N. P. and Bond, W. J. In press. Grassland restoration after afforestation: No

direction home? - Austral Ecology, no. doi: 10.1111/j.1442-9993.2010.02158.x

Wal, R. v. d. 2006. Do herbivores cause habitat degradation or vegetation state

transition? Evidence from the tundra.- Oikos, 114: 177–186.

Warman L. and Moles A.T. 2009. Alternative stable states in Australia’s Wet Tropics: a

theoretical framework for the field data and a field-case for the theory

Landscape Ecology 24: 1-13.

189 Chapter Six- General discussion and conclusions

Warman L., Moles A.T. and Edwards W. In press. Not so simple after all: searching for

ecological advantages of compound leaves. Oikos, no. doi: 10.1111/j.1600-

0706.2010.19344.x

Warman, L., Bradford, M. G. and Moles, A. T. In review. A broad approach to abrupt

boundaries: Looking beyond the boundary at soil attributes within and across

tropical vegetation.

Williams P. 2000. Fire-stimulated rainforest seedling recruitment and vegetative

regeneration in a densely grassed wet sclerophyll forest of north-eastern

Australia. Australian Journal of Botany 48: 651-658.

Williams P., Parsons M. and Devlin T. 2006. Rainforest recruitment and mortality in

eucalypt forests of the wet tropics - refining the model for better management.

Life in a Fire-Prone Environment: Translating Science into Practice. Griffith

University, Brisbane, Australia.

Williams, S. E. 2006. Vertebrates of the Wet Tropics rainforests of Australia: species

distributions and biodiversity. - Cooperative Research Centre for Tropical

Rainforest Ecology and Management, p. 282 pages.

Williams, S. E., Pearson, R. G. and Walsh, P. J. 1996. Distributions and biodiversity of

the terrestrial vertebrates of Australia's Wet Tropics: a review of current

knowledge. - Pacific Conservation Biology 2: 327-362.

Williams, S. E., Bolitho, E. E. and Fox, S. 2003. Climate change in Australian tropical

rainforests: an impending environmental catastrophe. - Proceedings of the Royal

Society B: Biological Sciences 270: 1887-1892.

Wilson J.B. and Agnew A.D.Q. 1992. Positive-Feedback Switches in Plant-

Communities. Advances in Ecological Research 23: 263-336.

190

“There was no smoke, no sign of water, no sign of the neighbourhood of the sea coast;--

but all was one immense sea of forest and scrub.”

From Ludwig Leichhardt’s “Journal of an Overland Expedition in Australia : from

Moreton Bay to Port Essington, a distance of upwards of 3000 miles, during the years

1844-1845”

191