Fauna Recolonisation of Mine Rehabilitation Through the Example of Arboreal Marsupials, with a Particular Focus on the Koala Phascolarctos Cinereus

Home , Fauna, Minjerribah (fly)

Fauna recolonisation of mine rehabilitation through the example of arboreal marsupials, with a particular focus on the koala Phascolarctos cinereus

1999

2003

2007

2011

Romane H.A. Cristescu PhD thesis School of Biological, Earth and Environmental Sciences University of New South Wales APRIL 2011 “Conservation biology in the short-term and restoration ecology in the long-term are the complementary activities that form the basis of our belated (but not hopeless) attempt to salvage the biodiversity disaster”

Young, T. P. 2000. Restoration ecology and conservation biology. Biological Conservation 92:73-83

Cover pictures: 1998 Ibis dune rehabilitation (pictures courtesy of Sibelco/CRL)

THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: CRISTESCU First name: Romane Other name/s: Hélène Adriana Abbreviation for degree as given in the University calendar: PhD School: School of Biological, Earth and Environmental Sciences Faculty: Science Title: Fauna recolonisation after mine rehabilitation by arboreal marsupials with a particular focus on the koala Phascolarctos cinereus

Abstract

I first review the Australian literature on post-mining rehabilitation and highlight some common processes of fauna recolonisation after mining rehabilitation. I then report on an experimental assessment of the reliability of faecal pellets surveys for koalas, a methodology commonly used in koala studies. Next I test the assumption that successful rehabilitation based on flora criteria would reflect success in koala recolonisation. I found that flora will rarely be a perfect surrogate for fauna, and that this relationship cannot always be assumed but should be demonstrated. I then further investigate koala recolonisation of mining rehabilitation by comparing habitat characteristics, roosting trees, diet, and predation risk for koalas in undisturbed and rehabilitated areas. This chapter explores and rejects the possibility that rehabilitated areas could lure koalas into an ecological trap. Finally, I test the popular conservation paradigm that a single flagship fauna species like the koala can be used as an indicator for rehabilitation success of other fauna (e.g. arboreal marsupials). This hypothesis was rejected; gliders’ recolonisation of rehabilitated areas was variable whereas koala presence was more widespread. I conclude with an examination of how successful rehabilitation is for arboreal marsupials in my study site, and make general suggestions for developing fauna criteria for assessing the success of rehabilitation.

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 hereafter 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:

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 ……………………………………

iii

‘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 part in the University libraries in all forms of media, now or hereafter 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 abstract of my thesis in Dissertations 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 ……………………………………

Statement of contributions of co- authors

This thesis compiles five stand-alone papers (chapters two to six) that are ready to be submitted or currently in review. Therefore each chapter is self-contained and some repetition occurs. To avoid unnecessary repetitions of references, a single reference list for all chapters is provided at the end of the thesis. References are formatted following guidelines for Conservation Biology. Because each chapter is a stand-alone journal article, the tables and figures are not sequentially numbered throughout the thesis; rather each chapter is presented as it would appear as a published journal article. Supplementary data mentioned in one chapter can be found at the end of that chapter and are indicated by an “S” (e.g. Figure S1). Appendices are located at the end of the thesis.

This thesis is a compilation of my own work, with guidance from my supervisors Peter Banks, Céline Frère and Frank Carrick. Apart from the contributions of co-authors outlined below, I conceptualised the research, conducted all field work and data analysis, generated all maps and photographs (except Sibelco/CRL credits) and wrote all manuscripts included in this thesis. No other authors will be submitting this work as part of their thesis submissions. Co-authors proof-read and edited the final manuscript versions as is required for publications. The contributions of each co-author are detailed below.

Chapter 2 A review of mine rehabilitation and fauna: Current status and future directions Authors: Romane Cristescu, Céline Frère and Peter Banks RC can appropriately claim credit to more than 90% of the work since she gathered the literature, performed the data analyses that are reported in the paper and wrote the manuscript. Figure 1 was created with assistance of Russell Miller. CF and PB critically reviewed the manuscript.

Chapter 3 Persistence and detectability of faecal pellets in different environments: Implications for pellet-based census of fauna Authors: Romane Cristescu, Klaartje Goethals, Peter Banks, Céline Frère and Frank Carrick RC can appropriately claim credit to more than 90% of the work since she conceived the project that formed the basis for this manuscript, undertook all of the field work, the data analysis and wrote the manuscript. Statistical advice on the survival models was provided by KG. CF, FC and PB critically reviewed the manuscript.

Chapter 4 Is restoring flora restoring fauna? Developing fauna criteria for assessing restoration success Authors: Romane Cristescu, Jonathan Rhodes, Céline Frère and Peter Banks RC can appropriately claim credit to more than 80% of the work since she conceived the project that formed the basis for this manuscript, conducted the data analysis and wrote the manuscript. RC performed all of the field work regarding plot search for scats. Vegetation survey of 12 plots were done by RC and David Bowen; 42 were available from Sibelco/CRL database. RC reviewed, corrected and compiled Sibelco/CRL raw data to perform the analyses. Statistical advice on the paper was provided by JR and CF. CF, JR and PB critically reviewed the manuscript.

Chapter 5 Habitat quality and the ecology of an arboreal mammal, the koala Phascolarctos cinereus, in a rehabilitated landscape Authors: Romane Cristescu, Céline Frère, Frank Carrick and Peter Banks RC can appropriately claim credit to more than 70% of the work since she performed all of the field work for radio-tracking, plot search for scats, 2009 predator survey, conducted all data analyses and wrote the manuscript; 95% of the laboratory work was done by DB. Vegetation surveys of 24 plots were done by RC and DB, 12 were available from Sibelco/CRL database. Predator surveys in 2003 and 2005 were performed by Sibelco/CRL personal. RC reviewed, corrected and compiled

vii

Sibelco/CRL raw data to perform the analyses. CF, FC and PB critically reviewed the manuscript.

Chapter 6 Assessing rehabilitation success for arboreal marsupials, a reflection on the risk of surrogate species Authors: Romane Cristescu, Céline Frère and Peter Banks RC can appropriately claim credit to more than 90% of the work since she conceived the project that formed the basis for this manuscript, performed all of the field work, the data analysis and wrote the manuscript. CF and PB critically reviewed the manuscript.

Appendix A North Stradbroke Island: An island ark for Queensland’s koala population? Authors: Romane Cristescu, William Ellis, Deidré de Villiers, Kristen Lee, Olivia Woosnam-Merchez, Céline Frère, Peter Banks, David Dique, Simon Hodgkison, Helen Carrick, Daniel Carter, Paul Smith and Frank Carrick This is a collaborative work summarising 10 years of research on NSI koalas. RC compiled the data, carried out statistical analyses, and wrote the manuscript. All co- authors critically reviewed the manuscript.

viii

Acknowledgements

The author of this thesis was supported by an Endeavour Europe Award and an Endeavour International Postgraduate Research Scholarship, and would like to thank the Australian Government for sponsoring international students.

Sibelco Australia – Mineral sand (formerly Consolidated Rutile Limited, CRL) provided ongoing support to this research through the provision of logistical support, access to company sites and relevant maps/databases.

I would like to acknowledge and thank the traditional owners and custodians of Minjerribah, the Goenpul and the Noonuccal, and pay my respects to their elders both past and present.

The list is long of all the great persons I wish to thank (you can just skip to your name, I wrote them in bold!). The list needs to be long however, as any PhD project - and even more so a PhD based on field work in Ecology - is carried by many along the road...

In Academia...

Firstly, an unconditional thanks to Assoc. Prof. Peter Banks for rescuing me and the PhD when unforeseen events deprived me of my first supervisor. Thank you so much for having adopted us and fitted us in your busy schedule. Thanks for helping me when I couldn’t see the forest for the trees, for always pointing me in the right direction, and doing so in a patient and friendly manner. This PhD would never have been completed without your vision, your rigor and your efficiency.

I have been incredibly lucky and very overwhelmed to benefit from Dr. Céline Frère’s

ix rigorous advice in statistics and scientific writing. There is a fine line between being a friend and being a supervisor, and you walked it brilliantly. Thanks for caring enough to tell me things I didn’t want to hear when I needed to hear them.

An enormous thank-you to Prof. Frank Carrick for spoiling me with the project of my dream. Thanks for trusting me to carry out this work. Thanks for your expert opinion on all koala matters, for the always riveting koala and ecological theories and for general help and guidance in koala field work and ecology.

Thanks to Prof. Des Cooper who made my move to Australia possible and trusted me enough to let me follow my dream. Thanks to Dr. Catherine Herbert for going through the numerous projects I developed, always here to give/expand ideas with an unlimited patience for correcting my spelling and idioms.

Thanks to the members of the postgraduate committee, in particular Dr. Steve Bonser, Assoc. Prof. William Sherwin and Dr. Alistair Poore, for the constructive comments and general guidance.

Thanks to the amazing Library Staff at UNSW for being so quick at providing books and rare journals to me as a remote student. Thanks to Dr. Tim Salmon for great IT support.

Thanks to Prof. David Mulligan for welcoming me in his research team at the Centre for Mined Land Rehabilitation, University of Queensland. Thanks for the constant support, kindness, patience and advice during this PhD, and for facilitating exchanges with CMLR researchers. Thanks particularly to Mandy Gravina, Vanessa Glenn, Narelle MacCallum, Tracey Gregg and Laurelle Elliott for their help.

Thanks to Dr. Owen Nichols for great discussions and comments on mining rehabilitation. Thanks to other academics for sharing their knowledge and making research a cooperative effort, especially thanks to Dr. Jonathan Rhodes, Klaartje Goethals, Dr. Adrian Bradley, Dr. Jennifer Seddon, Dr. Harriet Preece and Prof. Barry Fox.

Thanks to Dr. Jan Aldenhoven for sharing knowledge on North Stradbroke Island and for providing relevant background information, and for many interesting discussions on koalas and on mining.

An enormous thank-you to Stan Miller and Michael Mills for their expert proof- reading of the manuscript. The extent to which this has improved the thesis cannot be understated!

In the study site...

Thanks to all Sibelco/CRL personnel, and particularly the Kounpee Tech Services. I would never have imagined nor hoped that I could possibly get so much support from a mining company, and you blatantly shattered my prejudices. Thanks for everyone that kept reporting koala sightings or just took interest in the results. Thanks more particularly to the following…

Thanks to Toni Burgess, my first contact at CRL, who made everything happen so easily. Big thanks to Chris Moyle who found me a little place in paradise to begin with, and subsequently made organisation go smoothly. Thanks for your kindness and nice chats on the porch looking at beautiful Moreton Bay.

Thanks to the surveyor team, problem-solving multi awards winners! Special thanks to David Cruickshank for getting me up and running with GIS maps. Thanks also for letting me borrow everything the surveyor team ever possessed. Thanks to Russell Miller a million times. I don’t really know where to begin, so higgledy-piggledy, thanks for: extracting all my elevation data; going around the bush in the middle of the night playing koala calls; giving me accurate positions on earth ± 3 cm; radio-tracking on weekends when I was too tired to go on my own; playing hide and seek koala poos with me; helping me dig fire-breaks to save experimental sites way into the night; looking at some more poo and recording how quickly it disintegrates; bearing all my discussions about... poo; collecting various pieces of material, brackets, koala traps and so on all around Brisbane and Straddie; trusting me with truck-osaurus; driving up and down Straddie in search of my missing green-full-of-data clipboard; walking up and down hills carrying cameras, tough book, brackets, drillers and more; spending valentine’s xi day collecting koala poo in Australia zoo; and all the thousands other exciting things that constitute a normal field ecologist routine…

Thanks to Roger Pysden for continuous IT support. I can’t count the numbers of crashes my computer had (although a brief approximation would be 10 to 20 times every day for 6 months...), nor the number of times you came around to try to fix it.

Thanks to Deb Olive for the day-to-day help, for the chocolate support and for brightening days up. Thanks to Mike Shilling for nice chats and precious pieces of wisdom.

Thanks to Ben Barker for hours showing me around mine sites, conversations about trees and rehabilitation methods, and for always letting me steal the rehab material (whatever couldn’t be stolen from the surveyors!), and with a big smile and good tips on the top of it!

A special thanks to Steve Rodenhausenburgenstein for the great company and cheeky comments, for stirring me up when I was down, and for the help in the field, in particular with the predator plots and the glider experiment – even if you subsequently told everyone I tried to kill you running up hill with brackets, buckets and drills.

A very special thanks to (Dr.) Craig Lockhart. I kept on being amazed by all the means you found to help me. You are just the smartest at findings ways to tackle problems and quickly act to solve them. Thanks for your invaluable help for vegetation and rehabilitation matters, thanks for always jumping in the car to show me around, building a fire-break, buying me internet access, or 200 batteries - on Christmas day -, or SD cards or whatever I needed to move forward. Thanks for always advising me, with me and my PhD interests first in your mind. Thanks for your lovely company and I look forward to more camping trips with you and Karen.

I must finish the Sibelco/CRL round of thank-yous by a huge thanks to Paul Smith. Your constant support of the whole project is what made it happen and kept it going. Thanks for always saving the day (Need a new receiver, I’ll buy it! Want to meet this

xii guy, I’ll call him!). Thanks for supporting every crazy idea I had, sharing your knowledge, and always taking the time you didn’t have to help me. Thanks for being so excited and passionate about wildlife and innovative ways to protect it. I wish the mining industry was packed with people like you.

In the field...

Thanks to Deidré de Villiers for being so immensely committed to koalas and all koala matters. Thanks in particular for never making me feel guilty, but on the contrary making me feel like it was almost normal to come and help me at any time, and more particularly on weekends, when I needed some expert koala skills. Thanks for always being available to give me advice and help in any way. Thanks for bringing along your wonderful team (thanks Mimi!) and cooking skills! I have the intimate conviction that with more people like you around, the problems of the planet could be easily solved.

Thanks to Dr. Bill Ellis and Dr. Sean Fitzgibbon for their help in the field with koala catching, thanks for their advice throughout the project, thanks for their comments on parts of the thesis, thanks for making conferences so entertaining, thanks for their interesting sense of humour and obviously for introducing me to the Squirrel game.

Thanks to Jenny Davis, Lisa Bailey and Dr. Simon Hodgkison for making hours of hard work feel like fun. Thanks to Jenny and Lisa for their help with koala catching and for data sharing. Thanks to Dr. Adrian Bradley and Sarah Bell for sharing glider knowledge and letting me borrow their nice-looking brackets.

Finally, an enormous thank-you to Dave Bowen for his incredible dedication to the diet analysis. This was a hard piece of work, and you worked so hard on it you were just amazing. Thanks also for the hard work in vegetation characterisation and improving my plant knowledge more than a million times (beginning from close to zero however). Many hours in the bush and a million or so lollies were necessary but we made it.

And in life...

Thanks to my lovely Australian friends for becoming my surrogate family. I have been

xiii very blessed since I arrived in this country, and this experience has been an unbelievable succession of wonderful friendships, including many that I thanked above and some that I acknowledge below. This is by no mean a complete list, as there are too many people I wish to thank and I do feel guilty about the trees...

Thanks first and foremost to Liz and Michael Oldfield. I am not sure I deserved all your love and help, but you have been so unbelievably supportive of me from the beginning. Words can certainly not express how lucky and grateful I am to have met you, but I hope you know it. Thanks for your good advice, your patience, and your support of any kind you could imagine. Thanks for making it possible for me to come to Australia to begin with, and making it so easy for me to stay. Thanks to all your lovely family too, a particular thought to Sue and Steve Walsh, and Sally and Ann Pittman.

An immense thanks to Olivia and Alex Dudkowski. Thanks for your always excellent advice on how to deal with PhD drawbacks and for always being excited about my results and my progress (even so small...). Special thanks, Olivia, for your help in improving manuscripts and for the thousand discussions around koalas and much more. Thanks both of you for always making me feel like I’m family, and thanks for falling in love with and moving to Brisbane. As far as I’m concerned, best choice you ever made (but I’m a bit biased maybe...).

Thanks to Céline Frère (again) for her incredible help as a friend too. Thanks for supporting me, welcoming me, and checking up on me every so often to see if I was still kicking... Thanks for reassuring me I would get there eventually, and making sure I did.

Thanks to my lovely-brainy-funky Heather Shilling. Thanks for “cowboy-up cupcake”, thanks for being a breath of fresh air, thanks for the indiscriminate kindness you pour onto this world. Even if you fight it with all your heart, you are a very special and important part of my (and many others’) happiness.

Thanks to my little favourite surfer Helen Penrose who shares with me the blessing of PhD life and its ups and downs. It was invaluable to have you following the same road and to keep each other motivated!

xiv

Thanks to my crazy girl-friends Adrienne Ward, Lisa Jackson, Helen Groombridge and Sonia Bryan, thanks to the whole Barker family (thanks Paddy for all your koala clippings!), to the Cruickshank family, thanks to Mandy and Rod, thanks to them and to many more Straddie crew for being so welcoming and fun to live with.

Thanks to the Miller family, Stan, Patricia, Ian, Sally & family, for welcoming me, and making me feel at home wherever I was invited. Thanks in particular to Erin, Ruby and Tina for making Christmases far from my French family a special moment nonetheless. A special thanks to Carmen and Andrew Barney, and Fiona and Corey Mol, for accommodating me and my strange food habits any time, and very often at the last minute!

Thanks to my French friends who do not forget me, and make me feel welcome when I return home for short and precious moments: Elodie and Boris; Fabien le renard; Aideen Jessenne, Johanna Trollé, Laetitia Taralle and Audrey Lombard; Charly Pignon; Julie Dewilde, Malivonne Seui and Sarah Ferry; Maeva Dewas; Laurence Riquelme; Claire Cayol; Catherine Faure; and petite Cath and her François. I am expecting your visit, now that I finally have some free time to travel around Australia with you. Thanks to Sandra Rude, Nico and Jenny Gonard for their visit! Thanks to Florence Faure for travelling the world and meeting me each time in a new country.

Thanks to my family for supporting me in all my choices even when my choices take me to the other side of the world. Being so far away from you is the most difficult thing I ever had to do in my life. Mum, Dad, Aline, Elouan, Ulysse, Aodrenn, Juliette, Frédéric, Manon, Démian, Moïra: you are a bit heavy, but I always carry you in my heart anyway. Anywhere. Anytime. Australia is the most beautiful place and only you are missing.

Russ you are obviously the single greatest discovery I made during this thesis. No need to say I wouldn’t be here but for you. I can’t thank you enough for being such a happy, supporting, loving partner. You made me laugh every day through this PhD and just with that, you kept me going.

Dedication

To my beloved grand-parents, who saw the beginning and not the end of this PhD. I know you supported my decision to come here but I can’t believe the price you have to pay sometimes to follow your dreams. I am so grateful to have had you on my side all these years, and I owe you much more than words can express.

Thanks to Mamie Hiette and Daddy Serge for their inspiration for rigour and achievement.

Thanks to Mamie Geo for my genetically inherited perseverance (which, in me, some might call stubbornness...).

To Chris Miller, taken too soon from our love...

xvi

Abstract

In the face of decreasing pristine areas worldwide, conservation efforts now regularly include actions to restore disturbed landscapes. The field of restoration ecology is relatively new and rapidly growing, but it is currently heavily biased towards studies of flora rather than fauna. This bias is also reflected in rehabilitation work conducted as part of legislated requirements of, for example, mining restoration. Fauna is probably overlooked because of the largely untested assumption that successful rehabilitation of flora will bring the recovery of fauna. However, ignoring fauna is problematic given that it represents an essential element of biodiversity that is likely to show unique responses to rehabilitation, and plays many crucial roles in ecosystem function. This thesis was conceived to challenge the notion that “flora equals fauna” in post-mining rehabilitation, by examining recolonisation characteristics for one species of particular interest, the koala Phascolarctos cinereus in an area of post-mining rehabilitation on North Stradbroke Island, Queensland. I first review the Australian literature on post-mining rehabilitation and highlight some common processes of fauna recolonisation after mining rehabilitation. I then report on an experimental assessment of the reliability of faecal pellets surveys for koalas, a methodology commonly used in koala studies. Next I test the assumption that successful rehabilitation based on flora criteria would reflect success in koala recolonisation. I found that flora will rarely be a perfect surrogate for fauna, and that this relationship cannot always be assumed but should be demonstrated. I then further investigate koala recolonisation of mining rehabilitation by comparing habitat characteristics, roosting trees, diet, and predation risk for koalas in undisturbed and rehabilitated areas. This chapter explores and rejects the possibility that rehabilitated areas could lure koalas into an ecological trap. Finally, I test the popular conservation paradigm that a single flagship fauna species like the koala can be used as an indicator for rehabilitation success of other fauna (e.g. arboreal marsupials). This hypothesis was rejected; only certain rehabilitated areas were recolonised by gliders, whereas koala presence was more widespread. I conclude with an examination of how successful rehabilitation is for arboreal marsupials at my study site, and make general suggestions for developing fauna criteria for assessing the success of rehabilitation. xvii

Table of Contents

Thesis/Dissertation Sheet ...... ii

Originality statement ...... iii

Authenticity statement ...... v

Statement of contributions of co-authors ...... vi-viii

Acknowledgements ...... ix-xv

Dedication ...... xvi

Abstract ...... xvii

Table of contents ...... xviii-xix

Chapter 1: Introduction ...... 1- 24

Chapter 2: A review of mine rehabilitation and fauna: Current status and future directions ...... 25-68

Chapter 3: Persistence and detectability of faecal pellets in different environments: Implications for pellet-based census of fauna ...... 69-92

Chapter 4: Is restoring flora restoring fauna? Developing fauna criteria for assessing restoration success ...... 93-109

xviii

Chapter 5: Habitat quality and the ecology of an arboreal mammal, the koala Phascolarctos cinereus, in a rehabilitated landscape ...... 111-142

Chapter 6: Assessing rehabilitation success for arboreal marsupials, a reflection on the risk of surrogate species ...... 143-162

Chapter 7: Conclusion ...... 163-175

Appendix A: North Stradbroke Island’s koala population: An island ark? ...... 177-226

Appendix B: Plot details ...... 227-236

Appendix C: Fine-scale movements of koalas in rehabilitated and undisturbed areas ...... 237-244

Appendix D: Blood test results from eight koalas included in Chapter 5 ...... 245-250

References ...... 251-296

xix

Chapter 1

Introduction

The importance of ecological restoration

All ecosystems on the planet are under humanity’s domination (Vitousek et al. 1997; Haberl et al. 2007), owing to increasing human population size and technological advances (Diamond et al. 1989). As a result, the rate of biodiversity loss is accelerating (Pimm et al. 1995). One attempt to counter biodiversity loss is to buy and indefinitely put aside some land to protect the flora and fauna diversity it contains (McNeely & Miller 1984). Yet, protection of biodiversity can no longer rely only on setting conservation areas aside (Sinclair et al. 1995). While the decline of biodiversity is slower in protected areas than in non-protected areas, the level of biodiversity inside protected areas still declines over time (Sinclair et al. 1995). This decline has been attributed to anthropogenic pressures such as tourism, illegal activities, or climate change (Liu et al. 2001; Curran et al. 2004; Lee & Jetz 2008), as well as isolation of the reserves (DeFries et al. 2005). Moreover, the size of the protected area is not always large enough to allow for long-term survival of species (Sinclair et al. 1995; Rosenzweig 2001). In addition, severe habitat destruction and/or fragmentation has already occurred in many parts of the world where habitat protection alone is not sufficient to sustain biodiversity (Sinclair et al. 1995; Hobbs & Harris 2001). Finally, conservation areas are too often granted outside fertile and coastal areas which are generally reserved for agriculture and urbanisation. This conservation bias means that not all ecosystems are equally represented in protected areas (Lunney & Matthews 1997; McAlpine et al. 2007). As a result, a large part of the planet’s biodiversity exists in ecosystems impacted by humans rather than in pristine reserves (Pimentel et al. 1992). The previously described insufficiencies of protected areas to preserve biodiversity are also predicted to increase (Sutherland et al. 2011).

On the basis of all the shortcomings of protected areas, it is a necessity to add new strategies to our conservation actions portfolio. Restoration is one of them (Cairns 1988;

- 1 -

Jordan et al. 1988; Sinclair et al. 1995; MacMahon & Holl 2001), and some have even argued it will soon become the most important (Young 2000). In this thesis, I focus on one particular area of restoration: the one that occurs following mining. This introduction will first define some of the many terms associated with restoration. Some characteristics of mine rehabilitation will be detailed, with an emphasis on the often overlooked yet crucial role of fauna. I also present my study area, describe the rehabilitation carried out on site and the reasons why this mining company is an especially good candidate for research. Finally I outline the thesis structure.

What is ecological restoration?

The Society for Ecological Restoration (SER) defines ecological restoration as “the process of assisting the recovery of damaged, degraded, or destroyed ecosystems” (SER 2004). However, there are many terms that are used (sometimes interchangeably) in relation with ecological restoration, which I discuss in details below.

Restoration is generally used to describe the comprehensive reassembly of a degraded system to its pre-degradation state (Hobbs 1998; SER 2004). This might, however, be an unrealistic goal (Davis 2000; Hobbs 2007). First, if the goal of restoration is to return the ecosystem to a pre-degradation state, then choosing such time in history can be an arbitrary decision. Indeed, questions such as what constitutes an undisturbed, pristine state of the ecosystem (pre-Europeans in USA and in Australia, or even pre-human arrival?), and what records are available to describe it, or even what constitutes a disturbance continue to be unresolved. Secondly, restoring an ecosystem to its undisturbed state seems to imply that this state is at a stable and permanent equilibrium; yet we know this not to be true. Ecosystems are dynamic and restoration may lead to several end points with alternative trajectories that are equally conceivable (Wallington et al. 2005). Finally, restoring a habitat to its historic state may be an unattainable goal in an ever-changing world where the climate is evolving, some keystone species have been removed and/or new species have been introduced (Hobbs & Norton 1996; Davis 2000; Hobbs & Harris 2001; Choi 2004; Wallington et al. 2005; Halle 2007; Hobbs 2007; Jentsch 2007; Choi et al. 2008).

Rehabilitation shares with restoration the use of other natural ecosystems - historical or - 2 - contemporary - as models for the final outcome of a site being rehabilitated. However, rehabilitation emphasises the re-establishment of ecosystem processes, productivity or services. The goal of rehabilitation is to create some form of functioning ecosystem, and not necessarily to match the exact pre-disturbance state (SER 2004).

The term reclamation was first defined in the U.S. Surface Mine Control and Reclamation Act of 1977, and remains quite tightly associated with mines. Reclamation is less stringent than restoration or rehabilitation (Jackson et al. 1995). Reclamation aims at stabilisation, limiting erosion, increasing safety and aesthetic quality, and globally to return the impacted land to a state considered beneficial by local stakeholders (SER 2004). Reclamation is more anthropocentric, while benefits for flora and fauna are not considered in priority.

Revegetation, which is normally a component of reclamation, often only entails mono- species plantation (SER 2004). Mitigation is used when protection or restoration of a part of the landscape will be provided to compensate for the permanent destruction of another part of the environment (SER 2004). Some authors also use the term reallocation to describe the transformation of a site into a more productive or otherwise beneficial use (Hobbs 1998).

Many more definitions exist (see, for example, Aronson et al. 1993; Bradshaw 1996; Cairns & Heckman 1996; Lubke & Avis 1998) and all definitions are still subject to debates. Instead of the classification given above, where restoration covers only the most stringent goal of perfectly recreating the pre-disturbance state, some argue that restoration should describe the entire continuum and encompass all the activities described above (Hobbs & Norton 1996).

In this thesis, unless otherwise stated, I used the term restoration in the general definition of Hobbs & Norton (1996). I use the term rehabilitation to describe the specific efforts aiming at re-establishing a functional ecosystem that tends to approach reference ecosystems while acknowledging that true replication might not be realistic (Hobbs 1998; Davis 2000; SER 2004). I consider Restoration Ecology to be the science of ecological restoration, with its underlying theories and research priorities (Cairns &

- 3 -

Heckman 1996).

Although restoration ecology is still only a recent field of research, since the 1970s it has stimulated an exponential interest (Cairns & Heckman 1996; Choi 2007; Choi et al. 2008). Restoration ecology rests upon two pillars: science and social science (Cairns & Heckman 1996; Davis & Slobodkin 2004; Winterhalder et al. 2004; Temperton 2007), which makes it appealing but also complex.

Restoration and fauna mutualism

Ecological restoration goals frequently focus on the flora component of ecosystems, whereas fauna has received much less attention (Majer 1990; Young 2000). In a review of 68 restoration projects worldwide, the most recurrent emphasis was on plant species diversity (79% of studies), followed by vegetation cover or density (62%). Attention devoted to flora monitoring was well ahead of that for the most common fauna measured, arthropod species diversity, in 35% of the projects (Ruiz-Jaen & Aide 2005). The major reasons for monitoring flora rather than fauna may include that measures of vegetation structure are relatively less labour-intensive and time-consuming, and that seasonal variation in flora measures is usually minor (Ruiz-Jaen & Aide 2005).

The paucity of information on fauna in rehabilitation is nonetheless a shortcoming in our understanding of the process of restoration, as fauna plays many crucial parts at the ecosystem level. For example, fauna strongly influences nutrient cycling, soil aeration and structure, plant composition and productivity, pollination, dispersion of seeds and spores and control of insect pests (Majer 1989; Wunderle 1997; Majer & Brown 1998; Nichols & Nichols 2003; Frouz et al. 2006). Fauna recolonisation will therefore enhance the rehabilitation value both directly and indirectly. The importance of the role that fauna plays in restoration has been highlighted by the growing number of studies which have focused on or included fauna, particularly since the year 2000 (Majer 2009). While this represents a positive step, the proportion of restoration papers referring to fauna still remains low. For instance, reviews of the journal Restoration Ecology revealed that 10% of papers published between 2002 and 2005 partly concerned fauna (Weiher 2007) increasing to only 15% in 2007 (Majer 2009).

- 4 -

There are also many potential benefits of ecological restoration for fauna (Kimber et al. 1999). Hobbs (1993) and Lambeck (1999) have described them in the case of agricultural landscape, but they stand for restoration in general. Restored areas can: 1. increase the amount of available habitat, either by increasing the size of existing patches or by creating new ones; 2. improve the quality of existing habitats, by increasing vegetation diversity or enhancing structure (e.g. adding an understorey layer to a community with only trees); 3. promote connectivity by providing stepping stones or corridors between two habitats; 4. act as a buffer for remaining natural patches, notably by countering edge effects and deleterious impacts from surrounding land uses (e.g. fertiliser contamination, dispersion of invasive species); and by stabilising the landscape so that degradation starting in disturbed areas does not propagate toward the natural environment (e.g. soil erosion and drift, water-logging and rising saline water tables).

The example of mine rehabilitation

A comparison of the impact of mining with other large human disturbances in Australia

Mining is not typically the main threat to biodiversity (Czech et al. 2000) nor does it have the largest footprint (Hobbs & Hopkins 1990; Bell 2001). The disturbance is however total, as it concerns all parts of the ecosystem (Hobbs & Hopkins 1990). Mining impacts range from exploration and prospecting activities to vegetation clearance, the development of mining infrastructure, generation of waste and pollution and the legacy of non- or poorly- rehabilitated sites (Lloyd et al. 2002). The problem of contaminated land can extend beyond the mining site (Marcus et al. 2001; Donato et al. 2007). As the demand for resources grows, the mining impact is also expected to rise (Sutherland et al. 2011).

In comparison, agriculture and forestry disturbances are large in terms of both impact and size (Hobbs 1993). The consequences of these activities include loss of native plants, propagules and seeds, disruption of hydrology with erosion and salinity problems, mass movement of soil, changes in soil composition by spreading fertilisers

- 5 - and pesticides, and soil acidification (Hobbs 1993; Dale et al. 2000). Agriculture usually creates monotonous structural and compositional habitats which entail a loss of fauna biodiversity (Perfecto et al. 1997; Fischer et al. 2007a). The impact from grazing involves, in Australia, the largest surface of all land uses (Hobbs & Hopkins 1990). Grazing has impacts such as vegetation modification, erosion and soil compaction (Hobbs & Hopkins 1990; Hobbs 1993).

In Australia, the urban footprint is relatively small (Hobbs & Hopkins 1990). However, where urbanisation occurs, the impact of the disturbance is close to a total and permanent destruction of biodiversity (McIntyre & Hobbs 1999; McKinney 2002).

Specificity of mine rehabilitation

Mining is a special case of disturbance in that land clearing is a necessary means to an end rather than the end itself. When mineral extraction is finished, the mine site can be returned to another use and this reconversion is a typical part of mining processes. Rehabilitation is thus a necessary outcome of mining activity. In agriculture or urbanisation, the land use is the reason for land clearing, and is intended to last. For this reason, restoration projects in urban environments are often more limited and mostly concern areas such as parks, golf, cemeteries and riparian areas (Savard et al. 2000; Burgin & Wotherspoon 2009). This duality in final land use also links to a difference in respective locations in the matrix of land use. For instance, in agricultural landscapes, there is generally a large proportion of the landscape that is cleared and devoted to production relative to a small amount of remnant and rehabilitated patches (Ryan 2000). In mining, the opposite occurs: the disturbed patch is limited to the economically viable extraction path and surrounded by undisturbed areas. Thus, mine rehabilitation may be more easily colonised by wildlife, whereas agricultural rehabilitation is surrounded by inhospitable land that provides more deleterious edge effects and no population source (Hobbs 1993). Moreover, the parts of the mining lease not directly mined are frequently in a more pristine condition than the land directly adjacent to the lease, owing to restricted access and uses (Lloyd et al. 2002). This creates a difference in the hostility matrix comprising mining and agricultural rehabilitation (McIntyre & Hobbs 1999; McAlpine et al. 2002).

- 6 -

Another specificity of mine rehabilitation is that it is probably the only area of restoration that commonly has a legislative requirement (e.g., U.S. Surface Mine Control and Reclamation Act of 1977, Environmental Protection Act 1994, Queensland Government 2007). As part of the process of mine closure or relinquishment, the post mine rehabilitation is commonly audited and may have to reach certain standards before surrender (Queensland Government 2004). The quality of the rehabilitation is defined in relation to the agreed final land use (e.g. grazing, cropping, forestry plantation, water supply, nature conservation). One part of the requirements to be granted an environmental authority to begin mining is the ability of the mining company, the government, the landowner and any interested stakeholder to reach a consensus on the final land use. Rehabilitation also plays an important role in the calculation of the compulsory financial assurance that mining companies pay the government as a security of compliance to the environmental authority (Queensland Government undated). The financial assurance for instance increases with the total amount of open area waiting for rehabilitation, and can decrease if progressive rehabilitation is carried out and if the mining company can prove its commitment to achieve best environmental performances (Queensland Government 2010b).

Finally, a peculiarity of mine rehabilitation is that, unlike many restoration projects, it is not implemented by groups of volunteers but by powerful international companies. Mining generates vast amounts of money. In Australia alone the gross operating profit was more than A$19 billion for the March 2009 quarter (Australian Bureau of Statistics 2009). Consequently, the mining industry has the economic means to create high- quality rehabilitation.

Mine rehabilitation standards

A challenge that concerns the entire restoration community is to establish standards to measure the degree of success or failure of restoration (Hobbs & Harris 2001; Hobbs 2003). In the mining industry these standards are referred to as “completion criteria”. So far, no state in Australia has adopted standardised completion criteria for mining (Queensland Government 2004). One reason is that completion criteria depend on the environmental, economic and social characteristics of the site, rendering any global criteria difficult (Queensland Government 2007). - 7 -

To complicate the matter, the scientific basis for choosing how to measure restoration success is still hotly debated (Block et al. 2001). In general, completion criteria should be a panel of indicators relevant to restoration objectives that have achieved a specific threshold (Queensland Government 2007). And there's the rub: how to choose relevant indicators? There is probably no one superior paradigm in selecting indicators for rehabilitation success (Ehrenfeld 2000). Indeed to gauge the success of rehabilitation, indicators will have to represent different levels of the ecosystem, from single species ecology to ecosystem processes and functions (Goldstein 1999; Tongway & Hindley 2003; Lindenmayer et al. 2007).

Even though this subject is still polemical within the scientific community, government and industries are being asked to make such decisions. For instance, the Western Australian Environmental Protection Agency proposed a list of eight abiotic and ten biodiversity criteria to measure post-mine rehabilitation success (Table 1, EPA 2006).

Table 1: Example of (a) eight abiotic and (b) ten biodiversity completion criteria given in a governmental document aiming at helping mining companies to achieve environmentally acceptable proposals (EPA 2006)

(a)

Completion criteria category details or examples

Safety, stability Ensure health and safety of humans, stability of soils, landforms and hydrology Sustainability and suitability Provide long-term sustainability without additional management inputs and suitability for agreed land uses Visual amenity Provide visual amenity as defined by community expectations Heritage Retain significant Aboriginal or European abiotic heritage values Pollution Manage pollution Off-site impacts Avoid adverse off-site impacts Hydrology Restore flows and availability of surface and groundwater Soils Maintain soil profiles and structures

- 8 -

(b)

Completion criteria category details or examples

Resilient and self-sustaining Ensure plant reproductive capacity and vegetation resilience to disturbance events Plant species diversity Achieve 60-80% of pre-existing species richness of local native plants Plant abundance and cover Achieve a defined relative cover (% of area) Weed management Manage declared weed, eradicate major environmental weeds Pests and diseases Manage alien or native species of animals, fungi or microbes Plant genetic diversity Restrict seed collection to a narrowly defined geographic region Dominant plant species and plant Restore vegetation structural complexity by strata setting different cover and diversity targets for biodiversity plants that belong to different strata and for keystone species Diversity of ecological Ensure variation in the spatial distribution of communities vegetation Animal diversity Compare diversity of indicative groups of animal species such as birds, reptiles, or invertebrates to pre-disturbance levels Habitat diversity Incorporate the return of structural habitat components, such as logs and rocks

In practice, particular completion criteria are developed by the mining company, and are then reviewed by government agencies in consultation with the public (Queensland Government 2007). Completion criteria, therefore, have to be stringent enough to convince both the government and stakeholders that if the criteria are achieved, the rehabilitated areas will develop in the sustainable ecosystem defined by the agreed final land use (EPA 2006).

To this day, in regulatory as well as in specific mining company documents, fauna criteria are rarely provided (BHP Billiton 2006; EPA 2006; McGlynn & Coutts 2006; Alcoa 2007; Queensland Government 2007; Ward 2008; BEMAX 2009). While the Western Australian EPA’s list of eight abiotic and ten biodiversity criteria (Table 1, EPA 2006) contains one fauna criterion, it also directly highlights that “in most cases it will not be feasible to establish that animal diversity has been restored. However,

- 9 - evidence should be provided to establish that an ecosystem can provide a suitable diversity of habitats for all components of biodiversity” (EPA 2006). This represents one of many examples where it is assumed that proper habitat characteristics will promote fauna return (Koch 2007). In other words, it is assumed that restoring flora is restoring fauna. Interestingly, the mechanism and timing of this process has seldom been studied and even less often demonstrated (Palmer et al. 1997; Bisevac & Majer 1999b; Block et al. 2001). This gap in our knowledge has been underlined as one of the research priorities of restoration ecology (Clewell & Rieger 1997).

Rehabilitation of mining on North Stradbroke Island

My study site was North Stradbroke Island (NSI), Queensland, Australia, where Sibelco (previously Consolidated Rutile Limited or CRL) conducts a heavy mineral extraction operation. The company exploits heavy minerals that are extracted from the sand by gravity and magnetism mechanisms, with no chemicals involved. The sand is dredged from an artificial pond (Figure 1), following a defined mine path. The dredge progresses through the dunes at approximately 1km per year.

Figure 1: Open cut sand mining, with dredge and concentrator on an artificial pond (picture courtesy of Sibelco/CRL)

Rehabilitation occurs continuously, with the area behind the dredge being re-shaped - 10 - according to pre-mining landscape (Figure 2). Aspect, slope and height are matched to previous contours.

Figure 2: Re-creation of the landform (pictures courtesy of Sibelco/CRL)

Topsoil, scraped from cleared areas and stock-piled for up to two years, is spread to a depth of 250 to 300mm (Figure 3). Fencing is installed to prevent adjoining tailings sand migrating onto the spread topsoil (Ben Barker, personal communication, 27/09/2010).

Figure 3: Topsoil is being spread (picture courtesy of Sibelco/CRL) - 11 -

A low-nitrogen fertilizer is broadcast at 300kg per hectare. Liquid lime and a wetting agent are concomitantly incorporated into the soil. While the soil is moist, native seeds, generally collected ahead of the mine path and always within a 30km radius of the rehabilitated site, as well as a hybrid sterile sorghum crop, are sown. Seed mixes vary considerably in accordance with pre-mine vegetation surveys. Seed mixes generally have between 70 and 90 species. The sorghum is put in lines for a fast growing wind break to protect the young native seedlings from wind exposure (Figure 4). Sorghum is an ephemeral plant that dies out after a couple of years. It then acts as a green manure and supplies nutrients to the young seedlings. Terolas (an anionic slow set bitumen emulsion) is being sprayed to stabilise the soil surface and prevent sand and soil movement and erosion (see Bell et al. 1986). Terolas is infused with oxygen, making it highly water soluble and permitting a complete breakdown after approximately two years. Any seed waste generated while processing seeds is broadcast back into rehabilitated areas, as the dried fruit still contains some seeds (Ben Barker, personal communication, 27/09/2010).

Figure 4: Stabilisation with ephemeral sorghum, with log piles and standing trees installed for fauna (pictures courtesy of Sibelco/CRL)

One to two years after direct seeding, nursery stock is planted. An average of 110,000 seedlings per year, at a rate of 1650-2000 seedlings per hectare, is returned to rehabilitated areas. Seedlings are mixed in accordance with pre-mine surveys. The nursery consists mainly of plants from seeds which cannot be mechanically seeded or are either rare or otherwise difficult to collect in large amounts, making mechanical seeding not viable. Other nursery species are grown to fill rehabilitation gaps where initial establishment has not been up to standards. The nursery typically contains 30-40 - 12 - species, depending on the site to be planted (Ben Barker, personal communication, 27/09/2010).

Plants of conservation significance such as orchids and grass trees Xanthorrhoea sp. in the mine path are collected and transplanted back into rehabilitated areas. Occasionally, locations of rare trees such as Endiandra sieberi are recorded so they can be replaced in the same position after mining (Ben Barker, personal communication, 27/09/2010). Fauna recolonisation is encouraged by mulching and deploying log piles, bird perches (standing trees, Figure 4) and nesting boxes.

After assessing the outcome of this routine rehabilitation process, some corrective actions are sometimes necessary. Specific operations on removal of Acacia concurrens (Black Wattle) from rehabilitated areas are carried out, as A. concurrens out-competes any plants growing in the vicinity. Plants are, as much as possible, removed before the first generation of seeds is dispersed as they generate vast quantities. Acacia concurrens is a legacy of old mine practice and can be an issue in areas that have been mined previously. Brush matting is generally used in smaller areas where establishment has not been up to standards or to help prevent erosion, but is not common practice (Ben Barker, personal communication, 27/09/2010).

So far, around 3000ha have been rehabilitated from the six main mines on the island (see details in Table 2 and Figure 5), with 75ha added every year. The older rehabilitation dates from 1975, and methods used at this time have been very different from current good practice standards. Rehabilitation techniques have evolved with the changing views of society, as well as those of mine employees. For example, seed species collected have increased by about 30 species since 2000. New methods are continuously tested and implemented (Ben Barker, personal communication, 28/09/2010). On the basis of these changes, the rehabilitated areas can be broadly classified in three categories. The older rehabilitation (until June 1987) was mainly for stabilisation purpose, and involved exotic plants as well as natives. In June 1987 a new rehabilitation strategy saw the cessation of the use of exotic plants and A. concurrens, and an improved quality of rehabilitation. Finally since 1998, the use of Allocasuarina sp. in rehabilitation has also ceased, and only endemic seeds collected on the island

- 13 - have been used (Paul Smith, personal communication, 12/02/2008).

Table 2: Date and size (in hectares) of rehabilitated areas of six main mines (see also Figure 5)

Rehabilitation dates rehabilitation size Gordon 1985 to 2000 704.5 Bayside 1975 to 1997 735.0 Amity late 70’s to 1992 249.5 Ibis 1998 to 2008 475.5 Enterprise 2008 to present day 161.0 Yarraman 2003 to present day 520.5

Four distinct mine sites have already been closed (Gordon, Bayside, Amity and Ibis) and their rehabilitation entirely completed. Two mines are still active (Enterprise and Yarraman, Table 2 and Figure 5). I included three mines in the study (Bayside, Amity and Ibis), based on prior knowledge of targeted fauna inhabiting the unmined surroundings, and ground-checking. As there were no pre-mining fauna surveys for these mined areas (i.e. fauna surveys of the undisturbed area before it was mined), ground-checking of the undisturbed surroundings of the rehabilitated mine path was the best data I found to suggest mining had occurred in specific fauna habitat and rehabilitation should thus recreate habitat for these specific fauna species. It also ensured remnant populations existed and could potentially recolonise.

Ground-checking in the undisturbed surroundings of Gordon rehabilitated areas did not return fauna signs. Pre-mining fauna surveys were available for Enterprise and Yarraman and did not find any sign of our model fauna either. Consequently, these three mines were not further investigated.

- 14 -

Figure 5: North Stradbroke Island mines projected on a 2008 airborne laser scan of the island (Sibelco/CRL, unpublished data, Köppen zone map from Bureau of Meteorology, Commonwealth of Australia 2007)

- 15 -

Rehabilitation completion criteria in my study site

This mining company is unique in the sense that it is only the second company in Australia to have agreed with the State Government on specific rehabilitation completion criteria. Moreover, it is the first company, to my knowledge, that contains a fauna criterion (Table 3, Environmental Authority from CRL 2007).

Table 3: Completion criteria of Sibelco Australia - Mineral Sand (previously CRL), compiled from the Environmental Authority No. MIM800088202 (mining activities) between the mining company and the Queensland Government Environmental Protection Agency in 2007. The fauna criterion is given as a direct quotation: number Examples of criteria General · agreed land use is bushland 4 · mining leases must be free of waste Infrastructure · all infrastructures must be removed 1 Landform · slopes < 25% 7 · aspect must be as pre-mining in 80% of final landform · volume difference index between pre and post mining landform between -2.5 and 5.5 Geotechnical stability · factor of safety >1.3 2 Erosion · soil loss and accumulated litter equivalent to undisturbed 1 area Revegetation · presence of specific native species 10 · density and richness above threshold (see Table 4) · absence of the worst invasive plant species Water quality · water quality not statistically different between pre and 3 post mining Contaminated land · absence of salinity and diesel pollution 6 and groundwater Water level · water level not statistically different between pre and 3 post mining Fauna · “the environmental authority holder must demonstrate 1 that populations of endangered, vulnerable, rare or near threatened wildlife, as specified in the Nature Conservation Act 1992 and subordinate legislation, on the authorised mining tenement(s) will return to levels equivalent to other similar habitats on North Stradbroke Island.” - 16 -

Table 4: An example of the precision of vegetation criteria, flora criterion 17 (one of the 10 flora criteria). NB: Table 4 is directly extracted from the Environmental Authority No. MIM800088202 “All land disturbed by the mining activity that has been revegetated post-30 June 1987 must comply with the criteria specified below:”

Category stratum pe rformance me asure Number trees · all native species present in the baseline environmental studies and of ESR* are present in the rehabilitation species trees and · the number of native species present in the rehabilitation is not understorey statistically significantly less than 75% of the native species present in the baseline environmental studies and ESR* for the vegetation community · all significant species listed in The Register of the National Estate must be present in the rehabilitation Density trees · the mean stem count of all native tree species greater than 2m in height in the rehabilitation is not statistically significantly less than 75% of the mean value recorded in the baseline environmental studies and ESR* for the vegetation community · for each native tree species present in the rehabilitation, the mean stem count of native trees greater than 2m in height in the rehabilitation is not statistically significantly less than 50% of the mean value recorded for the same native tree species in the baseline environmental studies and ESR* for the vegetation community trees and · the mean stem count of native species in the rehabilitation is not understorey statistically significantly less than 75% of the mean value recorded in the baseline environmental studies and ESR* for the vegetation community Cover trees · the mean projective foliage cover of native species in the rehabilitation is not statistically significantly less than 75% of the mean value recorded in the baseline environmental studies and ESR* for the vegetation community understorey · the mean projective foliage cover of native species in the rehabilitation is not statistically significantly less than 75% of the mean value recorded in the baseline environmental studies and ESR* for the vegetation community ground · the mean projective foliage cover of native species in the rehabilitation is not statistically significantly less than 65% of the mean value recorded in the baseline environmental studies and ESR* for the vegetation community * ESR = Environmental Studies Report describing the results of baseline environmental studies for the specific mine

- 17 -

Table 3 and Table 4 demonstrate that completion criteria are well defined, measurable, comparable to pre-mining or other reference measures, and have a threshold established for fulfilment. The exception concerns the fauna criterion, which is very general and lacks the numbers, precision and details of flora criteria for instance (Table 4). Indeed, for this fauna criterion, the list of concerned species is not given, nor is the method for selecting “similar habitats” or the monitoring techniques - while all these information is present for flora criteria. As recognised by the mining company, the stakeholders and the scientific community, the need for research on how to implement fauna completion criteria is pressing.

Study site

North Stradbroke Island is the largest of a group of massive sand islands in Southeast Queensland. NSI is located in the southern part of Moreton Bay (27º23’/27º45’S, 153º23’/153º33’E). The island is roughly the shape of a triangle of 38km by 11km at its widest point (Figure 5). It is principally composed of dunes following a north-west axis and reaches a height of 219m (Laycock 1978). Most of the island is fringed by low plains where swamps commonly occur (Ward 1978). Geologically, NSI is mainly constituted by unconsolidated Cainozoic sediments (Laycock 1978). The island is part of a drowned landscape formed by high areas of blown sand (Ward 1978), which has accumulated up to 90m deep (Laycock 1975). The ridges have been shaped north- westerly following the prevailing wind influences (Laycock 1978). The sand constituting NSI has been deposited in successive waves, with considerable variations in the amount of time that different sand dunes have been exposed to weathering (Thompson & Ward 1975). In the centre of the island, ancient Pleistocene dune are intensely leached, eroded and podsolised (Ward 1978). Ancient dunes have a first horizon of white siliceous sand lying on a more concrete brown horizon (Rogers 1975). Dunes deposited later during the Holocene on the fringe of the island are somehow less leached (Ward 1978), although all soils on the island are characterised by very low fertility and extremely good drainage (Rogers 1975). Quartz is the main constituent of the sand, with minor concentrations of heavy minerals such as rutile, zircon, ilmenite, monazite, magnetite, and garnet (Laycock 1975).

NSI has a wet-dry subtropical climate (Specht 2009). The winter is drier, with rainfall - 18 - peaking between January and March. The mean annual rainfall is 1550mm, and monthly mean temperatures range between 13ºC and 29ºC (Figure 6).

Figure 6: NSI main rainfall (Dunwich station, 1960 to 2009) and NSI mean maximum and minimum temperatures (Point Lookout station, 1997 to 2010). Data from the Bureau of Meteorology (2010)

The vegetation on NSI can be broadly separated into three groups: vegetation of the ancient dunes, of the more recent dunes, and of the freshwater swamps (Westman 1975). Species diversity is very high in the swamps. The flora diversity decreases with dune age, with the swamps containing 27% more species than the frontal dunes and 112% more than the inland dunes (Westman 1975). The high rainfall of NSI creates a continuous leakage of nutrients partly responsible for the associations between vegetation and dune age (Thompson & Ward 1975). Other influences are dune topography, linked to the distance to the water table, accumulation of organic matter in the swamps, gradients in salt spray, wind intensity and rainfall (Westman 1975).

Flora associations on NSI have been classified into main regional ecosystems by the Queensland Herbarium (2009, see Figure 7). The vegetation of the inland dunes is mainly composed of low woodland of Eucalyptus mallee. The vegetation of frontal dunes comprises Eucalyptus racemosa woodland, Eucalyptus pilularis open forest and - 19 -

Banksia aemula woodland. The lowest parts of the island include Corymbia spp. open to close forest, Melaleuca quinquenervia open forest to woodland, swamps of Baumea spp., Juncus spp. and Lepironia articulate and mangroves (Queensland Herbarium 2009).

Figure 7: Main vegetation communities of NSI (Queensland Herbarium 2009) - 20 -

Focal study species - the koala

The koala, Phascolarctos cinereus (Goldfuss, 1817), was chosen as a focal species for several reasons. Since 2004, the status of koalas in the Southeast Queensland is vulnerable under the Nature Conservation Act 1992. Therefore koalas will be one of the species concerned, by definition, by Sibelco fauna completion criteria.

Koalas are living on the ecological edge (Krockenberger 2003), and they are expected to cope poorly with disturbances (Cork et al. 2000). This could make them particularly vulnerable to mining activities. Koalas are considered as living on the edge principally because of their very poor diet. They are specialist folivores (Martin & Handasyde 1999), eating mainly Eucalyptus spp. leaves. They are also arboreal marsupials, thus their size is constrained to be relatively small so as not to hinder movements in and between trees. Yet to allow enough energy to be extracted from a poor quality and high fibre diet, the intestines and caecum of an animal need to be highly developed, which consequently constrains a minimum body size (Parra 1978). This paradox is exacerbated by the mammal nature of koalas, which portends a high energy demand when producing milk. Koalas are also slow breeding mammals (single young every year or two, Martin & Handasyde 1999), which may further influence their ability to (re-) colonise new habitats.

Finally, koalas focus public attention, and serve as a flagship species for raising stakeholder concerns about the mining impact on NSI. The general appearance of koalas gives them a special place in the socio-politico-economical landscape (Martin & Handasyde 1999) and creates the opportunity for a deeper insight into fauna behaviour in rehabilitation.

Research approach

The motivation for this thesis stems from the lack of knowledge about the fate of fauna in restoration ecology. It focuses on rehabilitation (sensu Hobbs 1998), using mine rehabilitation as a study case. This thesis is divided into seven chapters, the first being this introduction.

- 21 -

Chapter 2 is a formal literature review of the general principles of fauna recolonisation after mine rehabilitation, using Australia as an example. It summarises criteria found to influence fauna recolonisation and compiles characteristics of the species favoured and disfavoured by rehabilitation. Fauna taxa responded to rehabilitation in contrasting ways. Thus, I compared, by fauna taxa, a variety of criteria between rehabilitated and undisturbed areas. Limitations of the literature on fauna recolonisation of mine rehabilitation and future directions are discussed.

Chapter 3 assesses the validity of using temporary signs, such as faecal pellets, as an index for fauna presence. This technique, widely used to study cryptic species such as koalas, is used as a short-cut to index abundance and presence/absence in lieu of direct animal surveys. Indeed, this indirect method is used in Chapters 4, 5 and 6. When using indirect fauna signs for distribution analysis, two factors may be influential. First, sign detectability may vary across the surveyed area, for instance with vegetation complexity. Second, the decay rate of signs (and thus the amount of time that signs remain available for surveys) may also vary across time and space. I designed experiments to study the variations of detectability and decay rate in space, at the scale of my study area, North Stradbroke Island. This information is critical because management decisions are often taken at this scale (e.g. by local councils). The sum of these local decisions can determine the long-term survival of a species. However, local surveys might be biased as variations in scat detectability and decay rate on this scale (same climate, same geographic area, but different vegetation communities) are rarely accounted for. The variations in scat detectability and decay rate I found between vegetation communities at this scale confirm that scat detectability and decay rate can induce serious biases in survey results and should thus be taken into account.

Chapter 4 investigates the relationship between habitat criteria routinely used by mining companies to assess the success of their rehabilitation and fauna recolonisation. The quality of rehabilitation is a legislative requirement that is so far focuses mostly on flora. This relies on an assumption that if flora is appropriately recovering, fauna will naturally follow. This assumption is tested using koalas. I hypothesised that current mining criteria for flora would not correlate with my fauna model. On the contrary, I thought that more ecologically relevant habitat criteria could be developed based on

- 22 - specific koala knowledge. I indeed found a lack of congruence between success in current mining criteria for flora and koala recolonisation. Even when habitat criteria more relevant to koalas were developed, difficulties remained. Implications are discussed in terms of the relevance and challenges of developing appropriate fauna criteria.

Chapter 5 describes the ecology of my fauna model, the koala, for individuals using rehabilitated and undisturbed habitats. A return to the ecology characterising fauna in undisturbed areas should be the ultimate goal of rehabilitation. Notably it is essential to ensure that fauna recolonising rehabilitation finds the appropriate shelter and food and does not face an increased mortality. The risk here would be for rehabilitation to create habitats with an attractiveness disconnected from the real suitability value, i.e. an ecological trap (Dwernychuk & Boag 1972). Alternative hypotheses are that rehabilitation creates a population sink of low quality habitat, or that rehabilitation increases the amount of appropriate habitat available. Although ecological traps might be a real risk in any human-created habitat, on the basis of my preliminary results of the ecology of koalas in rehabilitated areas of NSI, I favour the hypothesis that rehabilitation at this place is able to create good-quality habitat.

Chapter 6 scrutinises the recolonisation of other arboreal marsupials and tests the paradigm of surrogate species; a paradigm commonly used in rehabilitation work. When a high profile species like the koala is present at a site, this species is likely to attract much public attention (i.e. by definition, to be a flagship species). When choosing what species to monitor for assessing rehabilitation success for fauna, a flagship species will always be an obvious candidate. The danger lies in using the same flagship species as an indicator for other species without having first guaranteed the relevance of this choice. Here, I hypothesised that koalas and glider recolonisation could follow the same pattern based on a shared reliance on Eucalyptus forests, but that this similarity would be limited by the hollow-dependent nature of gliders. I found that koalas and gliders presented different recolonisation patterns altogether. These results stand as a warning that koalas would not be an appropriate indicator for the other arboreal marsupials on the island.

- 23 -

Chapter 7 summarises the principal findings of this thesis regarding the ability of the rehabilitated habitat to cater for my fauna models on NSI. The general practical management implications and further avenues for research resulting from this study are underlined. A reflexion follows on how to proceed with the integration of fauna criteria in routine mining monitoring. Finally a call is made for more collaborative research with mining companies, as a means to rapidly advance the science of restoration ecology and its uptake by conservation managers.

Each of the chapters is written in the form of a stand-alone manuscript. As a result of this style of presentation, there is necessarily some minor repetition amongst the different chapters, mainly in the background to the study and details of the study system. Each chapter also has more than one author: thus I use “we” in all chapters (except the conclusion), but in all cases I am the senior author.

- 24 -

Chapter 2

A review of fauna in mine rehabilitation: Current state and future directions

Abstract

Restoration of degraded land has been identified as one of the top research priorities in conservation. Fauna plays a critical role in the re-establishment of a functioning ecosystem, yet fauna recolonisation of restored areas remains poorly understood. We reviewed the findings of the literature on fauna recolonisation, through the example of mine rehabilitation in the Australian continent, a stronghold of large-scale mining.

Rehabilitation favoured fauna opportunists, generalists and species commonly occurring in disturbed areas, while recolonisation by specialists was more problematic. Species densities and richness were frequently lower in rehabilitated areas compared to undisturbed levels, even more so when only endemic species were considered. Recovery of the pre-mining fauna community composition was the hardest to achieve. However, data on ecological characteristics of species, such as body size or reproduction, seemed similar in rehabilitated and undisturbed habitats. Research is needed to determine the duration of the association we found between introduced species and rehabilitated areas.

Limitations to this review included strong biases toward certain mining companies, as well as missing data, which decreased the power of meta-analyses. The publications did not evenly represent all fauna taxa and studies were short when compared to the time needed to re-construct a whole ecosystem. Our review highlights the efficiency of continuous improvement in rehabilitation methodology, the necessity for long-term monitoring and adaptive management. The development of comprehensive fauna standards for assessing rehabilitation success is critical. This could be the next challenge in restoration ecology.

Keywords: fauna; recolonisation; rehabilitation; restoration; disturbance; mine

- 25 -

Introduction

With the growing demands of an ever-increasing human population, all ecosystems on earth are now under anthropogenic pressures (Vitousek et al. 1997). Habitat degradation, driven by land clearance for agriculture, urbanisation, logging and mining, is recognised as being the main cause of biodiversity loss worldwide (Dobson et al. 1997). As a result, restoration of degraded land has become a critical priority for conservation efforts (MacMahon & Holl 2001). In this review, we outline some of the successes and short-comings of habitat rehabilitation (sensu Hobbs 1998) in the context of mining. Mining provides an ideal environment to study habitat rehabilitation, since one of its by-products is bare land that needs to be rehabilitated before the end of mine life, i.e. mine closure.

While mining may not have the largest footprint in comparison to other anthropogenic disturbances (<0.1% in Australia for instance, Hobbs & Hopkins 1990; Bell 2001), it is recognised to have a significant and growing environmental impact (Walker & Willig 1999; IUCN & ICMM 2004; Sutherland et al. 2011). The number of mine closures around the world is expected to substantially increase in the near future (World Bank and the International Finance Corporation 2002). It is thus worrying that the vast majority of mining countries lacks appropriate legislative and policy frameworks with regard to mine closure (Clark & Cook-Clark 2005). Exceptions include North America and Australia which, after a heavy legacy of abandoned mines, are now closely regulated (World Bank and the International Finance Corporation 2002). These countries will therefore be in the forefront of setting up worldwide standards of environmentally and socially adequate procedures for mine closure.

A critical part of mine closure is to ensure that mined sites are rehabilitated in accordance with public interest (Wilson 1999). To evaluate the success of mine rehabilitation, it is therefore important to establish performance standards and monitor the progress of rehabilitated habitats toward them (Smyth & Dearden 1998). To date, the measures of physical factors (e.g. water quality, land topography) and flora (e.g. plant density, richness) are most commonly the basis of monitoring standards. Fauna, on the other hand, is not often monitored (Smyth & Dearden 1998) because animals are assumed to return following the re-establishment of flora (Block et al. 2001; Thompson - 26 -

& Thompson 2004). Few empirical studies have, however, demonstrated that restoring flora leads to restoring fauna (Majer 1990; Clewell & Rieger 1997; Bisevac & Majer 1999b). This remains an important gap in research on habitat restoration, as fauna return is essential in more than one way. Not only is fauna an integral component of an ecosystem, but it also plays a key role in many processes that would enhance restoration success. These include nutrient cycling, soil aeration and structure, plant composition and productivity, pollination, dispersion of seeds and spores, and control of insect pests (Majer 1989; Topp et al. 2001; Nichols & Nichols 2003; Frouz et al. 2006).

Here, we reviewed the current state of knowledge in fauna recolonisation of mine rehabilitated areas. We used the Australian continent as an example of what mining companies can hope to achieve in a context where legislative and social frameworks are promoting good environmental practices. Australia is recognised as a worldwide leader in mine closure (for instance, see the work on mining certification in Rae & Rouse 2001). Australia also has a booming mining industry. Thus the interactions between fauna and mine rehabilitation are of growing importance for fauna conservation on the scale of the continent. We aimed to extract information on the success of rehabilitation regarding fauna, and the predictors of rehabilitation success. We also highlight potential limitations of fauna research so far and future directions worth investigating. We conclude by underlining the importance of developing a relevant assessment of fauna success in mine rehabilitation. This is applicable to all restoration projects and is becoming a crucial part of our conservation effort.

Methods

We undertook a detailed search of peer-reviewed literature relating to fauna recolonisation after mining activities. We only used peer-reviewed papers to avoid a bias of accessibility and to act as a quality control for the material used. We used every combination of “fauna” or “animal” or “recolonisation” and “mine” or “mining” or “rehabilitation” or “restoration” or “mining disturbance” as search terms in ISI World of Science and Google scholar (last searched in March 2011), then subsequent reference lists, as well as author bibliographies. We rejected one publication on the basis that no rehabilitation was performed at the study site (Fletcher 1987), and three that focused on the interaction between mine rehabilitation and unrelated industrial pollution (Letnic & - 27 -

Fox 1997a; Letnic & Fox 1997b; Madden & Fox 1997). We found 70 publications consisting of 38 journal articles, 13 proceedings of conferences and workshops, 10 published bulletins and government reports, and 9 book chapters (Table S1).

First, we summarised the general characteristics of publications (e.g. type of mines used as study site, taxa studied) and the criteria used by these publications to measure the success of rehabilitation with regard to fauna. Second, we described, across all publications, the level of rehabilitation success as assessed by each fauna criterion. Third, we identified, again across all publications, the most frequent predictors of rehabilitation success for fauna. At minimum, publications assessed fauna criteria at the scale of rehabilitation blocks of a given age or methodology, allowing for comparison, between rehabilitation blocks, of the influences of different predictors. A smaller number of publications also gave details of success in fauna criteria at the plot level. This enabled us to further investigate the success of each plot in function of the predictors (i.e. explanatory variables) identified in the first step (at the rehabilitation block level). Density and richness were used as response variables for measuring fauna success because they were the only success criteria present frequently enough to be studied with precision. Five explanatory variables were available, two of them quantitative and three qualitative. The quantitative ones were time since rehabilitation in years, and rainfall at the study site. The qualitative three were: a. method of rehabilitation (0 = no rehabilitation performed, 1 = topsoil only, 2 = topsoil and plantation, 3 = topsoil and seeds, 4 = topsoil, seeds and seedlings); b. quality of the topsoil (0 = stockpiled, 1 = fresh); and c. taxa concerned by the study (taxa 1 = ants, taxa 2 = invertebrates, taxa 3 = beetles, taxa 4 = grasshoppers, taxa 5 = termites, taxa 6 = birds, taxa 7 = mammals, taxa 8 = invasive species3, taxa 9 = collembola, taxa 10 = amphibians, taxa 11 = reptiles, taxa 12 = arachnida, taxa 13 = crustacea, taxa 14 = chilopoda)

Plots for which density and richness were both available (18 publications, 204 plots) were analysed with a redundancy analysis, which models multiple response variables against multiple explanatory variables (Quinn & Keough 2006). Plots for which the only criteria available were either density (27 publications, 350 plots) or richness (25 publications, 292 plots) were also analysed in a multiple linear regression to compare

- 28 - the variables most influential for each response variable. We used a priori models (Johnson & Omland 2004) based on the five explanatory variables identified to influence recolonisation of fauna. This process allowed us to avoid step-wise selection techniques (Mac Nally 2000) and data dredging (Anderson 2001; Anderson et al. 2001a; Burnham & Anderson 2002). Global models comprised all five variables. The set of a priori models comprised models with the two most influential explanatory variables found by the redundancy analysis (see below) and every combination of these two variables with one, two or three (i.e. global model) of the three less influential variables. To account for model uncertainty we used the multi-model inference framework (Burnham & Anderson 2002).

Data analyses were performed in R 2.12.0 (R Development Core Team 2010). Variables were checked for outliers and skewed variables, and none were found. Explanatory variables were standardised to allow comparison of model parameter estimates (Quinn & Keough 2006). Prior to the inclusion of any explanatory variables in the models, collinearity was tested using variance inflation factors (VIF) and no collinearity was found (all VIF were <2). Density and richness were calculated as proportions (species density or richness in rehabilitated areas divided by total density or richness) and arcsin transformed. We estimated the global goodness of fit of the linear regressions with a Wald test using the “lmtest” package (Hothorn et al. 2010). Redundancy analyses were performed using the “Vegan” package (Oksanen et al. 2011).

Results

Publications overview

The 70 publications represented a total of 20 different mines, with 33 publications (n) based in a single mining company (Alcoa, Figure 1). Targeted minerals were bauxite (n=37), sand (n=21), uranium (n=5), coal (n=4), iron (n=2), manganese (n=2) and gold (n=2). Invertebrates were included in 40 publications and vertebrates in 36 publications (Table S1). Ants were the most commonly studied taxon (n=30). Across all publications, the criteria the most commonly assessing rehabilitation success were species presence and their density. All publications but two used direct fauna observations, and two used indirect signs (faecal pellets). Most publications (n=62/70)

- 29 - assessed more than one species and described for one or more taxa their richness (number of species), diversity, evenness or similarity in community or functional group composition. Eleven publications considered individual species characteristics, including body condition and/or reproductive status (n=10), movement patterns and behaviour (n=1) and nesting (n=1, Table S2). Hereafter, we use N to represent the number of instances where equal, better or similar criterion was reached in rehabilitated areas (numerator) relative to total instances examined (denominator), as most publications included data on more than one taxa, and some publications used the same dataset more than once (number of datasets = 88, Table S2).

(a)

(b)

- 30 -

Figure 1: Maps of Australia showing (a) all past and present mine sites and (b) the sites having published information on fauna recolonisation Data in (a) was supplied by Geoscience Australia, Commonwealth of Australia 2010. Black lines are major climate zones of Australia according to Köppen classification (Bureau of Meteorology, Commonwealth of Australia 2007) showing the different climatic zones mining occurs in Australia (Climate zones given in Map b: 1: Equatorial, 2: Tropical, 3: Grassland, 4: Desert, 5: Subtropical and 6: Temperate)

Success in fauna recolonisation

Less than half of the taxa studied in rehabilitated areas achieved equal or better density and species richness than undisturbed areas (N=58/134, Table S2). Diversity and evenness were not studied as often as density or richness, but when given they were almost half of the time equal or better in rehabilitated areas than in undisturbed areas (N=19/39). Community composition seemed the most difficult to re-establish (N=3/37). We found three important trends explaining the differences in community compositions in rehabilitated and undisturbed areas. Firstly, recolonising species were opportunists, generalists, cosmopolitans, and thermophilics, as well as species typical of other types of disturbed areas that were not present to the same extent in undisturbed areas (Greenslade & Majer 1980; Majer et al. 1984; Mawson 1986; Van Shagen 1986; Jackson & Fox 1996; Nichols & Nichols 2003; Moir et al. 2005). Secondly, introduced species were more frequently associated with rehabilitated areas than with the undisturbed surroundings (Table S3). Some studies found that introduced species recolonised rehabilitated areas before any endemic species (Fox & Fox 1984; Twigg et al. 1989; Wilson et al. 1990; Greenslade & Majer 1993; Fox 1996) and remained dominant (Majer 1984b; Majer 1985; Andersen 1992). Introduced species were also frequently found at higher densities in rehabilitated areas than in undisturbed areas (Kentish & Bourne 1982; Walker et al. 1986; McNee et al. 1995; Knight 1999). They were even present in rehabilitated areas when absent from adjacent undisturbed areas (Fox & Fox 1978; Kabay & Nichols 1980; Andersen 1993). Thirdly, species such as climate specialists, forest specialists, cryptic species, litter species, specialist predators, fauna with limited dispersal abilities (Greenslade & Majer 1980; Majer & Nichols 1998; Moir et al. 2005), although commonly found in undisturbed areas, were seldom

- 31 - recolonising rehabilitated areas. Rarely had all species present in undisturbed areas recolonised rehabilitated areas (N=2/19, although this might be biased by numerous plots being relatively young rehabilitation, i.e. less than 15 years old, Table S4).

Criteria influencing fauna recolonisation

At the rehabilitation block level

The factors found to influence rehabilitation success for fauna across publications were mainly the rehabilitation methodology and the age of the rehabilitated areas, then flora (which could be a consequence of some other factors as well). Other factors mentioned were climatic zone (rainfall in particular), distance from undisturbed area, species abundance in surrounding areas, mine shape, and connectivity between population source and rehabilitated areas (see each factor and its influence in Table 1).

Table 1: Most common rehabilitation characteristics found to influence recolonisation patterns within and between mine sites at: (a) mine scale and (b) rehabilitation scale

(a) Mine influence on recolonisation papers describing this characteristics influence Climate (particularly increased rate of recolonisation from Majer et al. 1982, Majer rainfall) dry to moist climate 1984b and 1984c

Active preservation of - persistence of a population source as - Majer 1989b fauna in undisturbed a condition sine qua non to areas surrounding recolonisation

rehabilitated areas - connectivity between population - Majer 1989b source and rehabilitated areas as a condition sine qua non to recolonisation

- trends in this population source will be - Nichols and Grant 2007 mirrored in recolonisation rate Shape impact of mine shape (localised impact Majer 1989b of pod; broader impact of linear mines through temporary division of local populations) can have long-term effects on the characteristics of population sources (and thus the recolonising individuals) - 32 -

(b) Rehabilitation influence on papers describing this influence characteristics re colonisation

Improved - increased rate of - Dunlop et al. 1985; Simmonds et al. rehabilitation recolonisation 1994; Majer and Nichols 1998; Nichols method 1998b; Bisevac and Majer 1999a

- increased density, - Majer 1978b and 1984c; Nichols and richness or evenness Watkins 1984; Collins et al. 1985; Armstrong and Nichols 2000

- community compositions - Nichols and Watkins 1984; Nichols more similar 1985; Nichols and Bamford 1985; Ward et al. 1990; Armstrong and Nichols 2000

Time since - increased density and - Nichols and Bunn 1980; Fox and Fox rehabilitation richness with time 1982; Dunlop et al. 1985; Mawson 1986; Greenslade and Majer 1993; Simmonds et al. 1994; Clements 1998; Cuccovia and Kinnear 1999; Majer et al. 2007; Brady and Noske 2009

- community compositions - Nichols and Watkins 1984; Ward et al. more similar with time 1990; Armstrong and Nichols 2000; Majer et al. 2007; Koch et al. 2010

Distance to - increased richness in the - Andersen 1992 and 1994 undisturbed areas edge of rehabilitated areas

- difference in species - Bragg et al. 2005 composition (edge avoiders) Improved flora parallel flora/fauna Fox & Fox 1978, Majer et al. 1982; (method, time and evolution of species Majer et al. 1984; Mawson 1986; Van climate could all richness, diversity or Shagen 1986; Twigg and Fox 1991; affect fauna composition Simmonds et al. 1994; Andersen et al. through their 2001; Moir et al. 2005; Brady and influence on flora) Noske 2009

Rehabilitation success for fauna was also variable between taxa. To describe this variability, below we present a summary of success criteria per taxa present in the literature. The specific references included in the trends below are detailed in Table S2 when too numerous to quote.

Arthropods. Ant studies showed that density was more often than not equal or superior in rehabilitated areas than in undisturbed areas (N=7/10, Figure 2). On the contrary, ant richness (N=5/17) as well as diversity and evenness (N=10/22) were - 33 - less often than not equal or better than in undisturbed areas (Figure 2). Overall, we found that composition of ant species (N=0/16) and functional groups (N=0/8) in rehabilitated areas bore little resemblance to those in undisturbed areas. Similarly to ants, density, richness, diversity and evenness of other insects had variable success across studies; and species composition was only once similar in rehabilitated and undisturbed areas (N=1/8). In rehabilitated areas compared to undisturbed areas, Collembola were equal or more abundant (N=3/3), but their richness, diversity and evenness were only once equal or better than in undisturbed areas (N=1/6). Crustaceans were not found to recolonise (Majer 1978a; Van Shagen 1986) and Myriapoda with difficulty (Majer 1981; Majer et al. 2007). Arachnidan recolonisation success was found variable across studies, with nine out of 21 criteria (density, richness, diversity and evenness) equal or better in rehabilitated areas compared to undisturbed areas (Figure 2). Species composition was again rarely similar in rehabilitated and undisturbed areas (N=1/5).

Herptiles. Densities and richness for herptiles were rarely equal or better in rehabilitated areas than in undisturbed areas (N=5/27, Figure 2), while diversity and evenness were rarely studied. However, Nichols and Bamford (1985) found that body conditions of herptiles did not differ between rehabilitated and undisturbed areas. Changes in behaviour between rehabilitated and undisturbed areas were found in lizards (Craig et al. 2007). In contrast to reptiles (Walker et al. 1986; Twigg & Fox 1991), amphibians present in rehabilitated areas showed no breeding signs (McNee et al. 1995).

Birds. Recolonisation by birds was the most successful, as density, richness, diversity and evenness were frequently equal or superior in rehabilitated areas than in undisturbed areas (N=19/25, Figure 2). While breeding was recorded in rehabilitated areas as early as five years post mining (Reeders 1985; Curry & Nichols 1986; McNee et al. 1995; Doyle & Kaeding 1998), some authors expressed concerns that hollow nesters may not be capable of breeding in rehabilitated areas (Collins et al. 1985).

- 34 -

Figure 2: Success criteria compared between rehabilitated and undisturbed areas across all original datasets (number of datasets=88) ordered from most (birds) to least (Myriapoda/Crustacea) successful

- 35 -

NB: Studies on different taxa had also varying numbers of years since rehabilitation (birds: 1-24, arachnida: 1-18, other insects/collembola: 0.5-20, mammals: 0-20, ants: 0- 20, amphibians: 4-14, reptiles: 2-23, myriapoda/crustacean: 0-13), which could influence the success criteria

Mammals. Similar to herptiles, density and richness of mammals in rehabilitated areas were rarely found equal or better than in undisturbed areas (N=6/19, Figure 2). While some studies found that females present in rehabilitated areas were carrying young (Nichols & Grant 2007), other studies of breeding did not find similar trends (Walker et al. 1986).

At the plot level

The arrangement of the 204 plots in the redundancy analysis ordination is shown in Figure 3.

Figure 3: Redundancy analysis ordination of two success criteria (density and - 36 -

richness) for 204 rehabilitated plots Qualitative explanatory variables are preceded by “f” (for factor), first level of each qualitative variable is used as a reference and does not appear in the figure

Method of rehabilitation: 0 = no rehabilitation performed, 1 = topsoil only, 2 = topsoil and plantation, 3 = topsoil and seeds, 4 = topsoil, seeds and seedlings Topsoil: 0 = stockpiled, 1 = fresh Taxa: taxa1 = ants, taxa2 = invertebrates, taxa3 = beetles, taxa4 = grasshoppers, taxa5 = termites, taxa6 = birds, taxa7 = mammals, taxa8 = invasive species, taxa9 = collembola, taxa10 = amphibians, taxa11 = reptiles, taxa12 = arachnida, taxa13 = crustacea, taxa14 = chilopoda

Eigen-values of first and second axis were 146.6 and 19.1, and the variables accounted for a cumulative variance explained of 45% (Table 2). Mainly two variables explained large amounts of variance in fauna rehabilitation success, although all five variables were significant (Table 2). The fauna taxon explained 20.3% of the variance and the method used for rehabilitation accounted for 15.5% of the variance.

Table 2: Percentage of variance in fauna rehabilitation success criteria (density and richness) explained by explanatory variables, and level of significance (significance results based on 1000 permutations)

Explanatory variance cumulative p values variables explained by variance single variable explained (%) (%) Taxa 20.3 20.3 0.001 Method 15.5 35.7 0.001 Years 4.9 40.6 0.002 Topsoil 3.5 44.0 0.002 Rainfall 1.4 45.5 0.024

When we studied the success criterion density and richness separately, our global models of plot success fitted the data better than the null models for the response variable density (Wald test F=14.75, p<2.2e-16) and richness (Wald test F=7.63,

- 37 - p<2.2e-16). For both success criteria (density and richness) several models were supported by the data (within two AIC units of the best model) and thus we calculated the model average estimates and the unconditional standard errors of each estimate for all variables across all models (Figure 4).

(a) 1.00

0.50

0.00 years seeds rainfall

-0.50 plantation collembola fresh topsoil arachnida arachnida birds invasive species invasive

invertebrates seeds from topsoil seeds and seedlings reptiles reptiles -1.00 mammals mammals beetles

chilopoda chilopoda termites termites crustacea crustacea grasshoppers grasshoppers amphibians -1.50

(b) 1

0.8

0.6

0.4

0.2

-0.2 years seeds rainfall

birds birds arachnida -0.4 plantation fresh topsoil invertebrates

termites termites -0.6 collembola amphibians amphibians reptiles seeds from topsoil seeds and seedlings invasive species species invasive beetles beetles mammals mammals -0.8 grasshoppers chilopoda chilopoda Figure 4: Average parameter estimates and unconditional standard errors of variables influencing (a) density and (b) richness of fauna in rehabilitated areas

- 38 -

For both density and richness, the methodology used in rehabilitation had the strongest positive effect. The effects given are in comparison to the reference level of method 0, where no rehabilitation was done. Methodologies using seeds or seeds and seedlings in particular had strong positive effects on both density and richness (Figure 4). The taxon effects for density and richness are given in comparison to reference values, which are ant density or richness. For density, only invasive species and collembola had a positive effect (i.e. they tended to increase density compared to ants). Invasive species however did not have an effect on richness. Most taxa studied had negative effects compared to ant density and richness (NB: discrepancies with Figure 2 are most likely due to differences in the sample of publications included in each analysis). The number of years post rehabilitation had a small positive effect on the richness of fauna, and almost no effect on the fauna density. The amount of rainfall at the study site had almost no effect on fauna density and richness in our sample of rehabilitated plots (Figure 4).

Discussion

Limitations of the current literature on fauna in rehabilitated mining areas

The distribution of the sources of the publications created a large bias towards one mining company (Alcoa) as well as one mineral (bauxite). Globally, the three major types of mines in Australia (coal, gold, iron, Geoscience Australia, 2009) are not represented in the trio for the highest number of publications (bauxite, sand, uranium). This strongly indicates that the literature available does not represent mine rehabilitation practices as a whole. The literature may in fact be only representative of mining companies with an interest in research; a characteristic probably linked with high standards in environmental practices. As long as mine rehabilitation research depends on mining companies for either access or funding, it may be inevitable that such research be biased towards best rehabilitation practices. Moreover, the two mine types constituting the bulk of the literature (bauxite and mineral sands) have fewer pollution risks than other mining industries [e.g. gold cyanide-bearing tailings (Donato et al. 2007) and coal acid drainage (Harries 1997)], further biasing the literature. As a consequence of these two biases, the results of this review might be giving an over- optimistic view of mine rehabilitation success for fauna.

- 39 -

Studying fauna recolonisation of mining sites poses both conceptual and logistical challenges (Michener 1997). The specific challenges encountered in studies constituting this review are summarised in Table 3. Some of these challenges such as confounding effects or pseudo replication are not easily avoided. However, the lack of consistency in data reporting could be corrected. For large scale experiments, meta-analysis is a powerful way to overcome lack of replication and to extract general principles (Oksanen 2001). However, even though we initially aimed to conduct a meta-analysis of factors influencing rehabilitation success for fauna, owing to numerous missing data we could only focus on fauna density and richness, perhaps not the best criteria for measuring fauna success (see below “The index paradigm”). We could also only compare a sub- sample of explanatory variables (taxa, methodology, topsoil, time since rehabilitation and rainfall), while others had to be ignored owing to their infrequent presence in original publications (flora characteristics, distance from plot to undisturbed areas, presence of population sources), as well as a sub-sample of our 70 publications. Missing explanatory variables could explain why our redundancy analysis only accounted for 45% of the total variance. We think missing data hampers potential meta-analyses and is a loss for the advancement of restoration ecology. We encourage researchers to publish details of their studies in supplementary materials (available online), which would enhance the quality of future reviews. We were also unable to establish a timescale describing the recolonisation by each taxon. Indeed, the age at which a fauna species was first recorded in rehabilitated areas was mostly determined by the age of the youngest plot included in the study (but with no information on whether even younger plots had been recolonised).

Compared to studies of agricultural rehabilitation (Ryan 2000; Munro et al. 2007), we found that a broader number of taxa was generally studied in the context of mine rehabilitation. While this is a positive finding, some taxa were nonetheless overlooked, and future research should endeavour to include them. Bats were only opportunistically reported (Knight 1999), and taxa such as Mollusca (e.g. snails) and Annelida (e.g. earthworms) were ignored. This happened despite them playing a considerable role in rehabilitation success (e.g., earthworms, Boyer & Wratten 2010). We may add that this is not a weakness particular to mine rehabilitation (e.g., the same limitations were found in a review of fauna in plantations by Lindenmayer & Hobbs 2004).

- 40 -

Table 3: Characteristics of the study designs encountered in this review, and some potential issues related to them (number of original datasets potentially concerned, when available) Elements of in this review potential issues associated study design

Sample size - fewer than 10 plots (27 datasets) - small sample size

- no undisturbed controls (5 - no reference for comparisons datasets) - more replicates in rehabilitated - unbalanced design rarely than in undisturbed areas (42 accounted for in the subsequent datasets); i.e., mean number of analysis (e.g., comparison of rehabilitated plots = 9 (1-30); mean richness between rehabilitated number of undisturbed plots = 4 and undisturbed areas) (1-22), see Appendix D too Chrono-sequence simultaneous comparison of post- confounding effects (e.g., change (space-for-time mining rehabilitated areas of of method and time since substitution) different ages rather than rehabilitation indistinguishable) comparison of same plot through extended periods of time (47 datasets) Repeated studies 19 repeated studies, with mean few long term studies number of visits = 3.9 (2-14), spread over a mean of 9.9 years (2- 27) Rehabilitation 1965 to 2000 latest improved methods not date assessed yet, assessment of rehabilitated areas that used outdated methods Time between 0 to 27 years still short relative to the time an rehabilitation and ecosystem needs to regenerate survey after a major disturbance

Location - often in the same mine - pseudo replication more than true replicates

- different geographical positions - confounding effects inside the mine path often have different times since rehabilitation and different rehabilitation methods Negative results unknown unreported number publication bias toward positive results Missing data on many plot data missing, such as modelling of success of methodology, distance to undisturbed areas (a rehabilitation as a function of results and plots critical measure for animals with these characteristics was limited limited dispersal abilities) due to missing cases

- 41 -

Species selection for conservation and management plans rests on public and funding priorities, which results in certain taxa being consistently overlooked (Tisdell et al. 2006). Interestingly, some of these taxa were more appropriately studied in rehabilitated mines in other parts of the world (readers interested in these taxa should consult Dunger et al. 2001; Wanner & Dunger 2001, 2002; Ganihar 2003; Hohberg 2003; Watters et al. 2005; Andres & Mateos 2006).

Publications on fauna after mine rehabilitation are more numerous than for any other rehabilitation types (Ruiz-Jaen & Aide 2005; Munro et al. 2007). However, there are 328 operating mines and 1113 historic mines in Australia (Geoscience Australia 2009) and we found publications on fauna monitoring relating to only 20 of them (Figure 1). This suggests the potential for many more research projects. Fauna recolonisation of restored areas still has many missing pieces. Mine rehabilitation may be able to address many of these knowledge gaps because it is both under high public scrutiny and mandatory. Thus, a great deal about restoration ecology could be learnt if more mining sites were part of research projects, and unpublished information became available through scientific publications.

Finally, all the results published so far, even the longest studies that represent a whole researcher career (e.g., 50-years project in Germany, Dunger et al. 2004) are short in relation to the time an ecosystem needs to regenerate after such a major disturbance (Wali 1999). Thus, the publications on the degree of success of rehabilitated areas we reviewed are still preliminary results. More time is needed to resolve whether fauna communities will converge towards those of undisturbed areas or achieve a new balance. We need to secure long-term collaborative projects with mining companies to ensure we can study restoration at the appropriate time scale (Oksanen 2001).

Implications of our findings for enhancing rehabilitation impact for fauna conservation

Improvement in rehabilitation methodology was the most often found cause for improvement of fauna recolonisation (Table 1) and had the strongest effects on fauna density and richness (Figure 4). Since rehabilitation practices (at some mines) have improved dramatically over the last 20 years (Nichols 1998a; Bell 2001), rehabilitation - 42 - success in the future should be higher. In particular, new ways of enhancing fauna recolonisation have been promoted (Brennan et al. 2005). Creating a fauna-friendly rehabilitated landscape starts with improving vegetation structure and composition, and can be accompanied with specific actions such as (1) implementing feral predator control to decrease the impact of their predation on recolonising fauna, (2) catering for hollow-dependent fauna by accelerating the creation of natural hollows and/or providing nest boxes, (3) increasing landscape complexity by adding dead stags, rocks, log piles and coarse woody debris, and (4) replacing resources provided by long-lived, slow- growth trees such as Xanthorrhoea sp. by transplanting them from ahead of the mine path to rehabilitated areas (Brennan et al. 2005). Worth noting are studies from Europe underlining that abandoned mines with no rehabilitation have higher biodiversity values than mines rehabilitated using poor methodologies (Hodacová & Prach 2003; Hendrychová 2008). Consequently, when rehabilitation is performed, it is crucial to ensure that only current best practices are implemented.

Fauna criteria (e.g. density, richness, composition) in rehabilitated areas generally improved with time (Table 1). The influence of time on fauna criteria in Table 2 is probably decreased by the strong influence of taxa and method. If taxa and method were standardised across studies, the influence of time since rehabilitation (through the development of habitat, the colonisation by new species, etc.) could become more apparent. However, we were concerned that some rehabilitated areas seem to evolve towards undisturbed areas only to a certain point and then become stagnant [e.g., Andersen (1993), Craig et al. (Craig et al. 2010), see also Davis et al. (2003)]. Some further rehabilitation management (e.g. thinning, burning) might be necessary as rehabilitated areas age (Ross et al. 2004). This underlines the need for long-term monitoring, adaptive management and for securing funding for rehabilitation even after mine closure.

Only fauna species present in undisturbed surrounding areas and with survival and reproductive rates sufficient to replace their populations and produce dispersing individuals, can be potential colonisers of rehabilitated areas. Thus, any rehabilitation project has to be integrative and preserve species in the areas not directly impacted by mining. For example, implementing predator control in the surroundings of an Alcoa

- 43 - bauxite mine was found to boost the numbers of small marsupials and improved recolonisation of rehabilitated areas (Nichols & Nichols 2003). A positive consequence of restricted access and uses of mining leases is that areas managed within the mining leases (but not directly impacted by mining operations) are frequently in a more pristine condition than the land outside the leases (Lloyd et al. 2002). This should enhance the protection of population sources.

Our review showed that invasive species were frequently associated with rehabilitation (Table S3) and were found at a higher density than most native species (Figure 4a). Future research is warranted for assessing the potential impact of invasive species, i.e. whether they are early or enduring colonisers. For example, several studies found that while house mouse density is initially high in rehabilitated areas, it decreases with time and in some cases the density of native mammals increases in parallel (Fox & Fox 1978; Fox & Twigg 1991; Fox 1996; Nichols & Nichols 2003). Long-term management of rehabilitated landscape may need to integrate an invasive species control plan if invasive species are enduring colonisers.

Suggestions to improve how success is assessed in ecological restoration

So far the impact of restoration projects on fauna is usually not included in the assessment of restoration success (Young 2000; Ruiz-Jaen & Aide 2005). We argue that if restoration is to fulfil its role in curbing habitat loss and improving conservation of biodiversity (Young 2000), appropriate fauna criteria need to be urgently adopted as part of routine monitoring and assessment of the success of restoration projects. Only scientifically-sound monitoring programs, such as advocated in adaptive management, can allow efficient corrective actions to be taken and improve conservation benefits (see Keith et al. 2011 and rest of the special issue).

The flora equals fauna paradigm

As noted earlier, it is often assumed that fauna will return if flora is re-established. While some studies found a congruence between the development of flora and fauna in rehabilitated areas (Majer et al. 1984; Van Shagen 1986), they did not necessarily take into account species identity (but see Fox & Fox 1978; Fox & Fox 1984). Not

- 44 - controlling for species identity when investigating links between flora and fauna recovery could lead to erroneous results (Goldstein 1999). For instance, species richness could increase for both fauna and flora, but such increases may not reflect changes in endemic species. Moreover, some studies found that vegetation diversity does not correctly reflect fauna diversity (Crisp et al. 1998; Longcore 2003; Andersen et al. 2004; Fleishman & Murphy 2009). This is unsurprising as, for instance, different beetle species correlate to the same flora characteristics in the opposite direction (Azeria et al. 2009).

Consequently, flora might be a necessary condition but not sufficient to promote fauna recolonisation. If the goal of restoration is to re-establish a functioning ecosystem similar to pre-disturbance, fauna species are part of the target and should therefore be part of the monitoring.

The index paradigm

From this literature review, a current trend in research on fauna recolonisation is to focus on many species pooled in higher taxa indices (e.g. ant density, reptile richness). This can introduce a spurious effect if specific identities are not determined (e.g. identification only to the genus or family) or if species are used interchangeably (Goldstein 1999; Hilty & Merenlender 2000). To be precise, when comparing species richness in rehabilitated and undisturbed habitats, the value can be similar even if some species from undisturbed areas are still absent in rehabilitated areas (because some other species are in the opposite position, see results above). For instance in our review, 29% of studies found similar richness between rehabilitated and undisturbed areas, while only 10% found that all species found in undisturbed areas had recolonised rehabilitated areas (Table S2 and S4).

Not making this distinction when assessing richness will be a problem in many cases. Particularly, if the desired outcome is to create an ecosystem that resembles the pre- disturbance state, then the appropriate measure to be compared is the percentage of initial species present in undisturbed areas having recolonised rehabilitated areas (Table S4). If, on the contrary, the final outcome is to create any functioning ecosystem, then reaching similar levels of richness might be appropriate (whatever the identity of these - 45 - species). This might be the case when the disturbance is such that the initial conditions can no longer be recreated (e.g., creation of wet grassland where a woodland was present, Lannoo et al. 2009). Even in that case, species richness might not be pertinent, as it includes invasive species (such an example can be found in Bennett 1990). Moreover, greater species richness is not necessarily synonymous with a better functioning ecosystem (see examples in Lindenmayer 1999). Thus, indices of success have to be very clearly defined towards meeting some a priori goals. Improperly defined indices might be one of the reasons similar richness and density were more often achieved in rehabilitated areas than community composition in this review. Comparing community similarity, which is by definition based on species identities, could be more effective for assessing rehabilitation success for fauna when the final desired outcome is to re-create the initial ecosystem. Alternatively, when the desired outcome of rehabilitation is to provide a safe, stable and sustainable ecosystem, reaching an exact match in fauna species between rehabilitated and undisturbed areas is not necessary and fauna success criteria should be defined accordingly.

Furthermore, a different picture of the success of each fauna taxon emerged from all success criteria taken together (i.e. birds are the most successful in Figure 2, which encompasses density, richness, diversity, evenness, community composition and ecological characteristics) compared to the picture given by criteria such as density or richness only (i.e. on the basis of Figure 4, ants and collembola are most successful). This casts doubts on the ability of one or two indices alone to represent the degree of fauna success in rehabilitated areas. Adopting a set of complementary success criteria might be necessary to appropriately reflect the specificities of fauna recolonisation.

The indicator paradigm

In ecological restoration, it is common practice to resort to indicator species to monitor the progress of a project toward the final goals (Lindenmayer et al. 2000). An indicator species is selected on the basis of its ability to represent a larger group of species, or sometimes overall biodiversity or environmental health (McKenzie et al. 1992). Many taxa included in this review, particularly invertebrates, have been proposed as potential indicators, either for other species, or more frequently for the rehabilitated landscape as a whole (Bisevac & Majer 1999b; Andersen et al. 2004; Greenslade 2007). However, - 46 - the use of one indicator to describe rehabilitation as “successful” is not trivial. For fauna in particular, the heterogeneity of their recolonisation patterns means that two fauna groups can have very different fates even in the same rehabilitated area (Nichols & Nichols 2003). This is supported by this review, as for example the influence of taxa explained most of the variation in rehabilitation success (Figures 3 and 4). This variation in success across taxa has been found in other contexts (Lindenmayer et al. 2008b). For example, reptile, mammal and bird responses to agricultural rehabilitation have been shown to be different (Cunningham et al. 2007; Cunningham et al. 2008). Thus, the use of a particular indicator to monitor rehabilitated areas can only be reasonable when avoiding general terms such as “indicator of rehabilitation success”. Instead, the precise part of the ecosystem the indicator is supposed to represent should be defined (Fleishman & Murphy 2009), in addition to what constitutes “success”.

The selection of indicators is further complicated by the heated debate around their general usefulness in restoration ecology (Walker 1995; Lambeck 1997; Simberloff 1998; Lindenmayer et al. 2002). Part of the solution may lie in the acceptance that there is not one paradigm for setting goals and monitoring restored habitats (Ehrenfeld 2000), which means we need to develop a multi-discipline and multi-level approach (Corbett 1999; SER 2004; Ruiz-Jaen & Aide 2005). This could integrate some elements of the species approach, of the community approach, and of the ecosystem function and process approach (Ehrenfeld & Toth 1997; Palmer et al. 1997; Goldstein 1999).

For instance, if the goal of restoration is to create a functioning ecosystem similar to its pre-disturbance state, the criteria for success regarding fauna should include data on: (1) keystone and flagship species for this particular ecosystem (Simberloff 1998); (2) taxa involved in ecosystem processes (e.g., soil mesofauna for soil quality, Andres & Mateos 2006); (3) community composition of as many taxa as logistically feasible, to try to avoid unsupported shortcuts (such as one group being an indicator for other groups, Lovell et al. 2007). However, as not all groups can be monitored, taxa with distant phylogeny (i.e. both invertebrates and vertebrates) and taxa that serve other agendas (i.e. included in 1 or 2) should have priority (Hilty & Merenlender 2000).

- 47 -

Finally, the collection of genetic and ecological characteristics, such as behaviour or survival and reproductive rates, needs to be incorporated when assessing the success of fauna in restored areas. This information is critical for assessing population viability of species in the long term (Fox 1998; Bellairs 1999; Miller & Hobbs 2007), and is greatly lacking in restoration ecology (Goldstein 1999).

Obviously, within the limitations of reality, selecting a good compromise between scientific relevance, public expectations and industry endorsement amongst the infinity of available criteria will be a challenge. Nevertheless, the stakes are high: relevant criteria will ensure our costly efforts to restore ecosystems are a useful card in our portfolio of conservation actions.

Acknowledgement

Thanks to Dr. O. Nichols and Prof. B. Fox for commenting on earlier drafts. Thanks to Geoscience Australia for providing data of past and present mines of Australia, and to Russell Miller for assisting in creating Figure 1. R.H.A.C. was supported by an Endeavour Europe Award and an Endeavour International Postgraduate Research Scholarship. Sibelco Australia – Mineral sand provided ongoing support to this research.

Supplementary material

The 70 publications used in this review (Table S1), the trends in criteria of fauna in rehabilitated compared to undisturbed areas, ordered by taxon and time since rehabilitation (Table S2), the invasive species presence, density and richness in rehabilitated areas compared to undisturbed areas, ordered by taxon and time since rehabilitation (Table S3) and the comparison of the number of species having recolonised rehabilitated areas from undisturbed areas for different taxa (Table S4) are available below.

- 48 -

Table S1: The 70 publications used in this review, ordered by first author and date

Authors date publication species mineral name of paper type Andersen, 1992 Book chapter ants uranium The use of ants to A.N. assess restoration success following mining, a case study at Ranger Uranium mine Andersen, 1993 Restoration ants uranium Ants as indicators of A.N. Ecology restoration success at a uranium mine in tropical Australia Andersen, 1994 Proceedings of ants uranium Ants as indicators of A.N. the AusIMM restoration success Conference following mining in Northern Australia Andersen, A. 1998 Supervising ants, beetles, uranium The role of ants in N., S. scientist report grasshoppers, mine site restoration Morrison, and termites and in the Kakadu region L. Belbin more of Australia's Northern Territory, with particular reference to their use as bioindicators Andersen, 2001 Austral Ecology grasshoppers uranium Grasshopper A.N., J. A. biodiversity and Ludwig, L. M. bioindicators in Lowe, and D. Australian tropical C. F. Rentz savannas: Responses to disturbance in Kakadu National Park Andersen, 2003 Ecological ants coal Ants as indicators of A.N. Management minesite restoration, and Restoration community recovery at one of eight rehabilitation sites in central Queensland

- 49 -

Authors date publication species mineral name of paper type Armstrong, K. 2000 Forest Ecology birds bauxite Long-term trends in N. and O. G. and avifaunal Nichols Management recolonisation of rehabilitated bauxite mines in the jarrah forest of south- western Australia Bisevac, L., 1999 Restoration ants mineral Comparative study and J. D. Ecology sands of ant communities Majer of rehabilitated mineral sand mines and heathland, Western Australia Brady, C. J., 2009 Restoration birds bauxite Succession in bird and R. A. Ecology and plant Noske communities over a 24-year chronosequence of mine rehabilitation in the Australian monsoon tropics Bragg, J.G., 2005 Austral Ecology reptiles mineral Distributions of J.E. Taylor, sands lizard species across and B.J. Fox edges delimiting open-forest and sand- mined areas Brennan, K. 2003 Ecological spiders bauxite Using fire to E., J. D. Management facilitate faunal Majer, and J. and Restoration colonization M. Koch following mining: An assessment using spiders in Western Australian jarrah forest Clements , A. 1998 Workshop ants, birds, gold / sand Rebuilding fauna proceedings reptiles, habitats "Fauna habitat mammals reconstruction after mining"

- 50 -

Authors date publication species mineral name of paper type Collett , K. 2003 Australasian scorpion bauxite Recolonisation of Arachnology Urodacus rehabilitated bauxite planimanus mine pits by the scorpion Urodacus planimanus Collins , B. G., 1985 Book chapter birds bauxite Recolonization of B. J. Wykes, restored bauxite and O. G. minelands by birds in Nichols Southwestern Australia Craig, M. D., 2007 Australian western bauxite Ecology of the M. J. Journal of bearded dragon western bearded Garkaklis, G. Zoology Pogona minor dragon (Pogona E. S. J. minor) in unmined Hardy, A. H. forest and forest Grigg, C. D. restored after Grant, P. A. bauxite mining in Fleming, and south-west Western R. J. Hobbs Australia Craig, M. D., 2010 Restoration reptiles, small bauxite Do thinning and R. J. Hobbs, Ecology mammals burning sites A. H. Grigg, revegetated after M. J. bauxite mining Garkaklis, C. improve habitat for D. Grant, P. terrestrial A. Fleming, vertebrates? and G. E. S. J. Hardy Cuccovia, A., 1999 Book chapter acarina bauxite Acarine (mite) and A. communities Kinnear colonizing rehabilitated bauxite mine pits in the jarrah forest of Western Australia Curry, P. J., 1986 Australian birds bauxite Early regrowth in and O. G. forestry rehabilitated bauxite Nichols minesites as breeding habitat for birds in the jarrah forest of south- western Australia

- 51 -

Authors date publication species mineral name of paper type Doyle, F., and 1998 Workshop birds mineral Bird colonization of G. Kaeding proceedings sands constructed "Fauna habitat wetlands at Capel, reconstruction WA after mining" Dunlop, J. N., 1985 Mulga ants iron A preliminary J. D. Majer, Research assessment of C. J. Morris, Center Journal minesite and K. J. rehabilitation in the Walker pilbara iron ore province using ant communities as ecological indicators Fox, B. J., and 1978 Australian New Holland mineral Recolonization of M. D. Fox Journal of mouse sands coastal heath by Ecology Pseudomys Pseudomys novaehollan- novaehollandiae diae (Muridae) following sand mining Fox, M. D., 1982 Book chapter ants mineral Evidence for and B. J. Fox sands interspecific competition influencing ant species diversity in a regenerating heathland Fox, B. J. 1982 Proceedings of New Holland mineral The influence of the symposium mouse, ants sands disturbance (fire, on dynamics mining) on ant and and small mammal management of species diversity in Mediterranean Australian heathland type ecosystems Fox, B. J., and 1984 Australian New Holland mineral Small mammal M. D. Fox Journal of mouse sands recolonization of Ecology open forest following sand mining

- 52 -

Authors date publication species mineral name of paper type Fox, B. J., and 1991 Australian mice mineral Experimental L. E. Twigg Journal of sands transplants of mice Ecology (Pseudomys and Mus) on to early stages of postmining regeneration in open forest Fox , B. J. 1996 Book chapter small mammals mineral Long term studies of sands small mammal communities in disturbed habitats of eastern Australia Greenslade, 1980 Book chapter collembola bauxite Collembola of P., and J. D. rehabilitated Majer minesites of Western Australia Greenslade, 1993 Australian collembola bauxite Recolonization by P., and J. D. Journal of Collembola of Majer Ecology rehabilitated bauxite mines in Western Australia Jackson, G. 1996 Australian ants mineral Comparison of P., and B. J. Journal of sands regeneration Fox Ecology following burning, clearing or mineral sand mining at Tomago, NSW: II. Succession of ant assemblages in a coastal forest Kabay, E. D., 1980 Alcoa of mammals, bauxite Use of rehabilitated and O. G. Australia birds, bauxite mined areas Nichols Limited amphibians, in the jarrah forest Environmental reptiles by vertebrates Bulletin Kentish, K., 1982 Bulletin of mammals not detailed Return of fauna to and A. R. Australian heathland Boume Mammal regenerating after Society mining

- 53 -

Authors date publication species mineral name of paper type Knight, R. 1999 Workshop mammals, coal Recalcitrant proceedings birds, recolonisers: "Indicators of amphibians, assessment of ecosystem reptiles rehabilitation rehabilitation success through success" study of small faunal assemblages Koch, J.M., 2010 Annals of collembola, bauxite Arthropods in coarse A.H. Grigg, Forest Science acarina, woody debris in R.K.Gordon, araneida, jarrah forest and and J.D.Majer coleoptera, rehabilitated bauxite diptera mines in Western Australia Majer, J. D. 1978 Book chapter arachnids, bauxite The importance of a crustacea, invertebrates in diplopoda, successful land chilopoda, reclamation with collembola, particular reference insecta to bauxite mine rehabilitation Majer, J. D. 1978 Forest Ecology arachnids, bauxite Preliminary survey b and crustacea, of the epigaeic Management diplopoda, invertebrate fauna chilopoda, with particular collembola, reference to ants, in insecta areas of different land use at Dwellingup, Western Australia Majer, J. D. 1978 Alcoa of ants bauxite Studies of c Australia invertebrates in Limited relation to bauxite Environmental mining activities in Bulletin the Darling Range - a review of the first 18 months research Majer, J. D. 1981 Forests arachnids, bauxite The role of Department of crustacea, invertebrates in Western diplopoda, bauxite mine Australia chilopoda, rehabilitation Bulletin collembola, insecta

- 54 -

Authors date publication species mineral name of paper type Majer, J. D. , 1982 Reclamation ants and soil mineral Colonization by ants M. Sartori, R. and fauna sands and other Stone, and W. Revegetation invertebrates in S. Perriman Research rehabilitated mineral sand mines near Eneabba, WA Majer, J. D. 1984 Book chapter ants mineral Ant return in a sands rehabilitated mineral sand mines on North Stradbroke Island Majer, J. D. 1984 Proceedings of ants mineral Ant return in b the 4th sands / rehabilitated mines – International bauxite / an indicator of Conference on manganese ecosystem resilience Mediterranean Ecosystems Majer, J. D. 1984 Reclamation ants bauxite / Recolonisation by c and manganese ants in rehabilitated Revegetation opencut mines in Research northern Australia Majer, J. D., 1984 Journal of ants bauxite Recolonization by J. Day, E. Applied ants in bauxite mines Kabay, and Ecology rehabilitated by a W. Perriman number of different methods Majer, J. D. 1985 Australian ants mineral Recolonization by Journal of sands ants of rehabilitated Ecology mineral sand mines on North Stradbroke Island, Queensland, with particular reference to seed removal Majer, J. D. , 1998 Journal of ants bauxite Long-term and O. Applied recolonization Nichols Ecology patterns of ants in Western Australian rehabilitated bauxite mines, with reference to use as indicators of restoration success

- 55 -

Authors date publication species mineral name of paper type Majer, J. D. , 2007 Restoration ants, bauxite Invertebrates and K. E. Ecology collembola, the restoration of a Brennan, and acarina, forest ecosystem: 30 M. L. Moir termites, years of research hemipteran, following bauxite arachnida and mining in Western more Australia Mawson, P. 1986 WAIT School arachnids bauxite A comparative study of Biology of Arachnid Bulletin communities in rehabilitated bauxite mines McNee, S. A., 1995 WAIT School mammals, mineral Population ecology A. Zigon, and of Biology birds, sands of vertebrates in B. G. Collins Bulletin amphibians, undisturbed and reptiles rehabilitated habitats on the northern sandplain of Western Australia Moir, M. L., 2005 Forest Ecology hemipteran bauxite Restoration of a K. E. and forest ecosystem: Brennan, J. Management The effects of M. Koch, J. vegetation and D. Majer, and dispersal capabilities M. J. Fletcher on the reassembly of plant dwelling arthropods Nichols, O. 1980 Alcoa of termites bauxite Termite utilisation of G., and S. Australia rehabilitated bauxite Bunn Limited mined areas Environmental Bulletin Nichols, O. 1984 Biological birds bauxite Bird utilisation of G., and D. Conservation rehabilitated bauxite Watkins minesites in Western Australia Nichols, O. G. 1985 Forest Ecology arachnids, bauxite Recolonisation of and chilopods, revegetated bauxite Management insects mine sites by predatory invertebrates

- 56 -

Authors date publication species mineral name of paper type Nichols , O. 1985 Biological amphibians, bauxite Reptile and frog G., and M. J. Conservation reptiles utilisation of Bamford rehabilitated bauxite minesites and dieback affected sites in Western Australia's Jarrah Eucalyptus marginata forest Nichols, O. 1989 Book chapter invertebrates, bauxite The return of G., B. J. birds, reptiles, vertebrate and Wykes, and carpet snakes invertebrate fauna to D. Majer bauxite mined areas in south-western Australia Nichols, O. G. 1998 Workshop ants, arachnids, bauxite Long-term a proceedings birds, reptiles monitoring of fauna "Fauna habitat return in Bauxite- reconstruction mined areas of the after mining" Darling range Nichols, O. G. 1998 Proceedings of ants, bauxite The development of b the fourth arachnids, a rehabilitation International birds, reptiles program designed to Conference of restore a jarrah the forest ecosystem International following bauxite Affiliation of mining in south- Land western Australia Reclamationists Nichols, O. 2003 Restoration ants, mammals, bauxite Long-term trends in G., and F. M. Ecology birds, reptiles faunal recolonization Nichols after bauxite mining in the Jarrah forest of southwestern Australia Nichols, O. 2007 Restoration mammals, bauxite Vertebrate fauna G., and C. D. Ecology birds, reptiles recolonization of Grant restored bauxite mines - key findings from almost 30 years of monitoring and research

- 57 -

Authors date publication species mineral name of paper type Reeders, A. 1985 Proceedings of mammals, bauxite Vertebrate fauna in P. F. the North birds, regenerated mines at Australian Mine amphibians, Weipa, North Regeneration reptiles Queensland Simmonds, S. 1994 Restoration spiders bauxite A comparative study J., J. D. Ecology of spider (Araneae) Majer, and O. communities of G. Nichols rehabilitated bauxite mines and surrounding forest in the southwest of Western Australia Simpson, J. 1998 Workshop koalas mineral Reconstruction of proceedings sands koala habitat after "Fauna habitat titanium minerals reconstruction mining in the after mining" Tomago Sandbeds, NSW Taylor, J. E., 2001 Austral Ecology lizards mineral Disturbance effects and B. J. Fox sands from fire and mining produce different lizard communities in eastern Australian forests Thompson, S. 2005 Pacific reptiles, gold Mammals or reptiles, A., and G. G. Conservation mammals as surveyed by pit- Thompson biology traps, as bioindicators of rehabilitation success for mine sites in the goldfields region of Western Australia? Twigg, L.E., 1989 Austral Ecology small mammals mineral The modified B.J. Fox, and sands primary succession L. Jia following sand mining: A validation of the use of chronosequence analysis

- 58 -

Authors date publication species mineral name of paper type Twigg, L. E., 1991 Australian lizards mineral Recolonization of and B. J. Fox Journal of sands regenerating open Ecology forest by terrestrial lizards following sand mining Van Shagen, 1986 WAIT School arachnida, coal Recolonisation by J. of Biology crustacea, ants and other Bulletin collembola, invertebrates in acarina rehabilitated coal chilapoda, mine sites near insecta, ants Collie, Western Australia Walker, K., J. 1986 Workshop reptiles, small iron Vertebrate Osborne, and proceedings mammals colonization after J. Dunlop "Australian revegetation of a Mining Industry rehabilitated iron ore Council waste dump, Environmental Newman, Western Workshop" Australia Ward, S. C., 1990 Proceedings of predatory bauxite Bauxite mine J. M. Koch, the Ecological invertebrates, rehabilitation in the and O. G. Society of birds Darling range, Nichols Australia Western Australia Wilson, B. A., 1990 Proceedings of small mammals coal Factors affecting D. Robertson, the Ecological small mammal D. J. Society of distribution and Moloney, G. Australia abundance in the R. Newell, Eastern Otway and W. S. Ranges, Victoria Laidlaw Woodward, 2008 Wildlife koalas mineral Koalas on North W., W. A. research sands Stradbroke Island: Ellis, F. N. Diet, tree use and Carrick, M. reconstructed Tanizaki, D. landscapes Bowen, and P. Smith Wykes, B. J. 1985 WAIT School birds bauxite The Jarrah forest of Biology avifauna and its re- Bulletin establishment after bauxite mining

- 59 -

Table S2: Trends in criteria comparing fauna in rehabilitated and undisturbed plots, ordered by taxon and time since rehabilitation

Density: number of animals/area Richness: number of species Diversity, Evenness and Community: based on a variety of indexes depending on the publication “+”: higher in rehabilitated than undisturbed areas “-”: lower in rehabilitated than undisturbed areas “=”: same “≠”: different “+/-”: same composition to unhealthy undisturbed forest but different from healthy forest Age: number of years after rehabilitation of the older plot studied NT: Northern Territory, WA: Western Australia, QLD: Queensland, NSW: New South Wales, Vic: Victoria

* Koch et al. 2010 compared community composition for all arthropods together, thus the difference in community is counted once and not four times

Note on the table: the trends are based on data available through the publications. Sometimes trends had to be calculated differently owing to the different results available. For instance, if richness is available only per plot, the total richness cannot be calculated (as richness is not additive), so mean richness had to be compared in these cases

- 60 -

Taxa others location age References density richness diversity evenness community Formicidae Majer 1978a, Darling ----≠ Biomass - 3 Majer 1978b Range WA Majer et al. Eneabba --- 3 1982 WA Groote Majer 1984c +=++≠ 6 Eylandt NT Andersen 1992, Functional group -- ≠ Kakadu NT 8 Andersen 1993 compositions ≠ Majer 1984c ====≠ Gove NT 8 Majer 1984c =--=≠ Weipa QLD 8 Nichols & Functional group Darling == ≠ 8 Nichols 2003 compositions ≠ Range WA Van Shagen Darling +--≠ 9 1986 Range WA Functional group Andersen 1994 - ≠ Kakadu NT 10 compositions ≠ Functional group Andersen 2003 - ≠ Biloela QLD 10 compositions ≠ Hawks nest Fox & Fox 1982 ++ - - ≠ 11 NSW Andersen et al. Functional group -- ≠ Kakadu NT 12 1998 compositions ≠ Dunlop et al. --≠ Pilbara WA 12 1985 Majer et al. Darling =-== 12 1984 Range WA Majer 1984a, Stradbroke ---≠ 15 Majer 1985 Is. QLD Majer & Nichols Functional group Darling - ≠ 15 1998 compositions ≠ Range WA Jackson & Fox Functional group Tomago - ≠ 18 1996 compositions ≠ NSW Bisevac & Functional group Eneabba =+++≠ 20 Majer 1999a compositions ≠ WA Hemipterans Darling Moir et al. 2005 == ≠ 10 Range WA

- 61 -

Taxa others location age References density richness diversity evenness community Grasshoppers Andersen et al. -- ≠ Kakadu NT 12 1998 Andersen et al. Functional group -= ≠ Kakadu NT 12 2001 compositions = Coleoptera Andersen et al. +- ≠ Kakadu NT 12 1998 Koch et al. Darling == ≠* 15 2010 Range WA Termites Andersen et al. -- ≠ Kakadu NT 12 1998 Nichols & Bunn Darling - 13 1980 Range WA Majer et al. Darling == 19 2007 Range WA Insecta (various species) Majer 1978a, Darling -- ≠ 3 Majer 1981 Range WA Koch et al. Darling ++ ≠* 15 2010 Range WA Collembola Greenslade & Darling +--- ≠ 3 Majer 1980 Range WA Van Shagen Darling +- ≠ 9 1986 Range WA Greenslade & Darling + ≠ 13 Majer 1993 Range WA Koch et al. Darling =- ≠* 15 2010 Range WA Arachnida Functional group Darling Mawson 1986 =--=≠ 8 compositions = Range WA Araneae Brennan et al. Darling = ≠ 9 2003 Range WA Koch et al. Darling =+ ≠* 15 2010 Range WA Simmonds et al. Darling -+ = 18 1994 Range WA

- 62 -

Taxa others location age References density richness diversity evenness community Scorpions Grouped with Darling Majer 1981 -- Pseudoscorpionida 2 Range WA and Opiliones Nichols et al. Darling 1985, Majer et ++ 13 Range WA al. 2007 Acarina Van Shagen Darling +- 9 1986 Range WA Koch et al. Darling -- ≠ 15 2010 Range WA Cuccovia & Darling ---- 20 Kinnear 1999 Range WA Myriapoda Nichols et al. Darling -- ≠ 13 1985 Range WA Crustacea Darling Majer 1978a None found 3 Range WA Van Shagen Darling None found 9 1986 Range WA Birds Reeders 1985 - Weipa QLD ? Darling Wykes 1985 = = 7 Range WA Nichols & Darling == ≠ 8 Nichols 2003 Range WA McNee et al. Eneabba - Breeding 10 1995 WA Curry & Nichols Darling Breeding / nesting 10 1986 Range WA Nichols & Grant Darling == = 11 2007 Range WA Kabay & Darling - 14 Nichols 1980 Range WA Nichols & Darling === +/- 14 Watkins 1984 Range WA Doyle & -- Capel WA 17 Kaeding 1998 Doyle & == Breeding Capel WA 17 Kaeding 1998

- 63 -

Taxa others location age References density richness diversity evenness community Birds - continued Ward et al. Darling === +/- 20 1990 Range WA Armstrong & Darling Nichols 2000, ++/-= = ≠ 23 Range WA Nichols 1998 Brady & Noske =- ≠ Gove NT 24 2009 Reptiles Reeders 1985 - Weipa QLD ? Nichols & Darling - 8 Nichols 2003 Range WA Walker et al. - - - Body size = Pilbara WA 9 1986 Thompson & Ora Banda -- 9 Thompson 2005 WA Nichols & Darling -- +/- Body condition = 10 Bamford 1985 Range WA McNee et Eneabba -- 10 al.1995 WA Myall Lakes Bragg et al. == ≠ National 12.5 2005 Park NSW Clermont Knight 1999 - 14 QLD Kabay & Darling - 14 Nichols 1980 Range WA Twigg & Fox Tomago --= Biomass + ; breeding 16 1991 NSW Craig et al. Darling + Behaviour ≠ 17 2007 Range WA Taylor & Fox Tomago +-- ≠ Body size = 20 2001 NSW Nichols & Grant Darling -- 23 2007 Range WA

- 64 -

Taxa others location age References density richness diversity evenness community Amphibians Reeders 1985 - Weipa QLD ? McNee et Body size + (no Eneabba -- 10 al.1995 juveniles) WA Nichols & Darling - 10 Bamford 1985 Range WA Clermont Knight 1999 - 14 QLD Kabay & Darling = 14 Nichols 1980 Range WA Mammals Kentish & Anglesea - ? Boume 1982 Vic Reeders 1985 - Weipa QLD ? Blayney Clements 1998 = 4 NSW Wilson et Anglesea -- 5 al.1990 Vic Walker et al. -- Pilbara WA 9 1986 Thompson & Ora Banda += 9 Thompson 2005 WA McNee et Eneabba -- Body size = 10 al.1995 WA Tomago Simpson 1998 = 10 NSW Kabay & Nichols 1980, Darling = Breeding 10 Nichols & Grant Range WA 2007 McNee et al. Eneabba -- 10 1995 WA Clermont Knight 1999 - 14 QLD Myall Lakes Fox 1996 -= National 20 Park NSW Woodward et Stradbroke - 20 al. 2008 Is. QLD

- 65 -

Table S3: Comparison of invasive species between rehabilitated and undisturbed areas, ordered by taxon and time since rehabilitation Richness: “+” for one species indicates it is present in rehabilitated areas and not in undisturbed areas, “=” present in both; others as Table S2

Taxa species name location age re marks References density richness Formicidae Majer 1978b Cardiocondyla Darling Range 2 species most nuda WA consistently associated with rehabilitated areas Majer 1984c = Quadristruma Groote Eylandt 6 emmae NT Andersen ++Paratrechina Kakadu NT 8 1992, 1993, longicornis 1994, Andersen et al. 1998 Majer 1984c = Technomyrmex Eylandt NT 8 albipes, Gove NT Anoplolepis Weipa QLD longipes, Cardiocondyla spp Van Shagen ++Cardiocondyla Darling Range 9 1986 nuda WA Majer et al. ++Cardiocondyla Darling Range 12 seem to disappear 1984 nuda WA after 15 years Majer 1984a, +=Pheidole Stradbroke Is. 15 abundant in Majer 1985 megacephala QLD rehabilitated areas older than 6 years Collembola Majer et + Entomobrya Eneabba WA 3 al.1982 unostrigata Greenslade & ++9 species Darling Range 13 only one species Majer 1980, including WA present in undisturbed 1993 Entomobrya areas unostrigata

- 66 -

Taxa species name location age re marks References density richness Amphibians Reeders 1985 Bufo marinius Weipa NT ? (cane toad) + Bufo marinius Clermont 14 Knight 1999 (cane toad) QLD Mammals Nichols & +=house mouse Darling Range ? Bamford 1985 WA

Kentish & +=house mouse Anglesea Vic ? Boume 1982 Reeders 1985 cattle, horse, pig, Weipa ? dog, cat Clements = rabbit, fox, house Blayney NSW 4 1998 mouse, hare Wilson et +=house mouse Anglesea Vic 5 al.1990 Fox & Fox ++house mouse Hawks nest 6 1978 NSW Walker et al. +=house mouse Pilbara WA 9 1986 Thompson & house mouse Ora Banda 9 Thompson WA 2005 McNee et +=house mouse Eneabba WA 10 al.1995 Kabay & = black rat, cat, Darling Range 10 mouse density initially Nichols 1980, house mouse, WA high, none at 6 years Nichols & rabbit, fox, pig Nichols 2003 Knight 1999 + dog, cat, hare, Clermont 14 mice only were more rabbit, fox, QLD abundant in house mouse, rehabilitation black rat Fox & Fox = house mouse Myall Lakes 26 replaced by native 1984, Fox National Park after 5 years 1996, Twigg NSW et al. 1989

- 67 -

Table S4: Comparison of the number of species having recolonised rehabilitated areas from undisturbed areas for different taxa

This table does not give the richness of rehabilitated areas: instead it indicates the number of species rehabilitated areas share with undisturbed areas; in some studies, certain species have been recorded in rehabilitated but not in undisturbed sites (see text)

Taxa years since richness in number of species re fe re nce rehabilitation undisturbed having areas (number recolonised of plots) (number of plots) Formicidae 1 to 3 57 (4) 13 (12) Majer et al. 1982 3 19 (1) 4 (1) Majer 1978a 2 to 8 52 (2) 25 (6) Andersen 1993 0 to 15 46 (3) 26 (12) Majer 1984a, Majer 1985 Collembola 1 to 13 28 (3) 19 (30) Greenslade & Majer 1993 Termites 3 to 13 11 (27) 10 (2) Nichols & Bunn 1980 Amphibians 4.5 to 10 8 (2) 6 (3) Nichols & Bamford 1985 5 to 10 6 (3) 5 (2) McNee et al. 1995 4 to 14 3 * 3 (30) Kabay 1980 Reptiles 4.5 to 10 23 (2) 17 (3) Nichols & Bamford 1985 4 to 14 18 * 12 (30) Kabay & Nichols 1980 5 to 10 12 (2) 5 (3) McNee et al. 1995 2 to 12 24 * 21 * Nichols & Grant 2007 9 22 (1) 10 (1) Walker et al. 1986 Birds 6 to 23 34 (2) 25 (3) Armstrong & Nichols 2000 4 to 14 67 * 50 (30) Kabay & Nichols 1980 5 to 10 16 (3) 13 (2) McNee et al. 1995 2 to 32 70 * 66 * Nichols 1998 Mammals 0 to 10 10 * 10 * Nichols & Grant 2007 * unspecified in publication

- 68 -

Chapter 3

Persistence and detectability of faecal pellets in different environments: Implications for pellet-based census of fauna

Abstract

Knowing the distribution of a species and its population trends are the basic requirements for determining its conservation status and appropriate management strategies. Yet, acquiring this fundamental information is often difficult, particularly for cryptic species. Indirect survey methods that rely on signs of recent activity, such as presence of faecal pellets (scats), can overcome some of these difficulties but are not without their own challenges. In particular, variation in scat detectability and decay rate across vegetation communities can lead to incorrect interpretations of habitat use. Here we studied how vegetation communities affect the detectability and decay rate of scats from the koala Phascolarctos cinereus, a cryptic marsupial of great conservation significance. We first calculated the percentage recovery of a known number of scats across different ground layer complexities. Detectability of scats was highly dependent on ground layer complexity, introducing up to 16% non-detection bias. The consistency of this bias suggests that a correction factor could be developed. We then examined scat decay rates across different vegetation communities. Scat decay rates were highly heterogeneous inside vegetation communities. On the basis of a priori generalised linear mixed models, exposure of scats to surface water (flooding) most strongly accelerated scat decay rate. The occurrence of flooding was associated with the location of scats in swamps. Finally, we experimentally compared decay rates of scats protected and unprotected against invertebrate attack. Attack of scats by invertebrates strongly accelerated scat decay rate, but was found to be unpredictable - which may explain the high variability of scat decay rate within a single vegetation community. Increasing the sampling effort in each vegetation community and developing a correction factor for vegetation communities subject to flooding could decrease decay bias of scat surveys. The combined effect of detectability and decay biases, both stronger in certain vegetation communities such as those found in swamps, has the potential to introduce

- 69 - substantial errors in scat surveys. These could in turn lead to the adoption of erroneous management decisions.

Key words: indirect sign survey, scat, scat survey, faecal pellets, detectability, decay, bias, koala

Introduction

Knowledge of species abundance and distribution represents the foundation of conservation and management decisions (IUCN 2001; MacKenzie 2005). However, acquiring such information requires a far from trivial effort (Royle & Nichols 2003; Tyre et al. 2003). This is particularly true for cryptic and elusive animals (Kéry 2002; Piggott & Taylor 2003), where indirect methods are often needed (Wilson & Delahay 2001). These indirect methods include, but are not limited to, remote photo- identification (Carbone et al. 2001) and sign surveys [e.g. faecal pellets (Harestad & Bunnell 1987; Barnes 1996), tracks (Stanley & Bart 1991; Kendall et al. 1992) and nests (Newton et al. 1977; Walsh & White 2005)].

Scat (faecal pellet) survey is one of the earliest and most widely used indirect methods employed to gain information about species’ distribution and sometimes abundance (Bennett et al. 1940; Putman 1984). The documented variability of both scat detectability and decay rate has however led to concerns regarding the reliability of such surveys in estimating occurrence and especially abundance (Block et al. 2001). Scat detectability can vary across observers (Van Etten & Bennett 1965; Neff 1968) and ground layer complexity (Pahl 1996; Buij et al. 2007). While the variability in scat detectability between observers can be minimised by developing personal correction factors or standardised by using the same observer (Neff 1968), variation of detectability with ground layer complexity is less well understood and catered for. As for scat decay rate, variation has been found across seasons (Murray et al. 2005), habitat types (Dzieciolowski 1976; Prugh & Krebs 2004; Kuehl et al. 2007) and diets (Van Etten & Bennett 1965; White 1995). Notably, the effects of season and habitat can be mediated by the influence of moisture (Harestad & Bunnell 1987), litter accumulation (Eaton 1993, cited in Prugh and Krebs, 2004), temperature (Hone & Martin 1998), rainfall (Wallmo et al. 1962) and detritivore attack (Neff 1968; Masunga et al. 2006), - 70 - which will ultimately influence scat decay rate. While impacts of seasonal variability on scat decay rate have been thoroughly described for a variety of species (Wigley & Johnson 1981; White 1995; Barnes et al. 1997; Hone & Martin 1998; Massei et al. 1998; Nchanji & Plumptre 2001; Murray et al. 2005), fewer studies have quantified the variability of scat decay rate across habitat types. The extent to which environmental factors can influence scat detectability and decay rate is not usually resolved, yet more often than not, some degree of environmental heterogeneity is present in study sites (Wilson & Delahay 2001). Since researchers cannot control the occurrence of such environmental heterogeneity, it is critical to quantify and account for the variability in scat detectability and decay rate across environments.

Koalas, Phascolarctos cinereus, are good candidates to investigate whether, and how, different environmental factors influence scat detectability and decay rate. Koalas use a variety of environments (vegetation communities, exposure, soil types, etc.) and are difficult to survey directly because of their cryptic nocturnal habits and their low population density. Scat surveys have thus been the widely used method to study koala distribution (e.g., Lunney et al. 1999; Sullivan et al. 2003b; Rhodes et al. 2006; McAlpine et al. 2008), habitat use (Phillips et al. 1995; Lunney et al. 2000; Sullivan et al. 2003a), and abundance (Sullivan et al. 2004). Moreover, for koalas, variation in the rate of scat decay due to changes in diet composition should be negligible, allowing us to focus on other sources of variability. Indeed, koalas are folivores with a relatively homogeneous diet all year round, relying mainly on a few genera in the family Myrtaceae (i.e., Eucalyptus, Corymbia, Melaleuca, Lophostemon, Martin & Handasyde 1999). This contrasts with the situation in other species, where diet can be composed of very different food items which induced a variability in scat decay rates (e.g., the diet was constituted of 100% of corn, or 100% of commercial pellets or 100% of soy bean in Cochran & Stains 1961).

Recent studies have demonstrated the need to incorporate information about how koala scat decay rate (Rhodes et al. 2011) and detectability (Sullivan et al. 2004) influence scat surveys. While Rhodes et al. (2011) showed that seasonal and geographical differences in climate significantly influence the rate of scat decay rate, they also emphasised that most of the variation in scat decay rates remained unexplained and

- 71 - called for more research. Both detectability and decay biases usually result in an underestimate of scat density. If detectability and decay rate are consistently different across different environments, the importance of some specific habitats for koalas might be overlooked. It will also be problematic if detectability and decay rates are inconsistent between similar environments in different locations, or inconsistent overall. All these scenarios could result in inappropriate management decisions being taken. This highlights the importance of understanding the basis of variation in koala scat detectability and decay rate across environments.

Robust scat survey methodology for ecological studies of koalas is particularly critical because management decisions taken at the scale of local governments have often relied on such surveys (Lunney et al. 1999; Lunney et al. 2000). These decisions affect localised extinction risk (Preece 2007) and the fate of a species is often determined by the sum of these local extinctions (Caughley 1994). For the koala and many other species, long-term survival depends on their protection outside reserves. This is especially the case for koalas, due to their dependence on riparian and/or coastal ecosystems that are not well represented in reserves owing to competing land use priorities (Lunney & Matthews 1997). This gives local government management decisions a key role in species conservation (Lunney et al. 1998) and decision-makers must, therefore, be confident that koala survey methodologies underpinning their decisions are reliable.

In this study, we investigated the effect of different environments on detectability and decay rate of koala scats in three experiments. We first studied the influence of ground layer complexity on scat detectability. Next, we analysed scat decay rates in different vegetation communities available to koalas and the relative influence on decay rate of local environmental variables identified from the literature. Lastly we compared decay rates between scats protected or unprotected from the decaying action of invertebrates. We discuss whether biases arising from variations in detectability and decay rate in different environments have the potential for introducing errors large enough to influence management decisions. We also examined how studies based on scat surveys may be refined by the integration of variability in scat detectability and decay rate.

- 72 -

Materials and Methods

The field work was conducted on North Stradbroke Island (NSI), Queensland, Australia (27º23’/27º45’S, 153º23’/153º33’E). This is an approximately 27,500ha island, where koalas occur in a mosaic of vegetation communities (classification based on regional ecosystems or REs identified by the Queensland Herbarium 2009). Six remnant REs were selected on the basis of their use by koalas on NSI (see Appendix A): RE 12.2.10, mallee Eucalyptus spp. low woodland; RE 12.2.15, swamps; RE 12.2.5, Corymbia spp. open to low closed forest; RE 12.2.6, E. racemosa woodland; RE 12.2.7, Melaleuca quinquenervia open forest to woodland; and RE 12.2.8, E. pilularis open forest. In addition, two disturbed vegetation communities (mine rehabilitation) were differentiated: one characterised by a complex litter layer, the other by a simple litter layer (see full description below and Figure 1). We conducted three experiments as described below.

Experiment 1: Faecal pellet detectability in litter layers of different complexities

To measure variation in the detectability of scats with environmental variables, 30 plots (1x5m) were established in vegetation communities of varying ground layer complexity. One researcher hid a random number of scats (1-22) in each plot. We used old scats collected from the field, as fresh ones have a more conspicuous colour and patina that makes them easier to find. This fresh condition lasts only a few days, consequently it does not characterise most scats naturally found during surveys (RC, personal observation). A second researcher (RC), who conducted all searches to standardise observer bias (Neff 1968), searched the plot without prior knowledge of the number of scats in the plot. The search had no time limit and ended when the second researcher was confident she had thoroughly searched the plot. For each plot, we recorded the percentage of scats found, the time needed to find each scat and the total time taken to search the plot to achieve this level of confidence.

Plots were established in vegetation communities classified in three groups on the basis of ground layer complexity: (1) the simple litter group (N=10) which had a flat litter layer composed of Allocasuarina needles, with little to no plants or woody debris (<5%) and varying amounts of bare ground (0 to 20%); (2) the complex litter group (N=10)

- 73 - which was composed of a three-dimensional litter of leaves and bark, no bare ground and some plants and woody debris (between 20% and 90%, Figure 1); and (3) the highly complex litter group (N=10), in which the substrate was mostly covered with plants and woody debris (>90%). For the simple and complex litter groups, all plots were searched for scats beforehand to ensure no scat was present prior to the experiment. For the highly complex litter group, however, pre-searching the plots would have disturbed the plots to an extent that would have compromised the experiment; thus they were not pre-searched. In order to minimise unwanted presence of scats prior to our experiment in the highly complex plots, we located the plots outside the known koala distribution on the island. Each of the three litter groups was replicated at two locations (five plots at each location). At each location, plots were placed 50m apart from one another.

(a) (b)

Figure 1: Example of two sites showing plot, cap and bag, on two different ground layer types (a) complex three dimensional litter, (b) simple A. littoralis litter

Experiment 2: Scat decay rate in different vegetation communities

Fresh scats were placed in different vegetation communities and their decay rate was recorded. We used a total of 1,980 fresh scats collected from the Australian Wildlife Hospital from 15 females and 31 males aged from 1 to 10 years. Koalas had been in the hospital for less than 2.5 months and none had received treatments that could alter scat decay rate (e.g. worm treatment). Hospital cages are cleaned daily so all scats used in this experiment were less than 24 hours old. Scats from all 46 koalas were mixed and ten scats were randomly selected to form a group. Each scat group was weighed and groups >10g or <6g were discarded to ensure homogeneity of groups. Scats were kept in

- 74 - a refrigerator overnight.

Scat groups were placed in each of the eight different vegetation communities previously described, with three replicates per vegetation community (24 locations in total). Replicates were several kilometres apart (mean=6.6km, SD=3.9). At each location, two pseudo replicates (50m apart) were laid. This was conducted on 14 and 15 February 2010. Each scat group (N=48) was placed on 10x10cm plots on the ground directly below the canopy of potential koala fodder or roosting trees of genera Eucalyptus, Corymbia, Melaleuca or Lophostemon.

During the first 48 hours after scat placement, NSI was struck by extraordinarily heavy rainfall (Withey 2010). Sites were checked the next day and some scats had already disappeared. It was feared that the rain had either soaked and disintegrated, or washed away some scats. Consequently, 10 randomly selected new scats were added in an additional plot next to each of the initial 48. Hereafter, the initial plots will be referred to as rained-plots and the ones deployed just after the rain will be simply referred to as plots. Rainfall events out of the normal range were not recorded again during the rest of the experiment.

Scat locations were visited once a week. We recorded the number and condition of remaining scats. The condition of scats was described as either (1) intact: the scat was complete; (2) surface eaten: the scat presented a rough surface; (3) partly eaten: parts of the scat were missing; (4) half eaten: at least half of the scat had disappeared; (5) fibrous: the inner matrix had disappeared, only fibres remained visible, but the scat shape remained present; and (6) mass of fibre: scats constituted of a shapeless mass of fibre. Rare states of scats were: (7) scats melted to a shapeless entity or (8) partially buried by invertebrates. When the weekly observations indicated decay rate had slowed down, scats were checked every two weeks, then less often. The survey lasted 36 weeks in total, by which time 50% of the scats had disappeared. Again, counts and classifications of the condition of scats were conducted by a single researcher to standardise observer bias (Neff 1968).

Variables characterising the local environment in which the scats were observed were

- 75 - also recorded at each visit. Scat moisture (wet or dry), accumulation of litter fallen on top of scats (presence/absence) and the activity of detritivores (defined here as the presence/absence of invertebrates or fungal-type organisms, Figure 2) on scats were recorded as a binary score. At the end of the experiment the results were averaged. Site elevations were extracted using Terramodel Version 10.61 from a 2008 airborne laser scan of the island (Sibelco/CRL, unpublished data). A final environmental variable was flooded. The flooded variable indicated, for individual plots, the occurrence of surface water in the vicinity of scats at any time during the experiment.

(a) (b)

Figure 2: Examples of scats found with (a) fungal bloom and (b) evidence of invertebrate attack (invertebrates were seen but are not visible in pictures)

Although, as indicated above, in some species scat decay rate can vary with diet, we did not expect that scats from hospitalised koalas would decay significantly differently from scats of NSI island koalas. Indeed, the food items consumed were closely related, even if differences in the particular tree species eaten occurred (Table 1). However, we tested this hypothesis by comparing the decay rate of scats from hospitalised koalas and wild koalas from NSI. Scats from wild koalas were collected from two females. Groups of ten randomly selected scats were deposited besides plots representing six out of the eight vegetation communities previously described. We were not able to collect sufficient fresh scats from wild koalas to include all eight communities.

- 76 -

Table 1: Eucalypt species fed to hospitalised koalas between 1 and 6 days before their scats were collected in comparison to species found in the diet of wild koalas on NSI (only E. tereticornis, E. robusta and E. resinifera are present on NSI, their percentage is taken from Chapter 5)

4% 12% 11% wild koalas % eaten NSI by 1999) et al. 5% 5% 4% 2% 1% 1% 1% 41% 23% 17% % fed between 6 days last 1 to E. tereticornis tereticornis E. E. dunnii E. robusta E. propinqua, E. punctata,E. propinqua, E. major camaldulensis E. E. grandis E. resinifera crebra E. E. deanei maculata E. Blue gum gum Blue White gum Swamp mahogany Common nameCommon Grey gum scientific name River red gum gum Flooded Red stringybark Ironbark Mountain Blue gum Spotted gum NB: food takes between 1and 6 days to appear in the scats of koalas (Ellis

- 77 -

Experiment 3: Variability of scat decay rate in relation to invertebrates

To investigate the effect of invertebrates on scat decay rate, we partially or totally protected some scats from their influence. Next to the 48 plots described earlier, two groups of 10 random scats were added (Figure 1). One group of 10 scats was placed on the ground, covered with a fly-screen secured into the ground to protect the scats from ground-dwelling and flying invertebrates. These groups are hereafter referred to as caps. The second group of 10 scats was placed into a sealed fly-screen bag and placed directly on the litter. This protected the scats from any invertebrate damage (i.e. no invertebrate was observed inside the bags during any visits, while some invertebrates were recorded on the scats in the plots, see Experiment 2). These groups are hereafter referred to as bags. All the scats in caps and bags were deposited prior to the heavy rainfall event. Scats in caps and bags were checked at Week 1 and Week 12 and the number of scats remaining and their condition were compared to the unprotected scats in the rained- plots and plots. Scats in caps and bags were observed wet and thus the treatments do not seem to have protected the scats from the rain.

Data analysis

All variables were tested for normality and homoscedasticity (Levene’s test) and appropriate parametric or non-parametric tests were performed in PAWS Statistics 18.0 (IBM 2009). Significance was taken to be p<0.05 (except when accounting for Bonferroni’s adjustment), standard errors of mean (SEM) and standard deviations (SD) are given when appropriate (Altman & Bland 2005).

Relationships between response and explanatory variables were investigated using a generalised linear mixed model (GLMM), with the 48 sites as a random effect. Rainfall is known to increase scat decay rate (Wallmo et al. 1962; Barnes et al. 1997), but since we were primarily interested in other potential factors in this study, we included rainfall as a random effect. Data were fitted with a quasi-Poisson distribution (Zuur et al. 2009), the standard distribution for count data with small over-dispersion (c=1.48).

For model selection, we used a priori models to avoid data dredging (Anderson et al. 2001a; Burnham & Anderson 2002) and to reduce the possible selection of noise

- 78 - variables (Flack & Chang 1987). Explanatory variables were chosen based on the literature quoted in the introduction and more koala-specific literature (Achurch 1989; Common & Horak 1994; Melzer et al. 1994; Jurskis & Potter 1997; Curtin et al. 2002; New 2004). Explanatory variables were vegetation communities and four local environmental variables: detritivores, litter, elevation and flooded (scat moisture was measured but not included, as it was correlated with other explanatory variables). We constructed a priori models based on each variable separately, then on the four possible combinations of vegetation communities with each environmental variable. These models were chosen to investigate the main variable of interest (vegetation communities) and to see if any of the environmental variables had an effect on its own or in combination with vegetation communities.

The GLMM models were constructed, compared and validated using R 12.10.0 (R Development Core Team 2010). Data were graphically analysed for skewed explanatory variables and none was observed. Explanatory variables were standardised to allow comparisons of model parameter estimates (Quinn & Keough 2006). Collinearity was tested with a variance inflation factor (VIF). The GLMMs were created using “glmmML” (Broström 2003) and “lme4” (Pinheiro et al. 2010). We estimated the global goodness of fit between the global model and the null model (with random effects only) using a likelihood ratio test (Mundry 2011), which is preferred to a Wald test in the context of small sample sizes (Pawitan 2001). The global models were validated by visual inspection of residuals as in Rhodes et al. (2009). Spatial correlations in the residuals were analysed with R-package “ncf” (Bjornstad 2009). Models were ranked on the basis of QAICc, the Akaike’s information criterion (Akaike 1973) corrected for small sample size and over-dispersion (Hurvich & Tsai 1989, 1995).

We also fitted our data with an interval-censored survival model (Peto 1973), with a random effect (more often referred to as a frailty effect in survival analysis, Bellamy et al. 2004; Goethals et al. 2009). While we acknowledge that survival models are more appropriate for our data, assessing the goodness-of-fit in interval censored survival models with frailty effects is still an area of active and somewhat controversial statistical research (Gomez et al. 2009). We found that both the GLMM and the survival analyses provided concordant results. As a consequence, we present only the GLMM

- 79 - results.

Results

Experiment 1: Faecal pellet detectability in litter layers of different complexities

The percentage of scats found decreased with the increase in litter complexity (Kruskal- Wallis=14.85, df=2, p=0.001). It varied from 100% in simple litter to 97.7% (SEM=1.1) in complex litter and down to 83.9% (SEM=4.4) in highly complex litter (Table 2). The total time taken to search the plot increased with litter complexity (Kruskal- Wallis=24.56, df=2, p<0.001). Over the duration of the experiment it took an average of 3.4 minutes (SD=1.2) to search plots in simple litter, 15.8 minutes (SD=6.6) in complex litter and 34.3 minutes (SD=10.0) in highly complex litter. The mean time taken to find a scat increased with litter complexity (Kruskal-Wallis=16.29, df=2, p<0.001), while the mean percentage of scats found at 2 minutes decreased with litter complexity (Kruskal- Wallis=23.06, df=2, p<0.001). However, the mean time to find the first scat was similar across litter complexity (Kruskal-Wallis=0.01, df=2, p=0.995).

Table 2: Percentage of scats detected and time needed to detect them, in three different ground layer complexities (SEM: standard error of mean)

Characteristics mean time of search (min) mean mean mean time of search percentage of time percentage to find the scats found mean minimum maximum taken by found at 1st scat (± SEM) scat (sec) 2min (sec) Simple litter 100.0 3.4 2.1 6.2 0.73 87.3 22 Complex litter 97.7 (±1.1) 15.8 7.1 25.2 1.05 30.3 10.1 Highly complex 83.9 (±4.4) 34.3 15.4 48.6 2.61 11.2 17.5 litter

Experiment 2: Scat decay rate in different vegetation communities

Before performing any analysis, we ensured that characteristics of our experimental design did not compromise the results. We first confirmed that the average weight of each group of ten scats was not different between vegetation communities (ANOVA, df=47; rained-plots: F=1.007, p=0.441, plots: F=2.107, p=0.065). We also confirmed that the origin of scats (wild/hospitalised koalas) had no significant effect on scat decay

- 80 - rate by comparing a model with the origin of scats as an explanatory variable to a null model including only the random effects (likelihood ratio test, χ2=0.24, p=0.61).

Survival curves showed that scats placed in swamps (RE 12.2.15) decayed faster than scats placed in any other vegetation community investigated (Figure 3a). Rain also accelerated scat decay rate (Figure 3b).

(a)

12.2.6 rehab complex 12.2.5 12.2.10 rehab simple 12.2.8 12.2.7

12.2.15 (swamps)

NB: Vegetation communities (12.2.10: mallee Eucalyptus spp. low woodland, 12.2.15: swamps, 12.2.5: Corymbia spp. open to low closed forest, 12.2.6: E. racemosa woodland, 12.2.7: Melaleuca quinquenervia open forest to woodland, and 12.2.8: E. pilularis open forest) follow Queensland Herbarium (2009) except for the addition of simple and complex rehabilitation

- 81 -

(b)

Figure 3: Survival curves of scats, displayed (a) by vegetation community and (b) by rainfall occurrence

Within each vegetation community, the number of scats remaining after 36 weeks was highly variable (every possibility between zero and ten scats, Figure 4a). When averaged across the 36 weeks duration of the experiment, all vegetation communities had a median of between seven and nine remaining scats, except RE 12.2.15 (swamps) which had a median of five (Figure 4b).

(a)

(b)

- 82 -

(b)

Figure 4: Boxplots (minimum, first quartile, median, third quartile, and maximum, with outliers ° and extreme values *) showing the variability of the average number of scats remaining by vegetation community (a) after 36 weeks (b) all 36 weeks combined (grey lines in (b) include max and min medians of all but swamps)

Boxplots are ordered from the vegetation community with the fewest remaining scats (12.2.15) to the one with most remaining scats (12.2.6)

Our global GLMM model at week 36 fitted the data significantly better than a model with just the random effects (likelihood ratio test, χ2=57.43, p=2.77e-08). Residuals plotted against each explanatory variable showed no patterns. No spatial correlation was found using the spline correlograms of the data or the residuals. When we compared all models, only one model was strongly supported by the data (evidence ratio between this

- 83 - model and the second best model=642, Table 3). This final model contained the variable flooded. Interestingly, it did not contain vegetation communities. The variable flooded increased scat decay rate, i.e. it decreased the number of remaining scats (β=-2.04 SE=0.59). It is noteworthy that most plots in swamps (RE 12.2.15) had been consistently and selectively flooded (all but one of the swamp plots was flooded, whereas only one plot outside swamps was flooded, in RE 12.2.7).

Table 3: Model selection for explaining scat decay rate after 36 weeks, all models were fitted with GLMM with the 48 sites and the rainfall event as random effects (N=102, c=1.48) K: number of parameters, QAICc: Akaike information criterion corrected for small sample size and over-dispersion, ΔQAICc: QAICc differences, see descriptions in text

Experiment 3: Variability of scat decay rate in relation to invertebrates

After one week, the number of scats remaining in the rained-plots was similar to the number of scats protected from invertebrates by caps placed above them (Mann- Whitney test=932.0, p=0.077); both were lower than the number of scats protected in bags (see Table 4 for details). After 12 weeks, rained-plots contained significantly fewer scats than caps, which in turn contained fewer scats than bags (see Table 4 for values and p-values).

- 84 -

Table 4: Number of scats remaining at Week 1 and Week 12 in the different treatments (N=48 per treatment, see text), and statistical differences between treatments (p values and Mann-Whitney U test statistics, Bonferroni’s adjustment α=0.004) min= minimum number of scats, max=maximum number of scats

mean SD min max rained-plot cap bag p p p U U U value value value plot 9.85 0.46 8 10 612.5 <0.001 845.5 0.002 1078.0 0.229 rained-plot 8.17 2.43 0 10 932.0 0.077 557.0 <0.001 cap 8.75 2.37 2 10 781.0 <0.001 Week 1 Week bag 9.96 0.2 9 10 plot 7.63 2.66 0 10 680.0 <0.001 1055.0 0.467 438.0 <0.001 rained-plot 5.94 2.75 0 10 643.0 <0.001 93.5 <0.001 cap 7.67 3.12 0 10 575.5 <0.001 Week 12 Week bag 9.75 0.93 5 10

The condition of the scats also varied with the treatment (Kruskal-Wallis tests, see Table 5). The best preserved scats were the ones in the bags, while the worst were the rained-plots (see percentages of intact scats at Week 1 and partly eaten at Week 12 in Table 5). The comparisons with scats in plots (which were the only groups of scats deposited after the rain) are also given, although these results are less relevant as the scats had a different history. Overall, the scats protected from all invertebrate activity in bags were the best preserved in terms of both quantity and quality, while the rained- plots were the worst preserved on both accounts.

Discussion

Scat detectability and bias

This study has demonstrated that scat detectability varies with ground layer complexity, with up to 16% variation in the proportion of scats detected between plots of different ground layer complexities. Importantly, this 16% figure might be an underestimate of the detectability bias in many scat surveys.

- 85 -

Table 5: Condition of scats at Week 1 and Week 12 in the different treatments Scat number: in bold, results are significantly different across treatments, Kruskal- Wallis tests (df=3): **: p≤0.001, *: p<0.01

0 0 4 1 fibrous melted Week 12 Week half half eaten part part eaten 220** 114**165** 25** 118**241** 19** 111**454** 3** 23** 1** 07 50 19 10 melted buried fibre mass of of mass part part eaten Week 1 Week eaten surface surface intact fibrous 270** 107**132** 129** 32*217** 44*272** 72** 63** 183** 113** 27* 1* 6* 18* 93** 14** 4* 2*

Treatments plot rained-plots caps bags test statisticsp values 32.11plot 17.62rained-plots <0.001 14.32caps 0.001bags 29.9 43.10 0.002 56.3 14.39 29.3 <0.001 0.002 6.20 22.3 51.3 10.0 0.102 56.9 7.50 6.7 17.0 25.6 0.057 88.65 38.3 <0.001 13.1 6.4 4.1 44.90 <0.001 1.3 0.001 16.51 0.2 22.0 1.1 0.570 2.01 2.9 0.9 0 0.0 0.4 0.2 54.6 1.5 0.2 39.1 2.1 61.3 0.0 6.3 67.1 31.8 94.8 30.9 7.0 0 4.8 0.8 0 0.2 1.1 0.2

scat number scat percentage

- 86 -

Firstly we searched the plots for a long period of time, whereas many scat surveys have to be performed within the limits of time and budget constraints. The amount of time we dedicated to some plots (up to 48 minutes) would most likely be incompatible with such constraints, thus more scats could be missed in routine scat surveys. Further, we did not impose a maximum time limit for scat searches, allowing plots with more complex ground layers to be searched for longer. With a constant amount of time of search per plot, we suspect that proportionally more scats would be missed in complex ground layer. Nonetheless, many studies relying on scat surveys for koalas use the Spot Assessment Technique or SAT (Phillips et al. 1995; Phillips et al. 2000), where the search for scat lasts a maximum of two person minutes (or until the first scat is found). This time constraint can decrease the probability of finding scats in more complex ground layers and lead to underestimating scat density in vegetation communities characterised by complex ground layer (Table 2).

The limitations of the SAT might be lessened by the fact that in SAT, only the first scat matters because the search is terminated when the first scat is found. In all the plots in our study, the first scat was found within the first two minutes of search and the time to find the first scat was independent of the litter complexity. These are positive insights for the validity of SAT (Table 2). Nonetheless, we suspect the time of search to the first scat would be increased in many real-world situations, compared to our survey sites where scats were deposited just before the search (for example, no litter had fallen above them). Maybe even more importantly, since a major aim of our experiment was to investigate the percentage recovery of scats, the number of scats per plot was higher than would be found naturally in many instances. From the data collected in 36 plots with koala scats on the island (Chapter 5), the average natural scat density was 1,098 per hectare, while in the plots used in this chapter, the (artificial) density was 27,333 per hectare. This means the time to find one scat could potentially be around 25 times longer in natural conditions than in our experiment; hence it will often require much longer than two minutes to find one scat in natural conditions. The use of consistent sampling techniques for koala scat surveys is critical to any attempt to evaluate the koala habitat values of sites with rigorous statistical analysis – some elements of the approach described in SAT are useful in the context of developing a consistent sampling method. However, given the potential detectability bias found in our study, we propose

- 87 - that either the space (plots of a fixed size) or the number of trees for scat searches be constrained in exchange for relaxation of the time constraint. This may help reduce the bias due to the variability of scat detectability between ground layers of different complexity.

We cannot be complacent about a bias of the size we found in detectability. Even slight variations in detection probability across different vegetation communities can seriously skew study results (Gu & Swihart 2004; MacKenzie 2005). A simulation study modelling species occurrence as a function of habitat covariates, described the errors resulting from imperfect detections (Gu & Swihart 2004). A detectability bias of 15 to 20% (as found in our study) resulted in up to 200% relative-bias when estimating parameters (i.e. an over or under estimation of 200% of the effect of a habitat variable on a species’ occupancy). Apart from Sullivan et al. (2004), few studies using koala scat surveys have measured or accounted for variation in scat detectability. Similarly to the Sullivan et al. (2004) study, we found that scat detectability varied consistently across different ground layer complexities. This consistency in variability could allow for the measurement and incorporation of a correction factor. While it may not be logistically feasible for each project to develop its specific correction factor, presence/absence models could be weighted by ground layer complexity based on available studies (e.g., in our study, the more complex ground layers reduced detectability by 16%, see also Sullivan et al. 2004).

Alternative methods have been developed to deal with imperfect detection using zero- inflated processes (MacKenzie et al. 2002; Tyre et al. 2003; Stauffer et al. 2004; Wintle et al. 2004). While these methods are based on direct animal surveys, they would also be applicable to indirect animal surveys. Some of these approaches even allow the probability of detection to vary with characteristics of the vegetation communities being surveyed (MacKenzie et al. 2002). Such methods, however, rely on repeated surveys over a short time (MacKenzie 2005; Wintle et al. 2005). This may not be appropriate for koala scat surveys, as koalas often occur at low density. Thus, in many instances, koala scat surveys repeated over intervals of a few weeks would not record new scats. As suggested by MacKenzie (2005), repeated samples could be collected by multiple observers carrying searches on the same plot and comparing scat detection. This would

- 88 - however introduce observer variability (Neff 1968). Despite potential shortcomings of these approaches for koalas, they still might be relevant for other species.

Scat decay rate and bias

In comparison to scat detectability, variability in decay rates was not consistently associated with variation in vegetation communities (see Figure 4). For instance, after 36 weeks extreme outcomes were seen for two plots inside the same vegetation community despite those plots being only 50m apart: one had 100% scats remaining and the other had 0%. Similar results were also found in a study by Rhodes et al. (2011), where a high and unexplained proportion of decay rate variability resulted from the variation between plots found within a site. In the current study, it is worth noting that the treatments preventing invertebrates to access the scats greatly influenced scat decay rate. Lepidopteron larvae develop in koala scats, while adult coleopteran exploit koala scats (Common & Horak 1994; Melzer et al. 1994; New 2004); all increasing the rate of scat decay. The fine-scale variation of invertebrate densities and the stochastic chances of invertebrates finding scats could explain the small-scale heterogeneity of scat decay rate. Whatever the reasons, the lack of consistency in scat decay rate makes the development of a correction factor difficult. One way of compensating for variability in scat decay rate could be to increase the intensity of sampling within each vegetation community in order to average fine-scale heterogeneity (Lunney et al. 1998; Rhodes et al. 2011).

One exception to the inconsistency in decay rate relates to the vegetation community associated with swamps. In swamps, the decay rate was consistently faster than that observed in other vegetation communities (Figure 3a and 5). Plots in swamps were consistently and selectively flooded, and flooding was found to accelerate scat decay rate (Table 3). This suggests that strong biases could be introduced when landscapes with surface water present (or probably even very moist substrates) are being surveyed. Koalas actually tend to favour wet habitats (Gordon et al. 1988; Munks et al. 1996; Cork et al. 2000; Lunney et al. 2000; Sullivan et al. 2003a) and so there is the paradoxical likelihood of a low density of detectable scats in areas supporting a relatively high density of koalas; however, few studies have attempted to control for decay bias in koala scat surveys. Even these studies have tried to account for scat decay - 89 - by excluding scats older than a certain threshold (Hasegawa 1995; Sullivan et al. 2003a; Sullivan et al. 2004). Scat aging, however, has been found to be unreliable (Prugh & Krebs 2004), or at best limited to highly skilled and experienced researchers (Sullivan et al. 2002). In other species, more objective criteria have been developed. For instance, gorilla Gorilla gorilla dung age has been reliably correlated with dung pile height (Kuehl et al. 2007). However, such a reliable and objective criterion to age koala scats is not yet available. Alternatively, developing a correction factor by measuring scat decay rate in sites where an accelerated decay rate can be expected might be achievable (see an example with deer in Brodie 2006).

Another method proposed to limit decay bias is to clear plots yearly (Prugh & Krebs 2004). However, we observed substantial decay heterogeneity over a period much shorter than one year. On the basis of our results, clearing plots five weeks prior to the survey could remove bias arising from heterogeneous decay rates. This method has been used in a study of macropods, where plots were cleared one month before the survey (Johnson & Jarman 1987). Again, however, owing to the typical low density of koala populations, clearing sites one month prior to surveys might result in scats not being deposited in the interval. In any case, researchers have warned that it might not be possible to remove 100% of the scats when clearing plots, resulting in further errors (Stanley & Royle 2005, but see avenues for potential solutions in Rhodes et al. 2010).

While decay bias may be problematic for scat counts, it may not affect presence/absence data to the same extent (Lunney et al. 1998). One argument is that many species, such as koalas, show long-term fidelity to their home range and would deposit scats frequently in the same spots (Lunney et al. 1998). More recent research, however, highlighted that koalas’ site fidelity not only tends to be short-term but that they also infrequently revisited the same tree (Ellis et al. 2009), thus reopening the debate.

Worth noting is the high impact of the substantial rain event on scat decay (Figure 3b). The high level of scats decay we observed means that surveying a site for scats following an important rain event (and before new scats have been deposited) has the potential to under-estimate scat number and thus habitat use.

- 90 -

Combination of detectability and decay rate variability

We compared the percentages of scats that would be found in a survey after accounting for detectability and decay biases between two vegetation communities (Table 6). On the one hand, swamps have the worst detectability owing to very dense ground vegetation and the quickest decay rate owing to their proneness to flooding. On the other hand, rehabilitated vegetation communities with simple litter recorded the best detectability and a decay rate not significantly different from average (see above). For the same density of scats deposited in swamps and in rehabilitated areas with simple litter ground layers, the proportion of scats found in a survey would vary widely (Table 6). In the worst case after 9 months, a survey would detect 8% of the total original scats in swamps against 42% in rehabilitated vegetation. In this example, if the actual koala utilisation rate of both areas were equal, swamps could be wrongly classified as five times less used by koalas.

Table 6: Comparison of the percentage of predicted scat recovery in two extreme vegetation communities based on observed detectability and decay rates

% % scats remaining due to % scats found due to both detected decay biases 3 months 9 months 3 months 9 months Habitat no rain rain no rain rain no rain rain no rain rain Swamp 83.9 40.0 40.0 18.3 10.0 33.6 33.6 15.4 8.4 Simple litter 100.0 73.3 50.0 68.3 42.9 73.3 50.0 68.3 42.9 rehab

Conclusion

Overall, our study demonstrates the potential biases inherent to using indirect signs of a species’ presence for assessing the species’ distribution, let alone its abundance. Scat detectability bias could be controlled because of its consistency, thus it should be accounted for in the future. Scat decay bias acts in more complex ways and may also be an even more problematic bias for koala studies. Indeed, scat decay rate is higher in wet habitats, which are also favoured by koalas. Thus, decay bias might selectively decrease the estimated value of prime koala habitat when the assessment of this value relies on scat surveys. Determining exactly under which conditions scat decay introduces a bias - 91 - sufficient to warrant additional corrections (e.g. based on scat decay trials) is still an area of active research (Rhodes et al. 2011).

The magnitude of potential detectability and decay biases combined could seriously impede scat survey reliability. These biases can occur even on a relatively small scale because most study areas are heterogeneous. Such biases associated with particular habitats are likely to introduce errors in scat surveys which in turn could lead to inappropriate management decisions.

Acknowledgement

We wish to thank Russell Miller for continuous assistance in the field, Olivia Woosnam-Merchez for commenting on earlier drafts and the Lone Pine Koala Sanctuary and the Australian Wildlife Hospital for providing the fresh koala scats. The first author is supported by an Endeavour Europe Award and an Endeavour International Postgraduate Research Scholarship. Sibelco Australia – Mineral sand provided ongoing support for this research through the provision of logistical support and access to company sites.

- 92 -

Chapter 4

Is restoring flora restoring fauna? Developing fauna criteria for assessing restoration success

Abstract

Restoration of degraded and disturbed landscapes has become increasingly necessary to curb net habitat loss. Success of restoration projects is often evaluated based on abiotic and flora criteria. While fauna is an important part of restored habitats, it is typically not monitored. Instead, a common assumption is that if flora recovers, fauna will return. We tested this paradigm by investigating the recolonisation patterns of koalas (Phascolarctos cinereus) in rehabilitated mine sites. Overall, we found that rehabilitation success based on current flora criteria calculated at two different scales (rehabilitation blocks of a common age and plots) did not correlate with koala presence. We investigated the possibility of developing new criteria, still based on abiotic and flora characteristics, that would be more relevant to koalas. Using zero inflated a priori models, we found that Eucalyptus and Corymbia richness as well as canopy cover had a positive effect on koala recolonisation, while elevation had a negative impact. Several difficulties in predicting koala recolonisation remained, such as model uncertainty, choice of appropriate scales and geographic specificity of relevant variables. This study provides strong evidence that to ensure the return to a fully functional ecosystem, fauna should be directly monitored in rehabilitated sites.

Key words: fauna monitoring, rehabilitation, restoration, indicators, completion criteria, koala, recolonisation, disturbed habitat, mining

Introduction

As a result of the anthropogenic domination of the planet (Vitousek et al. 1997), most biodiversity is contained in ecosystems already impacted to some extent by humans (Pimentel et al. 1992). Conservation of biodiversity thus can no longer rely only on the parts of the landscape set aside in protected areas (Sinclair et al. 1995). Instead, - 93 - restoration of impacted landscapes has become a necessary tool in conservation (Cairns 1988; Jordan et al. 1988; Sinclair et al. 1995; Bennett 2000). The need for restoration arises from many situations, including post-war damages (Hong 2001), industrial catastrophes (Ipatyev 2001), or daily human activities such as agriculture, logging or mining (Dobson et al. 1997). Restoration projects are costly interventions that need explicit performance standards and goals, so that the degree of success can be evaluated (Jackson et al. 1995; Cairns & Heckman 1996). Assessing success is also essential to developing adaptive management strategies (Gibbs et al. 1999; Block et al. 2001).

Criteria used to assess restoration success typically focus on abiotic and flora characteristics (Kearns & Barnett 1999; Tongway & Hindley 2003; Ruiz-Jaen & Aide 2005). Common criteria include objectives for landform, erosion, stability, water quality and level, contaminated land and flora density and diversity. In contrast to fauna, these criteria are relatively easy to measure and show little seasonal variation (Ruiz-Jaen & Aide 2005). Once abiotic and flora criteria are deemed satisfactory, then it is usually assumed that fauna will follow the same trends. Many researchers have questioned this assumption and argued for it to be tested (Palmer et al. 1997; Bisevac & Majer 1999b; Block et al. 2001; Koch 2007). Indeed, flora criteria assessing success of flora may be defined differently from flora criteria relating to the success for fauna (e.g. for certain fauna, trees must contain hollows, not merely be present). Moreover, fauna could depend on many other variables besides flora, including interaction with other species (Soulé et al. 2005), social structure (McAlpine et al. 2006b), dispersal abilities (Andersen 1994) and presence of population source (Majer 1989). Demonstrating to what extent satisfactorily restoring flora equals restoring fauna has been identified as a key challenge for restoration ecology (Clewell & Rieger 1997).

This study investigates this paradigm in mine rehabilitation on North Stradbroke Island (NSI), Australia. Mine rehabilitation (sensu Hobbs 1998) is a good example of this restoration paradigm. Although fauna is typically part of the final goal, it is again not explicitly considered in rehabilitation criteria (e.g., BHP Billiton 2006; EPA 2006; Alcoa 2007; BEMAX 2009). For example, governmental guidelines on mine-site rehabilitation state that in most cases it is too difficult to directly measure fauna and thus habitat variables, in particular flora, should be used as proxies (EPA 2006). This

- 94 - again underlines that flora criteria serve the double goal of reflecting their own state as well as that of fauna.

We wondered if rehabilitation success, based on fulfilment of currently implemented flora criteria, was reflecting the recolonisation pattern of a specific fauna species, the koala Phascolarctos cinereus. In a second step we defined other habitat criteria, more relevant to koalas, and tested if any of these habitat variables were better correlated to the recolonisation by koalas. The koala was chosen as a model fauna species because the factors influencing koala distribution have been intensely studied, which provided us with good working hypotheses for factors influencing koala recolonisation of rehabilitated landscapes. Also, the koala is a flagship species whose conservation on NSI is considered crucial by all stakeholders. We tested the efficiency of both sets of criteria to reflect koala recolonisation success based on faecal pellet surveys.

Materials and methods

Study site

Open mining for sand occurs on NSI, Moreton Bay, Queensland. Mining is followed by progressive rehabilitation, with 3000ha rehabilitated so far. NSI is an island formed predominantly of unconsolidated Cainozoic sediments (Laycock 1978), with a wet-dry subtropical climate (Specht 2009). We conducted ground checking for koala faecal pellets (scats) in the undisturbed surroundings of rehabilitated areas. This ensured a remnant population of koalas existed and could potentially recolonise (e.g. one mine was not included in this study because surrounding populations could not be confirmed). We did not however establish koala density in the undisturbed surroundings for logistic reasons; we considered the presence of scats a sufficient proof of the availability of population sources. Koala scats are easily identifiable (Triggs 1996).

Study plots

As our goal was to relate flora success criteria with koala presence, we studied plots used by the mine for assessing flora success criteria (Sibelco/CRL, unpublished data). These mine monitoring plots, each measuring 50x10m, were compatible in size with plots used in previous koala scat surveys (Lunney et al. 2000; Curtin et al. 2002). Plots - 95 - were established (at 200m intervals) along transects. Transects were established perpendicular to the mine path (i.e. across the width as mine paths are ribbon-like), with a random start between 0 to 100m from the undisturbed surroundings, then evenly spaced (at 100m intervals) across rehabilitated areas. All mining monitoring plots were included, except those younger than 6 years post-rehabilitation, which were excluded owing to time constraints. The age of plots varied from 7 to 31 years (see description of each plot in Appendix B). As practice for rehabilitation evolved over time, plots were separated into three groups on the basis of key methodological changes. Group ‘Pre- 1987’ included rehabilitation methodology mainly designed to stabilise landforms (N=12). Group ‘1988-1997’ included rehabilitation that used refined seed mixes to allow a more diverse flora to develop (N=15). And group ‘Post-1998’ included rehabilitation that used yet another improved seed mix and only endemic seeds collected on the island (N=27). Plot coordinates were recorded by GPS (Garmin, eTrex®H, USA, accuracy ±7m, AMG 84) and delimitated with measuring tapes. The ground was entirely searched for koala scats, as scat deposition is not limited to the base of the trees (Ellis et al. 1998). This took up to four hours, and was conducted by a single researcher (RC) to standardise observer bias (Neff 1968). In each plot, two koala presence variables were recorded: the number of scats and the number of scat locations (i.e. the number of trees under which scats were found). Plots were considered as unoccupied if no scat was found. As a quick validation test, three randomly-chosen 50x10m areas were searched within the home ranges of radio-tracked koalas. Pellets were found inside all three plots and they could not have been misclassified as unoccupied.

For each plot, environmental variables were recorded following the methodology of the mining criteria monitoring or using the mining database (Sibelco/CRL, unpublished data). All trees in the plot were counted. The percentage of canopy and ground cover (plants, bare or litter) was estimated every 2m along the two transects forming the longer borders of the plots. These are part of the data routinely collected when assessing flora rehabilitation success (main flora criteria are presented in Table 4 of the Introduction Chapter). We also calculated more koala-relevant variables using the same environmental data or readily available data on habitat characteristics, which could be easily calculated by the mine. Koala-relevant variables measured were: species richness, density, circumference at breast height and tree height for koala staple trees (Eucalypt

- 96 - and Corymbia, Martin & Handasyde 1999). Landform variables, comprising slope, aspect and elevation, were extracted using Terramodel Version 10.61 from a 2008 airborne laser scan of the island (Sibelco/CRL, unpublished data). Other koala variables included (1) distance from the plot to the swamps, which represent primary koala habitat and could constitute a source for recolonisation; and (2) distance to undisturbed habitat, a measure of easiness of recolonisation (Table 1).

Table 1: Description of the explanatory variables contained in the different models

Explanatory variables description

Method factor with three levels based on rehabilitation method (see text) Plants percentage of plants covering the ground in the plot Bare square root of percentage of bare ground in the plot Elevation elevation in meters above sea level of the plot Density log of mean density of Eucalyptus and Corymbia species in the plot (trees/ha) Richness mean number of Eucalyptus and Corymbia species in the plot Circumference (CBH) square root of mean circumference in cm of Eucalyptus and Corymbia species in the plot Canopy percentage of canopy cover in the plot Slope percentage of slope of the plot Distance to undisturbed Euclidean distance in meters from the plot to the edge of closest undisturbed areas Distance to swamps square root of Euclidean distance in meters from the plot to the edge of swamps P, N and K in soil quantity of P (mg/kg), N (percentage of weight) and K (mg/kg)

Data analysis

Comparing mining flora criteria to koala presence

Criteria assessing rehabilitation success for flora include species diversity, density, ground cover and presence of weeds (see Table 4 of the Introduction Chapter, from CRL 2007), with each criterion having a specific threshold. For example, tree density in rehabilitated areas must not be significantly less than 75% of density of mining reference sites (these reference sites come either from the pre-mining survey of the mine path or from the mine surroundings). Success in rehabilitation as assessed by the mine is classified by block of rehabilitation age. Depending on the rehabilitated block, 23 to 63 flora criteria were assessed. Variation was largely due to different areas having different numbers of species in the reference sites that had to be re-established. We classified - 97 - rehabilitation success in two ways: on the basis of flora criteria as defined by the mining company (see Table 4 of the Introduction Chapter) and on the basis of the number of koala scats or scat locations. We compared the correlation between the flora and fauna rankings using Spearman’s ρ and Kendall’s τ tests in PAWS Statistics 18.0 (IBM 2009). In case the scale at which rehabilitation success is assessed (block) by the mine was not fine enough, we calculated flora success for the same criteria at the plot level. We compared flora success at the plot level to koala success by the same methods as above.

Searching for relevant habitat criteria for koalas

We constructed models with koala-relevant explanatory variables, and investigated if these variables were better related to koala presence in term of scat locations. We did not include analyses of the number of scats (i.e. we used scat locations) because correlation between scats could create over-dispersion in the data. To avoid step-wise selection techniques (Mac Nally 2000) and data dredging (Anderson 2001; Anderson et al. 2001a; Burnham & Anderson 2002), we used a priori models comprising koala relevant variables to draw inferences (Johnson & Omland 2004). This method decreases the possible selection of noise variables (Flack & Chang 1987).

We fitted our response variable using a generalised linear model with a Poisson distribution, the standard distribution for counts (McCullagh & Nelder 1989). The frequency of zeros was greater than expected under a standard Poisson distribution (Zuur et al. 2009), a situation very common in ecology (Cunningham & Lindenmayer 2005). To avoid underestimated variance and false positive errors due to this zero- inflation (Lambert 1992; Martin et al. 2005), we used a zero-inflated mixture modelling approach. This produces zeros from two different processes: a binomial process (corresponding to false zeros) and a count process (true zeros and positive values, Zuur et al. 2009; Ainsworth & Dean 2011). Our models consequently comprised two parts, the binomial part and the count part, each part successively defined by a set of a priori models.

To determine the binomial part, a priori models with the same count part constituted by all the variables and different binomial parts were compared (hereafter called zero models). The binomial part of zero models included variables influencing the - 98 - probability of not finding scats when in fact koalas were using a plot (i.e. false zeros). False zeros can result from variation in scat detectability in relation to the complexity of the ground layer (Chapter 3, Sullivan et al. 2004). Thus, the percentage of plant coverage and bare ground were included in the zero models (litter itself was excluded because it correlated with the other two). The method used for rehabilitation, which influences the litter characteristics, was another variable of the binomial part. No plot was located in wet areas, which should decrease the variability in scat decay rate (Chapter 3). However, plot elevation was included as a surrogate for temperature and humidity, which were found to influence the rate of scat decay in another study (Rhodes et al. 2011), and hence could also lead to false zeros. All models with one single variable and combinations of two variables were compared, giving a total of 10 zero models (Table 2). The best model for the binomial part was then used for each of the count models described below.

Table 2: Model ranks for the zero part of the models based on AICc, all models were fitted with the full model count part Example: zero2: Scat location~density+richness+CBH+canopy+elevation+slope+swamp+undisturbed | plants K: number of parameters, AICc: Akaike information criterion corrected for small sample size, ΔAICc: AICc differences, see descriptions in text

Model binomial part K AiCc Δ AICc evidence AICc we ight ratio zero5 plants+bare 12 142.0 0 0.62 1 zero8 bare+elevation 12 144.6 2.6 0.17 3.7 zero2 plants 11 145.6 3.6 0.10 6.1 zero6 plants+elevation 12 148.1 6.2 0.03 21.8 zero10 method+elevation 13 148.8 6.9 0.02 31.1 zero1 bare 11 149.1 7.1 0.02 34.7 zero0 1 10 149.1 7.1 0.02 34.7 zero7 plants+method 13 149.3 7.4 0.02 39.7 zero4 method 12 150.9 8.9 0.01 86.1 zero3 elevation 11 151.6 9.6 0.01 121.6 zero9 bare+method 13 154.4 12.4 0.00 489.0

For the count part of the models, ecologically-meaningful variables for koala presence

- 99 - were identified to build a priori models. Owing to correlation between explanatory variables, tree height, age of rehabilitation and method were excluded from all count parts (see selected variables in Table 1). Two to three different variables were included in each count part to minimise the risk of spurious effects due to our small sample size (Burnham & Anderson 2002; Babyak 2004). The specifics of each model are detailed below, and the general literature on which the models are based is given in Table 3. Time since fire and rainfall have been found to influence koala distribution but probably not at our scale of time and space (Cork et al. 1997; Sullivan et al. 2003a; Matthews et al. 2007), so were not included.

Table 3: Different variables found in the literature to influence the presence of koalas (or arboreal marsupials)

Approach variables influencing koala literature (and others) presence Small scale density and richness of food and Bryan 1997; Cork et al. 2000; Hindell & shelter trees, size of trees and Lee 1987; Lunney et al. 2000; McAlpine quantity of foliage et al. 2006a, b; McAlpine et al. 2008; Moore & Foley 2005; Munks et al. 1996; Rhodes et al. 2006; Sullivan et al. 2003

Landscape elevation Crowther et al. 2009; Kavanagh et al. scale 1995 slope Pausas et al. 1995 proximity to creek-bed and Cork et al. 2000; Gordon et al. 1988; waterholes Lunney et al. 2000; Munks et al. 1996; Sullivan et al. 2003 Fragmentation ease of immigration McAlpine et al. 2006a, b; McAlpine et al. 2008 distance from population Braithwaite et al. 1993 sources Multilevel combination of above variables McAlpine et al. 2006a A model aside: foliar content (water, nutrients Ellis et al. 1995; Freeland & Janzen leaf and toxins) 1974; Freeland & Winter 1975; Moore & components Foley 2005; Pausas et al. 1995

Models 1 to 10 incorporated every combination of two and three fine-scale variables related to food and shelter trees for koalas (density, richness, circumference and canopy). Models 11 to 14 were composed of every combination of two to three landscape features (elevation, slope, and distance to swamps). Model 15 emphasised a - 100 - fragmentation approach, where distance to undisturbed populations and population sources (i.e. swamps) would be influential. Models 16 to 33 investigated multilevel models. Habitat quality was emphasised by including in each model a combination of two fine-scale variables, then adding in turn variables from landscape and fragmentation models that have been most recurrently found to influence koala distribution in the literature (elevation, distance to source or undisturbed area). Model 34 consisted only of these three variables.

A model aside was constructed to represent another influence on koala presence. Foliar content, and particularly toxins, have been found to influence koala distributions (Moore & Foley 2005). Within a species, individual trees also vary in toxin concentrations (Lawler et al. 1998), which can create patchiness in nutritional quality of foliage (Lawler et al. 2000). While mining companies are not routinely monitoring foliar content, soil characteristics are commonly collected. It is thought that the amount of foliar toxin is correlated with soil composition (Braithwaite et al. 1984; Cork & Sanson 1990; Crowther et al. 2009). Soil data was only available for a subset of 32 sites, so models with concentration of nitrogen, phosphorus and potassium in plot were tested in a parallel model selection. Only the global model was tested for this subset (see results for explanations). It contained the same binomial part as all other models.

Comparing koala-specific flora criteria to koala presence

Finally, for the most relevant koala variables found in the preceding step, we reiterated our first step by calculating the correlation between the habitat and fauna rankings using Spearman’s ρ and Kendall’s τ tests. This time the habitat criteria used were based on mean values of koala-relevant variables, calculated by block of rehabilitation, similarly to the procedure used by the mine to currently assess rehabilitation success.

Model selection and validation in R

Models were constructed, compared and validated using R 2.12.0 (R Development Core Team 2010). Explanatory variables were graphically analysed and skewed variables were square or log transformed, then standardised to allow comparison of model parameter estimates (Quinn & Keough 2006). Prior to the inclusion of any variables in

- 101 - the models, collinearity was tested using variance inflation factors (VIF). Any VIF superior to three was examined and eliminated if it was theoretically sound (O’Brien 2007).

Zero-inflated Poisson were created for all models using the R-package “pscl” (Jackman et al. 2010). Spatial correlations in the data and the residuals were analysed with R- package “ncf” (Bjornstad 2009). We estimated the global goodness-of-fit between the global model, and the null model with a likelihood ratio test (Mundry 2011), which is preferred to a Wald test in the context of small sample sizes (Pawitan 2001). To rank the models, we used Akaike’s information criterion (Akaike 1973; Anderson et al. 2001b; Burnham & Anderson 2002) corrected for small sample size, AICc (Hurvich & Tsai 1989).

Multi-model inference methods were used to determine the relative importance of explanatory variables based on our set of models (Anderson 2001; Anderson et al. 2001a). We calculated, based on AICc, Akaike differences (Δ) between each model and the most parsimonious model; Akaike weights, a measure of the weight of evidence of each model; and the evidence ratios (Burnham & Anderson 2002). To account for model uncertainty, we calculated the model average parameter estimates and the unconditional standard error of each estimate (Burnham & Anderson 2002).

Results

Comparing mining flora criteria to koala presence

In the 54 searched plots, 24 contained scats (1 to 444 scats per plot; scat, vegetation and landform characteristics of plots are shown in Table 4). Ranking of rehabilitation blocks based on flora criteria used by the mine did not correlate with ranking based on koala scats (Spearman’s ρ =0.369, p=0.295, Kendall’s τ-b =0.277, p=0.291). The same was found for scat locations (Spearman’s ρ =0.356, p=0.313, Kendall’s τ-b =0.277, p=0.291). In fact, the best rehabilitated area in flora ranking had no koala scats. Likewise, ranking of plots based on flora criteria used by the mine did not correlate with ranking based on koala scats (Spearman’s ρ =0.127, p=0.424, Kendall’s τ-b =0.098, p=0.421) or scat locations (Spearman’s ρ =0.086, p=0.588, Kendall’s τ-b =0.061,

- 102 - p=0.623).

Table 4: Characteristics of the plots (N=54) searched for koala scats

mean SEM minimum maximum Year of rehabilitation 1993 1 1978 2002 Number of scat locations in plot 1.3 0.3 0 8 Number of scats in plot 32.2 11.9 0 444 Time of search (minutes) 132.8 6.4 60 240 Density Eucalypt&Corymbia sp per ha 1501.1 201.0 200 9540 Richness in Eucalypt&Corymbia sp 4.4 0.2 1 8 Mean CBH Eucalypt&Corymbia sp 23.6 2.1 6.2 76.1 (cm) Mean height Eucalypt&Corymbia sp 5.4 0.4 2.4 15.0 (m) Canopy cover (%) 48.3 3.3 10 92 Ground cover: plants (%) 21.8 4.0 0 100 Ground cover: bare (%) 17.5 2.4 0 72 Elevation (m) 71.8 3.6 21.3 145.6 Aspect (degree) 187.8 13.6 4.0 329.5 Slope (%) 20.4 1.7 2.5 47.6 Distance to swamps (m) 515.3 58.9 70 1920 Distance to undisturbed (m) 163.4 14.7 20 435

Searching for relevant habitat criteria for koalas

The goodness-of-fit of the global model was significantly better than a model with just the intercept (likelihood ratio test, χ2=55.35, p=2.70e-08). Residuals plotted against each explanatory variable (included or not included in the models, see list in Table 4) showed no patterns. Very few spatial correlations were found using the spline correlograms of the data and even less in the residuals (Bjornstad 2009).

When zero models were compared (Table 2, Zero0 to Zero10), we found that the most parsimonious model based on AICc, included plants and bare ground (ΔAICc > 2, Table 2). As a result, we fitted the binomial part of the 34 models including different count parts with plants and bare ground.

The ranking of the 34 count models did not strongly support any particular model (Table 5). We found that the best model had an AICc weight of 0.38 (M22), and a total of six models were needed to achieve the 95% confidence set (ΣAICc weight=0.95).

- 103 -

AICc differences and evidence ratio confirmed that the first six models were closely supported by the data, while all models ranked after M9 were not supported.

Table 5: Model ranks for 34 models relating koala scat locations to koala-relevant habitat variables based on AICc, summing AICc weights from models M22 to M9 is necessary to achieve 95% (see text for details)

Model variables in count part K AiCc Δ AICc AICc evidence we ight ratio M22 richness+canopy+elevation 7 131.5 0.0 0.38 1.0 M23 richness+canopy+swamp 7 132.3 0.8 0.25 1.5 M3 richness+canopy 6 133.0 1.5 0.18 2.1 M8 richness+canopy+density 7 135.4 3.9 0.05 7.1 M24 richness+canopy+undisturbed 7 135.5 4.0 0.05 7.3 M9 richness+canopy+CBH 7 135.7 4.1 0.05 7.9 M2 richness+CBH 6 141.2 9.7 0 127 M28 richness+CBH+elevation 7 142.0 10.5 0 191 M17 density+richness+elevation 7 142.6 11.1 0 251 M7 density+richness+CBH 7 142.7 11.1 0 263 M30 richness+CBH+undisturbed 7 143.6 12.1 0 415 M29 richness+CBH+swamp 7 143.7 12.2 0 453 M1 density+richness 6 144.1 12.5 0 529 M18 density+richness+swamp 7 146.3 14.8 0 1636 M16 density+richness+undisturbed 7 146.6 15.1 0 1917 M19 canopy+CBH+elevation 7 147.8 16.3 0 3397 M12 elevation+swamp 6 148.1 16.6 0 4002 M0 1 4 148.4 16.9 0 4561 M25 density+canopy+elevation 7 148.9 17.4 0 5910 M11 elevation+slope 6 149.3 17.8 0 7305 M14 elevation+slope+swamp 7 149.5 18.0 0 8006 M26 density+canopy+swamp 7 149.8 18.3 0 9383 M34 elevation+swamp+undisturbed 7 150.1 18.5 0 10617 M27 density+canopy+undisturbed 7 150.9 19.4 0 16230 M4 density+canopy 6 151.0 19.5 0 17225 M31 density+CBH+elevation 7 151.1 19.6 0 17912 M6 canopy+CBH 6 151.7 20.2 0 23896 M5 density+CBH 6 151.9 20.4 0 26279 M20 canopy+CBH+swamp 7 151.9 20.4 0 27233 M15 swamp+undisturbed 6 152.5 20.9 0 35384 M13 slope+swamp 6 152.9 21.4 0 43594 M21 canopy+CBH+undisturbed 7 153.2 21.7 0 52229 M33 density+CBH+undisturbed 7 153.4 21.9 0 57449 M10 density+CBH+canopy 7 153.6 22.1 0 64030 M32 density+CBH+swamp 7 153.7 22.2 0 67238 - 104 -

Across all models, Eucalyptus and Corymbia species richness had a positive estimator (β=0.70 SE=0.16) and the highest relative importance (0.98, Table 6). Similar to richness, canopy cover had a positive estimator (β=0.67 SE=0.21), and a relative importance of 0.97. In contrast, elevation had a negative estimator (β=-0.31 SE=0.16), and a relative importance of 0.39. Elevation was also only included in one of the six closely supported models. For other estimates, we found that the sign of the estimator was unstable across models (undisturbed, swamp, CBH) and/or the importance was close to zero (density, undisturbed, CBH, slope). As a result, the influence of these parameter estimates could not be ascertained.

Table 6: Relative variable importance across all models (ordered by decreasing importance) with model-averaged estimates and their unconditional standard error estimators

Variable model-averaged unconditional relative variable estimate standard e rror importance estimator Richness 0.70 0.16 0.98 Canopy 0.67 0.21 0.97 Elevation -0.31 0.16 0.39 Swamp -0.27 0.14 0.25 Density 0.08 0.16 0.06 CBH 0.05 0.19 0.06 Undisturbed -0.05 0.13 0.05 Slope 0.10 0.15 0.00

In addition, likelihood ratio tests showed no significant differences between the global soil model investigated and a null model (χ2=3.68, p=0.29). Thus, none of our soil variables was supported.

Comparing koala-specific-flora criteria to koala presence

The two most important variables, koala tree richness and canopy, averaged across each block of rehabilitation, did not correlate with averaged number of koala scats across the same blocks (richness: Spearman’s ρ =0.510, p=0.054, Kendall’s τ-b =0.629, p=0.051 ; canopy: Spearman’s ρ =0.176, p=0.501, Kendall’s τ-b =0.239, p=0.506). The same was found for scat locations (richness: Spearman’s ρ =0.408, p=0.123, Kendall’s τ-b =0.584, - 105 - p=0.077 ; canopy: Spearman’s ρ =0.176, p=0.501, Kendall’s τ-b =0.291, p=0.415).

Discussion

On the flora equals fauna paradigm

This study examined the paradigm that restoring flora restores fauna, using the koala as a model species. We acknowledge that rehabilitation creating quality vegetation, in terms of structure, complexity and species diversity, is a condition sine qua non for fauna recolonisation. As such, some have argued that monitoring vegetation characteristics is more relevant than monitoring fauna species (Lindenmayer et al. 2000). In case of mine rehabilitation however, the approach is different, as flora is always monitored, but we propose that fauna should become an integral part of monitoring rehabilitation success. Indeed, we found that, in the current mining assessment, the best plots based on flora criteria were not correlated with koala presence. Our results indicated that flora criteria more relevant to koalas could be developed, such as richness in food trees (Eucalyptus and Corymbia) and quantity of food available (canopy). This further emphasises the dependence of koalas on their food trees (Hindell et al. 1985; Martin 1985). However, there seem to be uncertainties between models and we were unable to detect ‘one’ best model for koala recolonisation.

Moreover, the scale at which success criteria are calculated by the mines (i.e. per rehabilitation block of a common age) does not appropriately reflect the scale at which variables seemed to influence koala recolonisation. Indeed, while we detected important variables at the plot level, when we calculated criteria at the scale used by the mining company (the rehabilitation block), we found no correlation between these variables and koala presence. Determining the adequate scale in space (and time) to measure fauna success is actually one of the additional difficulties of fauna when compared to flora (Weaver 1995; Ruiz-Jaen & Aide 2005; Golet et al. 2011). In addition, there might be more than one relevant scale depending on the variables chosen (Cale & Hobbs 1994; Lindenmayer 2000; Cunningham et al. 2007).

Another problem in finding the best habitat variables to explain fauna recolonisation results from the difficulty in identifying which variables matter for the population of

- 106 - interest. Indeed, for koalas, relevant variables are often geographically specific (McAlpine et al. 2008). For other marsupials, the broader context of the landscape has been found to influence relevant variables (McAlpine et al. 2002). Thus, the relation between habitat variables and the fauna species of interest in different geographic areas would have to be studied locally before choosing which habitat variables to use as proxies. Furthermore, behaviour, notably social structure, will influence distribution pattern irrelevantly to landscape characteristics, even in a species defined as solitary such as the koala (McAlpine et al. 2006b; Ellis et al. 2009).

Given all these difficulties, it seems that using flora or other habitat proxies, without appropriate and local prior knowledge, may not accurately reflect fauna recolonisation. There is a decision matrix in the choice between using flora proxies or fauna criteria. On the one hand, one needs to weigh the costs of assessing the relevance of flora proxies against using a direct fauna criterion. On the other hand, one needs to assess the consequences of using a criterion uncoupled from what it is meant to represent, which could result in an inappropriate management decision. The final choice will depend on the fauna species of concern. For species of particular interest, such as the koala, we predict that fauna criterion will be chosen, as the cost of taking an inappropriate decision will override any other cost. Fauna monitoring data will also have the second advantage of being the most effective in terms of communication with stakeholders. An indirect method, such as scat count, would be a good candidate for a fauna criterion regarding the fate of koalas in rehabilitated areas.

Limitations of this study

The main limitation of this study was the relatively low number of plots compared to the number of habitat variables we were interested in. Babyak (2004) suggested that the number of explanatory variables should be scaled by the number of plots available by a minimum factor of ten. This limited the number of explanatory variables we could include in our models, and probably the ability of our models to detect all relevant variables.

Our low number of plots could explain why some variables that we thought would influence koala recolonisation did not. Here we examine other limitations in our study - 107 - design that may also explain this. Distance to undisturbed area was not a limitation to koala recolonisation. Koalas radio-tracked on NSI move an average distance of 125m per day, with some koalas moving as much as 441m (Appendix A). This means that, given the narrowness of the mine path, even the more “remote” parts of rehabilitated areas may be easily accessed by koalas (see mean and maximum distances from plots to undisturbed areas in Table 4). We also found no positive correlation between tree size and koala presence. It may be that most rehabilitated trees already exceeded the minimum size for koala use (Matthews et al. 2007). The lack of influence of koala tree density may result from the fact that density could be better modelled by an additive relation instead of a linear one. At high tree density, the competition for nutrients increases and more carbon-based defences are produced by trees (Cork & Braithwaite 1996). This could decrease the palatability of leaves for koalas. Similar to other studies (e.g., Samedi 1995), soil composition did not correlate with koala presence, which might reflect dissimilarities between soil and leaf characteristics (Moore et al. 2004). For instance, E. viminalis can grow on poor soils, such as the NSI soils derived from siliceous sand, yet show rather high foliar nutrients (Ladiges & Ashton 1974). Thus, foliar content could be having an influence on koala presence but our proxy, soil composition, might not have appropriately reflected it.

Another potential limitation of this study is the presence of false positives. Koala scats are easy to identify, and in the absence of possums on the island, there could not be scats from other species misidentified as a koala scat (Triggs 1996). However, there is a risk inherent to any fauna survey, where presence could be solely due to dispersing or “floating” individuals. Although it is possible that some of our plots belonged to this category, other studies in this rehabilitated area have proven that koalas can permanently inhabit rehabilitated areas (Chapter 5, Appendix C). We thus tend to think that most of our plots would be reflecting established koalas’ choices.

Conclusion

This study demonstrated that flora success in restoration does not automatically lead to fauna success. In fact, measuring flora criteria and assuming that fauna will follow the same trajectory has the potential to misrepresent fauna’s actual fate in restoration. This has also been demonstrated in invertebrates (Crisp et al. 1998; Longcore 2003; - 108 -

Andersen et al. 2004), amphibians (Mazerolle et al. 2006) and birds (Buffington et al. 2000). The flora/fauna mismatch could be exacerbated (1) for fauna limited by other factors than these routinely included in mining criteria (e.g. hollows-dependent species, fauna with limited dispersal abilities) or (2) for fauna less studied than koalas, and for which relevant habitat characteristics are unknown, making it difficult to find relevant proxies. The obvious danger with assuming that restoring flora is restoring fauna is to declare a site restored when only one component of biodiversity is actually returned, while the fate of fauna remains unknown. This represents a serious threat, not only for the conservation of biodiversity in general, but for the long-term resilience of the ecosystem (Fischer et al. 2006). Fauna plays, amongst other things, many crucial roles in ecosystems processes and functions (Majer 1989; Nichols & Nichols 2003). Indeed, even in cases where fauna species are not the direct target of rehabilitation (e.g. where there is no flagship species to re-establish), certain fauna groups should still become an integral part of assessing rehabilitation success in general. These could include, as mentioned above, fauna groups involved in ecosystem functions (e.g., pollinators, detritivores, Walker 1992; Andres & Mateos 2006), as well as keystone species (Simberloff 1998) or species particularly sensitive to threats (Lambeck 1997). If the lack of congruence between flora and fauna success in rehabilitation found in this study is general, developing cost-effective, relevant and feasible fauna criteria is crucial. This may well be the next challenge of Restoration Ecology.

Acknowledgements

The first author is supported by an Endeavour Europe Award and an Endeavour International Postgraduate Research Scholarship. Sibelco Australia – Mineral sand provided ongoing support for this research through the provision of logistical support, access to company sites and relevant maps/databases.

- 109 -

- 110 -

Chapter 5

Habitat quality and the ecology of an arboreal mammal, the koala

Phascolarctos cinereus, in a rehabilitated landscape

Abstract

The continuing increase in anthropogenic habitat disturbances forces restoration to achieve high quality. If restoration is to provide new habitat, fauna populations in restored habitat have to have the same survival and reproductive rates as populations found in undisturbed surroundings. If animals recolonise restored areas but later experience a decreased fitness, restored areas could transform into ecological sinks or traps. We investigated this hypothesis using koalas Phascolarctos cinereus in mining rehabilitated areas on North Stradbroke Island, Australia. We compared rehabilitated and undisturbed areas on the basis of the attractiveness of vegetation characteristics, koala condition, roosting trees and diet, and finally predation risk. Rehabilitated areas provided a very attractive habitat indeed in terms of food tree diversity, density and availability of foliage. Koalas were using rehabilitated areas, with a substantial amount of their home range inside it. Koalas were of good condition and had high reproductive output. Koalas were able to find roosting and food trees in rehabilitated areas. Predator density was not higher in rehabilitated areas compared to undisturbed surroundings. To reach a more thorough picture of rehabilitation success, further research should assess the quality of the foliage in rehabilitated areas, as well as predation risk. Variation in predation risk in rehabilitated areas could result from factors such as (1) the time spent by koalas on the ground while changing tree, and/or (2) the predator foraging efficiency, which is linked with densities of tracks and understorey thickness. The hypothesis of ecological sinks or traps is not supported by the results of this study so far. In future projects, the evaluation of restoration success should include criteria on fauna fitness to ensure functioning ecosystems, and not ecological sinks or traps, are created.

Key words: fauna, rehabilitation, restoration, recolonisation, koala, mine disturbance

- 111 -

Introduction

We are living in a world dominated by human beings (Vitousek et al. 1997). As a result, pristine environments available for protection are decreasing, and degraded environments in need of restoration are constantly increasing (MacMahon & Holl 2001). While it would clearly be preferable to avoid impacts on remaining undisturbed areas (Hobbs 2004), the demand for resources limits this approach and increasingly restoration of degraded land is coming to play a crucial role in the conservation of biodiversity (Bennett 2000).

Restoration projects, as defined by the Society for Ecological Restoration, aim to return damaged areas to their pre-disturbance state, which includes the return of all the biological components (SER 2004). There seems, however, to be a bias toward monitoring the recovery of flora (Ruiz-Jaen & Aide 2005), whereas research related to restoration of fauna remains sparse (Morrison 2001; Majer 2009). In particular, few studies to date have investigated the impact that restored areas have on the ecology of fauna species that use them (Goldstein 1999). This represents a critical gap in our understanding of how and how much ecological restoration can compensate habitat destruction. The need for future research to encompass fundamental ecological characteristics such as fauna diet and shelter, reproductive success, health status, disease susceptibility and inter-species interactions in restored habitats has been stressed (Fox 1998; Bellairs 1999; Miller & Hobbs 2007). Researchers have advocated that animal behaviour should also be an integral part of assessing impacts of management actions (Robertson & Hutto 2007).

A lack of fauna monitoring is counter-productive in any management program that impacts on wildlife, and this is not different for restoration. Monitoring provides vital feedback which is essential to the development of effective adaptive management strategies (Gibbs et al. 1999; Block et al. 2001). Moreover, a deeper analysis of the species’ characteristics in restored habitat can aid in the detection of potential ecological sinks (Pulliam 1988; Dias 1996) or, even more detrimental over time, ecological traps (Dwernychuk & Boag 1972; Delibes et al. 2001; Kristan 2003). Source/sink dynamics are often used to describe the heterogeneity of habitats available to a species: good- quality habitats that produce a surplus of animals are named sources, while poor-quality - 112 - habitats where a demographic deficit occurs are named sinks (Pulliam 1988; Dias 1996). An ecological trap arises when animals express a maladaptive choice: they settle in a habitat where their fitness is decreased relatively to what it could be in other available habitats (Robertson & Hutto 2006). There are other theories and frameworks for habitat selection, but we will not cover them in this chapter. For instance, under the ideal free distribution, individuals settle first in the habitat where their fitness will be the highest (on the basis of habitat quality and conspecific density). Settlement in the best habitat continues until the density in this habitat decreases individual fitness to the level of fitness in the second most suitable habitat, at which stage individuals settle in the second most suitable habitat. This occurs without their choice ever being pressured by other conspecifics or being inadequate (i.e. individuals are able to assess habitat suitability reliably). On the contrary, in the ideal despotic distribution, individual choices are restrained by the territorial behaviour of other conspecifics, so that individuals may be forced into settling in a less suitable habitat before their fitness in this habitat equals that of the most suitable habitat (Fretwell & Lucas 1970; Fretwell 1972). We chose not to work in these frameworks because we were interested in theories where individuals are not always able to make the best choice for their fitness. Furthermore, for koalas Phascolarctos cinereus, conspecifics might not be a strong constraint for habitat selection as koalas coexist with large home range overlaps (Ellis et al. 2009).

Restoration of disturbed habitats has the ultimate goal of supporting self-sustaining faunal assemblages that characterised the habitat prior to disturbance. Thus, even if fauna does recolonise an impacted site, restoration can only be regarded as a success if the population is stable in the long term and its individuals have equal fitness in the restored habitat as in the pre-disturbance habitat (this is often approximated by studying reference sites in the adjacent undisturbed habitats that provide the source for the recolonising population). That is, rehabilitated areas should not be population sinks nor should they function as ecological traps. To identify such processes, the success of restoration should not only be measured in terms of the return of communities (e.g. ant diversity) and the return of each individual species (as defined in Goldstein 1999), but ultimately should be measured in terms of the re-establishment of populations whose vital demographic rates (survival and reproductive rates) are similar to those of

- 113 - undisturbed populations.

Habitat restoration associated with post-mine rehabilitation is rapidly expanding and improving, driven in some places by the impetus of a legislative framework (e.g. in USA with the U.S. Surface Mine Control and Reclamation Act of 1977), public concern and the industry itself (Coaldrake 1979; Ellis 2003; Grant & Koch 2007). Post-mine rehabilitation provides an ideal framework to study the fate of fauna populations in restoration projects, as it is readily available and extended, and there is a legislative requirement to monitor rehabilitated areas and achieve regulatory standards (Smyth & Dearden 1998).

This study investigates and compares some ecological characteristics of our fauna model, the koala, in rehabilitated post-mining landscapes and nearby undisturbed landscapes on North Stradbroke Island (NSI), Australia. The koala, an arboreal marsupial, was chosen because it is a flagship species and its presence and welfare is important for all stakeholders. On NSI, a sand mining company is progressively rehabilitating its mine path with more than 3000ha currently available for fauna recolonisation. In order to test predictions from the theories of population sinks and ecological traps, we hypothesised possible scenarios. A population sink could occur if the rehabilitated habitat is of inferior quality. Individuals would not choose to settle in the sink but more would overflow from high-quality habitats, and animals inhabiting rehabilitated habitats would have lower survival and reproductive rates (Dias 1996). Lower survival rate could be associated with a higher predator density in rehabilitated areas. Indeed, some studies have found an association between disturbances and feral species, including feral predators such as European foxes Vulpes vulpes and wild dogs Canis lupus (Chapter 2, May & Norton 1996). Although adult koalas (>3kg weight) are probably not vulnerable to fox attacks (although anecdotal reports on fox attacks of adults have been reported, Melzer 2011), juvenile koalas fall within the fox prey-size range (NPWS 2003) and feral dogs are known to attack koalas of any size (Appendix A). In addition, small trees in younger rehabilitated areas may not provide sufficient shelter from aerial predators such as eagles and owls (Jurskis & Potter 1997; Melzer et al. 2003), or may require increased movements between trees owing to lower leaf mass per tree for feeding (see discussion). Lower reproductive rate could be a result of lower

- 114 - food quantity and/or quality, as well as an increased energy expenditure associated with movements to find appropriate food trees.

In ecological traps, contrary to sinks, animals choose to settle in lower-quality habitat despite high-quality habitat being available. An equal-preference trap can occur in rehabilitated habitat if individuals do not discriminate between rehabilitated and undisturbed habitats, but these two habitats have different suitability (Robertson & Hutto 2006). This could happen if rehabilitated habitats present the same cues of suitability as undisturbed habitat but with lower survival or reproductive rates (for the same potential reasons as above). A worse kind of ecological trap that could potentially occur for koalas on NSI is the severe trap, where animals favour the less suitable habitat instead of more suitable habitats (Robertson & Hutto 2006). A severe trap could be created if rehabilitated areas are more attractive than undisturbed areas, while reproduction or survival rates are decreased. Young trees in rehabilitated areas may be more attractive than surrounding undisturbed habitat as koalas favour leaves that have a high concentration of crude protein and lower fibre content (Ullrey et al. 1981; Zoidis & Markowitz 1992; but see Moore & Foley 2000), typical of fast-growing trees (Hume 1990).

On the basis of these population source/sink and ecological trap hypotheses, we can make some general predictions: i) if the rehabilitation area functions as a population sink, we predict that the habitat would be of low quality (e.g. low tree density and richness, particularly for specific koala food trees, high predator density), that the koalas residing there would be at low density, have poor general condition and low reproductive rate; ii) if the rehabilitation area functions as an ecological trap, we predict that although the habitat quality may be low as in the sink scenario above, the density of resident koalas would be equal to or higher than the density of those residing in undisturbed areas (as koalas would not avoid the rehabilitated habitat), but that koalas would present the same signs of low survival and reproductive rate as in the sink hypothesis; and iii) if the rehabilitated area provides good quality habitat, we predict that the habitat would present high levels of koala tree density and richness, and that resident koalas would present signs of survival and reproductive rate comparable to those in undisturbed areas, i.e. equal or lower predator density, equal or higher koala

- 115 - density, and individuals of good condition and breeding.

This chapter provides a preliminary attempt to test the above hypotheses by investigating different ecological components relevant to the habitat selection theories of source/sinks and traps for koalas on NSI. It compares, in undisturbed and rehabilitated koala habitats, key elements of habitat quality such as vegetation composition and structural characteristics, roosting trees, and the presence of feral predators; as well as providing information on koala diet, health and reproduction. Demonstrating the actual occurrence of ecological sinks or traps is notoriously difficult (Dias 1996; Robertson & Hutto 2006), and beyond the aims of this chapter, however, here we present a first step toward the investigation of the potential of restored landscape to act as ecological sinks or traps.

Materials and Methods

Study site and characteristics

North Stradbroke Island is a sand island located in Moreton Bay (27°34’S, 153°28’E) off the coast of south-eastern Queensland, Australia. It is predominantly formed of unconsolidated Cainozoic sediments (Laycock 1978), with a wet-dry subtropical climate (Specht 2009). Open-cut sand mining has been conducted on the island since the late 1940s to extract heavy minerals from dredged sand. The mined area is progressively rehabilitated: dune landform is recreated, topsoil is spread and stabilised, then seeds and tube-stock trees are planted out. Previously, mining activity has been undertaken in koala habitat, with these areas now being in various stages of rehabilitation. Preliminary surveys in undisturbed areas contiguous with those rehabilitated areas have confirmed their use by koalas (RC, unpublished data). This means remnant populations exist in proximity to rehabilitated areas and koalas can potentially recolonise them.

Koala habitat characteristics

Rehabilitated habitats of different ages were surveyed for signs of koala presence (faecal pellets, i.e. scats, and distinctive scratch marks on smooth barked trees) to determine the timescale of recolonisation in relation to successional stages of regrowth. - 116 -

Plots (50x10m) were searched for scats, with a total of 36 plots used by koalas selected to compare vegetation characteristics in rehabilitated and undisturbed koala habitats. Plot coordinates were recorded by hand-held GPS (Garmin, eTrex®H, USA, accuracy ±7m) using UTM in AMG 84 projection. The number of koala scats per plot was also recorded, as an index of intensity of use. Previous experiments found variation in scat detectability and decay rates between different habitats on the island (Chapter 3). Thus, the scat count was multiplied by a correction factor for the lower detectability characteristic of undisturbed plots (16.1% lower detectability, Chapter 3). No plot was located in zones subject to flooding, so no correction for higher scat decay rate was necessary (Chapter 3). A single researcher (RC) conducted all surveys to standardise bias resulting from heterogeneity in observer skills (Neff 1968).

There were no pre-mining vegetation studies available. The vegetation characteristics of koala habitat in undisturbed areas were therefore described by sampling all remnant vegetation communities surrounding the mine path. These remnants indicated which vegetation communities were most likely to be present before mining, and were replaced by rehabilitated areas. Vegetation communities were based on Regional Ecosystem (RE) maps (Queensland Herbarium 2009). Twelve plots in four vegetation communities were selected (Table 1).

Table 1: Regional Ecosystems (RE) potentially replaced by mining rehabilitated areas and their floristic description

RE community 12.2.6 Eucalyptus racemosa , Corymbia intermedia , C. gummifera , Angophora leiocarpa and E. pilularis shrubby or grassy woodland to open-forest 12.2.7 Melaleuca quinquenervia open-forest to woodland with E. tereticornis , C. intermedia , E. robusta , Lophostemon sp ; 12.2.8 E. pilularis and E. resinifera open-forest 12.2.10 Mallee forms of C. gummifera , E. racemosa and E. planchoniana ± Banksia aemula low shrubby woodland

To compare the vegetation characteristics in rehabilitated areas we assigned plots to one of three categories, based on rehabilitation methodology. Prior to 1987, rehabilitation was designed to stabilise landforms and involved exotic as well as native plant species

- 117 -

(6 plots). After 1987, a new rehabilitation policy was developed in which only endemic species were used and the use of Acacia sp. (black wattle) ceased, as it had been found that the acacias were outcompeting other species such as eucalypts (10 plots). After 1998, the method once again was refined: the previous extensive use of Allocasuarina sp., a species that was reaching densities too high, ceased. Moreover, only seeds collected on the island were used (8 plots). The improvements in methodology (including increases in the number of endemic species used, translocation of flora species of interest, etc.) were deliberate attempts by the company to improve the quality of the rehabilitated areas and succeed in re-establishing a functioning ecosystem (see Introduction Chapter for a full description of the methodology and its improvement, including increases in the number of endemic species used, translocation of species of interest, etc.). The improvements of methodology were based on benchmarking (Sutherland & Peel 2010), i.e. the comparisons of the efficiency of other methodologies used in the industry as well as the results of research on rehabilitation success (Ben Barker, personal communication, 28/09/2010).

In the 36 plots described above, environmental and vegetation variables were recorded using the methodology developed for the mine’s environmental monitoring program, or using the mining database, so that the observations of this study could be integrated with the information in the mine’s database (Chapter 4, Sibelco/CRL, unpublished data). Native tree species richness and density were calculated. A tree was defined as any live stem greater than 2m high and each tree was marked to ensure each was only counted once. Density and species richness for the main koala food trees of the genera Eucalyptus and Corymbia (E+C) were extracted (Martin & Handasyde 1999). For these genera, the circumference at breast height (CBHp for plot) was measured. When multiple stems occurred, all CBHp were added. The percentages of canopy and ground cover were estimated every two meters along the two transects forming the longer borders of the plots (Woodward et al. 2008). Elevation, slope and aspect of the plots were extracted by Terramodel Version 10.61 from a 2008 airborne laser scan of the island (Sibelco/CRL, unpublished data).

Koala condition

Eight koalas (six females, two males) were caught according to standard procedures - 118 -

(Ellis et al. 1995); seven of these were captured near rehabilitated areas and the eighth was captured inside one. Tooth-wear classes (Martin 1981; Gordon 1991) and body condition (Ellis & Carrick 1992) were assessed, and blood samples were collected (5ml, cephalic vein). Reproductive status was determined for females by pouch checks.

Movement patterns and habitat use

Koalas were fitted with radio-tracking collars, each transmitter being set to a different VHF frequency between 150–152MHz (Titley Electronics, Australia) and released back into the tree where they had been captured. Collared koalas were radio-tracked every one or two weeks from July 2008 to February 2010. Their location was recorded by hand-held GPS. The positions were plotted on the map of vegetation communities of the island (Queensland Herbarium 2009) superimposed with contours of rehabilitated areas by year (Sibelco/CRL, unpublished data). Home ranges were plotted using the Home Range Tools for ArcGIS® 1.1. (Rodgers et al. 2007). We used the kernel density estimation method, with the standard Gaussian curve (Worton 1989). Based on the Schoener index (Schoener 1981), the variances of our coordinates were unequal so the data were standardised. We used a fixed kernel (Seaman & Powell 1996), with a smoothing factor calculated by least squares cross validation (Worton 1995). Home range areas were calculated from isopleths of the volume contours. The percentages of each vegetation community inside the 95% isopleth of the home ranges were extracted in ArcGIS 9.3.1.

Roosting trees

Each tree where a collared koala was found was classified as belonging to undisturbed or rehabilitated habitat. Trees were tagged with a unique number to allow recording of reuse by the same or other koalas. Roosting tree species and circumference at breast height (CBHk for koala) were recorded.

Diet

Koala diet was evaluated by identification of the cuticle characteristics of leaf fragments remaining in the scats (Tun 1993; Hasegawa 1995; Ellis et al. 1999). When a collared koala was located, the area directly under it and the base of its tree were searched for - 119 - fresh scats (i.e. covered in mucous and smelling of eucalypts oil). Only fresh scats (i.e. less than a day or two old) were collected, to ensure the scats belonged only to the radio-tracked koala and reflected browse usage of the season in which they were collected.

For each individual koala included in the diet analysis (N=5), we selected four groups of scats produced when the koala was in rehabilitated habitat and an additional four groups of scats produced when the koala was in undisturbed habitat. We selected the groups of scats as equally as possible across seasons (Summer=11, Autumn=6, Winter=11, Spring=12). Each group of scats represents five scats collected during the same occasion and homogenised for analysis of leaf fragments (Tun 1993; Hasegawa 1995).

A NSI leaf library was prepared to assist in dietary scat analysis. Every tree species found during the study was sampled (one area sampled by trees, on average 20 leaves per sample), including species not typically defined as koala food, to avoid bias from preconceptions. Where possible, tree species were sampled four times: in rehabilitated and undisturbed areas at two separate geographic locations. The precise characteristics of leaf cuticle, for stomata in particular, were described and compared. Reference specimens from the leaf library enabled the identification of 100 leaf fragments extracted from each koala scat group. This procedure was repeated twice and the percentage of tree species present in each scat group was calculated.

Predation risk

An Activity Index (Allen et al. 1996) was calculated for foxes or feral dogs. Sand was flattened and smoothed in 2x2m plots, 1km apart, along tracks at the study sites at dusk. Plots in undisturbed habitats (n=16) and in rehabilitated habitats (n=14) were monitored at dawn on three consecutive days over three survey periods (October 2003, December 2005 and November 2009). Plots that were unreadable owing to weather conditions (e.g. heavy rain) or tyre tracks were excluded from calculations using the criteria of the Activity Index method.

- 120 -

Data analysis

All variables were tested for normality and homogeneity of variances (Levene’s test of homoscedasticity), and were compared between the three groups of rehabilitated areas and the undisturbed area by appropriate parametric or non-parametric tests in PAWS Statistics 18.0 (IBM 2009). Significance level was taken to be p<0.05 (except when accounting for Bonferroni’s adjustment), effect size (as defined in Hurlbert 1994), standard deviation (SD) or standard error of mean (SEM) being calculated when appropriate (Altman & Bland 2005).

Similarity and dissimilarity matrices were constructed using the Bray-Curtis measure (Bray & Curtis 1957) on the plant species in the plots. Square root transformation was used to down-weight the importance of abundant species. Differences between the three rehabilitated areas and undisturbed area were tested with analyses of similarities (ANOSIM, Clarke & Gorley 2006), a multivariate equivalent of the analysis of variance (ANOVA) based on similarity matrices (Clarke 1993). Browse species in the diet were compared between koalas, for different seasons and locations (rehabilitated/undisturbed habitats) with an ANOSIM based on a squared transformed Bray-Curtis matrix. These analyses were performed using Primer 6.1.12 (Primer-E Ltd. 2009).

- 121 -

Figure 1: Boxplots of the number of scats and selected vegetation characteristics of plots in rehabilitated (classified by method) and undisturbed koala habitats (dashed line represents undisturbed value)

- 122 -

Results

Recolonisation pattern

Signs of use by koalas were found in rehabilitated areas as young as 6 years since rehabilitation (the youngest age checked) and in plots representing all rehabilitation methods. The number of scats per plot ranged from 1 to 444 (Figure 1). The average number of scats, corrected for variation of scat detectability, was similar in each rehabilitated habitat and undisturbed habitat (undisturbed habitats: 23.8 SEM=7.8, pre- 87 habitats: 20.8 SEM=6.9; 88-97 habitats: 104.5 SEM=43.2; post-98 habitats: 70.9 SEM=49.8; Kruskal-Wallis test=2.27, df=3, p=0.518).

Koala habitat characteristics

The differences of vegetation characteristics between rehabilitated and undisturbed habitats are presented in detail in Figure 1 and Table 2. In rehabilitated compared to undisturbed habitat, tree density and richness were greater, E+C (i.e. Eucalyptus and Corymbia) density and richness were either equal or greater, percentages of E+C were either equal or superior and trees were smaller (even in 31-year-old rehabilitated habitats). Canopy cover and bare ground were similar for all habitats rehabilitated before 1997 and undisturbed habitats, while habitat rehabilitated after 1998 had less canopy and more bare ground. Elevation, aspect and slope were similar between the three rehabilitated habitats and undisturbed habitats (Kruskal-Wallis tests respectively: 3.51, df=3, p=0.319; 1.95, df=3, p=0.583; 4.62, df=3, p=0.201).

Species composition (NB: all trees present in this analysis were used by koalas as roosting and/or food trees) in the three rehabilitated habitat groups and undisturbed habitats was significantly different (ANOSIM, R=0.391, p<0.001). The undisturbed plots, which comprised different RE vegetation communities, were less similar to one another than were the rehabilitated plots. Rehabilitated habitats of different methods were more similar to each other than to undisturbed habitats (Table 3). However, the dissimilarity between rehabilitated and undisturbed habitats decreased with the change of rehabilitation methods, with the more recent methods producing vegetation associations more similar to adjacent undisturbed areas..

- 123 -

Table 2: Characteristics of vegetation composition and structure in rehabilitated koala habitats classified by method (R) compared to undisturbed (U) koala habitats t s e t f test statistics pd R to U to compared compared 1.5 superior 0.017 10.18 3 Kruskal-Wallis 1.7 equal 0.059 2.75 35 ANOVA 0.7 superior1.3 superior <0.001 24.15 <0.001 22.37 35 ANOVA 35 ANOVA -0.7 inferior <0.001 19.16 35 ANOVA size effect effect pre-87 -0.1pre-87 equal 0.892 34.00U Mann-Whitney pre-87 -0.1pre-87 equal 0.453 28.00U Mann-Whitney pre-87 0.1 equal 0.265 1.16test 16 T 88 to 97 1.4post-98 1.8 superior 0.001 superior 0.002 4.0088 to 97 15.00U -0.1post-98 Mann-Whitney equal -0.5 Mann-Whitney U inferior88 to 97 0.505 <0.001 0.0post-98 -0.68 7.9 equal -4.67 superior 20 18 <0.001 T test T test 0.620 0.00 52.50 U Mann-Whitney Mann-Whitney U 1 5 69 15 0.5 0.3 145 0.06 SEM U 4 74 5.3 1.9 902 207 114 0.26 mean 4 2 5 0.3 0.4 406 337 0.07 SEM post-98 32 19 37 9.6 5.1 884 0.73 1513 mean 1 3 5 0.5 0.4 376 144 0.07 SEM 88 to 97 88 to 4 27 69 5.1 501 0.62 10.0 1572 mean rehabilitated areas (R) areas rehabilitated 8 3 1 85 0.5 0.4 0.05 1315 SEM pre-87 pre-87 3 45 83 7.5 3.0 310 0.22 3617 mean

Density of trees of Density Density of E+C trees Richness of Richness E+C of E+C Percentage CBHp % cover Canopy % ground Bare - 124 -

Table 3: Average tree species composition similarity inside a group (shading) and dissimilarity between groups for rehabilitated compared to undisturbed koala habitats. The tree species given explain 50% of similarity/dissimilarity

post-98 sp + sp Allocasuarina Banksia + E. sp + Corymbia planchoniana Callitris sp + Allocasuarina sp sp + Corymbia Allocasuarina sp + Corymbia sp + Callitris sp + E. racemosa 60.76 Allocasuarina sp+ Corymbia pilularis + E. sp rehabilitated Allocasuarina sp + Banksia sp + Callitris sp Callitris sp + Allocasuarina sp racemosa + E. 61.40 Allocasuarina sp + Callitris sp 42.5 pre-87 88 to 97 to 88 pre-87 Allocasuarina sp + Callitris sp + Banksia sp 54.83 Allocasuarina sp + Callitris sp 53.72 73.86 69.26 46.02 undisturbed 42.48 Banksia sp + sp Corymbia 67.51 pre-87 post-98

Areas Undisturbed Rehabilitated 88 to 97

- 125 -

The main tree species contributing to the similarities were Corymbia sp. and Banksia sp. in undisturbed habitat, and Allocasuarina sp. and Callitris sp. in pre-97 rehabilitated areas, whereas similarities in post-98 rehabilitated areas were driven by Allocasuarina sp., Corymbia sp. and E. pilularis (other details in Table 3)

Koala condition

Based on tooth wear, captured koalas were between 2 to 14 years of age (Martin 1981; Gordon 1991). All koalas but one were in good body condition (condition index scores of 7 to 9). Only one koala caught was in a medium body condition. Blood analysis indicated that haematology and biochemistry values were in normal ranges, some had minor changes of no biological significance (Appendix D). Out of the six females caught, five had pouch or back young, the other was immature (<3 years).

Habitats used

Six out of seven koalas caught near rehabilitated habitats were subsequently found in those habitats. The signal from the last koala was lost, suggesting VHF transmitter failure. The eighth koala was captured and subsequently found in rehabilitated habitats. Across all koalas, home ranges (95% kernel) were mainly composed of rehabilitated areas (44.5% SD=18.7), then of Eucalyptus and Corymbia woodland RE12.2.6 (17.5% SD=9.4), Melaleuca woodland RE12.2.7 (13.5% SD=8.9), swamps RE12.2.15 (12.2% SD=3.6), and Eucalyptus open-forest RE12.2.8 (7.6% SD=3.5). The details for each koala are presented in Figure 2.

(a)

- 126 -

(b)

Figure 2: Percentages of rehabilitated areas (by decreasing order) in the home ranges of six koalas (a) Nareeba (65.9%) and Binang (46.9%), (b) Jundall (43.9%), Mirrigan (35.3%), Callitris (22.8%) and Dakabin (22.7%); together with the different remnant vegetation communities present in their home ranges (based on REs, RE12.2.15: swamps, other REs: see Table 1)

Remark 1: percentage of rehabilitated areas in the home range of Christine, the 7th koala, was 100% and thus is not drawn in the Figure

Remark 2: numbers of fixes were: Nareeba (female) = 41, Binang (male) = 37, Jundall (female) = 40, Mirrigan (female) = 26, Callitris (female) = 40, Dakabin (female) = 39 and Christine (female) = 44

Roosting trees

Individual collared koalas were found in trees inside rehabilitated habitats on 23% to 100% (mean=51%, SD=26%) of observations. Results for these seven koalas using rehabilitated habitats were pooled to calculate the following results, except as otherwise stated. There were 258 observations of koalas in identified roosting trees: 109 in undisturbed habitats and 149 in rehabilitated habitats (Table 4). - 127 -

Table 4: Number of times each tree species and each unique tree were used or reused in undisturbed and rehabilitated habitats by radio-tracked koalas (N=7) (5.1%) 4 (2) 2 (1) 2 12.7% (7%) 15 (7) 19 (7) number of times reused reused times of number (number of reused trees) reused of (number 13.7% undisturbed rehabilitated 149 57.80% rehabilitated 921.3 6921.3 11 7.4 109 number of times each species has been used been species has each times of number 42.20% undisturbed 123 273 154 13565.5 24.8 13654.6842.7 13.8 116 11.9 37 11.9 27 10.1 7 5821.8 16 9 48 23 1 4910.9 4.6 5 10.7910.9 3.7 2 15.4 2 47 3.7 3.4 5 7 (3) 1.3 4 1.8 31.5 10 2 (1) 12 13 (4) 2 (1) 13 6 4 (2) 1 8.1 8.7 11 0.7 7.4 rank numberrank % rank number % Tree species Banksia sp Banksia resiniferaE. tindaliae E. TOTAL E. racemosaE. sp Lophostemon sp Callitris tereticornis E. Corymbia sp pilularis E. Allocasuarina sp Angophora Acacia sp sp Melaleuca sp Duboisia tree) (umbrella sp Schefflera planchoniana E. E. robusta E. PERCENTAGE

- 128 -

Roosting trees were different on the basis of tree species in undisturbed and the three rehabilitated groups (χ2=144.37, df=48, p<0.001, contingency coefficient=0.604, p<0.001). In pair-wise comparisons, the most significant differences in roosting trees were between undisturbed habitats and pre-1987 rehabilitated habitats; then between pre-1987 and post-1998 rehabilitated habitats (Table 5, Bonferroni’s adjustment α=0.008).

Table 5: Significance level of the difference in roosting trees used by radio-tracked koalas (N=7) in rehabilitated and undisturbed habitats (Mann-Whitney U tests)

Areas undisturbed rehabilitated pre-87 88 to 97 test test test p values p values p value statistics statistics statistic Rehabilitated pre-87 2704.5 0.001 88 to 97 2473.5 0.003 1767.5 0.450 post-98 505.5 0.809 133.0 0.005 192.0 0.199

We found no difference between roosting trees in areas rehabilitated after 1998 and undisturbed habitats (these two groups were also the most similar in tree species composition, as indicated previously). At the individual animal level, out of the six koalas using both rehabilitated and undisturbed habitats, five were using different species of roosting trees in each (Table 6). The seventh individual used only habitats rehabilitated after 1998, which were the most similar to undisturbed habitats in terms of tree species composition.

Table 6: Significance of the results of similarity between roosting tree species used in rehabilitated and undisturbed habitats, for the six koalas using both habitats (Bonferroni’s adjustment α=0.008) χ2 df p

Binang 21.585 8 0.006 Callitris 31.912 11 0.001 Dakabin 6.130 7 0.525 Jundall 24.175 11 0.012 Mirrigan 18.750 9 0.027 Nareeba 15.608 6 0.016 - 129 -

Koalas used 14 roosting tree species in undisturbed habitats and 13 in rehabilitated habitats, with 10 species common to both habitats (Table 4). The roosting species most often used in undisturbed habitats was E. robusta (25%), and Callitris sp. was the most commonly used in rehabilitated habitats (32%). Two of the three most frequently used trees were the same in undisturbed and rehabilitated habitats (E. robusta and E. racemosa). The number of roosting trees that koalas used multiple times was quite low and similar (χ2=0.567, df=3, p=0.903) in undisturbed (7%) and rehabilitated habitats (5.1%). Mean CBHk of the trees used in undisturbed habitats (138cm, SEM=9.6) was unsurprisingly larger than in rehabilitated habitats (89cm, SEM=5.7, Mann-Whitney U test=4999, p<0. 001), reflecting the size of the trees available.

Diet

From the leaf library, we determined that leaf cuticle characteristics were the same between trees of the same species that had grown in rehabilitated or undisturbed areas (Figure 3, Mann-Whitney U test statistics and p values are given in Table S1).

Figure 3: Mean stomatal lengths and SEM for some of the NSI tree species in the leaf library (black: undisturbed, grey: rehabilitated). Comparison between two individual trees in rehabilitated areas, two in undisturbed areas, 10 stomata measured per sample

This confirmed that we could compare leaf fragments in scats collected in rehabilitated

- 130 - and undisturbed areas (Table 7). Tree species found in koala scats were the same across seasons (ANOSIM, R=0.01, p=0.363) and individual koalas (ANOSIM, R=0.096, p=0.051). As the difference between individuals was almost significant, we compared, for each koala separately, the food trees eaten in rehabilitated habitats to those eaten in undisturbed habitats. Taking koala identity into account, food trees differed between rehabilitated and undisturbed habitats (2-way crossed ANOSIM, R=0.24, p=0.018).

The main species eaten in undisturbed areas were E. racemosa and E. tereticornis (see mean percentages, SD and SEM in Table 7), with many other species well represented. In contrast, in rehabilitated habitats E. pilularis, E. tindaliae and E. racemosa were the three major species eaten. For each of the three rehabilitated habitats and the undisturbed habitats, the percentages of each tree species in the diet were calculated separately (Figure 4). We compared diet evenness based on Shannon evenness index (Shannon & Weaver 1949; Pielou 1975). The diet in undisturbed habitat was more evenly distributed across tree species than the diet in habitats rehabilitated before 1997

(Jundisturbed=0.84; Jpre-1987=0.72; and J88-97=0.73). In turn, the diet in habitats rehabilitated before 1997 was more even than the diet in post-1998 rehabilitated habitats (Jpost-

1998=0.62). However, this could just reflect the difference in the number of group of scats analysed for each habitat (N=4 to N=20): homogenised scat groups contained on average 69% of a single species, and that single species was highly variable between scat groups (see SD in Table 7). This indicates that results based on only four groups of scats (e.g. diet in post-1998) will be more uneven than results based on 20 groups of scats (i.e. diet in undisturbed habitat). Interestingly, the species the most often present in scats was Lophostemon (77.5%), then E. tindaliae (55.0%) and E. racemosa (52.5%, Table 7).

Table 7: Main percentages of tree species eaten by koalas in undisturbed (N=20 groups of scats) and rehabilitated (N=20 groups of scats) habitats (see next page) (NB: E. planchon is used in lieu of E. planchoniana in the following Table and Figure)

- 131 -

Unknown 0.9 1.7 0.2 1.1 0.2 0.4 Melaleuca Angophora 3.3 3.2 0.7 Lopho- stemon E. E. planchon E. E. 14.6 8.5 7.6 1.1 0.9 2.2 tindaliae E. E. pilularis E. E. tereticornis E. E. 9.9 3.8 30.3 26.1 4.3 3.8 7.9 5.7 19.2 17 35.5 25.4 resinifera E. E. robusta E. E. 52.5% 7.5% 22.5% 25.0% 40.0% 55.0% 2.5% 77.5% 10.0% 5.0% racemosa MeanSDSEM 21.9MeanSD 30.6 8.4 6.8SEM 25.2 Mean 20.9 11.4SD 4.7 34.3 7.7 SEM 23.4 23.5 21.2 5.2 32.1 4.2 5.1 29 16.5 15.2 6.5 10.6 2.4 8.5 30.4 21.1 6.8 12.5 3.3 1.9 25.1 3.3 23.4 8.5 4 33.4 17.3 1.9 0.5 22.3 5.3 1 1.7 6 0.2 3.5 5.8 0.3 6.4 1.4 1 0.3 0.2 0.8 1 0.2 0.5 0.9 1.3 0.1 2 0.1 0.3

Species Undisturbed Rehabilitated Total scat containing % of species each

- 132 -

Figure 4: Comparison of the percentages of each tree species in koala diets between undisturbed areas (N=20) and rehabilitated areas (pre 1987 N=12; 1988- 1997 N=4; post 1998 N=4)

On the basis of the tree species found in koala diet, the density of food trees per plot in undisturbed habitat (7.5 SD=8.6) was lower than food tree density in rehabilitated habitats (Mann-Whitney tests, pre-87: U=94, p=0.038; 88-97: U=22, p=0.011; post-98: U=13.5, p=0.005), whereas the food tree density for the three rehabilitated habitats (pre- 87: 23.6 SD=21.2; 88-97: 21.8 SD=24.8; post-98: 26.1 SD=25.1) was similar (Kruskal- Wallis test=0.925, df=2, p=0.630).

Predation risk

No significant difference was found in feral predator indices between rehabilitated and undisturbed areas for any year (Mann-Whitney tests, 2003: U=363, p=0.061; 2005: U=490, p=0.736; 2009: U=775, p=0.673, Bonferroni’s adjustment α=0.008). To compare predator indices between years, disturbed and undisturbed areas were pooled. The predator index in 2009 was significantly higher than in 2003 and 2005 (Mann- Whitney tests, 2003: U=1908, p=0.001; 2005: U=1520, p<0.001, see means, SEM and effect sizes in Table 8, Bonferroni’s adjustment α=0.008).

Table 8: Allen index for feral predator monitoring in 2003, 2005 and 2009 me an SEM effect size Ye ar 2003 2005 2003 0.41 0.08 2005 0.12 0.05 -0.71 2009 1.22 0.19 1.98 9.17 - 133 -

Discussion

Restored habitats have the potential to create sub-optimal habitat for fauna. For instance, it was feared that the restoration of roadsides might create ecological sinks or traps by increasing mortality of butterflies (Ries et al. 2001). Some interesting examples come from the association found between particular native birds and invasive plant species. In these specific cases, restoration projects removing the invasive plants were actually threatening native birds (Jones & Bock 2005; Sogge et al. 2008). Human- modified landscapes in general have often been associated with population sinks and ecological traps (Kerley et al. 2002; Kristan 2003; Battin 2004).

Fauna density and habitat quality are often linked, such that population sinks often have lower animal density than population sources (Goertz 1964; Beshkarev et al. 1994). The correlation between habitat quality and fauna density, however, does not attract unanimous support, and it has been argued that density can be an indicator of habitat quality only if accompanied with data on survival and reproductive rates (van Horn 1983; Robertson & Hutto 2006). For example, ecological traps may be characterised by lower reproductive success, lower survival rate and smaller body size and condition (Lloyd & Martin 2005). However for large and long-lived mammals, data assessing fitness would necessitate a substantial amount of time to collect, so until enough data are available we might have to rely on indirect measures. So far we have gathered information on koala population characteristics for a small sample size and on indirect measures that could influence survival rate, such as availability of food and shelter and predation risk. These are preliminary results and we hope they will trigger further interest and long-term research. On this basis we will now investigate the predictions made, from ecological sink and trap theories, on koala densities, availability of food and roosting trees, koala condition and reproductive rate, and predator density in rehabilitated and undisturbed areas. We highlight the hypothesis on the quality of rehabilitated habitat which is the most strongly supported so far.

Our first hypothesis, that rehabilitated areas could be sinks, does not seem likely. We did not find evidence that habitat was low-quality (in term of tree richness and density, or predators for instance) nor that koalas were of low density, poor condition, or low reproduction in rehabilitated areas. We found a similar density of scats in rehabilitated - 134 - and undisturbed habitats. Moreover, in our study, the radio-tracked koalas were found half of the time in rehabilitated habitats. On the basis of our small sample size, these koalas did not appear to be old, sick or dispersing animals that may have been evicted from primary habitat. Instead, koalas of various ages, of good general and breeding condition were observed using rehabilitated areas. The majority of females using rehabilitated habitats were carrying young (5/6), and the one koala located only in rehabilitated habitats successively carried two back young during the study. As the radio-tracking of koalas occurred on a weekly basis, it opens the possibility that short excursions out of rehabilitated areas were being missed. For the koala found 100% in rehabilitated areas on the basis of VHF data, GPS data (one position recorded every four hours) confirmed that she did not leave rehabilitated areas, even for short excursions in the undisturbed areas (Appendix C). Findings thus far on koala density, their general condition and their reproductive output do not suggest that rehabilitated areas are acting as population sinks, but suggest that koalas choose to establish part of their home range in rehabilitated areas.

On the basis of vegetation characteristics, rehabilitated habitats could indeed be as attractive to koalas as undisturbed areas, if not even more attractive. All rehabilitated habitats used by koalas had higher tree density and richness, and a similar canopy cover (for all habitats rehabilitated before 1997) than undisturbed habitats, along with a higher density of food trees than undisturbed habitats. In particular, the habitats rehabilitated after 1998 contained young, fast-growing trees, which could enhance leaf quality for koalas (Ullrey et al. 1981; Zoidis & Markowitz 1992). Thus, the rehabilitated habitats possessed characteristics likely to make them attractive koala habitats, as evidenced by their use by koalas. This outcome in terms of fauna recolonisation was indeed the goal of the mining company and the reason for their continuous work on improving rehabilitation methodology.

Nonetheless, could the cues that attracted koalas to the rehabilitated areas be disconnected from the net value of rehabilitated habitats for the species in term of reproductive or survival rates, thus opening the potentiality for an ecological trap (Battin 2004; Delibes et al. 2009)? A well-documented example of misleading cues is found in mayflies and dragonflies, for which oil and asphalt roads have superior visual

- 135 - attraction than water, diverting them from their breeding sites (Horváth & Zeil 1996; Horváth et al. 1998; Kriska et al. 1998). So far, we found that the reproductive output of koalas spending some time (and up to 100%) in rehabilitated areas might not be a limiting factor, and below we investigate some indirect measures that could impact on survival rates.

We assessed the likelihood of our second hypothesis, the ecological trap, on the basis of study findings on the processes able to create ecological traps. Ecological traps that provide lesser foraging quality and/or quantity can have dramatic consequences on fauna fitness, like slower growth rates and smaller adult size (Lloyd & Martin 2005) or even starvation (Thomas et al. 1996). As koalas depend on a low-nutrient diet (Tyndale- Biscoe 2005), readily accessible food is crucial to keep the energy ratio of food intake / travel cost in a viable range (Ellis et al. 2009). In this study, food quantity and diversity are unlikely to limit habitat suitability, as native tree density and species richness, as well as food tree density, were higher in rehabilitated than undisturbed habitats. However, on the basis of the diet analysis, koalas may rely on fewer food species in rehabilitated than in undisturbed areas, and this warrants further investigation. The number of tree species eaten by a folivore is critical, and may reflect the physiological constraints imposed on the individual (Wiggins et al. 2003). Indeed, different Eucalyptus species can have different chemical defences, or toxins (Eschler et al. 2000) and the food intake of folivores is limited by these toxins (Wiggins et al. 2003). However, folivores are known to be able to increase food consumption by switching between species with different toxins, which in turn use different detoxification pathways (Wiggins et al. 2006). Thus, if the lower number of species eaten in some rehabilitated areas is confirmed, it would be important to understand if this results from trees in rehabilitated areas possessing a lower level of toxins which could enable a larger quantity of the same species to be eaten, or from other causes. Furthermore, the nutritional value of the foliage for koalas, which depends principally on these toxins and on nutrients (Freeland & Janzen 1974; Pausas et al. 1995; Moore & Foley 2005; Moore et al. 2005), can be influenced by soil characteristics (Coley et al. 1985; Cork & Sanson 1990), even though this is not always the case (Ladiges & Ashton 1974). As the soil in mine rehabilitation is stripped, stockpiled and spread again with the addition of fertilisers, the composition might be different from undisturbed soil (Graham & Haynes

- 136 -

2004; Van Gorp & Erskine 2011). Foliage quality in reconstructed landscapes, where the soil has often been disturbed in some way, is an important path for future research.

Shelter is important to protect animals from predators and against the elements. In particular, thermoregulatory constraints could influence the selection of roosting trees by koalas (Clifton et al. 2007; Ellis et al. 2009). Ecological traps have been known to result from improper or disrupted shelter from predators (Hawlena et al. 2010) or from inappropriate temperature (Packard et al. 1989; Kolbe & Janzen 2002). In our study, rehabilitated habitats seemed to provide suitable shelter. The fact that koalas were not re-using the same roosting trees more frequently in rehabilitated habitats than in undisturbed habitats tends to indicate that shortage of roosting trees was not a problem in rehabilitated areas. Koalas also used the same number of roosting species in rehabilitated and undisturbed habitats. A potential problem could come from the size of roosting trees used by koalas: they were smaller in rehabilitated than in undisturbed areas, but koalas also use small tree size in natural habitats (Matthews et al. 2007).

Increased predation may be the ecological trap most frequently found in the literature (e.g., Dwernychuk & Boag 1972; Gates & Gysel 1978; Shochat et al. 2005; Weldon & Haddad 2005). In particular, landscapes disturbed by humans have created ecological traps by increasing predation for birds (Lloyd & Martin 2005; Robertson & Hutto 2007) and lizards (Hawlena & Bouskila 2006; Hawlena et al. 2010). From our results on predator presence in the area, it appears that predator density is not higher in rehabilitated than in undisturbed areas. However, the general increase in predator index across the years is of concern. Further research should also focus on whether terrestrial predator movements are facilitated in rehabilitated areas. For instance, rehabilitated areas could be linked with increased track density and/or bush penetrability (via a less complex ground layer for instance). Moreover, predation risk for koalas increases with the amount of time spent on the ground (White 1999a). In rehabilitated habitats, trees are smaller, which means that to gain access to the same quantity of foliage, koalas could have to change trees more often than in undisturbed habitats. Alternatively, if koalas feed selectively on tip growth (Degabriele 1981) and the proportion of tip growth tends to be higher (and the tips more accessible) in younger trees, koalas may be able to meet their browse intake more easily in the smaller trees, even though the larger trees in

- 137 - the undisturbed areas have bigger canopies with a higher foliar mass (but proportionately less tip growth). Also, trees are closer together in the rehabilitated areas, which means koalas are likely to have less distance to go from one tree to the next. Increased predation risk is a serious threat, and ongoing research currently attempts to quantify what impact, if any, living in rehabilitated habitats has on koala movement pattern (see Appendix C for details).

On the basis of the preliminary results of this study on population characteristics, food, shelter and predation risk, we postulate that our third hypothesis is so far the most likely: rehabilitation of previously mined areas seems to provide suitable new koala habitat. We found no evidence that the creation of a population sink or an ecological trap is likely. This thus seems to be a timely example of successful koala habitat restoration in the context of the Queensland Government’s commitment to achieve a net increase in koala habitat in Southeast Queensland by 2020 (Queensland Government 2010a). More research, particularly on predation risk and long-term survival rates, is needed to confirm this conclusion.

Several limitations of our study make these results only preliminary. First, due to the amount of time necessary to gather the information presented above, only small sample sizes could be included (see, notably, the number of koalas and the number of scats for diet). Another strong limitation for proving/disproving population sinks or ecological traps is that we did not directly study population demographics (e.g. survival rate) or we had very few individuals (e.g. reproductive output). Finally, for an ecological trap to be appropriately described, data on animal behaviour are necessary (i.e. to prove that a koala chose rehabilitated areas instead of available undisturbed areas). Proper ways to study animal choices include choice experiments or settlement patterns in migratory species (Robertson & Hutto 2006) and might be difficult for koalas. The other limitations can be resolved with longer studies than permitted by our time framework.

Interestingly, our results tend to indicate that koalas could be more adaptable than previously thought (Hume 1990; Cork et al. 2000). Notably, koalas were observed to change their habits to colonise new available habitat. Indeed, koalas were found using rehabilitated habitats aged from 6 to 31 years. These rehabilitated habitats differed in

- 138 - structure (e.g. tree size, bare ground, canopy) and species composition from undisturbed koala habitat. Moreover, radio-tracked koalas roosted in and ate a different suite of tree species in undisturbed and rehabilitated areas. Despite our small sample size (and a certain sex ratio bias), it appears that these comparisons of behaviour were reasonably accurate. Indeed, they were performed for the same individual in both habitats, thus controlling for bias resulting from differences in individual preference. Relative adaptability of koalas to disturbances mirror results found in logging areas (Jurskis & Potter 1997; Kavanagh et al. 2007) and fragmented landscapes (Gordon et al. 1990; White 1999a; Rhodes et al. 2008). Disturbed areas might thus still retain a conservation value for koalas. This supports a non-Manichean view of the landscape matrix, where a landscape can contain many shades of disturbance intensity and habitat suitability (McIntyre & Hobbs 1999).

An earlier study described koala diet and tree use on NSI (Woodward et al. 2008) with rather different conclusions than parts of this study. We found that tree species present differed between rehabilitated and undisturbed koala habitat in contrast to findings from Woodward et al. (2008). However, their sample size was smaller (8 plots rather than the 36 recorded in this study) and their analysis used simple ranking tests while we used statistics specifically designed for community composition analysis (i.e. Bray-Curtis matrices). We also found koalas had a more diversified diet, compared to half of koala diet comprised of E. robusta (swamp Mahogany) as reported by Woodward et al. (2008). This might be linked to the individuals they observed frequenting swamps more often (Figure S1). Another possibility is short-term seasonal variation, as they collected a high proportion of scats during one week (32 out of 80 scats were collected between 14 and 20 February 2005 - Ellis, unpublished data). Yet another explanation could be annual variation. The difference in diet between the two studies indicates that even at a relatively small scale, spatial and temporal variations can dramatically affect our understanding of key ecological resources for koalas. It also stands as a warning about using results of small-scale studies to infer broader habitat requirements. Finally, in contrast to Woodward et al. (2008), we found that the diet of koalas differed between rehabilitated and undisturbed habitats. However, their results were based on a comparison between scats from a single rehabilitated location and scats from 80 locations in undisturbed areas (Woodward et al. 2008).

- 139 -

Conclusion

The results of this study are promising for the restoration of NSI. The mine rehabilitation we studied has won environmental awards and is carried by an environmentally aware company, at the cutting edge of post-disturbance restoration in Australia and internationally (see description of the rehabilitation methodology in the Introduction Chapter). This study strengthens the hypothesis that when the commitment to rehabilitation is strong enough, suitable habitat for fauna can be created, even after the extreme disturbance produced by mining. The success achieved so far demonstrates that using benchmarking to continuously improve rehabilitation is worth the effort. We hope that this will encourage more mining companies to follow this example and set the benchmark for a new legislative framework that includes fauna in assessing rehabilitation success.

As human beings try to salvage some of the ever-increasing areas of land disturbed by their activities, the challenge is not only to put in the substantial effort required to restore fauna habitat attributes, but also to ensure that these areas will support a population with similar reproductive and survival rates as in comparable undisturbed original habitat. This will ensure restoration is increasing wildlife populations and not the reverse (Robertson & Hutto 2007). More research is needed to determine if we are able to succeed in the “acid test” of our ecological understanding of the world (Bradshaw 1983) and the extent to which Nature can be put back together again. In the mean time, it must be emphasised that though habitat restoration is important, it is no substitute for the protection of adequate amounts of undisturbed environment (Young 2000; Hobbs & Harris 2001; Hobbs 2004).

Acknowledgement

Thanks to Dr. Bill Ellis and Dr. Sean Fitzgibbon from the KEG and Olivia Woosnam- Merchez for commenting on earlier drafts. Thanks to Redland City Council for blood test analyses. Thanks to all volunteers for assistance in the field. Thanks for expert assistance from Deidré De Villiers and Dr. Bill Ellis. The first author is supported by an Endeavour Europe Award and an Endeavour International Postgraduate Research Scholarship. Sibelco Australia – Mineral sand provided ongoing support for this

- 140 - research through the provision of logistical support, access to company sites and relevant maps/databases.

This project was carried out under the Queensland Environmental Protection Agency wildlife permits (WISP00491302 and WITK05609808) and the University of Queensland animal ethics (permit project ID 206/07 and 314/08).

Table S1: Results of Mann-Whitney U tests comparing stomata lengths for some of the NSI tree species in the leaf library between trees sampled in rehabilitated areas and trees sampled in undisturbed areas

E. E. E. E. E. racemosa tindaliae robusta planchoniana pilularis Test statistics 194 71 98 183 199.5 p values 0.883 0.631 0.948 0.659 0.989

E. C. Angophora Lophostemo Melaleuca tereticornis intermedia leiocarpa n confertus quinquinervia Test statistics 94 163 164 168.5 90 p values 0.812 0.327 0.341 0.398 0.681

- 141 -

Figure S1: Locations where scats were collected for dietary analysis in this paper and in Woodward et al. (2008) (Some scats, circled in red, were inappropriately classified as belonging to undisturbed areas in Woodward et al. 2008)

- 142 -

Chapter 6

Assessing rehabilitation success for arboreal marsupials, a reflection on the risk of surrogate species

Abstract

With limited time and money, conservationists must often rely on surrogate species as a shortcut to inform their decisions. Controversially, flagship species, i.e. charismatic animals chosen for socio-cultural reasons rather than for their intrinsic biological or ecological qualities, are sometimes used as indicator or umbrella species. We studied a case of fauna recolonisation success after mine rehabilitation where a flagship species, the koala Phascolarctos cinereus, is currently monitored. We investigated the extent to which the koala was a potential surrogate for other arboreal marsupials. We studied the recolonisation of gliders in four rehabilitated areas based on rehabilitation methods, including the provision of artificial nest boxes. Based on a priori models, rehabilitation method was the main explanatory variable influencing glider distribution. Glider density in rehabilitated areas was significantly higher when nest boxes were provided, and marginally higher in rehabilitated areas with improved method. No gliders were found in the best-method rehabilitated area, which was also the youngest, and this habitat may not be suitable for gliders yet. Koalas on the other hand showed no difference in their recolonisation pattern across areas rehabilitated using different methods. These are only qualitative results, however even within this limitation it is obvious that koala and glider recolonisation patterns are inconsistent. This casts doubt on the usefulness of the koala as an indicator species for other arboreal folivores. It underlines that flagship species may only be used as such, unless specific research proves they also possess appropriate indicator or umbrella species characteristics.

Key words: flagship species, indicator species, surrogates, rehabilitation success, mining, and fauna recolonisation.

- 143 -

Introduction

Conservation planning and action are typically limited by time and/or funding. Often researchers and land managers have to rely on shortcuts to measure and achieve their conservation goals (Roberge & Angelstam 2004). Such shortcuts include the use of a few selected species as surrogates for other species, such as indicator, umbrella and flagship species (Simberloff 1998). Indicator species are monitored to describe global ecosystem health or biodiversity and reflect other species’ fluctuations (McKenzie et al. 1992; Lindenmayer et al. 2000). Umbrella species are animals characterised by large home ranges, thus the protection of one umbrella species is thought to extend to the many other animals in its home range (Simberloff 1988; Caro 2003). Flagship species are typically large attractive mammals used to arouse public awareness, interest and donations (Leader-Williams & Dublin 2000; Bowen-Jones & Entwistle 2002; Clucas et al. 2008).

There is much debate about the conservation benefits of using surrogate species (Walker 1995; Lambeck 1997; Andelman & Fagan 2000; Fleishman et al. 2001a; Fleishman et al. 2001b; Rubinoff 2001; Roberge & Angelstam 2004). Notwithstanding the criticisms, surrogates have proven useful in some situations and are still popular (Shrader-Frechette & McCoy 1993; Suter et al. 2002; Caro 2003; Pakkala et al. 2003; Betrus et al. 2005; Rondinini & Boitani 2006; Shokri et al. 2009). This is particularly true when public attention is focused toward a flagship species (Walpole & Leader-Williams 2002). If surrogates are to be useful, they need to be carefully selected to match the task at hand (Tulloch et al. 2011). This selection process should be based on the characteristics that define their classification as an indicator, an umbrella or a flagship species. Yet these three terms are often used interchangeably (Caro & O'Doherty 1999). For instance, because they are charismatic animals, flagship species are often monitored for their own sake, sometimes with the hope that they can act as an indicator or an umbrella species (Simberloff 1998; Frazier 2005). But a flagship species may not be a good indicator or umbrella species; thus its use as such is controversial (Simberloff 1998; Caro et al. 2004; Fleishman & Murphy 2009).

On North Stradbroke Island (NSI), mine rehabilitation endeavours to re-establish native bushland, including endemic flora and fauna species (CRL 2006). Fauna recolonisation - 144 - is a key component of rehabilitation success, but is notoriously difficult to monitor (Bisevac & Majer 1999b) and not every species can be individually scrutinised. In choosing which species to monitor, public support is amongst the many criteria considered (Caro & O'Doherty 1999; Lackey 2004). Koalas, Phascolarctos cinereus, occur naturally on NSI and have drawn much public concern. Koalas are the quintessential flagship species for conservation in Australia (Lunney & Matthews 1997; Jackson 2007). Although they are not amongst the most vulnerable marsupials, they attract a lot of conservation effort and public attention (Tisdell et al. 2006; Tisdell & Nantha 2007). Hence they are typically a priority for inclusion in a fauna monitoring plan. There is then a temptation to use koalas as an indicator species, and assume that if koalas are doing well in an ecosystem then other fauna will too.

Koalas are known to use mining rehabilitated areas of NSI (see Chapter 4, Chapter 5 and Woodward et al. 2008). In this study, we investigated whether the flagship koala could be a suitable indicator for forest-dependent fauna. To do so, we compared recolonisation patterns of koalas to those of other arboreal marsupials which share the forested habitats of NSI with them: the sugar glider Petaurus breviceps, the feathertail glider Acrobates pygmaeus and the squirrel glider Petaurus norfolcensis. Koalas and gliders are dependent on eucalyptus forests (Moore et al. 2004; Tyndale-Biscoe 2005). Koalas rely mostly on eucalyptus foliage (Martin & Handasyde 1999), whereas glider species present on NSI have a more varied diet, including nectar, pollen, sap, gum, honey-dew, as well as insects (Goldingay 1990; Sharpe & Goldingay 1998; Ball et al. 2009). Another important difference in habitat requirements is that gliders, unlike koalas, nest in tree hollows (Lindenmayer et al. 1991; Beyer et al. 2008). Trees develop natural hollows in mature forest (Gibbons & Lindenmayer 2002) and thus are not yet present in recently rehabilitated areas.

We studied the respective presence of gliders and koalas in rehabilitated areas, and compared the distribution patterns of these arboreal marsupials. Due to the different survey methods used to detect gliders and koalas, we provide here a qualitative comparison of their respective recolonisation success.

- 145 -

Materials and Methods

Study site

The surveys took place in rehabilitated sand mines of NSI. Rehabilitated areas can be classified in three groups based on the rehabilitation method used. Pre-June 1987, the rehabilitation was mainly for stabilisation purposes, and involved exotic as well as native plants. After June 1987, a new rehabilitation policy was developed. The use of the seeds from highly competitive Acacia concurrens (Black Wattle) ceased, which improved the restoration of other species such as Eucalyptus and Corymbia. After 1998, the use of another outcompeting tree species, the Allocasuarina sp., was ceased and only endemic seeds collected on the island were used thereafter (Paul Smith, personal communication, 12/02/2008, see Figure 1 and 2).

Survey methods

Measuring koala presence

The presence of koalas in rehabilitated areas was assessed from 44 plots (50x10m) spread across the three different rehabilitation groups (Figure 1). We surveyed six plots from the pre-1987 rehabilitation method, 13 plots from the 1988-1997 rehabilitation method, and 25 plots from the post-1998 rehabilitation method. Koala plots were placed at equal distances along parallel (100m apart) transects with a random start. This design is part of the mine’s long-term vegetation monitoring (Sibelco/CRL, unpublished data). The entire surface of the plot was examined (Ellis et al. 1998) ad libitum for koala faecal pellets (scats, see Chapter 4 for details) and smooth bark trees were inspected for scratch marks. Plots were searched between May and October 2009. The presence of koalas was defined based on two indices: (1) the presence of scats, and/or (2) the presence of scratches, which is a longer-lasting index as scratches remain visible for longer than scats. A potential limitation to koala recolonisation, regardless of rehabilitation quality, is the proximity of populations to source the recolonisation. Ground-checks in the undisturbed areas in the immediate surroundings of rehabilitated areas confirmed koala presence: thus connectivity should not influence the recolonisation outcome.

- 146 -

Figure 1: Localisation of glider and koala surveys across rehabilitated areas, classified by method and provision of nest boxes - 147 -

Figure 2: Selected vegetation characteristics for the three rehabilitation methodologies (based on Chapter 4 and Sibelco/CRL, unpublished data) (a) tree density and richness, (b) density of gliders trees (Eucalyptus, Corymbia, Angophora and Banksia spp, species selected based on Jackson 2000a; Smith & Murray 2003; Dobson et al. 2005), (c) Acacia concurrens density, (d) Allocasuarina littoralis density and (e) percentage of ground cover

- 148 -

Measuring glider presence

The presence of gliders was assessed using infra-red remote-recording photo cameras (Scoutguard 550v) which were baited with containers of honey to attract gliders (Brearley et al. 2010). This was undertaken in January 2010.

The first part of the sampling design involved placing cameras at five different distances from the separation between rehabilitated and undisturbed areas (see complete design in Table 1). This variable will be referred to as the distance to edge (the edge being the limit between rehabilitated/undisturbed areas). Cameras were placed more than 50m apart to ensure some independence of cameras (i.e. that cameras were not triggered by the same glider). Cameras could not be placed further apart owing to the limited size of one surveyed area (7 ha, area pre-1987 with nest boxes, see description below) and the need for consistency across all sites. Cameras were placed to detect:

(1) Glider presence in undisturbed areas We placed cameras at the edge between rehabilitated and undisturbed areas as well as 50m inside undisturbed areas. These cameras were placed to confirm the presence of gliders, thus the possibility of recolonisation (hereafter referred to as controls).

(2) Glider presence in rehabilitated areas Cameras were placed 50m, 100m and 200m inside rehabilitated areas, to examine if and how far gliders penetrated rehabilitated areas. We could not place cameras further inside rehabilitated areas than 200m, because this often represented the most remote part of rehabilitated areas. This is due to the shape of the mine path, which forms a ribbon sometimes not larger than 400m in width, with undisturbed areas on both sides. Cameras in rehabilitated areas should be easily accessible by gliders nesting in undisturbed areas, based on distance travelled by foraging gliders (up to 400m in 15min, Sharpe & Golgingay 2007)

This sampling design was deployed in four different parts of the rehabilitated area, which were similar except for the technique used for their rehabilitation (which is unavoidably confounded with age of rehabilitation). Otherwise, the four different parts of the rehabilitated area experienced similar methods of mining and recreation of - 149 -

landform, low occurrence of roads and human infrastructures, and similar undisturbed surroundings comprising a mix of vegetation communities.

Table 1: Number of cameras used to study glider presence in rehabilitated areas; each camera was run on five consecutive nights

Rehabilitated areas pre- 1988- post- pre-1987 with Distance to edge 1987* 1997* 1998 nest boxes 50 m in undisturbed areas in continuity with 5 5 5 5 rehabilitated areas Edge between undisturbed and rehabilitated 5 5 5 5 areas 50 m inside rehabilitated areas 5 5 5 5 100 m inside rehabilitated areas 5 5 5 5 200 m inside rehabilitated areas 5 5 5 5 TOTAL 25 25 25 25 * in 2 sites, see Figure 1

The four parts were characterised by (1) the three different rehabilitation methods used by the mining company (as described above), and (2) the provision of artificial nest boxes. Twenty-five cameras were placed per area: area pre-1987 without nest boxes (cameras divided between two sites), area pre-1987 with nest boxes, area 1988-1997 (cameras divided between two sites), and area post-98 (Figure 1). Nest boxes found in the pre-1987 area (N=175) were 13cm diameter x 41cm length fibre-reinforced-cement cylinders with an entry hole of 32 to 40mm (James 2007). The nest boxes were deployed in 2007 in a 350m by 200m zone (James 2007). This achieved a density of 25 boxes/hectare, approximating the density of hollow-bearing trees in undisturbed environment (Gibbons & Lindenmayer 2002). We acknowledge that there is limited replication in this experimental design owing to the absence of suitable sites and logistical constraints. In large-scale experiments, however, there are some compelling arguments that choosing the proper scale can sometimes and with some conditions override replication when both cannot be achieved together (Oksanen 2001, see discussion).

Infra-red cameras and honey baits were fixed on a bracket to tree trunks approximately 1.90m from the ground (1.57 to 2.21m). Tree species, tree size (CBH, circumference at breast height) and the aspect of the cameras were recorded. Tree species were selected

- 150 - to be used by gliders (Jackson 2000b; Smith & Murray 2003; Dobson et al. 2005) and camera aspect was randomly chosen. Cameras were set to record images at one-minute intervals when movement was detected. In other glider studies, gliders were tracked over four nights (Brearley et al. 2010); consequently we left each camera running for five consecutive nights. The number of pictures taken (gliders or other animals), the time at which pictures were taken, and the remaining quantity of honey were recorded each day. If the honey pot was found empty in the morning with pictures of animals other than gliders, the night was excluded from analysis and the camera recording was rerun for one additional night. This catered for the eventuality that birds or ants had emptied the honey before night commenced (n=6 events). For each camera, we calculated a variable presence/absence, as a binary variable recording gliders being pictured at any time during the five nights.

In the pre-1987 rehabilitated area with nest boxes, we used a non-invasive snake camera (Ridgid SeeSnake micro 9mm) to monitor the content of nest boxes. The presence of glider nests or gliders was recorded for all nest boxes during one single day to avoid gliders changing locations. We defined a glider nest as a pile of leaves arranged at the bottom of the nest box. Owing to logistic constraints this was conducted only in September 2010.

Data analysis

To study the patterns of koala or glider presence we used generalised linear models (McCullagh & Nelder 1989; Zuur et al. 2009). We used an a priori approach to avoid problems associated with stepwise selection based on p-values (Mac Nally 2000; Burnham & Anderson 2002; Johnson & Omland 2004). Koala response variables (presence of scats or scats and scratches) were fitted by generalised linear models with a binomial distribution, as a function of the two explanatory variable rehabilitated areas and distance from the plot to undisturbed areas. No further koala models were tested (see results). The glider data (i.e. the presence/absence variable, calculated by camera), were fitted by generalised linear models with a binomial distribution.

We constructed a total of six a priori models to investigate the relationship between glider presence and explanatory variables measured. In the first two models, we - 151 - incorporated information about the way the cameras were set up. Model 1 comprised camera height and aspect. Model 2 included the species and size of trees where cameras were fitted. Model 3 included the distance to edge (i.e. undisturbed, edge, 50, 100 and 200m in rehabilitated areas). Model 4 included the four rehabilitated areas (including rehabilitated areas provided with nest boxes). Model 5 included the rehabilitated areas and distance to edge. In Model 6, we integrated an interaction term between rehabilitated areas and distance to edge.

All models were analysed using R (R Development Core Team 2010). No collinearity was found between explanatory variables using variance inflation factors. We estimated the global goodness of fit between the global model (a model containing all variables mentioned above, i.e. two variables for koalas and six for gliders) and the null model. We used a likelihood ratio test, which is preferred to a Wald test in the context of small sample sizes (Pawitan 2001). Spatial correlations in the data and the residuals were analysed with R-package “ncf” (Bjornstad 2009). Spline correlograms presented very little spatial correlation for glider data, none for glider residuals, koala data and residuals. Model validation was based on graphical analysis of the residuals as in Rhodes et al. (2009).

The competing models for a species were compared based on Akaike’s information criterion (Akaike 1973) corrected for small sample size, AICc (Hurvich & Tsai 1989). We calculated, based on AICc, Akaike differences (Δ) between each model and the most parsimonious model; Akaike weights, a measure of the weight of evidence of each model; and the evidence ratios calculated as the ratio of the weight of the most parsimonious model by the weight of each other model (Anderson 2001; Anderson et al. 2001a; Burnham & Anderson 2002). We calculated the relative importance of each variable by adding up the Akaike weights of all models in which the variable appeared (Burnham & Anderson 2002).

For gliders, we also compared results across the different types of rehabilitated areas (pre-87 with and without nest boxes, 88-97 and post-98), using pairwise Mann-Whitney U tests. These statistic tests were performed using PAWS Statistics 18.0 (IBM 2009). Significance was taken to be p <0.05.

- 152 -

Results

Koala patterns

On the basis of the goodness-of-fit of our koala models, we found no significant relationship between the distance to undisturbed areas or the type of rehabilitated areas (pre-87, 88-97 and post-98) and measured signs of koala presence (scats: χ2=4.75, p=0.09; scats and scratches: χ2=1.20, p=0.54). This pattern was also confirmed when the AICc values of models for (1) scats and (2) scats and scratches were each compared to null models (scats: evidence ratio for model against null=1.13, scats and scratches: evidence ratio for null against model=5.21).

Glider patterns

Validation of the study design

Over the 500 camera nights, only once was a glider detected for the first time on the last night, confirming five nights of monitoring was an appropriate length. Two types of gliders were easily identified: the feathertail glider and a much larger type of glider which could have been either sugar or squirrel glider, however, it appeared too difficult to distinguish the glider species from photos (Figure 3).

(a) (b)

Figure 3: Two gliders recorded by infra-red remote recording photo cameras and classified as (a) squirrel glider and (b) feathertail glider

- 153 -

For simplicity in the rest of the chapter we will call the larger glider “squirrel glider” as they are by far the most common on the island (James 2007). We will also use the general term “glider” to refer to all species combined. Gliders visited the camera between 7pm and 5am (Figure 4). The cameras were also triggered 24 times by birds, three times by rats and once by a koala.

12 feathertail glider 10 squirrel glider 8 6 4

Number of visits of Number 2 0

Time of visits

Figure 4: Time and number of visits to cameras by gliders (for this figure all visits more than 20 minutes apart are included)

Results of the study

In contrast to koalas, the goodness-of-fit of the global glider model (including all six variables) was significantly better than the null model (χ2 =33.2, p<0.001). We ranked all models based on AICc (Table 2). On the basis of AICc weights, two models were supported (sum of AICc weights almost one). The variable rehabilitated areas was included in the two best models, thus its relative importance was close to one. When incorporated with other variables, distance to the edge had a relative importance of 0.36 (Table 2). However when it was investigated in isolation it had no support (AIC weight=0). We found no relationship between camera set-up variables (height, aspect, tree size and species) and gliders’ presence (Models 1 and 2, Table 2).

- 154 -

Table 2: Glider models ranked by AICc K: number of parameters, AICc: Akaike’s information criterion corrected for small sample size, Δ AICc: AICc differences, see descriptions in text

Model explanatory variables K AiCc Δ AICc AICc evidence we ight ratio 4 rehab area 4 45.22 0.0 0.64 1.0 5 rehab area + edge distance 5 46.45 1.2 0.34 1.9 6 rehab area * edge distance 8 52.30 7.1 0.02 34.5 3 edge distance 2 68.75 16.4 0.00 3724.9 2 tree species + CBH 3 67.08 21.9 0.00 55708.0 null model 1 67.26 22.0 0.00 61130.8 1 camera height + aspect 3 67.76 22.5 0.00 78360.8

Thus, the pattern of glider presence differed between rehabilitated areas (see Table 3 for details). In particular, the presence of gliders was higher in rehabilitated areas with nest boxes compared to the other areas (Table 4, compared to post-98 or pre-87 rehabilitated areas: U=37.5, p<0.001; compared to 88-97 rehabilitated areas: U=67.5, p=0.031). Glider presence was also higher in the area rehabilitated in 88-97 than in the pre-87 rehabilitated areas without nest boxes or the post-98 rehabilitated area (U=82.5, p=0.035).

Table 3: Location and species of gliders pictured by cameras

Rehabilitated areas distance to glider total total edge species nights Pre-87 rehabilitated areas undisturbed squirrel 1 feathertail 1 edge squirrel 2 4 Pre-87 rehabilitated areas 50m in rehab squirrel 10 with nest boxes 100m in rehab squirrel 9 200m in rehab squirrel 10 29 88-97 rehabilitated areas edge feathertail 2 squirrel 3 50m in rehab feathertail 1 200m in rehab feathertail 1 squirrel 5 12 Post-98 rehabilitated all none 0 0 areas

- 155 -

Table 4: Mean glider presence and standard errors of the means between the different rehabilitated areas (Mann-Whitney U tests)

Rehabilitated areas mean SEM U p value compared with Pre-1987 0.00 0.00 Pre-1987 with nest boxes 0.67 0.13 37.5 <0.001** pre 1987 37.5 <0.001** post 1998 67.5 0.031* 1988-1997 1988-1997 0.27 0.12 82.5 0.035* pre 1987 82.5 0.035* post 1998 Post-1998 0.00 0.00 112.5 1 pre 1987

The daytime monitoring of nest boxes (in the pre-87 rehabilitated area with nest boxes) confirmed that gliders were nesting in the rehabilitated area rather than simply visiting it during night time forays from undisturbed areas. Of the 132 nest boxes checked for glider use, glider nests were found in 94 boxes and squirrel gliders in 16 boxes (Table 5). We estimated glider density to be of 2.29 gliders per hectare. Owing to the non- invasive method used, we were unable to determine the number of gliders per box. Glider density was instead measured based on the number of nest boxes containing at least one glider in the 7ha of pre-87 rehabilitated areas with nest boxes. This estimate is probably conservative given that previous monitoring found that 30% of boxes contained more than one individual (James 2007).

Table 5: Glider monitoring of 175 nest boxes deployed in 2007 in a rehabilitated area (2007 data from James, 2007) 2007 2010 Number of nest boxes still up the trees 144 132 Number of nests inside nest boxes 25 94 Number of nests with at least one glider 10 16 Minimum density of gliders per hectare 1.43 2.29

Discussion

On the use of flagship species as indicator species

Here we demonstrate that the flagship koala is not a good indicator species for other arboreal marsupial species such as gliders. Although both koalas and gliders recolonised rehabilitated areas, they clearly had different recolonisation patterns. Koalas were

- 156 - detected in all types of rehabilitated areas, while gliders were not. No gliders were found inside either the pre-87 rehabilitated areas without nest boxes (NB: some were found at the edge of undisturbed/rehabilitated areas) or inside the post-98 rehabilitated areas. This is in line with some studies that found one species can be a poor indicator for other species because the species were limited by ecological factors that were irrelevant to the proposed indicator species (Roberge & Angelstam 2004). Even for species chosen in the same guild, and sharing the same ecological demands as the proposed indicator species, one species need not be a good indicator for the entire guild (Cushman et al. 2010).

Mammal, bird and butterfly species, as charismatic flagship animals, are often proposed as indicators for other species (WallisDeVries & Raemakers 2001; Mac Nally & Fleishman 2002; Halme et al. 2009; Sebastiao & Grelle 2009). This stems from the reasonable statement that not all species of interest can be monitored and managed individually (Wiens et al. 2008). The development of methodologies to select appropriate indicators is an area of active research (e.g., Favreau et al. 2006; Azeria et al. 2009; Tulloch et al. 2011). Yet in the possible criteria to select an efficient indicator, being charismatic is never considered. Some have hypothesised that the acclaim of charismatic species as indicators has resulted from being beloved by the public but has not been supported by data (Andelman & Fagan 2000; Fleishman & Murphy 2009). For instance, previous studies on birds have found no indicator species even inside groups of similar ecological characteristics (Cushman et al. 2010). Likewise, the usefulness of flagship butterflies (Fleishman & Murphy 2009) and mammals (Williams et al. 2000; Caro et al. 2004) as indicator species has been questioned.

One frequent explanation put forward for poor performance of indicator species is competition between species (Cushman et al. 2010). Niche theory predicts that two species cannot have exactly the same ecological requirements (Hutchinson 1957; Pulliam 2000). Thus, even very similar species are predicted to specialise or displace other species to decrease competition (MacArthur 1967). For instance, a study of closely-related arboreal marsupials described their slightly different requirements (Lindenmayer 1997). Consequently, their pattern of occurrence was different, underlining that none of the species could be used as an indicator species for the

- 157 - presence of other members of the arboreal marsupial guild (Lindenmayer & Cunningham 1997; Lindenmayer et al. 1999). This is even more so the case for koalas and gliders, which despite sharing the use of some tree species, depend on different parts of the trees, decreasing direct competition and possibly resulting in some differences in recolonisation patterns.

Consequently, koala return can only be used as an indicator of rehabilitation success regarding koalas, not of rehabilitation success of arboreal marsupials in general. Koalas will always be the focus of public concern and a priority in fauna monitoring in rehabilitated areas, as well as in any conservation areas where they occur. This is probably worthwhile as long as the flagship koala is not fitted into the umbrella or the indicator suit.

Other consequences of this study for glider recolonisation

This study is the first to describe glider recolonisation of mine rehabilitation. Some rehabilitation characteristics seemed beneficial for gliders and are given in the following sections. Recolonisation of mining rehabilitated areas by gliders seemed enhanced by the provision of nest boxes (Tables 3 and 4). This is in line with the literature on arboreal marsupials, where nest boxes have been found to support glider population in forests devoid of natural hollows (Smith & Agnew 2002; Beyer & Goldingay 2006; Goldingay et al. 2007). As natural hollows might take more than 120 years to develop naturally (Gibbons & Lindenmayer 2002), providing nest boxes in rehabilitated habitat is a useful management option. However, a successful nest box program needs maintenance, mainly against three ailments: attrition, pest invasion and tearing-off (Beyer & Goldingay 2006; Lindenmayer et al. 2009). In our study area, the key management issue is nest box tearing-off (Table 5), the other two ailments being rare. In the long term, management of rehabilitated areas after mine closure should thus include the provision for a nest box maintenance program, if return of hollow-dependent fauna is wished. In addition, when trees reach a sufficient size the development of natural hollows can be accelerated by different techniques (e.g., inoculation with specific fungi, cutting holes with chainsaws, Brennan et al. 2005).

The improvement of rehabilitation methodology could enhance the value of - 158 - rehabilitated areas for gliders. This could explain why, when comparing rehabilitated areas without nest boxes, no gliders were found in the pre-87 rehabilitated areas, while some were found in the 88-97 rehabilitated areas. The tree balance between trees preferred or avoided by gliders changes with the methodology. Preferred glider trees are Corymbia, Eucalyptus, Angophora and Banksia spp.; while Allocasuarina spp. and non- pinnate-leaved Acacia spp. (such as the ones present in rehabilitated areas) are negatively correlated with glider densities (Smith & Murray 2003). We found that preferred glider trees increased with the improvement of methods (Figure 2b) while globally Allocasuarina and Acacia spp. decreased (Figures 2c and 2d). (NB: The Acacia trees seem to be higher in the 88-97 rehabilitated areas than in pre-87 rehabilitated areas because most Acacia trees initially present in pre-87 rehabilitated areas are dying or dead and get replaced by other species.) The 88-97 rehabilitated areas might also provide better insect diversity than the pre-87 rehabilitated areas, based for instance on the plants available on the ground layer (Figure 2e), as ground dwelling insects and ground layer plants in mine rehabilitated areas can increase in parallel (Moir et al. 2005). Furthermore, artificially high tree density greatly reduces tree girth and growth rate, and delays the occurrence of large boughs and tree hollows by decades (Vesk et al. 2008). This means rehabilitation methodology should endeavour to continue decreasing tree density (Figure 2a).

Consequently, on the basis of tree density, richness, and ground layer complexity, the newer rehabilitation methodology should create even better glider habitat than older approaches. It is surprising that no glider was detected in post-98 rehabilitated areas. Although this rehabilitated area is young, gliders have been found to recolonise two- year-old plantations in other parts of Queensland (Smith & Agnew 2002). Further research is needed to understand the absence of gliders in the post-98 rehabilitated areas.

Limitations of the study design

We encountered important limitations in our survey. First, the glider cameras placed in each different rehabilitated area can only be considered as pseudo replicates (Hurlbert 1984). Our design however falls into the category where the scale of the question studied (fauna recolonisation) makes it difficult to have true replicates (Oksanen 2001; - 159 -

Cottenie & De Meester 2003). For instance, there is so far only one area providing nest boxes, which means no adequate replication can be provided for this area. One solution proposed is to replicate the controls, which could be possible in our experiment but has other drawbacks (described in Oksanen 2001). The other solution is to have pseudo replicates and accept the limitations of the design. Particularly, generalisation to other populations is unwise and more definite results will rise from meta-analysis of many (replicated or not) experiments (Oksanen 2001; Cottenie & De Meester 2003; Oksanen 2004).

Second, two different methods and sampling designs were used to monitor koala and glider presence in rehabilitation. We could not use the same sites to detect koala and glider presence owing to the use of two different methods which had their own constraints (e.g. the cameras needed to be checked every day which restrained our ability to spread them out as much as koala sites). This meant that results could only be directly compared at the scale of area (i.e. we did not use koala presence as an explanatory variable in glider presence models). Spotlighting may have provided directly comparable results between koalas and gliders, but low koala density on the island would make such survey methods extremely time-consuming and thus not compatible with our framework. Indeed, we showed that five observers can walk up to several days in strip transect before encountering one koala (see Appendix A).

It is possible that some of the presences we recorded were of animals dispersing through the rehabilitated areas and not of resident animals (for koalas, see discussion Chapter 4). It was not possible to avoid this risk for gliders as dispersal could be occurring most of the year because births happen through an extensive period (in the Brisbane area most of the year, from March to November (Sharpe & Goldingay 2010), throughout the year or bimodally in other parts of Queensland (Quin 1995; Jackson 2000a; Millis & Bradley 2001; Lindenmayer 2002) and the young disperse after a period of 10 to 12.5 months (Quin 1995). However, the high density of nest boxes occupied and their long-term occupancy (2007 till 2010) makes us believe the gliders surveyed were part of an established population, at least in the rehabilitated area with nest boxes.

The presence of glider population sources and connectivity between these population

- 160 - sources and rehabilitated areas could have biased the results. Indeed, gliders were absent from the post-98 rehabilitated areas, but they were also not found in the undisturbed area directly adjacent. This could suggest that gliders cannot access post-98 rehabilitated areas. However, in the pre-87 rehabilitated areas with nest boxes, the fact that no gliders were detected in adjacent undisturbed areas did not prevent us from detecting gliders inside the rehabilitated area.

Conclusion

“Flagship species” is a popular term and is frequently used in the media and by practicing conservationists (Fleishman & Murphy 2009). The uses and misuses of the term flagship species in the literature underline the advantages and drawbacks of this concept. Again, flagship species are, by their definition, chosen on socio-cultural and strategic grounds, rather than being selected based on biological or ecological characteristics like indicator or umbrella species (Leader-Williams & Dublin 2000). The usefulness of flagship species relies then on them being used inside the boundaries of this definition (Walpole & Leader-Williams 2002; Frazier 2005; Fleishman & Murphy 2009).

Indeed, flagship species remain a very good tool to raise awareness. In our case study, koalas are very appropriately used by activists to raise public concern about potential adverse impacts of mining. Wanting to monitor and protect a flagship species based on ethic, aesthetic and human values is justifiable. Ultimately, the worth of restoration should be adjudicated not only on ecological but also on historical, social, cultural, and moral grounds (Higgs 1994). But promoting flagship species to umbrella or indicator species without prior scientific knowledge that they are effective in this role, will undoubtedly lead to some inappropriate management decisions (Hilty & Merenlender 2000; Frazier 2005). Thus, a potential danger lies in management and conservation plans being pushed by political and public agendas into focusing on flagship species, while more biologically and ecologically relevant species are being neglected.

Acknowledgement

Thanks to Dr. Adrian Bradley and Olivia Woosnam-Merchez for commenting on earlier

- 161 - drafts. The first author is supported by an Endeavour Europe Award and an Endeavour International Postgraduate Research Scholarship. Sibelco Australia – Mineral Sand provided ongoing support to this research through the provision of logistical support, access to company sites and relevant maps/databases.

- 162 -

Chapter 7

Conclusion

Restoration is becoming an increasingly important part of our attempt to protect biodiversity (Young 2000). Yet it has been highlighted that restoration attempts will always be more costly and less effective than minimising – or, when possible, preventing – the initial environmental damage (Young 2000; Hobbs & Harris 2001; Hobbs 2004). Legacies from major disturbances are yet to be fully understood. For instance, the characteristics and resilience of ecosystems might be affected in the long term by major disturbances (Aronson & Floc'h 1996; Foster et al. 1998; Bellemare et al. 2002). However, a lot of damage has already been done. Furthermore, it is unreasonable to pretend we can stop further damage in the near future, not unless population growth and, perhaps more importantly, its aspirations to unrestrained consumerism are under control. Meanwhile, we have the critical task of ensuring restoration is delivering up to all stakeholders’ expectations.

In this final chapter I consider the results of my research in a holistic manner and ask how good the mine rehabilitation is in my study system, what should the criteria for successful rehabilitation be in regards to fauna and how the science of restoration ecology can be further progressed through collaborative research such as mine.

How successful is mine rehabilitation in my study system?

The assessment of success in any project can be influenced by many criteria including emotional, philosophical or ethical values. In the case of success regarding wildlife conservation, it is essential to avoid assessing the landscape from our biased anthropogenic point of view, and to let fauna tell us how successful restoration is (McIntyre & Hobbs 1999; McAlpine et al. 2002; Brooks et al. 2004; Manning et al. 2004).

In Chapter 2, I reviewed studies that have described the fauna story in rehabilitation success. My Australian-focussed review showed that rehabilitation of mines has had - 163 - variable successes in providing new fauna habitat. It also showed that our knowledge of fauna responses to rehabilitated areas remains limited, particularly regarding ecological data. Bearing this in mind, the rest of my thesis set out to add a supplementary piece to the puzzle by describing the fauna recolonisation of NSI mine rehabilitation, through the example of arboreal marsupials. As a result of my investigation, a piece of knowledge is added, some management implications are proposed (Table 1), and many more questions are raised (Table 2).

Research areas that can be considered as priorities are, first and foremost, how to integrate fauna into the mine closure process by the implementation of fauna criteria (see below). In the process, more research could be done to ensure (1) that fauna populations occupying rehabilitated areas have survival and reproductive rates comparable to those of populations in undisturbed areas, and (2) that taxa which are currently overlooked (e.g. snails, bats) are not negatively affected by mining and mine rehabilitation. If some taxa are found to have difficulties recolonising rehabilitated areas, research should investigate potential causes and develop corrective actions.

Such corrective actions could include: - the addition or improvement of key missing habitat characteristics, if fauna species rely on these essential habitat characteristics (e.g. log piles, tree hollows, specific flora species providing food or shelter), - feral predator control if they are overabundant, - thinning and/or burning the vegetation if tree density in rehabilitated areas is superior to natural density, - or even translocation of fauna unable to recolonise, if connectivity is the identified problem.

- 164 -

Table 1: Key management implications of this thesis per chapter

Chapter key management implications 2 monitoring cannot rely on one group as a surrogate for all fauna, because response of fauna groups to rehabilitation is variable improvement in best practices is mirrored by improvement in fauna recolonisation and should be continued data on fauna recolonisation of mine rehabilitation should be made more available through publication of mining reports and in scientifuc journals long-term monitoring and adaptive management are necessary, as the evolution of rehabilitation toward reference sites can progress up to a point then stop better understanding of rehabilitation success can be reached by identifying species and comparing community composition instead of adding them interchangeably in higher taxa indices (e.g., family richness) 3 scat survey is a useful, cost effective and efficient way to monitor koala presence if scat surveys are to be compared between different habitats, scat detectability and decay rate must be accounted for: one option is to develop correction factors for ground layer complexity and wet habitats

4 flora criteria cannot be used to assume fauna follows the same trajectory

for fauna groups of interest, targeting that group is required 5 good practices in rehabilitation should be encouraged, as they successfully promote fauna return, as demonstrated by our model, the koala feral predator control should be considered as part of the rehabilitation process 6 providing nest boxes is a useful rehabilitation technique for hollow- dependent fauna until natural hollows develop species of interest should be monitored directly (by direct or indirect sign surveys) if proven relevant indicators are not available maintenance and replacement of nest boxes to maintain a threshold density, even after mine closure, can be a good candidate for fauna completion criteria monitoring nest boxes could also be part of fauna completion criteria regarding hollow-dependent fauna. However, photo-id is a good alternative and can be used even where no nest boxes are deployed (thus allows comparison to reference sites in undisturbed areas), as well as detect not only species that use the nest boxes but others (as nest boxes might not be used by some species that nonetheless use rehabilitated areas)

- 165 -

Table 2: Some areas of interest for further research derived from this thesis

Chapter further directions for research

2 fate of under-studied taxa in rehabilitated areas (e.g., snails, bats) collect a more representative sample of mining companies (i.e., send questionnaires)

demographic, behavioural, genetic data on fauna in rehabilitated areas

feral species: opportunist early colonisers or enduring colonisers? 3 ways to integrate imperfect detectability in the calculation of fauna distribution (such as with zero-inflated models) relevant to low density fauna sampling strategies to account for the high variability of scat decay rate between plots, as this can bias distribution or density survey results

4 framework for selecting relevant fauna criteria 5 density, reproduction, survival and behaviour (such as choice of home range location) of koalas in rehabilitated areas compared to undisturbed areas: all these are the basis of definitely proving/disproving ecological traps fine scale movements of koalas between rehabilitated and undisturbed habitats, as an estimation of time and energy spent in travelling predation risk in rehabilitated areas compared to undisturbed areas (for instance, foraging ability of predators) leaf quality (water, nutrients and secondary compounds) in rehabilitated areas compared to undisturbed areas 6 density, reproduction, survival of gliders in rehabilitated areas compared to undisturbed areas

diet of gliders in rehabilitated areas compared to undisturbed areas behaviour, particularly recolonisation pattern after nest-boxes are provided, dispersal, family group size, number of nest-boxes per family group genetics of gliders living in rehabilitated areas compared to undisturbed areas extensive studies of gliders in rehabilitated areas that are not provided with nest boxes would be useful to understand glider recolonisation when no nest boxes are added, as nest boxes are difficult to install and maintain and it might not be the preferred option to install them over the entire rehabilitated area

- 166 -

Chapters 4 and 5 investigated the recolonisation by koalas based on scat surveys (after having corrected this indirect method for detectability and decay rate based on Chapter 3) and on direct observations. The scat survey used in vegetation monitoring plots was conducted to model koala presence based on environmental data. This did not yield the expected results but showed that many plots contained koala scats, far more than previously expected. This is the first time that areas of mine rehabilitation are shown to be used by koalas more than on an opportunistic basis (Simpson 1998; Woodward et al. 2008). Chapter 5 confirmed this, by providing the surprising result that koalas captured nearby the rehabilitated areas all subsequently entered them. Indeed, the collared koalas in my study could spend between 25-100% of their time in rehabilitated areas. The facts that koalas could spend a substantial amount of their time in rehabilitated habitat, and were in good condition and breeding (one female found 100% of the time in rehabilitated habitat raised two back young during the study) constitute strong evidence that rehabilitation can provide adequate koala habitat. To what degree of certitude the mining company will need to prove this is still unknown. One possible criterion would be to compare scat density surveys pre- and post-mining. However there is a lack of pre- mining data on koala densities in different parts of the island. An alternative would be to compare scat density surveys between undisturbed and surrounding rehabilitated areas. This, however, poses some problems given that mining occurs on dunes and that the majority of surrounding areas occur on lower parts of the island where the vegetation types are likely to be different and thus koala presence is difficult to compare. Long-term studies such as this thesis, in addition to previous monitoring beginning in 2002, provide a more detailed answer to the fate of koalas in rehabilitated areas. Even then, ten years of data collection is relatively short for a long-lived species making it difficult to collect survival and reproduction rates to allow comparisons between undisturbed or rehabilitated areas. Survival and reproduction rates would be the best way to answer the ultimate question: “is rehabilitation successfully increasing koala habitat and ensuring long-term population viability?” Definite success criteria for koalas would require extensive consultation with all stakeholders on Stradbroke Island.

In Chapter 6, glider recolonisation was studied with camera traps, and again provided encouraging results that mine rehabilitation on NSI can be effective for arboreal marsupials. On average, 14.3% of the cameras in rehabilitated areas were triggered (the

- 167 - percentage of camera-nights with gliders), with high variations between areas. Although the study was not designed to compare glider density between rehabilitated and undisturbed habitats, results from undisturbed habitats from this study and others indicate that densities of gliders in rehabilitated areas could at least achieve those of undisturbed areas (Table 3). Worth noting, the unexpected low number of cameras triggered in undisturbed habitat might be a bias from my design. Indeed, my undisturbed sites were chosen because they were adjacent to the rehabilitated areas, and not because they were optimal glider habitat. However, trap success data for optimal glider habitat in undisturbed areas on NSI is available (Table 3, Sarah Bell, unpublished data) and is lower than camera success in rehabilitated areas in my study. Caution though needs to be employed, as the results may not be strictly comparable. Indeed, in my study gliders were attracted by honey and then pictured, whereas in Bell (unpublished data) gliders were attracted by honey and then trapped, which means probably fewer gliders were detected in Bell (unpublished data).

Table 3: Camera or trap success in two different studies of gliders on NSI

Study method landscape number number of sites success success in sites of sites with gliders all sites with gliders Bell traps undisurbed 6 4 3.5% 10.4% (unpublished) undisurbed 6 2 1.0% 2.0% This study cameras rehabilitated 6 3 14.3% 28.6%

Finally, I found a minimum density of 2.29 squirrel gliders per hectare in the rehabilitated area that provides nest boxes. Typical densities of squirrel gliders in natural environments vary between 0.01 to 1.6 squirrel glider per hectare (Lindenmayer 2002; Smith & Murray 2003; Sharpe & Goldingay 2010). Again these estimates are based on different survey methods and therefore warrant caution, but they reinforce the same encouraging message that arboreal marsupials are found in rehabilitated habitats at densities not too dissimilar to undisturbed habitats. My survey method was useful in detecting gliders. This method could be used by mine managers wanting to compare density estimates between rehabilitated and undisturbed areas and ensure that rehabilitation is successful for gliders. Other possibilities include spotlighting, collection of hair using hair tubes or even detecting glider calls by remote recording - 168 - systems. Monitoring nest boxes would be useful to gather more information on population structure, but cannot be used to compare rehabilitated and undisturbed areas because alternative hollows are present in natural habitats making gliders less dependent on nest boxes (Smith & Agnew 2002). Similar to koalas, some problems in assessing rehabilitation success toward gliders remain unresolved. In particular, difficulties associated with developing appropriate final targets not only because all stakeholders have to agree but also given the lack of pre-mining records. As for the koala, survival and reproduction rates would have to be compared between rehabilitated and undisturbed areas to gather the ultimate proof of population viability. While the timeframe for collecting survival and reproduction data for gliders would be more feasible than for koalas, the logistic and budget involved would be large. Without strong incentives it is uncertain how many mining companies would be willing to conduct such studies. This is where collaborative projects become useful (see “A synergetic opportunity” below).

On the basis of this study, arboreal marsupials present on NSI are able to recolonise rehabilitated areas. Koalas seem more spread throughout rehabilitated areas, whereas gliders are mostly present where nesting boxes are provided. For arboreal marsupials, the mining company seems to be on the right path to fulfil its fauna criteria, as arboreal marsupials most likely “will return to levels equivalent to other similar habitats on North Stradbroke Island” (Sibelco/CRL completion criteria as stated in Environmental Authority No. MIM800088202, see Table 3 in the Introduction Chapter). For koalas, this completion criterion might be close to being reached for several parts of the rehabilitated area. For the hollow-dependent gliders, this might be the case either rapidly if nest boxes are provided, or in the long term when natural hollows develop. However, the next question relates to the appropriateness of this fauna criterion. The mining company on NSI is already a precursor in the sense that it is one of the few mining companies to date that has included a fauna criterion in their rehabilitation practices. As highlighted in Chapter 5, however, ensuring fauna groups are present in mining rehabilitated areas is a necessary first step but may not be the only step. Ultimately, success for fauna in rehabilitation will be to return to density, reproduction and survival levels equivalent to those of other similar habitats on North Stradbroke Island.

- 169 -

Developing generic fauna criteria to measure rehabilitation success

The mining industry still seems to suffer from a general lack of mandated standards and systematic monitoring approach regarding fauna recolonisation (Thompson & Thompson 2004). For instance, in the forestry industry worldwide, fauna criteria include listing forest-dwelling species in the logging area together with their protection status, and monitoring any changes in their distribution and density (Montréal Process Working Group 2007; Anonymous 2008). The Montréal Process includes countries such as Australia, USA, Canada, China, Russia, Argentina and many more, and all agreed on the necessity of including fauna in their assessment of sustainable management of forests. Thus, it seems straightforward that fauna should be included in a sustainable mining industry.

Furthermore, it is apparent that flora criteria used so far by mining companies to measure rehabilitation success cannot be consistently used as surrogate for fauna (Chapter 4). Thus, it is time that fauna be integrated in mining monitoring. As not all fauna can be monitored, the important question is how to select which species to study. This selection is critical because no single fauna group is suitable for use as an indicator for all fauna groups’ recolonisation (see Chapter 2 and Nichols & Nichols 2003).

As previously discussed (see Discussion of Chapters 2, 4 and 6), monitoring fauna success may be addressed by monitoring a selection of single species of particular interests. Flagship species (Walpole & Leader-Williams 2002), keystone species (Simberloff 1998), invasive species (Palmer et al. 1997), species particularly sensitive to threats (Lambeck 1997), umbrella species (Caro & O'Doherty 1999) have all been proposed. In this thesis, I mainly focused on single species research, through the flagship koala.

In the rehabilitation of disturbed land, the goals are often to re-establish a functional ecosystem, and it has been argued that monitoring should be less interested in the fate of individual species than processes and functions (Ehrenfeld 2000). Thus, community and ecosystem functions and services are becoming very popular areas of research (see, for example, Schläpfer 1999; Groot et al. 2002; Balvanera et al. 2006). However, some have cautioned against a total shift from species-based science towards a process-and- - 170 - function-based science (Goldstein 1999; Moore et al. 2009). One of the reasons against a complete replacement is that ecosystem functions and processes are still subject to animated debates in the scientific community (McIntyre et al. 2002). Moreover, ecosystem functions and processes are not straightforward to communicate to the general public (McIntyre et al. 2002). The main concern of a process-and-function- based science is that conservation actions could become focused on communities or ecosystem functions and processes for their own sake and lose the link with the fundamental reason for conservation: the protection of individual species (Goldstein 1999). Indeed, the caveat of distancing oneself from species-specific information is that sometimes indices that are monitored no longer reflect the original subject of interest (Goldstein 1999). Far from being mutually exclusive, research on both species and ecosystems is needed to successfully understand and manage the multiple ecological scales of restoration ecology (Lindenmayer et al. 2008a).

Part of the assessment of rehabilitation success for fauna should thus use the ecosystemic approach, which focuses on using fauna communities, either for their intrinsic values or more frequently to reflect broader ecological processes and functions (Walker 1992; Caro & O'Doherty 1999; Andres & Mateos 2006). Invertebrates in general, and ants in particular, are the most commonly proposed indicators for rehabilitation success in regards to specific ecosystem functions (e.g., Majer 1983; Majer 1984c; Bisevac & Majer 1999b; Andersen et al. 2002; Longcore 2003; Andersen et al. 2004; Ottonetti et al. 2006). Others have proposed the use of Acarina (Cuccovia & Kinnear 1999), Collembola (Greenslade 2007) or grasshopper assemblages (Andersen et al. 2001). The advantages of invertebrates as monitoring tools include (1) their ubiquity, abundance and diversity which aid toward robust data analyses and more meaningful trends; (2) their ease of sampling (although identification can take a long time, Rohr et al. 2007); (3) their involvement in many ecosystem processes; and (4) the fact that their community structure responds to environmental stress and thus provides a measure of ecosystem changes (e.g., Majer 1983; Andersen 1994; Andersen & Sparling 1997; Andersen et al. 1998; Majer & Brown 1998; Thoday 1998; Bisevac & Majer 1999b). Other monitoring tools using fauna communities involve functional groups (Andersen 1997; Ottonetti et al. 2006), trophic groups (Corbett 1997) or fauna of higher trophic levels (Knight 1999). Although all of these have been proposed as indicators of

- 171 - ecosystem rehabilitation, more research is needed to define the specific range of their usefulness (e.g. which specific community can be an indicator for each ecosystem function).

Some choices of what species to monitor will inevitably be politically driven, especially for species that are highly valued by the public. Indeed, the success of any ecological restoration project ultimately rests on its social context as well as good science. The choice of the species to monitor has to integrate historical, social, cultural, political, aesthetic, philosophical and moral aspects (Higgs 1997; Hobbs 2007). Chapter 6 warns us to be cautious on how we subsequently use these species, as they may not be the best measures of the effectiveness of rehabilitation for other fauna groups (see Block et al. 2001).

One additional difficulty in our attempt to select criteria representing rehabilitation success for fauna results from the dynamic nature of ecosystems and the interactions of rehabilitated areas with the surrounding landscape (Wallington et al. 2005). One important consequence is that measures of rehabilitation success have to integrate the natural variability in fauna measures across space and time, in both undisturbed references and rehabilitated habitats (Weaver 1995; White & Walker 1997; Hobbs 2007; Golet et al. 2011). Thus, choosing the proper temporal and spatial scales to measure the chosen fauna criteria is an essential part of defining relevant fauna criteria.

Choosing the combination of suitable fauna criteria, monitored at the proper scales, will inevitably be site-specific. However, governments would be stepping forward if they were to (1) promote the need for direct fauna monitoring in their guidelines for post- mine rehabilitation and completion criteria, instead of the current perspective that habitat criteria should suffice; and (2) include a list of fauna-specific questions to be investigated by mining companies when developing their environmental management plan (and other documents related to rehabilitation and mine closure). This list of questions could contain the following points of investigation:

- What fauna species are present in the vegetation communities comprised in the proposed mining path (as defined by pre-mining surveys and or historical

- 172 -

records)? Ö These constitute the list of candidate fauna species for monitoring.

- What are their statuses (endangered/flagship/keystone species etc.)? And/or are there species of particular ecological, cultural, political, economical, etc. interest? Ö These species are the primary candidates for direct/indirect monitoring. Ö Examples: flagship species: koalas, species of Aboriginal cultural importance: goannas Varanus sp., vulnerable species (Queensland Nature Conservation Act 1992): glossy black cockatoo Calyptorhynchus lathami.

- What degree of knowledge is each of these species warranted (presence, density, ecological data, genetics, etc.)? Ö This conditions the data to be collected and the survey methods. If presence only needs to be established, indirect signs (photo-id, scats or call recording) might suffice. Acquiring more precise ecological data will require individual captures and sampling. Ö Example: three levels of knowledge required, three methodologies: (1) scat surveys of koala to ensure their presence, (2) visual surveys to determine the reproductive status of the population, (3) capture, fitting radio-collars and tracking to collect behavioural data.

- Are there specific threatening processes expected to increase (pollution, invasive species, etc.), and are there known sensitive species to these processes? Ö Sensitive species or the direct threat can become fauna criteria. Ö Example of criteria: density of feral predators such as foxes or cats has to be inferior to undisturbed areas on the basis of the Activity Index (Allen et al. 1996).

- What are the key ecosystem processes and functions to be re-instated? What are the known communities associated with them? Ö These communities can be monitored to ensure key processes and functions develop appropriately in rehabilitated areas.

- 173 -

Ö Examples: earthworms and soil structure (Boyer & Wratten 2010).

- What time and space scales are relevant to measure each criterion and compare it to reference sites while accounting for natural variations? Ö Once the species/communities to be included as fauna criteria and the survey methods have been decided, specific scales for their monitoring will need to be determined. Ö Example: monitoring of species presence by auditory survey (e.g. calls) needs to be done at the appropriate season; this season needs to be consistent across surveys (undisturbed/rehabilitated area surveys, surveys at time T, time T+X years, etc.).

Developing this list to achieve a generic and comprehensive list that can result in adopting relevant fauna criteria for any mining project is a practical research area critical for ecological restoration. This would answer the call for researchers to become more proactive, in every possible way (including politics, legislation, education), in shaping the society they believe in; hopefully, a society where humans and their economies are functioning within the boundaries set by the sustainable use of Earth resources (Fischer et al. 2007b; McDonald 2010).

A synergetic opportunity

The application of principles of wildlife ecology to fauna restoration has lagged behind the use of advances in vegetation ecology for flora restoration (Morrison 2001), but the call for more fauna to be integrated in restoration ecology in general and in mine rehabilitation in particular is getting stronger. Although fauna is notoriously difficult to monitor (Bisevac & Majer 1999b), science in general could benefit from such an effort.

Indeed, restoration has been viewed as the ultimate experimental tool of Ecology, or even an experimental ecologist’s dream (Young et al. 2001). Ecological restoration may be useful for testing important hypotheses in many unique ways. For example, large- scale experimentation, including manipulations, may be more acceptable in a habitat to be restored than in an undisturbed one (Palmer et al. 1997). Restoration studies can increase the rate of data collection, compared to undisturbed habitat where successional - 174 - changes might be comparatively slow (Bell et al. 1997). The nature of restoration projects provides a unique opportunity for research on many ecosystemic perspectives (described in Ehrenfeld & Toth 1997; White & Walker 1997). Comparisons between endogenous and exogenous disturbances further increase the interest of restoration research (Fox 1990).

All these possibilities for research should attract many more scientists to develop long- term partnerships with mining companies. There is a unique synergetic opportunity that is largely under-exploited. As highlighted before, social and political considerations will ultimately determine the success of rehabilitation (Higgs 1997; Bellairs 1999; White 1999b; Hobbs 2007). Consequently, integrating university researchers and students in rehabilitation projects may not only be a unique experimental opportunity for ecology, as well as allowing the development of a cost-effective yet highly scientifically-based methodology for the industry; it may well be the only way the broader community will trust and embrace the whole process of mining relinquishment.

Blinky Bill, by Dorothy Wall

- 175 -

- 176 -

Appendix A

NORTH STRADBROKE ISLAND: AN ISLAND ARK FOR QUEENSLAND’S KOALA POPULATION?

Romane Cristescu1,2, William Ellis2,3, Deidré de Villiers4, Kristen Lee2, Olivia Woosnam-Merchez5, Celine Frere6, Peter Banks7, David Dique8, Simon Hodgkison8, Helen Carrick9, Daniel Carter10, Paul Smith11 and Frank Carrick2,9

Institutions 1 School of Biological, Earth and Environmental Sciences, University of New South Wales, Kensington, NSW 2052, Australia 2 Koala Study Program, Centre for Mined Land Rehabilitation, Sustainable Minerals Institute, The University of Queensland, St Lucia, QLD 4072, Australia 3 Koala Ecology Group, School of Biological Sciences, The University of Queensland, St Lucia, QLD 4072, Australia 4 Koala Policy and Operations Branch, Department of Environment and Resource Management, PO Box 5116 Daisy Hill, QLD 4127, Australia 5 Biodiversity Assessment And Management Pty Ltd, PO Box 1376, Cleveland, QLD 4163, Australia 6 School of Land, Crop and Food Sciences, The University of Queensland, St Lucia, QLD 4072, Australia 7 School of Biological Sciences, The University of Sydney, Camperdown, NSW 2006, Australia 8 GHD Pty Ltd GPO Box 668, Brisbane, QLD 4001, Australia 9 EcoIndig Resources Pty Ltd, PO box 498, Kenmore, QLD 4069, Australia 10 Redland City Council, PO Box 21, Cleveland 4163, Australia 11 Sibelco, PO Box 47, Dunwich 4183, Australia

- 177 -

ABSTRACT

Southeast Queensland (SEQ) is experiencing the fastest human population growth in Australia, with the attendant challenges of expanding urbanisation. The documented dramatic decline of koalas Phascolarctos cinereus on mainland SEQ raises the possibility of North Stradbroke Island (NSI) becoming an “island ark” for koalas. Besides the intrinsic ecological features of this population, interactions between koalas and sand mining activities and subsequent habitat rehabilitation have considerable direct and general relevance to koala conservation.

The principal mining company on the island is thus contributing to a koala research program. Since 2002, 32 koalas have been fitted with VHF collars and radio-tracked to determine habitat use and home ranges. Population characteristics have been gathered from radio-tracked koalas and from databases of the Department of Environment and Resource Management. Direct and indirect visual surveys were conducted, including the first whole-island koala survey, which was a joint effort between the Redland City Council, Sibelco - Mineral Sand, The University of Queensland and The University of New South Wales.

Findings of radio-tracking and visual surveys concur that koalas predominantly occupy the northern two-thirds of the western side of the island, that koalas occur in rehabilitated mining areas, and that the urban koala population forms a vital component of the island’s koala population. The ecological characteristics of NSI koalas, including body size, breeding season, reproductive output, home ranges and movements, are consistent with the typical values for Queensland koalas. Special features of NSI koalas are: (1) a diet that might be more reliant on Eucalyptus robusta than that of mainland SEQ koalas; (2) health and mortality status that might be characteristic of a population experiencing less habitat alienation than other SEQ populations; and (3) a lower genetic diversity than other SEQ populations.

We argue that owing to several characteristics including their low genetic diversity, NSI koalas are unique and the location and population should not be considered an island ark for the rest of SEQ, but be conserved and managed as a separate entity.

KEY WORDS: koala, North Stradbroke Island, koalas, mine site rehabilitation, motor vehicle injuries, dog attacks, genetic diversity - 178 -

INTRODUCTION

Australia’s highest human population growth rate and consequential urban expansion is currently occurring in Southeast Queensland (SEQ, Australian Bureau of Statistics, 2010). Habitat loss and fragmentation has been identified as the primary causal factor in the decline of populations of koala, Phascolarctos cinereus, in Queensland and New South Wales (ANZECC, 1998; Melzer et al., 2000; Gordon et al., 2006). This has resulted in the classification of the koala as “Vulnerable” within the SEQ bioregion in 2004 and in the drafting of the Nature Conservation (Koala): Conservation Plan 2006 and Management Program 2006 – 2016 (Queensland Koala Plan). Despite these measures, koala populations in their SEQ stronghold continue to decline dramatically (EPA, 2007; Preece, 2007; DERM, 2009; Rhodes et al., 2011). The situation is so critical that a nomination for listing koala populations in coastal SEQ is currently under consideration (Carrick, F., personal communication). The SEQ koala population is still being described as having both the highest density and the greatest threats for any population in Queensland (Queensland Government, 2006). However, there is one region in SEQ that displays a very different koala situation: Minjerribah (North Stradbroke Island, hereafter NSI).

This koala population is unusual in that the mix of threatening processes on the island is quite different to those impacting other populations in SEQ. Urbanisation is limited on NSI, with less than 2% urban footprint and few bitumen roads. As a result, both vehicle- and dog-induced mortality could be more restricted than in most other areas where koalas occur in SEQ. Habitat clearing on NSI has resulted primarily from ongoing mining activities on the island, though current clearing for mining activities is not occurring in the areas of the island most used by koalas. However, as mining areas on NSI are progressively rehabilitated, previous clearing which did impact on koala habitat can be reversible, providing rehabilitated areas reach a standard that allows koalas to reoccupy them and sustain a viable population. This is in stark contrast to the permanent destruction of koala habitat produced by urbanisation or most agricultural or industrial activities.

The NSI population of koalas is also unusual because it is thought to be an endemic island population, one of only two reported naturally occurring island populations of koalas in Australia (the other being Rabbit Island, Ellis, W., unpublished manuscript) - 179 - and the only extant koala population inhabiting a sand island. No one knows for sure how long koalas have occupied NSI, but they have undergone some period of isolation and they show genetic differentiation from mainland populations (Lee, 2009; Lee et al., 2010). Their likely survival for a long period of time on a relatively small island is interesting for genetic theories of isolation, inbreeding and population viability. In contrast to southern Australian island populations of koalas, neither NSI koalas nor any of the other populations (natural or artificial) on Queensland islands have ever been described as overabundant; this is a characteristic frequently associated with the southern island populations (Masters et al., 2004). The reasons behind the demographic differences of island koala populations remain unclear and more needs to be learned about island koala ecology.

The unique NSI koala population is, therefore, crucial for koala conservation both in SEQ, as a relatively protected population and nationally, as a key to understanding koala population growth and equilibrium in situations of geographical isolation. The NSI koala population could act as an island ark: by definition, a healthy population, isolated from threats, representing the global population and safe in the long term. Consequently, the NSI koala population has attracted considerable interest. Research began in 2002, through a collaborative project initiated by Sibelco (formerly CRL), the mining company currently operating on NSI, and The University of Queensland Koala Study Program. More recently, the Redland City Council (RCC), the Department of Environment and Resource Management (DERM) and The University of New South Wales have joined this partnership. The research has investigated the feasibility of mapping koala distribution and density on NSI and aspects of koala ecology and genetics. This paper analyses data gathered during the past ten years of koala research on NSI together with historical data, to help understand the ecology of this population and to investigate the potential role of NSI koalas as an island ark for the conservation of koalas in SEQ.

MATERIALS AND METHODS

HISTORICAL KOALA RECORDS ON NSI

Historical references to koalas on NSI were sought through an exhaustive search of both - 180 - primary and secondary sources, particularly those in the collections of the John Oxley Library (State Library of Queensland), the Fryer Library (The University of Queensland); the Queensland State Archives and the Archives of the Queensland Museum. Extensive searches were also made online of local (SEQ) historical newspapers via the National Library of Australia's Digitised Newspapers site. Particular emphasis was paid to examination of reports by early explorers, settlers and naturalists and also contemporary reports of the proceedings of meetings of the Royal Society of Queensland, the Natural History Society, the Queensland Acclimatisation Society and the monthly meetings of the Trustees of the Queensland Museum. As a follow-up to some of the information retrieved in these initial searches, contact was made with the Australian Museum (Sydney), Museum Victoria (Melbourne) and the Natural History Museum (London.) For more recent articles referring to koalas on NSI, we examined the Stradbroke Island, Moreton Bay, Clipping Book 1901-1972 compiled by Mrs G. Rahnsleben.

In order to ascertain the knowledge of koala presence on NSI from the traditional owner perspective, an interview with Minjerribah Moorgumpin Elder Auntie Margaret Islen was conducted on 22 February 2011.

NSI KOALA DISTRIBUTION

Data from a variety of sources were compiled to study koala distribution on the island: (1) surveys of koala faecal pellets (scats) were conducted either in plots (50x10m, N=66 from Cristescu, unpublished manuscript, or 50x50m, N=8 from Woodward et al., 2008); or at the base of as many trees as could be searched in 30min (N=54 from Woosnam-Merchez, 2008, and N=13 from Woosnam-Merchez O., Hodgkison S. and Cristescu R., unpublished data). (2) A koala survey of bushland and urban areas conducted on NSI in September 2008. The bushland survey consisted of strip transects (as described in Buckland et al., 1993). Five trained observers walked 10m apart, following a fixed compass bearing, for the length of the transects (N=117, surveyed area=149ha) looking for koalas in trees (and opportunistically for scats on the ground). The surveyed sites were chosen to allow all major Regional Ecosystems (RE, Queensland Herbarium, 2009) to be sampled. The predominant REs of NSI are

- 181 - eucalyptus woodlands, mallee low woodlands in the centre of the island and swamps on the fringe (see details in Fig. 1). The urban koala survey was conducted in one day by total count. The survey included searches of street trees, parks and as many private yards as possible, when access was granted by property owners. Trained observers accompanied by volunteers conducted the urban surveys. Urban surveys were repeated in 2009 and 2010 (RCC, unpublished data). (3) Another source of koala presence was recorded by the first author who opportunistically collected koala sights and signs of their presence (scats and skulls). (4) Finally, sightings were extracted from DERM WildNet and Koala Hospitals databases (1973 to 2007). All GPS locations recorded by the authors (Garmin, eTrex®H, USA) or extracted from DERM databases were projected on an RE map (Queensland Herbarium, 2009) using ArcGIS® 9.3.1.

GENERAL ECOLOGICAL CHARACTERISTICS OF NSI KOALAS

KOALA SIZE - To study koala ecology on NSI, 32 koalas were captured between 2002 and 2010 according to standard procedures (Ellis et al., 1995). Koala chronological age (determined from tooth-wear classes, Martin, 1981; Gordon, 1991), gender, body mass, head length and head width were recorded from captured individuals. The same data were gathered from koalas found deceased on the island and necropsied by the first author, and were used in calculating koala mean size (only for koalas that died from traumatic injury and which were in good body condition). Variables assessing koala size respected the assumptions of normal distribution and homogeneity of variances (Levene’s test), and were compared by Analysis of Variance (ANOVA) with PAWS Statistics 18.0 (IBM, 2009). Significance was taken to be p<0.05, standard error of mean (SEM) and standard deviation (SD) are given when appropriate (Altman & Bland, 2005).

KOALA REPRODUCTION - Breeding information was collected from captured or injured/dead females with young. Pouch young were observed without removal while back young were weighed and measured to allow their age to be precisely calculated (Tobey et al., 2006). The timing of birth and therefore mating (breeding season) was then estimated.

KOALA HEALTH AND CAUSES OF MORTALITY - Koala body condition (Ellis - 182 -

& Carrick, 1992) was assessed and blood samples were collected (5 ml, cephalic vein, N=15). All male koalas handled by the authors were checked for abnormal testis. Koalas were tested for Chlamydia sp. (N=20, Ellis W., personal communication) and koala retrovirus (N=8, Lee, 2009). Injured or sick koalas found on NSI and evacuated to the mainland (Australian Wildlife Hospital, Australia Zoo or Moggill Koala Hospital, DERM) for treatment are recorded in a DERM database. Records of NSI koalas were available from 1997 to 2010. Koalas found dead on the island underwent a necropsy by a veterinarian to determine the cause of death (N=14, Cristescu R., unpublished data). Number of koala deaths by age, gender and cause of death were summed from these two sources (DERM and necropsies).

NSI KOALA HOME RANGES AND MOVEMENTS

Very high frequency (VHF) radio-collars (150-152MHz, Titley Electronics, Australia) were fitted onto the 32 koalas which were then released into the tree from which they had been caught, according to standard procedures (Ellis et al., 1995). All koalas included in this study were detected during standard searches predominantly based on the western side of NSI. Koalas were radio-tracked approximately weekly, their GPS locations and the tree species they used were recorded. Home ranges were calculated with the Home Range Tools 1.1 for ArcGIS® (Rodgers et al., 2007), using the kernel density estimation method, with the standard Gaussian curve (Worton, 1989). On the basis of the Schoener index (Schoener, 1981), the variances of the coordinates were unequal, thus the data were standardised. We used a fixed kernel (Seaman & Powell, 1996), with a smoothing factor calculated by least squares cross validation (Worton, 1995). Home range areas were calculated from isopleths of 50%, 90% and 95% of the volume contours. The areas of the home ranges, as well as the areas of each RE class composing these home ranges, were extracted using ArcGIS 9.3.1. Four koalas were located on seven consecutive days to study daily movements.

NSI KOALA DIET

Koala diet was determined from fresh scats collected under radio-tracked koalas using

- 183 - standard techniques (as described in Tun, 1993; Hasegawa, 1995; Ellis et al., 1999). General results have been provided elsewhere (Cristescu R., unpublished manuscript, Woodward et al., 2008), however here we analysed the diet of four koalas tracked for seven consecutive days. Comparisons of scat analysis and day roosting trees were performed based on similarity matrices. These were constructed using the Bray-Curtis measure (Bray & Curtis, 1957) on the plant species constituting the diet per koala per day. Differences between the four koalas and across the seven days were tested with two-way crossed Analyses of similarities (ANOSIM, Clarke & Gorley, 2006), a multivariate equivalent of ANOVA based on similarity matrices (Clarke, 1993). These analyses were performed using Primer 6 (PRIMER-E Ltd., 2001).

NSI KOALA GENETICS

A total of 36 ear tissue samples were collected using standard techniques (Ellis et al., 2002a) from captured koalas to enable genetic analyses. Genotypes based on six microsatellite loci (Houlden et al., 1996) were available for these 36 NSI koalas and for 769 koalas from adjacent mainland areas from Lee (2009; 2010).

The mean number of alleles, the number of private alleles, the expected and observed heterozygosity levels (Nei, 1978) and the unbiased estimator for FIS (Weir & Cockerham, 1984) were calculated using GENETIX version 4.05.2 (Belkhir et al., 1996-2004) and GENALEX 6 (Peakall & Smouse, 2006). Mean allelic richness (mean number of alleles corrected by sample size to allow comparisons) of NSI and two southern island koala populations (French and Kangaroo Island from Cristescu et al., 2009) were compared through the statistical technique of rarefaction (Kalinowski, 2004) by using the HP-rare program (Kalinowski, 2005).

We investigated genetic bottleneck signatures using two tests available in BOTTLENECK 1.2.02 (Piry et al., 1999). The sign test relies upon the propensity for bottlenecked populations to present greater heterozygosity than expected on the basis of the number of alleles (Cornuet & Luikart, 1996). We used the sign test hypothesising that mutations follow the infinite allele model, which could be the most relevant for marsupials (Cristescu et al., 2010). The shift test, or allele frequency distribution test, relies on the characteristic of bottleneck populations to lose rare alleles (Luikart et al., - 184 -

1998).

The level of inbreeding was assessed by calculating the individual Internal Relatedness (IR). IR is a coefficient that estimates the relatedness between the two unidentified parents of an individual by comparing the two haplotypes inside an individual genotype (Amos et al., 2001). IR values range from -1 for outbred individuals, to 1 for inbred individuals. The IR was calculated with an Excel macro from the formula given in Amos et al. (2001).

We calculated Nei’s unbiased genetic distances D between koalas of Local Government Areas in SEQ using Genetix 4.02 (Belkhir et al., 1996-2004). As the government areas do not represent biological populations, we compared the genetic structure of koalas from NSI and from adjacent mainland regions of the Gold Coast (n=48) and Redlands (n=62). These two mainland regions were selected because the Redlands is geographically the closest to NSI koalas, being directly opposite to and at some locations less than 5 km away from NSI; and the Gold Coast, because it was found to be most genetically similar to NSI (Lee et al., 2010). The genetic structure of koalas from NSI, Gold Coast and Redlands were determined based on their genotypes using the Bayesian clustering program STRUCTURE 2.2 (Pritchard et al., 2000). To infer the number of genetic clusters (K), twenty independent runs of models K = 1 to K = 7 were used, deduced by posterior probabilities [Ln P(D)] using 100,000 iterations after a 100,000 iteration burn-in period. The average Ln P(D) for each K was plotted to determine the highest likelihood and the number of population clusters was determined by calculating ΔK (Evanno et al., 2005). Koalas were assigned to a particular cluster if they had a probability of membership to that cluster (q-value) ≥ 0.8. Koalas with a q- value between 0.19 and 0.79 were regarded as mixed or hybrid animals (Lee, 2009).

RESULTS

HISTORICAL KOALA RECORDS ON NSI

As early as August 1831, William Holmes, collector from the newly established Colonial Museum in Sydney, visited Amity Point to acquire specimens. Unfortunately, he accidentally fatally shot himself during his visit and the Australian Museum does not

- 185 - possess any of his material (records or specimens), and no other records or reports of koalas on the island at this early stage of colonisation exist (e-mail from Vanessa Finney, Manager Archives and Records, Australian Museum 18 August 2009). No record of koalas (or most terrestrial vertebrates present on the island) was found from early records of other explorers, settlers and naturalists. Archibald Meston even asserted, in 1895, “On Stradbroke, wallabies of various kinds and kangaroos are numerous, but there is not a possum, squirrel or bear [i.e. koala] on the whole island… The absence of all tree-climbing animals on Stradbroke Island is an inscrutable mystery” (Meston, 1895). However, this assertion was not based on firsthand knowledge or actual field work and is of dubious scientific value, as koalas and various gliders do occur on NSI (Martin, 1975). No evidence was found of any translocation of koalas or other Australian native mammals onto NSI after European occupation, though records of attempts to introduce non-native species such as deer and rabbits were located in the archives of the Queensland Acclimatisation Society (Queensland Acclimatisation Society Report, 13 October 1869).

From more recent records, the earliest observation of a koala was reported by Ellie Durbridge in August 1943, at Point Lookout (Martin, 1975). An article from November 29, 1967, entitled “Salad Bowl Star” reported that on NSI “wildlife is abundant… the occasional koala [is observed]”.

The old grannies of Myora mission used to show Auntie Margaret Iselin as a child (in the 1940s) the “Boom-Berpees” (koalas). Auntie Margaret Islen was taught not to eat its bitter meat and to be careful not to approach Boom-Berpees as they could be quite “vicious” in their reaction. Interestingly, the grannies were always pointing toward the south of the island and saying that Boom-Berpees came from there. From the interview with Auntie Margaret, it cannot be confirmed from Aboriginal oral history that koalas have been present on the island since before the beginning of the 20th century, but interestingly the interview hinted that the historical distribution of koalas on the island could have been in areas not initially frequented by Europeans.

NSI KOALA DISTRIBUTION

In general, koalas and koala signs were detected in most of the sites searched on the - 186 - western side of the island, except in the south third of the island (Fig. 1). Koala scats were found in 49 plots, in 14 locations during the bushland survey and opportunistically on 52 occasions. Koalas were sighted on eight occasions during the bushland survey. The GPS locations of plots and transects searched without scats or a koala being located were also recorded (Fig. 1). Skulls were found on two occasions, while DERM databases provided a further 35 koala locations. Most of the sites on the eastern side of NSI returned no records of koalas or their signs, with the exception of Point Lookout and its immediate surroundings. The middle of the island, which corresponds mainly to mallee low woodland vegetation communities, had very few signs of koala usage.

No density or population size could be extrapolated from the bushland survey as there were not enough koala sightings to derive a reliable population estimate. Compared to the time and effort invested in the bushland survey to find eight koalas, urban areas provided better results for relatively less search effort and seemed favoured by koalas. Urban surveys in 2008, 2009 and 2010 detected 16, 29 and 21 adult koalas, accompanied by respectively 1, 3 and 7 young from the year (see details in Table 1) indicative of an average density of adult koalas in urban areas of NSI of some 0.13 koala/ha.

GENERAL ECOLOGICAL CHARACTERISTICS OF NSI KOALAS

KOALA SIZE - The mean adult size of koalas on NSI was calculated from 33 healthy individuals and ranged from 5.0 to 8.9 kg. Male koalas were larger than females (ANOVA, weight: F=59.6, df=32, p<0.001; head length: F=123.8, df=29, p<0.001; head width: F=58.8, df=25, p<0.001; Table 2).

KOALA REPRODUCTION - The breeding season on NSI was calculated to occur between September and March, on the basis of the age of dependent young (N=16, Fig. 2) and the time of gestation in koalas (35 days; Martin & Handasyde, 1999). The percentage of females carrying young, based on captured and necropsied female koalas, was 71% (N=24).

KOALA HEALTH AND CAUSES OF MORTALITY - Koalas captured during this study were generally in good condition; however, of 24 males examined, two presented

- 187 - an abnormal testicular morphology. Analyses of blood samples indicated that haematological and biochemical values were within normal ranges, while some had minor changes likely to be of no biological significance (IDEXX Laboratories). Koala retrovirus was found in all tested koalas and sub-clinical chlamydial infections were detected in four koalas.

Between 1997 and 2010, 104 koalas were found dead, sick or injured on NSI with 82 confirmed deaths (Fig. 3, Table 3a). The gender was known for 80 dead, sick or injured koalas (50 males and 30 females Table 3b). Three causes of koala morbidity and mortality could be determined: car strikes (N=35), disease (N=21) and dog attacks (N=16, Table 3c). One radio-tracked koala, an adult male of 7.5kg, was one of the dog- related fatalities.

NSI KOALA HOME RANGES AND MOVEMENTS

Koala locations were recorded on 795 occasions between September 2002 and November 2010. The number of fixes per animal was variable (2 to 79). This was mainly due to collar design, as a weak link was included to ensure collars would self- detach to preclude the possibility of koalas being collared for life; in some cases this safety feature gave way after an unexpectedly short period. The second reason for variable numbers of fixes was that some koalas moved to unreachable areas in swamps and could not be accurately located. The roosting tree species was identified on 791 occasions (Table 4): the two main roosting trees were Eucalyptus robusta and Callitris sp.

Home ranges of 28 koalas were projected on maps (Fig. 4), while four had too few fixes to make this meaningful. In order to maximise the accuracy of home range estimates, home range sizes were calculated only for animals having more than 20 fixes (N=16), and excluding a sub-adult koala for whom asymptote in range size was not achieved (i.e. Jundall had a range size of 132 ha at the time of estimation) and the resulting mean home range size was 36 ha (SD=22, Table 5).

The 16 koalas described above were found regularly in six of the 24 remnant REs present on NSI. The main remnant vegetation communities represented in koala home

- 188 - ranges were RE 12.2.7, 12.2.15 and 12.2.6 (Table 6). In particular, they used Melaleuca open forest (12.2.7) and swamps (12.2.15) respectively 5.2 and 2.6 times more than the availability of these REs would predict. However, the “RE” most frequented by the 16 koalas was mine rehabilitation. The home range results confirm that koalas favour urban areas, as radio-tracked koalas utilised urban areas 5.3 times more than expected based on availability (Table 6).

Daily movements ranged from 0 to 441m for the four koalas followed over seven consecutive days (Noonie: 164m SD=92; Suzy: 27m SD=42; Emma: 37m SD=17; Bundy: 272m SD=146), with a mean of 125m (SD=116). Large-scale movements were sometimes observed, for instance one koala captured near Bayside travelled 6.6km in 2.5 months before settling at Myora (Fig. 5). Jundall travelled 6.8km in 5 months from Amity swamp into Amity township and then to the rehabilitated area of the previous Amity mine site (as seen in Fig. 4).

NSI KOALA DIET

The diet of individual koalas was similar and did not vary over time (ANOSIM, koalas: R=0.124, p=0.12, days: R=0.079, p=0.24) Koalas favoured E. robusta (58% of the total leaf fragments present), then E. pilularis and E. tereticornis (13%). E. robusta was the only species present in 100% of the scats sampled. Interestingly, half of the time (14 out of 28) the daytime roosting tree in which a koala was found was not present in its diet. The daytime roosting tree was the major species in the diet less than 15% of the time (Fig. S1).

NSI KOALA GENETICS

The genetic variability among NSI koalas based on microsatellite DNA results is summarised in Table 7. The Fis value was indicative of random mating. The mean allelic richness of NSI koalas was 4.0, compared to 3.3 on French Island in Victoria and 2.4 on Kangaroo Island in South Australia.

No evidence of genetic bottlenecks could be detected (Sign test: p=0.14, Shift test:

- 189 - normal L-shaped distribution, see Fig. 6). The mean IR was close to zero (0.09), suggesting little to no inbreeding. However, IR coefficients were variable with six koalas having their IR greater than 0.5.

The mean of Nei’s unbiased genetic distances from NSI to other SEQ regions (D=0.863) was higher than between any other SEQ population studied (Table 8). The NSI koalas showed the least distance to the koalas of the Gold Coast (D=0.464). STRUCTURE clustering analysis of koalas from NSI and the adjacent Redlands and Gold Coast koalas identified K=2 clusters (Fig. 7). The NSI koalas clustered together with many koalas from the Gold Coast, while the majority of the Redlands koalas formed the second cluster. Although some koalas from the Gold Coast and Redlands showed a mixed assignment to the clusters, no koala from NSI did so.

DISCUSSION

HISTORICAL KOALA RECORDS ON NSI

Historical records of koalas on NSI are scarce, but this scarcity must be considered in the context of Europeans having difficulties detecting koala presence across Australia (Martin & Handasyde, 1999). However, there is an Aboriginal oral history involving koalas on the island, though this is not informative as the origins of the population. It is not clear from recollections from Auntie Margaret Iselin whether koalas were known to be resident on the island in times before European contact. The “grannies” told the young Auntie Margaret that koalas came from the south of NSI. This could be understood as either that the bulk of the population occurred south of the Myora mission, or that koalas came from South Stradbroke Island or the Southern bay Islands.

NSI KOALA DISTRIBUTION

Koala distribution on NSI seems to be mostly concentrated on the northern two-thirds of the western side of the island. Vegetation communities favoured by koalas are present both on the eastern and western sides of the island, hence on the basis of vegetation community alone, koalas could be equally found on the eastern side of the island;

- 190 - however, the evidence indicates that this is not the case. Soil (Thompson & Ward, 1975) and ground water (DERM, unpublished data) distribution patterns do not seem to match the koala distribution either. Koala presence could be related to dune age and koalas could be mainly occupying old dunes, while avoiding ancient dunes (as defined in Benussi, 1975). The central parts of the island tend to correspond to ancient dunes (formed half a million years ago). The old dunes (formed 180,000 years ago) are mostly found on the western side of NSI, while recent dunes (formed 40 to 20,000 years ago) are found east and north (Benussi, 1975). It could be that the combination of mineral leaching and accumulation of organic material (Thompson & Ward, 1975; Westman, 1975) makes the western coast more able to support koalas. But the congruence between age of dunes and soil types is not straightforward, thus this hypothesis is very difficult to test. One influence could be salt deposition on the eastern side of the island from the main wind (east/north-east, Australian Government Bureau of Meteorology, 2004). The current distribution of koalas might also be the legacy of past fire regimes. Fires on NSI have played a major role in shaping plant communities: for instance, between 1995 and 2005, 56% of the island was burnt (RCC, unpublished data). The effects of large wildfires on the koala population are unknown. However, given the intensity and extent of these fires, some parts of the island might have seen local extinctions and the koala population might still today be recovering and recolonising parts of the island. The current distribution of NSI koalas remains largely unexplained.

In relation to this distribution, an alarming concern for the future of NSI koalas is raised by the implications of NSI Future Vision Map (Queensland Government, Fig. S2). Although the entire area (including some important koala habitat) south of Dunwich would be preserved under this plan, some vital areas seem to be being opened up for development. Particularly, areas ear-marked as ‘mixed use’ (including residential and commercial) and tourism along the western coast from Dunwich to Amity (which already represent a serious “pinch point” for koalas), and north towards Point Lookout, are in prime koala habitat which constitutes much of the most critical habitat for the species on NSI. Maybe even more worrying, the area with the highest koala records (between Amity and Flinders), is ear-marked for township expansion. This time there is no excuse that “we didn’t know” we were destroying the key habitats on which koalas depend.

- 191 -

Another concern for the future of NSI koalas arises from the observation that vegetation communities favoured by koalas on NSI are mostly wet areas (e.g. swamps, Melaleuca open forests) or forests situated in the lowest part of the island (Fig. 1). This places koalas on NSI in a vulnerable position in the scenario of climate change resulting in rapid sea rise. Furthermore, a considerable amount of water is extracted from NSI to Brisbane without prior knowledge about the effects on swamp communities, particularly regarding sea salt intrusion.

One interesting characteristic of the NSI koala distribution is the relative strength of the urban koala population, which suggests that European settlers might have built the township in high value koala habitat. Koalas and people often compete for the same fertile parts of the landscape (Lunney & Matthews, 1997). Moreover, the use of fertilisers and artificial irrigation in urban areas might enhance the quality of eucalypt foliage, although this concept remains to be tested and can in no way justify the proposal to increase the development footprint at the expense of valuable koala habitat, rather than in areas of lower environmental value and / or greater extent and / or which have previously been disturbed.

Koalas also occur in post-mining rehabilitated areas. A large part of the western side of the island, which is likely to have been prime koala habitat in the past, has been mined. However, koala populations persisted in adjacent areas and recolonised the rehabilitated landscape. Their use of rehabilitated areas does not appear to be random, as radio- tracked koalas used rehabilitated areas twice as much as would be predicted by availability. This is an encouraging result for habitat restoration in general, and an important step towards demonstrating that rehabilitated areas can develop into functioning ecosystems. This relative adaptability of koalas to rehabilitated areas has been found in other disturbed habitats such as logging areas (Jurskis & Potter, 1997; Kavanagh et al., 2007) and landscapes fragmented by agricultural and other activities (Gordon et al., 1990; White, 1999; Rhodes et al., 2008), and suggests that koalas are more adaptable than previously thought (Hume, 1990; Cork et al., 2000). However, as is clear from the data for SEQ, there are finite limits to this adaptability and these limits have been exceeded in significant areas of SEQ.

Although no precise population size could be estimated, the number of koalas on NSI could be higher than previously expected (Barry & Campbell, 1977). It is likely that the - 192 -

NSI koala population is larger than that of Magnetic island, the Queensland island with the reputation of possessing the largest northern island koala population (last population estimation of Magnetic Island: 170 koalas, Cahill et al., 1999; Jackson, 2007).

GENERAL ECOLOGICAL CHARACTERISTICS OF NSI KOALAS

KOALA SIZE - The ecological characteristics of NSI koalas seem to fall within the ranges previously reported for the species. Adult body weight for Queensland koalas is 5.1kg on average (4.1-7.3) for females and 6.5kg (4.2-9.1) for males (Martin & Handasyde, 1999). Koalas on NSI could be larger than Queensland averages, but still within the normal range. In particular, koala size on NSI is comparable to that of another koala island population in Queensland, St Bees Island (females=6.2kg, males=7.1kg, Ellis & Bercovitch, 2011). The “island rule” regarding body size suggests that contrary to other insular animals that express dwarfism or gigantism, marsupials exhibit no consistent pattern (Lomolino, 1985). Koalas’ body size does not appear to vary widely with insularity.

KOALA REPRODUCTION - The koala breeding season on NSI fits into the classic period of October to May (Martin & Handasyde, 1999; McLean & Handasyde, 2006), with a peak between December and March (Ellis et al., 2010b; Whisson & Carlyon, 2010). The NSI koalas have a higher reproductive output compared to that reported for other island populations such as Kangaroo Island (South Australia) where, depending on the year, 52 to 67% adult females are observed to be breeding (Whisson & Carlyon, 2010). On Kangaroo Island, the koala population is said to have increased from 18 individuals to 27,000 in 80 years (Masters et al., 2004). Although there is no estimate of NSI population, it is clearly much lower than the Kangaroo Island population. This raises the question of why the population growth on both islands differs so widely. According to our results, reproductive output does not seem to provide the answer. Differences in habitat quality, rainfall predictability, or abundance of predators have been hypothesised (Masters et al., 2004). Rainfall on NSI is abundant (annual rainfall: 1550mm, Bureau of Meteorology, 2010) and probably not a limit to population growth. Feral foxes are present on NSI (Cristescu, R., unpublished manuscript), whilst absent from Kangaroo Island (DAFF, 2009). However, dogs (which are known to represent a

- 193 - threat to koalas, see data above) are present on both islands and, although no objective data exist, our subjective impression is that a koala is probably at least as likely to encounter a dog on Kangaroo Island as on NSI. Poor quality of soil on NSI (Rogers, 1975) could be reflected in poor quality of browse (Braithwaite et al., 1984; Cork & Sanson, 1990; Crowther et al., 2009), perhaps limiting the koala carrying capacity of NSI; although this might be expected to be reflected in reduced longevity and/or reduced reproductive output on NSI, which does not appear to be the case. Also, high- intensity fires can have long-term effects on distribution and density of koala populations (Melzer et al., 2000) ; so NSI fire history might have played an important role. More research is needed to explain such differences in population growth, particularly since it seems to characterise other northern/southern island koala populations (Ellis, W., unpublished manuscript).

KOALA HEALTH AND CAUSES OF MORTALITY - Koalas on NSI typically have a reasonable body condition and on average are probably in better shape than their neighbours from the mainland (de Villiers, D., personal communication). Koalas on NSI carry typical koala infections, i.e. bacteria of the family Chlamydiacae and koala retrovirus (Girjes et al., 1988; Tarlinton et al., 2006). Although most koalas carry latent chlamydial infections, active chlamydiosis is widely known to decrease koala reproductive output (Girjes et al., 1988); however, the percentage of NSI breeding females was among normal figures for Chlamydia-free populations (65.6 to 81.1% in McLean, 2003) and well above the reproductive output of populations with a high prevalence of genital chlamydiosis (32.3 to 38.8% in McLean, 2003, see Mitchell et al., 1988). It is interesting to note that other clinical signs of chlamydiosis, such as kerato- conjunctivitis and cystitis, were not observed in NSI koalas by any of the authors. Although clinical signs of chlamydiosis have been reported in NSI koalas (DERM database), this might be a rare occurrence on the island. Effects of chlamydial infections are thought to be exacerbated by stress (Weigler et al., 1988; Ellis et al., 1993) and the same is likely to be the case for koala retroviral infections. The fact that NSI koalas do not seem to have a significant rate of expression of clinical signs of the most common diseases in koalas could mean they are relatively protected from the pressures usually associated with habitat alienation factors.

On the basis of our small sample size, observed abnormal testicular morphology in NSI

- 194 - males is present and could be relatively high (around 8%). This places the NSI population between the incidences of abnormal testis observed on French Island (4.3% in Seymour et al., 2001) and on Kangaroo Island (10.6% in Cristescu et al., 2009). However, a larger sample size needs to be studied to confirm these results.

The main cause of admission to hospital for koalas found by people on NSI was vehicle-related trauma (32%, Table S1). There is an obvious bias as car victims are more easily found by people than naturally sick or injured koalas in the bushland. Nonetheless, we can assume a comparable bias for all koala records in the DERM databases. This allows comparisons of the relative proportions of different causes of admission to hospitals between NSI koalas and koalas from other parts of SEQ (Table S1). Outside NSI, car-strike is never the first cause of koala admission to hospital (Table S1). Dogs are a continuing threat to NSI koalas (14% of hospitalised koalas) and appear to pose a greater threat on NSI than anywhere in mainland SEQ. However, the proportion of koalas admitted for disease in the NSI population is the lowest in SEQ. Reasons for the low proportion of koalas showing clinical signs of chlamydial/retroviral disease has been discussed above. We have no explanation as to why the “Other” category for NSI is much higher than anywhere else in SEQ (almost 1/3 of “admissions”, which would influence the other proportions of causes of admission).

On the basis of high proportions of vehicle- and dog-related causes of admission of NSI koalas to wildlife hospitals, the protection of koalas from anthropogenic factors could dramatically reduce koala mortality on NSI. With regard to vehicle-related trauma, reducing the speed limit on the island’s main road between Dunwich and Point Lookout (the maximum speed limit is currently 100kph, which is arguably unwarranted for a total distance of about 19km) and outside Amity (80kph in prime koala habitat) is one constant plea from wildlife carers of NSI. Raising driver awareness, particularly on ferries, might have a positive outcome for all NSI wildlife. For indeed, research has found a direct correlation between the number of cars brought by ferries and wallaby road-kills on the island (Osawa, 1989). The problem of dog attacks comes from both feral dogs in the bushland and unrestrained domestic dogs. Many roaming dogs are of domestic origin and feral dogs are now widespread on the island and increasing in numbers (Cristescu, R., unpublished manuscript). Feral-predator control seems to be critically needed on NSI. Eradication is never simple, but is achievable for islands

- 195 -

(Veitch & Bell, 1990; Myers et al., 2000). Some feral-predator control actions were instigated by CRL (now Sibelco) between 2002 and 2004 (Smith, P., personal communication) and by the RCC in 2009. As part of the RCC feral-animal control, 47 foxes and three cats have been removed from the island (Carter, D., personal communication). However, it might be necessary to develop an island-wide plan. This sort of island-wide plan was considered in 2006 by a group of NSI stakeholders (RCC, CRL, DERM, SEQ catchment and the Quandamooka Land Council) but no agreement could be reached and the plan was aborted. For an island-wide plan to be successful, cooperation between all stakeholders and land tenants would be necessary. Meanwhile, education of islanders and visitors to reduce dog attacks on koalas is necessary. Distribution of “Help protect North Stradbroke Island’s unique wildlife” flyers from RCC has commenced; another flyer of interest that could be distributed is the “koalas and dogs” factsheet from DERM. With regard to human population and road densities, which are very low compared to other parts of SEQ, human-induced koala mortality seems disproportionately high on NSI. However, without knowing the size of the NSI koala population it is difficult to assess how serious anthropogenic threats are to the population’s viability, or to compare it with the dramatic human impact on other SEQ koala populations.

NSI KOALA HOME RANGES AND MOVEMENTS

Home ranges of koalas vary greatly from one study to the next. First, koalas in different habitats (e.g. semi-arid tropics versus subtropical SEQ or the colder climes of southern Australia) appear to respond by establishing different-sized home ranges; but estimates of home range size will also be highly dependent on the study design (Seaman et al., 1999) and the method used (Gallerani-Lawson & Rodgers, 1997; Gitzen et al., 2006). While this can make comparisons hazardous, they remain informative nonetheless and we give below some results from other koala populations. Home range size from a modified harmonic mean (Dixon & Chapman, 1980; Spencer & Barrett, 1984) is given from several studies. On French Island (Victoria), home ranges were as small as 1.2ha for females (N=18) and 1.7ha for males (N=20), based on observations of one breeding season (Mitchell, 1990). Using the same method in Queensland, the mean home range size in a three-year study was as large as 101.4ha for females (N=9) and 135.6ha for - 196 - males (N=8, Ellis et al., 2002b) in the semi-arid tropics. Closer to the methodology used in this study (i.e. 95% kernel method), over three years koalas in SEQ were found to use a home range between 5.3-45.1ha for females (N=16) and 6.4-91.4ha for males (N=8) in one study (White, 1999), while another study reported smaller home ranges of 7.3-7.9ha for females (N=27) and 13.1-22.5ha for males (N=16, Thompson, 2006). On the Queensland island of St Bees, annual mean home range sizes were even smaller with 7.9ha for females (N=33) and 8.6ha for males (N=22, Ellis et al., 2009). Home ranges of NSI koalas (mean 36ha) may thus be larger than home ranges of other island populations but in the range of Queensland mainland koala populations.

Koalas on NSI travelled similar distances per day (125 m, N=4) to Saint Bees Island koalas (117 m, N=15, Ellis et al., 2009). The relatively long movements observed could correspond to koalas dispersing, as other studies in SEQ demonstrated that dispersing individuals could travel up to 10.6km away from their birth place (Dique et al., 2003). As for Jundall, another possible explanation for her observed journey of about 6.8km to Amity township could have been to seek mating opportunities during the breeding season, as she was observed in the same tree as a large male (15/11/2008). On her way back to Amity swamps, Jundall was reportedly struck by a car (29/12/2008). Radio- tracking on the following days did not reveal signs of injury, but this is a reminder that a proportion of koalas struck by cars might not be accounted for in hospital statistics. An interesting behavioural observation was made, as the day following the car-strike, Jundall was found in a small Umbrella Tree Schefflera sp., 20cm below another female koala. This could be incidental, or could reflect some social interaction. A phenomenon, called social support, has been observed in other species, where stress is thought to be decreased by the presence of conspecifics (Sachser et al., 1998).

NSI KOALA DIET

As found in other studies (Ellis et al., 2002b), roost tree species used by NSI koalas often were not represented in their diet. Choices of daytime roosting trees can be related to thermoregulatory constraints (Clifton et al., 2007; Ellis et al., 2009; Ellis et al., 2010a), which could explain the high use of Callitris, a non-food tree characterised by very dense foliage compared to the more open canopies of Eucalyptus and Corymbia sp.

- 197 - on NSI.

Eucalyptus tereticornis seems to be consistently described as an important koala fodder species in Queensland (White & Kunst, 1990; Martin & Handasyde, 1999; Ellis et al., 2002b). For instance, based on faecal-pellet analysis, E. tereticornis constitutes over 90% of koala diet on St Bees Island (Tucker, 2008). It is interesting to note that, in our study, E. tereticornis did not seem to feature prominently as a browse species; instead, E. robusta was much more represented in koala diet. This could reflect differences in availability of E. tereticornis on NSI compared with other parts of Queensland. Furthermore, the data were collected over one week and the diet composition was similar across days – the establishment of which was the primary purpose of this sampling regime. A practical consequence is that scats for diet analysis should be collected with a minimal interval of a couple of weeks and over the different seasons. Another study of the diet of NSI koalas, where scats were sampled throughout the year at longer intervals, is useful in providing a more complete picture of NSI koala diet. That study (Cristescu, R., unpublished data) found a different, more diverse diet. Indeed, E. racemosa and E. tereticornis were the main species consumed in undisturbed areas, while in rehabilitated habitats E. pilularis and E. tindaliae were also eaten. Globally, 10 species were consumed, including 5.8% of Lophostemon. One koala was also observed feeding on Allocasuarina littoralis. In contrast to a study at Point Halloran, Queensland (across Moreton Bay in the mainland part of RCC), where Melaleuca quinquenervia comprised up to 7% of koalas’ diets (Hasegawa, 1995), Melaleuca did not seem to be an important part of NSI koala diet (< 1%). These differences in diet between NSI and other parts of mainland Queensland can be due to numerous factors including tree availability.

NSI KOALA GENETICS

Genetic diversity of NSI koalas was the lowest of all SEQ populations tested so far (Lee et al., 2010). This was expected, as island populations usually show less diversity than mainland populations. On the basis of mean allelic richness, the NSI population has substantially fewer alleles than the next more diverse SEQ population (Redlands mainland = 5.2, NSI = 3.5, Lee, 2009; Lee et al., 2010). Lower genetic variability of

- 198 -

NSI koalas than adjacent mainland population could be due to the combination of a low number of founders and/or genetic drift that is expected to be high for a small population size (Frankham, 1998; Keller & Waller, 2002).

On the basis of mean allelic richness, the NSI population shows a greater diversity than any koalas on southern Australian islands, which are notorious for their history of genetic bottlenecks and inbreeding (Seymour et al., 2001; Cristescu et al., 2009; Lee, 2009). For the NSI koala population, signatures of a genetic bottleneck were not found. This could be due either to the absence of any genetic bottleneck or to the long isolation of NSI koalas: given enough time, new mutations will appear and bottleneck signatures will be lost (Cornuet & Luikart, 1996). Thus, it seems very unlikely that the current NSI koala population could have been established by the introduction of a small number of individuals at some time after European occupation of NSI, which was one option suggested by Barry and Campbell (1977). Furthermore there is absolutely no recorded evidence that this occurred despite intensive scrutiny of the records of the Queensland Acclimatisation Society, etc.; finally, any such introduction would have been most likely to occur in the northern parts of NSI where early European occupation was undertaken and this is inconsistent with the Aboriginal oral history indicating that koalas were associated with the southern end of the island.

The other hypothesis suggested by Barry and Campbell (1977) to account for the occurrence of koalas on NSI is that they colonised NSI naturally from the Gold Coast through South Stradbroke Island (the south and north islands were connected to each other until 1896) and/or via the Southern Bay Islands when they were connected to the mainland during a period of lower sea levels. Currently there is relatively little potential koala habitat and no koalas on South Stradbroke Island; and while some of the Southern Bay Islands do have suitable vegetation present, they are relatively small and thus probably unable to sustain koala populations in the long term. The establishment of the NSI koala population when there was last a land bridge to the mainland is consistent with Nei’s unbiased genetic distances (and FST results in Lee et al., 2010) showing that the NSI population is the least differentiated from the Gold Coast koala population of all the mainland SEQ populations, as well as the STRUCTURE clustering of NSI koalas with Gold Coast rather than Redlands koalas. This colonisation route suggests NSI koalas could have been isolated from the mainland for 8,000 years (Sloss et al., 2007).

- 199 -

Consequently, and not surprisingly, the population of NSI koalas could have lost the bottleneck signature but kept the low-diversity characteristics of small isolated populations. The presence of a relict population of Agile Wallabies (Macropus agilis) and the Golden Swamp Wallaby (a distinctive colour variant of Wallabia bicolor) on NSI provides evidence for long isolation of populations of large marsupials and is consistent with the timescale proposed for isolation of the current koala population.

Lee (2009) suggested, however, that the level of heterozygosity (and allelic richness) identified in the NSI koala population could indicate either that the effective population size has been consistently large or that there has been some degree of ongoing geneflow from the mainland. While they were mainly concerned to account for the herpetofauna of NSI, Barry and Campbell (1977) suggested that the geographical characteristics of southern Moreton Bay and the several major rivers that flow into it could account for transport of animals to Stradbroke Island by rafting via trees and other vegetation during floods. Over millennia, low-frequency transport of small numbers of koalas by such a mechanism is not impossible and may be consistent with Lee’s suggestion of some ongoing gene flow, since there seems no other evidence that NSI has ever supported a large koala population and certainly not during the last two centuries.

The large genetic distance between NSI and other SEQ populations strengthens the hypothesis that the NSI koala population has been relatively isolated from other populations for some considerable time, with genetic differentiation driven by genetic drift. This confirmed the spatial clustering analysis by Lee et al. (2010) which showed NSI clustering independently from all other SEQ koalas. Thus, the NSI koala population should be considered at least as a separate management unit on the basis of microsatellite differentiation (Sherwin et al., 2000). Propositions for introduction of mainland koalas to the island in order to increase genetic diversity are misguided and contraindicated as a management response at this time: NSI koalas are not inbred and do not present signs of inbreeding. However, low genetic diversity decreases the evolutionary potential and the chances of adapting to dramatic events (e.g. climate change, new diseases, Frankham, 1995). Thus, genetically, NSI koalas might be a more fragile population than mainland koalas. Overall, NSI koalas have differentiated from nearby mainland populations, suggesting independent management and special protection are appropriate.

- 200 -

CONCLUSION

Considering the fast decline of the SEQ koala population, it is tempting to think about the NSI koala population as an island ark and as a relatively secure population that might provide a source for translocation if some mainland populations become extinct. However, owing to likely long-term isolation and a relatively small population size, the genetic variation of NSI koalas is the lowest found in SEQ (Lee et al., 2010). Consequently, koalas must be preserved on the mainland and the NSI population should not be relied upon as a source of koalas to repopulate mainland areas without running the risk of repeating the unintended consequences of genetic homogenisation experienced in southern Australia. There have been pleas to avoid translocating genetically depauperate animals from one area to another (Sherwin et al., 2000). Examples of mitigated outcomes of such translocations are already available (Cristescu et al., 2009). The NSI koalas also present other characteristics that differ from those in the rest of SEQ and cannot be thought of as a representative sample of SEQ koalas overall. Finally, although the NSI population might seem better preserved than those in mainland SEQ, NSI is by no mean a totally safe haven, NSI koalas face their own array of threats, particularly the potential vulnerability of their key habitats on the island to the effects of global warming and water-extraction to mainland SEQ. The NSI koala population is worth preserving for its own sake, for aesthetic, cultural and philosophical values, but not as an island ark.

ACKNOWLEDGEMENTS

We would like to thank the traditional owners of Minjerribah and particularly Minjerribah-Moorgumpin Elder Auntie Margaret Islen for sharing her knowledge with us.

Sibelco Australia – Mineral Sand initiated the koala research program and provided ongoing support for 10 years through direct funding, provision of logistical support and access to company sites. Thanks to Russell Miller, Craig Lockhart, Toni Burgess, Steve Rodenburg and Deb Olive for ongoing logistical support.

Thanks to Dan Carter and the Redland City Council for funding the blood-test analyses.

- 201 -

Thanks to Lisa Bailey and Jennifer Davis from IndigiScape, Redland City Council, for organising the NSI urban koala counts and providing us access to their data.

Thanks to William Woodward, Myuki Tanizaki, Maureen Myers and Katherine Crouch for access to their field work results and dietary analysis. Thanks to David Dique for expert koala catching.

Thanks to Jack Jackson, Carolyn Hahn and all members of Wildlife Rescue on NSI for their continuous dedication to help injured and sick wildlife.

Thanks to Sergei Karabut and Dot Lim for extracting koala information from DERM databases. Thanks to Melanie Shaw, DERM, for data on NSI groundwater.

The authors are most grateful for expert opinion and advice on genetic analysis provided by Dr. Jennifer Seddon, School of Veterinary Science, The University of Queensland.

AUTHOR PROFILE

Romane Cristescu is finishing a PhD on arboreal marsupials, particularly koalas, in areas rehabilitated following mining. She is part of the Koala Study Program at the Centre for Mined Land Rehabilitation CMLR at The University of Queensland. She has previously studied lowland gorillas in the Republic of the Congo.

- 202 -

LITERATURE CITED

ALTMAN, D. G. & BLAND, J. M., 2005. Standard deviations and standard errors. British Medical Journal 331: 903. AMOS, W., WORTHINGTON WILMER, J., FULLARD, K., BURG, T. M., CROXALL, J. P., BLOCH, D. & COULSON, T., 2001. The influence of parental relatedness on reproductive success. Proceedings of the Royal Society of London Series B-Biological Sciences 268: 2021-2027. ANZECC, 1998. Australia and New Zealand Environment and Conservation Council, National koala conservation strategy. (Canberra, Australian Capital Territory). AUSTRALIAN BUREAU OF STATISTICS, 2010. 3218.0 - Regional population growth, Australia, 2008-09. (Canberra). AUSTRALIAN GOVERNMENT BUREAU OF METEOROLOGY, 2004. Wind speed and direction rose. BARRY, D. H. & CAMPBELL, P. R., 1977. A survey of the mammals and herptiles of Fraser Island, with comments on the Cooloola Peninsula, North Stradbroke, Moreton and Bribie islands. Occasional Papers of Anthropology of the University of Queensland 8: 147-178. BELKHIR, K., BORSA, P., CHIKHI, L., RAUFASTE, N. & BONHOMME, F., 1996- 2004. GENETIX 4.05, logiciel sous Windows TM pour la génétique des populations. (Laboratoire Génome, Populations, Interactions, CNRS UMR 5171, Université de Montpellier II Montpellier, France). BENUSSI, G., 1975. Genesis of North Stradbroke Island. Proceedings of the Royal Society of Queensland 86: 3-8. BRAITHWAITE, L., TURNER, J. & KELLY, J., 1984. Studies on the arboreal marsupial fauna of eucalypt forests being harvested for wood pulp at Eden, N.S.W. III. Relationships between faunal densities, eucalypt occurrence and foliage nutrients, and soil parent materials. Australian Wildlife Research 11: 41-48. BRAY, J. R. & CURTIS, J. T., 1957. An ordination of the upland forest communities of southern Wisconsin. Ecological Monographs 27: 325-349. BUCKLAND, S. T., ANDERSON, D. R., BURNHAM, K. P. & LAAKE, J. L., 1993. Distance sampling: Estimating abundance of biological populations. (Chapman and Hall, reprinted 1999 by RUWPA, University of St. Andrews, Scotland. London). BUREAU OF METEOROLOGY, 2010. Climate data online. (Commonwealth - 203 -

Australia). CAHILL, V., LUNNEY, D., TIMMS, P. & COOPER, D. W., 1999. The status of the koala on Magnetic Island. Pp. 137–142. In Foundation, A. K. (ed), Proceedings of a conference on the status of the koala in 1999. Brisbane) CLARKE, K. R., 1993. Non-parametric multivariate analysis of changes in community structure. Australian Journal of Ecology 18: 117-143. CLARKE, K. R. & GORLEY, R. N., 2006. PRIMER v6: user manual/tutorial. (PRIMER-E Ltd. Plymouth, United Kingdom). CLIFTON, I., ELLIS, W., MELZER, A. & TUCKER, G., 2007. Water turnover and the northern range of the koala (Phascolarctos cinereus). Australian Mammalogy 29: 85- 88. CORK, S. J. & SANSON, G. D., 1990. Digestion and nutrition in the koala: A review. Pp. 129-144. In Lee, A., Handasyde, K. & Sanson, G. (eds), Biology of the Koala. (Surrey Beatty and Sons. Sydney, Australia). CORK, S. J., HUME, I. D. & FOLEY, W. J., 2000. Improving habitat models and their utility in koala conservation. Conservation Biology 14: 660-668. CORNUET, J. M. & LUIKART, G., 1996. Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics 144: 2001-2014. CRISTESCU, R., SHERWIN, W. B., HANDASYDE, K., CAHILL, V. & COOPER, D., 2010. Detecting bottlenecks using BOTTLENECK 1.2.02 in wild populations: the importance of the microsatellite structure. Conservation Genetics 11: 1043-1049. CRISTESCU, R., CAHILL, V., HANDASYDE, K., CARLYON, K., WHISSON, D., JOHNSON, G., HERBERT, C., WILTON, A. N., CARLSSON, B. L. J. & COOPER, D., 2009. Inbreeding and testicular abnormalities in a bottlenecked population of koalas, Phascolarctos cinereus. Wildlife Research 36: 299-308. CROWTHER, M. S., MCALPINE, C. A., LUNNEY, D., SHANNON, I. & BRYANT, J. V., 2009. Using broad-scale, community survey data to compare species conservation strategies across regions: A case study of the Koala in a set of adjacent ‘catchments’. Ecological Management & Restoration 10: S88-S96. DAFF, 2009. Kangaroo Island long term ecological research site. (Australian Government). DERM, 2009. Decline of the Koala Coast koala population: Population status in 2008.

- 204 -

Pp. 1-14. (Queensland Government). DIQUE, D. S., THOMPSON, J., PREECE, H. J., DE VILLIERS, D. L. & CARRICK, F. N., 2003. Dispersal patterns in a regional koala population in south-east Queensland. Wildlife Research 30: 281-290. DIXON, K. R. & CHAPMAN, J. A., 1980. Harmonic mean measure of animal activity areas. Ecology 61: 1040-1044. ELLIS, W. & CARRICK, F. N., 1992. Total body water and the estimation of fat in the koala (Phascolarctos cinereus). Australian Veterinary Journal 69: 229-231. ELLIS, W. & BERCOVITCH, F., 2011. Body size and sexual selection in the koala. Behavioral Ecology and Sociobiology: in press. ELLIS, W., MELZER, A. & BERCOVITCH, F., 2009. Spatiotemporal dynamics of habitat use by koalas: The checkerboard model. Behavioral Ecology and Sociobiology 63: 1181-1188. ELLIS, W., MELZER, A., CLIFTON, I. D. & CARRICK, F. N., 2010a. Climate change and the koala Phascolarctos cinereus: Energy and water. Australian Zoologist 35: 369- 377. ELLIS, W., CARRICK, F. N., LUNDGREN, P., VEARY, A. & COHEN, B., 1999. The use of faecal cuticle examination to determine the dietary composition of koalas. Australian Zoologist 31: 127 - 133 ELLIS, W., BERCOVITCH, F., FITZGIGGON, S., MELZER, A., DEVILLIERS, D. & DIQUE, D., 2010b. Koala birth seasonality and sex ratios across multiple sites in Queensland, Australia. Journal of Mammalogy 91: 177-182. ELLIS, W. A., HALE, P. T. & CARRICK, F., 2002a. Breeding dynamics of koalas in open woodlands. Wildlife Research 29: 19-25. ELLIS, W. A. H., GIRJES, A. A., CARRICK, F. N. & MELZER, A., 1993. Chlamydial infection in koalas under relatively little alienation pressure. Australian Veterinary Journal 70: 427-428. ELLIS, W. A. H., MELZER, A., CARRICK, F. N. & HASEGAWA, M., 2002b. Tree use, diet and home range of the koala (Phascolarctos cinereus) at Blair Athol, central Queensland. Wildlife Research 29: 303-311. ELLIS, W. A. H., MELZER, A., GREEN, B., NEWGRAIN, K., HINDELL, M. A. & CARRICK, F. N., 1995. Seasonal-variation in water flux, field metabolic-rate and food- consumption of free-ranging koalas (Phascolarctos cinereus). Australian Journal of

- 205 -

Zoology 43: 59-68. EPA, 2007. Report on Koala Coast Koala Surveys 2005-2006. (Environmental Protection Agency). EVANNO, G., REGNAUT, S. & GOUDET, J., 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14: 2611-2620. FRANKHAM, R., 1995. Conservation genetics. Annual Review of Genetics 29: 305- 327. FRANKHAM, R., 1998. Inbreeding and extinction: Island populations. Conservation Biology 12: 665-675. GALLERANI-LAWSON, E. J. & RODGERS, A. R., 1997. Differences in home-range size computed in commonly used software programs. Wildlife Society Bulletin 25: 721- 729. GIRJES, A. A., HUGALL, A. F., TIMMS, P. & LAVIN, M. F., 1988. Two distinct forms of Chlamydia psittaci associated with disease and infertility in Phascolarctos cinereus (koala). Infection and Immunity 56: 1897-1900. GITZEN, R. A., MILLSPAUGH, J. J. & KERNOHAN, B. J., 2006. Bandwidth selection for fixed-kernel analysis of animal utilization distributions. The Journal of Wildlife Management 70: 1334-1344. GORDON, G., 1991. Estimation of the age of the koala, Phascolarctos cinereus (Marsupialia: Phascolarctidae) from tooth wear and growth. Australian Mammalogy 14: 5-12. GORDON, G., MCGREEVY, D. G. & LAWRIE, B. C., 1990. Koala populations in Queensland: major limiting factors. Pp. 85-95. In Lee, A. K., Handasyde, K. A. & Sanson, G. D. (eds), Biology of the Koala. (Surrey Beatry & Sons. Sydney). GORDON, G., HRDINA, F. & PATTERSON, R., 2006. Decline in the distribution of Koala Phascolarctos cinereus in Queensland. Australian Zoologist 33: 345-358. HASEGAWA, M., 1995. Habitat utilisation by koalas (Phascolarctos cinereus) at Point Halloran, Queensland. (The University of Queensland. Brisbane). HOULDEN, B. A., ENGLAND, P. & SHERWIN, W. B., 1996. Paternity exclusion in koalas using hypervariable microsatellites. Journal of Heredity 87: 149-152. HUME, I., 1990. Biological basis for the vulnerability of koalas to habitat fragmentation. Pp. 32-38. In Lunney, D., Urquhart, C. A. & Reed, P. (eds), Koala

- 206 -

Summit: Managing Koalas in New South Wales. (NSW National Parks and Wildlife Service) IBM, 2009. PAWS. (SPSS Inc. Chicago). JACKSON, S., 2007. Koala: Origins of an icon. (Jacana Books. Crows Nest, Australia). JURSKIS, V. & POTTER, M., 1997. Koala surveys, ecology and conservation at Eden. Pp. 1-70. (Forest research and development division, State Forests of New South Wales). KALINOWSKI, S. T., 2004. Counting Alleles with Rarefaction: Private Alleles and Hierarchical Sampling Designs. Conservation Genetics 5: 539-543. KALINOWSKI, S. T., 2005. hp-rare 1.0: a computer program for performing rarefaction on measures of allelic richness. Molecular Ecology Notes 5: 187-189. KAVANAGH, R. P., STANTON, M. A. & BRASSIL, T. E., 2007. Koalas continue to occupy their previous home-ranges after selective logging in Callitris–Eucalyptus forest. Wildlife Research 34: 94-107. KELLER, L. F. & WALLER, D. M., 2002. Inbreeding effects in wild populations. Trends in Ecology & Evolution 17: 230-241. LEE, K., 2009. Conservation genetics of the koala (Phascolarctos cinereus) in Queensland and Northeast New South Wales. (The University of Queensland. Brisbane). LEE, K. E., SEDDON, J. M., CORLEY, S. W., ELLIS, W. A. H., JOHNSTON, S. D., VILLIERS, D. L. D., PREECE, H. J. & CARRICK, F. N., 2010. Genetic variation and structuring in the threatened koala populations of Southeast Queensland. Conservation Genetics 11: 2091-2103. LOMOLINO, M. V., 1985. Body Size of Mammals on Islands: The Island Rule Reexamined. The American Naturalist 125: 310-316. LUIKART, G., ALLENDORF, F. W., CORNUET, J. M. & SHERWIN, W. B., 1998. Distortion of allele frequency distributions provides a test for recent population bottlenecks. Journal of Heredity 89: 238-247. LUNNEY, D. & MATTHEWS, A., 1997. The changing roles of state and local government in fauna conservation outside nature reserves: a case study of koalas in New South Wales. Pp. 97-106. In Hale, P. & Lamb, D. (eds), Conservation Outside Nature Reserves. (Centre for Conservation Biology, University of Queensland. Brisbane).

- 207 -

MARTIN, J. H. D., 1975. A list of mammals from Stradbroke Island. Proceedings of the Royal Society of Queensland 86: 73-76. MARTIN, R. W., 1981. Age-specific fertility in three populations of the Koala, Phascolarctos cinereus Goldfuss, in Victoria. Wildlife Research 8: 275-283. MARTIN, R. W. & HANDASYDE, K., 1999. The Koala: Natural history, conservation and management, Australian Natural History Series. (University of New South Wales Press Ldt. Hong Kong). MASTERS, P., DUKA, T., BERRIS, S. & MOSS, G., 2004. Koalas on Kangaroo Island: from introduction to pest status in less than a century. Wildlife Research 31: 267-272. MCLEAN, N., 2003. Ecology and management of overabundant koala (Phascolarctos cinereus) populations. Pp. 1-334. (University of Melbourne. Melbourne). MCLEAN, N. & HANDASYDE, K. A., 2006. Sexual maturity, factors affecting the breeding season and breeding in consecutive seasons in populations of overabundant Victorian koalas (Phascolarctos cinereus). Australian Journal of Zoology 54: 385-392. MELZER, A., CARRICK, F., MENKHORST, P., LUNNEY, D. & JOHN, B. S., 2000. Overview, critical assessment, and conservation implications of koala distribution and abundance. Conservation Biology 14: 619-628. MESTON, A., 1895. Moreton bay and Islands. (Handbook of Excursions / Australasian Association for the Advancement of Science. Brisbane). MITCHELL, P., 1990. The home ranges and social activity of koalas - a quantitative analysis. Pp. 171-187. In Lee, A., Handasyde, K. & Sanson, G. (eds), Biology of the Koala. (Surrey Beatty and Sons. Sydney, Australia). MITCHELL, P. J., BILNEY, R. & MARTIN, R. W., 1988. Population structure and reproductive status of koalas on Raymond Island, Victoria. Australian Wildlife Research 15: 511-514. MYERS, J. H., SIMBERLOFF, D., KURIS, A. M. & CAREY, J. R., 2000. Eradication revisited: Dealing with exotic species. Trends in Ecology & Evolution 15: 316-320. NEI, M., 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89: 583-590. OSAWA, R., 1989. Road-kills of the Swamp Wallaby, Wallabia bicolor, on North Stradbroke Island, South-east Queensland. Australian Wildlife Research 16: 95-104. PEAKALL, R. O. D. & SMOUSE, P. E., 2006. Genalex 6: genetic analysis in Excel.

- 208 -

Population genetic software for teaching and research. Molecular Ecology Notes 6: 288- 295. PIRY, S., LUIKART, G. & CORNUET, J. M., 1999. BOTTLENECK: A computer program for detecting recent reductions in the effective population size using allele frequency data. Journal of Heredity 90: 502-503. PREECE, H. J., 2007. Monitoring and modelling threats to koala populations in rapidly urbanising landscapes: Koala Coast, South East Queensland, Australia. (University of Queensland. School of Geography, Planning and Architecture). PRITCHARD, J., STEPHENS, M. & DONNELLY, P., 2000. Inference of population structure from multilocus genotype data. Genetics 155: 945-959. QUEENSLAND GOVERNMENT, 2006. Nature Conservation (Koala) Conservation Plan 2006 and Management Program 2006-2016. Pp. 1-140. (Queensland Government - Environmental Protection Agency). QUEENSLAND HERBARIUM, 2009. Regional Ecosystem Description Database (REDD). (Department of Environment and Resource Management. Brisbane, Australia). RHODES, J. R., NG, C. F., VILLIERS, D. L. D., PREECE, H. J., MCALPINE, C. A. & POSSINGHAM, H. P., 2011. Using integrated population modelling to quantify the implications of multiple threatening processes for a rapidly declining population. Biological Conservation. RHODES, J. R., CALLAGHAN, J. G., MCALPINE, C. A., JONG, C. D., BOWEN, M. E., MITCHELL, D. L., LUNNEY, D. & POSSINGHAM, H. P., 2008. Regional variation in habitat-occupancy thresholds: A warning for conservation planning. Journal of Applied Ecology 45: 549-557. RODGERS, A. R., CARR, A. P., BEYER, H. L., SMITH, L. & KIE, J. G., 2007. HRT: Home Range Tools for ArcGIS. (Ontario Ministry of Natural Resources, Centre for Northern Forest Ecosystem Research. Thunder Bay, Ontario, Canada). ROGERS, R. W., 1975. North Stradroke Island: Resources and management. Proceedings of the Ecological Society of Australia 9: 296-306. SACHSER, N., DÜRSCHLAG, M. & HIRZEL, D., 1998. Social relationships and the management of stress. Psychoneuroendocrinology 23: 891-904 SCHOENER, T. W., 1981. An empirically based estimate of home range. Theoretical Population Biology 20: 281-325.

- 209 -

SEAMAN, D. E. & POWELL, R. A., 1996. An evaluation of the accuracy of kernel density estimators for home range analysis. Ecology 77: 2075-2085. SEAMAN, D. E., MILLSPAUGH, J. J., KERNOHAN, B. J., BRUNDIGE, G. C., RAEDEKE, K. J. & GITZEN, R. A., 1999. Effects of sample size on kernel home range estimates. The Journal of Wildlife Management 63: 739-747. SEYMOUR, A. M., MONTGOMERY, M. E., COSTELLO, B. H., IHLE, S., JOHNSSON, G., ST. JOHN, B., TAGGART, D. & HOULDEN, B. A., 2001. High effective inbreeding coefficients correlate with morphological abnormalities in populations of South Australian koalas (Phascolarctos cinereus). Animal Conservation 4: 211-219. SHERWIN, W. B., TIMMS, P., WILCKEN, J. & HOULDEN, B., 2000. Analysis and conservation implications of koala genetics. Conservation Biology 14: 639-649. SLOSS, C., MURRAY-WALLACE, C. & JONES, B., 2007. Holocene sea-level change on the southeast coast of Australia: a review. The Holocene 17: 999-1014. SPENCER, W. & BARRETT, R., 1984. An evaluation of the harmonic mean measure for defining carnivore activity areas. Acta Zoologica Fennica 171: 277-281. TARLINTON, R. E., MEERS, J. & YOUNG, P. R., 2006. Retroviral invasion of the koala genome. Nature 442: 79-81. THOMPSON, G. H. & WARD, W. T., 1975. Soil landscapes of North Stradbroke Island. Proceedings of the Royal Society of Queensland 86: 9-14. THOMPSON, J., 2006. The comparative ecology and population dynamics of koalas in the Koala Coast region of South-East Queensland. Pp. 1-300. (University of Queensland. Brisbane). TOBEY, J., ANDRUS, C., DOYLE, L., THOMPSON, V. & BERCOVITCH, F., 2006. Maternal effort and joey growth in koalas (Phascolarctos cinereus). Journal of Zoology 268: 423-431. TUCKER, G. M., 2008. Ecology, behaviour and growth of sub-adult koalas on St Bees Island, Central Queensland. Pp. 1-172. (Central Queensland University. Rockhampton). TUN, N. U., 1993. Re-establishment of rehabilitated koalas in the wild and their use of habitat in Sheldon, Redland Shire, south-east Queensland, with particular reference to dietary selection. (University of Queensland. Brisbane). VEITCH, C. R. & BELL, B. D., 1990. Eradication of introduced animals from the islands of New Zealand. Pp. 137-146. In Towns, D. R., Daugherty, C. H. & Atkinson, I.

- 210 -

A. E. (eds), Ecological restoration of New Zealand islands, vol. 2. Conservation Science Publication. (Department of Conservation) WEIGLER, B. J., GIRJES, A. A., WHITE, N. A., KUNST, N. D., CARRICK, F. N. & LAVIN, M. F., 1988. Aspects of the epidemiology of Chlamydiapsittaci infection in a population of koalas (Phascolarctos cinereus) in southern Queensland, Australia. Journal of Wildlife Diseases 24: 282-291. WEIR, B. S. & COCKERHAM, C. C., 1984. Estimating F-statistics for the analysis of population-structure. Evolution 38: 1358-1370. WESTMAN, W. E., 1975. Pattern and diversity in swamp and dune vegetation, North Stradbroke Island. Australian Journal of Botany 23: 339-354. WHISSON, D. A. & CARLYON, K., 2010. Temporal variation in reproductive characteristics of an introduced and abundant island population of koalas. Journal of Mammalogy 91: 1160-1167. WHITE, N. A., 1999. Ecology of the koala (Phascolarctos cinereus) in rural south-east Queensland, Australia. Wildlife Research 26: 731-744. WHITE, N. A. & KUNST, N. D., 1990. Aspect of the ecology of the koala in southeastern Queensland. Pp. 109-121. In Lee, A., Handasyde, K. & Sanson, G. (eds), Biology of the Koala. (Surrey Beatty and Sons. Sydney, Australia). WOODWARD, W., ELLIS, W. A., CARRICK, F. N., TANIZAKI, M., BOWEN, D. & SMITH, P., 2008. Koalas on North Stradbroke Island: Diet, tree use and reconstructed landscapes. Wildlife Research 35: 606-611. WOOSNAM-MERCHEZ, O., 2008. Determining presence/absence of koalas using a rapid assessment technique to assist in the design of large-scale survey. (University of Queensland. Brisbane). WORTON, B. J., 1989. Kernel methods for estimating the utilization distribution in home-range studies. Ecology 70: 164-168. WORTON, B. J., 1995. A convex hull-based estimator of home-range size. Biometrics 51: 1206-1215.

- 211 -

FIG. 1: Compilation of surveys of direct and indirect signs of koala usage and sites where no koala signs were recorded (Note: all REs are remnants unless otherwise stated)

- 212 -

TABLE 1: Results of the urban koala counts in 2008, 2009 and 2010 NA: not searched that year NB: in 2008, the count was restricted to a small part of each township and results are not directly comparable to 2009 and 2010.

density koala/ha No of of No koalas 2010 - 57 persons 2010 - ha area area surveyed surveyed density koala/ha No of of No koalas 2009 - 26 persons 2009 - ha area area surveyed surveyed density koala/ha 0 but scats >0 98 4 0.04 98 3 0.03 No of koalas 2008 - 25 persons 2008 - ha NA NA NA 4.1 4 0.98 NA NA NA 36.5 10 0.27 50 14 0.28 50 12 0.24 18.5 6 0.32 84 7 0.08 84 6 0.07 area area surveyed surveyed Point Lookout Point Urban area PointAmity Dunwich Flinders Beach

- 213 -

TABLE 2: Koala sizes calculated from healthy adult koalas captured during the program or necropsied (for koalas healthy prior to death)

Weight (kg) Head length (cm) Head width (cm) males females males females males females Mean 7.9 6.2 14.7 12.8 8.1 7.1 Range 6.9-8.9 5.0-7.1 14.1-15.3 11.7-13.8 7.6-8.8 6.7-7.6 SD 0.6 0.6 0.3 0.5 0.4 0.3 SEM 0.2 0.1 0.1 0.1 0.1 0.1 Number of 15 18 12 18 9 17 koalas

FIG. 2: Estimated month of birth for 16 koala young from NSI

- 214 -

FIG. 3: Koalas found sick or injured on NSI and evacuated to mainland hospitals and number of NSI koalas returned to and released on NSI after hospitalisation (per year)

25 hospitalised koalas 20 released koalas

year

TABLE 3: NSI koalas evacuated or found dead and total number of koala deaths (a) by age class (b) by gender and (c) by cause of evacuation when known (young were excluded from calculation of b and c)

(a) Adult Subadult Young Evacuated/ found dead 72 7 25 Total deaths 64 6 12

(b) Females Males Evacuated/ found dead 30 50 Total deaths 26 45 - 215 -

(c) Disease Dog attack Car strike confirmed assumed confirmed assumed Evacuated/ found dead 21 13 3 31 4 Total deaths 20 15 32

TABLE 4: Roosting trees used by koalas on NSI

Tree species number E. robusta 278 Callitris sp. 120 Melaleuca sp. 56 E. racemosa 53 Lophostemon sp. 49 E. tereticornis 46 Allocasuarina sp. 41 Eucalyptus pilularis 31 Banksia sp. 28 Corymbia sp. 17 Angophora sp. 16 E. resinifera 13 E. planchoniana 10 dead tree 9 Acacia sp. 7 Cinnamomum camphora 5 Elaeocarpus sp. 4 Alphitonia sp. 2 E. tindaliae 2 Schefflera sp. (umbrella tree) 2 Acmena sp. 1 Duboisia sp. 1 TOTAL 791

- 216 -

FIG. 4: Home ranges of radio-tracked koalas (N=28) projected onto a 2008 airborne laser scan of the island (Sibelco/CRL, unpublished data)

- 217 -

TABLE 5: Home range sizes for koalas having more than 20 fixes (HRT tools, kernel LSCV, scaled variances)

Koala gender home range number 50 90 95 of fixes Jundall female 27.7 102.8 132.4 40 Bundy male 22.3 79.5 85.5 23 Casanova male 14.3 54.1 66.2 48 Callitris female 11.4 43.7 57.2 40 Tubs female 8.4 39.7 51.6 41 Christine female 10.2 31.3 39.0 44 Dakabin female 8.8 28.3 34.4 39 Suzy female 8.2 26.3 33.4 79 Mirrigan female 5.5 22.9 31.6 26 Peggy-Sue female 7.2 24.9 31.3 20 Noonie female 6.5 21.8 27.4 23 Nareeba female 6.1 19.1 23.0 41 Binang male 4.9 16.3 20.3 37 Nala female 2.7 9.0 11.7 45 Becca female 2.2 6.6 8.1 45 Myora female 1.6 4.9 6.0 25 mean 8.2* 29.2* 36.0* 38.5 SD 5.4* 20.1* 22.7* 14.2 *: calculated without Jundall, which is regarded as an outlier

- 218 -

TABLE 6: Composition of koala (N=16) home ranges - breakdown by different habitats, based on RE

State of Description Area % in % Ratio vegetation (ha) home available community range on the island Disturbed mine rehabilitation 223.4 34.2 16.4 2.1 Remnant Melaleuca quinquenervia open- 121.4 18.6 3.6 5.2 RE 12.2.7 forest to woodland with E. tereticornis, C. intermedia, E. robusta, Lophostemon sp Remnant swamps 86.6 13.3 5.1 2.6 RE 12.2.15 Remnant E. racemosa, C. intermedia, C. 72.6 11.1 25.5 0.4 RE 12.2.6 gummifera, Angophora leiocarpa and E. pilularis shrubby or grassy woodland to open-forest Disturbed townships 55.2 8.5 1.6 5.3 Remnant Corymbia spp., Banksia integrifolia, 52.9 8.1 7.0 1.2 RE 12.2.5 Callitris columellaris, Acacia spp. open forest to low closed forest on beach ridges usually in southern half of bioregion Remnant E. pilularis and E. resinifera open- 25.6 3.9 3.4 1.2 RE 12.2.8 forest Remnant mallee forms of C. gummifera, E. 14.8 2.3 17.4 0.1 RE 12.2.10 racemosa and E. planchoniana ± Banksia aemula low shrubby woodland

- 219 -

FIG. 5: Example of a long distance movement by Koala 781, which travelled 6.6 km in 2.5 months, from Bayside to Brown Lake to Myora

TABLE 7: Number of alleles, heterozygosity, internal relatedness and other genetic characteristics of NSI koalas (N=36)

Standard Mean Error No of alleles/locus 3.67 0.42 No of alleles/locus more frequent than 2.83 0.31 5% No of effective alleles/locus 2.32 0.23 Expected Heterozygosity 0.55 0.04 Observed Heterozygosity 0.50 0.11

Unbiased FIS 0.09 0.00 Internal Relatedness 0.09 0.05

- 220 -

FIG. 6: Allele frequency showing a normal L-shaped distribution, where the rare alleles (< 0.1) are the most frequent (Note that no alleles are found at 1 due to all our loci being polymorphic)

- 221 -

TABLE 8: Measures of Nei’s unbiased genetic distances D between koalas from Local Government Areas in SEQ (N: number of koalas sampled) Redcliffe Redland Brisbane NSI Pine Rivers Pine 0.329 0.326 0.320 0.257 0.317 0.384 0.354 0.416 0.373 0.314 0.863 N Beaudesert Caboolture Esk Coast Gold Ipswich Logan 18 4650 0.238 48 0.20850 0.17498 0.11362 0.114 0.41518 0.266 0.305 0.317 0.427 0.200 62 0.439 0.25836 0.002 0.307 0.457 0.124 0.195 0.724 0.190 0.202 0.221 0.357 0.315 0.984 0.403 0.332 0.404 0.339 0.694 0.143 0.383 0.440 0.464 0.500 0.216 0.725 0.064 0.021 1.022 0.359 1.086 0.435 1.071 0.011 0.980 0.883 317 0.378 0.417 0.458 0.212 0.293 0.013 0.441 0.527 Mean Nei's Nei's Mean dis tance Beaudesert Caboolture Es k CoastGold Ips wich Logan Rivers Pine Redcliffe Redland Brisbane NSI

- 222 -

FIG. 7: Cluster analysis of NSI koalas, their closest geographic neighbour (Redlands) and their closest genetic neighbour (Gold Coast) based on their microsatellite genotypes (K=2) Two clusters are present, indicated in black and grey. Each koala is represented by a vertical bar composed of black and grey indicating their relative assignment to each cluster. A bar mostly composed of one colour can be considered belonging to one cluster.

- 223 -

FIG. S1: Percentage of tree species in koala scats during seven consecutive days and concurrent tree use on day 1 to 7

- 224 -

FIG. S2: NSI Future Vision Map from Queensland Government http://www.derm.qld.gov.au/stradbroke/maps.html (accessed 28 March 2011)

- 225 -

TABLE S1: Causes of admission of koalas to wildlife hospitals from Local Government Areas in SEQ (2002 to 2009) extracted from DERM databases

Percentages do not add up to 100% as koalas could present more than one cause of admission (disease and car strike)

NSI Shire Redland Shire Pine Rivers Pine City Logan City Ipswich Ipswich Shire Caboolture 6% 13% 7% 11% 13% 7% 14% 60%19%18% 44%22% 25%43% 67% 8% 12% 22% 90% 34% 23% 31% 22% 50% 29% 36% 30% 24% 23% 95% 7% 19% 35% 24% 32% 3% 30% 30% 25% 2% 19% 32% 16% 17% 14% 14% 20% 15% 31% City Brisbane Brisbane Disease Cystitis Conjunctivitis Wasted Cars Dogs Other Total koalasTotal year per Koalas 778 97 1103 138 467 58 673 84 2792 349 3081 90 385 6

- 226 -

Appendix B

Plot details

Plot Group Description Date of Coordinates (south Number Number of scats Number (mine/RE) rehab corner of plot) of scats corrected for of scat in plot decay and locations AMG 84 E AMG 84 N detectability in plot 1 Undisturbed 12.2.6 N/A 543960 6966371 25 30.0 3 2 Undisturbed 12.2.6 N/A 544916 6965151 12 14.4 1 3 Undisturbed 12.2.10 N/A 541992 6954282 5 6.0 1 4 Undisturbed 12.2.10 N/A 541980 6954129 17 20.4 2 5 Undisturbed 12.2.10 N/A 541646 6950349 5 6.0 1 6 Undisturbed 12.2.10 N/A 541322 6950518 16 19.2 1 7 Undisturbed 12.2.6 N/A 541230 6947550 14 16.8 2 8 Undisturbed 12.2.6 N/A 541390 6947601 6 7.2 2 9 Undisturbed 12.2.7 N/A 541208 6948296 79 110.9 6 10 Undisturbed 12.2.7 N/A 541283 6948115 9 12.6 2 11 Undisturbed 12.2.8 N/A 540935 6948179 51 61.2 4 12 Undisturbed 12.2.8 N/A 541133 6947525 1 1.2 1 13 Rehabilitation pre 1987 Bayside 1979 541610 6949283 0 0 0 14 Rehabilitation pre 1987 Bayside 1979 541682 6949225 0 0 0 15 Rehabilitation pre 1987 Bayside 1979 541767 6949155 0 0 0 16 Rehabilitation pre 1987 Bayside 1979 541695 6949282 0 0 0 17 Rehabilitation pre 1987 Bayside 1978 541116 6955398 48 48 3 18 Rehabilitation pre 1987 Bayside 1978 541056 6955153 3 3 2 19 Rehabilitation pre 1987 Bayside 1985 540868 6950506 9 9 1 20 Rehabilitation pre 1987 Bayside 1983 540900 6950141 9 9 1 21 Rehabilitation pre 1987 Bayside 1980 541652 6949141 27 27 3 22 Rehabilitation pre 1987 Bayside 1981 541390 6948919 29 29 2 23 Rehabilitation 1987-1997 Bayside 1987 540921 6951548 4 4 1 24 Rehabilitation 1987-1997 Bayside 1987 540741 6951462 107 107 8 25 Rehabilitation 1987-1997 Bayside 1988 541210 6951687 0 0 0 26 Rehabilitation 1987-1997 Bayside 1988 541030 6951600 0 0 0 27 Rehabilitation 1987-1997 Bayside 1988 541253 6952151 0 0 0 28 Rehabilitation 1987-1997 Bayside 1988 541073 6952064 0 0 0 29 Rehabilitation 1987-1997 Bayside 1990 541139 6952981 2 2 2 30 Rehabilitation 1987-1997 Bayside 1990 540997 6952881 8 8 3 31 Rehabilitation 1987-1997 Bayside 1990 541054 6953300 207 207 7 32 Rehabilitation 1987-1997 Bayside 1990 540877 6953214 98 98 2 33 Rehabilitation 1987-1997 Bayside 1995 542675 6952824 0 0 0 34 Rehabilitation 1987-1997 Bayside 1995 542499 6952731 0 0 0 35 Rehabilitation 1987-1997 Bayside 1995 542333 6952629 0 0 0 - 227 -

Plot Group Description Date of Coordinates (south Number Number of scats Number (mine/RE) rehab corner of plot) of scats corrected for of scat in plot decay and locations AMG 84 E AMG 84 N detectability in plot 36 Rehabilitation 1987-1997 Amity 1990 544461 6966291 119 119 5 37 Rehabilitation 1987-1997 Amity 1988.5 544627 6966011 444 444 6 38 Rehabilitation 1987-1997 Bayside 1995 542325 6953389 17 17 2 39 Rehabilitation 1987-1997 Bayside 1995 542350 6953206 39 39 3 40 Rehabilitation post 1998 Ibis 1999.5 541638 6947039 16 16 2 41 Rehabilitation post 1998 Ibis 2001 541357 6947112 6 6 3 42 Rehabilitation post 1998 Ibis 1998 542734 6947016 0 0 0 43 Rehabilitation post 1998 Ibis 1998 542589 6947150 0 0 0 44 Rehabilitation post 1998 Ibis 1998 542447 6947283 0 0 0 45 Rehabilitation post 1998 Ibis 1999 542327 6947021 0 0 0 46 Rehabilitation post 1998 Ibis 1999 542196 6946878 0 0 0 47 Rehabilitation post 1998 Ibis 1999 542059 6946730 0 0 0 48 Rehabilitation post 1998 Ibis 1999 541922 6947084 0 0 0 49 Rehabilitation post 1998 Ibis 1999 541782 6946946 0 0 0 50 Rehabilitation post 1998 Ibis 2000 542076 6947252 0 0 0 51 Rehabilitation post 1998 Ibis 2000 541728 6947435 0 0 0 52 Rehabilitation post 1998 Ibis 2000 541589 6947291 0 0 0 53 Rehabilitation post 1998 Ibis 2000 541523 6947520 0 0 0 54 Rehabilitation post 1998 Ibis 2000 541371 6947232 1 1 1 55 Rehabilitation post 1998 Ibis 2000 541511 6947000 0 0 0 56 Rehabilitation post 1998 Ibis 2000 541409 6946712 0 0 0 57 Rehabilitation post 1998 Ibis 2000 541617 6946690 0 0 0 58 Rehabilitation post 1998 Ibis 2001 541116 6946791 0 0 0 59 Rehabilitation post 1998 Ibis 2001 541380 6946485 0 0 0 60 Rehabilitation post 1998 Ibis 2001 541200 6946399 8 8 1 61 Rehabilitation post 1998 Ibis 2001 541020 6946312 38 38 1 62 Rehabilitation post 1998 Ibis 2001 541267 6946120 0 0 0 63 Rehabilitation post 1998 Ibis 2001 541043 6946013 416 416 5 64 Rehabilitation post 1998 Ibis 2001 541467 6945747 61 61 3 65 Rehabilitation post 1998 Ibis 2002 541534 6945586 21 21 1 66 Rehabilitation post 1998 Ibis 2002 541462 6945163 0 0 0

- 228 -

Plot Number of trees Acacia Allocasuarina Allocasuarina Angophora Banksia Callitris Corymbia concurrens littoralis torulosa leieocarpa sp sp sp

1 28 154 2 31 8 3 29 5 4 59820 5 2314 6 47 47 7 41371 8 20 1 57 9 11 14351 10 24 731 11 71510 12 1351 13 10 4 2 18 14 49 15 13 12 9 16 22 21 17 16 312 6 3 7 203 18 7 46 4 136 1 19 49 11 3 1 20 65 8 1 21 34 12 3 62 1 22 19 5 1 32 1 23 19 1 1 6 1 24 10 17 3 25 61 82 26 40 1 1 1 5 1 27 163 2 106 28 116 1114 29 2241 11106 30 8 16 1 12 16 31 20 4 3 2 32 1302 11123 33 22 16 2 4 34 433 411 35 7194 811

- 229 -

Plot Number of trees Acacia Allocasuarina Allocasuarina Angophora Banksia Callitris Corymbia concurrens littoralis torulosa leieocarpa sp sp sp

36 242 4106 37 31 6 7 38 1 122 1 3 1 2 39 3 104 1 1 40 36 3 6 78 41 6161 33 42 1112 21247 43 1201 143 44 2121 14 8 45 12 1 412 46 20 5 4 4 1 47 1187 61 48 1361 4 1 49 1161 3 3 50 51 17 11 18 51 15 24 3 2 52 15 2 7 53 33 4 54 32 2 1 7 55 22 11 9 56 32 6 1 9 57 1324 6 58 56 2110 59 24 7 1 4 1 17 60 19 13 1 33 1 21 61 13 10 1 11 9 62 25 17 2 3 19 63 11219 19 64 20 8 1 1 5 65 21 11112 66 16 41 1

- 230 -

Plot Number of trees E. E. E. E E. E. E Lophostemon Melaleuca pilularis plancho- racemosa resinifera robusta tereticornis tindaliae confertus niana 1 1 2 26 3 34 4 62 1 5 4 6 2 7 5 8 21 9 2 10 15 11 14 12 3 13 112 14 15 8 15 531 16 10 2 4 1 17 32 1 18 12 48 19 23 7 2 20 11 21 2163 22 531 23 11 1 24 2213 25 1 26 1 27 21 28 1 29 221 1 30 545 1 31 2141 1 2 1 32 451 1 33 253 34 8165 1 35 774

- 231 -

Plot Number of trees E. E. E. E E. E. E Lophostemon Melaleuca pilularis plancho- racemosa resinifera robusta tereticornis tindaliae confertus niana 36 334 2 1 1 37 375 3 1 38 72639 39 15 23 16 2 7 40 16 32 22 8 7 41 13 8 5 1 42 3106 1 1 43 6 1 44 55 45 6 46 26 47 6179 48 9203 1 49 372 1 50 10 22 7 4 51 6135 1 52 5251 53 2121 54 4872 1 55 14 21 10 1 2 56 11 25 8 1 3 57 8158 3 2 58 6322 1 59 653 1 60 5521 61 33 4 62 482 1 63 7582 64 94131 3 65 31 66 131 1

- 232 -

Plot Native Native Native tree Native Tree Proportion Density of Richness of CBH tree tree dens ity Richness of Eucalypt Eucalypt Eucalypt Eucalypt dens ity richness (without A. (without A. &Corymbia &Corymbia &Corymbia &Corymbia by littoralis nor littoralis nor by hectare by hectare in cm hectare Acacia ) by Acacia ) hectare 1 600 5 560 4 0.14 80 1 242.8 2 940 4 940 4 0.34 320 3 52.1 3 1360 3 1360 3 0.57 780 2 47.1 4 1020 8 920 7 0.63 580 4 56.2 5 820 4 820 4 0.20 160 2 90.4 6 1880 3 960 2 0.04 40 1 134.5 7 960 5 880 4 0.14 120 2 110.5 8 1600 5 1200 4 0.05 60 2 131.8 9 360 8 320 6 0.06 20 1 120.0 10 460 7 420 6 0.05 20 1 143.0 11 560 6 420 5 0.52 220 2 123.0 12 260 5 240 4 0.33 80 2 121.9 13 730 7 530 6 0.81 430 4 49.0 14 440 4 360 3 0.50 180 2 49.0 15 820 6 560 5 0.59 330 4 49.0 16 410 8 380 7 0.84 320 5 49.0 17 9540 9 5040 7 0.13 660 2 22.6 18 5080 7 4020 5 0.06 260 2 76.1 19 1560 8 580 7 0.41 240 3 20.2 20 1520 5 220 4 0.27 60 3 61.0 21 2660 8 1980 7 0.22 440 4 48.6 22 1340 8 960 7 0.21 200 4 44.4 23 570 8 190 7 0.26 50 4 24.9 24 720 9 530 8 0.38 200 6 49.3 25 350 6 240 5 0.25 60 2 29.5 26 950 8 150 7 0.13 20 2 29.0 27 1680 10 420 8 0.52 220 4 22.4 28 650 8 320 6 0.03 10 1 16.0 29 930 11 430 9 0.53 230 5 29.6 30 1290 10 820 8 0.73 600 5 33.4 31 760 12 330 10 0.70 230 8 20.2 32 1150 12 540 10 0.44 240 5 20.9 33 1040 8 720 6 0.39 280 4 18.8 34 1400 9 670 7 0.87 580 4 13.2 35 1120 10 600 8 0.58 350 4 8.0

- 233 -

Plot Native Native Native tree Native Tree Proportion Density of Richness of CBH tree tree dens ity Richness of Eucalypt Eucalypt Eucalypt Eucalypt dens ity richness (without A. (without A. &Corymbia &Corymbia &Corymbia &Corymbia by littoralis nor littoralis nor by hectare by hectare in cm hectare Acacia ) by Acacia ) hectare 36 1560 11 680 9 0.53 360 5 31.8 37 1260 8 640 7 0.78 500 5 28.3 38 4040 10 1580 8 0.94 1480 4 14.7 39 3440 9 1300 7 0.86 1120 4 18.7 40 4160 10 3440 9 0.91 3120 6 20.3 41 1660 8 1220 6 0.97 1180 4 14.0 42 1120 15 880 13 0.56 490 6 20.4 43 860 9 450 7 0.24 110 1 12.1 44 740 10 460 8 0.24 110 4 27.9 45 310 8 250 6 0.60 150 3 14.5 46 790 8 400 7 0.40 160 3 6.2 47 1250 8 880 6 0.69 610 3 8.9 48 1470 10 750 8 0.87 650 4 7.9 49 700 10 370 8 0.78 290 5 19.5 50 2740 9 1720 8 0.65 1110 5 13.8 51 1350 9 1060 8 0.49 520 5 7.5 52 700 8 410 7 0.90 370 6 7.9 53 820 7 160 6 1.00 160 6 10.8 54 1220 10 580 9 0.91 530 6 13.8 55 1770 9 1330 8 0.80 1070 6 9.1 56 1880 10 1250 9 0.85 1060 6 7.4 57 1530 10 890 9 0.87 770 6 11.5 58 680 11 590 10 0.73 430 6 17.7 59 1310 11 840 10 0.69 580 5 15.0 60 1330 10 950 9 0.69 660 6 21.6 61 1050 9 790 8 0.35 280 4 26.9 62 1580 11 1090 10 0.60 650 5 16.5 63 1240 10 990 8 0.60 590 6 17.9 64 1240 11 850 10 0.71 600 6 19.0 65 200 9 160 7 0.69 110 3 17.1 66 350 9 220 7 0.45 100 4 10.2

- 234 -

Plot Heigth Canopy Ground Ground Ground El e vati on Aspect Slope Distance Distance to Eucalypt cover cover: cover: cover: in m in % to undis turbed &Corymbia % litter plants bare degrees swamps in m in m % % % in m 1 31.7 100 93 4 3 24.8 309.5 26.0 120 0 2 7.0 72 92 5 3 41.4 144.5 19.0 1380 0 3 7.3 90 80 20 0 94.7 328.0 11.5 1240 0 4 7.0 94 89 11 1 94.9 287.0 27.5 1235 0 5 12.1 80 86 14 0 53.1 286.5 5.5 970 0 6 14.2 88 71 28 1 42.1 166.5 10.5 680 0 7 13.1 66 86 2 12 64.7 293.5 17.5 240 0 8 11.8 34 82 4 14 94.4 253.0 9.5 285 0 9 13.0 64 92 8 0 29.7 127.0 9.5 190 0 10 20.1 54 76 19 5 28.3 127.0 7.5 145 0 11 16.1 80 86 14 0 14.8 315.5 18.5 40 0 12 15.4 66 86 10 5 36.9 268.5 22.0 100 0 13 10.0 64 78 7 15 56.1 299.0 40.5 750 50 14 10.0 60 75 22 3 67.8 87.5 3.0 820 150 15 10.0 64 86 14 0 69.2 212.0 5.6 900 110 16 10.0 32 89 11 0 63.9 194.0 20.2 830 95 17 7.3 88 98 0 2 48.9 180.5 23.5 260 200 18 15.0 92 97 0 3 24.2 68.5 9.0 220 100 19 5.2 78 93 1 7 75.9 278.0 17.5 195 185 20 10.7 90 97 0 3 58.4 324.5 34.5 200 200 21 10.3 82 95 0 4 70.8 302.5 10.0 255 255 22 8.5 70 90 9 1 68.6 66.5 2.5 530 250 23 5.9 40 84 14 2 78.4 329.5 23.4 215 190 24 11.1 72 88 0 12 21.3 311.5 41.6 70 60 25 5.3 26 2 98 0 85.5 322.0 17.9 445 180 26 6.9 76 0 100 0 80.1 9.5 13.5 310 270 27 5.3 56 0 100 0 102.9 129.0 15.5 430 70 28 4.2 54 6 92 2 61.9 276.5 38.0 280 80 29 6.7 60 75 22 4 74.0 117.0 46.2 380 50 30 6.8 68 94 6 0 67.4 260.5 24.7 250 210 31 5.5 46 88 4 8 76.6 157.0 18.5 340 60 32 7.9 88 96 2 2 52.0 275.5 17.9 170 110 33 5.1 74 82 11 7 112.7 101.5 34.2 1920 70 34 3.5 32 57 18 25 145.6 166.5 9.0 1730 220 35 2.9 34 57 37 5 130.6 267.5 38.7 1550 95

- 235 -

Plot heigth Canopy Ground Ground Ground El e vati on Aspect Slope Distance Distance to Euc al ypts cover cover: cover: cover: in m % to undis turbed &Corymbia % litter plants bare % swamps in m in m % % in m 36 6.2 76 99 0 1 46.1 143.5 11.5 585 290 37 4.7 84 92 6 2 54.1 4.0 4.0 790 310 38 4.7 64 95 1 5 100.2 212.5 5.0 1575 435 39 4.9 90 97 1 3 119.7 47.0 20.5 1590 370 40 4.8 58 61 0 39 51.7 235.0 13.0 425 380 41 3.2 26 61 1 38 49.4 200.0 5.5 125 70 42 4.5 50 50 0 50 68.6 151.5 47.6 850 50 43 3.2 28 42 4 54 108.2 15.5 34.0 780 160 44 3.2 36 82 4 14 46.2 19.0 43.3 600 20 45 3.5 12 52 27 21 89.3 201.5 26.5 690 260 46 2.4 18 13 58 29 112.4 210.0 20.4 530 220 47 2.9 16 16 60 24 63.7 204.0 39.9 430 40 48 2.6 10 20 57 23 116.2 244.0 13.8 540 280 49 4.4 34 6 74 19 61.1 286.0 36.8 550 330 50 3.9 68 50 24 26 76.1 61.5 23.6 395 60 51 2.7 24 72 10 18 102.1 62.5 35.7 230 80 52 2.7 18 58 28 14 92.2 259.0 22.8 375 270 53 3.4 34 16 72 12 100.3 310.5 13.7 250 50 54 3.3 24 64 10 26 57.9 269.0 17.1 160 170 55 3.0 30 36 52 12 55.2 288.5 9.4 275 310 56 2.7 42 56 14 30 57.6 302.0 24.8 360 350 57 3.1 20 62 8 30 69.3 98.5 9.8 510 230 58 4.3 16 46 6 48 58.2 89.0 21.7 210 70 59 3.7 54 28 0 72 52.0 118.5 9.2 480 180 60 4.4 40 68 2 30 68.3 15.5 4.9 322 220 61 5.2 56 70 12 18 39.6 285.0 9.1 100 30 62 3.6 24 44 6 50 69.7 204.0 16.3 255 60 63 4.5 46 60 0 40 38.2 295.0 16.0 90 30 64 4.2 22 46 2 52 54.4 127.5 4.4 270 90 65 2.9 26 45 41 14 63.7 144.0 12.6 295 60 66 3.4 14 44 30 26 43.6 301.5 27.2 140 90

- 236 -

Appendix C

Fine scale movements of koalas in rehabilitated and undisturbed areas

As highlighted in Chapter 5, more research needs to be conducted on fine scale movements of koalas in rehabilitated and undisturbed areas. This will enable us to be more definite in our assertion that rehabilitated areas can provide new quality habitat for koalas. Fine time-scale data on positions of koalas can allow us to deepen the analysis of the patterns of use of rehabilitated and undisturbed areas. We could determine for instance, the frequency of movements between the two areas and the time spent in each area, the use of the two areas in relation to koala circadian rhythm, or the travel time within each area. These are the avenues the research program will be focusing on from now on. So far, only two koalas found to use rehabilitated areas have been studied to answer these questions, using technology allowing recording of koala locations on earth at very frequent interval compared to VHF tracking. In the future, the research program goal is to integrate 10 koalas as part of the fine scale movement study.

Material and methods

The two koalas, Christine and Nareeba, were fitted with collar-mounted GPS loggers (Sirtrack, Havelock North, New Zealand), programmed to record and store positional fixes at 4-hourly intervals from 2AM, as in Ellis et al. (2011). GPS collared were deployed the 9/10/10 and recovered the 29/03/11 (both collars were found on the ground).

GPS collars had previously been calibrated (Ellis et al. 2011), and highest dilution of precision (HDOP) inferior to 2.1 were found to be accurate at 8.58m (SEM 0.64). Hence we selected and analysed only fixes with HDOP<2.1.

GPS coordinates were plotted on a 2008 airborne laser scan of the island superimposed with the contours of rehabilitated areas (Sibelco/CRL, unpublished data). The GPS - 237 - fixes of the two koalas were classified accordingly to (1) their occurrence in rehabilitated and undisturbed areas and (2) koala circadian time (day: 6AM to 6PM, night: 6PM to 6AM on the basis of koala activity in Queensland found in Ellis et al. 2011). Distances between GPS fixes were calculated by Hawth's Analysis Tools 3.27 (Beyer 2002-2006)

All variables were tested for normality and homogeneity of variances (Levene’s test of homoscedasticity), and were compared between rehabilitated and undisturbed areas by appropriate non-parametric tests in PAWS Statistics 18.0 (IBM 2009). Significance level was taken to be p<0.05, the standard deviation (SD) was calculated to describe the variability inside the samples (Altman & Bland 2005).

Results

We recorded 224 GPS locations (62 days) for Christine (Figure 1) and 287 GPS locations (138 days) for Nareeba (Figure 2).

For Christine, the data collected with the GPS collar confirmed that this koala was living 100% in rehabilitated areas. In the 62 days of GPS radio-tracking, Christine only used parts of the area used in 840 days.

- 238 -

Figure 1: GPS locations of Christine from VHF (44 points, 840 days) and GPS collars (224 points, 62 days)

For Nareeba, the data collected with the GPS collar confirmed the use of both rehabilitated and undisturbed areas, the areas used during the 138 days of GPS radio- tracking matched the areas found with 840 days of VHF radio-tracking.

- 239 -

Figure 2: GPS locations of Nareeba from VHF (41 points, 840 days) and GPS collars (287 points, 138 days)

For Nareeba, the integral range of GPS abilities can be used as Nareeba occupied both rehabilitated and undisturbed areas. Nareeba alternated between rehabilitated and undisturbed areas 46 times in 138 days, spending on average 1.6 (SD=1.6) days in rehabilitated areas and 3.1 (SD=3.0, Figure 3) days in undisturbed areas, these average did not significantly differ (Mann-Whitney U test= 186.5, p=0.051).

- 240 -

Figure 3: Boxplots comparing the time spend between rehabilitated and undisturbed areas for Nareeba

The association between rehabilitated and undisturbed areas and the time at which the fixes was recorded was weak but significant (Cramer’s V=0.156, p=0.008), and suggested more fixes at night were taken in undisturbed areas (Figure 4).

Figure 4: Comparison between the numbers of fixes recorded during day time and night time in rehabilitated and undisturbed areas for Nareeba

Travel distances between fixes were classified in three categories: (1) point of departure

- 241 - and arrival inside rehabilitated area, (2) point of departure and arrival inside undisturbed area, and (3) point of departure and arrival in a different area. The mean distances travelled increased from 31m (SD= 39) in rehabilitated area, to 48m (SD=51) in undisturbed area and to 154m (SD=82, Figure 5) when moving between the two areas (Kruskal-Wallis test=79.51, df=2, p<0.001). This trend held when only rehabilitated and undisturbed areas were compared (Mann-Whitney U test=5354, p=0.006).

Figure 5: Boxplots comparing the distances travelled between two fixes in rehabilitated areas, in undisturbed areas and between the two for Nareeba

Discussion and conclusion

GPS collars can be used to investigate more deeply the use of rehabilitated areas by koalas. One concern when we used VHF radio-tracking (Chapter 5) was the amount of time between two fixes, in between which the whereabouts of koalas were unknown. Although no generalisation can be given on the basis of the limited sample in this Appendix, some interesting results already emerged.

Koala choice for roosting trees can be influenced by thermoregulatory constraints (Clifton et al. 2007; Ellis et al. 2009; Ellis et al. 2010). Thus, by demonstrating rehabilitated areas are used by koalas based only on daytime VHF tracking, it could be argued that koalas use rehabilitated areas to protect themselves from the elements in the

- 242 - dense foliage of Callitris trees for example (as Callitris was the tree most frequently used by koalas in rehabilitated area, see Chapter 5), and move to undisturbed areas at night to feed. In this scenario, rehabilitated areas would not provide additional habitat, as not all the necessary elements for koala survival would be found in rehabilitated areas. Contrary to this hypothesis, we confirmed that koalas could survive whilst spending 100% of their time in rehabilitated areas (Christine, Figure 1). For Nareeba, which used both rehabilitated and undisturbed areas, the use of the rehabilitated area was not restrained to daytime activity (Figure 4). Nareeba could spend several consecutive days in each area. However, on the basis of our short term study (138 days), Nareeba could be spending more night time in undisturbed areas. This is an important trend to confirm or disprove, and hopefully the data collected on 10 animals in the future will help resolve this matter.

We made several hypotheses regarding the relative distances travelled in rehabilitated and undisturbed areas in Chapter 5 (Table 1).

Table 1: Hypotheses for relative distances travelled by koalas in rehabilitated and undisturbed areas

In rehabilitated areas consequences relative distances smaller trees change trees more often to gain access rehabilitated>undisturbed to the same quantity of foliage proportionately more tip growth change trees less often to gain access to rehabilitated

For Nareeba, the relative distances could be slightly smaller in rehabilitated than in undisturbed areas, however considering the precision of the GPS collars, our hypothesis so far is that the distances are probably equivalent. The longer distances travelled when changing area from rehabilitated to undisturbed areas (or the opposite) can be due to chance. Indeed, when Nareeba travelled a longer distance she was more likely to end her travel in another area than if she only travelled a short distance.

- 243 -

All the results presented above are only given to illustrate what questions can be investigated by using GPS collars. Obviously, these are only examples, and more koalas need to be included in this research before answering our questions relative to fine scale movements of koalas.

Acknowledgements

Thanks to Dr. Bill Ellis and Dr. Sean Fitzgibbon from the KEG for providing me access to their GPS data. Sibelco Australia – Mineral sand provided ongoing support to this research through the provision of logistical support, access to company sites and relevant databases/maps.

References

Altman, D.G., and J.M. Bland, 2005. Standard deviations and standard errors. British Medical Journal 331: 903. Beyer, H.L., 2002-2006. Hawth's Analysis Tools. Version 3.27. Clifton, I., W. Ellis, A. Melzer, and G. Tucker, 2007. Water turnover and the northern range of the koala (Phascolarctos cinereus). Australian Mammalogy 29: 85-88. Ellis, W., F. Bercovitch, S. FitzGibbon, P. Roe, J. Wimmer, A. Melzer, and R. Wilson, 2011. Koala bellows and their association with the spatial dynamics of free-ranging koalas. Behavioral Ecology doi:10.1093/beheco/arq216. Ellis, W., A. Melzer, and F. Bercovitch, 2009. Spatiotemporal dynamics of habitat use by koalas: The checkerboard model. Behavioral Ecology and Sociobiology 63: 1181- 1188. Ellis, W., A. Melzer, I.D. Clifton, and F.N. Carrick, 2010. Climate change and the koala Phascolarctos cinereus: Energy and water. Australian Zoologist 35: 369-377. IBM, 2009. PAWS. Version 18.0. SPSS Inc., Chicago.

- 244 -

Appendix D

Blood test results from eight koalas included in Chapter 5

Koala 1 = Dakabin

HAEMATOLOGY Reference Range RBC 3.0 x1012/L(2.7-4.2) HAEMOGLOBIN 106 g/L (88-140) HAEMATOCRIT 0.36 L/L (0.29-0.44) MCV 120 H fL (94-117) MCH 35 pg MCHC 294 L g/L (298-330) PLATELETS Clumped and adequate PLATELET COUNT 153 L x109/L (222-558) WCC 4.6 x109/L (2.8-11.2) NEUTROPHIL 56% 2.6 x109/L (0.5-6.3) LYMPHOCYTE 39% 1.8 x109/L (0.2-5.8) MONOCYTE 0% < 0.1 x109/L (< 0.7) EOSINOPHIL 6% 0.3 x109/L (< 1.2) BASOPHIL 0% < 0.1 x109/L (< 0.3) NUCLEATED RBCS 31 H /100wbc (< 21)

White blood cells are partially degenerate/disintergrated, hence the differential count may be in error. Red cell morphology normal.

BIOCHEMISTRY Reference Range SODIUM 138 mmol/L (132-145) POTASSIUM 5.1 mmol/L (3.5-6.8) CHLORIDE 91 L mmol/L (93-107) BICARBONATE 20 mmol/L (12-30) NA:K RATIO 27.1 ANION GAP 32.1 mmol/L (11.3-32.3) GLUCOSE, SERUM 6.5 mmol/L UREA < 0.8 mmol/L (0.2-6.6) CREATININE 0.09 mmol/L (0.08-0.15) CALCIUM 2.4 mmol/L (2.3-3.0) PHOSPHATE 0.5 L mmol/L (0.8-2.0) CA:P RATIO 4.8 PROTEIN, TOTAL 68 g/L (58-83) ALBUMIN 42 g/L (34-50) GLOBULIN 26 g/L (18-39) BILIRUBIN, TOTAL 4 umol/L (< 9) ALP 271 H IU/L (25-219) AST 24 IU/L (< 135) ALT 16 IU/L (< 237) CK 247 IU/L CHOLESTEROL 2.0 mmol/L (1.1-3.1) GAMMA GT 7 IU/L (< 17) SAMPLE APPEARANCE Normal

No significant abnormalities.

- 245 -

Koala 2 = Christine

BIOCHEMISTRY Reference Range SODIUM 137 mmol/L (132-145) POTASSIUM 4.5 mmol/L (3.5-6.8) CHLORIDE 93 mmol/L (93-107) BICARBONATE 16 mmol/L (12-30) NA:K RATIO 30.4 ANION GAP 32.5 H mmol/L (11.3-32.3) GLUCOSE, SERUM 6.5 mmol/L UREA 4.2 mmol/L (0.2-6.6) CREATININE 0.10 mmol/L (0.08-0.15) CALCIUM 2.6 mmol/L (2.3-3.0) PHOSPHATE 1.4 mmol/L (0.8-2.0) CA:P RATIO 1.9 PROTEIN, TOTAL 73 g/L (58-83) ALBUMIN 41 g/L (34-50) GLOBULIN 32 g/L (18-39) BILIRUBIN, TOTAL 1 umol/L (< 9) ALP 584 H IU/L (25-219) AST 23 IU/L (< 135) ALT 21 IU/L (< 237) CK 99 IU/L CHOLESTEROL 2.3 mmol/L (1.1-3.1) GAMMA GT 6 IU/L (< 17) SAMPLE APPEARANCE Normal

Significance of mildly high ALP in this species is doubtful. Koala 3 = Callitris

BIOCHEMISTRY Reference Range SODIUM 137 mmol/L (132-145) POTASSIUM 4.4 mmol/L (3.5-6.8) CHLORIDE 92 L mmol/L (93-107) BICARBONATE 19 mmol/L (12-30) NA:K RATIO 31.1 ANION GAP 30.4 mmol/L (11.3-32.3) GLUCOSE, SERUM 7.1 mmol/L UREA 0.9 mmol/L (0.2-6.6) CREATININE 0.11 mmol/L (0.08-0.15) CALCIUM 2.7 mmol/L (2.3-3.0) PHOSPHATE 0.6 L mmol/L (0.8-2.0) CA:P RATIO 4.5 PROTEIN, TOTAL 68 g/L (58-83) ALBUMIN 42 g/L (34-50) GLOBULIN 26 g/L (18-39) BILIRUBIN, TOTAL 5 umol/L (< 9) ALP 191 IU/L (25-219) AST 21 IU/L (< 135) ALT 8 IU/L (< 237) CK 515 IU/L CHOLESTEROL 2.5 mmol/L (1.1-3.1) GAMMA GT 8 IU/L (< 17) SAMPLE APPEARANCE Normal

Minor changes of doubtful significance.

- 246 -

Koala 4 = Binang

HAEMATOLOGY Reference Range RBC 4.0 x1012/L(2.7-4.2) HAEMOGLOBIN 123 g/L (88-140) HAEMATOCRIT 0.42 L/L (0.29-0.44) MCV 105 fL (94-117) MCH 31 pg MCHC 293 L g/L (298-330) PLATELETS Clumped and adequate PLATELET COUNT 170 L x109/L (222-558) WCC 4.2 x109/L (2.8-11.2) NEUTROPHIL 65% 2.7 x109/L (0.5-6.3) LYMPHOCYTE 27% 1.1 x109/L (0.2-5.8) MONOCYTE 2% < 0.1 x109/L (< 0.7) EOSINOPHIL 6% 0.3 x109/L (< 1.2) BASOPHIL 0% < 0.1 x109/L (< 0.3) NUCLEATED RBCS 9 /100wbc (< 21)

Red cell morphology normal. White blood cells are partially degenerate/disintergrated, hence the differential count may be in error.

BIOCHEMISTRY Reference Range SODIUM 133 mmol/L (132-145) POTASSIUM 4.2 mmol/L (3.5-6.8) CHLORIDE 91 L mmol/L (93-107) BICARBONATE 24 mmol/L (12-30) NA:K RATIO 31.7 ANION GAP 22.2 mmol/L (11.3-32.3) GLUCOSE, SERUM 6.4 mmol/L UREA 2.0 mmol/L (0.2-6.6) CREATININE 0.08 mmol/L (0.08-0.15) CALCIUM 2.5 mmol/L (2.3-3.0) PHOSPHATE 1.1 mmol/L (0.8-2.0) CA:P RATIO 2.3 PROTEIN, TOTAL 68 g/L (58-83) ALBUMIN 40 g/L (34-50) GLOBULIN 28 g/L (18-39) BILIRUBIN, TOTAL 5 umol/L (< 9) ALP 117 IU/L (25-219) AST 18 IU/L (< 135) ALT 10 IU/L (< 237) CK 133 IU/L CHOLESTEROL 1.6 mmol/L (1.1-3.1) GAMMA GT 4 IU/L (< 17) SAMPLE APPEARANCE Normal

No significant abnormalities.

- 247 -

Koala 5 = Nareeba

BIOCHEMISTRY Reference Range SODIUM 139 mmol/L (132-145) POTASSIUM 4.7 mmol/L (3.5-6.8) CHLORIDE 94 mmol/L (93-107) BICARBONATE 14 mmol/L (12-30) NA:K RATIO 29.6 ANION GAP 35.7 H mmol/L (11.3-32.3) GLUCOSE, SERUM 6.3 mmol/L UREA 2.6 mmol/L (0.2-6.6) CREATININE 0.10 mmol/L (0.08-0.15) CALCIUM 2.6 mmol/L (2.3-3.0) PHOSPHATE 1.5 mmol/L (0.8-2.0) CA:P RATIO 1.7 PROTEIN, TOTAL 69 g/L (58-83) ALBUMIN 41 g/L (34-50) GLOBULIN 28 g/L (18-39) BILIRUBIN, TOTAL 4 umol/L (< 9) ALP 224 H IU/L (25-219) AST 41 IU/L (< 135) ALT 13 IU/L (< 237) CK 442 IU/L CHOLESTEROL 2.2 mmol/L (1.1-3.1) GAMMA GT 6 IU/L (< 17) SAMPLE APPEARANCE Normal

No significant abnormalities. Koala 6 = Mirrigan

BIOCHEMISTRY Reference Range SODIUM 141 mmol/L (132-145) CHLORIDE 84 L mmol/L (93-107) BICARBONATE < 5 L mmol/L (12-30) GLUCOSE, SERUM 6.6 mmol/L UREA 2.7 mmol/L (0.2-6.6) CREATININE 0.09 mmol/L (0.08-0.15) PHOSPHATE 1.3 mmol/L (0.8-2.0) PROTEIN, TOTAL 74 g/L (58-83) ALBUMIN 45 g/L (34-50) GLOBULIN 29 g/L (18-39) BILIRUBIN, TOTAL 5 umol/L (< 9) AST 31 IU/L (< 135) ALT 20 IU/L (< 237) CK 370 IU/L CHOLESTEROL 2.8 mmol/L (1.1-3.1) GAMMA GT 7 IU/L (< 17) SAMPLE APPEARANCE Normal

It is suspected that this extremely low bicarb result is spurious (eg prolonged exposure to air or storage artefact).

- 248 -

Koala 7 = Jundall

HAEMATOLOGY Reference Range RBC 3.7 x1012/L(2.7-4.2) HAEMOGLOBIN 119 g/L (88-140) HAEMATOCRIT 0.39 L/L (0.29-0.44) MCV 105 fL (94-117) MCH 32 pg MCHC 305 g/L (298-330) PLATELETS Normal PLATELET COUNT 251 x109/L (222-558) WCC 4.0 x109/L (2.8-11.2) NEUTROPHIL 70% 2.8 x109/L (0.5-6.3) LYMPHOCYTE 25% 1.0 x109/L (0.2-5.8) MONOCYTE 3% 0.1 x109/L (< 0.7) EOSINOPHIL 1% < 0.1 x109/L (< 1.2) BASOPHIL 1% < 0.1 x109/L (< 0.3)

Red cell morphology normal. White blood cells are partially degenerate/disintergrated, hence the differential count may be in error.

BIOCHEMISTRY Reference Range SODIUM 136 mmol/L (132-145) POTASSIUM 5.0 mmol/L (3.5-6.8) CHLORIDE 97 mmol/L (93-107) BICARBONATE 14 mmol/L (12-30) NA:K RATIO 27.2 ANION GAP 30.0 mmol/L (11.3-32.3) GLUCOSE, SERUM 5.3 mmol/L UREA 2.4 mmol/L (0.2-6.6) CREATININE 0.10 mmol/L (0.08-0.15) CALCIUM 2.5 mmol/L (2.3-3.0) PHOSPHATE 1.6 mmol/L (0.8-2.0) CA:P RATIO 1.6 PROTEIN, TOTAL 65 g/L (58-83) ALBUMIN 42 g/L (34-50) GLOBULIN 23 g/L (18-39) BILIRUBIN, TOTAL 4 umol/L (< 9) ALP 679 H IU/L (25-219) AST 31 IU/L (< 135) ALT 27 IU/L (< 237) CK 1215 IU/L CHOLESTEROL 2.4 mmol/L (1.1-3.1) GAMMA GT 7 IU/L (< 17) SAMPLE APPEARANCE Normal

Minor changes only.

- 249 -

Koala 8 = Swampy

BIOCHEMISTRY Reference Range SODIUM 137 mmol/L (132-145) POTASSIUM 6.0 mmol/L (3.5-6.8) CHLORIDE 92 L mmol/L (93-107) BICARBONATE 16 mmol/L (12-30) NA:K RATIO 22.8 ANION GAP 35.0 H mmol/L (11.3-32.3) GLUCOSE, SERUM 7.3 mmol/L UREA 2.3 mmol/L (0.2-6.6) CREATININE 0.11 mmol/L (0.08-0.15) CALCIUM 2.3 mmol/L (2.3-3.0) PHOSPHATE 1.6 mmol/L (0.8-2.0) CA:P RATIO 1.4 PROTEIN, TOTAL 69 g/L (58-83) ALBUMIN 39 g/L (34-50) GLOBULIN 30 g/L (18-39) BILIRUBIN, TOTAL 2 umol/L (< 9) ALP 400 H IU/L (25-219) AST 24 IU/L (< 135) ALT 20 IU/L (< 237) CK 165 IU/L CHOLESTEROL 2.3 mmol/L (1.1-3.1) GAMMA GT 6 IU/L (< 17) SAMPLE APPEARANCE Normal

Minor changes of doubtful significance.

- 250 -

References

Achurch, H.M. 1989. Tree utilisation by koala (Phascolarctos cinereus) populations with regard to their geographic distribution within the close proximity to the City of Brisbane. Honours Thesis. University of Queensland, Brisbane. Ainsworth, L.M., and C.B. Dean 2011. Zero inflated spatial models. Environmental and Ecological Statistics in press. Akaike, H. 1973. Information theory as an extension of the maximum likelihood principle. Pages 267-281 in B.N. Petrov and F. Csaki, editors. Second International Symposium on Information Theory. Akademiai Kiado, Budapest. Alcoa 2007. Alcoa world alumina Australia darling range bauxite mine rehabilitation completion criteria. Departement of Industry and Resources Western Australia. Available from http://www.dsd.wa.gov.au/documents/Bauxite_Mine_Rehabilitation(1).pdf (accessed 10/04/2011). Allen, L., R. Engeman, and H. Krupa 1996. Evaluation of three relative abundance indices for assessing dingo populations. Wildlife Research 23: 197-206. Altman, D.G., and J.M. Bland 2005. Standard deviations and standard errors. British Medical Journal 331: 903. Andelman, S.J., and W.F. Fagan 2000. Umbrellas and flagships: Efficient conservation surrogates or expensive mistakes? Proceedings of the National Academy of Sciences of the United States of America 97: 5954-5959. Andersen, A.N. 1992. The use of ants to assess restoration success following mining: A case study at Ranger uranium mine. Pages 232-238 in I. Moffatt and A. Webb, editors. Conservation and development issues in northern Australia. North Australian Research Unit, Darwin. Andersen, A.N. 1993. Ants as indicators of restoration success at a uranium mine in tropical Australia. Restoration Ecology 1: 156-167. Andersen, A.N. 1994. Ants as indicators of restoration success following mining in northern Australia. Pages 465-467 Proceedings of the AusIMM Annual Conference 1994, Australian Mining Looks North - ‘the Challenges and Choices’, Darwin, Australia. Australian Institute of Mining and Metallurgy, Darwin.

- 251 -

Andersen, A.N. 1997. Ants as indicators of restoration success following mining: A functional group approach. Pages 319-325 in P. Hale and D. Lamb, editors. Conservation Outside Nature Reserves Centre for Conservation Biology, University of Queensland, Brisbane. Andersen, A.N., and G.P. Sparling 1997. Ants as indicators of restoration success: Relationship with soil microbial biomass in the Australian seasonal tropics. Restoration Ecology 5: 109-114. Andersen, A.N., S. Morrison, and L. Belbin 1998. The role of ants in minesite restoration in the Kakadu region of Australia's Northern Territory with particulcar reference to their use as bioindicators. Supervising Scientist Report 130. CSIRO, Canberra, Australia. Andersen, A.N., J.A. Ludwig, L.M. Lowe, and D.C.F. Rentz 2001. Grasshopper biodiversity and bioindicators in Australian tropical savannas: Responses to disturbance in Kakadu National Park. Austral Ecology 26: 213-222. Andersen, A.N., B.D. Hoffmann, W.J. Müller, and A.D. Griffiths 2002. Using ants as bioindicators in land management: Simplifying assessment of ant community responses. Journal of Applied Ecology 39: 8-17. Andersen, A.N. 2003. Ants as indicators of minesite restoration, community recovery at one of eight rehabilitation sites in central Queensland. Ecological Management and Restoration 4: 12-19. Andersen, A.N., A. Fisher, B.D. Hoffmann, J.L. Read, and R. Richards 2004. Use of terrestrial invertebrates for biodiversity monitoring in Australian rangelands, with particular reference to ants. Austral Ecology 29: 87-92. Anderson, D.R. 2001. Commentary on models in ecology. Bulletin of the Ecological Society of America 82: 160-161. Anderson, D.R., K.P. Burnham, W.R. Gould, and S. Cherry 2001a. Concerns about finding effects that are actually spurious. Wildlife Society Bulletin 29: 311-316. Anderson, D.R., W.A. Link, D.H. Johnson, and K.P. Burnham 2001b. Suggestions for presenting the results of data analyses. The Journal of Wildlife Management 65: 373- 378. Andres, P., and E. Mateos 2006. Soil mesofaunal responses to post-mining restoration treatments. Applied Soil Ecology 33: 67-78.

- 252 -

Anonymous 2008. Australia’s sustainable forest management framework of criteria and indicators 2008 - policy guidelines. Commonwealth of Australia, Canberra. Available from http://adl.brs.gov.au/forestsaustralia/_pubs/ciframework.pdf (accessed 20/02/2011). Armstrong, K.N., and O.G. Nichols 2000. Long-term trends in avifaunal recolonisation of rehabilitated bauxite mines in the jarrah forest of south-western Australia. Forest Ecology and Management 126: 213-225. Aronson, J., C. Floret, E. Le Floc'h, C. Ovalle, and R. Pontanier 1993. Restoration and rehabilitation of degraded ecosystems in arid and semiarid regions. I. A view from the South. Restoration Ecology 1: 8-17. Aronson, J., and E.L. Floc'h 1996. Vital landscape attributes: Missing tools for restoration ecology. Restoration Ecology 4: 377-387. Australian Bureau of Statistics 2009. 5676.0 Business indicators Australia, March quarter 2009. Available from http://www.ausstats.abs.gov.au/ausstats/meisubs.nsf/0/94029704A60067A5CA2575 C500147097/$File/56760_mar%202009.pdf (accessed 20 May 2009). Azeria, E.T., D. Fortin, C. Hebert, P. Peres-Neto, D. Pothier, and J.-C. Ruel 2009. Using null model analysis of species co-occurrences to deconstruct biodiversity patterns and select indicator species. Diversity and Distributions 15: 958-971. Babyak, M.A. 2004. What you see may not be what you get: A brief, nontechnical introduction to overfitting in regression-type models. Psychosomatic Medicine 66: 411-421. Ball, T., E. Adams, and R.L. Goldingay 2009. Diet of the squirrel glider in a fragmented landscape near Mackay, central Queensland Australian. Journal of Zoology 57: 295- 304. Balvanera, P., A.B. Pfisterer, N. Buchmann, J.-S. He, T. Nakashizuka, D. Raffaelli, and B. Schmid 2006. Quantifying the evidence for biodiversity effects on ecosystem functioning and services. Ecology Letters 9: 1146-1156. Barnes, R.F.W. 1996. Estimating forest elephant abundance by dung counts. Pages 28- 37 in K. Kangwana, editor. Studying Elephants. African Wildlife Foundation, Nairobi, Kenya.

- 253 -

Barnes, R.F.W., B. Asamoah-Boateng, J. Naada-Majam, and J. Agyei-Ohemeng 1997. Rainfall and the population dynamics of elephant dung-piles in the forests of southern Ghana. African Journal of Ecology 35: 39-52. Battin, J. 2004. When good animals love bad habitats: Ecological traps and the conservation of animal populations. Conservation Biology 18: 1482-1491. Bell, D., D. Carter, and R. Hetherington 1986. Experimental assessment of wind erosion after soil stabilization treatments at Eneabba, Western Australia. Environmental Geochemistry and Health 8: 99-104. Bell, L.C. 2001. Establishment of native ecosystems after mining - Australian experience across diverse biogeographic zones. Ecological Engineering 17: 179-186. Bell, S.S., M.S. Fonseca, and L.B. Motten 1997. Linking restoration and landscape ecology. Restoration Ecology 5: 318-323. Bellairs, S.M. 1999. Determining ecological indicators for native vegetation and wildlife habitat rehabilitation success at the Blair Athol and Tarong mines. Pages 105-118 in C.J. Asher and L.C. Bell, editors. Proceedings of the Workshop on Indicators of Ecosystem Rehabilitation Success, Melbourne. Australian Centre for Mining Environmental Research, Brisbane. Bellamy, S.L., Y. Li, L.M. Ryan, S. Lipsitz, M.J. Canner, and R. Wright 2004. Analysis of clustered and interval censored data from a community-based study in asthma. Statistics in Medicine 23: 3607-3621. Bellemare, J., G. Motzkin, and D.R. Foster 2002. Legacies of the agricultural past in the forested present: An assessment of historical land-use effects on rich mesic forests. Journal of Biogeography 29: 1401-1420. BEMAX 2009. Status of completion criteria at the Ludlow mine site, February 2009. Available from http://www.bemax.com.au/EnviroDocs/Ludlow%20Rehab%20Progress%20Report% 202008.pdf (accessed 06/04/2011). Bennett, A.F. 1990. Land use, forest fragmentation and the mammalian fauna at Naringal, south-western Victoria. Australian Wildlife Research 17: 325-347. Bennett, A.F. 2000. Future directions in revegetation: Will biodiversity issues receive greater consideration? An Occasional Publication of the Australian Research Centre For Urban Ecology 2: 101-107.

- 254 -

Bennett, L.J., P.F. English, and R. McCain 1940. A Study of deer populations by use of pellet-group counts. The Journal of Wildlife Management 4: 398-403. Beshkarev, A.B., J.E. Swenson, P. Angelstam, H. Andren, and A.B. Blagovidov 1994. Long-term dynamics of hazel grouse populations in source- and sink- dominated pristine taiga landscapes. Oikos 71: 375-380. Betrus, C.J., E. Fleishman, and R.B. Blair 2005. Cross-taxonomic potential and spatial transferability of an umbrella species index. Journal of Environmental Management 74: 79-87. Beyer, G.L., and R.L. Goldingay 2006. The value of nest boxes in the research and management of Australian hollow-using arboreal marsupials. Wildlife Research 33: 161-174. Beyer, G.L., R.L. Goldingay, and D.J. Sharpe 2008. The characteristics of squirrel glider (Petaurus norfolcensis) den trees in subtropical Australia. Australian Journal of Zoology 56: 13-21. BHP Billiton 2006. BHP Billiton iron ore Jimblebar - Wheelarra Hill mine: Decommissioning and final rehabilitation plan. Available from http://www.bhpbilliton.com/bbContentRepository/docs/wheelarraHillAppendixB.pdf (accessed 06/04/2011). Bisevac, L., and J. Majer 1999a. Comparative study of ant communities of rehabilitated mineral sand mines and heathland, Western Australia. Restoration Ecology 7: 117- 126. Bisevac, L., and J.D. Majer 1999b. Invertebrates as success indicators for mine site rehabilitation. Pages 47-52 in C.J. Asher and L.C. Bell, editors. Proceedings of the Workshop on Indicators of Ecosystem Rehabilitation Success, Melbourne. Australian Centre for Mining Environmental Research, Brisbane. Bjornstad, O.N. 2009. Spatial nonparametric covariance functions. Version 1.1-3. Block, W.M., A.B. Franklin, J. James P. Ward, J.L. Ganey, and G.C. White 2001. Design and implementation of monitoring studies to evaluate the success of ecological restoration on wildlife. Restoration Ecology 9: 293-303. Bowen-Jones, E., and A. Entwistle 2002. Identifying appropriate flagship species: The importance of culture and local contexts. Oryx 36: 189-195. Boyer, S., and S.D. Wratten 2010. The potential of earthworms to restore ecosystem services after opencast mining – a review. Basic and Applied Ecology 11: 196-203.

- 255 -

Bradshaw, A.D. 1983. The reconstruction of ecosystems: Presidential address to the British Ecological Society, December 1982. Journal of Applied Ecology 20: 1-17. Bradshaw, A.D. 1996. Underlying principles of restoration. Canadian Journal of Fisheries and Aquatic Sciences 53: 3-9. Brady, C.J., and R.A. Noske 2009. Succession in bird and plant communities over a 24- year chronosequence of mine rehabilitation in the Australian monsoon tropics. Restoration Ecology: doi: 10.1111/j.1526-1100X.2008.00511.x. Bragg, J.G., J.E. Taylor, and B.J. Fox 2005. Distributions of lizard species across edges delimiting open-forest and sand-mined areas. Austral Ecology 30: 188-200. Braithwaite, L., J. Turner, and J. Kelly 1984. Studies on the arboreal marsupial fauna of eucalypt forests being harvested for wood pulp at Eden, N.S.W. III. Relationships between faunal densities, eucalypt occurrence and foliage nutrients, and soil parent materials. Australian Wildlife Research 11: 41-48. Braithwaite, L.W., L. Belbin, J. lve, and M. Austin 1993. Land use allocation and biological conservation in the Batesman’s Bay forests of New South Wales. Australian Forestry 56: 4-21. Bray, J.R., and J.T. Curtis 1957. An ordination of the upland forest communities of southern Wisconsin. Ecological Monographs 27: 325-349. Brearley, G., A. Bradley, S. Bell, and C. McAlpine 2010. Influence of contrasting urban edges on the abundance of arboreal mammals: A study of squirrel gliders (Petaurus norfolcensis) in southeast Queensland, Australia. Biological Conservation 143: 60- 71. Brennan, K., J. Majer, and O. Nichols 2005. Innovative techniques for promoting fauna return to rehabilitated sites following mining. Australian Centre for Minerals Extension and Research (ACMER), Brisbane, and Minerals and Energy Research Institute of Western Australia (MERIWA report 258), Perth. Brennan, K.E., J.D. Majer, and J.M. Koch 2003. Using fire to facilitate faunal colonization following mining: An assessment using spiders in Western Australian jarrah forest. Ecological Management and Restoration 2: 145-147. Brodie, J.F. 2006. An experimentally determined persistence-rate correction factor for scat-based abundance indices. Wildlife Society Bulletin 34: 1216-1219. Brooks, T., G.A.B. Da Fonseca, and A.S.L. Rodrigues 2004. Species, data, and conservation planning. Conservation Biology 18: 1682-1688.

- 256 -

Broström, G. 2003. Generalized linear models with random intercepts. Available from http://cran.r-project.org/web/package/glmmML/index.html (accessed 3 November). Bryan, B.A. 1997. A generic method for identifying regional koala habitat using GIS. Australian Geographical Studies 35: 125-139. Buffington, J.M., J.C. Kilgo, R.A. Sargent, K.V. Miller, and B.R. Chapman 2000. Effects of restoration techniques on breeding birds in a thermally-impacted bottomland hardwood forest. Ecological Engineering 15: S115-S120. Buij, R., et al. 2007. Patch-occupancy models indicate human activity as major determinant of forest elephant Loxodonta cyclotis seasonal distribution in an industrial corridor in Gabon. Biological Conservation 135: 189-201. Bureau of Meteorology 2010. Climate data online. Commonwealth Australia. Available from http://www.bom.gov.au/climate/data/ (accessed 25 September 2010). Burgin, S., and D. Wotherspoon 2009. The potential for golf courses to support restoration of biodiversity for BioBanking offsets. Urban Ecosystems 12: 145-155. Burnham, K.P., and D.R. Anderson 2002. Model selection and multimodel inference. Second edition. Springer-Verlag, New York. Cairns, J. 1988. Increasing diversity by restoring damaged ecosystems. Pages 333-343 in E.O. Wilson and F.M. Peter, editors. Biodiversity. National Academy Press, Washington, DC. Cairns, J., and J.R. Heckman 1996. Restoration Ecology: The state of an emerging field. Annual review of energy and the environment 21: 167-189. Cale, P.G., and R.J. Hobbs 1994. Landscape heterogeneity indices: Problems of scale and applicability, with particular reference to animal habitat description. Pacific Conservation Biology 1: 183-193. Carbone, C., et al. 2001. The use of photographic rates to estimate densities of tigers and other cryptic mammals. Animal Conservation 4: 75-79. Caro, T., A. Engilis, E. Fitzherbert, and T. Gardner 2004. Preliminary assessment of the flagship species concept at a small scale. Animal Conservation 7: 63-70. Caro, T.M., and G. O'Doherty 1999. On the use of surrogate species in conservation biology. Conservation Biology 13: 805-814. Caro, T.M. 2003. Umbrella species: Critique and lessons from East Africa. Animal Conservation 6: 171-181.

- 257 -

Caughley, G. 1994. Directions in conservation biology. The Journal of Animal Ecology 63: 215-244. Choi, Y.D. 2004. Theories for ecological restoration in changing environment: Toward ‘futuristic’ restoration. Ecological Research 19: 75-81. Choi, Y.D. 2007. Restoration ecology to the future: A call for new paradigm. Restoration Ecology 15: 351-353. Choi, Y.D., V.M. Temperton, E.B. Allen, A.P. Grootjans, M. Halassy, R.J. Hobbs, M.A. Naeth, and K. Torok 2008. Ecological restoration for future sustainability in a changing environment. Ecoscience 15: 53-64. Clark, A.L., and J. Cook-Clark 2005. An international overview of legal frameworks for mine closure. Available from www.idrc.ca/en/ev-84059-201-1-DO_TOPIC.html (accessed 2/12/2010). Clarke, K.R. 1993. Non-parametric multivariate analysis of changes in community structure. Australian Journal of Ecology 18: 117-143. Clarke, K.R., and R.N. Gorley 2006. PRIMER v6: user manual/tutorial. PRIMER-E Ltd, Plymouth, United Kingdom. Clements, A. 1998. Rebuilding fauna habitats. Pages 57-64 in C.J. Asher and L.C. Bell, editors. Proceedings of Fauna Habitat Reconstruction after Mining Workshop, Adelaide. Australian Centre for Mining Environmental Research, Brisbane. Clewell, A., and J.P. Rieger 1997. What practitioners need from restoration ecologists. Restoration Ecology 5: 350-354. Clifton, I., W. Ellis, A. Melzer, and G. Tucker 2007. Water turnover and the northern range of the koala (Phascolarctos cinereus). Australian Mammalogy 29: 85-88. Clucas, B., K. McHugh, and T. Caro 2008. Flagship species on covers of US conservation and nature magazines. Biodiversity and Conservation 17: 1517-1528. Coaldrake, J.E. 1979. Coastal sandmining and landscape rehabilitation in Eastern Australia. Landscape Planning, Special Issue Australian Landscapes: a Melting Pot of Aboriginal and European Influences 6: 359-374. Cochran, G.A., and H.J. Stains 1961. Deposition and decomposition of fecal pellets by cottontails. The Journal of Wildlife Management 25: 432-435. Coley, P.D., J.P. Bryant, and F.S. Chapin 1985. Resource availability and plant antiherbivore defense. Science 230: 895-899.

- 258 -

Collett, K. 2003. Recolonisation of rehabilitated bauxite mine pits by the scorpion Urodacus planimanus. Australasian Arachnology 68: 12. Collins, B.G., B.J. Wykes, and O.G. Nichols 1985. Recolonization of restored bauxite minelands by birds in Southwestern Australia. Pages 341-354 in A. Keast, H.F. Recher, H. Ford and D. Saunders, editors. Birds of eucalypt forests and woodlands: Ecology, conservation and management. Surrey Beatty and Sons, Sydney. Common, I.F.B., and M. Horak 1994. Four new species of Telanepsia Turner (Lepidoptera: Oecophoridae) with larvae feeding on koala and possum scats. Invertebrate Taxonomy 8: 809-828. Corbett, L. 1997. Environmental monitoring at Ranger mine: A whole ecosystem approach. Pages 240-254. 22nd Annual Environmental Workshop Proceedings, Adelaide. Minerals Council of Australia, Dickson, ACT. Corbett, M. 1999. Revegetation of mined land in the wet-dry tropics of northern Australia: A review. Report to the Alligator Rivers Region Technical Committee. Supervising Scientist, Darwin. Cork, S., and L. Braithwaite 1996. Resource availability, eucalypt chemical defences, and habitat quality for leaf-eating marsupials. Pages 9-16 in G. Gordon, editor. Koalas - Research for management, Brisbane, Australia. World Koala Research. Cork, S., E. Cawsey, A. Nicholls, M. Yialeloglou, and M. Austin 1997. Improving prediction and assessment of diet and habitat quality for koalas. Report to the AKF. Brisbane, Australia. Cork, S.J., and G.D. Sanson 1990. Digestion and nutrition in the koala: A review. Pages 129-144 in A. Lee, K. Handasyde and G. Sanson, editors. Biology of the Koala. Surrey Beatty and Sons, Sydney, Australia. Cork, S.J., I.D. Hume, and W.J. Foley 2000. Improving habitat models and their utility in koala conservation. Conservation Biology 14: 660-668. Cottenie, K., and L. De Meester 2003. Comment to Oksanen (2001): Reconciling Oksanen (2001) and Hurlbert (1984). Oikos 100: 394-396. Craig, M.D., M.J. Garkaklis, G.E.S.J. Hardy, A.H. Grigg, C.D. Grant, P.A. Fleming, and R.J. Hobbs 2007. Ecology of the western bearded dragon (Pogona minor) in unmined forest and forest restored after bauxite mining in south-west Western Australia. Australian Journal of Zoology 55: 107-116.

- 259 -

Craig, M.D., R.J. Hobbs, A.H. Grigg, M.J. Garkaklis, C.D. Grant, P.A. Fleming, and G.E.S.J. Hardy 2010. Do thinning and burning sites revegetated after bauxite mining improve habitat for terrestrial vertebrates? Restoration Ecology 18: 300-310. Crisp, P.N., K.J.M. Dickinson, and G.W. Gibbs 1998. Does native invertebrate diversity reflect native plant diversity? A case study from New Zealand and implications for conservation. Biological Conservation 83: 209-220. CRL 2006. Rehabilitation outcomes report - Proposed rehabilitation acceptance criteria. unpublished report. CRL 2007. Environmental Authority No. MIM800088202 (mining activities). unpublished report. Crowther, M.S., C.A. McAlpine, D. Lunney, I. Shannon, and J.V. Bryant 2009. Using broad-scale, community survey data to compare species conservation strategies across regions: A case study of the Koala in a set of adjacent ‘catchments’. Ecological Management & Restoration 10: S88-S96. Cuccovia, A., and A. Kinnear 1999. Acarine (mite) communities colonizing rehabilitated bauxite mine pits in the jarrah forest of Western Australia. Pages 54-59 in W. Ponder and D. Lunney, editors. The other 99%: The conservation of biodiversity of invertebrates. Royal Zoological Society of New South Wales, Mosman, Australia. Cunningham, R.B., and D.B. Lindenmayer 2005. Modeling count data of rare species: Some statistical issues. Ecology 86: 1135-1142. Cunningham, R.B., D.B. Lindenmayer, M. Crane, D. Michael, and C. McGregor 2007. Reptile and arboreal marsupial response to replanted vegetation in agricultural landscapes. Ecological Applications 17: 609-619. Cunningham, R.B., D. Lindenmayer, C. M., M. D., M. C., R. Montague-Drake, and J. Fischer 2008. The combined effects of remnant vegetation and tree planting on farmland birds. Conservation Biology 22: 742-752. Curran, L.M., S.N. Trigg, A.K. McDonald, D. Astiani, Y.M. Hardiono, P. Siregar, I. Caniago, and E. Kasischke 2004. Lowland forest Loss in protected aeas of Indonesian Borneo. Science 303: 1000-1003. Curry, P.J., and O.G. Nichols 1986. Early regrowth in rehabilitated bauxite minesites as breeding habitat for birds in the jarrah forest of south-western Australia. Australian Forestry 49: 112-114.

- 260 -

Curtin, A., D. Lunney, and A. Matthews 2002. A survey of a low-density koala population in a major reserve system, near Sydney, New South Wales. Australian Mammalogy 23: 135-144. Cushman, S.A., K.S. McKelvey, B.R. Noon, and K. McGarigal 2010. Use of abundance of one species as a surrogate for abundance of others. Conservation Biology 24: 830- 840. Czech, B., P.R. Krausman, and P.K. Devers 2000. Economic associations among causes of species endangerment in the United States. BioScience 50: 593-601. Dale, V.H., S. Brown, R.A. Haeuber, N.T. Hobbs, N. Huntly, R.J. Naiman, W.E. Riebsame, M.G. Turner, and T.J. Valone 2000. Ecological principles and guidelines for managing the use of land. Ecological Applications 10: 639-670. Davis, A.L.V., R.J. Van Aarde, C.H. Scholtz, and J.H. Delport 2003. Convergence between dung beetle assemblages of a post-mining vegetational chronosequence and unmined dune forest. Restoration Ecology 11: 29-42. Davis, M.A. 2000. Letters. "Restoration" - a misnomer? Science 287: 1203. Davis, M.A., and L.B. Slobodkin 2004. The science and values of restoration ecology. Restoration Ecology 12: 1-3. DeFries, R., A. Hansen, A.C. Newton, and M.C. Hansen 2005. Increasing isolation of protected areas in tropical forests over the past 20 years. Ecological Applications 15: 19-26. Degabriele, R. 1981. A relative shortage of nitrogenous food in the ecology of the koala (Phascolarctos cinereus). Austral Ecology 6: 139-141. Delibes, M., P. Gaona, and P. Ferreras 2001. Effects of an attractive sink leading into maladaptive habitat selection. The American Naturalist 158: 277-285. Delibes, M., S. Cabezas, B. Jiménez, and M.J. González 2009. Animal decisions and conservation: The recolonization of a severely polluted river by the Eurasian otter. Animal Conservation 12: 400-407. Diamond, J.M., N.P. Ashmole, and P.E. Purves 1989. The present, past and future of human-caused extinctions [and discussion]. Philosophical Transactions of the Royal Society of London. B, Biological Sciences 325: 469-477. Dias, P.C. 1996. Sources and sinks in population biology. Trends in Ecology & Evolution 2: 326-330.

- 261 -

Dobson, A.P., A.D. Bradshaw, and A.J.M. Baker 1997. Hopes for the future: Restoration ecology and conservation biology. Science 277: 515-522. Dobson , M., R. Goldingay , and D. Sharpe 2005. Feeding behaviour of the squirrel glider in remnant habitat in Brisbane. Australian Mammalogy 27: 27-35. Donato, D.B., O. Nichols, H. Possingham, M. Moore, P.F. Ricci, and B.N. Noller 2007. A critical review of the effects of gold cyanide-bearing tailings solutions on wildlife. Environment International 33: 974-984. Doyle, F., and G. Kaeding 1998. Bird colonization of constructed wetlands at Capel, WA. Pages 145-150 in C.J. Asher and L.C. Bell, editors. Proceedings of Fauna Habitat Reconstruction after Mining Workshop, Adelaide. Australian Centre for Mining Environmental Research, Brisbane. Dunger, W., et al. 2001. Development of soil fauna at mine sites during 46 years after afforestation. Pedobiologia 45: 243-271. Dunger, W., H.-J. Schulz, B. Zimdars, and K. Hohberg 2004. Changes in collembolan species composition in Eastern German mine sites over fifty years of primary succession. Pedobiologia 48: 503-517. Dunlop, J.N., J.D. Majer, C.J. Morris, and K.J. Walker 1985. A preliminary assessment of minesite rehabilitation in the pilbara iron ore province using ant communities as ecological indicators. Mulga Research Centre Journal 8: 25-31. Dwernychuk, L.W., and D.A. Boag 1972. Ducks nesting in association with gulls - an ecological trap? Canadian Journal of Zoology 50: 559-563. Dzieciolowski, R. 1976. Roe deer census by pellet-group counts. Acta Theriologica 21: 351-358. Eaton, B.R. 1993. Estimates of snowshoe hare abundance from pellet plot counts: A critical evaluation. Thesis. Acadia University, Wolfville, Nova Scotia, Canada. Ehrenfeld, J.G., and L.A. Toth 1997. Restoration ecology and the ecosystem perspective. Restoration Ecology 5: 307-317. Ehrenfeld, J.G. 2000. Defining the limits of restoration: The need for realistic goals. Restoration Ecology 8: 2-9. Ellis, W., and F.N. Carrick 1992. Total body water and the estimation of fat in the koala (Phascolarctos cinereus). Australian Veterinary Journal 69: 229-231.

- 262 -

Ellis, W., A. Melzer, B. Green, K. Newgrain, M.A. Hindell, and F.N. Carrick 1995. Seasonal-variation in water flux, field metabolic-rate and food-consumption of free- ranging koalas (Phascolarctos cinereus). Australian Journal of Zoology 43: 59-68. Ellis, W., F.N. Carrick, P. Lundgren, A. Veary, and B. Cohen 1999. The use of faecal cuticle examination to determine the dietary composition of koalas. Australian Zoologist 31: 127 - 133 Ellis, W. 2003. Integration of natural resources management under Queensland mining legislation. Environmental and Planning Law Journal 20: 66-77. Ellis, W., A. Melzer, and F. Bercovitch 2009. Spatiotemporal dynamics of habitat use by koalas: The checkerboard model. Behavioral Ecology and Sociobiology 63: 1181- 1188. Ellis, W.A.H., B.J. Sullivan, A.T. Lisle, and F.N. Carrick 1998. The spatial and temporal distribution of koala faecal pellets. Wildlife Research 25: 663-668. EPA 2006. Guidance for the assessment of environmental factors Western Australia (in accordance with the Environmental Protection Act 1986). 6: 1-53. Environmental Protection Authority, Perth. Eschler, B.M., D.M. Pass, R. Willis, and W.J. Foley 2000. Distribution of foliar formylated phloroglucinol derivatives amongst Eucalyptus species. Biochem Syst Ecol 28: 813-824. Favreau, J.M., C.A. Drew, G.R. Hess, M.J. Rubino, F.H. Koch, and K.A. Eschelbach 2006. Recommendations for assessing the effectiveness of surrogate species approaches. Biodiversity and Conservation 15: 3949-3969. Fischer, J., D.B. Lindenmayer, and A.D. Manning 2006. Biodiversity, ecosystem function, and resilience: Ten guiding principles for commodity production landscapes. Frontiers in Ecology and the Environment 4: 80-86. Fischer, J., D.B. Lindenmayer, S.P. Blomberg, R. Montague-Drake, A. Felton, and J.A. Stein 2007a. Functional richness and relative resilience of bird communities in regions with different land use intensities. Ecosystems 10: 964-974. Fischer, J., et al. 2007b. Mind the sustainability gap. Trends in Ecology & Evolution 22: 621-624. Flack, V.F., and P.C. Chang 1987. Frequency of selecting noise variables in subset regression analysis: A simulation study. The American Statistician 41: 84-86.

- 263 -

Fleishman, E., R.B. Blair, and D.D. Murphy 2001a. Empirical validation of a method for umbrella species selection. Ecological Applications 11: 1489-1501. Fleishman, E., D.D. Murphy, and R.B. Blair 2001b. Selecting effective umbrella species. Conservation in Practice 2: 17-23. Fleishman, E., and D.D. Murphy 2009. A realistic assessment of the indicator potential of butterflies and other charismatic taxonomic groups. Conservation Biology 23: 1109-1116. Fletcher, D. 1987. Ant succession in Pilbara borrow pits. Mulga Research Centre Journal 10: 1-8. Foster, D.R., D.H. Knight, and J.F. Franklin 1998. Landscape patterns and legacies resulting from large, infrequent forest disturbances. Ecosystems 1: 497-510. Fox, B.J., and M.D. Fox 1978. Recolonization of coastal heath by Pseudomys novaehollandiae (Muridae) following sand mining. Australian Journal of Ecology 3: 447-465. Fox, B.J. 1982. The influence of disturbance (fire, mining) on ant and small mammal species diversity in Australian heathland. Pages 213-219 Proceedings of the symposium on dynamics and management of Mediterranean type ecosystems USDA Forest Service, General Technical Report, PSW-58, Riverside California. Fox, B.J., and M.D. Fox 1984. Small-mammal recolonization of open-forest following sand mining. Austral Ecology 9: 241-252. Fox, B.J. 1990. Two hundred years of disturbance: How has it aided our understanding of succession in Australia? Proceedings of the Ecological Society of Australia 16: 521-529 Fox, B.J., and L.E. Twigg 1991. Experimental transplants of mice (Pseudomys and Mus) on to early stages of post-mining regeneration in open forest. Australian Journal of Ecology 16: 281-287. Fox, B.J. 1996. Long term studies of small mammal communities in disturbed habitats of eastern Australia. Pages 467-501 in M.L. Cody and J.A. Smallwood, editors. Long-term studies of vertebrate communities. Academic Press, Orlando, Florida. Fox, B.J. 1998. Fauna habitat reconstruction after mining. Pages 151-160 in C.J. Asher and L.C. Bell, editors. Proceedings of Fauna Habitat Reconstruction after Mining Workshop, Adelaide. Australian Centre for Mining Environmental Research, Brisbane.

- 264 -

Fox, M.D., and B.J. Fox 1982. Evidence for interspecific competition influencing ant species diversity in a regenerating heathland. Pages 99-110 in R. Buckley, editor. Ant plant interactions in Australia. W. Junk, The Hague. Frazier, J. 2005. Flagging the flagship: Valuing experiences from ancient depths. Journal Maritime Studies 4: 273-303. Freeland, W.J., and D.H. Janzen 1974. Strategies in herbivory by mammals: The role of plant secondary compounds. The American Naturalist 108: 269-289. Fretwell, S.D., and H.L.J. Lucas 1970. On territorial behavior and other factors influencing habitat distribution in birds. Acta Biotheoretica 19: 16-36. Fretwell, S.D. 1972. Populations in Seasonal Environments. Princeton University Press, Princeton, NJ. Frouz, J., D. Elhottova, V. Kuraz, and M. Sourkova 2006. Effects of soil macrofauna on other soil biota and soil formation in reclaimed and unreclaimed post mining sites: Results of a field microcosm experiment. Applied Soil Ecology 33: 308-320. Ganihar, S.R. 2003. Nutrient mineralization and leaf litter preference by the earthworm Pontoscolex corethrurus on iron ore mine wastes. Restoration Ecology 11: 475-482. Gates, J.E., and L.W. Gysel 1978. Avian nest dispersion and fledging success in field- forest ecotones. Ecology 59: 871-883. Geoscience Australia 2009. Operating Mines. Commonwealth of Australia. Available from http://www.australianminesatlas.gov.au/mapping/downloads.jsp (accessed 20 may 2010). Gibbons, P., and D. Lindenmayer 2002. Tree hollows and wildlife conservation in Australia. CSIRO Publishing, Collingwood. Gibbs, J.P., H.L. Snell, and C.E. Causton 1999. Effective monitoring for adaptive wildlife management: Lessons from the Galápagos Islands. The Journal of Wildlife Management 63: 1055-1065. Goertz, J.W. 1964. The influence of habitat quality upon density of cotton rat populations. Ecological Monographs 34: 359-381. Goethals, K., B. Ampe, D. Berkvens, H. Laevens, P. Janssen, and L. Duchateau 2009. Modeling interval-censored, custered cow udder quarter infection times through the shared gamma frailty model. Journal of Agricultural, Biological, and Environmental Statistics 14: 1-14.

- 265 -

Goldingay, R.L. 1990. The foraging behaviour of a nectar feeding marsupial, Petaurus australis. Oecologia 85: 191-199. Goldingay, R.L., M.J. Grimson, and G.C. Smith 2007. Do feathertail gliders show a preference for nest box design? Wildlife Research 34: 484-490. Goldstein, P.Z. 1999. Functional ecosystems and biodiversity buzzwords. Conservation Biology 13: 247-255. Golet, G.H., T. Gardali, J.W. Hunt, D.A. Koenig, and N.M. Williams 2011. Temporal and taxonomic variability in response of fauna to riparian restoration. Restoration Ecology 19: 126-135. Gomez, G., M.L. Calle, R. Oller, and K. Langohr 2009. Tutorial on methods for interval-censored data and their implementation in R. Statistical Modelling 9: 259- 297. Gordon, G., A.S. Brown, and T. Pulsford 1988. A koala (Phascolarctos cinereus Goldfuss) population crash during drought and heatwave conditions in south-western Queensland. Austral Ecology 13: 451-461. Gordon, G., D.G. McGreevy, and B.C. Lawrie 1990. Koala populations in Queensland: major limiting factors. Pages 85-95 in A.K. Lee, K.A. Handasyde and G.D. Sanson, editors. Biology of the Koala. Surrey Beatry & Sons, Sydney. Gordon, G. 1991. Estimation of the age of the koala, Phascolarctos cinereus (Marsupialia: Phascolarctidae) from tooth wear and growth. Australian Mammalogy 14: 5-12. Graham, M.H., and R.J. Haynes 2004. Organic matter status and the size, activity and metabolic diversity of the soil microflora as indicators of the success of rehabilitation of mined sand dunes. Biology and Fertility of Soils 39: 429-437. Grant, C., and J. Koch 2007. Decommissioning Western Australia’s first bauxite mine: Co-evolving vegetation restoration techniques and targets. Ecological Management & Restoration 8: 92-105. Greenslade, P., and J.D. Majer 1980. Collembola of rehabilitated minesites of Western Australia. Pages 397-408 in D.L. Dindal, editor. Soil Biology as Related to Land Use Practices. Environmental Protection Agency, Washington. Greenslade, P., and J.D. Majer 1993. Recolonization by Collembola of rehabilitated bauxite mines in Western Australia. Austral Ecology 18: 385-394.

- 266 -

Greenslade, P. 2007. The potential of Collembola to act as indicators of landscape stress in Australia. Australian Journal of Experimental Agriculture 47: 424-434. Groot, R.S.d., M.A. Wilson, and R.M.J. Boumans 2002. A typology for the classification, description and valuation of ecosystem functions, goods and services. Ecological Economics 41: 393-408. Gu, W., and R.K. Swihart 2004. Absent or undetected? Effects of non-detection of species occurrence on wildlife-habitat models. Biological Conservation 116: 195- 203. Haberl, H., K.H. Erb, F. Krausmann, V. Gaube, A. Bondeau, C. Plutzar, S. Gingrich, W. Lucht, and M. Fischer-Kowalski 2007. Quantifying and mapping the human appropriation of net primary production in earth's terrestrial ecosystems. Proceedings of the National Academy of Sciences 104: 12942-12947. Halle, S. 2007. Science, art, or application - the ‘‘karma’’ of restoration ecology. Restoration Ecology 15: 358-361. Halme, P., M. Monkkonen, J.S. Kotiaho, A.-L. Ylisirnio, and A. Markkanen 2009. Quantifying the indicator power of an indicator species. Conservation Biology 23: 1008-1016. Harestad, A.S., and F.L. Bunnell 1987. Persistence of black-tailed deer fecal pellets in coastal habitats. The Journal of Wildlife Management 51: 33-37. Harries, J. 1997. Acid mine drainage in Australia: Its extent and potential future liability. Supervising Scientist Report 125, Canberra, Australia. Hasegawa, M. 1995. Habitat utilisation by koalas (Phascolarctos cinereus) at Point Halloran, Queensland. Master thesis. The University of Queensland, Brisbane. Hawlena, D., and A. Bouskila 2006. Land management practices for combating desertification cause species replacement of desert lizards. Journal of Applied Ecology 43: 701-709. Hawlena, D., D. Saltz, Z. Abramsky, and A. Bouskila 2010. Ecological trap for desert lizards caused by anthropogenic changes in habitat structure that favor predator activity. Conservation Biology 24: 803-809. Hendrychová, M. 2008. Reclamation success in post-mining landscapes in the Czech Republic: A review of pedological and biological studies. Journal of Landscape Studies 1: 63 - 78.

- 267 -

Higgs, E.S. 1994. Expanding the scope of restoration ecology. Restoration Ecology 2: 137-146. Higgs, E.S. 1997. What is good ecological restoration? Conservation Biology 11: 338- 348. Hilty, J., and A. Merenlender 2000. Faunal indicator taxa selection for monitoring ecosystem health. Biological Conservation 92: 185-197. Hindell, M.A., K.A. Handasyde, and A.K. Lee 1985. Tree species selection by free- ranging koala populations in Victoria. Wildlife Research 12: 137-144. Hindell, M.A., and A.K. Lee 1987. Habitat use and tree preferences of koalas in a mixed eucalypt forest. Wildlife Research 14: 349-360. Hobbs, R. 1998. Restoring the health and wealth of ecosystems. A Conference on Ecological Restoration in New Zealand, 28-30 September 1998, Christchurch, New Zealand. Available from http://www.landcareresearch.co.nz/news/conferences/ecorestoration/ (accessed 01/05/2010). Hobbs, R., and J. Harris 2001. Restoration ecology: Repairing the Earth's ecosystems in the new millennium. Restoration Ecology 9: 239-246. Hobbs, R.J., and A.J.M. Hopkins 1990. From frontier to fragments: European impact on Australia's vegetation. Proceedings of the Ecological Society of Australia 16: 93- 114. Hobbs, R.J. 1993. Can revegetation assist in the conservation of biodiversity in agricultural areas? Pacific Conservation Biology 1: 29-38. Hobbs, R.J., and D.A. Norton 1996. Towards a conceptual framework for restoration ecology. Restoration Ecology 4: 93-110. Hobbs, R.J. 2003. Ecological management and restoration: Assessment, setting goals, and measuring success. Ecological Management & Restoration 4: S2-S3. Hobbs, R.J. 2004. Restoration ecology: The challenge of social values and expectations. Frontiers in Ecology and the Environment 2: 43-48. Hobbs, R.J. 2007. Setting effective and realistic restoration goals: Key directions for research. Restoration Ecology 15: 354-357. Hodacová, D., and K. Prach 2003. Spoil heaps from brown coal mining: Technical reclamation versus spontaneous revegetation. Restoration Ecology 11: 385-391.

- 268 -

Hohberg, K. 2003. Soil nematode fauna of afforested mine sites: Genera distribution, trophic structure and functional guilds. Applied Soil Ecology 22: 113-126. Hone, J., and W. Martin 1998. A study of dung decay and plot size for surveying feral pigs using dung counts. Wildlife Research 25. Hong, P.N. 2001. Reforestation of mangroves after severe impacts of herbicides during the the Viet Nam war: the case of Can Gio. Unasylva 52: 57-60. Horváth, G., and J. Zeil 1996. Kuwait oil lakes as insect traps. Nature 379: 303-304. Horváth, G., B. Bernáth, and G. Molnár 1998. Dragonflies find crude oil visually more attractive than water: Multiple-choice experiments on dragonfly polarotaxis. Naturwissenschaften 85: 292-297. Hothorn, T., A. Zeileis, G. Millo, and D. Mitchell 2010. Package ‘lmtest’, Testing Linear Regression Models. Version 0.9-27. Hume, I. 1990. Biological basis for the vulnerability of koalas to habitat fragmentation. Pages 32-38 in D. Lunney, C.A. Urquhart and P. Reed, editors. Koala Summit: Managing Koalas in New South Wales, Sydney. NSW National Parks and Wildlife Service, Sydney. Hurlbert, S.H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54: 187-211. Hurlbert, S.H. 1994. Old shibboleths and new syntheses. Trends in Ecology and Evolution 9: 495-496. Hurvich, C.M., and C.-L. Tsai 1989. Regression and time series model selection in small samples. Biometrika 76: 297-307. Hurvich, C.M., and C.-L. Tsai 1995. Model selection for extended quasi-likelihood models in small samples. Biometrics 51: 1077-1084. Hutchinson, G.E. 1957. Concluding remarks, Cold Spring Harbor Symposium. Quantitative Biology 22: 415-427. IBM 2009. PAWS. Version 18.0. SPSS Inc., Chicago. Ipatyev, V.A. 2001. Healing the damage of Chernobyl: Radiation-contaminated forests and their rehabilitation. Unasylva 52: 37-42. IUCN 2001. IUCN Red List Categories and Criteria: Version 3.1. IUCN Species Survival Commission, Gland, Switzerland and Cambridge, UK.

- 269 -

IUCN, and ICMM 2004. Integrating mining and biodiversity conservation: Case studies from around the world. IUCN, Gland, Switzerland and Cambridge, UK and ICMM, London, UK. Jackman, S., A. Tahk, A. Zeileis, C. Maimone, and J. Fearon 2010. Political Science Computational Laboratory, Stanford University. Version Version 1.03.6. Jackson, G.P., and B.J. Fox 1996. Comparison of regeneration following burning, clearing or mineral sand mining at Tomago, NSW: II. Succession of ant assemblages in a coastal forest. Austral Ecology 21: 200-216. Jackson, L.L., N. Lopoukhine, and D. Hillyard 1995. Commentary ecological restoration: A definition and comments. Restoration Ecology 3: 71-75. Jackson, S.M. 2000a. Population dynamics and life history of the mahogany glider, Petaurus gracilis, and the sugar glider, Petaurus breviceps, in north Queensland. Wildlife Research 27: 21-37. Jackson, S.M. 2000b. Habitat relationships of the mahogany glider, Petaurus gracilis, and the sugar glider, Petaurus breviceps. Wildlife Research 27: 39-48. Jackson, S.M. 2007. Koala: Origins of an icon. Jacana Books, Crows Nest, Australia. James, T. 2007. The use of fibre reinforced cement nest boxes by gliders on North Stradbroke Island. Bachelor of Environmental Management. University of Queensland, Brisbane. Jentsch, A. 2007. The challenge to restore processes in face of nonlinear dynamics - on the crucial role of disturbance regimes. Restoration Ecology 15: 334-339. Johnson, C.N., and P.J. Jarman 1987. Macropod studies at Wallaby creek VI. A validation of the use of dung-pellet counts for measuring absolute densities of populations of Macropodids. Australian Wildlife Research 14: 139-145. Johnson, J.B., and K.S. Omland 2004. Model selection in ecology and evolution. Trends in Ecology and Evolution 19: 101-108. Jones, Z.F., and C.E. Bock 2005. The Botteri's sparrow and exotic Arizona grasslands: An ecological trap or habitat regained? The Condor 107: 731-741. Jordan, W.R., R.L. Peters, and E.B. Allen 1988. Ecological restoration as a strategy for conserving biological diversity. Environmental Management 12: 55-72. Jurskis, V., and M. Potter 1997. Koala surveys, ecology and conservation at Eden. Research paper No.34. Forest research and development division, State Forests of New South Wales, Sydney.

- 270 -

Kabay, E.D., and O.G. Nichols 1980. Use of rehabilitated bauxite mined areas in the jarrah forest by vertebrates. Alcoa of Australia Limited Environmental Bulletin 8: 1- 11. Kavanagh, R., S. Debus, T. Tweedie, and R. Webster 1995. Distribution of nocturnal forest birds and mammals in north-eastern New South Wales: Relationships with environmental variables and management history. Wildlife Research 22: 359-377. Kavanagh, R.P., M.A. Stanton, and T.E. Brassil 2007. Koalas continue to occupy their previous home-ranges after selective logging in Callitris–Eucalyptus forest. Wildlife Research 34: 94-107. Kearns, A., and G. Barnett 1999. Use of ecosystem function analysis in the mining industry. Pages 31-46 in C.J. Asher and L.C.Bell, editors. Proceedings of the Workshop on Indicators of Ecosystem Rehabilitation Success, Melbourne, Victoria. Australian Centre for Mining Environmental Research, Brisbane. Keith, D.A., T.G. Martin, E. McDonald-Madden, and C. Walters 2011. Special Issue Editorial: Adaptive management for biodiversity conservation in an uncertain world - Uncertainty and adaptive management for biodiversity conservation. Biological Conservation 144: 1175-1178. Kendall, K.C., L.H. Metzgar, D.A. Patterson, and B.M. Steele 1992. Power of sign surveys to monitor population trends. Ecological Applications 2: 422-430. Kentish, K.A., and A.R. Bourne 1982. Return of fauna to heathland regenerating after mining. Bulletin of the Australian Mammal Society 7: 29. Kerley, L.L., J.M. Goodrich, D.G. Miquelle, E.N. Smirnov, H.B. Quigley, and M.G. Hornocker 2002. Effects of roads and human disturbance on Amur tigers. Conservation Biology 16: 97-108. Kéry, M. 2002. Inferring the absence of a species: A case study of snakes. The Journal of Wildlife Management 66: 330-338. Kimber, S.L., A.F. Bennett, and P.A. Ryan 1999. Revegetation and wildlife: What do we know about revegetation and wildlife conservation in Australia? A report to Environment Australia. School of Ecology and Environment Deakin University. Knight, R. 1999. Recalcitrant recolonisers: Assessment of rehabilitation success through study of small faunal assemblages. Pages 53-64 in C.J. Asher and L.C. Bell, editors. Proceedings of the Workshop on Indicators of Ecosystem Rehabilitation Success, Melbourne. Australian Centre for Mining Environmental Research, Brisbane.

- 271 -

Koch, J.M. 2007. Alcoa's mining and restoration process in south western Australia. Restoration Ecology 15: S11-S16. Koch, J.M., A.H. Grigg, R.K. Gordon, and J.D. Majer 2010. Arthropods in coarse woody debris in jarrah forest and rehabilitated bauxite mines in Western Australia. Annals of Forest Science 67: 106. Kolbe, J.J., and F.J. Janzen 2002. Impact of nest-site selection on nest success and nest temperature in natural and disturbed habitats. Ecology 83: 269-281. Kriska, G., G. Horváth, and S. Andrikovics 1998. Why do mayflies lay their eggs en masse on dry asphalt roads? Water-imitating polarized light reflected from asphalt attracts Ephemeroptera. Journal of Experimental Biology 201: 2273-2286. Kristan, W.B. 2003. The role of habitat selection behavior in population dynamics: Source-sink systems and ecological traps. Oikos 103: 457-468. Krockenberger, A. 2003. Meeting the energy demands of reproduction in female koalas, Phascolarctos cinereus: Evidence for energetic compensation. Journal of Comparative Physiology B: Biochemichal, Systemic, and Environmental Physiology 173: 531-540. Kuehl, H.S., A. Todd, C. Boesch, and P.D. Walsh 2007. Manipulating decay time for efficient large-mammal density estimation: Gorillas and dung height. Ecological Applications 17: 2403-2414. Lackey, R.T. 2004. A response to the article (Hobbs 2004): Restoration ecology: The challenge of social values and expectations. Frontiers in Ecology and the Environment 2: 43-48. Ladiges, P.Y., and D.H. Ashton 1974. Variation in some Central Victoria populations of Eucalyptus viminalis Labill. Australian Journal of Botany 22: 81-102. Lambeck, R.J. 1997. Focal species: A multi-species umbrella for nature conservation. Conservation Biology 11: 849-856. Lambeck, R.J. 1999. Landscape planning for biodiversity conservation in agricultural regions: A case study from the wheatbelt of Western Australia. Department of the Environment and Heritage, Canberra. Lambert, D. 1992. Zero-inflated Poisson regression, with an application to defects in manufacturing. Technometrics 34: 1-14.

- 272 -

Lannoo, M.J., V.C. Kinney, J.L. Heemeyer, N.J. Engbrecht, A.L. Gallant, and R.W. Klaver 2009. Mine spoil prairies expand critical habitat for endangered and threatened amphibian and reptile species. Diversity 1: 118-132. Lawler, I.R., W.J. Foley, B.M. Eschler, D.M. Pass, and K. Handasyde 1998. Intraspecific variation in Eucalyptus secondary metabolites determines food intake by folivorous marsupials. Oecologia 116: 160-169. Lawler, I.R., W.J. Foley, and B.M. Eschler 2000. Foliar concentration of a single toxin creates habitat patchiness for a marsupial folivore. Ecology 81: 1327-1338. Laycock, J.W. 1975. Hydrogeology of North Stradbroke Island. Proceedings of the Royal Society of Queensland 86: 15-19. Laycock, J.W. 1978. North Stradbroke Island. Papers of the Department of Geology, University of Queensland 8: 89-96. Leader-Williams, N., and H. Dublin 2000. Charismatic megafauna as 'flagship species'. Pages 53-81 in A. Entwistle and N. Dunstone, editors. Priorities for the Conservation of Mammalian Diversity. Has the Panda Had Its Day? Cambridge University Press, Cambridge. Lee, T.M., and W. Jetz 2008. Future battlegrounds for conservation under global change. Proceedings of the Royal Society B: Biological Sciences 275: 1261-1270. Letnic, M., and B.J. Fox 1997a. The impact of industrial fluoride fallout on faunal succession following sand mining of dry sclerophyll forest at Tomago, N.S.W.: I Lizard recolonization. Biological Conservation 80: 63-81. Letnic, M., and B.J. Fox 1997b. The impact of industrial fluoride fallout on faunal succession following sand mining of dry sclerophyll forest at Tomago, N.S.W.: II Myobatrachid frog recolonization. Biological Conservation 82: 137-146. Lindenmayer, D. 2000. Factors at multiple scales affecting distribution patterns and their implications for animal conservation – Leadbeater’s Possum as a case study. Biodiversity and Conservation 9: 15-35. Lindenmayer, D., et al. 2008a. A checklist for ecological management of landscapes for conservation. Ecology Letters 11: 78-91. Lindenmayer, D.B., R.B. Cunningham, M.T. Tanton, A.P. Smith, and H.A. Nix 1991. Characteristics of hollow-bearing trees occupied by arboreal marsupials in the montane ash forests of the Central Highlands of Victoria, south-east Australia. Forest Ecology and Management 40: 289-308

- 273 -

Lindenmayer, D.B. 1997. Differences in the biology and ecology of arboreal marsupials in forests of Southeastern Australia. Journal of Mammalogy 78: 1117-1127. Lindenmayer, D.B., and R.B. Cunningham 1997. Patterns of co-occurrence among arboreal marsupials in the forests of central Victoria, southeastern Australia. Austral Ecology 22: 340-346. Lindenmayer, D.B. 1999. Future directions for biodiversity conservation in managed forests: Indicator species, impact studies and monitoring programs. Forest Ecology and Management 115: 277-287. Lindenmayer, D.B., R.B. Cunningham, M.L. Pope, and C.F. Donnelly 1999. The response of arboreal marsupials to landscape context: A large-scale fragmentation study. Ecological Applications 9: 594-611. Lindenmayer, D.B., C.R. Margules, and D.B. Botkin 2000. Indicators of biodiversity for ecologically sustainable forest management. Conservation Biology 14: 941-950. Lindenmayer, D.B. 2002. Gliders of Australia: A natural history. University of New South Wales Press Ltd, Sydney. Lindenmayer, D.B., A. Manning, P.L. Smith, M.A. McCarthy, H.P. Possingham, J. Fischer, and I. Oliver 2002. The focal species approach and landscape restoration: A critique. Conservation Biology 16: 338-345. Lindenmayer, D.B., and R.J. Hobbs 2004. Fauna conservation in Australian plantation forests - a review. Biological Conservation 119: 151-168. Lindenmayer, D.B., et al. 2007. The complementarity of single-species and ecosystem- oriented research in conservation research. Oikos 116: 1220-1226. Lindenmayer, D.B., J. Fischer, A. Felton, M. Crane, D. Michael, C. Macgregor, R. Montague-Drake, A. Manning, and R.J. Hobbs 2008b. Novel ecosystems resulting from landscape transformation create dilemmas for modern conservation practice. Conservation Letters 1: 129-135. Lindenmayer, D.B., A. Welsh, C. Donnelly, M. Crane, D. Michael, C. Macgregor, L. McBurneya, R. Montague-Drake, and P. Gibbons 2009. Are nest boxes a viable alternative source of cavities for hollow-dependent animals? Long-term monitoring of nest box occupancy, pest use and attrition. Biological Conservation 142: 33-42. Liu, J., M. Linderman, Z. Ouyang, L. An, J. Yang, and H. Zhang 2001. Ecological degradation in protected areas: The case of Wolong Nature Reserve for giant pandas. Science 292: 98-101.

- 274 -

Lloyd, J.D., and T.E. Martin 2005. Reproductive success of Chestnut-collared Longspurs in native and exotic grassland. The Condor 107: 363-374. Lloyd, M.V., G. Barnett, M.D. Doherty, R.A. Jeffree, J. John, J.D. Majer, J.M. Osborne, and O.G. Nichols 2002. Managing the impacts of the Australian minerals industry on biodiversity. Australian Centre for Mining Environmental Research. Longcore, T. 2003. Terrestrial arthropods as indicators of ecological restoration success in coastal sage scrub (California, U.S.A.). Restoration Ecology 11: 397-409. Lovell, S., M. Hamer, R. Slotow, and D. Herbert 2007. Assessment of congruency across invertebrate taxa and taxonomic levels to identify potential surrogates. Biological Conservation 139: 113-125. Lubke, R.A., and A.M. Avis 1998. A review of the concepts and application of rehabilitation following heavy mineral dune mining. Marine Pollution Bulletin 37: 546-557. Lunney, D., and A. Matthews 1997. The changing roles of state and local government in fauna conservation outside nature reserves: A case study of koalas in New South Wales. Pages 97-106 in P. Hale and D. Lamb, editors. Conservation Outside Nature Reserves. Centre for Conservation Biology, University of Queensland, Brisbane. Lunney, D., S.S. Phillips, J. Callaghan, and D. Coburn 1998. Determining the distribution of koala habitat across a shire as a basis for conservation: A case study from Port Stephens, New South Wales. Pacific Conservation Biology 4: 186-196. Lunney, D., C. Moon, A. Matthews, and J. Turbill 1999. Coffs Harbour City koala plan of management. Part B, Coffs Harbour koala study. NSW National Parks and Wildlife Service, Hurstville, Australia. Lunney, D., A. Matthews, C. Moon, and S. Ferrier 2000. Incorporating habitat mapping into practical koala conservation on private lands. Conservation Biology 14: 669- 680. Mac Nally, R. 2000. Regression and model-building in conservation biology, biogeography and ecology: The distinction between – and reconciliation of – ‘predictive’ and ‘explanatory’ models. Biodiversity and Conservation 9: 655-671. Mac Nally, R., and E. Fleishman 2002. Using "indicator" species to model species richness: Model development and predictions. Ecological Applications 12: 79-92. MacArthur, R.H. 1967. Limiting similarity, convergence and divergence of coexisting species. The American Naturalist 101: 338-345.

- 275 -

MacKenzie, D.I., J.D. Nichols, B.L. Gideon, S. Droege, J.A. Royle, and C.A. Langtimm 2002. Estimating site occupancy rates when detection probabilities are less than one. Ecology 83: 2248-2255. MacKenzie, D.I. 2005. What are the issues with presence-absence data for wildlife managers? The Journal of Wildlife Management 69: 849-860. MacMahon, J.A., and K.D. Holl 2001. Ecological restoration, a key to conservation biology's future. Pages 245-269 in M.E. Soulé and G.H. Orians, editors. Conservation Biology: Research priorities for the next decade. Island Press, Washington, DC. Madden, K.E., and B.J. Fox 1997. Arthropods as indicators of the effects of fluoride pollution on the succession following sand mining. The Journal of Applied Ecology 34: 1239-1256. Majer, J.D. 1978a. Preliminary survey of the epigaeic invertebrate fauna with particular reference to ants, in areas of different land use at Dwellingup, Western Australia. Forest Ecology and Management 1: 321-334. Majer, J.D. 1978b. The importance of invertebrates in successful land reclamation with particular reference to bauxite mine rehabilitation. Pages 47-61 in J. Fox, editor. Rehabilitation of Mined Lands in Western Australia. Western Australia Institute of Technology, Bentley. Majer, J.D. 1978c. Studies of invertebrates in relation to bauxite mining activities in the Darling Range - a review of the first 18 months research. Alcoa of Australia Limited Environmental Bulletin 3: 1-19. Majer, J.D. 1981. The role of invertebrates in bauxite mine rehabilitation. Forests Department of Western Australia Bulletin 93: 1-29. Majer, J.D., M. Sartori, R. Stone, and W.S. Perriman 1982. Colonization by ants and other invertebrates in rehabilitated mineral sand mines near Eneabba, Western Australia. Reclamation and Revegetation Research 1: 63-81. Majer, J.D. 1983. Ants: Bio-indicators of minesite rehabilitation, land-use, and land conservation. Environmental Management 7: 375-383. Majer, J.D. 1984a. Recolonisation by ants in rehabilitated open-cut mines in northern Australia. Reclamation and Revegetation Research 2: 279-298.

- 276 -

Majer, J.D. 1984b. Ant return in rehabilitated mineral sand mines on North Stradbroke Island. Pages 325-332 in J. Covacevich, editor. Focus on Stradbroke. Boolarong Press, Brisbane. Majer, J.D. 1984c. Ant return in rehabilitated mines - an indicator of ecosystem resilience. Pages 105-108 in B. Dell, editor. Proceedings of the 4th International Conference on Mediterranean Ecosystems. University of Western Australia, Perth. Majer, J.D., J. Day, E. Kabay, and W. Perriman 1984. Recolonization by ants in bauxite mines rehabilitated by a number of different methods. The Journal of Applied Ecology 21: 355-375. Majer, J.D. 1985. Recolonization by ants of rehabilitated mineral sand mines on North Stradbroke Island, Queensland, with particular reference to seed removal. Australian Journal of Ecology 10: 31-48. Majer, J.D. 1989. Animals in primary succession; the role of fauna in reclaimed lands. Cambridge University Press, Cambridge. Majer, J.D. 1990. Rehabilitation of disturbed land: Long-term prospects for the recolonisation of fauna. Proceedings of the Ecological Society of Australia 16: 509- 519. Majer, J.D., and E. Brown 1998. The role of invertebrates in ecological functioning. Pages 11-30 in C. Asher and L. Bell, editors. Proceedings of Fauna Habitat Reconstruction after Mining Workshop, Adelaide. Australian Centre for Mining Environmental Research, Brisbane. Majer, J.D., and O. Nichols 1998. Long-term recolonization patterns of ants in Western Australian rehabilitated bauxite mines, with reference to use as indicators of restoration success. The Journal of Applied Ecology 35: 161-182. Majer, J.D., K.E.C. Brennan, and M.L. Moir 2007. Invertebrates and the restoration of a forest ecosystem: 30 years of research following bauxite mining in Western Australia. Restoration Ecology 15: S104-S115. Majer, J.D. 2009. Animals in the restoration process; progressing the trends. Restoration Ecology 17: 315-319. Manning, A.D., D.B. Lindenmayer, and H.A. Nix 2004. Continua and umwelt: Novel perspectives on viewing landscapes. Oikos 104: 621-628.

- 277 -

Marcus, W.A., G.A. Meyer, and D.R. Nimmo 2001. Geomorphic control of persistent mine impacts in a Yellowstone Park stream and implications for the recovery of fluvial systems. Geology 29: 355-358 Martin, R.W. 1981. Age-specific fertility in three populations of the Koala, Phascolarctos cinereus Goldfuss, in Victoria. Wildlife Research 8: 275-283. Martin, R.W. 1985. Overbrowsing, and decline of a population of the koala, Phascolarctos cinereus, in Victoria. I. Food preference and food tree defoliation. Wildlife Research 12: 355-365. Martin, R.W., and K. Handasyde 1999. The Koala: Natural history, conservation and management, Australian Natural History Series. University of New South Wales Press Ldt, Hong Kong. Martin, T.G., B.A. Wintle, J.R. Rhodes, P.M. Kuhnert, S.A. Field, S.J. Low-Choy, A.J. Tyre, and H.P. Possingham 2005. Zero tolerance ecology: Improving ecological inference by modelling the source of zero observations. Ecology Letters 8: 1235- 1246. Massei, G., P. Bacon, and P.V. Genov 1998. Fallow deer and wild boar pellet group disappearance in a Mediterranean area. The Journal of Wildlife Management 62: 1086-1094. Masunga, G.S., Ø. Andresen, J.E. Taylor, and S.S. Dhillion 2006. Elephant dung decomposition and coprophilous fungi in two habitats of semi-arid Botswana. Mycological Research 110: 1214-1226. Matthews, A., D. Lunney, S. Gresser, and W. Maitz 2007. Tree use by koalas (Phascolarctos cinereus) after fire in remnant coastal forest. Wildlife Research 34: 84-93. Mawson, P. 1986. A comparative study of Arachnid communities in rehabilitated bauxite mines. WAIT School of Biology Bulletin 14: 1-50. May, S.A., and T.W. Norton 1996. Influence of fragmentation and disturbance on the potential impact of feral predators on native fauna in Australian forest ecosystems. Wildlife Research 23: 387-400. Mazerolle, M.J., M. Poulin, C. Lavoie, L. Rochefort, A. Desrochers, and B. Drolet 2006. Animal and vegetation patterns in natural and man-made bog pools: implications for restoration. Freshwater Biology 51: 333-350.

- 278 -

McAlpine, C.A., D. Lindenmayer, T.J. Eyre, and S.R. Phinn 2002. Landscape surrogates of forest fragmentation: Synthesis of Australian Montreal Process case studies. Pacific Conservation Biology 8: 108-120. McAlpine, C.A., M.E. Bowen, J.G. Callaghan, D. Lunney, J.R. Rhodes, D.L. Mitchell, D.V. Pullar, and H.P. Possingham 2006a. Testing alternative models for the conservation of koalas in fragmented rural-urban landscapes. Austral Ecology 31: 529-544. McAlpine, C.A., J.R. Rhodes, J.G. Callaghan, M.E. Bowen, D. Lunney, D.L. Mitchell, D.V. Pullar, and H.P. Possingham 2006b. The importance of forest area and configuration relative to local habitat factors for conserving forest mammals: A case study of koalas in Queensland, Australia. Biological Conservation 132: 153-165. McAlpine, C.A., T.A. Spies, P. Norman, and A. Peterson 2007. Conserving forest biodiversity across multiple land ownerships: Lessons from the Northwest Forest Plan and the Southeast Queensland regional forests agreement (Australia). Biological Conservation 134: 580-592. McAlpine, C.A., J.R. Rhodes, M. Bowen, D. Lunney, J.G. Callaghan, D.L. Mitchell, and H.P. Possingham 2008. Can multiscale models of species' distribution be generalized from region to region? A case study of the koala. Journal of Applied Ecology 45: 558–567. McCullagh, P., and J. Nelder 1989. Generalized linear models. Second edition. Chapman and Hall, London, UK. McDonald, T. 2010. Restoration and rehabilitation start with stopping the damage. Ecological Management and Restoration 11: 1. McGlynn, J., and A. Coutts 2006. Guidelines to the mining, rehabilitation and environmental management process. 1-30. NSW Department of Primary Industries - Mineral Resources. McIntyre, S., and R. Hobbs 1999. A framework for conceptualizing human effects on landscapes and its relevance to management and research models. Conservation Biology 13: 1282-1292. McIntyre, S., J. McIvor, and K. Heard 2002. Managing and conserving grassy woodlands. CSIRO Publishing, Melbourne. McKenzie, D.H., D.E. Hyatt, and V.J. McDonald 1992. Ecological indicators. Volumes 1 and 2. Elsevier Science Publishers, New York.

- 279 -

McKinney, M.L. 2002. Urbanization, biodiversity, and conservation. BioScience 52: 883-890. McNee, S.A., A. Zigon, and B.G. Collins 1995. Population ecology of vertebrates in undisturbed and rehabilitated habitats on the northern sandplain of Western Australia. WAIT School of Biology Bulletin 16: 1-132. McNeely, J., and K. Miller 1984. National Parks conservation and development: The role of protected areas in sustaining society. Proceedings of the World Congress of National Parks. Smithsonian Institution Press, Washington, D.C. Melzer, A., M.A. Schneider, and D. Lamb 1994. Insects associated with the faecal pellets of the koala, Phascolarctos cinereus Goldfuss. Australian Entomologist 21: 69-70. Melzer, A., G. Tucker, J. Hodgson, and B. Elliott 2003. A note on predation on koalas, Phascolarctos cinereus, by raptors, including wedgetailed eagles, Aquila audax, in Queensland. Queensland Naturalist 41: 38-41. Melzer, A. 2011. Fox predation on koalas. Available from http://www.aph.gov.au/senate/committee/ec_ctte/koalas/submissions.htm (accessed 24/10/2011). Michener, W.K. 1997. Quantitatively evaluating restoration experiments: Research design, statistical analysis, and data management considerations. Restoration Ecology 5: 324-337. Miller, J.R., and R.J. Hobbs 2007. Habitat restoration - do we know what we’re doing? Restoration Ecology 15: 382-390. Millis, A.L., and A.J. Bradley 2001. Reproduction in the squirrel glider, Petaurus norfolcensis (Petauridae) in south-east Queensland. Australian Journal of Zoology 49: 139-154. Moir, M.L., K.E.C. Brennan, J.M. Koch, J.D. Majer, and M.J. Fletcher 2005. Restoration of a forest ecosystem: The effects of vegetation and dispersal capabilities on the reassembly of plant dwelling arthropods. Forest Ecology and Management 217: 294-306. Montréal Process Working Group 2007. Criteria and indicators for the conservation and sustainable management of temperate and boreal forests. The Montréal Process, Third Edition. Available from http://www.rinya.maff.go.jp/mpci/meetings/an-6.pdf (accessed 27/1/2011).

- 280 -

Moore, B.D., and W.J. Foley 2000. A review of feeding and diet selection in koalas (Phascolarctos cinereus). Australian Journal of Zoology 48: 317-333. Moore, B.D., I.R. Wallis, K.J. Marsh, and W.J. Foley 2004. The role of nutrition in the conservation of the marsupial folivores of eucalypt forests. Pages 549 - 575 in D. Lunney, editor. Conservation of Australia’s Forest Fauna. 2nd edition. Royal Zoological Society of New South Wales, Mosman, Australia. Moore, B.D., and W.J. Foley 2005. Tree use by koalas in a chemically complex landscape. Nature 435: 488-490. Moore, B.D., W.J. Foley, I.R. Wallis, A. Cowling, and K.A. Handasyde 2005. Eucalyptus foliar chemistry explains selective feeding by koalas. Biology Letters 1: 64-67. Moore, S.A., et al. 2009. Diversity in current ecological thinking: Implications for environmental management. Environmental Management 43: 17-27. Morrison, M.L. 2001. Introduction: Concepts of wildlife and wildlife habitat for ecological restoration. Restoration Ecology 9: 251-252. Mundry, R. 2011. Issues in information theory-based statistical inference - a commentary from a frequentist’s perspective. Behavioral Ecology and Sociobiology 65: 57-68. Munks, S., R. Corkrey, and W. Foley 1996. Characteristics of arboreal marsupial habitat in the semi-arid woodlands of Northern Queensland. Wildlife Research 23: 185-195. Munro, N.T., D.B. Lindenmayer, and J. Fischer 2007. Faunal response to revegetation in agricultural areas of Australia: A review. Ecological Management & Restoration 8: 199-207. Murray, D., E. Ellsworth, and A. Zack 2005. Assessment of potential bias with snowshoe hare fecal pellet-plot counts. The Journal of Wildlife Management 69: 385-395. Nchanji, A.C., and A.J. Plumptre 2001. Seasonality in elephant dung decay and implications for censusing and population monitoring in south-western Cameroon. African Journal of Ecology 39: 24-32. Neff, D.J. 1968. The pellet-group count technique for big game trend, census, and distribution: A review. The Journal of Wildlife Management 32: 597-614. New, T.R. 2004. Concluding comment: Looking to the future for moths. Journal of Insect Conservation 8: 275-276.

- 281 -

Newton, I., M. Marquiss, D.N. Weir, and D. Moss 1977. Spacing of sparrowhawk nesting territories. Journal of Animal Ecology 46: 425-441. Nichols, O., and S. Bunn 1980. Termites utilisation of rehabilitated bauxite mined areas. Alcoa of Australia Limited Environmental Bulletin 9: 1-9. Nichols, O. 1985. Recolonisation of revegetated bauxite mine sites by predatory invertebrates. Forest Ecology and Management 10: 49-64. Nichols, O., B.J. Wykes, and D. Majer 1989. The return of vertebrate and invertebrate fauna to bauxite mined areas in south-western Australia. Pages 397-422 in J.D. Majer, editor. Animals in Primary Succession: the Role of Fauna in Reclaimed Lands. Cambridge University Press, Cambridge. Nichols, O.G., and D. Watkins 1984. Bird utilisation of rehabilitated bauxite minesites in Western Australia. Biological Conservation 30: 109-131. Nichols, O.G., and M.J. Bamford 1985. Reptile and frog utilisation of rehabilitated bauxite minesites and dieback-affected sites in Western Australia's Jarrah Eucalyptus marginata forest. Biological Conservation 34: 227-249. Nichols, O.G. 1998a. The development of a rehabilitation program designed to restore a jarrah forest ecosystem following bauxite mining in south-western Australia. Pages 315-325 in H.R. Fox, H.M. Moore and A.D. McIntosh, editors. Land reclamation - achieving sustainable benefits. Proceedings of the Fourth International Conference of the International Affiliation of Land Reclamationists, Nottingham. Balkema, Rotterdam. Nichols, O.G. 1998b. Long-term monitoring of fauna return in Bauxite-mined areas of the Darling range. Pages 47-56 in C.J. Asher and L.C. Bell, editors. Proceedings of Fauna Habitat Reconstruction after Mining Workshop, Adelaide. Australian Centre for Mining Environmental Research, Brisbane. Nichols, O.G., and F.M. Nichols 2003. Long-term trends in faunal recolonization after bauxite mining in the Jarrah forest of southwestern Australia. Restoration Ecology 11: 261-272. Nichols, O.G., and C.D. Grant 2007. Vertebrate fauna recolonization of restored bauxite mines - key findings from almost 30 years of monitoring and research. Restoration Ecology 15: S116-S126. NPWS 2003. Draft recovery plan for the koala. New South Wales National Parks and Wildlife Service, Hurstville, NSW.

- 282 -

O’Brien, R. 2007. A caution regarding rules of thumb for variance inflation factors. Quality & Quantity 41: 673-690. Oksanen, J., F.G. Blanchet, R. Kindt, P. Legendre, R.B. O'Hara, G.L. Simpson, M. Peter Solymos, H.H. Stevens, and H. Wagner 2011. Package ‘vegan’, Community Ecology Package. Version 1.17-8. Oksanen, L. 2001. Logic of experiments in ecology: Is pseudoreplication a pseudoissue? Oikos 94: 27-38. Oksanen, L. 2004. The devil lies in details: Reply to Stuart Hurlbert. Oikos 104: 598- 605. Ottonetti, L., L. Tucci, and G. Santini 2006. Recolonization patterns of ants in a rehabilitated lignite mine in central Italy: Potential for the use of mediterranean ants as indicators of restoration processes. Restoration Ecology 14: 60-66. Packard, J.M., R.K. Frohlich, J.E.R. Iii, and J.R. Wilcox 1989. Manatee response to interruption of a thermal effluent. The Journal of Wildlife Management 53: 692-700. Pahl, L. 1996. Koala and bushland survey of west and central Logan City. Pages 82-92 in G. Gordon, editor. Koalas: Research for management. Proceedings of the Brisbane Koala Symposium, 1990, Brisbane, Queensland. World Koala Research. Pakkala, T., J. Pellikka, and H. Linden 2003. Capercaillie Tetrao urogallus - a good candidate for an umbrella species in taiga forests. Wildlife Biology 9: 309-316. Palmer, M.A., R.F. Ambrose, and N.L. Poff 1997. Ecological theory and community restoration ecology. Restoration Ecology 5: 291-300. Parra, R. 1978. Comparison of foregut and hindgut fermentation in herbivores. Pages 205-229 in G. Montgomery, editor. The ecology of arboreal folivores. Smithsonian Institution Press, Washington, DC. Pausas, J.G., L.W. Braithwaite, and M.P. Austin 1995. Modelling habitat quality for arboreal marsupials in the south coastal forests of New South Wales, Australia. Forest Ecology and Management 78: 39-49. Pawitan, Y. 2001. In all likelihood. Oxford University Press. Perfecto, I., J. Vandermeer, P. Hanson, and V. Cartín 1997. Arthropod biodiversity loss and the transformation of a tropical agro-ecosystem. Biodiversity and Conservation 6: 935-945. Peto, R. 1973. Experimental survival curves for interval-censored data. Journal of the Royal Statistical Society, Series C 22: 86-91.

- 283 -

Phillips, S., J. Callaghan, and D. Staines 1995. The Koala habitat atlas - field survey guidelines, data analysis and habitat modelling details. Pages 63-67. Proceedings from "A conference on the status of the Koala in 1996", Greenmount, Coolangatta, Queensland. Australian Koala Foundation, Brisbane. Phillips, S., J. Callaghan, and V. Thompson 2000. The tree species preferences of koalas (Phascolarctos cinereus) inhabiting forest and woodland communities on Quaternary deposits in the Port Stephens area. Wildlife Research 27: 1-10. Pielou, E.C. 1975. Ecological diversity. Wiley, New York. Piggott, M., and A. Taylor 2003. Remote collection of animal DNA and its applications in conservation management and understanding the population biology of rare and cryptic species. Wildlife Research 30: 1-13. Pimentel, D., U. Stachow, D.A. Takacs, H.W. Brubaker, A.R. Dumas, J.J. Meaney, J.A.S. O'Neil, D.E. Onsi, and D.B. Corzilius 1992. Conserving biological diversity in agricultural/forestry systems. BioScience 42: 354-362. Pimm, S.L., G.J. Russell, J.L. Gittleman, and T.M. Brooks 1995. The future of biodiversity. Science 269: 347-350. Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R Development Core Team 2010. nlme: Linear and Nonlinear Mixed Effects Models. Version 3.1-97. R Foundation for Statistical Computing, Vienna. Preece, H.J. 2007. Identifying hotspots for threats to koalas using spatial analysis. MODSIM 2007 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand. Available from http://www.mssanz.org.au/MODSIM07/papers/21_s46/IdentifyingHotspots_s46_Pre ece_.pdf (accessed 20/02/2011). Primer-E Ltd. 2009. Primer 6.1.12 & Permanova + 1.0.2, Plymouth, United Kingdom. Prugh, L.R., and C.J. Krebs 2004. Snowshoe hare pellet-decay rates and aging in different habitats. Wildlife Society Bulletin 32: 386-393. Pulliam, H.R. 2000. On the relationship between niche and distribution. Ecology Letters 3: 349-361. Pulliam, R. 1988. Sources, sinks, and population regulation. The American Naturalist 132: 652-661. Putman, R.J. 1984. Facts from faeces. Mammal Review 14: 79-97.

- 284 -

Queensland Government 2004. Policy framework to encourage progressive rehabilitation of large mines. Available from http://www.derm.qld.gov.au/register/p01471aa.pdf (accessed 20/02/2011). Queensland Government 2007. Guideline 18: Rehabilitation requirements for mining projects. Available from http://www.derm.qld.gov.au/register/p01673aa.pdf (accessed 20/02/2011). Queensland Government 2010a. Offsets for Net Gain of Koala Habitat in South East Queensland Policy. Sustainable Communities and Landscapes Department of Environment and Resource Management Available from http://www.derm.qld.gov.au/wildlife- ecosystems/wildlife/koalas/koala_crisis_response_strategy/documents/offset- netgain.pdf (accessed 8/04/11). Queensland Government 2010b. Guideline Environmental Protection Act 1994: Calculating financial assurance for mining projects. Department of Environment and Resource Management. Available from http://www.derm.qld.gov.au/services_resources/item_details.php?item_id=200507 (accessed 24/01/2011). Queensland Government undated. Mining fact sheet 10 - Financial assurance. Environmental Protection Agency. Available from http://www.derm.qld.gov.au/register/p00322aa.pdf (accessed 14/10/2010). Queensland Herbarium 2009. Regional Ecosystem Description Database (REDD). Version Version 6.0b, updated November 2009. Department of Environment and Resource Management, Brisbane. Quin, D. 1995. Population ecology of the squirrel glider (Petaurus norfolcensis) and the sugar glider (P. breviceps) (Maruspialia : Petauridae) at Limeburners Creek, on the central north coast of New South Wales. Wildlife Research 22: 471-505. Quinn, G., and M. Keough 2006. Experimental design and data analysis for biologists. 5th edition. Cambridge University Press, Cambridge. R Development Core Team 2010. R: a language and environment for statistical computing. Version 2.12.0. R Foundation for Statistical Computing, Vienna. Rae, M., and A. Rouse 2001. Mining certification evaluation project: Independent certification of environmental and social performance in the mining sector. WWF Australia, Melbourne.

- 285 -

Reeders, A.P.F. 1985. Vertebrate fauna in regenerated mines at Weipa, North Queensland. Pages 105-118 in J.W. Lawrie, editor. Proceedings of the North Australian Mine Regeneration Workshop, Weipa. Comalco Aluminium, Weipa. Rhodes, J.R., T. Wiegand, C.A. McAlpine, J. Callaghan, D. Lunney, M. Bowen, and H.P. Possingham 2006. Modeling species' distributions to improve conservation in semiurban landscapes: Koala case study. Conservation Biology 20: 449-459. Rhodes, J.R., J.G. Callaghan, C.A. McAlpine, C.d. Jong, M.E. Bowen, D.L. Mitchell, D. Lunney, and H.P. Possingham 2008. Regional variation in habitat-occupancy thresholds: A warning for conservation planning. Journal of Applied Ecology 45: 549-557. Rhodes, J.R., C.A. McAlpine, A. Zuur, G.M. Smith, and E. Ieno 2009. GLMM applied on the spatial distribution of koalas in a fragmented landscape. Pages 469-492 in A. Zuur, E. Ieno, N. Walker, A. Saveliev and G. Smith, editors. Mixed Effects Models and Extensions in Ecology with R. Springer, New York. Rhodes, J.R., D. Lunney, C. Moon, A. Matthews, and C.A. McAlpine 2011. The consequences of using indirect signs that decay to determine species’ occupancy. Ecography 34: 141-150. Ries, L., D.M. Debinski, and M.L. Wieland 2001. Conservation value of roadside prairie restoration to butterfly communities. Conservation Biology 15: 401-411. Roberge, J.M., and P. Angelstam 2004. Usefulness of the umbrella species concept as a conservation tool. Conservation Biology 18: 76-85. Robertson, B.A., and R.L. Hutto 2006. A framework for understanding ecological traps and an evaluation of existing evidence. Ecology 87: 1075-1085. Robertson, B.A., and R.L. Hutto 2007. Is selectively harvested forest an ecological trap for Olive-sided Flycatchers? Condor 109: 109-121. Rodgers, A.R., A.P. Carr, H.L. Beyer, L. Smith, and J.G. Kie 2007. HRT: Home Range Tools for ArcGIS. Version 1.1. Ontario Ministry of Natural Resources, Centre for Northern Forest Ecosystem Research, Thunder Bay, Ontario, Canada. Rogers, R.W. 1975. North Stradroke Island: Resources and management. Proceedings of the Ecological Society of Australia 9: 296-306. Rohr, J.R., C.G. Mahan, and K.C. Kim 2007. Developing a monitoring program for invertebrates: Guidelines and a case study. Conservation Biology 21: 422-433.

- 286 -

Rondinini, C., and L. Boitani 2006. Differences in the umbrella effects of African amphibians and mammals based on two estimators of the area of occupancy. Conservation Biology 20: 170-179. Rosenzweig, M.L. 2001. Loss of speciation rate will impoverish future diversity. Proceedings of the National Academy of Sciences of the United States of America 98: 5404-5410. Ross, K.A., J.E. Taylor, M.D. Fox, and B.J. Fox 2004. Interaction of multiple disturbances: Importance of disturbance interval in the effects of fire on rehabilitating mined areas. Austral Ecology 29: 508-529. Royle, J.A., and J.D. Nichols 2003. Estimating abundance from repeated presence- absence data or point counts. Ecology 84: 777-790. Rubinoff, D. 2001. Evaluating the California gnatcatcher as an umbrella species for conservation of southern California coastal sage scrub. Conservation Biology 15: 1374-1383. Ruiz-Jaen, M.C., and T.M. Aide 2005. Restoration success: How is it being measured? Restoration Ecology 13: 569-577. Ryan, P. 2000. The use of revegetated areas by vertebrate fauna in Australia: A review. Pages 318-335 in H. R.J. and Y. C.J., editors. Temperate Eucalypt Woodlands in Australia: Biology, Conservation, Management and Restoration. Surrey Beatty & Sons, Chipping Norton, NSW. Samedi, I. 1995. Modelling habitat suitability for koalas and application of UET methods in planning for koala habitat. Doctor of Philosophy. The University of Queensland, Brisbane. Savard, J.-P.L., P. Clergeau, and G. Mennechez 2000. Biodiversity concepts and urban ecosystems. Landscape and Urban Planning 48: 131-142. Schläpfer, F. 1999. Expert estimates about effects of biodiversity on ecosystem processes and services. Oikos 84: 346-352. Schoener, T.W. 1981. An empirically based estimate of home range. Theoretical Population Biology 20: 281-325. Seaman, D.E., and R.A. Powell 1996. An evaluation of the accuracy of kernel density estimators for home range analysis. Ecology 77: 2075-2085.

- 287 -

Sebastiao, H., and C.E.V. Grelle 2009. Taxon surrogates among Amazonian mammals: Can total species richness be predicted by single orders? Ecological Indicators 9: 160 - 166. SER 2004. The SER international primer on ecological restoration. Society for Ecological Restoration International (www.ser.org) & Tucson. Shannon, C.E., and W. Weaver eds. 1949. The mathematical theory of communication. Univiversity of Illinois Press, Urbana. Sharpe, D.J., and R.L. Goldingay 1998. Feeding behaviour of the squirrel glider at Bungawalbin Nature Reserve, north-eastern New South Wales. Wildlife Research 25: 243-254. Sharpe, D.J., and R.L. Golgingay 2007. Home range of the Australian squirrel glider, Petaurus norfolcensis (Diprotodontia). Journal of Mammalogy 88: 1515-1522. Sharpe, D.J., and R.L. Goldingay 2010. Population ecology of the nectar-feeding squirrel glider (Petaurus norfolcensis) in remnant forest in subtropical Australia Wildlife Research 37: 77-88. Shochat, E., M.A. Patten, D.W. Morris, D.L. Reinking, D.H. Wolfe, and S.K. Sherrod 2005. Ecological traps in isodars: Effects of tallgrass prairie management on bird nest success. Oikos 111: 159-169. Shokri, M.R., W. Gladstone, and J. Jelbart 2009. The effectiveness of seahorses and pipefish (Pisces: Syngnathidae) as a flagship group to evaluate the conservation value of estuarine seagrass beds. Aquatic Conservation: Marine and Freshwater Ecosystem 19: 588-595. Shrader-Frechette, K., and E. McCoy 1993. Methods in ecology: Strategies for conservation. Cambridge University Press, Cambridge, U.K. Simberloff, D. 1988. The contribution of population and community biology to conservation science. Annual Review of Ecology and Systematics 19: 473-511. Simberloff, D. 1998. Flagships, umbrellas, and keystones: Is single-species management passé in the landscape era? Conservation Biology and Biodiversity Strategies 83: 247-257. Simmonds, S.J., J.D. Majer, and O.G. Nichols 1994. A comparative study of spider (Araneae) communities of rehabilitated bauxite mines and surrounding forest in the southwest of Western Australia. Restoration Ecology 2: 247-260.

- 288 -

Simpson, J. 1998. Reconstruction of koala habitat after titanium minerals mining in the Tomago Sandbeds, NSW. Pages 73-80 in C.J. Asher and L.C. Bell, editors. Proceedings of Fauna Habitat Reconstruction after Mining Workshop, Adelaide. Australian Centre for Mining Environmental Research, Brisbane. Sinclair, A.R.E., D.S. Hik, O.J. Schmitz, G.G.E. Scudder, D.H. Turpin, and N.C. Larter 1995. Biodiversity and the need for habitat renewal. Ecological Applications 5: 579- 587. Smith, A.P., and M. Murray 2003. Habitat requirements of the squirrel glider (Petaurus norfolcensis) and associated possums and gliders on the New South Wales central coast. Wildlife Research 30: 291-301. Smith, G.C., and G. Agnew 2002. The value of ‘bat boxes’ for attracting hollow dependent fauna to farm forestry plantations in southeast Queensland. Ecological Management and Restoration 3: 37-46. Smyth, C.R., and P. Dearden 1998. Performance standards and monitoring requirements of surface coal mine reclamation success in mountainous jurisdictions of western North America: A review. Journal of Environmental Management 53: 209-229. Sogge, M.K., S.J. Sferra, and E.H. Paxton 2008. Tamarix as habitat for birds: Implications for riparian restoration in the southwestern United States. Restoration Ecology 16: 146-154. Soulé, M.E., J.A. Estes, B. Miller, and D.L. Honnold 2005. Strongly interacting species: Conservation policy, management, and ethics. BioScience 55: 168-176. Specht, R.L. 2009. Structure and species richness in wetland continua on sandy soils in subtropical and tropical Australia. Austral Ecology 34: 761-772. Stanley, T.R., and J. Bart 1991. Effects of roadside habitat and fox density on a snow track survey for foxes in Ohio. Ohio Journal of Science 91: 186-190. Stanley, T.R., and J.A. Royle 2005. Estimating site occupancy and abundance using indirect detection indices. The Journal of Wildlife Management 69: 874-883. Stauffer, H.B., C.J. Ralph, and S.L. Miller 2004. Ranking habitat for marbled murrelets: New conservation approach for species with uncertain detection. Ecological Applications 14: 1374-1383. Sullivan, B.J., G.S. Baxter, and A.T. Lisle 2002. Low-density koala (Phascolarctos cinereus) populations in the mulgalands of south-west Queensland. I. Faecal pellet sampling protocol. Wildlife Research 29: 455-462.

- 289 -

Sullivan, B.J., G.S. Baxter, and A.T. Lisle 2003a. Low-density koala (Phascolarctos cinereus) populations in the mulgalands of south-west Queensland. III. Broad-scale patterns of habitat use. Wildlife Research 30: 583-591. Sullivan, B.J., W. Norris, and G.S. Baxter 2003b. Low-density koala (Phascolarctos cinereus) populations in the mulgalands of south-west Queensland. II. Distribution and diet. Wildlife Research 30: 331-338. Sullivan, B.J., G.S. Baxter, A.T. Lisle, L. Pahl, and W.M. Norris 2004. Low-density Koala (Phascolarctos cinereus) populations in the mulgalands of southwest Queensland. IV. Abundance and conservation status. Wildlife Research 31: 19 - 29. Suter, W., R.F. Graf, and R. Hess 2002. Capercaillie (Tetrao urogallus) and avian biodiversity: Testing the umbrella-species concept. Conservation Biology 16: 778- 788. Sutherland, W.J., and M.J.S. Peel 2010. Benchmarking as a means to improve conservation practice. Oryx 45: 56-59. Sutherland, W.J., et al. 2011. Horizon scan of global conservation issues for 2011. Trends in Ecology and Evolution in press. Taylor, J.E., and B.J. Fox 2001. Disturbance effects from fire and mining produce different lizard communities in eastern Australian forests. Austral Ecology 26: 193- 204. Temperton, V.M. 2007. The recent double paradigm shift in restoration ecology. Restoration Ecology 15: 344-347. Thoday, M. 1998. Ants: Bio-indicators of minesite rehabilitation. Groundwork 4: 16-17. Thomas, C.D., M. Singer, and D.A. Boughton 1996. Catastrophic extinction of population sources in a butterfly metapopulation. The American Naturalist 148: 957- 975. Thompson, G.G., and S.A. Thompson 2005. Mammals or reptiles, as surveyed by pit- traps, as bio-indicators of rehabilitation success for mine sites in the Goldfields region of Western Australia? Pacific Conservation Biology 11: 268-286. Thompson, G.H., and W.T. Ward 1975. Soil landscapes of North Stradbroke Island. Proceedings of the Royal Society of Queensland 86: 9-14. Thompson, S.A., and G.G. Thompson 2004. Adequacy of rehabilitation monitoring practices in the Western Australian mining industry. Ecological Management & Restoration 5: 30-33.

- 290 -

Tisdell, C., C. Wilson, and H.S. Nantha 2006. Public choice of species for the ‘Ark’: Phylogenetic similarity and preferred wildlife species for survival. Journal for Nature Conservation 14: 97-105. Tisdell, C., and H. Nantha 2007. Comparison of funding and demand for the conservation of the charismatic koala with those for the critically endangered wombat Lasiorhinus krefftii. Biodiversity and Conservation 16: 1261-1281. Tongway, D., and N. Hindley 2003. Indicators of ecosystem rehabilitation success. Stage Two – Verification of EFA indicators. Final Report. Australian Centre for Minesite Environmental Research. Topp, W., M. Simon, G. Kautz, U. Dworschak, F. Nicolini, and S. Pruckner 2001. Soil fauna of a reclaimed lignite open-cast mine of the Rhineland: Improvement of soil quality by surface pattern. Ecological Engineering 17: 307-322. Triggs, B. 1996. Tracks, scats and other traces: A field guide to Australian mammals. Oxford University Press, South Melbourne. Tulloch, A., H.P. Possingham, and K. Wilson 2011. Wise selection of an indicator for monitoring the success of management actions. Biological Conservation 144: 141- 154. Tun, N.U. 1993. Re-establishment of rehabilitated koalas in the wild and their use of habitat in Sheldon, Redland Shire, south-east Queensland, with particular reference to dietary selection. Master thesis. University of Queensland, Brisbane. Twigg, L.E., B.J. Fox, and L. Jia 1989. The modified primary succession following sand mining: A validation of the use of chronosequence analysis. Austral Ecology 14: 441-447. Twigg, L.E., and B.J. Fox 1991. Recolonization of regenerating open forest by terrestrial lizards following sand mining. Australian Journal of Ecology 16: 137-148. Tyndale-Biscoe, H. 2005. Life in the trees: Koala, greater glider and possum. Pages 219-266. Life of Marsupials. CSIRO Publishing, Collingwood, Australia. Tyre, A.J., B. Tenhumberg, S.A. Field, D. Niejalke, K. Parris, and H.P. Possingham 2003. Improving precision and reducing bias in biological surveys: Estimating false- negative error rates. Ecological Applications 13: 1790-1801. Ullrey, D.E., P.T. Robinson, and P.A. Whetter 1981. Composition of preferred and rejected browse offered to captive koalas, Phascolarctos cinereus (Marsupialia). Australian Journal of Zoology 29: 839-846.

- 291 -

Van Etten, R.C., and C.L. Bennett 1965. Some sources of error in using pellet-group counts for censusing deer. The Journal of Wildlife Management 29: 723-729. Van Gorp, L., and P.D. Erskine 2011. The influence of topsoil management on sand mine rehabilitation: Implications for ecosystem recovery. Proceedings of the Royal Society of Queensland in review. van Horn, B. 1983. Density as a misleading indicator of habitat quality. The Journal of Wildlife Management 47: 893-901. Van Shagen, J. 1986. Recolonisation by ants and other invertebrates in rehabilitated coal mine sites near Collie, Western Australia. WAIT School of Biology Bulletin 13: 1-26. Vesk, P.A., R. Nolan, J.R. Thomson, J.W. Dorrough, and R.M. Nally 2008. Time lags in provision of habitat resources through revegetation. Biological Conservation 141: 174-186. Vitousek, P.M., H.A. Mooney, J. Lubchenco, and J.M. Melillo 1997. Human domination of earth's ecosystems. Science 277: 494-499. Wali, M. 1999. Ecological succession and the rehabilitation of disturbed terrestrial ecosystems. Plant and Soil 213: 195-220. Walker, B. 1995. Conserving biological diversity through ecosystem resilience. Conservation Biology 9: 747-752. Walker, B.H. 1992. Biodiversity and ecological redundancy. Conservation Biology 6: 18-23. Walker, K., J. Osborne, and J. Dunlop 1986. Vertebrate colonization after revegetation of a rehabilitated iron ore waste dump, Newman, Western Australia. Pages 113-150 in D. Whitrow, editor. Proceedings of the 1986 Australian Mining Industry Council Environmental Workshop, Launceston. Australian Mining Industry Council, Canberra. Walker, L.R., and M.R. Willig 1999. An introduction to terrestrial disturbances. Pages 1-16 in L. Walker, editor. Ecosystems of disturbed ground. Ecosystems of the world, Elsevier, Amsterdam. Wallington, T.J., R.J. Hobbs, and S.A. Moore 2005. Implications of current ecological thinking for biodiversity conservation: A review of the salient issues. Ecology and Society 10: 15.

- 292 -

WallisDeVries, M.F., and I. Raemakers 2001. Does extensive grazing benefit butterflies in coastal dunes? Restoration Ecology 9: 179-188. Wallmo, O.C., A.W. Jackson, T.L. Hailey, and R.L. Carlisle 1962. Influence of rain on the count of deer pellet groups. The Journal of Wildlife Management 26: 50-55. Walpole, M.J., and N. Leader-Williams 2002. Tourism and flagship species in conservation. Biodiversity and Conservation 11: 543-547. Walsh, P.D., and L.J.T. White 2005. Evaluating the steady state assumption: Simulations of gorilla nest decay. Ecological Applications 15: 1342-1350. Wanner, M., and W. Dunger 2001. Biological activity of soils from reclaimed open-cast coal mining areas in Upper Lusatia using testate amoebae (protists) as indicators. Ecological Engineering 17: 323-330. Wanner, M., and W. Dunger 2002. Primary immigration and succession of soil organisms on reclaimed opencast coal mining areas in eastern Germany. European Journal of Soil Biology 38: 137−143. Ward, S. 2008. Assessment of capability to rehabilitate areas disturbed by mining at the Bemax Resources Ltd Happy Valley mineral sands operations. URS, Bunbury, Australia. Ward, S.C., J.M. Koch, and O.G. Nichols 1990. Bauxite mine rehabilitation in the Darling range, Western Australia. Proceedings of the Ecological Society of Australia 16: 557-565. Ward, W.T. 1978. Notes on the origin of Stradbroke Island. Papers of the Department of Geology, University of Queensland 8: 97-104. Watters, G.T., T. Menker, and S.H. O'Dee 2005. A comparison of terrestrial snail faunas between strip-mined land and relatively undisturbed land in Ohio, USA – an evaluation of recovery potential and changing faunal assemblages. Biological Conservation 126 166-174. Weaver, J.C. 1995. Indicator species and scale of observation. Conservation Biology 9: 939-942. Weiher, E. 2007. On the status of restoration science: Obstacles and opportunities. Restoration Ecology 15: 340-343. Weldon, A.J., and N.M. Haddad 2005. The effects of patch shape on indigo buntings: Evidence for an ecological trap. Ecology 86: 1422-1431.

- 293 -

Westman, W.E. 1975. Pattern and diversity in swamp and dune vegetation, North Stradbroke Island. Australian Journal of Botany 23: 339-354. White, L.J.T. 1995. Factors affecting the duration of elephant dung-piles in rain forest in the Lopé Reserve, Gabon. African Journal of Ecology 33: 142-150. White, N.A. 1999a. Ecology of the koala (Phascolarctos cinereus) in rural south-east Queensland, Australia. Wildlife Research 26: 731-744. White, P.S., and J.L. Walker 1997. Approximating Nature's variation: Selecting and using reference information in restoration ecology. Restoration Ecology 5: 338-349. White, T. 1999b. A community perspective of ecosystem rehabilitation success. Pages 65-70 in C.J. Asher and L.C. Bell, editors. Proceedings of the Workshop on Indicators of Ecosystem Rehabilitation Success, Melbourne. Australian Centre for Mining Environmental Research, Brisbane. Wiens, J.A., G.D. Hayward, R.S. Holthausen, and M.J. Wisdom 2008. Using surrogate species and groups for conservation planning and management. BioScience 58: 241- 252. Wiggins, N.L., C. McArthur, S. McLean, and R. Boyle 2003. Effects of two plant secondary metabolites, cineole and gallic acid, on nightly feeding patterns of the common brushtail possum. Journal of Chemical Ecology 29: 1447-1464. Wiggins, N.L., C. McArthur, N.W. Davies, and S. McLean 2006. Behavioral responses of a generalist mammalian folivore to the physiological constraints of a chemically defended diet. Journal of Chemical Ecology 32: 1133-1147. Wigley, T.B., and M.K. Johnson 1981. Disappearance rates for deer pellets in the southeast. The Journal of Wildlife Management 45: 251-253. Williams, P.H., N.D. Burgess, and C. Rahbek 2000. Flagship species, ecological complementarity and conserving the diversity of mammals and birds in sub-Saharan Africa. Animal Conservation 3: 249-260. Wilson, B.A., D. Robertson, D.J. Moloney, G.R. Newell, and W.S. Laidlaw 1990. Factors affecting small mammal distribution and abundance in the Eastern Otway Ranges, Victoria. Pages 379 - 396 in D.A. Saunders, A.J.M. Hopkins and R.A. How, editors. Australian Ecosystems: 200 years of utilization, degradation and reconstruction, Proceedings of the Ecological Society of Australia. Surrey Beatty & Sons Pty Ltd, NSW.

- 294 -

Wilson, G.J., and R.J. Delahay 2001. A review of methods to estimate the abundance of terrestrial carnivores using field signs and observation. Wildlife Research 28: 151- 164. Wilson, I.H. 1999. A government view of indicators of ecosystem rehabilitation success. Pages 71-82 in C.J. Asher and L.C. Bell, editors. Proceedings of the Workshop on Indicators of Ecosystem Rehabilitation Success, Melbourne. Australian Centre for Mining Environmental Research, Brisbane. Winterhalder, K., A.F. Clewell, and J. Aronson 2004. Values and science in ecological restoration - a response to Davis and Slobodkin. Restoration Ecology 12: 4-7. Wintle, B.A., M.A. McCarthy, K.M. Parris, and M.A. Burgman 2004. Precision and bias of methods for estimating point survey detection probabilities. Ecological Applications 14: 703-712. Wintle, B.A., R.P. Kavanagh, M.A. McCarthy, and M.A. Burgman 2005. Estimating and dealing with detectability in occupancy surveys for forest owls and arboreal marsupials. The Journal of Wildlife Management 69: 905-917. Withey, A. 2010. Heavy rain soaks south-east Queensland. Available from http://www.abc.net.au/news/stories/2010/02/16/2820535.htm (accessed 30/07/2010). Woodward, W., W.A. Ellis, F.N. Carrick, M. Tanizaki, D. Bowen, and P. Smith 2008. Koalas on North Stradbroke Island: Diet, tree use and reconstructed landscapes. Wildlife Research 35: 606-611. World Bank and the International Finance Corporation 2002. Mining and development, it’s not over when it’s over: Mine closure around the world. International Finance Corporation, Washington, DC. Worton, B.J. 1989. Kernel methods for estimating the utilization distribution in home- range studies. Ecology 70: 164-168. Worton, B.J. 1995. A convex hull-based estimator of home-range size. Biometrics 51: 1206-1215. Wunderle, J.M. 1997. The role of animal seed dispersal in accelerating native forest regeneration on degraded tropical lands. Forestry Ecology and Management 99: 223- 235. Wykes, B.J. 1985. The Jarrah forest avifauna and its re-establishment after bauxite mining. WAIT School of Biology Bulletin 11: 1-164.

- 295 -

Young, T.P. 2000. Restoration ecology and conservation biology. Biological Conservation 92: 73-83. Young, T.P., J.M. Chase, and R.T. Huddleston 2001. Community succession and assembly. Ecological Restoration 19: 5-18. Zoidis, A.M., and H. Markowitz 1992. Findings from a feeding study of the koala (Phascolarctos cinereus adustus) at the San Francisco Zoo. Zoo Biology 11: 417- 431. Zuur, A., E. Ieno, N. Walker, A. Saveliev, and G. Smith 2009. Zero-truncated and zero- inflated models for count data. Pages 261-293 in A. Zuur, E. Ieno, N. Walker, A. Saveliev and G. Smith, editors. Mixed Effects Models and Extensions in Ecology with R. Springer, New York.

- 296 -