Factors determining the spread and impact of the exotic grass Indian couch (Bothriochloa pertusa) into native ecosystems Gabrielle Lebbink Bachelor of Science Honours 1

A thesis submitted for the degree of Doctor of Philosophy at The University of Queensland in 2020 School of Biological Sciences Abstract

The invasion by exotic plant species into native plant communities is a major threat to biodiversity and ecosystem function across the globe. In many cases, their widespread establishment and spread is facilitated by their deliberate introduction for pasture, lawn and soil conservation. This thesis discusses the impact, spread and management of invasive pasture species within Queensland, Australia, with a particular focus on the exotic grass species Bothriochloa pertusa (Indian couch). Bothriochloa pertusa was introduced in 1939 and in recent decades its spread and observed dominance throughout the landscape has raised concerns about the impacts of this species on native biodiversity. I investigated the impacts of B. pertusa on floristic diversity and discuss these results in the context of Queensland’s recent and rapid intensification of land-use.

In Chapter 2, I used comprehensive mapping data and results from field surveys to investigate changes in land-use and the cover of prominent invasive species over time (1996 – 2018), within a focus area of Queensland. I found that despite policy developments aimed at reducing land-clearing and protecting threatened ecosystems during this time, land-clearing still occurred at a rate of 36,960 ha per year in the study area, mainly for the expansion of grazing and cropping agricultural systems. Remaining fragments of native vegetation experienced a significant increase in the cover and occupancy of the invasive pasture species, B. pertusa and Cenchrus ciliaris (buffel grass). I discuss the role of propagule pressure altered disturbance regimes and habitat loss in promoting the spread of these species in Queensland. The results from this chapter highlight the significant and ongoing impacts of land-use change for native biodiversity. Further they highlight the need to implement consistent, effective and well-regulated policies to protect species and habitats in the future.

In Chapter 3, I present a comprehensive account of the introduction history of B. pertusa and discuss factors associated with its spread throughout the landscape. Using B. pertusa presence/absence data I also built habitat suitability models, to predict its potential distribution and cover within Queensland. Results from this chapter highlight climatic extremes and livestock grazing as key mechanisms facilitating the establishment and spread of B. pertusa. In Chapter 4, I used fence-line comparisons, within two conservation reserves and their neighbouring pastoral properties, to further assess the role of grazing in facilitating

ii the spread of B. pertusa. Within both reserves and across the three vegetation types examined, the cover of B. pertusa was higher in the grazed pastoral property than the conservation reserve. Diversity did not differ significantly between the pastoral properties and reserves, although this result was likely confounded by high C. ciliaris cover within the un-grazed reserve. Results suggest grazing protected areas and conservative stocking rates can reduce the spread and impact of B. pertusa within Queensland and elsewhere in Australia. Other invasive species however, such as C. ciliaris may be favoured by grazing protection. In Chapter 5, I comment on the use of controlled livestock grazing to manage the invasive grass Cenchrus ciliaris and discuss the implications of this for the spread of B. pertusa.

In Chapter 6, I investigated the impact of B. pertusa on native floristic diversity in the iron- bark woodlands of sub-coastal Queensland, a habitat particularly vulnerable to invasion. Specifically, I assessed its impact across multiple spatial-scales and found that diversity was lower at both small and large spatial scales in invaded areas. By examining changes to root- traits with increasing invader cover we found that competition for belowground space and resources was a likely mechanism contributing to its broad-scale impact on diversity. Results from these two chapters highlight the capacity for B. pertusa to continue to spread and intensify in Queensland, with the potential for significant and large-scale declines in floristic diversity.

In Chapter 7, I bring together my results and those of previous studies, to provide recommendations and discuss strategies to reduce the impacts of B. pertusa, but more broadly the impacts of invasive pasture species and land-use change. With the majority of land-used for livestock grazing, ensuring the implementation of adaptive and conservative grazing strategies is crucial for maintaining ecosystem resilience and improving resistance to invasion and stochastic events. Further, encouraging the protection and restoration of habitat by ensuring its economic viability for producers is essential and there is considerable scope for this to be achieved (i.e. premium carbon credits, environmental stewardship). For the successful application of these initiatives however, they must be coupled with ongoing research and adequate financial, technical and administrative support.

iii

This thesis examined the spread and impact of invasive pasture species and in doing so highlighted the interactions between land-use change, the spread of invasive species and biodiversity decline. The findings presented here suggest the current approach to land-use is not compatible with the conservation of native species and habitats, and considerable changes are necessary if we are to ensure their persistence in the future.

iv

Declaration by author

This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, financial support and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my higher degree by research candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis and have sought permission from co- authors for any jointly authored works included in the thesis.

v

Publications included in this thesis No publications included

vi

Submitted manuscripts included in this thesis

Lebbink, G., Dwyer, J., Fensham, R. J. (2020). ‘Invasion credit’ after extensive land-use change: An example from eastern Australia. Journal of Environmental Management (in review).

Lebbink, G., Dwyer, J., Fensham, R. J. (2021). An invasive grass species has both local and broad scale impacts on diversity: Potential mechanisms and implications. Journal of Vegetation Science 32:e12972.

Lebbink, G., Dwyer, J., Fensham, R. J. (2021). Controlled livestock grazing for conservation outcomes in a fragmented landscape. Ecological Management and Restoration 22:5-9.

Other publications during candidature

Peer-reviewed papers

Lebbink, G., Fensham, R. & Cowley, R. (2018) Vegetation responses to fire history and soil properties in grazed semi-arid tropical savanna. The Rangeland Journal 40:271-285.

Conference abstracts

September 2017 ‘Vegetation responses to fire history and soil properties in grazed semi-arid tropical savanna’ presented at Australian Rangeland Society conference, Port Augusta

November 2018 ‘The impact of an invasive perennial grass (Bothriochloa pertusa) on floristic diversity across multiple spatial scales’ presented at Ecological Society of Australia conference, Brisbane

June 2019 ‘Invasive perennial grass has both small- and large-scale effects on diversity’ presented at International Congress of Community Ecology, Bologna, Italy

vii

August 2019 ‘The impact of an invasive perennial grass, Bothriochloa pertusa on floristic diversity across multiple spatial scales’ presented at Queensland Herbarium seminar, Brisbane

September 2019 ‘Modelling the current and potential spread of exotic grass Bothriochloa pertusa throughout sub-coastal Queensland’ presented at Australian Rangeland Society Conference, Canberra

July 2019 ‘Less lawn, more grass’ presented at the University of Queensland Three-Minute Thesis, Faculty of Science Finals.

Contributions by others to the thesis

The multidisciplinary nature of this thesis has warranted input from a broad spectrum of academics and government agencies. The most important contributions are listed below.

The Queensland Department of Agriculture and Fisheries provided B. pertusa presence data from their Q-graze long-term monitoring sites. These points were used in the Chapter 3.

The Queensland Herbarium provided B. pertusa presence and absences data from their CORVEG and other long-term monitoring sites.

Peter O’Reagain, John Bushell and the department of primary industries provided the data for Figure 3-1.

viii

Statement of parts of the thesis submitted to qualify for the award of another degree No works submitted towards another degree have been included in this thesis.

Research Involving Human or Animal Subjects No animal or human subjects were involved in this research.

ix

Acknowledgements

Wowee…this has all been quite the adventure and one which has been sustained by the generosity, humour, knowledge of family, friends and colleagues.

The adventure crew, the Appel St and Victoria st mob who I have lived with these past few years and shared many an amazing meal, a walk in the bush, a camp in a creek bed or in the ocean, a climb up a mountain, a sunset, a sunrise and disco ball lit boogies. You are such a big beautiful group of humans which I am so fortunate to know. Your support has been unwavering throughout this big ol’ couchy adventure and I couldn’t have done it without you, so thank you!

My family, have provided unfailing love and support has motivated and inspired me my whole life and this journey was no exception. Mum, your regular check ins (making sure I was eating well, sleeping, not drinking too much) made sure I was keeping healthy and happy. You were always there when I needed to complain or cry and always motivated me to keep going and not to give up. Dad you have been not only been an integral part to this research through supporting me in the field but by always encouraging me to try my best in everything I do you have been helped me to become a competent and skilled scientist. Our time in field together has taught me many skills - we would win the competition for fastest tarp set up for sure.

My amazing colleagues and friends from all over but particularly the University of Queensland, The Queensland Herbarium, the Department of Agriculture and Fisheries. Peter O’reagain and John Bushell, working with you both on the wambianas grazing trail was an amazing and rewarding experience, which I owe much of my plant knowledge to! Thanks to the soils lab in the eco-science precinct for their assistance with particle size measurement. The Queensland Herbarium, I spent many a day behind your walls and was always greeted and helped where needed by several friendly quirky, extremely knowledgeable botanists. The plant identification for this thesis would not have been possible without the resources, skills and knowledge readily shared by the people at the Herbarium.

x

Renee Rossini, ever since we first met when you were my tutor for 3rd year outback ecology, you have inspired me with your wisdom and knowledge about the natural world. You have supported this journey in countless ways; helping in field, working through research ideas and always offering to read and edit my writing! You are a skilled ecologist and all-round amazing human and I am so thankful for your friendship!

Rachael Collett, I am so grateful you decided to approach me in the hall at uni that day; within a five-minute conversation we had decided to set off to the beach and go camping for the weekend, having probably only talked to each other once or twice before. We have shared many a crazy unforgettable adventure together since then and these have kept me sane and productive during these last few years. I am looking forward to many more adventures together!

Thanks to all the people who shared their homes during the writing phase of this thesis. Pete and Kel along the mighty Mary river in Conondale – the river is in good hands with you at its door. I wouldn’t have made it through these last few months without your amazing support and friendship (those covid cookups and foot spa days). Renee Rossini and Jen Silcock your beautiful mountain homes on Jiniburra country were crucial to many stages of this thesis – thank you! Other special places; Mawair (the Brisbane river), the Enogerra reservoir, Yaroomba, the Mary River and the D’Aguilar national park; running, reading and writing by your side has kept me sane during this big adventure. Also, to the many landholders who let me scout around for couch and camp on their properties. Thank you for sharing your stories and knowledge of the land, these have been integral to understanding the implications of this thesis. Your passion for caring for country is obvious and this gives me hope we can tackles some of the tricky management problems outlined in this thesis.

Finally, my supervisors, Rod Fensham and John Dwyer. We make a pretty good team I reckon! Rod, from traipsing around lost in mulga on outback ecology to measuring couch in the back block of central Queensland you have been a continual source of inspiration, entertainment and knowledge. Your support has been unwavering throughout my whole academic journey and I have gained so much knowledge and learnt so many skills as a result. John, your structured and calculated approach to everything you do has helped to keep this thesis on track! You have taught me so much about statistics, writing, experimental design

xi and how to run a straight tape and these skills have been invaluable to the outcomes of this thesis. I am also so grateful to have been able to teach with you and have learnt much knowledge and skills through these experiences.

xii

Financial support This research was supported by an Australian Government Research Training Program Scholarship.

Field work was supported The Holsworth Wildlife Research Endowment Grant (The Ecological Society of Australia) and the Reef Guardian Research Grants 2018 (Grant number: rgrg1800002).

Keywords Cenchrus ciliaris, competition, environmental policy, fire-sensitive vegetation, grasslands, invasive grass, invasive species management, land-use change, livestock grazing

xiii

Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 050103 Invasive Species Ecology, 45 % ANZSRC code: 050202 Conservation and Biodiversity, 30 % ANZSRC code: 060208 Terrestrial Ecology, 25 %

Fields of Research (FoR) Classification

FoR code: 0602, Ecology, 50% FoR code: 0501, Ecological Applications, 30%

xiv

Table of Contents Chapter 1 – Introduction ...... 1

Process and mechanisms of invasion ...... 1 Impacts of invasion ...... 3 Invaders in Australia ...... 4 Bothriochloa pertusa (indian couch) ...... 5 Outline of thesis ...... 6 Chapter 2 – ‘Invasion debt’ after extensive land-use change: An example from eastern Australia...... 10

Introduction ...... 10 Methods ...... 15 Results ...... 21 Discussion ...... 28 Chapter 3 – The ‘lawnification’ of Queensland’s grassy woodlands: Mapping the past, current and likely future extent of the exotic grass Bothriochloa pertusa ...... 36

Introduction ...... 36 Methods ...... 37 Results and Discussion ...... 41 Conclusions ...... 52 Chapter 4 – Damned if you do, damned if you don’t: Variable effect of grazing and vegetation type on the success of two invasive grass species ...... 54

Introduction ...... 54 Methods ...... 56 Results ...... 61 Discussion ...... 63 Chapter 5 – Controlled livestock grazing for conservation outcomes in a fragmented landscape ...... 68

Introduction ...... 68 Restoring and conserving fire-sensitive vegetation in Queensland ...... 70 Conclusion ...... 74 Chapter 6 – An invasive grass species has both local and broad-scale impacts on diversity: Potential mechanisms and implications ...... 77

Introduction ...... 77 xv

Methods ...... 81 Results ...... 86 Discussion ...... 90 Chapter 7 – Discussion ...... 96

Slowing down the ‘lawnification’– perspectives on managing Bothriochloa pertusa ..... 96 A balancing act: approaches to conservation on private land ...... 100 Conclusion ...... 104 References ...... 105

Appendices ...... 127

Appendix 2-1 – Detailed results from landscape scale data validation analysis used in Chapter 2...... 127 Appendix 3-1 – Details and methods for data sources used in Chapter 3 to model the B. pertusa’s current and potential distribution...... 129 Appendix 3-2 – Habitat suitability model methods ...... 132 Appendix 4-1 – Methods for soil particle size analysis...... 134 Appendix 5-1 – Extent of fire-sensitive ecosystems within conservation reserves and their fire frequency ...... 135

xvi

List of Figures

Figure 1-1. Photographs of Bothriochloa pertusa (indian couch). …………………………7 Figure 2-1. The distribution of remnant vegetation in 2017...... 24 Figure 2-2. The total number of sites each invader (B. pertusa (a), C. ciliaris (b) and P. hysterophorus (c)) ...... 28 Figure 3-1. Change in the frequency of B. pertusa (black line) and its native congener B. ewartiana (blue line) over time...... 45 Figure 3-2. Total B. pertusa presence records within 30 km2 raster grid cells ...... 46 Figure 3-3. Partial dependence plots for the four most influential variables in the model for B. pertusa occurance (presence/absence) ...... 50 Figure 3-4. Partial dependence plots for the four most influential variables in the model for B. pertusa cover ...... 51 Figure 3-5. Probability of occurrence (presence/absence) (left) and predicted broad cover (right) of B. pertusa in Queensland...... 52 Figure 4-1. a) The location of Homevale and Dipperu National Parks in Queensland, Australia...... 60 Figure 4-2. The mean cover of (a) B. pertusa and (b) C. ciliaris in grazed (green) and long- ungrazed (blue) plots...... 62 Figure 6-1. Examples of uninvaded (solid) and invaded (dashed) ...... 81 Figure 6-2. (a) The location of the 41 plots (circles) and the property where root cores were collected ...... 84 Figure 6-3. (a) observed SARs for all plots based on total species richness and ...... 87 Figure 6-4. Relationships between plot-scale B. pertusa cover and the estimated SAR intercepts and SAR slopes ...... 88 Figure 6-5. Relationships between plot-scale B. pertusa cover and the richness of rare (a) and common (b) species in the community ...... 89 Figure 6-6. Relationships between B. pertusa cover and (a) SRL, (b) RTD, (c) root biomass and (d) total shoot biomass ...... 90

xvii

List of Tables

Table 2-1. The dominant broad vegetation groups and their preclearing extent (km2) within the study area...... 16 Table 2-2. The dominant broad vegetation groups within the catchment area and the percent remnant remaining in 1997 and 2020...... 25 Table 2-3. Total remnant vegetation cleared and recovered between 2017 and 2020 for each land-use category ...... 26 Table 2-4. Total number of sites cleared (remnant to non-remnant), recovered (non-remnant to remnant), ...... 27 Table 3-1. Total number of B. pertusa presence, absence and cover records collated from a variety of data sources between 1941 – 2019 ...... 39 Table 3-1. Relative importance of environmental variables to models predicting B. pertusa occurrence (presence/absence) and cover ...... 49 Table 4-1. Difference in mean percentage clay, silt and sand between the three surveyed vegetation types ...... 57 Table 4-2. Total number of paired line-transects and floristic plots sampled in each national park and vegetation type...... 59

List of Abbreviations

DTW – Distance to waterway EPBC – Environmental Protection and Biodiversity Act FPC – Foliage protective cover LUC – Land-use change MGSR – Mean growing season rainfall MGST – Mean growing season temperature RTD – Root tissue density RVI – Rainfall variability index SAR – Species area relationships SRL – Specific root length

xviii

Chapter 1 – Introduction

For centuries, humans have assisted the dispersal of plant species across the globe, either deliberately for food, resources or amenity, or accidently through the movement of people and goods (Boivin et al. 2017). This has usually occurred with little regard or understanding of the consequences for recipient native communities (Cook and Dias 2006b). Today, the potential and realised impacts of exotic plant species are well-recognised, with many becoming invasive and disrupting ecosystem function (Driscoll et al. 2014). Due to their sometimes-extreme consequences for biodiversity and ecosystem processes, managing the impact and spread of invasive species is a priority for land managers across the globe (Butchart et al. 2010, Maxwell 2016). To inform management and improve efficacy of control strategies, an understanding of the invasion process and the mechanisms which facilitate the spread and impact of invasive species is crucial.

Process and mechanisms of invasion

The fundamental process of invasion can be simplified into three stages; arrival, establishment and persistence, and spread (Mack et al. 2000, Shea and Chesson 2002, Lockwood et al. 2005). The environmental conditions of the site, in addition to competition and disturbance are key factors which influence the success of the invader at each stage (Fargione et al. 2003, Tilman 2004). Disentangling the relative importance of these factors in facilitating the dominance of invasive species is a key focus of invasion ecology and necessary to understand in order to guide their appropriate management (MacDougall and Turkington 2005).

Competition is an important determinant of community assembly and species coexistence (Tilman 2004). In many invasion scenarios, the superior ability of the invader to compete for resources has facilitated their dominance and subsequent impact on native populations (Seabloom et al. 2003, Funk and Vitousek 2007, Schmidt et al. 2007, Rossiter - Rachor et al. 2009). The competitive superiority of some invaders has been demonstrated using manipulative field and glasshouse trails which examine differences in resource use efficiency between co-occurring native and invasive species (Seabloom et al. 2003, Ens et al. 2015, Broadbent et al. 2018). For instance by measuring the resource use of native and invasive 1 plant species across three habitats in Hawaii, Funk and Vitousek (2007) found that invasives were generally more efficient at using limited resources, and depleted these resources quicker than natives.

More commonly, resource competition between species is inferred by analysing the functional composition of a community and comparing the functional traits of invasives with those of natives (Funk and Wolf 2016). Functional traits can describe the efficiency of resource acquisition or use and when coupled with an understanding of environmental conditions can be used to understand competitive hierarchies among species (Firn et al. 2010). High-impact invasive species commonly possess novel traits which confer a competitive advantage over natives. For example, the ability of Australian Acacia species to attract nitrogen fixing bacteria has allowed them to invade low-nutrient coastal dunes in Portugal (Rodríguez-Echeverría et al. 2009). More commonly, the traits of invasive species differ quantitatively to traits already present in existing native communities. For instance, the invasive perennial grass Andropogon gayanus (gamba grass) has far greater root biomass and depth than native perennial grass species and this is associated with its higher and more efficient uptake of nitrogen (Rossiter - Rachor et al. 2009). This has likely been a key factor facilitating its invasion into the nutrient poor savannas of northern Australian. Species which become invasive also tend to have greater trait plasticity than resident species, which can improve their performance and adaptability to environmental change and disturbance (Funk 2008, Turcotte and Levine 2016).

The importance of disturbance in facilitating the establishment and spread of invasive species is well-established (MacDougall and Turkington 2004). In particular, changes to disturbance regimes, from those which natives are well-adapted, can reduce the success of natives and favour the establishment of opportunistic invasive species (Hobbs and Huenneke 1992). New and altered disturbance regimes are a key consequence of human land-use and this has been associated with the spread and impact of many invasive species across the globe (DeGasperis and Motzkin 2007, Chytrý et al. 2008, Mosher et al. 2009, Pyšek et al. 2010, Vilà and Ibáñez 2011). For example, in California, intense grazing regimes combined with periods of drought, led to the replacement of 9.2 million ha of perennial grasslands by annual exotic grass species (Seabloom et al. 2003). Even in areas were grazers have now been removed, native perennial grasses have failed to recover, despite using resources more efficiently than exotics. In this

2 instance, as well as in other invasion scenarios (Foster and Tilman 2003, Corbin and D' Antonio 2004), recruitment limitation is reducing the capacity for native species to recover. Anthropogenic disturbances are often coupled with extensive and repeated introductions of exotic species and this has been implicated in the exacerbated spread of some invasive species (Lockwood et al. 2005), such as the invasive pasture species Cenchrus ciliaris (buffel grass) in Australia (Fensham et al. 2013) or the invasive vine, Pueraria montana (Kudzu) used for soil conservation in United States (Forseth and Innis 2004).

In most cases, invader dominance and subsequent impact is determined by an interacting array of factors relating to competition, disturbance, environmental conditions and propagule pressure and the relative importance of these factors varies with time since invasion and invasion stage. For example, long-term suppression of fire in the garry oak (Quercus garryana) savannas of south-western Canada, led to the establishment and maintained dominance of invasive perennial grass species, Poa pratensis and Dactylis glomerata (MacDougall and Turkington 2004). Their negative impact on native plant communities however is associated with their superior ability to compete for light, which in this system is limited. Re-introducing fire, to help reduce the competitive superiority of dominants and increase the abundance of sub-ordinate native species, has been suggested as a possible restoration strategy. Habitat loss and competitive exclusion however has resulted in the recruitment limitation of many native taxa and this is thought to limit their capacity to recover. Examples such as this and those discussed previously highlight the need to evaluate plant invasions within the context of the broader environment as to help disentangle the interacting drivers of invader dominance.

Impacts of invasion

Aside from the direct impacts to species diversity and composition through competition, the dominance of some invasive species can have dramatic consequences for the structure and function of ecosystems, sometimes leading to irreversible ecosystem transformation (Levine et al. 2003). This often occurs when invaders introduce novel or extreme trait values (Richardson et al. 2000). For example, woody species which invade naturally treeless ecosystems significantly alter vegetation structure and consequentially reduce species diversity by increasing competition for light (Griffin et al. 1989, Firn et al. 2008).

3

Alternatively, novel strategies of resource acquisition, such as nitrogen fixation, can drastically alter nutrient dynamics and facilitate changes to species composition (Holmes and Cowling 1997, Miller 2004). Similarly, some invaders alter hydrological cycling by altering evapotranspiration rates (Zavaleta 2000), run-off (Koci et al. 2020), or soil water availability (Dyer and Rice 1999, Ens et al. 2015). In the fynbos of South-Africa, the continued invasion by exotic pine trees is reducing catchment water supply, with signification implications for biodiversity, including the residing human population (Le Maitre et al. 1996). The impact of invasive flora on fire regimes has received considerable attention in the invasion literature (D’Antonio 2000, Levine et al. 2003, Grice 2006). In particular, there are many examples of increased fire frequency and intensity as a result of invasion from high- biomass exotic grass species (D'Antonio and Vitousek 1992). Where they invade wooded ecosystems, this can sometime result in their conversion to grassland (D'Antonio and Vitousek 1992, Keeley et al. 2005). Even in fire-dependent ecosystems, such as the savannas of northern Australia increased frequency and intensity of fire due to the invasion of A. gayanus is reducing the success of woody recruitment and increasing stem mortality (Setterfield et al. 2010). In the case of A. gayanus and some other invasive grass species (Miller et al. 2010), a positive feed-back cycle is established, whereby fire reduces woody- recruitment which further promotes invasive grass the establishment and increases fire risk.

Invaders in Australia

As with most invaders across the globe, the introduction of invasive plant species in Australia largely occurred as a result of deliberate introductions for agricultural, ornamental or amenity purposes (Cook and Dias 2006b). Managing the environmental, agricultural and social impacts of these invasive species is a national priority, with their management estimated to cost 4.9 billion per year to the agricultural industry alone. According to the Australian Weed Strategy 2017 – 2027, around 500 of the estimated 3207 naturalised weed species are declared noxious and under some form of legislative control, including 32 which are listed as a weed of national significance. For the purpose of this strategy a ‘weed’ is considered pragmatically as a plant that requires some form of action to reduce its negative effects on the economy, the environment, human health and amenity.

4

Unfortunately, many of the most damaging invasive species for biodiversity are not controlled and sometimes deliberately established over large areas, due to their benefits for pastoral productivity (Cook and Dias 2006b, McAlpine et al. 2009, Marshall et al. 2011, Driscoll et al. 2014). The exotic perennial grass C. ciliaris (buffel grass) is a typical example, whereby despite its devastating impacts for biodiversity being well-known, the species continues to be deliberately established across vast tracts of land, often in conjunction with broad scale clearing of native woody species (Fairfax and Fensham 2000, Butler and Fairfax 2003, McAlpine et al. 2009). Several other invasive pasture species have been introduced and have naturalised across Australia, often with similar consequences for native biodiversity. Reducing the impact of these often wide-spread and well-established invasive species is challenging, particularly within highly modified landscapes, with conflicting management priorities. With exotic pasture introductions expected to increase and the distributions of existing invasive pasture species expected to change in response to climate change (Jackson and Sax 2010, Driscoll et al. 2014), understanding the factors influencing their spread and impact on native communities is crucial.

This thesis investigates the impact, spread and management of the invasive pasture species Bothriochloa pertusa (Indian couch). More broadly I critically assess and discuss the challenges associated with managing invasive pasture species and conserving biodiversity within modified landscapes.

Bothriochloa pertusa (indian couch)

Bothriochloa pertusa is a perennial grass species native to south-east Asia and India. It has also naturalised in regions of north and south America, Africa and Mexico (Bisset 1980). Although established throughout Queensland for several decades (see Chapter 3 for introduction history), the dominance and rapid spread of B. pertusa across large areas of Queensland in recent years has raised concerns over its impact for both conservation and pastoral imperatives (Spiegel 2016).

B. pertusa does not produce large amounts aboveground biomass but forms a dense mat of stolons (Figure 1-1), that allows for a much more continuous ground cover of culms than is typical of tussock grasses (McIvor and Gardener 1994). The network of stolons renders B.

5 pertusa resistant to grazing, and it spread and dominance in the landscape is thought to be facilitated by livestock grazing, (McIvor 2007a; McIvor et al. 1982). Because of its grazing tolerance and its efficient colonising ability, B. pertusa is sometimes regarded as a valuable fodder (Walker and Weston 1990, Jones 1997). Discussions with producers (opportunistically and at producer information sessions) and producer surveys (Spiegel 2016) however suggest there is concern over the dominance of this species, particularly where it displaces more productive species, such as its native congener Bothriochloa ewartiana (desert blue grass) or the African species, C. ciliaris (buffel grass).

Bothriochloa pertusa has been associated with significant declines in the diversity of native plant species (Kutt and Kemp 2012). In uncleared Queensland rangelands, Kutt and Kemp (2012) examined the response of native species to increasing cover of the exotic B. pertusa and the native B. ewartiana and found the number of native species declined with increasing B. pertusa cover but remained stable with increasing B. ewartiana cover (See Figure 1-1c for photo of B. pertusa invading B. ewartiana grassland). This study also reports on a significant negative relationship between B. pertusa cover and total ground cover, forbs and perennial grass richness and cover and the cover of nine native perennial grasses. Because B. pertusa produces less biomass than many native species, its invasion has led to dramatic changes to understorey canopy structure and associated shifts in faunal assemblages (Kutt and Fisher 2011). Although providing some ground cover during good seasons, B. pertusa is less drought tolerant than most native species and its dominance has been associated with increased run-off and soil erosion (Bastin et al. 2008, Bartley et al. 2014, Koci et al. 2020).

Outline of thesis

The interaction between land-use change and the spread of invasive species is well- established (Chytrý et al. 2008, Pyšek et al. 2010). Over the last 60 years, large areas of Queensland, have undergone significant land transformation, to allow for the expansion of agricultural, mainly livestock, production systems. Managing the spread and impact of invasive species within such modified landscapes is challenging. To appreciate these complexities, in Chapter 2 I use comprehensive vegetation and land-use mapping to assess temporal trends in land-use within central Queensland, with a focus on the last 20 years (during which the spread of B. pertusa has been rapid). Within the same area I also examine

6

Figure 1-1 Photographs of Bothriochloa pertusa (indian couch). a) B. pertusa seed head; b) B. pertusa stolons; c) B. pertusa (in foreground) invading Bothriochloa ewartiana (larger tussock in background) dominant grassland near Gayndah, central Queensland; d) B. pertusa monoculture in Eucalyptus orgadophila (Mtn. coolabah) woodland in near Nebo, central Queensland. All photographs taken by Gabrielle Lebbink. 7 changes to the cover and occupancy of three prominent invasive species (B. pertusa, C. ciliaris and Parthenium hysterophorus) using site-scale data collected in 1996 and 2018. The results from this chapter highlight ‘invasion debt’ as a major consequence of land-use change that will likely exacerbate the already extreme loss of biodiversity due to habitat loss and fragmentation.

Having a thorough understanding of the invasion history is important for understanding the factors associated with their spread (Müllerová et al. 2005, Puth and Post 2005, Vilà and Ibáñez 2011).

In Chapter 3 I use presence/absence records alongside information obtained from the literature and discussions with land holders to present a comprehensive account of the introduction history of B. pertusa and discuss factors related to its spread. Using this data, I also build habitat suitability models to predict and map the potential distribution and cover of B. pertusa across Queensland. By identifying where and how B. pertusa invades, the findings from this chapter will help to inform and prioritise the management of B. pertusa in Queensland and elsewhere in Australia.

Findings from Chapter 3 suggest the spread of B. pertusa has been facilitated by livestock grazing. To examine this hypothesis, I measured differences in B. pertusa cover between two long-ungrazed conservation reserves and their neighbouring pastoral properties (Chapter 4). These findings suggest grazing protected areas are more resilient to invasion by B. pertusa but are sometimes more vulnerable to invasion by another invasive perennial grass, C. ciliaris. Indeed, grazing is sometimes suggested as a useful tool for managing the impact of C. ciliaris and some other high-biomass exotic grass species. In Chapter 5, I comment on use of grazing as a management tool for C. ciliaris and discuss implications of this for the spread of B. pertusa. Given the dominance of these two invaders across vast areas of Queensland, managing their impact on native communities represents a major challenge for conservation managers.

B. pertusa invasion has been associated with declines in native flora species at the local scale (Kutt and Kemp 2012). The impact of invasive species can vary with spatial scale however, with some invaders having very large impacts at small scales but progressively smaller

8 impacts as scale increases (Powell et al. 2013). To accurately assess and predict invader- induced biodiversity decline, it is important to examine impact-scale relationships. Further, identifying the traits associated with broad-scale impact is important for prioritising invasive species management. In Chapter 6, I assess the impacts of B. pertusa on floristic diversity across multiple spatial scales in the iron-bark woodlands of Queensland. I also examine changes to community root traits across an invasion gradient so as to evaluate belowground competition as a mechanism contributing to broad scale invader impact.

This thesis highlights the challenges and complexities of managing invasive species and conserving biodiversity within modified landscapes, with conflicting management priorities. These problems are not unique to Australia and represent a major challenge for conservation across the globe. In Chapter 7, I critically examine some of the approaches taken to maintain biodiversity alongside agricultural imperatives thus far. In doing so I highlight areas for improvement, recommend new approaches and suggest avenues for future research.

9

Chapter 2 – ‘Invasion debt’ after extensive land-use change: An example from eastern Australia

Abstract

Land-use change, and associated land clearing/conversion and fragmentation are major drivers of biodiversity decline across the globe. The spread of invasive species is a well- recognised consequence of land-use change. The extent and intensity of invasion however is often difficult to assess due to a lack of temporal data. Using detailed mapping information for 130, 950 km2 of sub-coastal Queensland, Australia and results from field surveys we investigated changes to land-use, the extent of remnant (intact) vegetation and the spread of prominent invasive plant species over time (1997 – 2018). In the 50 years prior to 1997 the area underwent significant land development (mostly for livestock grazing and crops), resulting in a reduction of 45% of its remnant vegetation. Despite key policy developments aimed at protecting the remaining vegetation and species, land clearing/conversion still occurred at a rate of 370 km2 per year between 1997 and 2017, mainly for the expansion of grazing and cropping lands. Vegetation types specifically listed for national protection under these policies were some of the greatest affected, highlighting the need for a major review into the implementation and regulation of these control measures. Within remaining fragments of remnant vegetation, the cover and occupancy of two invaders (out of the three examined), Bothriochloa pertusa (Indian couch) and Cenchrus ciliaris (buffel grass) increased significantly during this time period. The spread of these species within the landscape likely reflects an ‘invasion debt’, incurred from an intense history of land-use within the region and we predict this trend will continue to degrade remnant ecosystems.

Introduction

In recent decades, the rate of land-use change (LUC) across the globe has accelerated as a result of an increasing human population and consumption rate (Myers and Kent 2004). Industrial advancements have also improved our capacity to rapidly modify ecosystems over large spatial scales. It has become increasingly apparent that the current rate of LUC is not sustainable for the conservation of global biodiversity and ecosystem processes, with the extinction and decline of many plant and animal species attributed to LUC and associated

10 habitat loss and fragmentation (Sala et al. 2000, Foley et al. 2005, Brook et al. 2008, Maxwell 2016).

Increasing the extent and profitability of agricultural production systems is a major driver of rapid LUC across the globe (McAlpine et al. 2009). In many areas this has led to the broad scale conversion of native forests into managed grazing lands, as well as the expansion of crops to produce food and resources for humans and livestock. Aside from the destruction and displacement of species, the expansion of agricultural systems and the loss of native forests has been associated with increased carbon emissions, altered hydrological functioning and changes to climate (Asner et al. 2004). The impacts of forest loss are further compounded by fragmentation and edge effects which increase the vulnerability of remaining forest fragments to secondary impacts, such as dispersal limitation, altered fire and nutrient regimes and invasion by exotic species (Vilà and Ibáñez 2011, Chytrý et al. 2012). These impacts often occur over large spatial and temporal scales and their realised consequences for biodiversity may take several decades to eventuate (Kuussaari et al. 2009, Gilbert and Levine 2013).

Disturbance and fragmentation associated with LUC are well recognised drivers of the invasive species spread across the globe (Vilà and Ibáñez 2011). Their spread is further exacerbated by the large-scale development of pastures dominated by exotic species (Driscoll et al. 2014). The extreme propagule pressure and inherent resilience of these large well- established populations facilitate their continued spread throughout the landscape (Mack et al. 2000, Fensham et al. 2013, Warren et al. 2013). Furthermore, the deliberate selection of pasture species with rapid growth rates and tolerance to environmental stress enhances their capacity to outcompete resident native species (Lavergne and Molofsky 2007). Despite the link between LUC and the spread of invasive species being well known, these two key drivers of global biodiversity decline are often examined independently (Domènech et al. 2005).

Queensland, Australia presents an interesting case study for assessing LUC and the spread of invasive species as most land development has occurred relatively recently (within the last 50 years) and has resulted in the conversion of ~ 50% of its forested area into managed agricultural systems (predominantly pasture and cropping) (Wilson et al. 2002b). In addition,

11 detailed monitoring and mapping of Queensland’s vegetation has been conducted and continually improved since 1997, which coincides with the implementation of key government policies to reduce land clearing and protect native vegetation. Evaluating changes to land-use in Queensland in recent decades should provide insights into the success of environmental policies in protecting native species and habitats and may help to inform the development of vegetation management policy in Australia and elsewhere in the world. Furthermore, assessing changes to the cover and occupancy of invasive species within remaining fragments of native vegetation is essential for managing their impact and is useful for predicting the ongoing consequences of LUC for biodiversity.

History of land-use and environmental policy in Queensland

The most dramatic change to occur in Queensland was the conversion of large areas of wooded ecosystems into grassy pasture dominated by exotic grass species. This largely occurred within woodlands dominated by Acacia species, particularly A. harpophylla (brigalow) between 1960 and 1990. Less than 10% of this previously widespread ecosystem now remains (Fensham, et al. 2017). Intact these Acacia woodlands have little production value, however occur on fertile soils and when cleared can support healthy swards of grass (Fairfax and Fensham 2000). The species most frequently sown is Cenchrus ciliaris (buffel grass), a robust perennial native to Africa. Across the landscape generally however many other exotic grass and legume species have been introduced on the premise of improved pastoral production (Cook and Dias 2006b). Native grasslands occurring on fertile clay soils have also experienced significant declines in the region (Fensham 1999), predominantly due to the expansion of sorghum and other grain crops used in livestock feed, but also for cotton. In addition to agriculture, mining, mainly for coal is another major use of land and cause of habitat destruction, with 63% of Australia’s coal resources occurring in Queensland (Australia 2020).

In response to national and international concern over the extreme deforestation occurring in Queensland, from the late 1990s the government progressively introduced legislative measures to help protect ecosystems and reduce vegetation clearing. The implementation of the state legislated Vegetation Management Act 1999 aimed to restrict clearing of remnant woody vegetation, and initially slowed the rates of clearing considerably (Reside et al. 2017).

12

Multiple changes of government and several amendments to the legislation however, allowed clearing to continue at more than 200,000 hectares per year until 2006, when a tightening of restrictions saw a significant and maintained reduction in clearing in Queensland (Kehoe 2009). Another change in government however in 2011 saw the suspension of investigations into illegal clearing and a weakening of clearing restrictions, resulting in another spike in woody clearing between 2012 and 2014.

Another key policy to be implemented during this time and one we will specifically evaluate in this study, was the Environmental Protection and Biodiversity Conservation Act 1999 (EPBC) which enlists threatened species and ecological communities for federal protection. This level of protection means new or intensified activities that may have a significant impact on the listed threatened species or community requires additional impact assessments and approval by the Australian Governments Minister for the Environment. When considering a projects approval, the Minister considers the potential impact on the threatened ecosystem or species as well as the social and economic impacts of the projects.

Goals

Using Queensland’s detailed mapping information and data from floristic surveys we assess changes to land-use, the extent of remnant (intact) vegetation and the cover of invasive species, within a 130, 950 km2 focus area of central Queensland. In 1996 the grassland and grassy woodland ecosystems within this area were subject to detailed floristic surveys (including assessments of invasive species cover) to provide advice on their conservation into the future (Fensham (1999)). These surveys included assessments on the presence and cover of all invasive species, with a particular focus on evaluating the spread and dominance of the invasive perennial grass Cenchrus ciliaris (buffel grass) and the invasive sub-shrub Parthenium hysterophorus (parthenium weed).

In this study we first examine landscape-scale (whole study area) land clearing and LUC using detailed vegetation mapping data from 1997 and 2017 and land-use mapping data from 2017. At the landscape-scale we also assess the effectiveness of the Environmental Protection and Biodiversity Conservation Act 1999 (EPBC) in reducing land clearing/conversion in two threatened vegetation types, native grasslands on fertile soils and

13

A. harpophylla woodlands, which were listed under the EPBC in 2009 and 2001 respectively. Using the same mapping data, we then assess site-scale (at sites established by Fensham 1999) land clearing and LUC. Also, at the site-scale, we assess changes in the abundance and frequency of three focal invasive species C. ciliaris, P. hysterophorus and another invasive grass species, Bothriochloa pertusa (Indian couch), using floristic data collected in 1996 (Fensham 1999) and again in 2018 (for a subset of sites only). All three species are well known for their detrimental impacts on biodiversity and their spread has been correlated with the decline of many species of flora and fauna in the region (Chippendale and Panetta 1994; Fairfax and Fensham 2000; Jackson 2005; Kutt and Fisher 2011; Kutt and Kemp 2012; Nguyen, et al. 2010). Results are discussed and evaluated in the context of Queensland’s land-use and environmental policy history.

We will address the following key research questions:

Landscape scale: 1. How much of and what type of vegetation has been cleared, and for what land-use between 1997 and 2017? 2. How did the rates of clearing in focal threatened vegetation types change after they were listed for protection under the federal Environmental Protection and Biodiversity Act 1999?

Site-scale: 2. Between 1997 and 2017, how many sites have been cleared of remnant (intact) vegetation and for what land-use? 3. Between 1996 and 2018, how has the cover and abundance of the three focal invasive species changed?

We predict trends in land-use and land clearing within the study area assessed here to reflect those of Queensland at large. Specifically, we expect significant clearing of intact vegetation within the study area, and this to be largely associated with the expansion and intensification of agricultural production systems, particularly livestock grazing. Acacia harpophylla woodlands and native grasslands on fertile soils are likely to have been most affected, however we would expect the rates of vegetation clearing within in these systems to have

14 reduced after they were listed under the EPBC in 2001 and 2009 respectively. Disturbance from grazing and vegetation clearing are well known facilitators of invasive alien species and this has been demonstrated for the focal invaders examined here; B. pertusa (Kutt and Fisher 2011), C. ciliaris (Butler and Fairfax 2003), P. hysterophorus (Fensham et al. 1999). We therefore expect the cover and occupancy of these species to have increased within the study area.

Methods

Study area and site locations

The study area extends across 130, 950 km2 of central Queensland. Mean annual rainfall varies from 500 mm in the west to 900 mm in the east (Bureau of Meteorology, 2018) (Figure 2-1). The dominant land types and associated broad vegetation groups and their extent (based on pre-clearing estimates) in the study area are outlined in Table 2-1. Pre-clearing here refers to the vegetation condition pre-European invasion in 1750 (Data Sources for methodology). The A. harpophylla dominant woodlands, which form part of Acacia on clay vegetation group were listed for protection under the Environmental Protection and Biodiversity Act 1999 (EPBC) in 2001. All of the native grasslands on clay within the study area were listed for protection under the Environmental Protection and Biodiversity Act 1999 (EPBC) in 2009.

Sites (established by Fensham (1999) examined in site-scale analyses were well-spread within the study area, however, were restricted to the fertile grasslands or grassy woodland vegetation groups; Eucalyptus on texture contrast soils (n = 22), woodland on alluvium (n = 21), grasslands on clay (n = 107) and Eucalyptus on clay (n = 57) (Figure 2-1). Sites were selected in 1996 to represent intact vegetation community’s, with structure and diversity considered typical of that particular vegetation type.

15

Table 2-1. The dominant broad vegetation groups and their preclearing extent (km2) within the study area.

Total pre-clear Land Type Vegetation, landform, soil description (km2) Eucalyptus on texture- Eucalyptus populnea and E. melanophloia dry woodlands to open contrast soils** woodlands on sandplains or depositional plains. Soils of low-moderate 42669.70 fertility, dispersible sub-soils prone to erosion Acacia on clay Acacia harpophylla (brigalow) (dominant in study region), A. cambagei, A. argyrodendron open forests to woodlands on heavy clay soils. High fertility well-structured soils. Low production value when uncleared, 36207.80 high production value when cleared. A. harpophylla woodlands specifically, listed for protection under the EPBC* in 2001. Eucalyptus on hills Eucalyptus crebra and other E. spp. open forest and woodland on hills. 13720.37 Variable soils, but often rocky and steep. Grasslands on clay** Astrebla spp. (mitchell grass), Dichanthium spp. (bluegrass) tussock grasslands on heavy clay soils. High fertility well-structured soils. High 8205.96 production and cropping value. Listed for protection under the EPBC* in 2009. Woodland on Eucalyptus coolibah woodland and other woodland and open forest in alluvium** frequently flooded situations associated with large watercourses. 7186.69 Generally fertile soils with high production value, although flood-prone. Eucalyptus on clay** Eucalyptus orgadophila and other E. spp. open woodland on level or gently undulating clay soils derived from basalt or shales. Fertile, well- 5695.34 structured soils, although prone to rockiness. Suitable for cultivation in some instances Acacia on sand Acacia shirleyi and A. catenulata open forest on eroded edges of sandstone and on sand sheets. Low fertility, poorly structured soils. Low 5473.20 production value, prone to erosion when cleared. Eucalyptus on sand Dry eucalypt woodlands to open woodlands on sandy plateaus and plains. Mostly E. crebra and Corymbia spp. (bloodwoods) dominant 3632.51 within study area. Low fertility, poorly structured soils

*Environmental Protection and Biodiversity Act 1999 (EPBC)

** Sites (established by Fensham 1999) assessed in site-scale analysis were within these vegetation groups. 16

Spatial data sources

To map changes to land-use and vegetation at the landscape-scale (whole study area) and site-scale we used the Queensland Government’s Vegetation Management (Queensland Herbarium 2017) and Land-Use spatial data series (Queensland Government 2017). These data series use information from satellite imagery, extensive ground truthing, historical data, and consultation with experts to attribute vegetation and land-use attributes to polygons.

Queensland’s vegetation management spatial data series classifies and maps vegetation into broad vegetation groups and defines their distribution on a pre-clearing coverage, ‘remnant’ and ‘non-remnant’ layer (Neldner 2015). The pre-clearing coverage layer uses current and historical data and imagery, to map the distribution and extent of vegetation groups before the modification of land by non-indigenous peoples (in Australia this equates to before European invasion in 1750). The remnant layer maps the distribution of uncleared or intact (see criteria below) vegetation, while the non-remnant layer maps the distribution of cleared vegetation. The term clearing in this paper is used to describe both clearing of trees in wooded ecosystems as well as the loss of understorey vegetation (grasses and forbs) in non-wooded ecosystems. Queensland’s vegetation mapping data series was first made available in 1997 and has been updated approximately every two years to incorporate improvements to mapping and changes to the extent of remnant and non-remnant vegetation.

Vegetation is mapped as remnant if; for wooded vegetation types, the dominant canopy has greater than 70% of the height and greater than 50% of the cover relative to the undisturbed height and cover of that stratum, and is dominated by species characteristic of the vegetations undisturbed canopy; for non-wooded vegetation types, the area has not been cultivated for 15 years, contains native species normally found in that vegetation type and is not dominated by non-native perennial species. In most cases vegetation that is mapped as remnant has not been cleared, and its composition, structure and function, reflective of a pre-clearing condition. However, the definitions described above also allow for vegetation that has been cleared, thinned or cropped in the past, but then substantially regrown and restored, to be classified as remnant. Vegetation that has been classified as non-remnant in the past may therefore return to remnant status given sufficient time to recover.

17

The land-use mapping series classifies Queensland’s land-use according to the Australian Land-use and Management (ALUM) Classification (Version 8). In this study we consolidated the ALUM categories into six broad categories; grazing, cropping, mining, conservation and minimal use, forestry and other intense uses (mostly farm infrastructure/residential). The grazing category includes grazing of both native and modified pasture as these are not well defined in the database. Forestry includes native forests and plantation forests, however land- used for plantations is very minimal within the study area. The main cropping conducted in the region is for grains (mostly sorghum) and cotton. Land-use categories which did not fall into these six categories (including water) contributed to < 1% cover and were included in total area calculations, however, are not discussed further. In this study we used the most current land-use mapping available from 2017.

To assess vegetation clearing and associated land-use change at the landscape scale (Question 1) and site-scale (Question 3) we used the 1997 and 2017 vegetation management spatial series and the 2017 land-use series. To assess the effectiveness of the EPBC Act 1999 in protecting the threatened A. harpophylla woodlands and native grasslands (Question 2) we also used the 2001 and 2009 vegetation management spatial series, which corresponds to the year these vegetation types were listed for protection respectively. We conducted all analyses in ArcGIS 10.3 (Environmental Systems Research Institute 2015).

Landscape scale mapping methods

Assessment of vegetation clearing and land-use change (Question 1)

To address Question 1, we used the pre-clearing coverage data and the 1997 and 2017 vegetation management data to calculate the remnant percentage remaining of each dominant vegetation group (Table 2-1) in 1997 and 2017, relative to their pre-clearance extent within the study area (Table 2-2). The difference between the 1997 and 2017 percentage remnant reflects the net change, as it includes vegetation which has recovered from past clearing events (re-gained remnant status). To provide a clearer depiction of actual vegetation clearing we intersected the 1997 and 2017 data layers and calculated the total area of remnant vegetation cleared (remnant to non-remnant) and the total area recovered (non-remnant to remnant) between 1997 and 2017 for each vegetation group. This combined data frame was

18 then intersected with the current land-use mapping, to calculate the total remnant area cleared (remnant to non-remnant) and recovered (non-remnant to remnant) for each land-use category (Table 2-3).

Assessment of policy in protecting focal threatened vegetation types (Question 2)

To address Question 2, we used the vegetation management data from 1997, 2001 (for A. harpophylla woodland), 2009 (for native grasslands on clay) and 2017, and calculated the total remnant area cleared and average rate of loss per year, before and after their respective protection. Although the A. harpophylla woodlands are classified within the Acacia on clay broad vegetation group, the A. harpophylla dominant areas have been mapped separately, allowing for their separate analyses.

For A. harpophylla the pre-protection period assessed is between 1997 and 2001 (4 years) and the post-protection period assessed is between 2001 and 2017 (16 years). For native grasslands the pre-protection period is between 1997 and 2009 (12 years) and the post- protection period assessed is between 2009 and 2017 (8 years).

Site Scale mapping methods

Assessment of vegetation loss and land-use change (Questions 3)

To examine site-scale (of those sites established by Fensham 1999 in 1996) land clearing (Question 3), all sites were intersected with the 1997 and 2017 vegetation mapping data and the current land-use mapping. Sites were then separated into four vegetation status groups; sites which had been cleared (remnant to non-remnant), sites which had recovered from past clearing (non-remnant to remnant), sites which remained non-remnant and sites which remained remnant. Within each group the total number of sites for each vegetation group and land-use category was calculated.

Assessment of invasive species spread (Question 4)

19

Floristic surveys were conducted at all sites in 1996 and repeated for a subset of sites (n = 54) in 2018. Site location was determined using their Global Positioning System (GPS) location recorded in 1996. The same methods were implemented at both survey times, although by different observers. For these resurveyed sites, we explored site-scale changes in the cover of three focal invasives species; Bothriochloa pertusa (Indian couch), Cenchrus ciliaris (buffel grass) and Parthenium hysterophorus (parthenium weed). At each site, the cover of B. pertusa, C. ciliaris and P. hysterophorus was visually estimated within a 10 × 2 m plot using broad cover categories; 0 = absent , 1 = < 1%, 2 = 1 – 5%, 3 = 5 – 20%, 4 = > 20%. If the invader was present within a 20 m radius of the centre of the plot, they were recorded with a cover score of 1 = < 1%. The plot start (0 m) was located on the GPS point and ran for 10 m perpendicular to the nearest road or track and 1 m either side (2 m total).

Invader cover and presence were modelled as a function of year and broad vegetation group (as determined from the pre-clear vegetation mapping) as interacting fixed effects using general linear models with either a binomial (presence model) or Poisson (cover model) error distribution.

Resurveyed sites were all within Eucalyptus on clay (n = 26), Eucalyptus on texture contrast soils (n = 8) and grasslands on clay (n = 20). No sites in woodland on alluvium were resurveyed. Only sites that were still being used for grazing, conservation or other minimal use (i.e. road reserves) were surveyed.

Data validation

Either due to satellite interpretation error or ground truthing limitations, there is a degree of error in remnant and land-use classification within the spatial data used. We aimed to quantify some of this error by examining polygons and sites using high resolution satellite imagery (30 – 15 cm resolution) (Environmental Systems Research Institute, 2020). This method will only detect obvious discrepancies in land-use and vegetation clearing status, such as if a polygon is reported as remnant but inspection of satellite imagery reveals obvious vegetation clearing. At the landscape-scale data validation is used only to assist interpretation of results. At the site-scale results have been corrected to reflect data validation results.

20

Landscape-scale

We assessed a subset of polygons reported to have been cleared (changed from remnant to non-remnant status) or recovered (changed from non-remnant to remnant status) between 1997 and 2017. We organised this analysis within the six land-use categories, resulting in 12 assessment groups: two remnant status groups × six land-use categories.

For each group we used the high-resolution satellite imagery to examine either every polygon greater than 1 ha or up to 100 polygons. Polygons which were interpreted to have been restored (changed from non-remnant to remnant status), were inspected for evidence of vegetation clearing or intense use of land (i.e. cropping or mining). If obvious modification of land was evident within more than 50 % of the polygon an error was recorded. For polygons reported to have been cleared (changed from remnant to non-remnant) an error was recorded if the polygon had canopy cover representative of its broad vegetation group and similar to surrounding patches of remnant vegetation within the same broad vegetation group. Across the whole study area (landscape-scale) we examined 918 polygons across 2176 km2. Overall, the remnant status and land-use categorisation were correct for 78 % of polygons, equating to 22 % error. To assist with interpretation of results, error for each assessment group is presented as a percentage of the total polygons inspected (Table 2-2). A detailed breakdown of these results can be found in Appendix 2-1.

Site-scale

In 2018 we visited 79 sites which helped to validate and improve remnant status and land-use results obtained from spatial data sets. To assess and validate the remaining 128 sites we used high resolution satellite imagery, as per the landscape scale analysis. In total there were five errors in land-use classification and eight errors in remnant classification; six sites falsely attributed to remnant and two falsely attributed to non-remnant. The results reflect the corrected values for these sites.

Results

21

Landscape-scale

How much of and what type of vegetation has been cleared, and for what land-use?

In 1997, only 55% of vegetation was classified as remnant within the study area (Figure 2-1). By 2017 a further 5.6% of remnant vegetation had been cleared, totalling 7392 km2. Vegetation groups with the greatest loss of remnant vegetation between 1997 and 2018 were also generally the most widespread and included Eucalyptus woodlands occurring on low- moderately fertile texture contrast soils (E. populnea or E. melanophloia dominant) and Acacia woodlands on fertile clay soils (Table 2-2). These vegetation groups in addition to grasslands on fertile clay soils also experienced the greatest decline in remnant extent prior to 1997. 1405 km2 of non-remnant vegetation recovered remnant status over this time period and this mostly occurred in Eucalyptus woodlands on texture contrast soils and Acacia woodlands on clay.

Land-used for livestock grazing accounted for 95% of the remnant vegetation cleared (7048 km2) (Table 2-3). Cropping and mining were the next most common, accounting for 1.9% (146 km2) and 1.8% (133 km2) of remnant vegetation cleared. Altogether, conservation and natural environments, forestry and other intensive uses made up less than 1% of remnant vegetation cleared.

Land which recovered remnant status was mostly within land-used for livestock grazing, however there was a small proportion within more intensive land-uses, such as cropping and mining (Table 2-3). The polygons that contributed to this result were often small and on the edge of land-used for these intensives uses. Data validation techniques using high resolution satellite imagery also show a high degree of remnant classification error for some of these land-uses, particularly for cropping. Cropping land-units are known to sometimes be mistaken for remnant grassland, as the digital signature is similar.

How did clearing rates in focal threatened ecosystems change after their listing for protection under federal environmental policy?

22

Within the study area only 20% of the A. harpophylla woodland and 36% of the native grassland pre-clearing extent remained in 1997. Between 1997 and 2001 (pre-protection), 937 km2 of A. harpophylla woodland was cleared, and a further 209 km2 between 2001 and 2017 (post-protection). This equates to an average rate of 234 km2/per year before protection (and after 1997) and 13 km2 /per year after protection. For native grasslands 176 km2 was cleared between 1997 and 2009 (pre-protection) and a further 109 km2 between 2009 and 2017 (post- protection). This equates to a rate of 14 km2/per year before protection (and after 1997) and 13.6 km2/per year after protection.

23

Figure 2-1. The distribution of remnant vegetation in 2017. Land which was cleared (became non-remnant) pre-1997 is shown in white. Areas which were cleared between 1997 and 2017 are shown in dark grey. Overlayed are all sites surveyed in 1996 (triangles); those resurveyed in 2018 are shown in red and those not resurveyed but included in overall analysis are shown in black. 24

Table 2-2. The dominant broad vegetation groups within the catchment area and the percent remnant remaining in 1997 and 2020. Total remnant area cleared and recovered (km2) between this time is also shown. Groups are ordered from greatest to smallest loss of remnant vegetation between this time period. Remnant Remnant Total lost Total gained Land Type remaining remaining (km2) (km2) 1997 (%) 2020 (%) Eucalyptus on texture-contrast 63.62 55.12 4107 437 soils Acacia on clay1 19.70 17.03 1353 352 Eucalyptus on hills 64.29 60.55 443 162 Woodland on 88.64 86.74 335 125 alluvium Grasslands on 36.61 33.89 285 51 clay2 Acacia on sand 83.83 80.85 214 51 Eucalyptus on sand 74.58 69.88 210 57 Eucalyptus on clay 71.23 68.32 205 52 1 937 km2 within Acacia harpophylla (brigalow) ecosystems. 209 km2 cleared after listed for protection under the EPBC in 2001.

2 109 km2 of remnant lost after listed for protection under the EPBC in 2009.

25

Table 2-3. Total remnant vegetation cleared and recovered between 2017 and 2020 for each land-use category. Percent error (brackets) is reflective of data validation results which assessed errors in remnant classification for a subset of polygons using high resolution satellite imagery (See Appendix 2-1 for detailed data validation results). Groups are ordered from greatest to smallest loss of remnant vegetation between 1997 and 2020.

Land-use Remnant lost (km2) Remnant gained (km2) Grazing 7048.1 (2%) 1348.4 (1%) Cropping 145.8 (0%) 16.8 (50%) Mining 133.1 (0%) 8.2 (12.5%) Conservation or minimal use 16.1 (17%) 20.6 (2%) Forestry 12.9 (10%) 0.5 (10%) Other intensive use 3.6 (0%) 3.7 (9%)

Site scale

How many sites have been cleared of vegetation and for what land-use between 1997 and 2017?

Of the 207 sites, 78 were already classified as non-remnant in 1997. By 2017, an additional 17 sites were classified as non-remnant and 36 sites (34 already non-remnant) were now being used for cropping (mostly in grasslands and Eucalyptus woodlands on clay) (Table 2- 4). Of the 109 sites which remained remnant, 105 were used for livestock grazing and seven sites were within conservation areas. These protected areas were all within grasslands on clay.

26

Table 2-4. Total number of sites cleared (remnant to non-remnant), recovered (non- remnant to remnant), remained non-remnant or remained remnant between 1997 and 2017 and their corresponding broad vegetation group and current land-use. In 1995 all sites were within either land-used for grazing or conservation. Sites converted to cropping indicate an intensification of land-use (*).

Eucalyptus Woodland on Eucalyptus on Grasslands Total on clay alluvium texture-contrast on clay (n=57) (n=21) soils (n=22) (n=107)

Cleared remnant

Cropping 2 2 * Grazing 1 3 1 10 15

Recovered remnant

Grazing 1 2 3 Remained non-remnant Cropping 12 2 1 19 34 * Grazing 16 4 3 21 44 Remained remnant Conservation 7 7 Grazing 28 11 17 46 102

27

How has the cover and abundance of the three focal invasive species changed between 1996 and 2018?

Across the 54 re-surveyed sites, there was a significant increase in the cover and presence of B. pertusa and C. ciliaris between 1996 and 2018 and this was irrespective of broad vegetation group (Figure 2-2). Bothriochloa pertusa was present on one site in 1997 and on 35 sites in 2018, with greater than 20% cover on 10 sites. Cenchrus ciliaris was present on 22 sites in 1995, with zero sites above 20% cover. In 2018 C. ciliaris was present on 40 sites, with greater than 20% cover on 12 sites. Parthenium hysterophorus cover and presence also increased between 1996 and 2017, although not significantly so.

Figure 2-2. The total number of sites each invader (B. pertusa (a), C. ciliaris (b) and P. hysterophorus (c)) was present and their broad cover values (light blue to dark green), in 1995 and 2018. Results from linear models with survey year as the predictor and either invader cover or presence as the response are also shown ((NS P > 0.5, *0.01 > P < 0.05, **0.001 > P 0.01, *** P < 0.001)).

Discussion

This study aimed to assess land clearing, changes to land-use and the invasion of exotic species within an area of Australia subject to rapid and extreme land-use change. We found that despite specific environmental policy and legislative developments aimed at reducing land clearing, significant clearing still occurred. In addition to this ongoing threat of clearing, our assessment highlights the vulnerability of remaining fragments of native vegetation to

28 invasion by high-impact exotic species. Continued loss of remnant vegetation will likely only further facilitate the spread and impact of these invasive species within these fragmented landscapes.

Land-use change, land clearing and the failure of environmental policy

During the last 50 years, Queensland, Australia has been a global hotspot for deforestation and land clearing (Reside et al. 2017) and this has largely been driven by the expansion of managed agricultural systems (Lepers et al. 2005, McAlpine et al. 2009). Within the study area assessed here, livestock grazing, and cropping were the predominant source of historical land clearing (pre 1997) and these remain the main threat to fragments of native vegetation in recent decades. These trends in land-use are not unique to Queensland or Australia but are consistent across the globe (McAlpine et al. 2009), with Steinfeld and Wassenaar (2007) estimating that managed grazing and fodder crops, occupy a third of the earth’s land surface.

Woodlands and grasslands existing on fertile soils, particularly the Acacia on clay and grasslands on clay experienced the greatest loss in remnant vegetation historically, with only 20% and 37% of these respective vegetation types remaining within the study area in 1997. These vegetation types also experienced some of the highest clearing rates of remnant vegetation in recent decades, particularly in the fertile Acacia woodlands and including significant clearing within Acacia harpophylla (brigalow) woodland after it was listed for federal protection under the Environmental Protection and Biodiversity Act 1999 (EPBC) in 2001. Similarly, 285 km2 of remnant grasslands on clay were cleared between 1997 and 2017, of which 109 km2 was cleared after its listing for protection under the EPBC in 2009. These fertile vegetation types are well suited to being cultivated as exotic pasture or crops, and the economic incentive for their development can outweigh the likelihood of being prosecuted or potential penalty from disregarding legislation. These findings are particularly pertinent given the recent audit into the EPBC Act 1999 which found severe negligence in the reporting, monitoring and enforcement of this legislation (Auditor-General 2020).

Remnant patches of Eucalyptus on texture contrast soils and Eucalyptus on hills also declined significantly, with 4107 km2 and 443 km2 cleared respectively during this time period. Although less fertile and not suitable for cropping, these vegetation types can sometimes

29 support greater grass growth and pasture development after clearing (Silcock et al. 2015). Although these vegetation types are not protected under the federal EPBC, restrictions on broad scale clearing of remnant vegetation implemented in 2006 under the state-controlled Vegetation Management Act 1999 (VMA), should have provided some protection from clearing in these vegetation types. Wavering implementation and amendments to the VMA with changes to government administration have likely led to continued clearing of remnant vegetation within these vegetation types.

Analysis of land-use at the site scale highlighted the significant contribution of land-use intensification to land clearing within the region. Thirty-four (16%) sites categorised as non- remnant in 1997, but still used for low intensity land-uses (i.e. grazing of native vegetation or conservation) had been converted for cropping. Similar trends in land-use intensification are occurring across the globe (particularly in wealthier nations) with considerable consequences for biodiversity (Donald et al. 2001, Söderström et al. 2003).

While the conversion of some vegetation types does enhance primary production, such intense land-use has been a primary driver of biodiversity decline in Australia, and across the globe (Maxwell 2016). McAlpine et al. (2002) suggest that where ecosystems have been reduced to 30% of their pre-clearing extant, 25-30% of vertebrate fauna will decline, with the full impact likely not realised for 50-100 years (McAlpine et al. 2002). This threshold of ecological collapse has already been reached for the fertile Acacia woodlands and grasslands examined here and evidence of biodiversity loss within these vegetation types is increasingly reported (Woinarski et al. 2006, Fensham et al. 2017, Silcock and Fensham 2018, Fensham et al. 2019). For example, between 1973 and 2002 there was a significant decline in many woodland species and a shift in composition to more disturbance-tolerant species within these heavily cleared landscapes of Queensland (Woinarski et al. 2006). Many plant species have also undergone significant range contractions, with some presumed extinct due to the intensity of LUC in region (Fensham et al. 2018, Fensham et al. 2019). The protection and restoration of remaining fragments of remnant vegetation is crucial to the persistence of native biota within these heavily cleared landscapes. The ecological legacy of such intense LUC however will likely continue provide many challenges for the conservation of biodiversity (Lindenmayer et al. 2008). The ongoing invasion of exotic species is as a key

30 consequence and one that greatly compromises the persistence of native species within remaining fragments of remnant vegetation.

LUC and the spread of invasive species

The spread of invasive species is a well-known consequence of LUC and this is particularly evident in regions where their establishment has been assisted by pasture development (Chytrý et al. 2008, McAlpine et al. 2009, Pyšek et al. 2010). In the current study we found a significant increase in two damaging invasive pasture grass species, between 1996 and 2018. As highlighted by our landscape-scale analysis, this time period captures both the end of a historic era of extreme land transformation and the start of a modern era of land-use reform. Land clearing, altered disturbance regimes, and propagule pressure are interacting factors which have likely facilitated the spread of the invaders examined here, as well as in other fragmented landscapes across the globe (Chytrý et al. 2008, Vilà and Ibáñez 2011).

Clearing of woodlands for the development of Cenchrus ciliaris (buffel grass) pastures has been a key enterprise in the region for more than 50 years and the remaining fragments of native vegetation are often surrounded by vast C. ciliaris pastures. As such, high propagule pressure has been identified as the main factor, aside from fire and grazing, that facilitates the spread of this species into fragments of remnant vegetation (Fensham et al. 2013). In addition, the species responds positively to fire and its high biomass can fuel intense and frequent wildfire. Acacia woodlands, where the establishment of C. ciliaris has been most extensive, are fire-sensitive and in their remnant condition fire-resistant due to their high canopy reducing the capacity for grass growth. On the edges of the these woodlands however, light is abundant and C. ciliaris can readily establish, increasing the incidence of fire, eroding the woodland edge and further promoting the establishment of C. ciliaris (Butler and Fairfax 2003). For vegetation types where light is not limited, such as many Eucalyptus dominated woodlands in the region, C. ciliaris can readily invade. Although Eucalyptus woodlands are generally fire tolerant, frequent high-intensity burns within these systems can compromise the persistence of many native species and further promote the establishment of C. ciliaris (Fensham et al. 2015b).

31

These impacts of pasture development are not unique to Australia, with similar losses to biodiversity occurring across the globe. Between 1970 and 2000 in Colombia, clearing rates exceeded 230,000 ha per year with the majority of clearing attributed the expansion of exotic grasslands, which increased by 50% during the 1990s (Etter et al. 2008). These exotic grasses now cover most of the cleared forest in sub-humid areas and akin to C. ciliaris are degrading native biodiversity and ecosystem processes by outcompeting native grasses, altering fire regimes and inhibiting the restoration of tree species (Williams and Baruch 2000, Mistry and Berardi 2005, Hoffmann and Haridasan 2008). Similarly, in Brazil, ~1.9 million ha per year was cleared between 1978 and 2007, largely for exotic pasture development (McAlpine et al. 2009). Remaining forest fragments are surrounded by exotic pastures and their persistence is threatened by fire and competition from exotics (Laurance et al. 2006).

Although B. pertusa was first introduced in the 1930s, its rapid spread throughout Queensland in recent decades is likely an example of an ‘invasion debt’ incurred from an intense history of land degradation in the region (Jackson and Sax 2010, Essl et al. 2011, Vilà and Ibáñez 2011). Unlike C. ciliaris, B. pertusa was not as extensively sown for pasture. This species is very grazing tolerant however and the intense use of land for livestock grazing, in conjunction with widespread land-clearing has likely facilitated the spread of this species in recent decades. Similar trends have been observed for other invaders across the globe, with many taking advantage of ‘windows of opportunity’ after landscape scale disturbance facilitated their spread (DeGasperis and Motzkin 2007, Mosher et al. 2009).

Within a similar environment in northern Queensland, comparable increases in the occupancy of B. pertusa were recorded during a long-term grazing trial (O’Reagain and Bushell 2015). At the beginning of the trial in 1998, B. pertusa was very infrequently recorded on less than < 1% of sites. From the mid-2000s however the species rapidly increased, occurring on 60% of the heavily grazed sites and 25% of the conservatively grazed sites in 2014. The increased proportion of edge habitat within the landscape has also likely facilitated the spread of B. pertusa, who’s dominance along roadsides has been associated with its spread into adjacent habitat (Bisset 1980). With the incurred ‘invasion debt’ of B. pertusa invasion, an associated delay in the response of native communities, an ‘extinction debt’, may also be expected (Gilbert and Levine 2013).

32

Unlike C. ciliaris and B. pertusa, P. hysterophorus is recognised as a weed of national significance due to its negative impacts to pasture productivity (Chippendale and Panetta 1994). Mandatory control measures (herbicide, biocontrol, vehicle wash-downs) have been implemented across Queensland since the 1970s and recently, expert consultation identified P. hysterophorus as the most damaging weed in Queensland (Osunkoya et al. 2019). The only small increase in P. hysterophorus cover and occupancy observed here, may be partly due to the implementation of these control measures. Parthenium hysterophorus however is strongly disturbance dependent and can be diminished by reducing disturbance and allowing native ground cover to re-establish (Fensham et al. 1999, Fensham et al. 2016). Despite this being well-known, managing grazing pressure is not considered under the Australian government’s management guidelines for P. hysterophorus (Weeds Australia, 2020). In contrast, perennial grasses, such as C. ciliaris and B. pertusa have far greater consequences for biodiversity than annual plants such as P. hysterophorus. Furthermore, it is extremely difficult to reverse the invasion of perennial grasses (Brooks et al. 2010). The omission of C. ciliaris and B. pertusa from the recent inventory of invasive plant species (Osunkoya et al. 2019) and the poor consideration of grazing management under the Australian weed management strategy, highlights the currently biased approach to weed management in Australia. To accurately assess and prioritise the management of invasive species, it is necessary to account for the impacts of invasive pasture species. Without doing so, we run the risk of underestimating the current and potential impact of invasive species and wasting resources on those less imperative.

Conclusions

Environmental policy has been essential in slowing the rates of land clearing in Queensland (Reside et al. 2017). However, it is clear from the results presented here that considerable rates of land conversion still occur, often within vegetation protected under national environmental legislation. This has largely been due to the inconsistent enforcement of key legislation and policies, and poor monitoring and penalisation of illegal clearing. If we are to limit LUC induced species declines and extinctions, improving the development, compliance and reporting of policy is essential, and this relies on the provision of adequate resources and support (Burnett 2020).

33

In addition to policy and legislative improvements, implementing programs which incentivise the protection and restoration of private land is crucial, and this is gaining momentum in Australia and elsewhere (Kleijn and Sutherland 2003, Hajkowicz and Collins 2009, Burns et al. 2016). Many of the most heavily modified ecosystems, such as the Acacia and grasslands on clay, have high potential for restoration (Dwyer et al. 2010, Fensham et al. 2016), however without compensating landholders for reduced production, there is little incentive for their protection. These ecosystems are also important carbon sinks (Fensham and Guymer 2009, Dwyer et al. 2010, Smith 2014), and due to their importance for conservation, carbon sequestered from their restoration has high potential to qualify for a premium price on the carbon market. As the willingness to buy premium carbon credits gains momentum, this presents a considerable opportunity to encourage the restoration of these and other ecosystems (Parnphumeesup and Kerr 2015). Incentives such as this provide other sources of income for landholders, improving their capacity to reduce the extent and intensity of land- use on their property.

The spread of invasive exotic species is a key threat to the persistence and restoration of remnant vegetation within these highly fragmented and modified landscapes. In many cases their spread is exacerbated by the widespread establishment of exotic pasture species for livestock grazing. Recognising and acknowledging the negative impact of invasive pasture species (such as B. pertusa and C. ciliaris) on biodiversity by listing them for management under national or state legislation will help to raise awareness about their impacts and allocate resources towards their management. Overall, to improve the management of invasive species within fragmented and modified landscapes, ongoing research, education and technical support is crucial. Importantly, ensuring both short-term (e.g. herbicide) and long- term (e.g. grazing management) control strategies are considered will lead to far more widespread management outcomes.

34

Chapter 2 is currently under review with the Journal of Environmental Management.

Lebbink, G.H., Dwyer, J., Fensham, R. J. (2020). ‘Invasion credit’ after extensive land-use change: An example from eastern Australia. Journal of Environmental Management (in review).

Writing, analysis, field work and mapping by GL, with editorial advice from RF and JD.

35

Chapter 3 – The ‘lawnification’ of Queensland’s grassy woodlands: Mapping the past, current and likely future extent of the exotic grass Bothriochloa pertusa

Abstract Many of today’s environmental weed species were deliberately introduced during pasture development trails. Despite the published evidence of negative impact of these species, there are few studies documenting their introduction history and spread. These stages of invasion are important to evaluate, as they not only inform management strategies, but help to highlight areas of improvement for pasture introduction trails in the future. Using collated presence/absence and cover data, alongside a review of the literature and discussions with land managers, we present a comprehensive analysis of the introduction history and spread of the invasive perennial grass species, Bothriochloa pertusa throughout Queensland. Using this data, we also develop habitat suitability models and use these to predict its potential distribution and local-scale cover across Queensland. We found that B. pertusa was introduced on multiple occasions and across a large area of Queensland, despite re-occurring doubts and poor evidence for its benefit to livestock production. Livestock grazing, associated disturbances (i.e. land clearing, soil erosion) and climatic extremes were commonly associated with its spread throughout the landscape. Today B. pertusa is common within most coastal and sub-regions of Queensland and dominates the groundcover in many areas. Habitat suitability models suggest the occurrence and local-scale cover of B. pertusa is largely determined by climate (mostly mean growing season temperature) and the foliage projective cover of trees. Based on these results B. pertusa still has considerable capacity to spread and increase in dominance across many areas of Queensland. We recommend reducing stocking rates and increasing grazing-protected areas to slow the spread of this species. Overall, to reduce the current and potential impacts of invasive pasture plants, a fundamental shift in ideologies - from enhancing pastoral productivity at the expense of the environment, to maintaining ecological values alongside sustainable pastoral production, needs to occur.

Introduction

The introduction of exotic plant species for the purpose of forage and browse has been a pervasive agricultural practice in Australia for several decades (Walker and Weston 1990,

36

Cook and Dias 2006a). The success of these species in increasing pastoral productivity has been variable, with many species achieving negligible benefits for production and commonly becoming problematic invasive species (Cook and Dias 2006b, Grice 2006, Firn and Buckley 2010, Setterfield et al. 2010, Driscoll et al. 2014). Despite the negative impact of many failed pasture species being well-established, their introduction history and initial spread is often overlooked or understated within the invasion biology literature (Puth and Post 2005, Cook and Dias 2006a), although there are exceptions, see; Adams et al. (2015); Cook and Dias (2006b); Lonsdale (1994)). This is concerning as the introduction and early spread of invaders provides important insights into the mechanisms that drive their expansion and are essential for developing strategies to reduce their spread and impact (Lonsdale 1999). Further, given the pressure to meet global food demands and the likely intensification of pasture development in Australia (McAlpine et al. 2009, Driscoll et al. 2014), having a thorough understanding of past introductions is important for improving the outcomes of pasture introduction programs in the future. This study presents a comprehensive analysis of the introduction history, spread and potential future distribution of an invasive pasture species in Queensland, Australia.

Since its introduction in 1939, the perennial grass species, Bothriochloa pertusa (B. pertusa) has spread widely and has been observed forming apparent monocultures across large areas of Queensland (McIvor 2007, Ash et al. 2011, O'Reagain et al. 2018). The continued spread of the species poses a significant threat to native biodiversity and ecosystem function (Kutt and Fisher 2011, Kutt and Kemp 2012, Koci et al. 2020), in a region already under threat from other invasive species, widespread habitat clearance and other anthropogenic disturbances (McAlpine et al. 2009, Reside et al. 2017). Understanding the past and current extent of B. pertusa is crucial for predicting and managing its ongoing spread and impacts on native biota. Although the findings presented here are specific to north-eastern Australia, the methods and concepts discussed are relevant for understanding the spread and impact of invaders across the globe.

Methods

Data sources

37

B. pertusa presence/absence records were collated from multiple data sources and used to describe and analyse the introduction history and spread of B. pertusa, and to predict its future distribution throughout the north-eastern Australian state of Queensland. Data included both publicly available records and records collated from field surveys (Table 3-1, Appendix 3-1). Once collated, the data was checked, and duplicate records removed. In total we collated 1449 presence records spanning 1939 – 2019. In addition, 759 contemporary absence records were collated from where B. pertusa was not recorded during comprehensive floristic surveys conducted between 2017 and 2020.

For 1262 of the 2208 total presence/absence records, a measure of B. pertusa cover was also recorded, either as a percentage or within the broad cover categories (1 = absent , 2 = < 1% , 3 = 1 – 10%, 4 = 10 – 20% , 5 = 20 – 50%, 6 = > 50%). We converted all records to broad cover categories for use in the habitat suitability model described below.

38

Table 3-1. Total number of B. pertusa presence, absence and cover records collated from a variety of data sources between 1941 – 2019. Contemporary absence records were collated from where B. pertusa was not recorded during floristic surveys conducted between 2017 and 2020. Further information on data collection methods can be found in Appendix 3- 1.

Source Presence/absence Cover Date range Detailed floristic surveys Queensland Herbarium CORVEG 774 738 1984 – 2019 Other Queensland Herbarium survey sites 411 1950 – 2019 Department of Agriculture and Fisheries Qgraze 79 1992 – 2001 Bothriochloa pertusa impact and spread surveys* 130 130 2017 – 2020 Online sources Australian Virtual Herbarium 271 1941 – 2018 Other Bothriochloa pertusa spread surveys* 390 357 2017 – 2020 Property evaluation vegetation maps** 153 1950 – 1997 Total 2208 1225 *Conducted by the author (G. Lebbink) **Sourced from the Queensland Department of Natural Resources, Mines and Energy

Early introduction and spread

We used only the B. pertusa presence records, alongside information obtained from within the literature, and opportunistic discussions with land managers to assess and describe the introduction history and spread of B. pertusa throughout Queensland. To illustrate its spread, we mapped the change in the total number of presence records (within 30 km grid cells) over time. We acknowledge that there are limitations to using species presence records to discuss trends in a species’ spread. In particular, the rate of species collections is not consistent over time and space. However, the trends in this data were verified by information obtained from the literature and discussions with long-term land holders who have witnessed the spread of B. pertusa.

39

Habitat suitability and predicted future spread

To assess the relationship between B. pertusa and key environmental predictors, and to predict its future spread, we performed two habitat suitability models (HSM); one with presence/absence (occurrence) and another with cover as the dependent variable. These models predict the potential distribution of B. pertusa based on its current occurrence and environmental preferences. For Queensland, the environmental predictors that were consistently mapped and appropriate for inclusion in a HSM were as follows: land-zones (12 classes summarising soil and geology), climate variability, wet-season rainfall, mean temperature during the growing-season (October to April), distance to a waterway, soil clay content, foliage projective cover (FPC) of trees and vegetation clearing index (3 classes; cleared, regrowth, remnant). Gridded climate data was obtained from Bureau of Meteorology, 2020. Land-zone information (Christian et al. 1953; Gunn et al. 1967; Story et al. 1967; Speck et al. 1968; Galloway, Story & Gunn 1970; Galloway et al. 1974; Nix & Gunn 1977) and other environmental data were obtained from the Queensland Government Spatial Catalogue 2020. Some land-zones were not well represented by the data and were excluded from both the cover and occurrence models. To ensure the cover models were representative of the current invasion potential, only records (both presence and absence) from the last 5 years (2015 – 2020) and those with a cover value > 20% (as this still reflects vulnerable habitat), were included (1226 records in total). After these data checks, 1433 presence records and 740 absence records (2174) were used in the occurrence models, and 466 presence (with broad cover 2-6) and 760 absence (with broad cover = 1) records were used in the cover model.

A boosted regression tree approach with ten-fold cross validation of training data (Elith et al. 2008) was used to build the HSMs. The resulting occurrence model explained 45% of cross- validated deviance, while the cover model explained 38%. The fitted HSM was used alongside the mapped grids of environmental conditions to predict the probability of occurrence and likely cover of B. pertusa within Queensland. Analyses were conducted in R (version 2.10.0, R development Core Team, 2009) using the “gbm” library supplemented with functions from Elith et al. (2008) (see Appendix 3-2 for complete modelling methods).

40

Results and Discussion

Early introductions

The first Australian Herbarium samples of B. pertusa are from Queensland and date back to 1939, where it was introduced to the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Fitrzroyvale research facility, in Rockhampton, and to the agricultural college facility in Gatton. At Fitzroyvale, B. pertusa was trialled within five 40 m2 plots; two sown with only B. pertusa and three with a mixture of B. pertusa, six other exotic grass species and the exotic leguminous shrub Stylosanthes gracilis (Miles 1949). At conclusion of the trial in 1942 the species was noted as producing a “soft growth to 15 inches (40 cm) high, with late summer flowering and was fairly well sought by stock”. B. pertusa was not included amongst the promising species enlisted for wider regional testing. Among the species considered ‘promising’ by Miles (1949) were a number that have since spread widely and had substantial negative environmental impacts; namely Andropogon gayanus (Rossiter - Rachor et al. 2009, Setterfield et al. 2010), Chloris gayana, Cenchrus ciliaris (Fairfax and Fensham 2000, Jackson 2005), Melinis multiflora (van Klinken and Friedel 2017) and Panicum maximum (van Klinken and Friedel 2017).

Despite not being suggested for wider testing, B. pertusa continued to be trialled at a number of different research facilities across the state. In the early 1950s, B. pertusa was being grown for seed at the CSIRO research facilities in Samford and Strathpine, in South East Queensland. In 1955, the species was trailed at the Department of Agriculture and Fisheries (DAF) research stations in Emerald and Biloela. At Emerald, B. pertusa was noted as “unimpressive for pastures because of its low productivity” (Bisset 1980). In 1978 B. pertusa was introduced to the DAF Brian Pastures research facility, near Gayndah in the Burnett region, and in the early 1990s B. pertusa was included in pasture trails at CSIRO Landsdown station in Townsville (Jones 1997).

B. pertusa was also regularly sown for amenity purposes, such as lawns, airstrips and along road verges. In the mid-1950s seed was made available to the Department of Civil Aviation for experimental plantings at the Cloncurry, Charleville and Bowen aerodromes (Bisset 1980). The Cloncurry planting didn’t survive, but the grass is still present at Charleville. In

41

1950 the species was recorded at the Bowen showgrounds, likely from a deliberate planting for lawn. Although the pasture trails in Emerald (in 1955) did not advocate B. pertusa as fodder, it was suggested as a promising lawn species for these drier areas of Queensland, so was regularly sown for this purpose within residential properties, but also for golf courses and showgrounds. Sowing of the species for amenity purposes was also occurring in the Northern Territory. In 1986 the species was recorded in old settlements and horse paddocks in Kakadu National Park, which was likely introduced before the park’s gazettal in 1979 (Cowie 1992). The species is now common throughout the Northern Territory and its dominance is increasing in some grazed ecosystems (Robyn Cowley pers. comm. 2019).

In the late 1970s, a number of trials were established to assess the value of B. pertusa for soil conservation (Bisset 1980, Truong and McDowell 1985). Initial trials at Amberly (South East Queensland) in 1978 noted B. pertusa as “the most promising grass amongst the species evaluated” (Truong and McDowell 1985). Based on the initial findings from this four-year trial, a series of larger trials were established in 1979 and 1980 across several regions of Queensland (Moreton, Darling Downs, Coastal Burnett and Central Highlands), to evaluate the effectiveness of B. pertusa in stabilising farm waterways. B. pertusa was quick to germinate and establish high cover and deemed useful for stabilising the riverbanks and reducing water erosion. Its use for other stabilising works such as mining overburden and road and railway embankments was also promoted in this study.

During the 1980s the species was sown for rehabilitation at Blackwater, Theodore and Callide coal mines (Truong and McDowell 1985) and was observed spreading naturally onto mine spoils at the Collinsville coal mine. The abundance of B. pertusa on gem-mine spoils near Rubyvale, central Queensland, is also likely associated with the rehabilitation of these mines in the late 1990s. It is difficult to determine how frequently the species was used for rehabilitation projects as detailed records were often not kept or were not made available (Silcock 1991). The Queensland Government’s Soil Conservation Guidelines (2015), however, suggest that B. pertusa, along with Cenchrus ciliaris (buffel grass) were used widely for rehabilitation projects on the lighter arid inland soils but “were no longer recommended due their weed potential”.

42

Although its value for fodder is now largely contested, during the late 1980s and into the 1990s the species was often recommended as a useful fodder crop, particularly in low fertility soils and to sustain high grazing pressure (Partridge and Miller 1991). A survey of 297 commercial beef producers in Queensland during 1996-97, found B. pertusa was used for pasture improvement on 2% of properties in the central coastal region and central highlands region, and 4% of properties in north Queensland (Bortolussi et al. 2005). Further, the species was observed naturally spreading onto several properties in these regions, particularly in north Queensland.

Spread

It wasn’t until the mid-1960s that the number of B. pertusa herbarium records in Queensland started to increase and its spread discussed within the literature. During the 1960s B. pertusa mostly occurred within the Bowen region, where the species was described to have ‘spread like wildfire’, during the 1960s and 1970s (Figure 3-2a) (Bisset 1980). The landholders of Salisbury Plains, near Bowen first saw the species in 1964 and were “initially concerned by its aggression”, but noted an “improvement in management and production” in comparison to pastures previously dominated by Heteropogon contortus (black spear grass) (Partridge and Miller 1991). Vegetation maps used for property evaluations in the 1960s and 1970s (Table 3-1, Appendix 3-1), also suggest the species was dominant across several properties in the Bowen region during this time.

By 1980 the species was observed forming “solid stands over whole paddocks between the coast and Bogie River”, which sits 50 km inland and runs parallel to the coast from Ayr to Bowen (Bisset 1980). The species was also noted as abundant and spreading along roadsides inland from Bowen and up towards Townsville. Transport of hay and cattle from the Bowen region is thought to have facilitated its spread into these regions (Bisset 1980). It was also during the 1980s when B. pertusa started to increase in occurrence inland of Mackay, and it is within this region today that B. pertusa is particularly abundant (Figure 3-2b, e). Coal mining is an extensive enterprise in this region, and the use of B. pertusa to rehabilitate mine spoils during this time (Truong and McDowell 1985) is also plausibly associated with this spike in occurrences.

43

Throughout the 1990s the number of records in the Charters Towers region (west of Townsville) increased considerably (Figure 3-2c). B. pertusa was observed naturally spreading onto several properties during this time, often from the edge of roads or from neighbouring sown pastures (Bortolussi et al. 2005, O’Reagain and Bushell 2015). A grazing trial initiated in the region in 1992, found a steady increase in B. pertusa after the mid-1990s, particularly on heavily grazed sites (Ash et al. 2011). Its increase after this time was suggested to be associated with a series of below average rainfall years prior to 1996, followed by a run of higher rainfall years leading up to the 2000s (Ash et al. 2011). Also during the 1990s, B. pertusa was observed replacing substantial areas of H. contortus grasslands in coastal and sub-coastal Queensland, with Walker and Weston (1990) suggesting 100,0000 ha in the Burdekin Shire, 200,000 ha in the Dalrymple Shire and 500,000 ha in the Bowen Shire had been colonised by B. pertusa by the early 1990s.

During the 2000s the species appeared to expand throughout the coastal and sub-coastal area between Mackay and the northern Burdekin (Figure. 3-2d). On a long-term (1998 to current) grazing trail near Charters Towers, the species was very infrequently recorded up until 2007, when it increased exponentially across all grazing treatments, but particularly on the heavily grazed treatments in the poplar box (Eucalyptus populnea) woodland (O’Reagain and Bushell 2015) (Figure 3-1). Similar to the initial spike in the region in the 1990s, this increase was thought to be associated with the few years of above average rainfall following drought culminating in 2008 (Fensham et al. 2015a). The significant increase in B. pertusa across the trail was coupled with significant declines in native perennial grass species, particularly of its native conger Bothriochloa ewartiana (Figure 3-1). In recent years, both B. pertusa and B. ewartiana have declined considerably across the whole trail, possibly also due to ongoing rainfall deficits in the region. Also during the 2000s, B. pertusa appeared to increase in the Cape York Peninsula, which aligns with findings from Bortolussi et al. (2005) who suggest the species was naturally spreading onto several pastoral properties in this region during this time.

44

B. ewartiana - Heavy stocking B. ewartiana - Moderate stocking B. pertusa - Heavy stocking B. pertusa - Moderate stocking 70% 1400

60% 1200

50% 1000

40% 800

30% 600

Rainfall (mm) Rainfall Frequency 20% 400

10% 200

0% 0

Figure 3-1. Change in the frequency of B. pertusa (black line) and its native congener B. ewartiana (blue line) over time (1998 – 2020) in heavily stocked (triangles) and moderately stocked (squares) poplar box (Eucalyptus populnea) woodlands on the Wambiana grazing trail, near Charters Towers. Heavy stocking rates equate to 4 ha per animal equivalent and moderate stocking rates to 8 ha per animal equivalent. Average rainfall (mm) over time is also shown. Data provided by Dr Peter O’Reagain and the Department of Agricultural and Fisheries. Further details of the trial and methods can be found in (O’Reagain 2006).

In the last 10 years (2010 – 2020) B. pertusa has continued to spread and increase in dominance in northern and central Queensland, and more recently has spread within the Burnett region (Figure. 3-1e). A producer survey conducted in central Queensland in 2016 suggests that only in the last 5 – 10 years has B. pertusa become particularly noticeable and problematic (Spiegel 2016). Surveys conducted in central Queensland’s grassy woodland

45 ecosystem (between Springsure and Moranbah) in 1995-6 found B. pertusa in only 1 of 207 survey sites. We resurveyed 92 of these sites in 2018 and found B. pertusa in 43 sites and at greater than 20% cover in nine sites (See Chapter 2). Its spread within the Burnett region has likely been even more recent with two producers near Gayndah suggesting that although it has been present for ~ 25 years (mostly along roadsides) its dominance within the pasture has only become noticeable in the last five years. Since its introduction to the Brian Pastures research station in 1978, B. pertusa has also spread considerably, replacing native Bothriochloa ewartiana pastures in some areas.

Figure 3-2. Total B. pertusa presence records within 30 km2 raster grid cells in 1980 (a), 1990 (b) 2000 (c), 2010 (d) and 2019 (e), within Queensland. Records were collated from a number of different data sources which are detailed in Appendix 3-1.

Habitat suitability and predicted spread

The occurrence and cover of B. pertusa across Queensland was largely predicted by climate (particularly mean temperature during the growing season (October to April)) and FPC of trees (Table 3-1, Figure 3-3 and 3-4). B. pertusa mostly occurred in areas with a mean 46 growing season temperature between 23 C˚ and 27 C˚ and in areas with low tree cover (< 40% FPC). Outside of these thresholds B. pertusa was very infrequently recorded. This temperature range is typical of most sub-humid and semi-arid regions of Queensland; as well as in India and south-east Asia, where B. pertusa is native.

Where it was most likely to achieve high cover, was in areas with a mean growing season temperature of 25 C˚ and in areas with < 10% FPC. Its dominance at this temperature may be in part associated with high germination success, with Howden (1988) finding 80% germination at 25 C˚, compared to only 10% at 20 C˚ and 30 C˚. These results align with most other plant species distribution models, which find climate rather than soil or terrain variables, are the most important predictors of plant species occupancy (Syphard and Franklin 2009). It is also well established that competition from trees (for light and nutrients) limits grass growth, particularly for shade intolerant species, such as B. pertusa (Jackson and Ash 1998, Setterfield et al. 2005).

For B. pertusa cover, rainfall variability was also a strong predictor, with high cover associated with moderate to high rainfall variability (index of variability ~1.0) (Table 3-1, Figure 3-4). This response aligns with anecdotal reports suggesting B. pertusa increases after cycles of drought, followed by a period of above average rainfall. Although B. pertusa is not considered particularly drought tolerant, its stoloniferous growth strategy and large seed bank enables it to quickly regain space and resources in response to improved growing conditions (Howden 1988, Ash et al. 2011). Conversely, many co-occurring native perennial grass species are considered drought tolerant. Livestock grazing severely compromises this tolerance however, with many studies finding a decline in basal area and survival of native grass species during drought, particularly on intensely grazed sites (McIvor 2007, Ash et al. 2011, Orr and Reagain 2011). Even in response to improved growing conditions, the rate of recovery and recruitment of these species was low, and they were often replaced by B. pertusa (Ash et al. 2011). Thus, drought and grazing-induced competitive release, combined with B. pertusa’s proficient colonising ability and high grazing tolerance, provides the ideal conditions for its proliferation.

Based on habitat suitability, B. pertusa still has considerable capacity to spread into new areas of Queensland and to increase in dominance within the coastal and sub-coastal regions

47 where it already proliferates. In particular, it may become more prevalent in the western (towards Longreach) and south-western (towards Roma) parts of the state (Figure 3-5), where its occurrence is still relatively rare (Figure 3-3e). It is not predicted to reach high cover in these areas however, and this is predominantly because these regions are outside of the optimal temperature and rainfall range for high B. pertusa cover (Figure. 3-4). Where it already occurs in western Queensland, these limitations to growth have been observed; for instance, on a conservation property near Longreach, B. pertusa is common but restricted to seasonally inundated and water run-on areas, suggesting the moister and likely cooler microclimates of these habitats allow for its establishment. Results from the HSMs also suggest B. pertusa is more likely to occur and achieve high cover closer to waterways (Table 3-1, Figure 3-3). Towards the Burnett region in the south, a combination of lower and less variable rainfall and high FPC in many areas has resulted in smaller and more dispersed patches of predicted suitable habitat, and lower predicted cover, than in northern Queensland. Predicted increases to temperature and rainfall variability under future climate change scenarios (IPCC, 2020) however, may improve the suitability of the Burnett and other areas of southern Queensland for B. pertusa occupation. Conversely in northern and western Queensland, temperatures may become too hot and rainfall too infrequent for B. pertusa to persist. Similar predictions have been made for C. ciliaris (Martin et al. 2015) and other tropical exotic grass species (Gallagher et al. 2013), that are expected to move southwards with the warmer winter temperatures predicted under climate change.

48

Table 3-1. Relative importance of environmental variables to models predicting B. pertusa occurrence (presence/absence) and cover. Variable are ordered in decreasing order of importance to the occupancy model. Variables assessed were; mean growing season temperature (MGST), foliage protective cover (FPC), distance to waterway (DTW), rainfall variability index (RVI), mean growing season rainfall (MGSR), land-zone, clay content (%) and clearing score. Variable Relative importance (%)

Occupancy Cover MGST (C˚) 36.6 29.4 FPC (%) 15.3 15.4 DTW (m) 12.5 10.1 RVI 11.6 17.7 MGSR (mm) 9.5 11.5 Land-zone 9.0 9.1 Clay content (%) 3.6 4.5 Clearing score 1.9 1.3

49

Figure 3-3. Partial dependence plots for the four most influential variables in the model for B. pertusa occurance (presence/absence). Variables are as follows; mean growing season temperature (MGST), foliage projective cover (FPC), rainfall variability index (RVI) and distance to waterway (DTW). Y-axes are on the logit scale and centred to have zero mean over the data distribution. Rug plots at inside top of plots show distribution of sites across that variable, in deciles. See Table 3-1 for the full list of environmental variables and their relative importance.

50

Figure 3-4. Partial dependence plots for the four most influential variables in the model for B. pertusa cover. Variable are as follows; mean growing season temperature (MGST), rainfall variability index (RVI), foliage protective cover (FPC) and mean growing season rainfall (MGSR). Y-axes are on the logit scale and centred to have zero mean over the data distribution. Rug plots at inside top of plots show distribution of sites across that variable, in deciles. See Table 3-1 for the full list of environmental variables and their relative importance.

51

Figure 3-5. Probability of occurrence (presence/absence) (left) and predicted broad cover (right) of B. pertusa in Queensland, as predicted by habitat suitability models, built using boosted regression trees. The resulting occurrence model explained 45% of cross-validated deviance, while the cover model predicted 38%.

Conclusions

This study presents a comprehensive analysis of the introduction, spread and potential distribution of an invasive exotic pasture species in Queensland, Australia. As with many invasive pasture species, B. pertusa was introduced on multiple occasions and across a large area of Queensland, despite re-occurring doubts and poor evidence for its benefit. It is clear from the evidence presented here that livestock grazing, associated disturbances (i.e. land clearing, soil erosion) and climatic extremes have facilitated the spread of B. pertusa in Queensland. Results from the HSMs suggest B. pertusa still has considerable capacity to spread and increase in dominance across Queensland. To curb these predictions, implementing sustainable grazing management practices (which allow adequate rest and rotation of pastures) and increasing grazing-protected areas are crucial. By helping to 52 maintain populations of native species, these strategies may help to improve the resistance and resilience of ecosystems to stochastic climatic events, and invasions by opportunistic invaders, such as B. pertusa. Changes to the frequency of B. pertusa and B. ewartiana on the Wambiana grazing trail, near Charters Towers, clearly highlights the potential for greater invasion resistance with more sustainable grazing management (Figure 3-1).

Bothriochloa pertusa is just one example whereby the poor evaluation and reporting of potential impacts enabled the widespread establishment and spread of an invasive species. Potential consequences for the environment were rarely considered and still today these impacts are poorly accounted for in pasture development programs (Cook and Dias 2006b, Driscoll et al. 2014). The implementation of a National Weed Risk Assessment System in 1997 has improved the regulation of new plant introductions into Australia. However, this system assesses the invasive potential of a species based on their life-history and ecology, and so data deficiencies may compromise the effectiveness of this approach. Further, to accurately ensure a species has low invasive potential, extensive post-introduction monitoring and control is required, and this is often where legislation falls shorts (Hulme 2012). Finally, this approach does not regulate the introduction and establishment of new cultivars and existing exotic species (Stone et al. 2008). Indeed, government agencies and research bodies still promote the establishment of high-impact environmental weed species, such as C. ciliaris and Leuceana leucocephala in Australia (The State of Queensland, 2021). The impacts of these species for ecosystems can be severe and improving their regulation and management should be a priority (Driscoll et al. 2014).

For B. pertusa and many other exotic pasture species, the pressure to enhance pastoral production, either by increasing stocking rates or improving live weight gains, has aided their widespread establishment and spread. Considering this, improving the capacity of landholders to manage for both conservation and pastoral imperatives is crucial and as hilighted in Chapter 2 and 7 there is considerable scope for where this can be achieved.

53

Chapter 4 – Damned if you do, damned if you don’t: Variable effect of grazing and vegetation type on the success of two invasive grass species

Abstract

Disturbance from livestock grazing can facilitate the spread of invasive plant species and declines in native plant diversity, particularly in regions with a short evolutionary history of grazing. This paper assessed the response of two widespread exotic invasive grass species, Bothriochloa pertusa (Indian couch) and Cenchrus ciliaris (buffel grass) to grazing using fence-line contrasts in long-ungrazed protected areas and their neighbouring pastoral properties, in central Queensland, Australia. We also examined the response of native floristic diversity to grazing and invasive grass cover. The surveys extended across two national parks and three distinct vegetation systems; the mountain coolabah and mixed Eucalyptus woodlands of Homevale national park and the Box (Eucalyptus populnea) woodlands of Dipperu. The cover of B. pertusa was consistently greater in the grazed pastoral properties than the protected reserves and its high cover in the Box and Mountain Coolabah woodland was associated with declines in native species richness. In contrast the cover of C. ciliaris was significantly greater in the grazing protected habitats of Homevale (not examined for Dipperu) and this was also associated with a significant decline in native diversity. Species richness overall did not differ between the protected areas and the pastoral properties; however, this result was likely confounded by contrasting impacts of the examined invaders. Managed livestock grazing has been proposed as a useful tool for managing C. ciliaris and our results support this. Importantly however we have highlighted that secondary invasion by B. pertusa is a potential consequence, particularly in susceptible ecosystems such as the mountain coolabah woodlands. Understanding how these two widespread invaders respond to different grazing regimes is an essential area of future research that will help to inform the use of grazing as a conservation management tool.

Introduction

Grazing is often found to facilitate the recruitment and dominance of invasive alien plant species over natives (Bock et al. 2007, Tozer et al. 2008, McAlpine et al. 2009). This is particularly true in systems with a short evolutionary history of grazing, whose native species

54 are less adapted to defoliation and easily displaced by grazing tolerant exotic species (Cingolani et al. 2005). In these systems, grazing resilience is minimal, and the introduction of grazers can lead to significant declines in native species diversity and irreversible changes to community composition and ecological function (Yates et al. 2000, Landsberg et al. 2003, Cingolani et al. 2005).

Having only been exposed to ungulate grazing in recent evolutionary history, Australia’s flora is considered particularly vulnerable to the intense grazing regimes now pervasive across most of the continent (Milchunas et al. 1988, Fensham and Skull 1999), and livestock grazing has been implicated in the spread of invasive species and declines in native flora across a variety of ecosystems (Abensperg-Traun et al. 1998, Fairfax and Fensham 2000, Yates et al. 2000, Hobbs 2001, Landsberg et al. 2002, Dorrough et al. 2004a, Grice 2006, Ash et al. 2011, Kutt and Fisher 2011). This inherent vulnerability to grazing is further compounded by the proliferation of exotic pasture species, whose introduction in Australia is tightly linked to pastoral practice (Cook and Dias 2006b). With livestock grazing the predominant land-use across Australia and with many of these exotic pasture species selected for their grazing tolerance and superior growth strategies (Driscoll et al. 2014), restricting their spread into and impact on native vegetation is an ongoing priority for conservation managers in Australia.

Because of the vulnerability of Australia’s flora to grazing, livestock grazing is usually prohibited within areas dedicated conservation reserves, although many have had long histories of grazing prior to their gazettal. These grazing protected areas provide important refuges for grazing-sensitive native plant communities and their associated fauna (Legge et al. 2011, Taylor et al. 2011, Frank et al. 2014, Craigie et al. 2015), and usually represent relatively intact vegetation fragments. Many reserves however exist within vast grazing lands (often dominated by exotic pasture) and the invasion by exotic species is ongoing. Understanding how grazing protection influences the dominance and impact of exotic invaders on native plant communities is important for improving conservation management within these fragmented landscapes. In this study we explore how protection from grazing has influenced invader cover and native species diversity within the fragmented agricultural landscape of central Queensland.

55

Specifically, we conduct surveys within two national parks, Homevale and Dipperu, and their adjacent pastoral properties in north Queensland, to examine differences in the cover of two prominent exotic invasive grass species in the region, Bothriochloa pertusa and Cenchrus ciliaris. This study also investigates the response of native species diversity to grazing protection and invasive grass cover. The survey extends across three distinct vegetation types and we explore results in relation to these. Prior to destocking, both Homevale and Dipperu were managed for livestock (mostly cattle) grazing. Although we don’t have data on the floristic diversity and composition in the reserves prior to livestock grazing, we can compare their current floristics with that of adjacent pastoral properties to assess their value in restoring native species diversity and reducing the establishment and spread of exotic species.

The following research questions are addressed:

1) How does long-term protection from grazing influence the cover of focal invasive grass species?

2) How does long-term protection from grazing and the cover of focal invasive grass species influence the abundance and richness of native flora?

3) How does the cover of focal invasive grass species influence the abundance and richness of native flora?

Methods

Study area

The study was conducted on the boundary of Homevale and Dipperu National Parks and their adjacent pastoral properties in north Queensland, Australia (Figure 4-1). In 1980 and 1995 livestock were removed from Dipperu and Homevale respectively and their management handed over to Queensland Parks and Wildlife Service for the preservation of their natural and cultural values. Dipperu protects a number of threatened ecosystems, including one of the largest remnants of Acacia harpophylla (brigalow) and Eucalyptus populnea (poplar box) woodland. Homevale also protects areas of A. harpophylla, as well endangered semi-

56 evergreen vine thicket and mixed Eucalypt woodlands. The climate at Homevale and Dipperu is sub-tropical, with a similar mean annual rainfall of approximately 650 ml and the majority of rain falling from October to March (Bureau of Meteorology, 2018). Floristic surveys were conducted in March 2018 for optimal floristic expression.

The survey at Dipperu was conducted on either side (grazed; long-ungrazed) of the northern boundary fence on texture-contrast soils dominated by Eucalyptus populnea. The survey at Homevale was conducted on either side (grazed; long-ungrazed) of the western boundary fence and extended across two adjacent vegetation types; an open woodland of Eucalyptus orgadophila (Mountain coolabah) and an open woodland with mixed Eucalyptus and Corymbia species (E. crebra, E. populnea, E. orgadophila and C. clarksoniana). The three vegetation types across the two national parks will hereafter be referred to as poplar box, Mtn. coolabah and mixed, respectively. Exploratory analyses of particle size (Appendix 4-1) revealed distinct differences in soil texture between the three sites (Table 4-1).

Table 4-1. Difference in mean percentage clay, silt and sand between the three surveyed vegetation types; mixed, Mtn. Coolabah and poplar box. Letters determine significantly different means as determined from post-hoc analysis of variance. Methods can be found in Appendix 4-1.

Homevale National park Dipperu National park Mixed Mtn. coolabah Poplar box Clay (%) 15.3 a 24.3 b 8.4 c Silt (%) 21.0 a 21.6 a 14.4 b Sand 63.6 a 53.7 b 77.2 c (%)

Study species

Two exotic invasive pasture grasses were studied, Cenchrus ciliaris and Bothriochloa pertusa, which were both introduced for pasture improvement in the mid-1900’s. Both species can co-occur and persist under a broad range of soil types.

57

C. ciliaris is a tall perennial tussock grass species, native to Africa and Asia. It is well adapted to the semi-arid and arid environment of Australia and has been sown extensively throughout these regions for fodder, usually in conjunction with broad scale clearing of woody vegetation. It forms dense swards and is often associated with declines in diversity and increased fire frequency and intensity (Jackson 2005, Grice 2006, Miller et al. 2010). There is evidence to suggest C. ciliaris increases in abundance with livestock grazing (Fensham et al. 2013), however in some protected areas grazing has been used to reduce its biomass and consequent flammability (Melzer and Melzer 2017). B. pertusa is a stoloniferous, perennial grass species native to eastern Asia. As a less productive species, B. pertusa has not been sown to the extent of C. ciliaris, however has successfully spread and is abundant throughout vast areas of north and central Queensland. The species has been associated with significant declines is floristic diversity and changes to faunal assemblages (Kutt and Fisher 2011, Kutt and Kemp 2012). B. pertusa is considered grazing tolerant and typically increases under intense grazing regimes (McIvor 2007a; McIvor et al. 1982). C. ciliaris and B. pertusa were prevalent across both Homevale and Dipperu and no active control measures were in place for either species at the time of survey. In the Box woodland at Dipperu however, C. ciliaris was in very low abundance and only the cover B. pertusa was assessed.

Experimental design

Within both national parks, a mostly contiguous patch of each vegetation type with a minimum area of 50 ha and encompassing both the grazed and ungrazed sides of the boundary fence was surveyed to assess the impact of long-term grazing protection on 1 ) invader cover and 2) native floristic diversity and abundance.

Invader response to grazing protection (Question 1)

To assess the impact of grazing protection on invader cover, a series of paired 20 m line transects were surveyed either side of the boundary (grazed; long-ungrazed) within each vegetation type. The number of paired transects varied depending on the extent of the vegetation type within the survey area (11 pairs in Mtn Coolabah, 13 pairs in mixed woodland, 32 pairs in poplar box) (Table 4-2). Transects ran perpendicular to the boundary

58

fence with the start point (0 m) positioned 50 m from the boundary fence. Paired transects were position 50 m apart. Invader cover was assessed using the point intercept method every meter along the transect.

Diversity response to grazing protection (Question 2)

To assess the impact of grazing protection on floristic diversity and abundance, a detailed floristic survey was conducted within paired 7 m × 2 m plots, either side of the boundary (grazed; long-ungrazed). Plots were positioned 50 m in from the fence and spread out along the extent of the line transects examined for Question 1 (Figure 4-1b). Again, the number of paired floristic plots surveyed varied depending on the extent of the vegetation type within the survey area (17 pairs in the Box, 9 pairs in the Mtn. Coolabah and 10 pairs in the Mixed woodland; Table 4-2).

Each plot was separated into four sub-plots of increasing size (0.3 m2, 1 m2, 3m2, 7m2 ) and data on total species richness and functional group richness and abundance collected (Figure 4-1c). Functional groups comprised; annual grasses, annual forbs, perennial grasses and perennial forbs. Plant species in the first subplot were assigned an abundance score of 4, additional species present in the second subplot an abundance of 3, additional species in the third subplot an abundance of 2 and the final subplot an abundance of 1. Functional group abundance values are the species cumulative abundance scores for each plot.

Table 4-2. Total number of paired line-transects and floristic plots sampled in each national park and vegetation type. National Park Vegetation Number of paired Number of paired 1 line-transects floristic plots2 Dipperu Poplar box woodland 32 17

Homevale Mixed Eucalyptus woodland 13 9

Homevale Mountain Coolabah woodland 11 10

1 Data used to address Question 1 2 Data used to address Question 2

59

Diversity response to invader cover (Question 3)

The assess the impact of B. pertusa and C. ciliaris on floristic diversity, the point intercept method was used to assess invader cover every meter around the perimeter and centre of each plot. A plot proportion cover was attained by dividing the total hits by the total possible hits (24).

Across the three vegetation types the occurrence of other exotic species was relatively low. As this study was only interested in the response the two focal invaders and native species to grazing protection, other exotic species were recorded but excluded from analyses.

Figure 4-1. a) The location of Homevale and Dipperu National Parks in Queensland, Australia. The cross indicates the survey locations. b) The layout of 20 m line transects (solid lines) used to assess differences in invader cover between the grazed pastoral property and the long-ungrazed national parks (Question 1), and the layout of detailed floristic plots (grey boxes) used assess the response of native diversity to grazing protection (Question 2). c) Each floristic plot was split into four subplots of increasing size and assigned species abundance scores as illustrated. Cover of B. pertusa and C. ciliaris was determined using the point intercept method every meter around the perimeter and centre of each plot (black dots) (Question 3). 60

Statistical analysis

To examine invader response to grazing protection (Question 1), cover (for both B. pertusa and C. ciliaris) was modelled as a two-column vector containing the number of point intercepts along the line transect that were covered by the invader, and the number not covered by the invader. Both cover vectors were then modelled separately as a function of grazing and vegetation type, and their two-way interaction using general linear mixed-effect models with a binomial error distribution, logit link function and transect pair included as a random effect.

To examine the response of native diversity to grazing protection (Question 2), total species and lifeform richness and abundance were modelled as a function of grazing and vegetation type and their two-way interaction, using a series of general linear mixed effect models with a negative binomial distribution transect pair as random effects. The abundance of annual grass was too low across all vegetation types to reliably determine a treatment effect and thus excluded from further analyses.

To analyse the impact of B. pertusa and C. ciliaris cover on native species diversity (irrespective of vegetation type) (Question 3), diversity indices were modelled as a function of invader cover using a series of general linear models with a negative binomial distribution to account for overdispersion of errors. Because only B. pertusa was recorded at Dipperu, we analysed invader impact on diversity separately for both national parks. Consequently, for each floristic response variable two models were performed; one with B. pertusa cover as in individual fixed effect (Dipperu only) and another with both C. ciliaris and B. pertusa cover as individual fixed effects (Homevale only). Their interaction was not included in the models as their co-occurrence in plots was relatively rare.

Results

Question 1) How does long-term protection from grazing influence the cover of two exotic grass species

61

B. pertusa cover was significantly less in the long-ungrazed Box and Mtn. Coolabah woodlands than the grazed (P = <0.001). This negative grazing response was far greater in the Mtn. Coolabah than in the Box woodland (P = <0.001). B. pertusa did not differ significantly between grazed and long-ungrazed in the Mixed woodland. C. ciliaris was significantly greater in both the long-ungrazed Mtn. Coolabah and Mixed woodlands than the grazed (P = <0.001).

Figure 4-2. The mean cover of (a) B. pertusa and (b) C. ciliaris in grazed (green) and long- ungrazed (blue) plots across the three vegetation types; poplar box, Mtn. coolabah and mixed. Points indicate the probability of occurrence with 95% confidence intervals.

Question 2) How does long-term protection from grazing influence the abundance and richness of native flora.

155 species were recorded across the three vegetation types; 105 species in the Box, 79 in the mixed and 68 in the Mtn. Coolabah woodlands. Of these an additional 13 exotic species were recorded across the three vegetation types, however at very low rates of occurrence compared to B. pertusa and C. ciliaris. These were excluded from analyses. Total and lifeform species richness did not significantly differ between grazed and long-ungrazed, across all vegetation types. The abundance of some lifeforms differed between grazed and long-ungrazed, however this was vegetation dependent. In the Box woodland the abundance of annual herbs was higher in the grazed than long-ungrazed (P = <0.05). Perennial herbs were significantly less abundant in the long-ungrazed Box woodland, which was in contrast to the Mtn. Coolabah were perennial herbs increased with grazing (P = <0.01). The abundance of perennial grass species did not significantly differ between grazed and long-ungrazed across 62 all vegetation types, however this result was highly confounded by the cover of B. pertusa and C. ciliaris.

Question 3) How does the cover of focal invasive grass species influence the abundance and richness of native flora.

Models exploring the impact of B. pertusa cover on diversity indices at Dipperu were not significant. At Homevale invader cover was associated with significant changes to native species assemblages. Total species richness was significantly negatively related to C. ciliaris cover (P = <0.01), but unrelated to B. pertusa cover. The richness and abundance of perennial grass species was significantly negatively related to the cover of both invaders (P = <0.05). The abundance, but not richness of perennial forb species was positively related to B. pertusa cover (P = <0.05). The richness and abundance of annual forbs was unrelated to the cover of either invader.

Discussion

This study has demonstrated a variable effect of grazing on the establishment of two prominent exotic grass species in Queensland, Australia. C. ciliaris cover was greatest in the grazing protected national park, across both vegetation types surveyed. In contrast, B. pertusa was generally facilitated by grazing, particularly in the Box and Mtn. Coolabah woodlands. Both invaders were associated with significant declines in native species richness and the richness of some functional groups. There was a negligible response of total species richness to grazing protection, likely reflecting the confounding effect of invader cover and perhaps the legacy of an intense grazing history.

Invader response to grazing protection

B. pertusa was generally facilitated by grazing, occurring at very low abundance in the grazing protected vegetation of the national parks. This result is consistent with previous findings, which have described B. pertusa as a ‘passenger of change’, usually increasing as a result of altered disturbance regimes rather than through competitive exclusion of native species (Scanlan et al. 1996, Kutt and Fisher 2011). As a low growing stoloniferous species

63

B. pertusa is well equipped to avoid herbivores and maintain abundance through rapid clonal growth. Although generally increasing across all vegetation types, it was considerably more abundant in the Mtn. Coolabah woodland. The Mtn. Coolabah soil contained significantly more clay than the sandier soils of the Mixed and Box woodland soils (Table 4-1), perhaps indicating a preference for the generally enhanced water and nutrient holding capacity of clay soils. Alternatively, the dominant native species in the Mtn. Coolabah woodland, Bothriochloa ewartiana is particularly vulnerable to grazing induced decline and commonly displaced by its grazing tolerant exotic cousin B. pertusa, throughout other areas of Queensland (Bartley et al. 2014).

The cover of C. ciliaris cover was significantly greater in the grazing protected national park (across both Mtn. coolabah and Mixed), than the grazed pastoral property. This is consistent with findings from reserves elsewhere in Queensland (Baxter et al. 2001, Melzer 2015). Grazing may be an effective management strategy to reduce the dominance of high-biomass palatable invasive grasses such as C. ciliaris, however as this study highlights, it is important to consider the potential for secondary invasion by other invasive species. Re-introduction of grazers into the Mtn. Coolabah woodland of this study for instance, may facilitate the establishment of B. pertusa, a result known to occur in other areas of Queensland. For example, on a conservation reserve, near Emerald, central Queensland, a regime of intense cattle grazing has been implemented to reduce the biomass of C. ciliaris and consequential fire risk, to support the regeneration of threatened A. harpophylla (brigalow) woodland. This has led to the desired reductions in C. ciliaris but an increase in B. pertusa, which is readily colonising the available niche space (R. Diete 2019, pers. comm). This provides a tricky predicament for land managers, and perhaps additional management strategies, such as re- populating the seed bank with native species is necessary to maintain diversity in the understorey (Buisson et al. 2019, Sampaio et al. 2019).

In the grazed Mixed woodland, the cover of C. ciliaris and B. pertusa was relatively low and compared to the national park, moderately more diverse. This may be an indication of grazing moderating the effect of C. ciliaris on diversity in this environment. Whether this system is less vulnerable to B. pertusa invasion cannot be confirmed with the data collected in this study. Additional floristic surveys, and information about grazing utilisation and history within the area would be necessary to conclude this. As there were some plots within the

64 iron-bark woodlands that were very heavily invaded by B. pertusa, it may be that over time B. pertusa will invade this system similarly to the Mtn. Coolabah.

Diversity response to grazing protection

Both Homevale and Dipperu national parks have had long histories of livestock grazing prior to their gazettal. This study has shown that despite their long-term protection from ungulate grazing, total species richness was similar to neighbouring pastoral properties. The abundance and richness of functional groups were also similar; although ephemeral species were positively associated with grazing, likely a response to the reduced abundance of perennial grass species and consequential increase in light and space for their establishment. The negligible effect of grazing on diversity is likely due to the confounding influence of the two focal invaders, which were collectively abundant across the national parks and pastoral properties. Other legacy effects from an intense grazing history may have also contributed to the minimal recovery of native species with grazing protection in these national parks. For instance, degradation to the soil surface can change water and nutrient dynamics, often creating conditions unsuitable for plant growth (Yates et al. 2000). Exposed soil horizons are very difficult to restore and can remain unvegetated even when livestock are removed (Zeidler et al. 2002, Jauffret and Lavorel 2003, Cingolani et al. 2005).

Similar negligible responses to grazing protection have been found in other degraded systems across Australia (Lunt 2007, Souter and Milne 2009). For instance, in an area of Eucalyptus camaldulensis (River red gum) riparian woodland in south-eastern Australian, Lunt (2007) found vegetation composition and structure to remain similar even after 12 years of grazing exclusion. This study suggest a history of heavy stocking and altered flooding and burning regimes has led to severe degradation of these systems and recommend more complex management strategies, such as re-planting native species, are necessary to improve the diversity and abundance of native species in these systems. In contrast, systems which have short degradation histories and relatively intact native plant communities, grazing exclusion has helped to maintain native species diversity and reduce the establishment of exotic invasive species (Legge et al. 2011, Frank et al. 2014). These findings highlight the need for different management strategies to either restore native species within degraded systems or conserve native species within intact systems. It may be that the areas examined in this study

65 have surpassed a threshold of degradation and require more intensive approaches to restore native floristic diversity.

Diversity response to invader cover

Both C. ciliaris and B. pertusa were associated with significant differences in native species assemblages. With increasing B. pertusa and C. ciliaris, perennial grass richness and abundance significantly declined, a result consistent with previous research (Jackson 2005, Kutt and Kemp 2012, Fensham et al. 2015b). Species with similar functional strategies compete more strongly for resources and are less likely to coexist, particularly when resources are scarce (MacDougall and Turkington 2005). In addition, preferential grazing of palatable native perennial grass species may have led to their decline and associated increase in the less palatable, grazing avoidant B. pertusa. The prostrate stoloniferous growth form of B. pertusa may favour weaker light competitors, possibly explaining the observed increase in perennial forb species with B. pertusa cover. Conversely, the high biomass producing C. ciliaris, led to significant declines in both perennial grass and overall species richness, an indication of this species superior ability to occupy niche space and reduce opportunities for subordinate species to establish, particularly in the absence of livestock grazing. Associated increases in fire frequency and intensity with C. ciliaris invasion has also likely led to changes in native species assemblages (Grice 2006, Miller et al. 2010).

Conclusion and management implications

This study has demonstrated a contrasting influence of grazing on the abundance of two prominent invaders in Queensland, Australia, presenting a challenging predicament for land managers. B. pertusa and C. ciliaris are both very abundant throughout large areas of Queensland, often co-occur and are associated with significant declines in native species diversity. Managing their spread is crucial for the conservation of biodiversity across the vast area and ecosystems they invade.

We have shown that protection from grazing can reduce the establishment of B. pertusa, however can also promote the dominance of C. ciliaris. The extent of C. ciliaris in these reserves is too large for traditional control strategies (i.e. broad scale herbicide application) to 66 be effective, not to mention expensive and labour intensive. As has been trailed in other conservation reserves, the re-introduction of a controlled grazing regime in these ecosystems may assist in the recovery of native biodiversity by reducing competition from C. ciliaris and reducing the incidence of intense fire. This also presents an opportunity to work collaboratively with neighbouring pastoralists, an outcome which can help establish social networks and foster knowledge exchange between land managers. This should not be the primary solution however, and needs to be integrated alongside other strategies, such as herbicide application, re-populating native seed banks and controlled use of fire (Delaney et al. 2016, Bryant et al. 2017, Sampaio et al. 2019, Wohlwend et al. 2019).

Areas protected from livestock grazing are essential for the conservation of native biodiversity. The scant resource budget of many of the reserves however limits their ability to implement appropriate management and monitoring regime. Under such resource limitations and in an extremely modified and threatened landscape, managing protected areas requires an adaptive and collaborative approach, or we risk losing what we set out to protect.

67

Chapter 5 – Controlled livestock grazing for conservation outcomes in a fragmented landscape Abstract

Protecting native biodiversity is a difficult prospect in extremely modified landscapes, especially where high-impact exotic species are widespread. Using new data and a review of the literature, this paper comments on the use of livestock grazing to manage the invasive and highly combustible pasture grass species, Cenchrus ciliaris (buffel grass) and thereby help conserve fire-sensitive Acacia harpophylla (brigalow) vegetation in reserves in Queensland, Australia. We cite evidence that shows that grazing is a potentially useful management tool in such cases and its use can be compatible with the protection of both fire-sensitive vegetation and other native plant species within the understorey. However, there are a number of limitations on the implementation of grazing within conservation reserves including the lack of a clear understanding of the influence of grazing on biodiversity and resource condition . Importantly, we highlight secondary invasion by the exotic grass Bothriochloa pertusa (Indian couch) as a key emerging threat that may undermine the biodiversity benefits gained by grazing in reserves. Grazing can be a useful tool for conservation management in particular scenarios, but the associated risks demand accompanying monitoring and reporting of positive and negative impacts to ensure the fundamental aim of biodiversity protection is being achieved.

Introduction

There is a long-standing debate about the compatibility of domestic livestock in conservation areas (WallisDeVries, et al. 1998). In Australia the argument would seem to be clear-cut. A diverse assemblage of megafauna existed during the Pleistocene, but these soft-footed marsupials have been extinct for at least 40,000 years (Roberts, et al. 2001). Smaller macropod species replaced them as the dominant herbivores in the landscape, however their densities were probably regulated by predation and water availability (Silcock, et al. 2013). As such, the continent of Australia is generally considered as having a light evolutionary history of grazing and its biodiversity potentially vulnerable to grazing by commercial livestock (Díaz, et al. 2007; Morton, et al. 1995).

68

The negative impact of livestock grazing on Australia’s biodiversity is frequently highlighted (Kutt and Fisher 2011; Kutt and Gordon 2012; Legge, et al. 2011; Reside, et al. 2017; Woinarski, et al. 2015) and has been implicated in the significant decline of small to medium-sized mammals (Fitzsimons, et al. 2010; Legge, et al. 2011) and some flora species (Silcock and Fensham 2018). As there is limited knowledge about the composition of flora pre-European contact, the full extent of biodiversity decline due to grazing is difficult to determine and it is likely many more species than those recorded have experienced declines and extinctions as a result of the introduction of high densities of hard-hooved grazers to Australia. The use of ungulate livestock therefore seems to be an anathema to conservation management in Australia.

However, many conservation reserves in Australia were grazed by livestock for long periods prior to their establishment and generally contain a mixture of native and exotic grazing- tolerant plant species (Cook and Dias 2006; Driscoll, et al. 2014). Furthermore, reserves often represent fragments of native vegetation in highly modified landscapes and invasion by exotic plant species is ongoing. Invaders in these systems are usually pasture species which have been introduced across the landscape for their grazing tolerance and rapid growth relative to native herbaceous species (Driscoll, et al. 2014). The impact of these invasive pasture species on native plant communities is diverse, with many leading to declines in native species diversity (Driscoll, et al. 2014; Grice 2006) and changes to ecological processes such as fire regimes (D'Antonio and Vitousek 1992; Setterfield et al. 2010) and nutrient cycling (Rossiter - Rachor et al. 2009). Restricting their spread and impact on native vegetation is challenging and traditional methods of control (e.g. herbicide) are often not practical. In these degraded systems, managed livestock grazing may be the most viable option for managing the impact of invasive exotic species and maintaining native biodiversity (Lunt, et al. 2007).

This paper comments on the use of livestock grazing to manage the invasive pasture grass species, Cenchrus ciliaris (buffel grass) and thereby help conserve fire-sensitive vegetation (FSV) within reserves in Queensland. We use examples from the literature, as well as new data to first highlight the extent of a fire management issue within Queensland, which tends to increase with increasing cover of C. ciliaris and also impact upon fire sensitive species. We then examine the compatibility of using grazing as an alternative management tool with

69 the conservation of native understorey species. Importantly, we highlight potential limitations to grazing in these instances, with particular consideration of the emerging problem of invasion by Bothriochloa pertusa (Indian couch), another prominent invader in Queensland.

Restoring and conserving fire-sensitive vegetation in Queensland

In recent decades, Queensland has been a global hotspot for deforestation and habitat fragmentation (McAlpine, et al. 2009). An increase in the extent and profitability of managed grazing systems has been the main driver of broad scale conversion of native woodlands to pastures dominated by exotic species. Fertile woodlands dominated by Acacia harpophylla (brigalow) have been the most affected, with only 10% of their former range remaining (Fensham, et al. 2017).

Where these woodlands are intact, they have little production value, however once cleared, burnt and sown to non-native pasture species they can support greater livestock productivity. The pasture species that has proved most successful is Cenchrus ciliaris (buffel grass) which is now a critical fodder resource, crucial to the prosperity of the cattle grazing industry in these cleared ecosystems. Clearing often leaves behind young trees and the soil disturbance can provoke root-suckering, particularly in A. harpophylla woodlands (Johnson 1964). The regrowth recovers to varying degrees and may eventually reform a structure akin to the original forest (Dwyer, et al. 2010a; Dwyer, et al. 2010b).

Forests and woodlands of A. harpophylla and associated species are however fire-sensitive, and they are, in their remnant condition, less prone to fire due to their high canopy cover and sparse grass cover. Although fire can often promote suckering in A. harpophylla, repeated hot fires can damage to root systems and eventually prevent regeneration (Johnson 1964). At the higher rainfall end of its distribution A. harpophylla woodland can also grade into semi- evergreen vine thickets which is also a highly fire-sensitive and threatened ecosystem. The persistence and restoration of these fire-sensitive ecosystems within fragmented landscapes can therefore be severely compromised by invasive grasses, such as C. ciliaris, which if left unmanaged can fuel intense wild-fires. Protecting remaining fragments and facilitating the restoration of these threatened ecosystems is a priority for many reserves within Queensland.

70

The extent of the problem

Using the Queensland’s vegetation mapping series alongside field inspections, we identified the national parks where the restoration of previously cleared (non-remnant) fire-sensitive vegetation (FSV) is likely threatened by C. ciliaris invasion and its associated fire risk (methodology found in Appendix 1). Within the FSV (remnant and non-remnant) of these national parks, we also calculated the total burnt area and total number of fire events that occurred between 2003 and 2020, using the Northern Australian and Rangeland Fire Information (NAFI) fire scar imagery.

These analyses suggest that the restoration of 7533 ha of non-remnant FSV, spanning across 19 National Parks is likely threatened by C. ciliaris. We also found that over the last 13 years, 38,292 ha of FSV burnt during 157 fire events, which included areas of remnant vegetation. Although we did not assess C. ciliaris invasion within remnant patches, the degree to which these areas can be invaded by C. ciliaris will vary but are usually adjacent to heavily invaded non-remnant patches. The occurrence of fire within non-remnant areas can erode the edges of remnant areas by killing trees and damaging root systems, potentially reducing their resilience to future fire events (Johnson 1964).

Eliminating ignitions as a means of reducing destructive fires is only feasible over small areas given the inevitability of lightning strikes (Kuleshov, et al. 2006). Considering cost, reducing fuels with slashing, firebreaks or herbicide is also only practical at small scales. The most efficient large-scale, fuel-reduction strategy is management with livestock grazing. Grazing to reduce fuel loads in C. ciliaris pastures and protect woody regrowth has been occasionally implemented across some reserves in Queensland (Melzer 2015). However, it remains unclear if the level of grazing needed to reduce fuel loads is compatible with the protection of other conservation values.

Is grazing Cenchrus ciliaris (buffel grass) compatible with understorey diversity?

Cenchrus ciliaris has the capacity to form dense swards and competes strongly with native plant species for light and nutrients, often reducing their abundance and diversity (Fairfax and Fensham 2000; Fensham, et al. 2015; Jackson 2005). In some environments native perennial

71 grass species can also form dense swards and historically their dominance has been regulated through fire regimes and macropod grazing (Bradstock, et al. 2012; Lunt and Morgan 2002; Morgan 1998; Tremont 1994). With these disturbances, space and resources are made available for less competitive, often ephemeral species (Andersen, et al. 2005; Lebbink, et al. 2018). These events replenish the seed bank and are crucial for the persistence of these lifeforms. In degraded systems where native perennial grasses have been displaced by C. ciliaris, management strategies that reduce the dominance of C. ciliaris, such as grazing, may similarly increase establishment opportunities for native species.

At Taunton National Park (Scientific) in central Queensland, intense short periods of grazing (pulse grazing) at the end of the summer growing season, has been implemented in some buffel-dominated areas to help restore fire-sensitive regrowth, but also to maintain habitat and food species for the critically endangered Bridled Nailtail Wallaby (Onychogalea fraenata) (Figure 5-1). Forb species constitute a significant portion of diet of Bridled Nailtail Wallaby (Dawson, et al. 1992) and these forbs occur less frequently in areas of high C. ciliaris cover presumably due to competition for light (Fairfax and Fensham 2000; Melzer, et al. 2014). By reducing the cover of C. ciliaris, both grazing and herbicide have been effective at promoting herbaceous species, however due to the extent of C. ciliaris invasion within the park, grazing is generally considered a more practical management option (Lowry and Ryan 2018; Melzer, et al. 2014; Melzer 2015). Similar grazing-induced increases in understorey species richness were found in A. harpophylla and Eucalyptus woodlands invaded by C. ciliaris in south-western Queensland (Baxter, et al. 2001). At these sites there was a higher diversity of grasses in areas variably grazed (generally a continuous grazing regime of one animal equivalent per four ha) by cattle than those where cattle had been excluded, and this higher diversity was associated with lower C. ciliaris cover. These two examples of managed grazing in A. harpophylla suggest that this may be a useful tool for managing C. ciliaris invasion and its associated fire risk, while also maintaining understorey floristic diversity.

72

Figure 5-1 Acacia harpophylla (brigalow) restoration monitoring site at Taunton National Park (Scientific) in 2012 (a) and 2017 (b). Managed pulse grazing commenced at the site in 2014 to reduce Cenchrus ciliaris (buffel grass) cover and its associated fire risk. Note dense C. ciliaris understorey in 2012 (a) and reduced C. ciliaris cover and maintained A. harpophylla regrowth in 2017 (b). Fire severely compromises the restoration of fire-sensitive ecosystems such as A. harpophylla. Photos reproduced from Melzer and Melzer 2017.

Implications and considerations

In our view, the area where grazing may be useful for managing C. ciliaris invasion and its associated fire risk in reserves is extensive. However, there are a number of detrimental impacts that may be associated with grazing in reserves.

Firstly, the costs involved with livestock management can be substantial. The installation of appropriate infrastructure (watering points, fences etc.), monitoring of animal welfare and mustering costs are all potentially considerable and ongoing. Secondly, even with careful development of infrastructure, livestock do not graze evenly across space and areas of high fuel loads will still persist within the landscape (Pringle and Landsberg 2004). Whether the level of grazing heterogeneity is sufficient to reduce the spread of wildfire will vary case-by- case and depends on fencing, watering points, the mosaic of vegetation types and the distribution and density of other herbivores (Pringle and Landsberg 2004). Generally however, fire-sensitive regrowth vegetation occurs on relatively fertile soils supporting dense grass swards (Fensham, et al. 2017) and will otherwise be selectively grazed where it occurs in paddocks with other vegetation types. Another problem is that in wet seasons when fuel is 73 abundant there is an abundance of pasture and little imperative to move livestock onto reserves under agistment arrangements.

An emerging consequence of managed livestock grazing in Queensland is the potential of secondary invasion by the exotic grass Bothriochloa pertusa (Indian couch). Similar to C. ciliaris, this species has been associated with significant declines in floristic diversity (Kutt and Kemp 2012) and changes to fauna assemblages (Kutt and Fisher 2011) and can successfully invade a broad range of habitats, particularly when heavily grazed (Scanlan, et al. 1996). Where grazing has been used for the management of C. ciliaris and fire-sensitive regrowth on a reserve in central Queensland, B. pertusa has been colonising the grazing induced niche gaps (pers comm. Rebecca Diete).

In Chapter 4, we assessed differences in C. ciliaris and B. pertusa cover and the diversity of native species within an ungrazed conservation reserve (Homevale National Park, central Queensland) and a neighbouring pastoral property managed for cattle production. There was significantly lower C. ciliaris but higher B. pertusa cover in the pastoral property than the reserve. This analysis also shows the richness and abundance of some native species, particularly perennial grasses, were negatively related to the cover of both invaders. With both grasses abundant across large areas of Queensland, this will represent a common dilemma, and may be an overwhelming disadvantage of using grazing as a management tool unless other solutions (such as early treatment of B. pertusa with herbicide) can be successfully implemented. Future research aimed at understanding the grazing response of B. pertusa and C. ciliaris across the diverse habitats they invade will help to guide the management of these problematic and wide-spread grasses.

Conclusion

Livestock grazing can be a useful tool for managing high-biomass exotic grass species and thereby maintaining native species and habitats within conservation reserves. In Queensland this is particularly relevant due to the extent of C. ciliaris invasion and the priority of many reserves to conserve FSV. Secondary invasion by grazing tolerant species is a potential consequence highlighted here, and this may limit the use of grazing for conservation management. Where grazing is considered an appropriate management tool, adaptive

74 management that includes opportunistic adjustments and ongoing monitoring and reporting of positive and negative impacts should be implemented. Complex management dilemmas such as this largely originate from, and are compounded by, intense human land-use and in these landscapes the conservation of native species will continue to represent a major challenge.

75

Chapter 5 is published in the Journal of Environmental Management and Restoration.

Lebbink, G.H., Dwyer, J., Fensham, R. J. (2020). Controlled livestock grazing for conservation outcomes in a fragmented landscape. Ecological Management and Restoration 22:5-9.

Written by GL, with input and reworking from RF and JD. The field work and analyses from the referenced appendices and Chapter 4 was conducted by GL.

76

Chapter 6 – An invasive grass species has both local and broad-scale impacts on diversity: Potential mechanisms and implications

Abstract

The impact of invasive plant species on native diversity varies with spatial scale, with some invaders leading to broad-scale diversity declines and others only local declines. These discrepancies may reflect the invaders capacity to reduce niche opportunities across spatial scales which can be associated with their functional traits. We investigated impact-scale relationships and trait-based mechanisms, in areas invaded by the exotic perennial grass species, Bothriochloa pertusa, in the ‘ironbark’ woodlands of eastern Queensland, Australia. We examine root traits specifically, as belowground competition was considered particularly important to the success of this species. To assess invader impact across spatial scale (up to 1000 m2) we analysed changes to the species area relationships (SARs) across sites variably invaded by B. pertusa (very low to very high cover). Changes to SARs were assessed in relation to the invaders effect on rare (low-patch-occupancy) and common (high-patch- occupancy) species in the community. In a separate analysis within the same habitat we collected root cores across a gradient of invader cover and analysed changes to community root traits that were considered important correlates of competition for space and nutrients. Invasion-induced reductions in diversity were pervasive at all scales investigated, and this was associated with a proportionally greater effect on rare species in the community. In the separate root analysis, changes in community root traits with increasing invader cover were indicative of more intense competition for resources rather than space. The observed regional-scale dominance of B. pertusa and associated declines in diversity warrant serious concern for the conservation of native plant communities and species in a region already at risk from other anthropogenic threats. Intense competition for belowground resources is likely a contributing mechanism to the success of B. pertusa in this study system. Experimental examination of this and other mechanisms would help to validate these findings.

Introduction

77

The impact of invasive flora is usually perceived as being negative for biodiversity and ecosystem processes (Grice 2006, Vilà et al. 2007, Gaertner et al. 2009). There are some instances however where invasions have led to an increase in species richness or had no influence, which has led some researchers to question whether most invasive species are really a leading threat to native biodiversity (Simberloff and Von Holle 1999, Davis et al. 2005, Gilbert and Levine 2013). It has been suggested that these contradictions may be a consequence of the spatial scale used for investigation (Davies et al. 2005, Towers and Dwyer 2018). For instance, invaded communities may have lower local richness but steeper species accumulation with increasing area than that of uninvaded communities, leading to proportionately fewer species lost at broader spatial scales (Powell et al. 2013). In some cases, however, invasion-induced richness declines may persist across large spatial scales, with significant consequences for diversity (Hejda et al. 2009, Fensham et al. 2015b). How the invader affects rare (low-patch-occupancy) and common (high-patch-occupancy) species in the community can help to explain these differences in invader impact across spatial scale (Powell et al. 2011).

Rare species often occupy specialised habitat refuges created by environmental heterogeneity, and the likelihood of encountering these refuges generally increases with spatial scale (Marvier et al. 2004, Davies et al. 2005). As the effect of the invader changes from proportionally greater effects on common species to rare species, the potential for large scale declines in diversity increases (Powell et al. 2011). This is because the number of rare species generally increase with spatial scale and these species are more vulnerable to local extinction because they occur in fewer patches (Gaston 1998, Hanski 1998). This is conditional on the overall number of rare and common species in the community however; communities with fewer rare species are less vulnerable to invasion-induced richness declines at large scales, as common species are more likely to persist, albeit likely at lower abundance, within invaded communities (Hejda and Pyšek 2006, Powell et al. 2011).

In situations where environmental conditions are suitable for the invader to maintain high patch occupancy across large spatial scales, the realised impact on species (regardless of their rarity) will depend on both the traits of the invader and traits of the recipient community. For instance, species with traits associated with rapid growth and high cover, such as clonal reproduction, tall stature and high specific leaf area, are inherently good competitors for

78 space and resources and can reduce opportunities for species coexistence more effectively than invaders with more conservative traits (Maskell et al. 2006, Van Kleunen et al. 2010, Pyšek et al. 2012, Keser et al. 2014, Canavan et al. 2019). For example, the invasion of the forb species Heracleum mantegazzianum (giant hog weed) into sub-montane meadow communities in the Czech Republic, led to declines in both local and landscape-scale species richness; a result attributed to its ability to form dense mono-specific stands, which are taller and higher in cover than native-dominated communities (Hejda et al. 2009). Presumably, shorter native species could not persist under such intense light competition, and niche opportunities for light-demanding native species became rare, even at the landscape scale. Alternatively, invaders with more conservative growth strategies or those inherently clustered in their distributions (e.g. annual species with high individual mortality), can sometimes leave space for other species to survive and are less likely to have broad scale impacts on diversity (Hejda and Pyšek 2006, Lai et al. 2015). Similar interactions can occur belowground, with spatial variation in root biomass and the presence of ‘root gaps’ identified as an important mechanism allowing poor belowground competitors to persist (Cahill and Casper 2000). This paper investigates impact-scale relationships and potential trait-based mechanisms in the ‘iron bark’ woodlands of sub-coastal Queensland, Australia. The grassy understories of these woodlands are variably invaded by the exotic perennial grass species, Bothriochloa pertusa, which was introduced into Australia in 1939 for pasture and lawn development. Although B. pertusa is well established across large areas of these woodlands, it has becoming noticeably dominant in many areas over the last decade, possibly due to the combined effects of grazing and drought.

There is evidence to suggest B. pertusa invasion can lead to declines in native forbs and perennial grass richness and cover at small spatial scales (1m2) (Kutt and Kemp 2012) and its observed dominance over large areas is suggestive of broad-scale diversity declines. However, unlike some of the previously mentioned examples of invaders with large-scale impact, B. pertusa is not a tall species and generally produces less aboveground biomass than most native perennial grasses in the same habitat. Field observations of invaded and uninvaded areas however suggest that B. pertusa invests in denser root structures than native species of the same habitat. Intense competition for belowground space and resources may therefore contribute to the observed dominance of B. pertusa at both small and large scales.

79

We therefore hypothesise that 1) B. pertusa does indeed have broad-scale impacts on diversity and 2) this may be associated with its belowground competitive ability.

First, we assess the impact of B. pertusa across spatial scale (up to 1000m2) by evaluating changes to species area relationships (SARs) with increasing invader cover. SARs can be examined using the equation S = cAz, where S is the species number, A is area and c and z are the intercept and the slope respectively. If the impact of B. pertusa on diversity is maintained with increasing area, we would expect a decrease in the intercept (c) and a similar slope value (z) with increasing invader cover (Powell et al. 2013) (Figure 6-1). Because this can be an indication that rare rather than common species are disproportionately affected, we may also expect rare species to decline more than common species with increasing B. pertusa cover (Powell et al. 2011).

In a separate analysis within the same habitat, we then evaluate the role of belowground competition for space and resources as a likely contributing mechanism for broad-scale impact in B. pertusa. Specifically, we investigate changes in community-level root biomass and other root traits that describe species’ nutrient acquisition strategies (Cahill and Casper 2000, Pérez-Harguindeguy et al. 2013). If competition for belowground space and resources increases with invasion, we would expect areas of low root biomass (root gaps) to decline, and overall root biomass to be maintained or increase with increasing invader cover. We would also expect changes towards more resource-acquisitive root traits (Pérez- Harguindeguy et al. 2013).

The following research questions are addressed:

1) How are species area relationships (SARs) for total richness and lifeform richness related to invader cover and is this associated with the invader’s effect on rare vs. common species?

2) How do the root traits of communities change along a gradient of invasion cover?

80

Figure 6-1. Examples of uninvaded (solid) and invaded (dashed) species area relationships (a, c) and their corresponding log-log formulations (b, d) in two communities representing different scale-impact relationships. In community one, local-scale richness, represented by the intercept ‘c’, is lower in invaded plots. The rate of species accumulation, represented by the slope ‘z’ is steeper in invaded plots, leading to proportionally fewer species lost at broader spatial scales. The invader in community one is therefore having a large negative impact on species richness at small scales, but a negligible impact at the largest scale. In community two, local-scale richness, ‘c’, is again lower in invaded plots. The rate of species accumulation, ‘z’, however remains the same. The proportional reduction in species richness is therefore maintained across spatial scales. The invader in community two is therefore having a large negative impact on species richness across all scales investigated.

Methods

81

Study species

Bothriochloa pertusa, native to southern Asia, was first introduced into eastern Australia in the 1930s, primarily to be trialled as a pasture and lawn grass (Truong and McDowell 1985). B. pertusa is now abundant throughout subtropical regions of Queensland and elsewhere in Australia and has been associated with declines in floristic diversity and changes to faunal assemblages (Kutt and Fisher 2011, Kutt and Kemp 2012). As with many introduced pasture species B. pertusa possesses high fecundity and rapid growth rates (McIvor et al. 1996, Driscoll et al. 2014). B. pertusa can produce a dense mat of stolons, allowing for a much more continuous cover of culms than is typical of native grass species in the same habitat. B. pertusa is resistant to grazing, and it typically increases under heavy stocking rates after colonizing bare ground (McIvor 2007a; McIvor et al. 1982).

Study site

To address Questions 1, 41 1000 m2 plots, at least 15 km apart, were surveyed from south west of Mackay to north of Charters Towers, Queensland, Australia (Figure 6-2). The climate is sub-tropical, with most rain occurring in summer, between October and March. The mean annual rainfall across all sites is 633 mm with a range of 500 to 800 mm. Sampling was conducted in March 2017, near the end of the wet season for the region. Plots were confined to the ‘iron bark’ woodlands, which exist on moderately fertile sandy-loam soils and are a dominant ecosystem within sub-coastal region of Queensland examined here. They are relatively uniform in their vegetation structure and composition, with an average tree height of 10-15 m high and canopy cover of ~20%. The dominant tree species are the ‘ironbark’ Eucalyptus species, E. crebra and E. melanophloia. Common native species in the understory include the perennial grass species, Heteropogon contortus, Chrysopogon fallax, Aristida calycina, Themeda triandra and Enneapogon polyphyllus and a number of different forb species including Neptunia gracilis, Indigofera linnaei and Melhania oblongifolia.

The invasive grass, B. pertusa has been present in the region for several decades, with herbarium records and publications indicating it has been prevalent since at least the early 1980s (Bisset 1980). However, in recent years the species has rapidly increased in cover in some areas within the ‘iron bark’ woodlands, possibly due to the combined effects of

82 livestock grazing and climatic extremes (Ash et al. 2011, O'Reagain and Scanlan 2013). The cover of B. pertusa throughout these woodlands is therefore variable and plots were selected systematically to represent a range of B. pertusa cover. Plots were all located on pastoral properties grazed by cattle. Information on the precise grazing history, disturbance history or time since original invasion of plots was not possible to obtain, however areas with obvious evidence of heavy grazing, recent vegetation clearing or fire where avoided.

To address Question 2, root cores were extracted from within the ‘iron bark’ woodland west of Mackay (Figure 6-2). Sample collection was conducted in a paddock grazed by cattle, although had been rested from grazing for 57 days and had received approximately 490 mm of rain in the previous 6 months, an average amount for the sub-tropical climate of the region (Bureau of Meteorology, 2019).

Survey methods

Each of the 41 plots were separated into four sub-plots of increasing size; 1 m2, 10 m2, 100 m2 and 1000 m2 (Figure 6-2). Species and functional group richness were recorded within each sub-plot. Functional groups comprised annual grasses, annual forbs, perennial grasses, perennial forbs, shrubs and trees. For each sub-plot, B. pertusa cover was also assessed using the point intercept method at 1 m intervals around the perimeter of each sub-plot. A measure of plot-scale (1000m2 ) cover was calculated by dividing the total hits across all sub-plots by the total number of possible hits, 188. The total number of rare and common species within each plot were calculated based on their occurrences across the 41 plots. Species were considered rare if they occurred in less than four plots and common if they occurred more than 19 out of the 41 plots. Plant species were identified by reference to voucher specimens lodged with the Queensland Herbarium with nomenclature following Bostock and Holland (2018).

Root cores were extracted from ten 50 × 50 cm quadrats, distributed across a 30 m2 area. The ten quadrats were spread across a range of B. pertusa cover, increasing in approximately equal intervals between zero and one hundred percent. B. pertusa cover was measured using the point intercept method at 10 cm intervals within the quadrat (36 points total). Within each quadrat four 10 cm diameter cores were randomly placed and extracted to 20 cm depth. Cores

83 were transported back to the laboratory for analysis in a refrigerator set at 4˚C. Shoots were cut from each core and placed separately into the drying oven at 60˚C for 48 hours. It was not possible to accurately separate the roots of B. pertusa from other species within the core and thus the roots from all species within the core were used for analysis. We assume however that with increasing B. pertusa cover, the roots present in each core will be increasingly dominated by B. pertusa and any change in root traits across the invasion gradient are likely associated with the invader’s roots.

Roots were carefully sieved and washed, before being placed in a waterbed scanner to be analysed for a range of root traits in WinRHIZO (Régent Instruments, Quebec, Canada). Roots were then placed in the drying oven at 60˚C for 48 hours. Specific root length (SRL) was calculated by dividing total root length by root biomass. Root tissue density (RTD) was calculated by dividing root biomass by the root volume (Pérez-Harguindeguy et al. 2013). Root biomass was the calculated from the combined dry root weight of all species within the core. High SRL is commonly associated with enhanced nutrient acquisition, but shorter root lifespan; while high RTD is associated with increased drought tolerance and resistance to pathogens, but less efficient nutrient acquisition compared to low RTD species (Pérez- Harguindeguy et al. 2013).

Figure 6-2. (a) The location of the 41 plots (circles) and the property where root cores were collected (cross). All sites are located in the ‘ironbark’ woodlands of sub-coastal Queensland, Australia; (b) plot sampling design, showing the increasing area of the four sub-plots in a typical ‘ironbark’ woodland.

84

Statistical analysis

All exploratory and statistical analysis were conducted using R statistical software (R Development Core Team 2019). To evaluate changes to SARs with invasion (Question 1), we examined how the slope and the intercept of species-area relationships (SARs) relate to plot- scale B. pertusa cover. First however, we assessed spatial autocorrelation in the plot-scale cover of B. pertusa by converting cover to a proportion, logit-transforming and then fitting an intercept-only generalised least squares model (using the gls function in the nlme package in R (Pinheiro 2016)). We then plotted a semi-variogram of the model residuals using the Variogram function, also in the nlme package. The semivariogram showed no evidence of spatial autocorrelation, so we proceeded with analysis of SARs.

We assumed the SAR equation S = CAz, where S is the number of species, A is area, and c and z are constants (the intercept and the slope respectively). For all analyses the linear form of the equation [log(S) = zlog(A) + log(c)] was used to allow for calculation of the intercept and slope. In this case, the intercept corresponds to species or functional group richness at the smallest spatial scale, while the slope refers to the rate of species accumulation with area. The influence of B. pertusa on the slope and intercept of SARs were explored for total species richness and for the richness of functional groups (total number of species with each functional group). Frequency of annual grasses at all scales were too low to reliably examine distribution patterns so were excluded from analyses. Specifically, we performed log-log regressions using linear mixed-effect models, with sub-plot area and the plot-scale cover of B. pertusa as interacting fixed effects. We adopted a random intercepts and slopes structure that allowed SARs to vary for each plot. Using this formulation, it is possible to examine how well the plot-level slopes and intercepts are explained by fixed effects, in this case the cover of B. pertusa (Gelman and Hill 2007). When modelling SARs for functional groups we used log(S+1) to deal with small numbers of zeros at the smallest spatial scales. The richness of rare and common species richness within each plot were modelled as function of B. pertusa cover using simple linear regression models. All linear mixed effects models were fitted using the lmer function from the lme4 package (Bates et al. 2015).

To assess changes to community root traits with invasion (Question 2), root traits were modelled as a function of B. pertusa cover using a series of linear regression models for each

85 response. To confirm our expectation that B. pertusa has low investment to aboveground biomass, we also examined the relationship between shoot biomass and B. pertusa cover using a linear model.

Explanatory variables were assessed for normality and ln- or sqrt- transformed as necessary to improve linearity with response variables.

Results

Question 1: How do species and functional group richness SARs relate to invader cover and is this associated with the disproportional effects on rare and common species in the community?

A total of 219 species were recorded during the survey. Perennial herbs (n=68) were the most species rich functional group, followed by annual herbs (n=45), perennial grasses (n=35), trees (n=32), shrubs (n=23) and annual grasses (n=16). Across the 41 plots there was a total of 133 rare species and 17 common species.

When total species richness was modelled (Figure 6-3ab), the intercept of the SAR was negatively related to the plot-scale B. pertusa cover (Figure 6-3c; P = 0.02), while the slope was unrelated (Figure 6-3d). This suggests local-scale richness declines with invasion, however the rate of species accumulation with area is similar to plots with low B. pertusa cover.

The intercept of the SAR for perennial grass and shrub species were unrelated to B. pertusa cover (Figure 6-4a, c), while the slope was significantly negatively related (Figure 6-4b, d; P = < 0.01). This suggests that the local-scale richness of these functional groups is similar irrespective of invasion, however the rate of species accumulation with area is much less in plots with high B. pertusa cover. The slope and intercept for annual and perennial forbs and tree species was not significantly related to B. pertusa cover. The richness of rare species was significantly negatively related to B. pertusa cover (Figure 6-5a, P = < 0.01). There was no relationship between the richness of common species and B. pertusa cover (Figure 6-5b).

86

Figure 6-3. (a) observed SARs for all plots based on total species richness and (b) corresponding fitted log-log relationships for each plot from the linear mixed-effects model. In (a) and (b) line thickness is proportional to the plot-scale cover of B. pertusa. Panels c) and d) show the relationships between plot-scale B. pertusa cover and the plot-level intercepts and slopes estimated by the linear mixed-effects model. Black lines are the fitted relationships from the plot-level part of the model and grey envelopes are the associated 95% confidence intervals. A solid or dashed line indicates if the relationship was significant (P <0.05) or not significant (P >0.05) respectively. Each point represents a plot, and the bars show the standard error of each plot’s (c) intercept and (d) slope.

87

Figure 6-4. Relationships between plot-scale B. pertusa cover and the estimated SAR intercepts and SAR slopes from the linear mixed-effects models for perennial grass richness (a, b) and shrub richness (c, d). Black lines are the fitted relationships from the plot-level part of the models and grey envelopes are the associated 95% confidence intervals. A solid or dashed line indicates if the relationship was significant (P <0.05) or not significant (P >0.05) respectively. Each point represents a plot, and the bars show the standard error of each plot’s (c) intercept and (d) slope.

88

Figure 6-5. Relationships between plot-scale B. pertusa cover and the richness of rare (a) and common (b) species in the community. Black lines are fitted relationships from linear models and grey envelopes are the associated 95% confidence intervals. A solid or dashed line indicates if the relationship was significant (P <0.05) or not significant (P >0.05) respectively.

Question 2: How do the root traits of communities change along a gradient of invasion intensity?

Root traits of communities generally became more resource-acquisitive with increasing plot- scale B. pertusa; SRL was positively related to B. pertusa cover, while RTD was negatively related (Figure 6-6a, b; P = <0.01). Root biomass was not significantly related to invader cover (Figure 6-6c). With the exception of the first uninvaded quadrat (0 % cover), there was also no discernible association with B. pertusa cover and the presence of areas with relatively low root biomass (Figure 6-6c). In moderately invaded plots (30-65 % cover), root biomass was notably more variable than relatively uninvaded (< 30 % cover) or heavily invaded plots (> 65 % cover). Although not significant, shoot biomass was moderately negatively related to B. pertusa cover, indicative of this species low allocation to aboveground biomass (Figure 6- 6d).

89

Figure 6-6. Relationships between B. pertusa cover and (a) SRL, (b) RTD, (c) root biomass and (d) total shoot biomass. Black lines are fitted relationships from linear models and grey envelopes are the associated 95% confidence intervals. A solid or dashed line indicates if the relationship was significant (P <0.05) or not significant (P >0.05) respectively.

Discussion

The impacts of invasive species on native plant communities are not consistent across spatial scales and understanding the mechanisms behind these scale-impact relationships is important for prioritising invasive species management. This study evaluated changes to SARs across sites variably invaded by B. pertusa (ranging from very low to very high cover). We also examined changes to community root traits with increasing invader cover, to evaluate the role of belowground competition as a potential mechanism driving broad-scale impact in this species. Local-scale species richness declined with increasing B. pertusa cover, however the rate of species accumulation with increasing area remained the same, indicating that this invasive grass has the potential to reduces diversity from small to large spatial scales in the ‘iron bark’ woodlands of eastern Queensland. The observed changes to SARs were

90 associated with a proportionally greater effect of B. pertusa on rare species in the community. Changes to community root traits with increasing B. pertusa were indicative of an increasingly resource-acquisitive belowground environment, however contrary to our hypothesis there was no association with invader cover and areas of low root biomass or ‘root gaps’. This suggests that although competition for root space does not increase with invasion, competition for resources is likely more intense and could contribute to a loss of diversity at broad spatial scales.

Are changes to SARs reflective of broad-scale invader impact?

The spatial scale of investigation has been used as a justification for both positive and negative relationships between native diversity and invasive exotic species (Davies et al. 2005, Fridley et al. 2007, Gaertner et al. 2009, Powell et al. 2011, Case et al. 2016). Investigating the impact of invasive species on SARs, Powell et al. (2013) found that invaded communities had lower local richness (SAR intercept), but steeper species accumulation (SAR slope) than uninvaded communities. This led to proportionally fewer species lost at broader spatial scales (up to 500 m2) than at local scales. Similar declines in effect size with increasing scale were observed in two meta-analyses investigating scale-dependent invader impacts (Gaertner et al. 2009, Powell et al. 2011).

Despite investigating spatial scales up to 1000 m2, we found no such increase in SAR slopes with invasion. In the case of B. pertusa, local-scale species richness was significantly lower in plots with high cover, but the rate of species accumulation was similar to uninvaded plots. This suggests that the proportional loss of species with increasing invader cover was similar at both small and large spatial scales. Furthermore, for perennial grass and shrub species, SAR slopes declined (became flatter) with increasing B. pertusa cover, suggesting that proportional reductions in the richness for these functional groups amplify with increasing scale. These results suggest that in this study system B. pertusa invasion can lead to both small- and large-scale declines in diversity.

In line with findings from simulated invaded communities (Powell et al. 2011), we found that the broad-scale impact of B. pertusa, was associated with a relatively high proportion of rare species (low patch occupancy) in the community which were affected more than common

91 species (high patch occupancy) with increasing invader cover. Because rare species occur in fewer patches, they are more likely to be encountered at larger spatial scales and are more vulnerable to regional extinctions. Thus, if the invader affects rare species more than or equal to common species, as has occurred in this study, large-scale declines in diversity are likely, particularly if there is a relatively high number of rare compared to common species in the community (Powell et al. 2011). This does not necessarily mean that B. pertusa is competing more strongly with rare species (although this may be the case), however it does indicate that they are at least equally as affected as common species and are not escaping invader impacts by occupying less invasable niches.

High proportions of rare species within communities, as was evident here, can arise as a result of spatial niche differentiation. If B. pertusa is indeed competing more strongly with rare species, it may be that the specialised niches which support rare species are also more invasable by B. pertusa (Powell et al. 2011). Alternatively, if B. pertusa is influencing both common and rare species equally, B. pertusa may have broad niche requirements, allowing it to successfully invade under a wide range of environmental conditions. High proportions of rare species, however, can also be a consequence of disturbances that reduce the patch occupancies of several species over a large area (Didham et al. 2007). As the sites examined here were all grazed by cattle, and are within a region subject to recent and widespread habitat loss (McAlpine et al. 2002, Wilson et al. 2002a), it is likely these disturbances are exacerbating the effect of B. pertusa by reducing the occupancy of native species across large spatial scales. Additionally, grazing is known to facilitate the spread and increase the dominance of B. pertusa (Jones 1997, Ash et al. 2011).

To disentangle the relative influence of B. pertusa on rare and common species, future research should experimentally examine changes to the abundance distributions of species with increasing invader cover. Common species for instance may experience proportionally similar declines in abundance to rare species but are more likely to persist and therefore contribute to diversity than rare species. Nevertheless, these findings highlight that within this study system, B. pertusa has the capacity to reduce diversity at both small and large spatial scales and this is associated with a decline in the richness of rare species in the community.

92

How does the belowground competitive environment change with invasion?

Although disturbance potentially exacerbates the impact of B. pertusa on diversity across spatial scales (by altering species abundance distributions), its capacity to maintain dominance over large areas is perhaps associated with its ability to compete for space and resources. We hypothesised that competition for belowground space and resources was potentially a key strategy driving its broad-scale impact. We found some evidence to support this; increased specific root length and decreased root tissue density with increasing invader cover suggest that B. pertusa invests in roots that are shorter-lived and less drought tolerant, but more efficient at resource acquisition than native species in the community (Wahl and Ryser 2000, Tucker et al. 2011, Laliberté et al. 2015). This acquisition strategy can reduce available nutrients and moisture within the soil profile and decrease the establishment success and growth of other species in the community, particularly in resource-limited environments (Mommer et al. 2011, Broadbent et al. 2018). Although these results are indicative of potential differences in the resource-acquisition strategies of B. pertusa and co-occurring native species, without examining the root traits for each species separately and assessing resource-use efficiency under relevant environmental conditions, this cannot be confirmed. This would be a useful area of future research to separate the relative importance of competition and disturbance in driving the spread and impact of this species.

Contrary to our expectations, the presence of ‘root gaps’ did not decline with increasing invader cover, suggesting competition for belowground space is similar irrespective of invader cover. In moderately invaded communities (30-65% cover) however, there was substantial variability in total root biomass between cores. This may reflect a change in community composition or trait expression in response to competition from the invader. For instance, some invaders can alter the environment so as to facilitate some species more than others, potentially leading to changes in functional composition with invasion (Simberloff and Von Holle 1999, MacDougall and Turkington 2005). Alternatively, phenotypic plasticity in response to competition can increase trait variability (Turcotte and Levine 2016, Conti et al. 2018). For example, Ballaré et al. (1990) showed that seedlings in the grassy woodlands of Estonia increased their rate of stem growth in response to increased competition for light. Similarly, some native species may respond to belowground resource competition with B. pertusa by increasing their root biomass allocation.

93

It is also important to consider other aspects of B. pertusa’s ecology which may contribute to its broad-scale impacts on diversity. Despite having low aboveground biomass, B. pertusa is stoloniferous, which can allow for a much more continuous ground cover of culms than is typical of native tussock grasses. As shown for other clonal species, this can enhance the species’ capacity to colonise and compete for inter-tussock space and provide access to a larger resource pool than non-clonal species (Keser et al. 2014). There is also evidence to suggest B. pertusa allelopathically inhibits the establishment of other species (Hussain et al. 1982).

Despite the obvious caveats which have been discussed, the community-level root analysis suggests that intense competition for belowground resources may be an important mechanism contributing to broad-scale impacts of this species.

Conclusions

Scale-impact relationships cannot be generalised across all invasion scenarios (Stohlgren and Rejmánek 2014) and are influenced by a number of interacting abiotic and biotic factors (Wardle et al. 2011, Pyšek et al. 2012, Stohlgren and Rejmánek 2014, Case et al. 2016). In the case of B. pertusa invading the ‘iron bark’ bark woodlands of eastern Queensland, diversity declines were pervasive from small to large spatial scales. We have highlighted the potential importance of both disturbance and below-ground competition in driving the broad- scale impacts of this species.

By reducing the patch-occupancy of native species and increasing the establishment and dominance of B. pertusa, disturbances such as livestock grazing and regional-scale habitat loss, have likely exacerbated the competitive effects of B. pertusa in these woodlands. Reducing the extent and intensity of these disturbances may therefore increase the resistance of these ecosystems to invasion and prevent large-scale species losses. To further inform the management of this species, future research should experimentally examine the relative importance of disturbance (grazing by livestock and climatic extremes) and competition in driving patterns of invasion and diversity decline in these woodlands.

94

Chapter 6 is published in the Journal of Vegetation Science.

Lebbink, G.H., Dwyer, J., Fensham, R. J. (2020). An invasive grass species has both local and broad scale impacts on diversity: Potential mechanisms and implications. Journal of Vegetation Science 32:e12972.

Field work, plant identification, and writing by GL, with input and reworking from RF and JD. Analysis by JD and GL.

95

Chapter 7 – Discussion

This thesis examined factors associated with the spread, impact and management of the invasive pasture species B. pertusa, in Queensland Australia. As with many invasive species across the globe, the spread of B. pertusa was facilitated by a combination of deliberate and repeated introductions and intensive agricultural land-use (Chapter 2 and 3). Livestock grazing in particular is thought to have facilitated the spread of this species and evidence of this can be found within the literature, observations by land holders, and fence-line contrasts between long-ungrazed reserves and their neighbouring pastoral properties (Chapter 4). Where B. pertusa achieves high cover, diversity declines persist at both small and large spatial scales (Chapter 6). Furthermore, habitat suitability modelling suggests there is considerable capacity for B. pertusa to spread and increase its dominance within Queensland (Chapter 3). Broad-scale diversity declines, coupled with it the potential to spread and increase in dominance throughout Queensland, highlights the management importance of this species for conserving native biodiversity.

In this final chapter I synthesise my results and those from previous studies to suggest strategies to mitigate the spread and impact of B. pertusa. As this thesis has highlighted however, B. pertusa is just one of the many threats to ecosystems within these modified landscapes, with habitat loss, fragmentation, livestock grazing and other invasive species presenting major and inter-related threats to biodiversity. Protecting native species and habitats within such modified and threatened landscapes, often with conflicting management priorities, is one of the most pressing challenges of the 21st century. Thus, in the second half of this chapter, I also critically examine some of the key approaches taken to conserve and restore biodiversity within modified agricultural landscapes.

Slowing down the ‘lawnification’– perspectives on managing Bothriochloa pertusa

Although there is some evidence that B. pertusa is a strong competitor for below-ground resources (Chapter 6), disturbance appears to be an important pre-requisite for B. pertusa invasion. In particular, findings from this and other research suggest that prolonged intense grazing, in combination with climatic extremes (drought followed by periods of above- average rainfall) are key factors associated with its establishment and spread (Ash et al. 2011,

96

O’Reagain and Bushell 2015). Grazing and drought compound to reduce the cover of native species (particularly perennial grass species), which creates niche opportunities for B. pertusa and other opportunistic invaders. B. pertusa’s stoloniferous habit (Ash et al. 2011), profuse seed supply (Howden 1988) and grazing tolerance (Jones 1997) make it functionally different from many co-occurring native species, and likely better equipped to rapidly colonise, acquire nutrients, and establish in response to improved growing conditions.

There is evidence to suggest that maintaining native species cover through conservative grazing strategies improves the resilience of native communities to invasion by B. pertusa even after periods of drought (Ash et al. 2011, Orr and Reagain 2011, Kutt and Kemp 2012, O'Reagain et al. 2018). Specifically, results from a long term-grazing trail (1993-2000) in north-east Queensland, found that pastures utilised at 25% with no spelling, or pastures utilised at 50% with early wet-season spelling were able to maintain native perennial grass cover during periods of drought and were less invaded by B. pertusa than paddocks utilised at 75% with no pasture spelling (Ash et al. 2011).

Similar results were found on another long-term grazing trail (1998 – Current) near Charters Towers in which B. pertusa increased across all grazing treatments, but most significantly within paddocks which were heavily grazed (4 ha per animal equivalent, no rest) (O’Reagain and Bushell 2015, O'Reagain et al. 2018). In comparison, in paddocks which were moderately stocked (8 ha per animal equivalent), either continually or with rotational wet- season spelling, the increase in B. pertusa was half that of the heavy stocking rate treatment. These paddocks also maintained the greatest native perennial grass cover over time, with little variation in perennial grass frequency, even during a sequence of dry years between 2002 and 2007 (O’Reagain and Bushell 2015, O'Reagain et al. 2018). For example, at the peak of drought in 2004, B. pertusa’s native congener Bothriochloa ewartiana declined from 40% to 20% in heavily stocked paddocks. Conversely, in the moderately stocked paddocks, this species maintained a frequency of around 50% during these dry years. Although observational and correlative data such as this are important, additional manipulative research is required to disentangle the relative influence of climate, grazing and disturbance. Investigating the response of floristic communities and specifically the dynamics between native and exotic species under scenarios of climate change is also an important area of future research.

97

It is clear from these long-term grazing trials that adapting stocking rates and resting regimes to maintain native species cover is likely a key strategy for reducing the establishment success of B. pertusa. Whether natives can recover after reducing the dominance of B. pertusa is unclear, however there is some anecdotal evidence of this occurring with careful grazing management. For example, on Mt. Pleasant, a grazing property near Bowen, central Queensland, switching from a continuous to a timed rotational grazing system, resulted in the notable recovery of native grass species in B. pertusa dominated pastures within 10 years of implementation. On this property stocking rates are adjusted according to land condition and climate, and cattle are frequently moved between several paddocks, resulting in multiple short grazing periods followed by a long period of pasture rest. The benefits of rotational grazing systems (which incorporate long resting periods) for biodiversity and invasion resistance has been demonstrated elsewhere in Australia (McDonald et al. 2019). However, due to the time and money needed to establish infrastructure and herding systems, this strategy is not always viable and it is unclear whether similar results could be achieved with less resource intensive management strategies. Despite these limitations, the results from Mt. Pleasant are promising and research into this and other strategies for restoring native species into B. pertusa dominated grasslands would be a useful area of future research and crucial for assessing the biodiversity potential of these degraded systems.

Reversing perennial grass invasions has proven challenging in the past, with recruitment limitation of native species a major constraint to restoration (Standish et al. 2007, Brooks et al. 2010). Although some natives may be inherently poor recruiters, habitat loss, fragmentation and competition interact to create dispersal and recruitment barriers and reduce the restoration capacity of ecosystems (Higgins et al. 2003, Fensham et al. 2016). Further, propagule pressure from exotic grasses can be immense in heavily fragmented landscapes (Fensham et al. 2013). Alleviating recruitment limitation by reintroducing native propagules has been effective in some grassland restoration projects and would be an important strategy to consider when trying to restore native species into B. pertusa monocultures (Ladouceur and Mayfield 2017, Cuneo et al. 2018). This may also be useful for improving the colonisation rates of native species when using grazing to manage C. ciliaris invasion. Grazing-protected areas are relatively resilient to invasion (Chapter 5) and are important refuges and source of propagules for native species. Expanding the network of reserves may

98 also help to overcome problems with recruitment limitation (D Antonio et al. 2001). Even with adequate propagule supply however, the environmental conditions may no longer be suitable for their establishment. In particular, the recovery of soil resources after a legacy of livestock grazing can take several decades and reseeding native species may need to be coupled with external inputs to adequately restore organic matter, soil carbon and soil nitrogen (Fuhlendorf et al. 2002, Prober et al. 2005, Lindsay and Cunningham 2011).

Managing grazing pressure and ground cover to maintain resilient ecosystems is not unique to managing B. pertusa invasion (Dorrough et al. 2004b, McIntyre et al. 2004, Müller et al. 2007). Indeed, this is a fundamental principle of maintaining diversity and ecosystem functioning within grazing management systems and has received considerable attention within the literature and from government-led research organisations (McIntyre et al. 2004, McIvor 2012, Hunt et al. 2014). Despite this, poor grazing management practices are still pervasive, and this is largely associated with actual or perceived declines in productivity (Greiner and Gregg 2011, Moon and Cocklin 2011).

There is emerging evidence however that sustainable grazing strategies are also economically viable, particularly over the long term. The for-mentioned long-term grazing trial near Charters Towers for example, found that between 1997 and 2009, the most profitable grazing strategy were those which either continuously moderately stocked or variably stocked according to pasture growth and climate predictions, exceeding the profits made with continuous heavy stocking by nearly 50% (O'Reagain et al. 2011). Although the heavily stocked sites were initially more profitable, over time these gains were negated by losses incurred from reduced animal performance and forced sales at the beginning of dry years. Moderately stocked paddocks also had higher native perennial grass cover (O'Reagain and Scanlan 2013) and more diverse fauna assemblages (Neilly et al. 2018, Neilly and Schwarzkopf 2018), highlighting the potential for both pastoral and conservation outcomes with this management strategy. Importantly, this study also highlights the importance of a premium price for well-conditioned cattle, to increase the economic viability of conservative stocking (O'Reagain et al. 2009). Currently, this is not relevant for all markets, including live- export which is one of the major avenues for selling cattle in northern Australia.

99

Maintaining native species cover through adaptive and conservative grazing strategies is a key strategy towards improving the resilience of ecosystems to invasion by B. pertusa and this is also compatible with pastoral productivity in some instances. The importance of long- term experimental trials, which highlight the economic and ecological value of different management strategies are required. By highlighting the often-dual benefits for biodiversity and production these trials are crucial for increasing the adoption of more sustainable grazing practices. It is clear from the results of this and other research however that increasing grazing-protected areas and the restoration of habitats is also essential if we are to mitigate the currently extreme rates of biodiversity decline due to invasive species and land-use change (Kamal et al. 2015). Despite this being well-established, initiatives towards this aim have proven challenging to implement due to the sometimes-necessary compromises to agricultural production, which is the dominant land-use of private land. In recent years however there has been an increased awareness and engagement in private land conservation strategies. In the following section I examine and discuss the key strategies taken to promote sustainable grazing practices, increase grazing protected areas and promote the restoration of ecosystems on private land.

A balancing act: approaches to conservation on private land

Although not mutually exclusive, strategies towards private land conservation generally fall within two broad approaches; the ‘duty of care’ approach which encourages land-holders to acknowledge their responsibility in protecting ecological values, and the incentivisation approach which provides monetary rewards for conservation management on private property. By helping to encourage and incentivise sustainable land management and conservation on private land, these strategies are important for mitigating the threats associated with land-use change and invasive species highlighted in this thesis.

The ‘duty of care’ approach

By definition a ‘duty of care’ requires people to take reasonable steps to avoid foreseeable harm to a specific class of objects (e.g. workers, students, natural resources) and are usually accompanied by a code of practice or guidelines to help understand what the duty entails (Earl et al. 2010). Although there have been considerable attempts to define what a ‘duty of

100 care’ for the environment might entail, a lack of consensus remains, and this is largely due to the inherent complexities of natural ecosystems and the conflicting values and ideologies of people (Bates 2001, Earl et al. 2010). Nevertheless, an acknowledgment of ‘duty of care’ has been the driving force for a number of successful private land conservation initiatives, such as Australia’s Landcare program, and conservation covenanting programs which are being increasingly implemented across the globe (Gallo et al. 2009, Pocewicz et al. 2011, Curtis et al. 2014).

These programs rely on voluntary participation from landholders to improve land management and protect conservation values on their property. Many landholders already consider themselves stewards of the land (Greiner and Gregg 2011, Moon and Cocklin 2011). These programs help to reinforce and encourage these inherent values. By doing so they also help change social norms, build social capital and empower communities to protect natural resources and the environment (Walker 2013). Although some financial and technical support is often provided (usually through government grants) these are usually delivered sporadically and rarely parallel the necessary commitments and sacrifices required by the landholders (Stephens 2001). The government led Queensland Nature Refuge program for instance has been hugely influential in increasing the protected area network in Queensland by encouraging the voluntary listing of properties under a conservation covenant which restricts land-use practices contrary to an established conservation agreement. To date Nature Refuges in Queensland is protect 4.4 million hectares (30% of the protected area estate) and 955 threatened species (Our Living Outback, 2018). A recent report however, suggests that the current funding for the project is insufficient, and an increase from the current investment of $4.6 million (only 23% of which is spent on incentives, grants and subsidies) to $28.6 million per year is necessary to better support landholders to protect and manage their land under the Nature Refuge program (Allen et al. 2018). Similarly, progressive defunding of the Australian Landcare program, which encourages and supports landholders (through education programs and monetary grants) to work collaboratively to overcome land management problems, has reduced the capacity of this program to provide effective land management solutions (Curtis et al. 2014, Cooke and Moon 2015).

Although there has been high participation from landholders in these programs (Walker 2013, Allen et al. 2018), it is becoming increasingly evident that we cannot rely on altruistic

101 motives alone, and in many cases, what is required to achieve conservation outcomes within such modified and threatened ecosystems goes above and beyond what is considered ‘duty of care’. This becomes particularly apparent when significant investments in infrastructure or losses to agricultural production are necessary to achieve conservation outcomes.

The incentivisation approach

In response to conflicting production and conservation priorities, providing monetary incentives for biodiversity outcomes is becoming increasingly adopted across all levels of government (Ansell et al. 2016). This commonly takes the form of one-off or ongoing payments to conduct specific management strategies or achieve specific conservation outcomes. Australia has implemented a number of programs under this framework, including the ‘Bush Bids’ program in South Australia, the ‘Forest Conservation Fund’ in Tasmania and more recently the Australia-wide ‘Environmental Stewardship Program’. These programs usually target specific threatened ecosystems and are delivered through a market-based process, whereby landholders are required to bid competitively with other land managers to have their project funded. Bids which achieve the greatest conservation outcome for the least cost are generally selected for funding. Although the contractual commitment usually extends over several years, the funding wavers with government administration and this has hugely influenced their success (Burns et al. 2016). Further, there is minimal ongoing support once the agreed commitment time expires. Although the initial investment is often the greatest, ongoing financial and technical support is necessary for enduring conservation success, and this is currently not well accounted for in the accounting of these programs (Stephens 2001, Moon and Cocklin 2011).

Recently, the Australian Government announced an Agricultural Biodiversity Stewardship Pilot Program to commence in 2021, which will provide grants to support small and medium- sized farms in the adoption of improved biodiversity practices on farms (outlined under the Agricultural Biodiversity Policy currently being developed). This pilot will also run alongside the Australian Farm Biodiversity Certification Scheme which will assist farmers to showcase their biodiversity practices and receive premium prices for the product. These programs have the potential to facilitate improved land management practices and private land conservation, however this will depend on the exact criteria and definitions assigned to improved

102 biodiversity management. Further, as with stewardship programs implemented in the past, their value in achieving biodiversity outcomes in the longer term will require ongoing monitoring and support.

By offering financial payments for reducing greenhouse gas emissions or increasing carbon sequestration and storage, emerging carbon markets also present a significant opportunity for incentivising improved land management and the protection and restoration of ecosystems (Venter et al. 2009, Dwyer et al. 2010, Waters et al. 2017). For instance, in the savannas of northern Australian, improved management (moderate stocking with pasture spelling) has the potential to raise approximately US $5 per hectare per year in carbon revenue (based on a carbon price of US $14 per tonne of carbon dioxide equivalent) and prevent the release of 1-2 billion tonnes of carbon dioxide equivalent over 90 years (Douglass et al. 2011). Similarly, as highlighted in Chapter 2, carbon based incentivisation schemes also present significant opportunities for the restoration of threatened Acacia harpophylla (brigalow) woodland and fertile grasslands in central Queensland (Dwyer et al. 2010, Fensham et al. 2016). In most cases, restoring A. harpophylla within grazing lands would involve considerable cost to production, and incentivisation through carbon is likely the most promising strategy for promoting the restoration of this ecosystem. As discussed in Chapter 6, this will likely also assist with the management of C. ciliaris (buffel grass) invasion.

Synthesis

Both the ‘duty of care’ and incentivisation approaches to private land conservation have considerable potential to improve land management, mitigate biodiversity decline, and build social capital and there is considerable evidence of where this has occurred (Greiner and Gregg 2011, Moon and Cocklin 2011, Allen et al. 2018). In the highly fragmented landscapes of Queensland, which have been the focus of this thesis, these initiatives present considerable opportunity to increase the restoration of threatened ecosystems, such as A. harpophylla and fertile grassland. Further, by compensating losses to production, they can improve the capacity for producers to increase the area of ungrazed land or reduce stocking rates, which as I have highlighted here is important for increasing ecosystem resilience to invasion and climatic extremes.

103

Despite there being a clear willingness for landholders to participate, inadequate financial and technical support is leading to compromised conservation and social outcomes. To ensure the outcomes already achieved from these programs are nurtured and sustained, and to increase their implementation more broadly, it is clear far greater investment is necessary. Although a number of issues with their implementation have been recognised, private land conservation initiatives remain a key strategy for mitigating the current biodiversity and climate crisis and their development should be a priority across all levels of government.

Conclusion

Over the last 70 years, Queensland, Australia has been a global hotspot for habitat loss and fragmentation. Such extreme land modification will no doubt have considerable aftershocks for the environment. The invasion of Bothriochloa pertusa is one of these aftershocks and as highlighted by this thesis, its continued spread throughout the landscape is a major threat to native flora species. Its negative impact for fauna and ecosystem services has also been recognised in previous research (Kutt and Fisher 2011, Koci et al. 2020). Although the exact prescription for its control is not clear and deserves further research, maintaining land condition through sustainable grazing strategies is important for maintaining functional ecosystems resilient to invasion and other stochastic events. Increasing grazing-protected areas and habitat restoration are likely important strategies for reducing the spread and impact of B. pertusa, but more broadly are crucial for managing large-scale processes of degradation which are contributing to declines in diversity and ecosystem function.

It is clear that the complexity of the environmental problems we now face requires a far more collaborative approach and increased financial, technical and administrative support. The Government’s rapid response to the COVID-19 pandemic has shown that Australia has significant capacity to respond to the enormity of such challenges. Importantly, there is considerable willingness to shift behaviours and values and these provide the necessary foundations for systemic change to occur. Only when this is met with equivalent top-down support from our government and leaders can we expect these changes to be sustained necessary changes and the desired outcomes to eventuate.

104

References

Abensperg-Traun, M., L. Atkins, R. Hobbs, and D. Steven. 1998. Exotic Plant Invasion and Understorey Species Richness: A Comparison of Two Types of Eucalypt Woodland in Agricultural Western Australia. Pacific conservation biology 4:21-32. Adams, V. M., A. M. Petty, M. M. Douglas, Y. M. Buckley, K. B. Ferdinands, T. Okazaki, D. W. Ko, and S. A. Setterfield. 2015. Distribution, demography and dispersal model of spatial spread of invasive plant populations with limited data. Methods in Ecology and Evolution 6:782-794. Allen, L., R. Ledgar, A. Leverington, and F. Leverington. 2018. Nature Refuges in Queensland: report to The Pew Charitable Trusts and Queensland Trust for Nature. Protected Area Solutions, Queensland, Australia Ansell, D., F. Gibson, and D. Salt. 2016. Learning from agri-environment schemes in Australia. Investing in biodiversity and other ecosystem services on farms. ANU Press, Canberra, Australia Ash, A. J., J. P. Corfield, J. G. McIvor, and T. S. Ksiksi. 2011. Grazing Management in Tropical Savannas: Utilization and Rest Strategies to Manipulate Rangeland Condition. Rangeland Ecology & Management 64:223-239. Asner, G. P., A. J. Elmore, L. P. Olander, R. E. Martin, and A. T. Harris. 2004. Grazing systems, ecosystem responses, and global change. 29:261-299. Auditor-General. 2020. Auditor-General Report No.47, Referrals, assessments and approvals of controlled actions under the environment protection and biodiversity conservation act 1999, Department of Agriculture, Water and the Environment, Queensland. Australia, G. 2020. Coal. Australian Government, Canberra, ACT. Ballaré, C. L., A. L. Scopel, and R. A. Sánchez. 1990. Far-red radiation reflected from adjacent leaves: an early signal of competition in plant canopies. Science 247:329- 332. Bartley, R., J. P. Corfield, A. A. Hawdon, A. E. Kinsey-Henderson, B. N. Abbott, S. N. Wilkinson, and R. J. Keen. 2014. Corrigendum to: Can changes to pasture management reduce runoff and sediment loss to the Great Barrier Reef? The results of a 10-year study in the Burdekin catchment, Australia. The Rangeland Journal 36:311. Bastin, G., B. Abbott, B. Karfs, C. Holloway, A. Liedloff, and V. Chewings. 2008. Landscape leakiness in the grazed rangelands of the Burdekin.in Proceedings of the

105

15th Australian Rangeland Society Biennial Conference. Australian Rangeland Society, Australia. Bates, D., M. Mächler, B. M. Bolker, and S. C. Walker. 2015. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67. Bates, G. 2001. A duty of care for the protection of biodiversity on land, consultancy report to the Productivity Commission, AusInfo, Canberra. Baxter, G. S., E. J. Moll, and A. T. Lisle. 2001. Pasture grazing by black-striped wallabies (Macropus dorsalis) in central Queensland. Wildlife Research 28:269-276. Bisset, W. 1980. Indian bluegrass has special uses. Queensland agricultural journal 106:507- 517. Bock, C. E., J. H. Bock, L. Kennedy, and Z. F. Jones. 2007. Spread of non-native grasses into grazed versus ungrazed desert grasslands. Journal of Arid Environments 71:229-235. Boivin, N., R. Crassard, and M. Petraglia. 2017. Human dispersal and species movement: from prehistory to the present. Cambridge University Press, Cambridge Bortolussi, G., J. G. McIvor, J. J. Hodgkinson, S. G. Coffey, and C. R. Holmes. 2005. northern Australian beef industry, a snapshot. 1. Regional enterprise activity and structure. northern Australian beef industry, a snapshot. 1. Regional enterprise activity and structure 45:1057-1073. Broadbent, A., C. Stevens, D. Peltzer, N. Ostle, and K. Orwin. 2018. Belowground competition drives invasive plant impact on native species regardless of nitrogen availability. Oecologia 186:577-587. Brook, B. W., N. S. Sodhi, and C. J. A. Bradshaw. 2008. Synergies among extinction drivers under global change. Trends in Ecology & Evolution 23:453-460. Brooks, K. J., S. A. Setterfield, and M. M. Douglas. 2010. Exotic grass invasions: Applying a conceptual framework to the dynamics of degradation and restoration in australia’s tropical savannas. Restoration Ecology 18:188-197. Bryant, D., M. J. Bruce, and S. J. Sinclair. 2017. Observations of responses to re-introducing fire in a Basalt Plains grassland after the removal of grazing: Implications for restoration. Ecological Management & Restoration 18:239-245. Buisson, E., S. Le Stradic, F. A. O. Silveira, G. Durigan, G. E. Overbeck, A. Fidelis, G. W. Fernandes, W. J. Bond, J.-M. Hermann, G. Mahy, S. T. Alvarado, N. P. Zaloumis, and J. W. Veldman. 2019. Resilience and restoration of tropical and subtropical grasslands, savannas, and grassy woodlands. Biological Reviews 94:590-609.

106

Bureau of Meteorology, 2018. Climate statistics for Australian locations, Bureau of Meteorology, Australian Government. Available at: http://www.bom.gov.au (Accessed 24 January 2018) Burnett, P. 2020. Let there be no doubt: blame for our failing environment laws lies squarely at the feet of government. The Conversation The Converstaion Media Group Ltd, Victoria. Burns, E., C. Zammit, S. Attwood, and D. Lindenmayer. 2016. The environmental stewardship program: lessons on creating long-term agri-environment schemes.in D. Ansell, F. Gibson, and D. Salt, editors. Learning from agri-environment schemes in Australia. ANU Press, Canberra Butchart, S. H. M., M. Walpole, B. Collen, A. van Strien, J. P. W. Scharlemann, R. E. A. Almond, J. E. M. Baillie, B. Bomhard, C. Brown, J. Bruno, K. E. Carpenter, G. M. Carr, J. Chanson, A. M. Chenery, J. Csirke, N. C. Davidson, F. Dentener, M. Foster, A. Galli, J. N. Galloway, P. Genovesi, R. D. Gregory, M. Hockings, V. Kapos, J.-F. Lamarque, F. Leverington, J. Loh, M. A. McGeoch, L. McRae, A. Minasyan, M. H. Morcillo, T. E. E. Oldfield, D. Pauly, S. Quader, C. Revenga, J. R. Sauer, B. Skolnik, D. Spear, D. Stanwell-Smith, S. N. Stuart, A. Symes, M. Tierney, T. D. Tyrrell, J.-C. Vié, and R. Watson. 2010. Global biodiversity: Indicators of recent declines. Science 328:1164. Butler, D. W., and R. J. Fairfax. 2003. Buffel grass and fire in a gidgee and brigalow woodland: A case study from central Queensland. Ecological Management and Restoration 4:120-125. Cahill, J. J. F., and B. B. Casper. 2000. Investigating the relationship between neighbor root biomass and belowground competition: field evidence for symmetric competition belowground. Oikos 90:311-320. Canavan, S., L. Meyerson, J. Packer, P. Pyšek, N. Maurel, V. Lozano, D. Richardson, G. Brundu, K. Canavan, A. Cicatelli, J. Čuda, W. Dawson, F. Essl, F. Guarino, W.-Y. Guo, M. Kleunen, H. Kreft, C. Lambertini, J. Pergl, H. Skálová, R. Soreng, V. Visser, M. Vorontsova, P. Weigelt, M. Winter, and J. Wilson. 2019. Tall-statured grasses: a useful functional group for invasion science. Biological Invasions 21:37-58. Case, E. J., S. Harrison, and H. V. Cornell. 2016. Do high‐impact invaders have the strongest negative effects on abundant and functionally similar resident species? Functional Ecology 30:1447-1453.

107

Chippendale, J. F., and F. D. Panetta. 1994. The cost of parthenium weed to the Queensland cattle industry. Plant Protection Quarterly 9:73-76. Chytrý, M., L. C. Maskell, J. Pino, P. Pyšek, M. Vilà, X. Font, and S. M. Smart. 2008. Habitat invasions by alien plants: a quantitative comparison among Mediterranean, subcontinental and oceanic regions of Europe. Journal of Applied Ecology 45:448- 458. Chytrý, M., J. Wild, P. Pyšek, V. Jarošík, N. Dendoncker, I. Reginster, J. Pino, L. C. Maskell, M. Vilà, J. Pergl, I. Kühn, J. H. Spangenberg, and J. Settele. 2012. Projecting trends in plant invasions in Europe under different scenarios of future land‐use change. Global Ecology and Biogeography 21:75-87. Cingolani, A. M., I. Noy - Meir, and S. Díaz. 2005. Grazing effects on rangeland diversity: A synthesis of contemporary models. Ecological Applications 15:757-773. Conti, L., S. Block, M. Parepa, T. Münkemüller, W. Thuiller, A. T. R. Acosta, M. Kleunen, S. Dullinger, F. Essl, I. Dullinger, D. Moser, G. Klonner, O. Bossdorf, and M. Carboni. 2018. Functional trait differences and trait plasticity mediate biotic resistance to potential plant invaders. Journal of Ecology 106:1607-1620. Cook, G. D., and L. Dias. 2006a. It was no accident: deliberate plant introductions by Australian government agencies during the 20th century. It was no accident: deliberate plant introductions by Australian government agencies during the 20th century 54:601-625. Cook, G. D., and L. Dias. 2006b. It was no accident: deliberate plant introductions by Australian government agencies during the 20th century. Australian Journal of Botany 54:601-625. Cooke, B., and K. Moon. 2015. Aligning ‘public good’environmental stewardship with the landscape-scale: Adapting MBIs for private land conservation policy. Ecological Economics 114:152-158. Corbin, J. D., and C. M. D' Antonio. 2004. Competition between native perennial and exotic annual grasses: implications for an historical invasion. Ecology 85:1273-1283. Cowie, I. D. 1992. Alien plant species invasive in Kakadu National Park, tropical northern Australia. Biological Conservation 63:127-135. Craigie, I., A. Grech, R. L. Pressey, V. M. Adams, M. Hockings, M. Taylor, and M. Barnes. 2015. Terrestrial protected areas of Australia. Austral Ark: The State of Wildlife in Australia and New Zealand:5.

108

Cuneo, P., P. Gibson-Roy, G. Fifield, L. Broadhurst, T. Berryman, A. Crawford, and D. Freudenberger. 2018. Restoring grassy woodland diversity through direct seeding: Insights from six ‘best-practice’ case studies in southern Australia. Ecological Management & Restoration 19:124-135. Curtis, A., H. Ross, G. R. Marshall, C. Baldwin, J. Cavaye, C. Freeman, A. Carr, and G. J. Syme. 2014. The great experiment with devolved NRM governance: lessons from community engagement in Australia and New Zealand since the 1980s. Australasian Journal of Environmental Management 21:175-199. D'Antonio, C. M., and P. M. Vitousek. 1992. Biological invasions by exotic grasses, the grass/fire cycle, and global change. Annual Review of Ecology and Systematics:63- 87. D Antonio, C., J. Levine, and M. Thomsen. 2001. Ecosystem resistance to invasion and the role of propagule supply: a California perspective. Journal of Mediterranean Ecology 2:233-246. D’Antonio, C. 2000. Fire, plant invasions, and global changes. Pages 65–95 in H. Mooney and R. Hobbs, editors. Invasive species in a changing world. Island Press: Covelo, CA. Davies, K. F., P. Chesson, S. Harrison, B. D. Inouye, B. A. Melbourne, and K. J. Rice. 2005. Spatial heterogeneity explains the scale dependence of the native-exotic diversity relationship. Ecology 86:1602-1610. Davis, M. A., K. Thompson, and J. Philip Grime. 2005. Invasibility: the local mechanism driving community assembly and species diversity. Ecography 28:696-704. DeGasperis, B. G., and G. Motzkin. 2007. Windows of opportunity: historical and ecological controls on berberis thunbergii invasions. Ecology 88:3115-3125. Delaney, J. T., R. A. Moranz, D. M. Debinski, D. M. Engle, and J. R. Miller. 2016. Exotic- Dominated Grasslands Show Signs of Recovery with Cattle Grazing and Fire. PloS one 11:e0165758. Didham, R. K., J. M. Tylianakis, N. J. Gemmell, T. A. Rand, and R. M. Ewers. 2007. Interactive effects of habitat modification and species invasion on native species decline. Trends in Ecology & Evolution 22:489-496. Domènech, R., M. Vilà, J. Pino, and J. Gesti. 2005. Historical land‐use legacy and Cortaderia selloana invasion in the Mediterranean region. Global Change Biology 11:1054-1064.

109

Donald, P. F., R. E. Green, and M. F. Heath. 2001. Agricultural intensification and the collapse of europe's farmland bird populations. Proceedings: Biological Sciences 268:25-29. Dorrough, J., J. Ash, and S. McIntyre. 2004a. Plant responses to livestock grazing frequency in an Australian temperate grassland. Ecography 27:798-810. Dorrough, J., A. Yen, V. Turner, S. Clark, J. Crosthwaite, and J. Hirth. 2004b. Livestock grazing management and biodiversity conservation in Australian temperate grassy landscapes. Australian Journal of Agricultural Research 55:279-295. Douglass, L. L., H. P. Possingham, J. Carwardine, C. J. Klein, S. H. Roxburgh, J. Russell- Smith, and K. A. Wilson. 2011. The effect of carbon credits on savanna land management and priorities for biodiversity conservation. PloS one 6:e23843. Driscoll, D. A., J. A. Catford, J. N. Barney, P. E. Hulme, Inderjit, T. G. Martin, A. Pauchard, P. Pyšek, D. M. Richardson, S. R, and V. V. 2014. New pasture plants intensify invasive species risk. Proceedings of the National Academy of Sciences 111:16622- 16627. Dwyer, J. M., R. J. Fensham, and Y. M. Buckley. 2010. Agricultural legacy, climate, and soil influence the restoration and carbon potential of woody regrowth in Australia. Ecological Applications 20:1838-1850. Dyer, A. R., and K. J. Rice. 1999. Effects of competition on resource availability and growth of a California bunchgrass. Ecology 80:2697-2710. Earl, G., A. Curtis, C. Allan, and S. McDonald. 2010. Social acceptability of a duty of care for biodiversity. Australasian Journal of Environmental Management 17:8-17. Elith, J., J. R. Leathwick, and T. Hastie. 2008. A working guide to boosted regression trees. Journal of Animal Ecology 77:802-813. Ens, E., L. B. Hutley, N. A. Rossiter-Rachor, M. M. Douglas, and S. A. Setterfield. 2015. Resource-use efficiency explains grassy weed invasion in a low-resource savanna in north Australia. Frontiers in Plant Science 6. Essl, F., S. Dullinger, W. Rabitsch, P. E. Hulme, K. Hülber, V. Jarošík, I. Kleinbauer, F. Krausmann, I. Kühn, and W. Nentwig. 2011. Socioeconomic legacy yields an invasion debt. Proceedings of the National Academy of Sciences 108:203-207. Etter, A., C. McAlpine, and H. Possingham. 2008. Historical patterns and drivers of landscape change in colombia since 1500: A regionalized spatial approach. Annals of the Association of American Geographers 98:2-23.

110

Fairfax, R. J., and R. J. Fensham. 2000. The effect of exotic pasture development on floristic diversity in central Queensland, Australia. Biological Conservation 94:11-21. Fargione, J., C. S. Brown, and D. Tilman. 2003. Community assembly and invasion: An experimental test of neutral versus niche processes. Proceedings of the National Academy of Sciences of the United States of America 100:8916. Fensham, R. J. 1999. Native grasslands of the central highlands, queensland, australia. Floristics, regional context and conservation. The Rangeland Journal 21:82-103. Fensham, R. J., A. J. W. Biggs, D. W. Butler, and H. J. MacDermott. 2017. Brigalow forests and associated eucalypt woodlands of subtropical eastern Australia. Pages 389-409 in D. A. Keith, editor. Australian vegetation. Cambridge University Press, Cambridge. Fensham, R. J., D. W. Butler, R. J. Fairfax, A. R. Quintin, and J. M. Dwyer. 2016. Passive restoration of subtropical grassland after abandonment of cultivation. Journal of Applied Ecology 53:274-283. Fensham, R. J., S. Donald, and J. M. Dwyer. 2013. Propagule pressure, not fire or cattle grazing, promotes invasion of buffel grass Cenchrus ciliaris. Journal of Applied Ecology 50:138-146. Fensham, R. J., J. Fraser, H. J. MacDermott, and J. Firn. 2015a. Dominant tree species are at risk from exaggerated drought under climate change. Global Change Biology 21:3777-3785. Fensham, R. J., and G. P. Guymer. 2009. Carbon accumulation through ecosystem recovery. Environmental Science & Policy 12:367-372. Fensham, R. J., J. Halford, C. Hansen, B. Laffineur, and B. Williams. 2019. Threatened species in a threatened ecosystem: the conservation status of four Solanum species in the face of ongoing habitat loss. Oryx 53:439-449. Fensham, R. J., J. E. Holman, and M. J. Cox. 1999. Plant species responses along a grazing disturbance gradient in Australian grassland. Journal of Vegetation Science 10:77-86. Fensham, R. J., B. Laffineur, and J. L. Silcock. 2018. In the wake of bulldozers: Identifying threatened species in a habitat decimated by rapid clearance. Biological Conservation 219:28-34. Fensham, R. J., and S. D. Skull. 1999. Before Cattle: A Comparative Floristic Study of Eucalyptus Savanna Grazed by Macropods and Cattle in North Queensland, Australia 1. Biotropica 31:37-47.

111

Fensham, R. J., J. Wang, and C. Kilgour. 2015b. The relative impacts of grazing, fire and invasion by buffel grass (Cenchrus ciliaris) on the floristic composition of a rangeland savanna ecosystem. The Rangeland Journal 37:227-237. Firn, J., and Y. M. Buckley. 2010. Impacts of Invasive Plants on Australian Rangelands. Rangelands 32:48-51. Firn, J., A. MacDougall, S. Schmidt, and Y. Buckley. 2010. Early emergence and resource availability can competitively favour natives over a functionally similar invader. Oecologia 163:775-784. Firn, J., T. Rout, H. Possingham, and Y. M. Buckley. 2008. Managing beyond the invader: manipulating disturbance of natives simplifies control efforts. Journal of Applied Ecology 45:1143-1151. Foley, J. A., R. Defries, G. P. Asner, C. Barford, G. Bonan, S. R. Carpenter, F. S. Chapin, M. T. Coe, G. C. Daily, H. K. Gibbs, J. H. Helkowski, T. Holloway, E. A. Howard, C. J. Kucharik, C. Monfreda, J. A. Patz, I. C. Prentice, N. Ramankutty, and P. K. Snyder. 2005. Global consequences of land use. Science 309:570-574. Forseth, I. N., and A. F. Innis. 2004. Kudzu (Pueraria montana): History, physiology, and ecology combine to make a major ecosystem threat. Critical Reviews in Plant Sciences 23:401-413. Foster, B. L., and D. Tilman. 2003. Seed limitation and the regulation of community structure in oak savanna grassland. Journal of Ecology 91:999-1007. Frank, A. S. K., G. M. Wardle, C. R. Dickman, and A. C. Greenville. 2014. Habitat- and rainfall-dependent biodiversity responses to cattle removal in an arid woodland– grassland environment. Ecological Applications 24:2013-2028. Fridley, J. D., J. J. Stachowicz, S. Naeem, D. F. Sax, E. W. Seabloom, M. D. Smith, T. J. Stohlgren, D. Tilman, and B. Von Holle. 2007. The Invasion Paradox: Reconciling Pattern and Process in Species Invasions. Ecology 88:3-17. Fuhlendorf, S. D., H. Zhang, T. R. Tunnell, D. M. Engle, and A. F. Cross. 2002. Effects of grazing on restoration of southern mixed prairie soils. Restoration Ecology 10:401- 407. Funk, J. L. 2008. Differences in plasticity between invasive and native plants from a low resource environment. Journal of Ecology 96:1162-1173. Funk, J. L., and P. M. Vitousek. 2007. Resource-use efficiency and plant invasion in low- resource systems. Nature 446:1079.

112

Funk, J. L., and A. A. Wolf. 2016. Testing the trait‐based community framework: Do functional traits predict competitive outcomes? Ecology 97:2206-2211. Gaertner, M., A. Den Breeyen, H. Cang, and D. M. Richardson. 2009. Impacts of alien plant invasions on species richness in Mediterranean-type ecosystems: a meta-analysis. Progress in Physical Geography 33:319-338. Gallagher, R. V., D. Englert Duursma, J. O’Donnell, P. D. Wilson, P. O. Downey, L. Hughes, and M. R. Leishman. 2013. The grass may not always be greener: projected reductions in climatic suitability for exotic grasses under future climates in Australia. Biological Invasions 15:961-975. Gallo, J. A., L. Pasquini, B. Reyers, and R. M. Cowling. 2009. The role of private conservation areas in biodiversity representation and target achievement within the Little Karoo region, South Africa. Biological Conservation 142:446-454. Gaston, K. J. 1998. Rarity as double jeopardy. Nature 394:229-230. Gelman, A., and J. Hill. 2007. Data analysis using regression and multi-level/hierarchical models. Cambridge University Press, New York. Gilbert, B., and J. M. Levine. 2013. Plant invasions and extinction debts. Proceedings of the National Academy of Sciences 110:1744-1749. Greiner, R., and D. Gregg. 2011. Farmers’ intrinsic motivations, barriers to the adoption of conservation practices and effectiveness of policy instruments: Empirical evidence from northern Australia. Land Use Policy 28:257-265. Grice, A. C. 2006. The impacts of invasive plant species on the biodiversity of Australian rangelands. The Rangeland Journal 28:27-35. Griffin, G., D. S. Smith, S. R. Morton, G. Allan, K. Masters, and N. Preece. 1989. Status and implications of the invasion of tamarisk (Tamarix aphylla) on the Finke River, Northern Territory, Australia. Journal of Environmental Management 29:297-315. Hajkowicz, S., and K. Collins. 2009. Measuring the benefits of environmental stewardship in rural landscapes. Landscape and Urban Planning 93:93-102. Hanski, I. 1998. Metapopulation dynamics. Nature 396:41-49. Hejda, M., and P. Pyšek. 2006. What is the impact of Impatiens glandulifera on species diversity of invaded riparian vegetation? Biological Conservation 132:143-152. Hejda, M., P. Pyšek, and V. Jarošík. 2009. Impact of invasive plants on the species richness, diversity and composition of invaded communities. Journal of Ecology 97:393-403.

113

Higgins, S. I., S. Lavorel, and E. Revilla. 2003. Estimating plant migration rates under habitat loss and fragmentation. Oikos 101:354-366. Hobbs, R. J. 2001. Synergisms among Habitat Fragmentation, Livestock Grazing, and Biotic Invasions in Southwestern Australia. Conservation Biology 15:1522-1528. Hobbs, R. J., and L. F. Huenneke. 1992. Disturbance, diversity, and invasion: implications for conservation. Conservation Biology 6:324-337. Hoffmann, W. A., and M. Haridasan. 2008. The invasive grass, Melinis minutiflora, inhibits tree regeneration in a Neotropical savanna. Austral Ecology 33:29-36. Holmes, P. M., and R. M. Cowling. 1997. The effects of invasion by acacia saligna on the guild structure and regeneration capabilities of south african fynbos shrublands. Journal of Applied Ecology 34:317-332. Howden, M. 1988. Some aspects of the ecology of four tropical grasses, with special emphasis on Bothriochloa pertusa. Griffith University, Brisbane. Hulme, P. E. 2012. Weed risk assessment: a way forward or a waste of time? Pages 10-19. Blackwell Publishing Ltd, Oxford, UK. Hunt, L., J. McIvor, A. Grice, and S. Bray. 2014. Principles and guidelines for managing cattle grazing in the grazing lands of northern Australia: stocking rates, pasture resting, prescribed fire, paddock size and water points–a review. The Rangeland Journal 36:105-119. Hussain, F., H. H. Naqvi, and I. Ilahi. 1982. Interference exhibited by Cenchrus ciliaris L. and Bothriochloa pertusa (L.) A. Camus. Bulletin of the Torrey Botanical Club 109:513-523. Jackson, J. 2005. Is there a relationship between herbaceous species richness and buffel grass (Cenchrus ciliaris)? Austral Ecology 30:505-517. Jackson, J., and A. Ash. 1998. Tree-grass relationships in open eucalypt woodlands of northeastern Australia: influence of trees on pasture productivity, forage quality and species distribution. An International Journal incorporating Agroforestry Forum 40:159-176. Jackson, S. T., and D. F. Sax. 2010. Balancing biodiversity in a changing environment: extinction debt, immigration credit and species turnover. Trends in Ecology & Evolution 25:153-160.

114

Jauffret, S., and S. Lavorel. 2003. Are Plant Functional Types Relevant to Describe Degradation in Arid, Southern Tunisian Steppes? Journal of Vegetation Science 14:399-408. Jones, R. J. 1997. Steer gains, pasture yield and pasture composition on native pasture and on native pasture oversewn with Indian couch (Bothriochloa pertusa) at three stocking rates. Australian Journal of Agricultural Research 37:755-765. Kamal, S., M. Grodzińska-Jurczak, and G. Brown. 2015. Conservation on private land: a review of global strategies with a proposed classification system. Journal of Environmental Planning and Management 58:576-597. Keeley, J. E., M. Baer-Keeley, and C. Fotheringham. 2005. Alien plant dynamics following fire in Mediterranean‐climate California shrublands. Ecological Applications 15:2109-2125. Kehoe, J. 2009. Environmental law making in Queensland: The Vegetation Management Act 1999 (Qld). Environmental and planning law journal 26:392-410. Keser, L. H., W. Dawson, Y.-B. Song, F.-H. Yu, M. Fischer, M. Dong, and M. van Kleunen. 2014. Invasive clonal plant species have a greater root-foraging plasticity than non- invasive ones. Oecologia 174:1055-1064. Kleijn, D., and W. J. Sutherland. 2003. How effective are European agri-environment schemes in conserving and promoting biodiversity? Journal of Applied Ecology 40:947-969. Koci, J., R. C. Sidle, A. E. Kinsey-Henderson, R. Bartley, S. N. Wilkinson, A. A. Hawdon, B. Jarihani, C. H. Roth, and L. Hogarth. 2020. Effect of reduced grazing pressure on sediment and nutrient yields in savanna rangeland streams draining to the Great Barrier Reef. Journal of Hydrology 582:124520. Kutt, A. S., and A. Fisher. 2011. Increased grazing and dominance of an exotic pasture (Bothriochloa pertusa) affects vertebrate fauna species composition, abundance and habitat in savanna woodland. The Rangeland Journal 33:49-58. Kutt, A. S., and J. E. Kemp. 2012. Native plant diversity in tropical savannas decreases when exotic pasture grass cover increases. The Rangeland Journal 34:183-189. Kuussaari, M., R. Bommarco, R. K. Heikkinen, A. Helm, J. Krauss, R. Lindborg, E. Öckinger, M. Pärtel, J. Pino, F. Rodà, C. Stefanescu, T. Teder, M. Zobel, and I. Steffan-Dewenter. 2009. Extinction debt: a challenge for biodiversity conservation. Trends in Ecology & Evolution 24:564-571.

115

Ladouceur, E., and M. M. Mayfield. 2017. The early response of subtropical tussock grasslands to restoration treatments. Restoration Ecology 25:689-695. Lai, H. R., M. M. Mayfield, J. M. Gay‐Des‐Combes, T. Spiegelberger, and J. M. Dwyer. 2015. Distinct invasion strategies operating within a natural annual plant system. Ecology Letters 18:336-346. Laliberté, E., H. Lambers, T. I. Burgess, and S. J. Wright. 2015. Phosphorus limitation, soil- borne pathogens and the coexistence of plant species in hyperdiverse forests and shrublands. New Phytologist 206:507-521. Landsberg, J., C. D. James, J. Maconochie, A. O. Nicholls, J. Stol, and R. Tynan. 2002. Scale‐related effects of grazing on native plant communities in an arid rangeland region of South Australia. Journal of Applied Ecology 39:427-444. Landsberg, J., C. D. James, S. R. Morton, W. J. Müller, and J. Stol. 2003. Abundance and composition of plant species along grazing gradients in Australian rangelands. Journal of Applied Ecology 40:1008-1024. Laurance, W. F., H. E. M. Nascimento, S. G. Laurance, A. C. Andrade, P. M. Fearnside, J. E. L. Ribeiro, and R. L. Capretz. 2006. Rainforest fragmentation and the proliferation of successional trees. Ecology 87:469-482. Lavergne, S., and J. Molofsky. 2007. Increased genetic variation and evolutionary potential drive the success of an invasive grass. Proceedings of the National Academy of Sciences 104:3883. Le Maitre, D. C., B. W. Van Wilgen, R. A. Chapman, and D. H. McKelly. 1996. Invasive plants and water resources in the western cape province, south africa: modelling the consequences of a lack of management. Journal of Applied Ecology 33:161-172. Legge, S., M. S. Kennedy, R. Lloyd, S. A. Murphy, and A. Fisher. 2011. Rapid recovery of mammal fauna in the central Kimberley, northern Australia, following the removal of introduced herbivores. Austral Ecology 36:791-799. Lepers, E., E. F. Lambin, A. C. Janetos, R. DeFries, F. Achard, N. Ramankutty, and R. J. Scholes. 2005. A synthesis of information on rapid land-cover change for the period 1981–2000. Bioscience 55:115-124. Levine, J. M., Vil, xe, Montserrat, C. M. D'Antonio, J. S. Dukes, K. Grigulis, and S. Lavorel. 2003. Review paper. Mechanisms underlying the impacts of exotic plant invasions. Proceedings: Biological Sciences 270:775-781.

116

Lindenmayer, D., R. J. Hobbs, R. Montague-Drake, J. Alexandra, A. Bennett, M. Burgman, P. Cale, A. Calhoun, V. Cramer, P. Cullen, D. Driscoll, L. Fahrig, J. Fischer, J. Franklin, Y. Haila, M. Hunter, P. Gibbons, S. Lake, G. Luck, C. MacGregor, S. McIntyre, R. M. Nally, A. Manning, J. Miller, H. Mooney, R. Noss, H. Possingham, D. Saunders, F. Schmiegelow, M. Scott, D. Simberloff, T. Sisk, G. Tabor, B. Walker, J. Wiens, J. Woinarski, and E. Zavaleta. 2008. A checklist for ecological management of landscapes for conservation. Ecology Letters 11:78-91. Lindsay, E. A., and S. A. Cunningham. 2011. Native Grass Establishment in Grassy Woodlands with Nutrient Enriched Soil and Exotic Grass Invasion. Restoration Ecology 19:131-140. Lockwood, J. L., P. Cassey, and T. Blackburn. 2005. The role of propagule pressure in explaining species invasions. Trends in ecology & evolution (Amsterdam) 20:223- 228. Lonsdale, W. M. 1994. Inviting trouble: Introduced pasture species in northern Australia. Australian Journal of Ecology 19:345-354. Lonsdale, W. M. 1999. Global patterns of plant invasion and the concept of invasibility. Ecology 80:1522-1536. Lunt, I. D., A. Jansen, D. L. Binns, and S. A. Kenny. 2007. Long-term effects of exclusion of grazing stock on degraded herbaceous plant communities in a riparian Eucalyptus camaldulensis forest in south-eastern Australia. Austral Ecology 32:937-949. MacDougall, A. S., and R. Turkington. 2004. Relative importance of suppression‐based and tolerance‐based competition in an invaded oak savanna. Journal of Ecology 92:422- 434. MacDougall, A. S., and R. Turkington. 2005. Are invasive species the drivers or passengers of change in degraded ecosystems? Ecology 86:42-55. Mack, R. N., D. Simberloff, W. Mark Lonsdale, H. Evans, M. Clout, and F. A. Bazzaz. 2000. Biotic invasions: causes, epidemiology, global consequences, and control. Ecological Applications 10:689-710. Marshall, N. A., M. Friedel, R. D. van Klinken, and A. C. Grice. 2011. Considering the social dimension of invasive species: the case of buffel grass. Environmental Science & Policy 14:327-338. Martin, T. G., H. Murphy, A. Liedloff, C. Thomas, I. Chadès, G. Cook, R. Fensham, J. McIvor, and R. D. van Klinken. 2015. Buffel grass and climate change: a framework

117

for projecting invasive species distributions when data are scarce. Biological Invasions 17:3197-3210. Marvier, M., P. Kareiva, and M. G. Neubert. 2004. Habitat Destruction, Fragmentation, and Disturbance Promote Invasion by Habitat Generalists in a Multispecies Metapopulation. Risk Analysis 24:869-878. Maskell, L. C., L. G. Firbank, K. Thompson, J. M. Bullock, and S. M. Smart. 2006. Interactions between non‐native plant species and the floristic composition of common habitats. Journal of Ecology 94:1052-1060. Maxwell, S. 2016. The ravages of guns, nets and bulldozers. Nature 536:143-145. McAlpine, C. A., A. Etter, P. M. Fearnside, L. Seabrook, and W. F. Laurance. 2009. Increasing world consumption of beef as a driver of regional and global change: A call for policy action based on evidence from Queensland (Australia), Colombia and Brazil. Global Environmental Change 19:21-33. McAlpine, C. A., R. J. Fensham, and D. E. Temple-Smith. 2002. Biodiversity conservation and vegetation clearing in Queensland: principles and thresholds. The Rangeland Journal 24:36-55. McDonald, S. E., N. Reid, R. Smith, C. M. Waters, J. Hunter, and R. Rader. 2019. Rotational grazing management achieves similar plant diversity outcomes to areas managed for conservation in a semi-arid rangeland. The Rangeland Journal 41:135-145. McIntyre, S., J. G. McIvor, and K. M. Heard. 2004. Managing and conserving grassy woodlands. CSIRO Publishing, Collingwood, Victoria. McIvor, J. 2012. Sustainable management of the Burdekin grazing lands-A technical guide of options for stocking rate management, pasture spelling, infrastructure development and prescribed burning to optimise animal production, profitability, land condition and water quality outcomes. Department Of Agriculture and Fisheries. State of Queensland. McIvor, J., and C. Gardener. 1994. Germinable soil seed banks in native pastures in north- eastern Australia. Australian Journal of Experimental Agriculture 34:1113. McIvor, J. G. 2007. Pasture management in semi-arid tropical woodlands: dynamics of perennial grasses. The Rangeland Journal 29:87-100. McIvor, J. G., V. Singh, J. P. Corfield, and R. J. Jones. 1996. Seed production by native and naturalised grasses in north-east Queensland: effects of stocking rate and season. Tropical Grasslands 30:262-269.

118

Melzer, R., and A. Melzer. 2017. Managing buffel grass to promote species richness and facilitate ecosystem recovery. Pages 88-93 14th Queensland Weed Symposium. Weed Society of Queensland. Melzer, R. I. 2015. When is stock grazing an appropriate ‘tool’ for reducing Cenchrus Ciliaris (buffel grass) on conservation reserves? Proceedings of the Royal Society of Queensland 120:53-68. Milchunas, D. G., O. E. Sala, and W. K. Lauenroth. 1988. A generalized model of the effects of grazing by large herbivores on grassland community structure. The American Naturalist:87-106. Miles, J. F. 1949. Plant introduction trials in central coastal Queensland 1936-46. Commonwealth Scientific and Industrial Research Organization, Melbourne. Miller, G., M. Friedel, P. Adam, and V. Chewings. 2010. Ecological impacts of buffel grass (Cenchrus ciliaris L.) invasion in central Australia does field evidence support a fire- invasion feedback? The Rangeland Journal 32:353-365. Miller, I. L. 2004. Uses for Mimosa pigra. , p. 63-67.in 3rd international symposium on the management of Mimosa pigra, Charles Darwin University, Darwin Australia. Mistry, J., and A. Berardi. 2005. Assessing fire potential in a brazilian savanna nature reserve. Biotropica 37:439-451. Mommer, L., E. Visser, J. Ruijven, H. Caluwe, R. Pierik, and H. Kroon. 2011. Contrasting root behaviour in two grass species: a test of functionality in dynamic heterogeneous conditions. Plant and Soil 344:347-360. Moon, K., and C. Cocklin. 2011. Participation in biodiversity conservation: motivations and barriers of Australian landholders. Journal of Rural Studies 27:331-342. Mosher, E., J. Silander, and A. Latimer. 2009. The role of land-use history in major invasions by woody plant species in the northeastern North American landscape. Biological Invasions 11:2317-2328. Müller, B., K. Frank, and C. Wissel. 2007. Relevance of rest periods in non-equilibrium rangeland systems – A modelling analysis. Agricultural Systems 92:295-317. Müllerová, J., P. Pyšek, V. Jarošík, and J. Pergl. 2005. Aerial photographs as a tool for assessing the regional dynamics of the invasive plant species Heracleum mantegazzianum. Journal of Applied Ecology 42:1042-1053. Myers, N., and J. Kent. 2004. The new consumers: the influence of affluence on the environment. Island Press, Washington, D.C.

119

Neilly, H., P. O’Reagain, J. Vanderwal, and L. Schwarzkopf. 2018. Profitable and sustainable cattle grazing strategies support reptiles in tropical savanna rangeland. Rangeland ecology & management 71:205-212. Neilly, H., and L. Schwarzkopf. 2018. Heavy livestock grazing negatively impacts a marsupial ecosystem engineer. Journal of Zoology 305:35-42. Neldner, V. J., Niehus, R.E., Wilson, B.A., McDonald, W.J.F., Ford, A.J. 2015. Descriptions of Broad Vegetation Groups, Version 2.0. Queensland Herbarium, Department of Environment and Science, Brisbane, Australia. O'Reagain, P., J. Bushell, C. Holloway, and A. Reid. 2009. Managing for rainfall variability: effect of grazing strategy on cattle production in a dry tropical savanna. Animal Production Science 49:85-99. O'Reagain, P., J. Bushell, and B. Holmes. 2011. Managing for rainfall variability: long-term profitability of different grazing strategies in a northern Australian tropical savanna. Managing for rainfall variability: long-term profitability of different grazing strategies in a northern Australian tropical savanna 51:210-224. O'Reagain, P., J. Bushelle, L. Pahl, and J. Scanlan. 2018. Wambiana grazing trial Phase 3: stocking strategies for improving carrying capacity, land condition and biodiversity outcomes. Meat and Livestock Australia, Sydney. O'Reagain, P. J., and J. C. Scanlan. 2013. Sustainable management for rangelands in a variable climate: evidence and insights from northern Australia. Animal: An International Journal of Animal Bioscience 7:68-78. O’Reagain, P. 2006. Testing and developing principles and management guidelines for the sustainable management of the seasonably variable tropical savannas Meat & Livestock Australia Limited, North Sydney, Australia. O’Reagain, P., and J. Bushell. 2015. Wambiana grazing trial Phase 2: Stocking and spelling strategies for improving carrying capacity and land condition in north Australian grazing lands. Meat and Livestock Australia, Sydney. Orr, D. M., and P. J. Reagain. 2011. Managing for rainfall variability: impacts of grazing strategies on perennial grass dynamics in a dry tropical savanna. The Rangeland Journal 33:209-220. Osunkoya, O. O., J. G. Froese, S. Nicol, C. Perrett, K. Moore, J. Callander, and S. Campbell. 2019. A risk‐based inventory of invasive plant species of Queensland, Australia: Regional, ecological and floristic insights. Austral Ecology 44:1123-1138.

120

Parnphumeesup, P., and S. A. Kerr. 2015. Willingness to pay for gold standard carbon credits. Energy Sources, Part B: Economics, Planning, and Policy 39:412-417. Partridge, I. J., and C. P. Miller. 1991. Sown pastures for the seasonally dry tropics. Queensland Dept. of Primary Industries, Brisbane. Pérez-Harguindeguy, N., S. Díaz, E. Garnier, S. Lavorel, H. Poorter, P. Jaureguiberry, M. S. Bret-Harte, W. K. Cornwell, J. M. Craine, D. E. Gurvich, C. Urcelay, E. J. Veneklaas, P. B. Reich, L. Poorter, I. J. Wright, P. Ray, L. Enrico, J. G. Pausas, A. C. de Vos, N. Buchmann, G. Funes, F. Quétier, J. G. Hodgson, K. Thompson, H. D. Morgan, H. ter Steege, L. Sack, B. Blonder, P. Poschlod, M. V. Vaieretti, G. Conti, A. C. Staver, S. Aquino, and J. H. C. Cornelissen. 2013. New handbook for standardised measurement of plant functional traits worldwide. Australian Journal of Botany 61:167-234. Pinheiro, J. C., D. M. Bates, S. DebRoy, D. Sarkar, and R Core Team. 2016. nlme: linear and nonlinear mixed effects models. R package version 3.1-128, http://CRAN.R- project.org/package=nlme. Pocewicz, A., J. M. Kiesecker, G. P. Jones, H. E. Copeland, J. Daline, and B. A. Mealor. 2011. Effectiveness of conservation easements for reducing development and maintaining biodiversity in sagebrush ecosystems. Biological Conservation 144:567- 574. Powell, K. I., J. M. Chase, and T. M. Knight. 2011. A synthesis of plant invasion effects on biodiversity across spatial scales. The American Journal of Botany 98:539-548. Powell, K. I., J. M. Chase, and T. M. Knight. 2013. Invasive plants have scale-dependent effects on diversity by altering species-area relationships. Science 339:316-318. Prober, S. M., K. R. Thiele, I. D. Lunt, and T. B. Koen. 2005. Restoring ecological function in temperate grassy woodlands: manipulating soil nutrients, exotic annuals and native perennial grasses through carbon supplements and spring burns. Journal of Applied Ecology 42:1073-1085. Puth, L. M., and D. M. Post. 2005. Studying invasion: have we missed the boat? Ecology Letters 8:715-721. Pyšek, P., S. Bacher, M. Chytrý, V. Jarošík, J. Wild, L. Celesti-Grapow, N. Gassó, M. Kenis, P. W. Lambdon, W. Nentwig, J. Pergl, A. Roques, J. Sádlo, W. Solarz, M. Vilà, and P. E. Hulme. 2010. Contrasting patterns in the invasions of European terrestrial and freshwater habitats by alien plants, insects and vertebrates. Global Ecology and Biogeography 19:317-331.

121

Pyšek, P., V. Jarošík, P. E. Hulme, J. Pergl, M. Hejda, U. Schaffner, and M. Vilà. 2012. A global assessment of invasive plant impacts on resident species, communities and ecosystems: the interaction of impact measures, invading species' traits and environment. Global Change Biology 18:1725-1737. Queensland Herbarium. 2017. Biodiversity status of pre-clearing, 1997 and 2017 remnant regional ecosystems of Queensland series, Version 11.0 (December 2017). State of Queensland (Department of Environment and Science), Brisbane. Queensland Government, 2017. Land Use Mapping Series (QLUMP). Brisbane. Reside, A. E., J. Beher, A. J. Cosgrove, M. C. Evans, L. Seabrook, J. L. Silcock, A. S. Wenger, and M. Maron. 2017. Ecological consequences of land clearing and policy reform in Queensland. Pacific conservation biology 23:219-230. Richardson, D. M., P. Pyšek, M. Rejmánek, M. G. Barbour, F. D. Panetta, and C. J. West. 2000. Naturalization and invasion of alien plants: concepts and definitions. Diversity and Distributions 6:93-107. Rodríguez-Echeverría, S., J. Crisóstomo, C. Nabais, and H. Freitas. 2009. Belowground mutualists and the invasive ability of Acacia longifolia in coastal dunes of Portugal. Biological Invasions 11:651-661. Rossiter - Rachor, N. A., S. A. Setterfield, M. M. Douglas, L. B. Hutley, G. D. Cook, and S. Schmidt. 2009. Invasive Andropogon gayanus (gamba grass) is an ecosystem transformer of nitrogen relations in Australian savanna. Ecological Applications 19:1546-1560. Sala, O., F. Chapin, J. Armesto, and E. Berlow. 2000. Global biodiversity scenarios for the year 2100. Science 287:1770-1774. Sampaio, A. B., D. L. M. Vieira, K. D. Holl, K. F. Pellizzaro, M. Alves, A. G. Coutinho, A. Cordeiro, J. F. Ribeiro, and I. B. Schmidt. 2019. Lessons on direct seeding to restore Neotropical savanna. Ecological Engineering 138:148-154. Scanlan, J. C., A. J. Pressland, and D. J. Myles. 1996. Grazing modifies woody and herbaceous components of north Queensland woodlands. The Rangeland Journal 18:47-57. Schmidt, C. D., K. R. Hickman, R. Channell, K. Harmoney, and W. Stark. 2007. Competitive abilities of native grasses and non-native (Bothriochloa spp.) grasses. Plant Ecology 197:69-80.

122

Seabloom, E. W., W. Stanley Harpole, O. J. Reichman, and D. Tilman. 2003. Invasion, competitive dominance, and resource use by exotic and native California grassland species. Proceedings of the National Academy of Sciences of the United States of America 100:13384. Setterfield, S. A., M. M. Douglas, L. B. Hutley, and M. A. Welch. 2005. Effects of Canopy Cover and Ground Disturbance on Establishment of an Invasive Grass in an Australia Savanna. Biotropica 37:25-31. Setterfield, S. A., N. A. Rossiter‐Rachor, L. B. Hutley, M. M. Douglas, and R. J. Williams. 2010. Biodiversity research: Turning up the heat: the impacts of Andropogon gayanus (gamba grass) invasion on fire behaviour in northern Australian savannas. Diversity and Distributions 16:854-861. Shea, K., and P. Chesson. 2002. Community ecology theory as a framework for biological invasions. Trends in Ecology and Evolution 17:170-176. Silcock, J. L., and R. J. Fensham. 2018. Using evidence of decline and extinction risk to identify priority regions, habitats and threats for plant conservation in Australia. Australian Journal of Botany 66:541-555. Silcock, R. G. 1991. Pastures, trees and shrubs for rehabilitating mines in Queensland : impediments to their use on opencut coal and alluvial mines. Queensland Department of Primary Industries, Brisbane, Queensland. Silcock, R. G., C. H. Finlay, D. S. Loch, and G. L. Harvey. 2015. Perennial pastures for marginal farming country in southern Queensland. 2. Potential new grass cultivar evaluation. Tropical Grasslands-Forrajes Tropicales 3:15-26. Simberloff, D., and B. Von Holle. 1999. Positive interactions of nonindigenous species: invasional meltdown? Biological Invasions 1:21-32. Smith, P. 2014. Do grasslands act as a perpetual sink for carbon? Global Change Biology 20:2708-2711. Söderström, B., S. Kiema, and R. S. Reid. 2003. Intensified agricultural land-use and bird conservation in Burkina Faso. Agriculture, Ecosystems and Environment 99:113-124. Souter, N. J., and T. Milne. 2009. Grazing exclusion as a conservation measure in a South Australian temperate native grassland. Grassland Science 55:79-88. Spiegel, N. 2016. Developing an RD&E project to address loss of productivity in Queensland pastures invaded by Indian couch (Bothriochloa pertusa), Department of Agriculture and Fisheries, State of Queensland

123

Standish, R. J., V. A. Cramer, S. L. Wild, and R. J. Hobbs. 2007. Seed dispersal and recruitment limitation are barriers to native recolonization of old-fields in western Australia. Journal of Applied Ecology 44:435-445. The State of Queensland, 2020. Selecting the right pasture species. Available at https://www.business.qld.gov.au/industries/farms-fishing-forestry/agriculture/grazing- pasture/improving/species (Accessed 24 January 2020). Steinfeld, H., and T. Wassenaar. 2007. The Role of Livestock Production in Carbon and Nitrogen Cycles. Annual Review of Environment and Resources 32:271-294. Stephens, S. 2001. Visions and viability: how achievable is landscape conservation in Australia? Ecological Management & Restoration 2:189-195. Stohlgren, T. J., and M. Rejmánek. 2014. No universal scale-dependent impacts of invasive species on native plant species richness. Biology Letters 10:20130939. Stone, L. M., M. Byrne, and J. G. Virtue. 2008. An environmental weed risk assessment model for Australian forage improvement programs. Australian Journal of Experimental Agriculture 48:568-574. Syphard, A. D., and J. Franklin. 2009. Differences in spatial predictions among species distribution modeling methods vary with species traits and environmental predictors. Ecography 32:907-918. Taylor, M., P. Sattler, J. Fitzsimons, C. Curnow, D. Beaver, L. Gibson, and G. Llewellyn. 2011. Building nature’s safety net 2011: the state of protected areas for Australia’s ecosystems and wildlife. Tilman, D. 2004. Niche tradeoffs, neutrality, and community structure: A stochastic theory of resource competition, invasion, and community assembly. Proceedings of the National Academy of Sciences of the United States of America 101:10854. Towers, I. R., and J. M. Dwyer. 2018. Regional climate and local-scale biotic acceptance explain native-exotic richness relationships in Australian annual plant communities. Proceedings of the Royal Society B 285:20181328. Tozer, K. N., D. F. Chapman, P. E. Quigley, P. M. Dowling, R. D. Cousens, G. A. Kearney, and J. R. Sedcole. 2008. Controlling invasive annual grasses in grazed pastures: population dynamics and critical gap sizes. Journal of Applied Ecology 45:1152- 1159.

124

Truong, P. N., and M. McDowell. 1985. Indian bluegrass for soil conservation and land stabilization in Queensland. Journal of the Soil Conservation Service of New South Wales 41:38-43. Tucker, S. S., J. M. Craine, and J. B. Nippert. 2011. Physiological drought tolerance and the structuring of tallgrass prairie assemblages. Ecosphere 2:1-19. Turcotte, M. M., and J. M. Levine. 2016. Phenotypic plasticity and species coexistence. Trends in Ecology & Evolution 31:803-813. Van Kleunen, M., E. Weber, and M. Fischer. 2010. A meta‐analysis of trait differences between invasive and non‐invasive plant species. Ecology Letters 13:235-245. van Klinken, R. D., and M. H. Friedel. 2017. Unassisted invasions: understanding and responding to Australia’s high-impact environmental grass weeds. Australian Journal of Botany 65:678-690. Venter, O., E. Meijaard, H. Possingham, R. Dennis, D. Sheil, S. Wich, L. Hovani, and K. Wilson. 2009. Carbon payments as a safeguard for threatened tropical mammals. Conservation letters 2:123-129. Vilà, M., and I. Ibáñez. 2011. Plant invasions in the landscape. Landscape Ecology 26:461- 472. Vilà, M., J. Pino, and X. Font. 2007. Regional assessment of plant invasions across different habitat types. Journal of Vegetation Science 18:35-42. Wahl, S., and P. Ryser. 2000. Root tissue structure is linked to ecological strategies of grasses. New Phytologist 148:459-471. Walker, B., and E. J. Weston. 1990. Pasture development in Queensland - a success story. Tropical Grasslands 24:257-268. Walker, D. 2013. Landcare: Linking Australia’s landscapes by linking its land managers. Pages 185-196 Linking Australia’s landscapes: Lessons and opportunities from large- scale conservation networks. CSIRO Publishing. Wardle, D. A., R. D. Bardgett, R. M. Callaway, and W. H. Van der Putten. 2011. Terrestrial ecosystem responses to species gains and losses. Science 332:1273-1277. Warren, R. J., T. Ursell, A. D. Keiser, and M. A. Bradford. 2013. Habitat, dispersal and propagule pressure control exotic plant infilling within an invaded range. Ecosphere 4:1-12.

125

Waters, C. M., S. E. Orgill, G. J. Melville, I. D. Toole, and W. J. Smith. 2017. Management of grazing intensity in the semi‐arid rangelands of southern Australia: effects on soil and biodiversity. Land Degradation and Development 28:1363-1375. Williams, D., and Z. Baruch. 2000. African grass invasion in the Americas: ecosystem consequences and the role of ecophysiology. Biological Invasions 2:123-140. Wilson, B. A., V. J. Neldner, and A. Accad. 2002a. The extent and status of remnant vegetation in Queensland and its implications for statewide vegetation management and legislation. The Rangeland Journal 24:6-35. Wilson, B. A., V. J. Neldner, and A. Accad. 2002b. The extent and status of remnant vegetation in Queensland and its implications for statewide vegetation management and legislation. The Rangeland Journal 24:6-35. Wohlwend, M. R., M. R. Schutzenhofer, and T. M. Knight. 2019. Long-term experiment manipulating community assembly results in favorable restoration outcomes for invaded prairies. Restoration Ecology 27:1307-1316. Woinarski, J. C. Z., J. C. McCosker, G. Gordon, B. Lawrie, C. James, J. Augusteyn, L. Slater, and T. Danvers. 2006. Monitoring change in the vertebrate fauna of central Queensland, Australia, over a period of broad-scale vegetation clearance, 19732002. Wildlife Research 33:263-274. Yates, C. J., D. A. Norton, and R. J. Hobbs. 2000. Grazing effects on plant cover, soil and microclimate in fragmented woodlands in south‐western Australia: implications for restoration. Austral Ecology 25:36-47. Zavaleta, E. 2000. Valuing ecosystem services lost to Tamarix invasion in the United States. Pages 261-300 Invasive species in a changing world. Island Press, Washington, D.C. Zeidler, J., S. Hanrahan, and M. Scholes. 2002. Land-use intensity affects range condition in arid to semi-arid Namibia. Journal of Arid Environments 52:389-403.

126

Appendices

Appendix 2-1 – Detailed results from landscape scale data validation analysis used in Chapter 2.

Either due to satellite interpretation error or ground truthing limitations the mapping databases used are prone to some errors in remnant classification. We quantified some of this error by assessing a subset of polygons (remnant to non-remnant and non-remnant to remnant) for each land-use category (12 groups; two remnant status groups and 6 land-use categories) using high resolution satellite imagery (30 – 15 cm resolution).

For each group we verified either every polygon greater than 1 ha or up to 100 polygons. Polygons which changed from non-remnant to remnant status, were inspected for evidence of vegetation clearing or intense use of land (i.e. cropping or mining) and if this was evident within more than 50 % of the polygon an error recorded. For polygons which changed from remnant to non-remnant an error was recorded if the polygon had canopy cover representative of its BVG and similar to surrounding patches of remnant vegetation within the same BVG. In total we verified 2176 km2 over 918 polygons. Overall, the remnant status and land-use categorisation was correct for 78 % of polygons, equating to 22 % error. The number of suspected errors for each group is presented as a percentage of the total polygons inspected (Table 1).

127

Table 1 Total number of polygons and total area inspected for errors using high resolution satellite imagery. Polygons were inspected in groups based on their land-use category and change in remnant status; R – NR = remnant to non-remnant and NR – R = non-remnant to remnant. The percent correct and percent error of total polygons inspected is shown.

Percent Percent Remnant Polygons Area Land-use correct error status inspected (km2) (%) (%) Conservation NR - R 100 16.6 98 2 Cropping NR - R 118 10.1 50 50 Forestry NR - R 10 0.4 90 10 Grazing NR - R 100 269.7 99 1 Mining NR - R 24 5.4 87.5 12.5 Intensive use NR - R 32 2.2 90.6 9.4 Conservation R - NR 100 12.8 83 17 Cropping R - NR 100 89.6 100 0 Forestry R - NR 100 12.6 90 10 Grazing R - NR 100 1648.6 98 2 Mining R - NR 100 105.5 100 0 Intensive use R - NR 34 2.7 100 0 Total 918 2176.4 77.6 22.4

128

Appendix 3-1 – Details and methods for data sources used in Chapter 3 to model the B. pertusa’s current and potential distribution.

Data sources used

Table 1. Total number of B. pertusa presence, absence and cover records collated from a variety of data sources between 1941 – 2019. Absence records were collated from where B. pertusa was not recorded during floristic surveys conducted between 2017 and 2020.

Source Presence/absence Cover Date range Detailed floristic surveys Queensland Herbarium CORVEG 774 738 1984 – 2019 Other Queensland Herbarium survey sites 411 1950 – 2019 Department of Agriculture and Fisheries Qgraze 79 1992 – 2001 Bothriochloa pertusa impact and spread surveys* 130 130 2017 – 2020 Online sources Australian Virtual Herbarium 271 1941 – 2018 Other Bothriochloa pertusa spread surveys (Vehicle 2017 – 2020 390 357 based) * Property evaluation vegetation maps** 153 1950 – 1997 Total 2208 1225 *Conducted by the author G. Lebbink **Sourced from the Queensland Department of Natural Resources, Mines and Energy

Methods for each data source

Queensland Herbarium CORVEG All vascular plants were recorded within 10 × 2 m plots. GPS point accuracy < 50 m (more accurate in recent years with improvements to GPS technology).

Other Queensland Herbarium survey sites

129

Detailed floristic survey conducted for a range of research projects. GPS point accuracy < 50 m (more accurate in recent years with improvements to GPS technology). Only B. pertusa presence records were used from these sources.

Department of Agriculture and Fisheries Q-GRAZE All vascular plants were recorded within 25, 50 × 50 cm quadrats evenly spaced along permanent 100 m transect lines. Per site, a total of 5 transect lines approximately 10 m apart were surveyed. GPS point accuracy < 50 m (more accurate in recent years with improvements to GPS technology).

Australian virtual herbarium B. pertusa specimens submitted to herbariums within Australia. GPS point accuracy < 50 - 1000 m (more accurate in recent years with improvements to GPS technology). Only B. pertusa presence records were used from these sources.

Floristic surveys 2017 – 2020 Records collated from detailed floristic surveyed conducted across Queensland to address questions related to B. pertusa spread and impact (conducted by the author Gabrielle Lebbink between 2017 and 2019). GPS point accuracy < 50 m (more accurate in recent years with improvements to GPS technology). Includes sites used in Chapters 2 and Chapter 6.

Roadside surveys Vehicle based assessments of B. pertusa presence and absence conducted by the author Gabrielle Lebbink between 2017 and 2019. This was accompanied by frequent stops to ground truth and ensure accurate identification. GPS point accuracy < 50 m (more accurate in recent years with improvements to GPS technology).

Property evaluation records Records sourced from detailed vegetation maps used for property evaluations between 1950 and 1990. Maps were drawn after a thorough examination of properties via horseback or by walking and identification of the dominant flora species. Land contours and vegetation boundaries were calculated using compass triangulation (Figure 1). Coupled with digital

130 imagery, these maps were detailed enough to allow for an accurate assessment of where B. pertusa was on the property.

Figure 1. Example of a property evaluation map from 1980. Sourced from the Queensland Department of Natural Resources, Mines and Energy.

131

Appendix 3-2 – Habitat suitability model methods

A boosted regression tree approach with ten-fold cross validation of training data was used to build two HSM’s for B. pertusa occupancy (presence/absence) and cover (Elith, Leathwick & Hastie 2008). This method creates an ensemble of regression trees based on a series of binary splits constructed from the predictor variables (Hastie, Friedman & Tibshirani 2001). The importance of predictor variables is calculating by averaging the number of times the variable is selected for splitting and the squared improvement to the model (reduction in deviance) as a result of each split (Friedman 2001).

Both models were fit to allow interactions using a tree complexity of 5 and a learning rate of 0.009 and 0.006 for the occupancy and cover models respectively. Occupancy was modelled using a binomial distribution, while cover was modelled using a Poisson distribution. Ten- fold cross validation of training data was used to determine the optimal number of trees necessary to minimize deviance and maximise predictive performance to the independent test set (Leathwick et al. 2006; Elith, Leathwick & Hastie 2008).

Model performance was assessed on prediction to the independent testing set, withheld during cross-validation. The occurrence model predicted 54 % of cross-validated deviance. While the cover model predicted 38 % of cross-validated deviance.

The fitted HSM was then used alongside the mapped grids of environmental conditions to predict the probability of occurrence and likely cover of B. pertusa within Queensland. Analyses were conducted in R (version 2.10.0, R development Core Team, 2009) and the “gbm” library supplemented with functions from Elith, Leathwick and Hastie (2008) (see Appendix 2 full modelling methods).

132

Table 2. Predictive performance of boosted regression tree models assessing the relationship between B. pertusa occupancy (presence/absence) and cover to environmental predictors. Models were fit with a tree complexity of 5 and with a learning rate of 0.009 and 0.006 for the occupancy and cover model respectively. The optimal number of trees were determined using ten-fold cross validation of training data which resulted 3950 and 1950 fitted trees for the occupancy and cover models respectively. Table values indicate for both models the total mean deviance, residual deviance, cross-validated residual deviance and the cross-validated proportion of the total deviance explained (D2).

Mean total Model Cross- D2 deviance residual validated deviance residual deviance (SE) Occupancy 1.283 0.326 0.579 (0.03) 0.54 Cover 1.312 0.269 0.809 (0.02) 0.38

References

Elith, J., Leathwick, J.R. & Hastie, T. (2008) A working guide to boosted regression trees. Journal of Animal Ecology, 77, 802-813. Friedman, J.H. (2001) Greedy Function Approximation: A Gradient Boosting Machine. The Annals of Statistics, 29, 1189-1232. Hastie, T., Friedman, J. & Tibshirani, R. (2001) The Elements of Statistical Learning Data Mining, Inference, and Prediction. Springer New York : Imprint: Springer, New York, NY. Leathwick, J.R., Elith, J., Francis, M.P., Hastie, T. & Taylor, P. (2006) Variation in demersal fish species richness in the oceans surrounding New Zealand: An analysis using boosted regression trees. Marine Ecology Progress Series, 321, 267-281. https://doi.org/10.3354/meps321267.

133

Appendix 4-1 – Methods for soil particle size analysis

Surface soil (1-5 cm) samples were collected by bulking six samples around the perimeter of each 7 × 2 m plot. Particle size distributions were determined by laser diffraction (Mastersizer 2000; Malvern Instruments Ltd.). The output of continuous particle size distribution was segmented as clay (particles <5 mm), silt (5–15 mm), fine sand (15–62.5 mm) and coarse sand (>62.5 mm), using cut-offs designed to replicate the traditional pipette method (Fred Oudyn, pers. comm.). Laser diffraction has been shown to be a robust and reproducible method, which is much more rapid than the traditional pipette method, however this method tends to overestimate silt particles and underestimate clay particles (Arriaga et al. 2006).

References

Arriaga, F.J., Lowery, B. & Mays, M.D. (2006) A fast method for determining soil particle size distribution using a laser instrument. Soil Science, 171, 663-674.

134

Appendix 5-1 – Extent of fire-sensitive ecosystems within conservation reserves and their fire frequency

Mapping methodology and analysis

Using the Queensland vegetation mapping database, we calculated the area of non-remnant fire-sensitive ecosystems (Table 1) within the National Park Estates, to identify candidate areas where fire management may be a problem due to Cenchrus ciliaris (buffel grass) invasion. We focussed on reserves within central and western Queensland where C. ciliaris has been widely sown for pasture development. The mapping database identifies 1351 regional ecosystems across the state and defines both their ‘pre-clearing’ and ‘remnant’ (remaining) distributions (Wilson, et al. 2002). The difference between these layers defines the ‘non-remnant’ distribution of the regional ecosystems. Areas were assessed by field inspections and discussions with managers to determine whether the proliferation of C. ciliaris posed a fire management problem. We assessed and report on the non-remnant area specifically as this is where C. ciliaris invasion is usually more severe and a greater threat to fire sensitive regrowth.

Using the Northern Australian and Rangeland Fire Information (NAFI)) fire scar imagery, we calculated the total burnt area and total number of fire events which occurred within fire sensitive vegetation (remnant and non- remnant) between 2003 and 2020. These burnt areas are mapped from satellite images of fire with a pixel size of 250m. Aerial validation surveys suggest this method of fire mapping is 85 – 95 % accurate (NAFI, 2016). As the total burnt area is the sum of all burnt polygons over time, it does not tell us how much was repeatedly burnt. To assess this, we merged all burnt polygons over time into a single polygon for each reserve. We then subtracted the total area burnt (sum of all burnt polygons) minus the area of the single polygon to get the total area burnt more than once.

Results

This analysis suggests that the restoration and persistence of 7553 ha of the non-remnant fire sensitive vegetation within the National Park Estate is likely threatened by C. ciliaris invasion and its associated fire risk (Table 2). Between 2003 and 2020, 38 292 ha of fire-

135 sensitive vegetation (remnant and non-remnant) burnt over 157 fire events. Of this total area 9553 ha burnt more than once over this period.

136

Table 1. Fire sensitive regional ecosystems (REs).

RE Code Description

10.3.1 Acacia argyrodendron low open woodland on alluvial plains (western) 11.3.8 Acacia argyrodendron woodland on alluvial plains 11.4.5 Acacia argyrodendron woodland on Cainozoic clay plains 6.4.1 Acacia cambagei ± Casuarina cristata low open forest on clay plains 6.3.4 Acacia cambagei ± Eucalyptus ochrophloia woodland on alluvium 6.3.6 Acacia cambagei low woodland on braided channels or alluvial plains 10.3.4a Acacia cambagei open-woodland (western) 11.3.5 Acacia cambagei woodland on alluvial plains 11.4.6 Acacia cambagei woodland on Cainozoic clay plains 6.3.25 Acacia harpophylla and/or A. cambagei low woodland to woodland on alluvial plains 11.7.1 Acacia harpophylla and/or Casuarina cristata and Eucalyptus thozetiana or E. microcarpa woodland on lower scarp slopes on Cainozoic lateritic duricrust 11.3.1 Acacia harpophylla and/or Casuarina cristata open forest on alluvial plains 11.9.5 Acacia harpophylla and/or Casuarina cristata open forest on Cainozoic fine-grained sedimentary rocks 11.4.3 Acacia harpophylla and/or Casuarina cristata shrubby open forest on Cainozoic clay plains 10.9.3b Acacia harpophylla low open-woodland to woodland on Cretaceous sediments. 11.11.14 Acacia harpophylla open forest on deformed and metamorphosed sediments and interbedded volcanics 11.11.13 Acacia harpophylla or A. argyrodendron, Terminalia oblongata low open forest on deformed and metamorphosed sediments and interbedded volcanics 11.4.9 Acacia harpophylla shrubby open forest to woodland with Terminalia oblongata on Cainozoic clay plains 11.9.11 Acacia harpophylla shrubland on Cainozoic fine-grained sedimentary rocks 6.9.3 Acacia harpophylla woodland with emergent Eucalyptus cambageana with stony soils derived from Cretaceous sediments 11.9.10 Acacia harpophylla, Eucalyptus populnea open forest on Cainozoic fine-grained sedimentary rocks 11.4.9a Acacia harpophylla, Lysiphyllum carronii ± Casuarina cristata open-forest to woodland. 11.9.1 Acacia harpophylla-Eucalyptus cambageana open forest to woodland on Cainozoic fine-grained sedimentary rocks 6.4.2 Casuarina cristata ± Acacia harpophylla open forest on clay plains 11.3.28 Casuarina cristata ± Eucalyptus coolabah open woodland on alluvial plains

137

11.4.11 Dichanthium sericeum, Astrebla spp. and patchy Acacia harpophylla, Eucalyptus coolabah on Cainozoic clay plains 10.9.3a Eucalyptus cambageana open-woodland on Cretaceous sediments. 10.3.3a Eucalyptus cambageana open-woodland with or without Acacia harpophylla understorey on alluvium. 11.4.8 Eucalyptus cambageana woodland to open forest with Acacia harpophylla or A. argyrodendron on Cainozoic clay plains 11.11.16 Eucalyptus cambageana, Acacia harpophylla woodland on old sedimentary rocks with varying degrees of metamorphism and folding. Lowlands 11.3.16 Eucalyptus largiflorens ± Acacia cambagei ± A. harpophylla woodland to low open woodland on alluvial plains 6.3.8 Eucalyptus largiflorens ± Acacia cambagei woodland on alluvium 11.4.10 Eucalyptus populnea or E. pillagaensis, Acacia harpophylla, Casuarina cristata open forest to woodland on margins of Cainozoic clay plains 11.3.17 Eucalyptus populnea woodland with Acacia harpophylla and/or Casuarina cristata on alluvial plains 6.4.3 Eucalyptus populnea, Casuarina cristata or Acacia harpophylla ± Geijera parviflora woodland on clay plains 11.11.19 Eucalyptus thozetiana, Acacia harpophylla woodland on old sedimentary rocks with varying degrees of metamorphism and folding. Lowlands and footslopes 12.12.18 Low microphyll vine forest ± Araucaria cunninghamii and semi-evergreen vine thicket on Mesozoic to Proterozoic igneous rocks. 11.9.8 Macropteranthes leichhardtii thicket on Cainozoic fine-grained sedimentary rocks. Lowlands 11.8.6 Macropteranthes leichhardtii thicket on Cainozoic igneous rocks 11.11.5 Microphyll vine forest ± Araucaria cunninghamii on old sedimentary rocks with varying degrees of metamorphism and folding 11.4.7 Open forest to woodland of Eucalyptus populnea with Acacia harpophylla and/or Casuarina cristata on Cainozoic clay plains 11.4.1 Semi-evergreen vine thicket ± Casuarina cristata on Cainozoic clay plains 11.12.4 Semi-evergreen vine thicket and microphyll vine forest on igneous rocks 11.3.11 Semi-evergreen vine thicket on alluvial plains 11.9.4 Semi-evergreen vine thicket on Cainozoic fine grained sedimentary rocks 11.8.3 Semi-evergreen vine thicket on Cainozoic igneous rocks. Steep hillsides 11.5.15 Semi-evergreen vine thicket on Cainozoic sand plains/remnant surfaces 11.9.4a Semi-evergreen vine thicket that occur on crests and mid-slopes of steep hills.

138

Table 2. Conservation reserves in sub-humid Queensland where C. ciliaris invasion is a potential problem for the protection of fire sensitive vegetation. The area of fire-sensitive vegetation (FSV) (remnant and non-remnant) protected within each reserve was estimated using the Queensland Governments pre-clearing and remnant vegetation mapping database. Within the FSV of each reserve, the total number of fire events, the total area burnt (ha), and the area burnt more than once (ha) between 2003 and June 2020 are also shown. Reserve Name Total reserve Non-remnant Remnant Total FSV burnt 2003 – 2020 FSV burnt more than once Total fire events within area (ha) FSV (ha) FSV (ha) FSV (ha) (ha) 2003 – 2020 (ha) FSV 2003 – 2020 Nairana National Park 19078 2257 9603 11859 4357 67 9 Taunton National Park (Scientific) 11381 1730 3636 5366 336 0 3 Carnarvon National Park 294265 987 35241 36229 19208 6204 49 Idalia National Park 137886 845 8629 9474 0 0 0 Homevale National Park 21163 546 871 1417 1603 334 17 Snake Range National Park 2615 208 63 271 0 0 0 Cudmore (Limited Depth) National Park 20811 172 44 216 221 81 4 Albinia National Park 7179 132 119 251 217 54 5 Nuga Nuga National Park 2869 127 383 510 198 0 2 Expedition (Limited Depth) National Park 18957 126 4352 4477 3184 1192 22 Dipperu National Park (Scientific) 11021 94 7732 7825 8 0 3 Goodedulla National Park 24971 90 3779 3870 1870 95 12 Mount Etna Caves National Park 567 70 493 562 0 0 0 Junee National Park 5310 52 1201 1253 982 0 2 Palmgrove National Park (Scientific) 24495 42 3324 3367 2843 1049 12 Mazeppa National Park 4054 39 3686 3725 557 0 4 Narrien Range National Park 7312 27 832 859 761 1 4 Epping Forest National Park (Scientific) 2646 8 1783 1791 299 0 3 Moorrinya National Park 32111 2 1811 1813 1649 473 6

Total 648690 7553 87582 95135 38292 9553 157

139

References Northern Australian and Rangeland Fire Information. (2016) The NAFI Booklet, Darwin Centre for Bushfire Research, Charles Darwin University, Australia. https://firenorth.org.au/nafi3/ (Accessed 24th of January 2020)

140