PLEASE NOTE: TWO OF THE FIGURES HAVE BEEN

REMOVED FROM THE DIGITAL VERSION OF THIS THESIS

DUE TO COPYRIGHT.

RESTORATION ECOLOGY IN THE SEMI-ARID

WOODLANDS OF NORTH-WEST VICTORIA

Fiona Alys Murdoch

B.Sc. (Melb), B.Sc.(Hons.) (Syd)

This thesis is submitted in total fulfilment of the requirements of the degree of Doctor of Philosophy.

School of Science and Engineering

University of Ballarat PO Box 663 University Drive, Mt Helen Ballarat, Victoria 3353 Australia

Submitted in December, 2005

-ii-

Abstract Arid areas are often overgrazed and dysfunctional with poor recruitment of desirable species, diminished control over resources and altered soil properties. Restoration ecology re-establishes these valued processes.

State-and-transition models summarise knowledge of vegetation dynamics and tools for restoration, and encourage the incorporation of new information. The model developed here for semi-arid woodlands of north-west Victoria highlighted the unknown cause of observed, natural recruitment and the need for a technique, other than direct seeding and handplanting, for enhancing the recruitment of desirable species. I pursued these knowledge gaps for two dominant, woodland trees: Allocasuarina luehmannii and Casuarina pauper.

Natural recruitment of juvenile C. pauper was found to be limited and primarily from root suckers. Extensive recruitment of A. luehmannii was shown to be mostly seedlings established following substantial reductions in grazing pressure since 1996. Seedlings were associated with areas devoid of ground flora near a female tree. The importance of competition between seedlings and ground flora, spatial variation in soil moisture and individual variation in the quantity of seed produced deserves further investigation to enhance future restoration success.

Root suckers of both C. pauper and A. luehmannii can be artificially initiated, albeit in low numbers and this was found to be a feasible, new tool for restoration. Suckers are preceded by the growth of callus tissue on exposed or damaged, living, shallow roots. Both male and female trees can produce suckers and spring treatments may be more successful.

Genetic fingerprinting of mature A. luehmannii and C. pauper trees in six populations did not identify any clonal individuals indicating that recruitment in the past has been from seedlings. Despite this, the high level of gene flow suggests that the impact of introducing small numbers of root suckers into existing populations is unlikely to impact negatively on the population genetics of these species.

- ii -

Executive Summary

Vegetation dynamics in arid areas are complex and unpredictable. Rainfall is a key driver of positive vegetation change but is infrequent and erratic. Overgrazing has strong negative impacts on landscapes but often, simply reducing grazing pressure does not lead to recovery of the vegetation: active management of the composition of the vegetation community and nutrient and water cycles may also be required. Restoration ecology offers a variety of tools to enable managers to enhance the condition of the vegetation, based around five primary components: reducing grazing pressure, reintroduction of component species, reducing competition from inappropriate species, improving soil properties and increasing control over resources such as water and nutrients.

The semi-arid woodlands of north-west Victoria, like many arid landscapes worldwide, are degraded and dysfunctional. There is an increasing interest in options for restoration of these landscapes, particularly in enhancing the recruitment of desirable species. In north-west Victorian national parks, dune reclamation, management of grazing pressure and reintroduction of component species through hand-planting and direct seeding are all conducted with varying degrees of success. This thesis explores and develops theories, models and tools of restoration ecology in arid areas, in particular semi-arid woodlands and the recruitment of two dominant woodland species: Casuarina pauper and Allocasuarina luehmannii.

State-and-transition (S-T) models provide a way to summarise knowledge of vegetation dynamics, tools for restoration and the impact of restoration activities. A theoretical S-T framework was developed which shows that vegetation passes through three, increasingly degraded states: loss of component species, dominance of inappropriate species and finally, loss of soil functioning. Restoration must address and reverse these transitions in a step-wise manner. The S-T framework was used to organise the history of degradation and restoration in the semi- arid woodlands of Hattah Kulkyne National Park (HKNP), north-west Victoria. This process highlighted two main opportunities to enhance restoration success.

One opportunity was exploring where, when and why natural recruitment of key species was occurring. Some recruitment of C. pauper root suckers was found in Murray Sunset National Park and extensive recruitment of A. luehmannii seedlings was detected in HKNP. The seedlings are believed to have established post-1996 following extensive reductions in grazing pressure and did not appear related to a particular rainfall event. This is contrary to the literature for this species and supports suggestions from others that continuous recruitment of arid plant species is necessary and can occur in the absence of high grazing pressure. The seedlings were associated positively with bare areas within 30 m of a female tree, implying that lack of competition with

- iii -

ground flora may be significant for their establishment. Although further research is required, better control of annual weeds or replacement with perennial grasses may be necessary to enable the establishment of seedlings following restoration activities. Seedlings were also correlated with topography that increased soil moisture and females trees which produced large quantities of seed. In contrast, variation in seed quality between trees, as determined by germination trials, was not correlated with the occurrence of seedlings although the trials were probably confounded by differing maturity of seed at the time of collection. Dark-colored, mature seed had a higher percentage germination and more rapid germination than light-colored seed. This sends a clear message that care should be taken to only collect mature seed for restoration activities.

The second opportunity was developing an alternative tool to enhance the regeneration of desirable species. Direct-seeding and hand-planting experiences a high failure rate due to browsing and low soil moisture. Initiating root suckers of C. pauper and A. luehmannii was investigated because both species are known to produce root suckers and there is evidence from other species that root suckers are better able to establish under these conditions than are seedlings. Although not previously considered in restoration ecology texts, root suckers were found to be a viable tool for restoration. They could be produced in low numbers by damaging and / or exposing shallow roots of male or female trees. There was some indication that spring was likely to be the best season for treatment.

No identical genotypes were detected in the genetic fingerprint (Amplified Fragment Length Polymorphisms) obtained for three mature woodlands each of A. luehmannii (82 individuals) and C. pauper (65 individuals). This suggests that historically, seedlings were the predominant mode of recruitment in these species and the presence of numerous juvenile root suckers was attributed to a higher frequency of root disturbance now than historically. However, I believe there would be minimal potential impact of reduced genetic diversity following introduction of root suckers into existing populations. Most genetic diversity was held within the A. luehmannii and C. pauper populations (94.4% and 87.7% respectively) indicating a high level of gene flow. Even small amounts of natural seedling recruitment would be sufficient to restore patterns of genetic variation. Stimulating root suckers also retains genetic diversity because it allows genotypes to persist in the population for longer, thus increasing the probability that they are exposed to conditions conducive to the establishment of seedlings. Nevertheless, a conservative approach is required, aiming for little more than a one to one replacement of existing trees rather than promoting extensive clonal stands. The technique is not viewed as a final solution for the restoration of semi-arid woodlands but rather an emergency measure for degraded woodlands with little natural recruitment. Future research needs in this area are developing a practical tool for broad-scale, mechanised initiation of root suckers in the field.

- iv -

Additional areas requiring further research, as identified by the S-T model, include understanding and developing techniques to enable regenerating or partially restored woodlands to move to the desirable state of a self-perpetuating, restored woodland. There is also an opportunity to improve the success of direct seeding, and a need to monitor and manage the threat posed by emerging weeds. New information can easily be incorporated into the S-T model developed earlier, demonstrating the utility of the conceptual model.

- v -

Statement of Authorship

Except where explicit reference is made in the text of the thesis, this thesis contains no material published elsewhere or extracted in whole or in part from a thesis by which I have qualified for or been awarded another degree or diploma.

No other person’s work has been relied upon or used without due acknowledgement in the main text and bibliography of the thesis.

Signed: ______Date: ______

Fiona A. Murdoch

- vi -

Acknowledgements

Many people and organisations made invaluable contributions to this research. This research would not have been possible without an Australian Research Council Linkage grant with the financial support of Parks Victoria (Australian Postgraduate Award Industry - LP0235347).

A number of people at the University of Ballarat provided moral support and advice on my infrequent visits and via email. I would particularly like to thank my supervisors, Prof. Martin Westbrooke and Dr. Florentine Singarayer for providing the impetus for this research, support throughout and trusting me to carry out this research as an off-campus student. I would also like to thank Dr. Kate Callister for sharing her knowledge and experience of the semi-arid woodlands and Marion O’Keefe for her expert tutorage in all aspects of MapInfo. The statistical advice received from Dr. Cameron Hurst for Chapter 5 is greatly appreciated. Wendy Cloke was always willing to assist with the purchase of equipment and related areas.

Assistance during field trips and advice was gratefully received from Parks Victoria Mallee staff. The endless digging in those early days was possible only with the assistance of Wayne Berry (Jex), Peter Teasdale, Lorraine Ludewigs, Bruce Summerfield and Travis Horsefall. Funny how no-one came back for a second day … The advice and logistic support from Russell Manning, David Christian, Jack Kelly, Pete Sandell, Rob McGlashan, Shaun Stephens, Shane Southon and Danielle Southon was much appreciated. Fantastic accommodation at the wonderful Shearers’ Quarters was facilitated by the Parks Victoria, Werrimull staff.

Staff at the Department of Primary Industries, Irymple were always willing to share their laboratory facilities. In particular Sue McConnell, who always assured me that it could be done and Dr. Mark Downey who saved the day with his experience in extracting DNA. Also Cathy Taylor, Dr. Mark Krstic, Narelle Nancarrow, Glenda Kelly, Linda Pollock, Lynn McMahon and Adam Wightwick assisted in various ways.

A number of staff at the University of Melbourne provided generous assistance when I came to them as an ex-student. I am indebted to Dr. Ed Newbigin of the School of Botany. I would not have been able to commence the genetics component of this research without his initial support. Dr. Joe Barrins provided cheerful advice and his expertise in running the AFLPs under contract allowed me to complete the genetics research. Dr. Peter Ades (School of Forest and Ecosystem Science) and Dr. Belinda Appleton (School of Genetics) helped me to unravel some of the mysteries of genetics. Assoc. Prof. David Morgan and Dr. Graeme Coulson (School of Zoology) both shared their experience of mallee ecology.

- vii -

Dr. Dave Spratt and colleagues at CSIRO Sustainable Ecosystems (Canberra) provided much needed advice, particularly in those early days. Their wise council and the knowledge I gained during my Honours research in the “Burrendong team” assisted me greatly.

Family and friends played an important role, in particular my mum and dad, Liz and Tony Cavanagh. Tom McNair, Ann Wragg and Sam Murdoch each took a turn at digging. My big sister, Dr. Jo Cavanagh provided much needed support and editing. Ken and Mary Murdoch offered a halfway house for all those trips to Ballarat.

A late arrival on the scene is our daughter Emily Murdoch. Emily slept soundly through the first months of her life which assisted me greatly whilst I addressed comments from examiners. I hope she will one day read this thesis.

Finally, and most importantly, my wonderful husband Phil who followed up a sterling performance during my Honours research with another amazing effort to get me through this PhD. I gratefully acknowledge his assistance every step of the way, from project design, to developing concepts, through to editing the final write-up. I would not have completed this research without his assistance in the field with digging, shooting, motivation, design of bits and pieces and overall being so practical. I was incredibly fortunate to have his unflagging support from start to finish.

- viii -

Table of Contents Abstract ...... ii Executive Summary ...... iii Statement of Authorship...... vi Acknowledgements ...... vii Table of Contents...... ix Chapter 1 Introduction to restoration ecology in semi-arid areas 1.1 Restoration ecology...... 1 1.1.1 What is restoration ecology? ...... 1 1.1.2 Why is restoration ecology needed in semi-arid areas?...... 1 1.2 Key drivers in arid and semi-arid ecosystems ...... 3 1.2.1 Overgrazing and landscape dysfunction...... 3 1.2.2 Rainfall, recruitment and overgrazing...... 4 1.3 Restoration techniques in semi-arid areas ...... 6 1.3.1 Root suckers...... 8 1.4 Outline of research ...... 9 1.4.1 Research needs in north-west Victoria...... 9 1.4.2 Focus of research...... 9 1.4.3 Objectives of research...... 10 Chapter 2 Study area and species 2.1 Study area ...... 12 2.1.1 Location ...... 12 2.1.2 Geomorphology and soils...... 12 2.1.3 Vegetation ...... 13 2.2 Climate ...... 17 2.2.1 North-west Victoria ...... 17 2.2.2 Rainfall events ...... 18 2.3 Vegetation change since European settlement ...... 25 2.3.1 The years 1800-1870...... 25 2.3.2 The years 1870-1920...... 26 2.3.3 The years 1920-1950...... 26 2.3.4 From the 1950s onwards...... 27 2.4 Establishing the national parks...... 28 2.4.1 Hattah Kulkyne National Park...... 28 2.4.2 Murray Sunset National Park...... 29 2.5 Management of national parks ...... 31 2.5.1 Management objectives...... 31

- ix -

2.5.2 Grazing pressure ...... 31 2.6 Study species - Casuarinaceae...... 36 2.6.1 Taxonomy...... 36 2.6.2 Description...... 36 2.6.3 Recruitment ...... 38 2.7 Data collection ...... 39 Chapter 3 Development of a state-and-transition model for semi-arid woodlands of north-west Victoria using qualitative data 3.1 Scope and objectives ...... 41 3.2 Development of models to describe semi-arid vegetation...... 42 3.2.1 Range succession models...... 42 3.2.2 State-and-transition models...... 42 3.3 Developing the framework of a state-and-transition model ...... 45 3.3.1 The framework...... 45 3.3.2 From framework to model...... 50 3.3.3 Defining vegetation communities...... 51 3.4 Developing a model for semi-arid, north-west Victoria ...... 53 3.4.1 Overview...... 53 3.4.2 Transitions ...... 55 3.4.3 Reverse transitions...... 55 3.4.4 Desirable states...... 57 3.4.5 Missing component species...... 60 3.4.6 Dominance of inappropriate species ...... 69 3.4.7 Alteration of abiotic (soil) factors...... 73 3.5 Using the model to better manage semi-arid woodlands...... 74 Chapter 4 Describing the occurrence of natural recruitment 4.1 Scope and objectives ...... 75 4.2 Introduction...... 76 4.2.1 Background...... 76 4.2.2 Where and when is seedling recruitment occurring?...... 76 4.2.3 Why has seedling recruitment occurred? ...... 77 4.3 Methods...... 79 4.3.1 Where is recruitment occurring?...... 79 4.3.2 When has seedling recruitment occurred? ...... 82 4.3.3 Why is seedling recruitment occurring?...... 83 4.3.4 Tree-based modelling of key factors influencing recruitment ...... 87 4.4 Results...... 88 4.4.1 Where is recruitment occurring?...... 88

- x -

4.4.2 When did recruitment occur?...... 95 4.4.3 Why is seedling recruitment occurring?...... 97 4.5 Discussion ...... 101 4.5.1 Recruitment of Casuarina pauper...... 101 4.5.2 Recruitment of Allocasuarina luehmannii ...... 101 4.5.3 Seed collection for revegetation ...... 106 Chapter 5 The use of root suckers to enhance recruitment 5.1 Scope and objectives ...... 107 5.2 Introduction...... 108 5.2.1 What is clonal growth?...... 108 5.2.2 Advantages and disadvantages of clonal growth ...... 108 5.2.3 Root suckers...... 110 5.3 Methods...... 116 5.3.1 Field characteristics of suckers...... 116 5.3.2 Experiment 1: Initiating callus and suckers...... 116 5.3.3 Experiment 2: Casuarina pauper trial ...... 118 5.3.4 Analysis of Experiments 1 and 2 ...... 119 5.4 Results...... 121 5.4.1 Field characteristics of suckers...... 121 5.4.2 Observations from Experiments 1 and 2 ...... 125 5.4.3 Experiment 1: Casuarina pauper trial ...... 126 5.4.4 Experiment 2: Initiating callus and suckers – Casuarina pauper ...... 131 5.4.5 Experiment 2: Initiating callus and suckers – Allocasuarina luehmannii.... 137 5.4.6 Statistical analyses ...... 145 5.4.7 Effects of exposure...... 149 5.5 Discussion ...... 151 5.5.1 The growth and development of root suckers...... 151 5.5.2 What influences the initiation of root suckers? ...... 154 5.5.3 Are root suckers a viable tool for restoration? ...... 157 Chapter 6 Conservation genetics 6.1 Scope and objectives ...... 161 6.2 Introduction to conservation genetics ...... 162 6.2.1 Genetic diversity and population structure...... 162 6.2.2 Amplified Fragment Length Polymorphisms...... 164 6.2.3 The mating system of Casuarina pauper and Allocasuarina luehmannii... 166 6.3 Methods...... 167 6.3.1 Collection of plants ...... 167

- xi -

6.3.2 Extraction of DNA ...... 168 6.3.3 AFLP analyses...... 169 6.3.4 Data analysis ...... 170 6.4 Results...... 171 6.4.1 Clonal genotypes...... 171 6.4.2 Genetic variation...... 172 6.5 Discussion ...... 173 6.5.1 Absence of clonal genotypes...... 173 6.5.2 Genetic structuring of the population...... 175 6.5.3 Impact of introducing clonal genotypes for restoration ...... 177 Chapter 7 Summary of results 7.1 Background information sources ...... 179 7.1.1 Restoration ecology in arid areas ...... 179 7.1.2 The semi-arid woodlands of north-west Victoria...... 181 7.2 Organisation of knowledge: State-and-transition modelling ...... 181 7.3 Research ...... 182 7.3.1 Lessons from natural recruitment ...... 182 7.3.2 Root suckers as a tool for restoration...... 185 7.4 Future research ...... 188 References ...... 189 Appendix 1: Australian trees and shrubs from the rangelands known to produce root suckers ...... 216

- xii - Chapter 1 Introduction

Chapter 1 Introduction to restoration ecology in semi-arid areas

1.1 Restoration ecology 1.1.1 What is restoration ecology?

Restoration ecology provides the concepts, models, methodologies and tools for practitioners to carry out ecological restoration in any landscape (SER 2002). Early definitions of ecological restoration described it as the process of assisting the recovery of an ecosystem that has been degraded, damaged or destroyed (SER 2002). More recently social, cultural, economic and ethical values have been incorporated to produce a more useful and flexible definition.

Ecological restoration is the process of restoring one or more valued processes or attributes of a landscape (Davis and Slobodkin 2004)

Depending on the purpose, valued processes could be defined as productive grasslands for grazing, a self-perpetuating species guild for aesthetic purposes, suitable habitat for a target species of fauna or restoring an ecosystem to near-historical conditions. Incorporating agricultural land is a recent development intended to manage landscape-scale problems such as restoring hydrological regimes (Hobbs and Norton 1996). Restoration has received greater emphasis as landscapes are acknowledged as increasingly degraded (Holl et al. 2003). Restoration is particularly important to enhance conservation values in protected landscapes such as national parks.

Effective restoration requires a goal or policy that is clear, articulated and accepted by researchers, practitioners and end-users. Whilst “the goal” is often defined in terms of restoring an ecosystem back to its original state, in most situations this is impossible because of the dynamic nature of ecosystems, the irreversibility of some changes and insufficient knowledge of past ecosystems to enable re-creation (Aronson et al. 1993; Hobbs and Harris 2001; Bestelmeyer et al. 2003; Choi 2004). Rather than unrealistic preconceptions focusing on the past, it is best to focus on ecosystems that are believed to meet desires and needs in the future (Oliver et al. 2002; Davis and Slobodkin 2004). Success is reintroducing “valued processes” with sufficient biotic and abiotic resources for development to continue without further assistance (SER 2002). However, success is both difficult to measure and difficult to achieve (Allison 2002; Anand and Desrochers 2004; Martin et al. 2005).

1.1.2 Why is restoration ecology needed in semi-arid areas?

Arid and semi-arid areas cover 30% of the global land surface and host around one quarter of the earth’s population (IUCN 1999). Much of this area has been substantially altered and as a result a significant proportion of arid landscapes has been degraded, desertified or fragmented (Vallejo et

- 1 - Chapter 1 Introduction al. 2006). This has resulted in a change in ecosystem processes, loss of biodiversity and widespread poverty. Around 60% of arid or semi-arid ecosystems are considered degraded and this is increasing at a rate of six million hectares annually (IUCN 1999).

In Australia, the “rangelands”, cover approximately 70% of the continent comprising mostly the arid and semi-arid areas (Figure 1.1). Rangelands is used in this thesis interchangeably with arid and semi-arid vegetation, worldwide. A defining feature of the rangelands is low rainfall, which is extremely variable in time and space and precludes regular cropping and allows only stock grazing of unimproved pasture (Stafford-Smith and Morton 1990). High evaporation rates that generally exceed annual rainfall are also common (Stanley 1983). Soils of Australia’s rangelands are highly leached and infertile (Stanley 1983). Low-productivity, slow growing perennials dominate because they are highly efficient in their use of nutrients (Aerts and van der Peijl 1993) and have the ability to “wait out” the poor seasons (Wiegand et al. 2004).

Figure 1.1 Map of Australia’s moisture zones. The white zone is arid, pale grey represents semi-arid, dark-grey is dry-subhumid and black is moist (from Smyth and James 2004). Land-use of the Australian rangelands is primarily grazing of sheep in the south and cattle in the north (Smyth and James 2004). Past and present management practices in the rangelands have resulted in accelerated soil erosion, invasion of exotic plants and animals, reduced water quality, soil salinity, sodicity and decreased fertility and loss or alteration of native plant communities, particularly loss of palatable species (ANZECC and ARMCANZ 1999). In the Australian rangelands 13% of the area is classed as degraded, 25% as significantly disturbed, 16% as being affected by erosion, and 17% as being invaded by woody weeds (Smyth and James 2004). Recommendation 2.2 of the National Principles and Guidelines for Rangeland Management (ANZECC and ARMCANZ 1999) is to “develop mechanisms for the restoration and future management of degraded lands.”

- 2 - Chapter 1 Introduction

1.2 Key drivers in arid and semi-arid ecosystems

In the rangelands, vegetation dynamics are often unpredictable, nonlinear, and strongly influenced by environmental drivers, such as climate. Rainfall in particular fluctuates widely from year to year affecting productivity, species composition and structure (Westoby et al. 1989; Bestelmeyer et al. 2003). The trigger-transfer-reserve-pulse framework (Figure 1.2) describes the key drivers of ecosystem processes in rangelands (Noy-Meir 1973; Ludwig et al. 1997). The “trigger” is typically rainfall, but may be nutrients in dust or leaf litter. The trigger is spatially redistributed (“transfer”) by runoff / runon or wind. Landscape patches capture these resources and store them as “reserves”. If the reserves exceed a threshold requirement then a “pulse” of growth occurs. Some of the pulse feeds the reserves (e.g. seeds) whilst some materials feed back into the landscape patches (e.g. new grass tussocks) and contribute to the functioning of spatial redistribution during the transfer. Some of the pulse is lost from the system by off-take such as runoff out of the system or grazers consuming the pulse and then moving on.

TRIGGER rainfall

Losses Outflow- TRANSFER from runoff run-off/runon system Feedback to landscape RESERVE Off-take- structures- fire, Ploughback- soil water patch density, grazing seeds width, height Recycling- nutrients PULSE growth

Figure 1.2 The trigger-transfer-reserve-pulse framework for arid and semi-arid landscapes (from Ludwig et al. 1997). 1.2.1 Overgrazing and landscape dysfunction

Changes in land use, particularly overgrazing of rangeland landscapes, can lead to a breakdown of the trigger-transfer-pulse framework, landscape dysfunction, collapse of ecosystem function and desertification (Ludwig et al. 1997). No single grazer is solely responsible; it is the total grazing pressure from domestic, feral and native species (Freudenberger et al. 1997).

The direct effects of overgrazing are the death of plants particularly grasses (Grice and Barchia 1992), the elimination of seedling recruits (Milton and Wiegand 2001), and preferential browsing

- 3 - Chapter 1 Introduction of palatable species resulting in altered species composition (Illius and O’Connor 1999; Friedel et al. 2003; Landsberg and Crowley 2004). Whilst high levels of grazing pressure are necessary for the collapse of vegetation biomass, much lower levels will suppress regrowth and maintain the vegetation in a degraded state (Noy-Meir 1975).

The indirect effect of overgrazing is a dysfunctional landscape with diminished control over limiting resources such as water and nutrients. In comparison to functional landscapes, overgrazed landscapes have fewer, smaller patches of vegetation and cryptogams and a long fetch length (bare ground) between each patch. Within the fetch, the soil is compacted and crusted which reduces infiltration and increases runoff (Tongway and Ludwig 1994; Greene et al. 1998). As a result, much of the “trigger” (water and nutrients) is lost rather than being captured within vegetation patches (Ludwig and Tongway 1995; Freudenberger et al. 1997; Sparrow et al. 2003; Tongway et al. 2003). The loss of vegetation also increases evaporation and removes the protective microclimate for seedlings (Tongway and Ludwig 1990; Eldridge and Robson 1997; Yates et al. 2000). Overall, overgrazing results in a lower soil productive potential: smaller, shorter and less frequent production pulses following rainfall (Bastin et al. 1993; Tongway and Ludwig 1997; Pickup et al. 1998; Holm et al. 2003) and reduced establishment of shrub and tree recruits (Anderson and Hodgkinson 1997).

1.2.2 Rainfall, recruitment and overgrazing

Water has a dominant role in arid and semi-arid environments and is the trigger for most events including growth, recruitment and regeneration (Noy-Meir 1973; Chesson et al. 2004; Ogle and Reynolds 2004). The latter terms are similar in meaning; recruitment being the establishment of new individuals, especially seedlings; regeneration the overall incorporation of new individuals into the population (Harper 1977). The term recruitment is used throughout.

The success in arid areas of natural recruitment or restoration attempts depends on rainfall events over a single storm, weeks, or seasons (Schwinning et al. 2003; Reynolds et al. 2004; Schwinning et al. 2005). “Biologically important” rainfall events are those which stimulate plant growth and recruitment (Noy-Meir 1973) and depend on plant functional type (microbial community, grass, tree, shrub etc), season, latitude and plant phenology and age (Golluscio et al. 1998; Schwinning et al. 2003; Chesson et al. 2004; Ogle and Reynolds 2004; Reynolds et al. 2004; Schwinning and Sala 2004). Small events of <2 mm may stimulate soil microbes and decomposition (Austin et al. 2004), 5 mm can stimulate the growth of grasses (Sala and Lauenroth 1982), whilst at least 10 mm may be required in summer to stimulate the growth of grasses and shrubs (Schwinning et al. 2003). The threshold event required to stimulate recruitment is generally higher than that required for growth and at least 15-25 mm may be required for the germination of many desert plants (Beatley 1974; Bowers 1996).

- 4 - Chapter 1 Introduction

Many perennial species from arid areas are believed to exhibit episodic recruitment following high rainfall events (section 2.2.2). This observation has arisen from both anecdotes and the analysis of the age structure of stands of plants estimated from the size of a plant. Usually the diameter of the stem is used to suggest age; individuals with a larger stem diameter are taken as being older. The various size classes are expressed as a frequency histogram (Harper 1977). The histogram for perennial plants from arid environments generally shows clusters of similar sized plants believed to reflect recruitment events (e.g. Crisp and Lange 1976; Walker et al. 1986; Boucher 1997; Wiegand et al. 2000) linked to high rainfall (Hall et al. 1964; Lange and Purdie 1976; Crisp 1978; Chesterfield and Parsons 1985; Cooke 1987; Woodell 1990; Read 1995; Wiegand et al. 2000) or a flooding event from remote rainfall (Florentine and Westbrooke 2005; Westbrooke et al. 2005).

There are a number of limitations of the size-age relationship (West et al. 1979). There is substantial variation in the growth rate of individual plants, hence size at a given age varies greatly. An age distribution dominated by older trees may arise by a variety of mechanisms, e.g. continuous recruitment but high mortality in the younger classes, or continuous recruitment with seedling growth suspended until significant growth occurs in response to an environmental factor such as release from grazing or favourable rainfall (Watson et al. 1997b). Indeed, when detailed studies have been conducted for some species, recruitment has been found to be more continuous than previously believed (Crisp and Lange 1976; Ireland 1997; Watson et al. 1997a). Simulation models of population dynamics have suggested that both episodic and continuous recruitment are necessary for rangeland species to persist (Wiegand et al. 2000; Wiegand et al. 2004). The apparent reliance on a high rainfall event may also be an artefact of overgrazing where grazing eliminates recruits in all years except those where recruitment is extremely high, such as following favourable rainfall (Milton and Wiegand 2001).

One indicator of poor condition in rangelands is the absence of recruitment of desirable species (Friedel et al. 2003). Since European settlement, little recruitment has been reported for many shrubs and trees in arid areas of Australia (Hall et al. 1964; Chesterfield and Parsons 1985; Auld 1990, 1993; Cheal 1993; Auld 1995c) and overseas (Obeid and Seif El Din 1970; Wiegand et al. 1999; Bowers and Turner 2002; Wiegand et al. 2004). The recruitment of juvenile plants into the adult population is a critical determinate of long-term population size and persistence (Woodell 1990; Auld 1993; Wiegand et al. 2004). In Australia, increased grazing pressure since European settlement is believed to have reduced recruitment to levels where the long-term survival of some species is threatened (Hall et al. 1964; Lange and Purdie 1976; Crisp 1978; Chesterfield and Parsons 1985). In north-west Victoria, the Victorian Land Conservation Council found “no evidence that the woodland species can regenerate under continuous grazing” (LCC 1989).

- 5 - Chapter 1 Introduction

The absence of a large recruitment event may be attributable to lack of suitable rainfall, however even in the high rainfall years of 1973-1975 there was no recruitment, apparently due to high grazing pressure (Chesterfield and Parsons 1985; Westbrooke et al. 1988). There is increasing evidence that some germination occurs in most years, however grazing eliminates the seedlings and prevents recruitment (Wellington and Noble 1985; Gardiner 1986; Milton and Wiegand 2001). The absence of these small recruitment events is a cause for concern because it increases the probability of extinction of some plant species (Wiegand et al. 2004). Only in high rainfall years are sufficient seedlings produced and sufficient alternative food available to allow some seedlings to survive (Milton and Wiegand 2001). Furthermore, the effects of overgrazing on the ground flora (increased annuals at the expense of perennials: Friedel et al. 2003) are only just becoming clear. Summer-growing annuals deplete soil moisture more rapidly than perennial grasses and have been shown to inhibit establishment of seedling, woody, perennial species. This effect can be overcome by high rainfall events (Davis et al. 1998; Gordan and Rice 2000).

1.3 Restoration techniques in semi-arid areas

The key feature of restoration in arid areas is encouraging the recruitment of desirable plant species. Restoration ecology offers a variety of tools to enable managers to achieve this, including five primary components: reducing grazing pressure, reintroduction of component species, reducing competition from inappropriate species, restoring soil function and increasing availability of soil water (Whisenant 1999; Vallejo et al. 2006).

Passive restoration involves removing the stress, e.g. grazing, and allowing natural succession to occur (Vallejo et al. 2006). However, the timescale over which recovery processes occur following reduction of total grazing pressure is measured in decades or centuries. In some cases, without intensive inputs, degraded conditions may persist despite removal of agents of degradation (Friedel 1991; Milton et al. 1994; Lovich and Bainbridge 1999).

Active management techniques include introducing missing component species. This can be achieved via direct seeding local seed mixes onto the site but although this technique is cost- effective when successful, it often fails in dry areas (Sluiter et al. 1997a; Schirmer and Field 2000; Thomas 2003). Hand-planting of seedlings can achieve greater success but is more costly. Successful establishment can be enhanced by using advanced tubestock in elongated pots which allow root development of up to one metre (Whisenant 1999; Anonymous undated) or by preconditioning the seedlings to drought (Vilagrosa et al. 2003; Vallejo et al. 2006). If possible, recruits should be protected from browsing with guards or fences (Schirmer and Field 2000) or through the use of a nurse plant (Maestre et al. 2001; Castro et al. 2002; Gomez-Aparicio et al. 2004). Facilitating natural dispersal of seeds into new areas can be achieved by encouraging agents of dispersal such as birds (Vallejo et al. 2006).

- 6 - Chapter 1 Introduction

Competition from weeds can be managed by via chemical treatment with herbicides pre- establishment and sometimes post establishment. Mechanical site preparation breaks up and buries weeds and can be successful although is more useful in an agricultural setting where treatment can be repeated, otherwise it can enhance establishment of weeds (Schirmer and Field 2000). Fires can also be used to control recruitment of undesirable woody species (Hodgkinson and Harrington 1985; Westoby et al. 1989).

Natural or artificial recruitment will not be observed following release from grazing pressure if the soil habitat is no longer suitable for the establishment of seedlings (Tongway and Ludwig 1997; Hobbs and Harris 2001; Tongway et al. 2003). Whether in fact a degraded landscape will recover depends on the degree of landscape dysfunction. In severe cases restoration must focus on restoring “transfer” processes: hydrology, energy capture and nutrient cycling (Whisenant 1999). The restoration of soil function includes dune reshaping, soil pitting or scalloping or erosion bunds to increase infiltration and / or interrupt the flow of wind or water across a surface or deep ripping to alleviate compaction of the soil (Whisenant 1999). Piles of cut brush improve patch structure, water infiltration, seed capture and soil nutrition (Ludwig and Tongway 1996). Erosion matting or a cover crop can also halt erosion. Australian soils are generally deficient in nitrogen and phosphorus and most is contained in the top ten cm; removal of topsoil by erosion results in a nutrient poor or saline subsoil (Attiwill and Leeper 1987; van den Berg and Kellner 2005). Degraded sites may require the addition of specific nutrients (Suding et al. 2004; Vallejo et al. 2006) or inoculation of seeds with symbiotic bacteria to improve establishment success (Thrall et al. 2005).

Salt-affected land is another symptom of altered soil function. Dryland salinity results from seepage upwards of saline groundwater; scalds occur where loss of vegetation and erosion removes the topsoil and exposed the saline or sodic subsoil beneath (Marcar and Crawford 2004). Some form of mechanical preparation (cultivation, mounding) of scalded or saline sites can increase surface roughness forming niches of lower salinity and safe sites for seedling establishment (Williams and Semple 2001; van den Berg and Kellner 2005). Waterponding can reduce salinity over a broader area to achieve complete cover of plants (Ringrose et al. 1989).

In arid areas, sufficient soil moisture is critical for seed germination, establishment and subsequent plant growth (Noy-Meir 1973; Whisenant 1999). Recruitment in average rainfall years may occur in areas where soil moisture is locally abundant, such as roadsides, drainage lines or depressions that hold water (Chesterfield and Parsons 1985; Ludwig 1987; Tiver and Andrew 1997). The addition of mulch can shade the soil, reducing evaporation (Vallejo et al. 2006). Supplementary watering is effective during the growth phase although it is expensive and often impractical. Water-conserving methods such as drip-irrigation or deep-pipe irrigation may be useful (Lovich and Bainbridge 1999). Water-harvesting is an under-utilised alternative,

- 7 - Chapter 1 Introduction particularly in Australia. Mechanical site preparation is used to shape the ground to capture and store rainfall and runoff (Whisenant 1999; Edwards et al. 2000). Water harvesting through microcatchments is used primarily in developing countries. Microcatchment designs include contour ridges for establishing crops, contour bunds and negarim (diamond-shaped) microcatchments for establishing trees, and semi-circular bunds for range and fodder (Critchley and Siegert 1991). Soil requirements may limit the application of this technique. Ideally the soil in the catchment area should have a high runoff coefficient while the soil in the cultivated area should be a deep, fertile loam (Critchley and Siegert 1991).

1.3.1 Root suckers

Developing unconventional restoration methods can achieve results when conventional techniques fail (Scheffer and Carpenter 2003). For example, using knowledge of interannual variability of rainfall as predicted by El Nino Southern Oscillation (ENSO) events can enhance the success of revegetation (Holmgren and Scheffer 2001). Similarly the use of root suckers may accelerate the recruitment of key plant species. Artificially stimulating root suckers is a special case of restoring component species that has not been addressed previously in restoration ecology texts. Root suckers, along with bulbs, rhizomes and stolons are a form of clonal growth (Klimes et al. 1997). They are more common amongst clonal trees and shrubs than herbs (Jenik 1994) but are uncommon in gymnosperms (Burrows 1990; Peterson and Jones 1997; Del Tredici 2001). Root suckers grow from buds within the roots of an existing plant, generally following disturbance (Schier 1981; Lacey and Johnston 1990; Del Tredici 2001). This allows for the spatial spread and multiplication of genetically identical individuals (ramets) that are, or have the capacity to be, independent (Peterson and Jones 1997; Del Tredici 2001).

Transplanting disconnected “plantlets” has been suggested when germination of seeds is poor (Whisenant 1999). However, root suckers have an enduring physical connection to the parent plant which allows the movement of photoassimilates, inorganic nutrients and water (Caraco and Kelly 1991). Whilst the connection eventually becomes redundant, in resource limited environments it enhances the survival of juvenile root suckers over seedlings (Callaghan 1984; De Steven 1989; Barsoum 2001). This is particularly important for long-lived woody plants where juveniles are the most vulnerable age class (Peterson and Jones 1997).

Root suckers are a very important form of recruitment for rangeland trees and shrubs because the plant is less reliant on seedlings which are often slow growing and susceptible to moisture stress (Noble 1986; Lacey and Johnston 1990). In central Australia 15% of 426 species of dicotyledons produced root suckers, most commonly those species on the sand-dune / sand plain system (Maconochie 1982). Although root suckers can only be stimulated from certain responsive species when there is an existing, living root system available, the technique may have merit where there

- 8 - Chapter 1 Introduction is no recruitment of seedlings of existing trees or shrubs (e.g. Hall et al. 1964; Chesterfield and Parsons 1985; Auld 1990). In the rangelands of Australia there are many species which naturally produce root suckers and could be targeted to produce root suckers for restoration (Appendix 1).

1.4 Outline of research 1.4.1 Research needs in north-west Victoria

The semi-arid woodlands mostly comprise Belah woodlands (Casuarina pauper F. Muell. ex LAS Johnson) or Pine Buloke woodlands (Callitris gracilis ssp. murrayensis (J Garden) KD Hill and Allocasuarina luehmannii (RT Baker) LAS Johnson). They are recognised as being the most degraded vegetation community in north-west Victoria (LCC 1987). There is generally less than 40% of these woodlands remaining due to timber-cutting, clearing, overgrazing, erosion and fire (section DNRE 1996b; White et al. 2003). Remaining woodlands suffer from lack of recruitment and senescence of the overstorey, low cover and decreased floristic diversity of woody perennials in the understorey, invasion of weeds and development of scalds on floodplains and erosion and mobilisation of sand dunes (Westbrooke et al. 1988; Cheal 1993; DNRE 1996b).

Management guidelines for north-west Victorian national parks require active restoration of this degraded community (section 2.5), and legislation (Flora and Fauna Guarantee Act 1988 and the Commonwealth Environment Protection and Biodiversity Conservation Act 1999) requires the Buloke communities be protected. The key techniques to enhance the recruitment of perennial plant species are hand-planting and direct seeding. Although much time and money is spent annually on this, little research has been conducted into effectiveness or whether better strategies are available. Reduction of grazing pressure is essential for the protection of these communities and there is anecdotal evidence that communities may be responding favourably.

1.4.2 Focus of research

This research explores further the options for the restoration of semi-arid woodlands, namely techniques to enhance the recruitment of perennial plant species of the semi-arid woodlands in north-western Victorian national parks. Reduction of grazing pressure and the use of root suckers are key areas. My underlying emphasis is on management actions that can encourage small amounts of recruitment because gradual change in a consistent direction can have a large effect on the structure of populations (Watson et al. 1996). Large recruitment events are primarily the result of high rainfall, which is not in the control of the land manager.

I have focused on the Pine-Buloke and Belah woodlands, in particular because these are degraded (Westbrooke et al. 2001; Gowans and Westbrooke 2002) and are the current focus of restoration activities (DNRE 1996b). A large proportion of these woodlands in Victoria are contained in Hattah Kulkyne (HKNP) and Murray Sunset National Parks (MSNP, DNRE 1996b) and these

- 9 - Chapter 1 Introduction national parks are my study area. C. pauper is a dominant woodland species in MSNP and A. luehmannii is a dominant woodland species in HKNP. These species are difficult targets for restoration because they are slow-growing and direct seeding and hand-planting often fail (P. Murdoch, Parks Victoria – Hattah, pers. comm. 2002). Both species can reproduce via seedlings or root suckers and are thus suitable for exploration of root suckers as a tool for restoration.

1.4.3 Objectives of research

The objectives of this research were to firstly, review thoroughly the factors influencing the current condition of semi-arid woodlands in north-west Victoria. This includes an examination of climate, in particular recent rainfall events, a thorough overview of the influence of European settlement on the vegetation and a description of the establishment and management of north-west Victoria’s national parks. A suitable strategy to present information on vegetation change in arid vegetation was sought and the state-and-transition (S-T) model (Westoby et al. 1989) was used for this purpose. The model needed to be able to highlight opportunities for management and identify gaps in the knowledge base.

The process of developing the S-T model highlighted two large gaps in our existing knowledge: why was natural recruitment occurring in some areas but not others, and is there an alternative to direct seeding and hand-planting? Further research aimed to address these questions. Firstly, I sought to confirm anecdotal evidence of seedling recruitment in semi-arid woodlands and establish the timeframe over which this occurred. I then explored maternal effects which may correlate with enhanced seedling recruitment with the aim of using this information to assist managers to enhance restoration success in other areas.

Secondly, I aimed to investigate whether root suckers could be a viable alternative to direct seeding and hand-planting for restoration work. Research was conducted to determine whether root suckers can be initiated when required, and to assess impacts of this on the population genetics of existing stands. The research used various combinations of damage and exposure of in situ tree roots to initiate the growth of root suckers. Genetic fingerprinting (Amplified Fragment Length Polymorphisms) was used to identify root suckers in natural stands and establish the genetic diversity of C. pauper and A. luehmannii. I intended to provide guidelines for initiating root suckers whilst maintaining genetic diversity.

- 10 - Chapter 1 Introduction

The chapters of this thesis address these topics:

• Introduction to restoration ecology in arid areas (this chapter), • Introduction to the study area and species (Chapter 2), • Review of the modelling of vegetation change in arid areas, and restoration efforts in north-west Victorian national parks. Development of a state-and-transition model for semi- arid woodlands in northwest Victoria (Chapter 3), • Identification of conditions where natural recruitment is occurring and determination of key factors that have contributed to this outcome (Chapter 4), • Description of techniques to initiate root suckers (Chapter 5), • Investigation of the genetic implications of encouraging the growth of root suckers (Chapter 6), and • Review of the key findings (Chapter 7).

- 11 - Chapter 2 Study area and species

Chapter 2 Study area and species

2.1 Study area 2.1.1 Location

The study area is two national parks in north-west Victoria: Hattah Kulkyne (HKNP) and the northern portion of Murray Sunset National Park (MSNP, Figure 2.1). The area was chosen because it contains large areas of semi-arid woodlands which have been negatively affected by past management practices. Woodlands are currently in a poor to fair (HKNP: Gowans and Westbrooke 2002) or degraded condition (MSNP: Westbrooke et al. 2001) and active restoration is underway (DNRE 1996b). Both parks are managed by Parks Victoria.

& MURRAY-SUNSET HATTAH-KULKYNE 01530 NATIONAL PARK NATIONAL PARK & kilometres

Figure 2.1 Map of the study area in north-west Victoria, showing Hattah Kulkyne and Murray Sunset National Parks. 2.1.2 Geomorphology and soils

The area consists primarily of marine sandy and clay sediments deposited in the Tertiary (Parilla Sands and Blanchetown Clay) and reworked by the wind in the Quaternary to form dune systems (Woorinen Formation and Lowan Sands). The following description is abbreviated from White et al. (2003).

- 12 - Chapter 2 Study area and species

2.1.2.1 The Riverine Plain and modern floodplains Modern floodplains occur along all major streams and are flanked by the Riverine Plain, alluvial sediments deposited by ancient and current stream courses. The Hattah Lakes in HKNP are fed by an anabranch of the Murray River and are a smaller remnant of a once larger floodplain system. The plains are dominated by grey, calcareous, sodic, clay soils without marked differentiation into horizons. The clays are generally alkaline, particularly at depth, and moderately to highly fertile due to incorporation of organic matter into deep cracks and gilgais formed by wet-dry cycles. Brown clays occur on less frequently flooded areas.

2.1.2.2 Siliceous Lowan Sand dunes and plains The tongue-shaped Lowan sand dune fields extend from west to east and comprise jumbled and parabolic dunes. The often steep dunes consist of loose, pale greyish-yellow to white quartz sand, neutral-acidic and containing little clay and relatively little lime (calcium carbonate). The soils are extremely infertile, have minimal accumulation of humus, are readily permeable and free draining making them unsuitable for cropping and prone to erosion; therefore much of the native vegetation has been retained.

2.1.2.3 Calcareous Woorinen Formation dunes and plains The regular, east-west, parallel, low-relief Woorinen dunes are composed of reddish-brown siliceous silty sands and red calcareous silty clay and sandy clay. The soils are relatively calcareous, and usually contain plates and nodules of carbonate that at depth grade into calcrete hardpans. Soils on the plains have a higher clay content and are sandy loams overlying sandy clays. The Woorinen sands are slightly more alkaline, less permeable and more consolidated than the Lowan sands. They have a higher clay content and are also more fertile, especially the redder sands, and are more suited to cropping and hence clearing

2.1.2.4 Groundwater discharge sites The Raak Plain groundwater discharge site is a shallow depression covering approximately 300 km2. It comprises sandplains, gypsum flats, gypsite (copi) hills, and salinas. Like other low-lying areas, it has relatively fertile clayey soil. Boinkas are sites of groundwater discharge where saline springs result in areas bare of vegetation. Lunettes or shoreline ridges are present on the eastern shores of boinkas, with wind forming clay and silt lunettes and wave action forming sandy lunettes.

2.1.3 Vegetation

Riverine and floodplain forests, mallee dune fields and semi-arid woodlands comprise the principal vegetation types of north-west Victoria. The riverine and floodplain forests are dominated by River Red Gum E. camaldulensis Dehnh. or Black Box E. largiflorens F. Muell.

- 13 - Chapter 2 Study area and species with an open understorey of small trees including Golden Wreath Willow Acacia saligna (Labill.) H.L. Wendl. and Eumong A. stenophylla A. Cunn. ex Benth. Grasses (in frequently flooded areas) or chenopod shrubs (in occasionally flooded areas) comprise the understorey.

The mallee dune fields of both the Woorinen and Lowan formations are dominated by multi- stemmed mallee eucalypts. On shallow sands the mallee is taller, more open and dominated by Oil Mallee Eucalyptus oleosa F. Muell. ex Miq. subsp. oleosa, White Mallee E. gracilis F. Muell. and others (nomenclature follows Anonymous 2004a; Ross 2004). The open shrub layer comprises mainly non-sclerophyllous shrubs. On deeper sands (Woorinen), the mallee is shorter and dominated by Grey Mallee E. socialis F. Muell. ex Miq., Dumosa Mallee E. dumosa A. Cunn. ex Oxley, and others, over a very dense layer of Porcupine Grass Triodia irritans N.T. Burb. with a few sclerophyllous shrubs. On deeper and strongly siliceous jumbled dunes (Lowan), the dominant species are Yellow Mallee E. incrassata Labill. and Narrow-leaf Mallee E. leptophylla F. Muell. ex Miq. A tall, sclerophyllous shrub layer and ground stratum are present, also dominated by T. irritans.

The semi-arid woodlands intergrade with mallee eucalypts but often occur on better soils. They are variously dominated by Belah Casuarina pauper F. Muell. ex LAS Johnson, Buloke Allocasuarina luehmannii (RT Baker) LAS Johnson , Silver Needlewood Hakea leucoptera R.Br., Sugarwood Myoporum platycarpum ssp. platycarpum R. Br., Slender Cypress Pine Callitris gracilis ssp. murrayensis (J Garden) KD Hill and Cattlebush Alectryon oleifolius ssp. canescens (Desf.) S. Reynolds. For simplicity, the latter three species will hereafter be referred to as M. platycarpum, C. gracilis and A. oleifolius.

Belah woodlands are dominated by Casuarina pauper and, in an ungrazed condition, have a well- developed and diverse understorey of chenopods and native herbs and in some cases grasses. Typical tall shrubs include Acacia spp., Leafless Ballart Exocarpus aphyllus, A. oleifolius,Twin- leaf Emubush Eremophila oppositifolia, Weeping Pittosporum Pittosporum phylliraeoides and Quandong Santalum acuminatum. Typical smaller shrubs include Desert Cassia Senna artemisioides subspp., Emubush E. glabra, Daisy Bush Olearia pimeleoides, O. muelleri and Prickly Fanflower Scaevola spinescens. The ground flora includes herbs, sub-shrubs, tussock grasses and lianes such as Purple Pentatrope Rhyncharrhena linearis, Bush Banana Marsdenia australis, and Climbing Twinleaf Zygophyllum eremaeum (LCC 1987). Belah woodlands tend to be found in the scattered east-west dunes on ridges and plains in the north, and fringing the Noora depression in the west. They are also found on locally high ridges without linear dunes such as around Yarrara and Robinvale and in plains formed by evaporative basins such as the Raak Depression and on lunettes (LCC 1987).

- 14 - Chapter 2 Study area and species

Gypseous Plain woodlands community occurs on undulating plains where powdery gypsum (“copi”) outcrops. Scattered old trees of M. platycarpum suggest this species may have once dominated and occasionally shrubs such as Maireana spp. remain of what was probably once a shrubby chenopod understorey. However only highly disturbed remnants remain; the original composition is unclear (LCC 1987).

Pine-Buloke woodlands are codominated by Allocasuarina luehmannii and C. gracilis with an open understorey containing Acacia spp., P. phylliraeoides or Santalum acuminatum. Like Belah woodlands, Pine Buloke woodlands are found in plains formed by evaporative basins such as the Raak Depression, and on lunettes. Unlike Belah woodlands, Pine-Buloke woodlands can be found on the higher alluvial plains on terraces beside the Murray River and some creek courses (LCC 1987).

In 1985-87 the Land Conservation Council (LCC) produced the first detailed descriptions and mapping of 31 vegetation communities in the broader area of north-west Victoria. The five semi- arid woodland communities described were Belah Woodland, Pine-Buloke Woodland and Savannah Woodland / Savannah Mallee / Grassland Mosaic of the lunettes and ridges and Gypseous Plain Grassland and Sandplain Grassland of the Evaporative Basin (Boinka) (LCC 1987).

The description of these vegetation communities was modified in 1999 to correspond with vegetation mapping conducted in other states. The communities were termed Ecological Vegetation Classes (EVCs): “floristic communities that appear to be associated with a recognisable environmental niche … described through a combination of its floristic, life-form and reproductive strategy profiles” (DNRE 1996a). Both the definitions and distribution of the communities was further modified in 2002-3 following more extensive mapping of extant native vegetation and modelling of pre-European vegetation (White et al. 2003). The terminology of the vegetation communities differed from 1999 to 2003 (

- 15 - Chapter 2 Study area and species

Table 2.1).

The 2003 descriptions were not readily available until I had established much of the methodology for this project. Therefore, for mapping and planning my project I used the 1999 vegetation communities.

- 16 - Chapter 2 Study area and species

Table 2.1 Relationship between the 1999 vegetation communities and the more recently mapped 2003 vegetation communities. Vegetation community: 19991 Vegetation community: 20032

Pine-Buloke Woodland Semi-arid woodland

Belah Woodland Semi-arid woodland

Semi-arid Chenopod woodland

Semi-arid Parilla woodland

Savannah Woodland / Savannah Chenopod Mallee Mallee / Grassland Mosaic Semi-arid woodland

Plains Savannah

Gypseous Plain Grassland Semi-arid Chenopod woodland

Sandplain Grassland Semi-arid Chenopod woodland

Sources: 1: LCC (1987), 2: White et al. (2003)

2.2 Climate 2.2.1 North-west Victoria

Approximately 60 years of climate records are available from Bureau of Meteorology Stations to the south (Walpeup, -35.1200o S 142.0039o E), and the north (Mildura Airport, -34.2306o S 142.0839o E) of HKNP (Anonymous 2004b). Regional temperatures are warm with mean daily maximum / minimum temperatures varying from 31.2oC / 15.9oC in summer to 16.1oC / 4.9oC in winter at Mildura and Walpeup. The warmest month is January and the coolest month is July. Annually there are 4.2 days in Mildura and 1.9 days in Walpeup where the minimum temperature is less than 0oC. Severe frosts occurred during the 1982 droughts and temperatures less than -5oC have been recorded at HKNP (P. Murdoch, Parks Victoria – Hattah, pers. comm. 2003).

Regional rainfall is very low with mean annual rainfall varying from 342 mm at Walpeup, to 268 mm at Mildura occurring mostly during the winter and spring. Thunderstorms during summer influenced by tropical systems from the north can provide significant local rainfall and monthly and daily rainfall extremes typically occur in the warmer months. The highest recorded monthly rainfall was 128 mm in March in Mildura and 134 mm in October in Walpeup. The highest recorded daily rainfall record was 94 mm in January in Walpeup and 91 mm in March in Mildura. Extended periods of above-average rainfall over large areas of south-eastern Australia have occurred in the years of 1832-1834, 1851-1853, 1871-1873, 1954-1956 and 1973-1975 (Westbrooke 1998; Bureau of Meteorology 2004). Droughts are natural features of the mallee and

- 17 - Chapter 2 Study area and species occur, on average, once per decade (Sluiter et al. 1997a). The most severe droughts since records began were in 1895-1902, 1913-1915, 1918-1920, 1925-1930, 1942-1945, 1966-1968, 1971- 1973, 1981-1983 (LCC 1987) and 1991-1995 (Bureau of Meteorology 2004).

Regional evaporation is high with the monthly mean evaporation exceeding rainfall in all months and by approximately 15-fold in January. Winds are highly variable, generally from the south through to the west, with strong winds predominantly from the west or north.

2.2.2 Rainfall events

The success of natural recruitment or restoration attempts in arid areas depends on the amount of rainfall (Whisenant 1999). However, plant growth in arid areas is not so much a direct response to rainfall, but rather to soil moisture (Reynolds et al. 2004). Analysis of simple cumulative monthly or yearly rainfall totals are misleading estimators of soil moisture (Paruelo and Sala 1995). Soil moisture depends on the size of a rainfall event, pre-existing soil moisture, the temporal separation of falls, evaporation, soil type and the effects of runon and runoff (Ogle and Reynolds 2004). Most rainfall events are small (< 5mm Hodgkinson and Freudenberger 1997; Sadras and Baldock 2003), and suffer from high evaporation and therefore may not stimulate all biological processes (Schwinning et al. 2003; Reynolds et al. 2004; Schwinning and Sala 2004; Snyder et al. 2004). Mid-summer rainfall of 5 mm only increased the soil water potential in the upper 50 mm for two days (Sala and Lauenroth 1982), whereas following complete saturation of the soil profile it was 39 days before a reduction in soil water potential at a depth of 60 cm was observed (Sala et al. 1981).

2.2.2.1 Rainfall events at Hattah Kulkyne National Park I characterised the size, frequency and distribution of rainfall events at HKNP. Daily rainfall records have been kept since 1963 from a rain gauge at the Rangers’ Office in the southern portion of HKNP (622,575E 6,153,146N), and monthly pan evaporation from 1969 to 1983.

The highest annual rainfall was 636 mm in 1973, and the lowest was 112 mm in 1963 and 1967. The median annual rainfall (1964-2004) was 299 mm (Figure 2.2). The largest daily rainfall event was 85 mm in May 1989. Rainfall is winter / spring dominated with a median total rainfall in these seasons of 180 mm versus a summer / autumn median of 124 mm (Figure 2.3).

There were 2697 days (21.7%) at HKNP between 1963 and 2004 where some rain fell. Most of these rainfall events were small with 71% of wet days receiving less than 5 mm of rain. Fifteen percent of rainfall events were between 5 mm and 10 mm, 10% were between 10 mm and 20 mm and a mere 4% of rainfall events were greater than 20 mm. This is a common pattern and can be modelled adequately using power laws (Sadras and Baldock 2003). The amount of rainfall contributed by small events (<5 mm) was relatively constant between the years (mean 75 ± SD 26

- 18 - Chapter 2 Study area and species mm, range 30 – 140 mm), whereas the amount of rainfall contributed by larger events (>5 mm) varied considerably (235 ± 89 mm, 64 - 529 mm). This may be a general phenomenon of semi- arid areas and suggests that high rainfall years are attributable to a few large rainfall events rather than many smaller events throughout the year (Golluscio et al. 1998).

700

600

500

400

300 Annual rainfall (mm) Annual 200

100

0 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 Year

Figure 2.2 Annual rainfall at HKNP from 1964 to 2004. The median annual rainfall (broken line) is 299 mm.

35

30

25

20

15 Median rainfall (mm) rainfall Median 10

5

0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month

Figure 2.3 Median monthly rainfall at HKNP showing a winter / spring dominance. Based on rainfall records from 1964 – 2004.

- 19 - Chapter 2 Study area and species

2.2.2.2 Modelling soil moisture Water balance modeling uses daily rainfall, evaporation and soil characteristics to follow patterns of soil moisture and rangeland productivity over time (Hobbs et al. 1994; Paruelo and Sala 1995; Kemp et al. 1997; Paruelo et al. 2000; Reynolds et al. 2000; Huxman et al. 2005), including the effect of soil moisture on recruitment (Harrington 1991; Gao and Reynolds 2003). Water balance is based on the law of conservation of mass – the amount of water added to the system via precipitation is equal to the amount lost. Losses are from evapotranspiration (ET), runoff from the soil surface and deep drainage into soil layers below the root profile. ET usually accounts for 90% of precipitation in dryland ecosystems (Wilcox et al. 2003 cited in Paruelo and Sala 1995; Huxman et al. 2005). ET is a combination of water intercepted by vegetation and evaporated directly back to the atmosphere, water evaporating from the soil surface, and water infiltrating the soil and being transpired by plants (Zhang et al. 2002). Runoff can be important in semi-arid regions when high rainfall intensity exceeds the infiltration capacity of the soil (Hortonian overland flow: Zhang et al. 2002). Deep-drainage is generally negligible in low rainfall areas (Paruelo and Sala 1995) and in the southern Victorian mallee, only daily rainfall greater than 75 mm is considered likely to result in losses to deep drainage (Zhang et al. 1999).

The recommended means for estimating ET is the Penman-Monteith equation (Allen et al. 1998). This equation uses radiation, air temperature, air humidity and wind speed data to estimate potential evapotranspiration (PET): the volume of water evaporated and transpired by a dense stand of actively growing short grass with unlimited water. However, as all the required meteorological data are seldom available, potential evapotranspiration can be approximated from an evaporation pan reading (Chiew and McMahon 1992). Pan evaporation (E) is the evaporation from a free water surface and integrates the effects of radiation, wind, temperature and humidity. E is related to but is substantially greater than PET (Kemp et al. 1997).

PET = kPE where kP is the pan coefficient, range 0.45 to 0.85 (Allen et al. 1998).

Furthermore, actual evapotranspiration (AET) is usually less than PET, as a real field may not be uniformly dense and well watered (Allen et al. 1998).

AET = kCPET where kC is the crop coefficient which averages soil evaporation and plant transpiration.

2.2.2.3 Simple bucket model of soil moisture for HKNP To better determine ecologically significant rainfall events, I used a simple bucket model of soil water-balance (Maidment 1993). I used daily rainfall records from HKNP (1964 to 2004) to fill

- 20 - Chapter 2 Study area and species the bucket and evapotranspiration based on pan evaporation rates (HKNP: 1969 to 1983) to empty the bucket. When the bucket was full, the extra water was lost to deep drainage and / or runoff.

Mt = M0 + P - L - ET where Mt is the soil moisture at time t, M0 is the stored soil moisture, P is precipitation, L is losses from runoff and deep drainage and ET is evapotranspiration. The simplified equation is based on the WATBAL model (Keig and McAlpine 1974).

The model does not consider factors such as soil hydraulic conductivity but was chosen because it avoids unnecessary complexity and requires minimal input data. The model is subject to systematic error resulting from simplifying assumptions (Walker and Langridge 1996; Zhang et al. 2002). However, I intended that the model merely suggest associations between soil moisture and success of restoration activities (Harrington 1991; Zhang et al. 2002).

Rainfall (P) was measured as mm / day. I assumed that total available water holding capacity in the 0-500 mm profile was 65 mm (Jensen et al. 1990). This is an upper limit estimation with other estimates for the Victorian mallee being in the vicinity of 50 mm (Sadras et al. 2003; Murphy and Leys 2004). Most of the annual water uptake by plants is from this layer (Reynolds et al. 2004) and summer rainfall, for all but the largest events, tends not to penetrate below this layer (Reynolds et al. 2000). For simplicity I assumed that the soil would continue to accept rainfall

(L=0) until the soil moisture (M0 + P) reached field capacity of 65 mm (Harrington 1991; Hobbs et al. 1994). This ignores runoff from high intensity rainfall which may be important but for which I had no data.

Evapotranspiration (ET) is one aspect of the soil-water budget that is subject to considerable uncertainty. I estimated actual ET from the long-term monthly average data for Class A pan evaporation (E) from the years 1969 - 1983. Long-term monthly averages improve the accuracy of this relationship (Chiew and McMahon 1992; Allen et al. 1998). E is usually much less than ET (Kemp et al. 1997) so E was multiplied by a soil moisture stress index which limits the potential evapotranspiration rate as a function of the available water in the soil (Shuttleworth 1993). A simple linear relationship between E and ET was adopted where ET was 0.7*E when the soil was at field capacity and 0.15*E when the soil was dry (Harrington 1991).

2.2.2.4 Rainfall and soil moisture at HKNP, 1963 -2004 The water-balance modeling is presented as soil moisture (mm in upper 500 mm soil profile) using a daily timestep (Figure 2.4). Summer and autumn are the critical periods for woody seedling establishment because water stress is highest at this time (e.g. Schwinning et al. 2005) and competition for water is greatest (Davis et al. 1998; Davis et al. 1999; Rey Benayas et al. 2001; Maestre et al. 2003). Harrington (1991) found that more than four consecutive weeks in

- 21 - Chapter 2 Study area and species autumn of soil moisture greater than 55% of field capacity and less than four consecutive weeks in the following summer with zero soil moisture was necessary for the establishment of Dodonaea viscosa. These prescriptions are likely to be site specific but do offer an indication of the magnitude of soil moisture effects. I examined the output of the soil moisture model produced for HKNP and looking for wet autumns (>30 days of soil moisture >50%) not followed by dry summers (>30 days of 0% soil moisture). Only the years 1971 and 1973 clearly met both the wet and dry soil moisture criteria for recruitment of woody seedlings:

• 1971 had 36 wet days in late autumn, not followed by a dry period until 53 days in summer 1972/73. • 1973 had 38 wet days in late autumn, followed by 110 wet days in winter / spring. There were no dry periods in summer 1973/74, and the winter of 1974 has 95 wet days before a period of 42 dry days in summer 1974/75. The years 1980 and 1992 met the wet criterion and the dry criterion however there was a dry period in the following autumn:

• 1980 had 33 wet days in late autumn, followed by 76 wet days in winter, however there were 40 and 32 dry days in autumn 1981. • 1992 had 34 wet days in late autumn but this was followed by 31 dry days in autumn 1993. The years 1974, 1986, 1989 and 1990 met the wet criterion but there was a dry period in the following summer, often as well as other dry periods:

• 1974 had 95 wet days in late autumn but this was followed by 42 and 32 dry days in summer 1974/75 • 1986 had 42 wet days in late autumn, followed by 106 wet days in winter, however there were 32 dry days during summer 1986/87. • 1989 had 96 wet days in late autumn but this was followed by a dry spring of 41 days, and three dry periods (36, 48 and 34 days) in summer / autumn 1989/90 • 1990 had 81 wet days in late autumn but 38 and 45 dry days in summer / autumn 1990/91 The longest wet periods tended to be in winter / spring (118 days in 1978, 110 days in 1973, 106 days in 1986). The longest dry period was in summer / autumn (112 days in 2004), summer (85 days in 2001/02, 78 days in 1964/65, 71 days in 1985) and autumn (72 days in 1997).

- 22 - Chapter 2 Study area and species

80 70 70 60 60 50 50 40 40 30 30 20 20 D aily rainfall (m m ) 10 10 Soil m oisture (m m ) 0 0 1964 1965 1966 1967 1968 1969 Year

80 70 70 60 60 50 50 40 40 30 30 20 20 D aily rainfall (m m ) 10 10 Soil m oisture(m m ) 0 0 1970 1971 1972 1973 1974 1975 Year

80 70 70 60 60 50 50 40 40 30 30 20 20 D aily rainfall (m m ) 10 10 Soil m oisture (m m ) 0 0 1976 1977 1978 1979 1980 1981 Year

- 23 - Chapter 2 Study area and species

80 70 70 60 60 50 50 40 40 30 30 20 20 Daily rainfall (m m ) 10 10 Soil m oisture (m m ) 0 0 1982 1983 1984 1985 1986 1987 Year

80 70 70 60 60 50 50 40 40 30 30 20 20 Daily rainfall (m m ) 10 10 Soil m oisture (m m ) 0 0 1988 1989 1990 1991 1992 1993 Year

80 70 70 60 60 50 50 40 40 30 30 20 20 Daily rainfall (m m ) 10 10 Soil m oisture (m m ) 0 0 1994 1995 1996 1997 1998 1999 Year

- 24 - Chapter 2 Study area and species

80 70 70 60 60 50 50 40 40 30 30 20 20 Daily rainfall(m m ) 10 10 Soil m oisture (m m ) 0 0 2000 2001 2002 2003 2004 Year

Figure 2.4 Soil moisture and rainfall at HKNP, 1964 – 2004. The blue line is the soil moisture in the upper 500 mm of the soil profile (maximum 65 mm). The black bars represent daily rainfall totals.

2.3 Vegetation change since European settlement

Vegetation change in north-west Victoria is characterised by progressive deterioration resulting from overgrazing and clearing combined with massive wind erosion following droughts.

2.3.1 The years 1800-1870

Early explorers of north-west Victoria included Captain Charles Sturt in 1830 (Kenyon 1914b), Major Thomas Mitchell in 1836 (Mitchell 1839). The first Europeans to explore HKNP were the overlanders Joseph Hawdon and Charles Bonney in 1838 (Coulson 1988). Little detail was given of the semi-arid woodlands by any of these parties.

Grazing of the mallee lands by domestic stock began with the large holdings of the squatters(Kenyon 1915a). By around 1847 the whole Murray River frontage and the Hattah Lakes area was stocked, mainly with sheep (Kenyon 1915a). Settlement was severely restricted by lack of water in areas not fringed by the Murray River but by 1865 much of the mallee had been parcelled out in runs, although many were speculative only (Kenyon 1914b). Clearing of native timber for housing and pasture also began at this time. Much timber was also felled to fuel the paddlesteamers which were the main form of transport along the Murray River from 1853 until the 1880s and persisted until the 1950s (LCC 1987).

Accounts of the vegetation at this time are lacking, probably due to the keen rivalry of run-seekers to secure the best land. Surveyors included E. R. White in 1849-1852, Pritchard in 1850-1, Baron von Mueller in 1853, Mr. Dallachy in 1858-60 and Dr. George Neumayer in 1860 but descriptions of the vegetation were largely qualitative. The latter is quoted as describing the area as “that

- 25 - Chapter 2 Study area and species horrible region covered with porcupine grass and stunted shrubs” and “Here and there the monotony of the mallee scrub was broken by groups of fine pine trees” (Kenyon 1914a, 1914b).

2.3.2 The years 1870-1920

From 1870-1880 the land was selected and settled (Kenyon 1915a) and grazing pressure increased dramatically. Rabbits arrived around 1866 and by 1876-78 had greatly affected the vegetation of the land (Kenyon 1915a). In 1880, reports came in of the “abomination of desolation” caused by thousands of rabbits which had consumed every blade of grass and stripped the trees bare of bark as far as they could reach (Kenyon 1914a). Sheep numbers peaked in the Mallee at 425,000 in 1871 (LCC 1987) and these heavy stocking rates are suggested to have had an even greater negative impact on the vegetation than the rabbit (Mitchell 1991). Short-term leases discouraged settlers from constructing fences to manage grazing pressure from stock, or controlling rabbits on their land (LCC 1987; White et al. 2003).

From 1850 to 1920, particularly from 1870 to 1900, substantial thinning of timber occurred to increase pasture, for fencing, housing, fuel, vineyard stakes, railway sleepers and during the 1890s, for wheat (Kenyon 1915b). Clearing of vegetation focused on the fertile clay-loam soils of the alluvial plains and avoided the siliceous Lowan sand country (White et al. 2003). The Wimmera-Mallee Water Scheme, introduced in the 1880s, and the mallee rail network established between 1903 and 1923 opened up more land for settlement, clearing and cropping. The invention of the “mallee roller” and “stump-jump plough” assisted the farmers to clear the land, despite the mallee lignotubers which had previously hampered their efforts. These machines were followed by even more efficient “chaining” using tractors (White et al. 2003).

The declining condition of semi-arid woodlands were compounded by the drought of 1877 to 1882 and the “Federation drought” of 1895-1902 (Bureau of Meteorology 2004). As early as 1906 “large clumps of Tobacco Bush, and immense fields of white and yellow Everlastings” were noted (Kenyon 1906). The occurrence of these species (probably Nicotiana glauca and Polycalymma stuartii respectively) points to substantial erosion because both occur in response to the mobilisation of sand dunes (Cunningham et al. 1992). This is supported by the 1912 observation of “hundreds of acres of wind-eroded red loam devoid of vegetation” (Williamson 1913). In 1915, naturalists visiting the area noted the lack of recruitment of the woody perennials (O'Donoghue 1915).

2.3.3 The years 1920-1950

Drought and the Great Depression (1929-1932) forced many farmers from their land and leases reverted to the Crown or were combined. But by the end of World War II, all the suitable land in the mallee had been settled and what was left is now public land (LCC 1987).

- 26 - Chapter 2 Study area and species

Dry years generally persisted from 1913 to 1929 in the mallee, with the drought of 1914 to 1915 being especially harsh (Bureau of Meteorology 2004). Drought, combined with poorly adapted farming practices and clearing of native vegetation, led to wide-scale erosion and irreparable damage to soil structure and remaining vegetation (Kenyon 1906; Chandler 1938; VNPA 1959). The World War II droughts of 1937-45, with a respite in late 1939, yielded dust storms throughout the “dirty thirties” and the summer of 1944-5, with further devastating losses of mallee topsoil (Bureau of Meteorology 2004).

Between the 1920s and 1940s most of the remaining, mature timber was cut for firewood for the boilers of irrigation pumps around Mildura and logged areas became grasslands (VNPA 1959). Survey plans in the 1920s noted that “all the good timber has been cut” (Callister 2004). Large stands of Callitris were felled for fencing materials, especially in the Millewa, until in the mid- 1930s the Lands Department forbade this on Crown leaseholds (LCC 1987). By 1941, the vegetation was clearly much altered and the open, parklike areas were attributed to European practices (Jones 1942). The loss of large belts of pine was noted in the 1940s, due to clearing, death from drought and insect attack, and loss due to windthrow in thinned stands. Pine was also fallen as feed for drought-stricken stock (Williamson 1913; Jones 1942; Zimmer 1944; VNPA 1959). Semi-arid woodlands showed a higher rate of clearance than surrounding vegetation types (Sluiter et al. 1997b). Over 90% of the woodlands outside Murray Sunset National Park were cleared and Callitris gracilis especially is substantially less common now than at the time of settlement (Callister 2004).

In the 1930s and 1940s the rabbit was in plague proportions and the area supported many rabbit trappers. The first field trial of the introduction of the myxomatosis virus in 1938-42 was successful but the virus did not spread due to the lack of the mosquito vector in the dry conditions.

2.3.4 From the 1950s onwards

The 1950s to the current day are characterised by the ongoing battle with grazing pressure. Myxomatosis spread rapidly in the early 1950s and rabbit numbers declined by as much as 95%. However resistance to the virus developed rapidly and by the late 1970s, the Kulkyne sandhills were considered the most degraded crown land in the district due to rabbits (Mueck 1981). Baiting, fencing and ripping and fumigating warrens were required to manage rabbit numbers in the 1980s in HKNP [section 2.5.2.1 \Williams, 1995 #1233]. Overabundance of kangaroos became an increasing problem in the 1980s and culling commenced in earnest in 1990 (Cheal 1986; Mueck 1987; Coulson 1988; Coulson and Norbury 1988; Coulson et al. 1990; DNRE 1996b; Sluiter et al. 1997a; Coulson 2001). Rabbit Haemorrhagic Disease (RHD) achieved widespread reductions in rabbit numbers following its arrival in the mallee in 1996 (Sandell 2002).

- 27 - Chapter 2 Study area and species

Droughts also continued with a dry decade post 1957, followed by another drought in 1965-68 (Bureau of Meteorology 2004). Severe dust storms and exceptionally severe frosts were experienced in the 1982-3 drought (Bureau of Meteorology 2004) causing much damage to vegetation. The 1990s and early 21st century have generally been in the grip of El Nino and have suffered prolonged dry periods (Bureau of Meteorology 2004). In contrast, widespread exceptional rainfall occurred in the years of 1954-56 and 1973-75. The former period coincided with low rabbit numbers and resulted in extensive recruitment for much semi-arid vegetation. The latter period however, was a time of high grazing pressure from rabbits and stock and little recruitment occurred (Chesterfield and Parsons 1985; Westbrooke 1998).

2.4 Establishing the national parks 2.4.1 Hattah Kulkyne National Park

In 1915, the area within 805 m (40 chains) of Lakes Hattah, Little Hattah, Lockie, Brockie, Mournpall, Yelwell and Konardin was declared a sanctuary for native game (Anonymous 1916). This was followed by further reservations of state forest in 1919 (1080 Ha), and 1924 (additional 30,500 Ha). In 1941, an additional 17,010 Ha of primarily mallee was set aside to protect the Lakes from sand-drift, and the Kulkyne National Forest was declared a sanctuary for native game (Morrison 1941; Jones 1942). In 1960, Hattah Lakes National Park was established. It included the southern section of the lakes system and the mallee in the west of the current park (17,820 Ha) despite protestations that the entire area should become national park (VNPA 1959). In 1980, following the recommendations of the Land Conservation Council (LCC 1977), the present-day HKNP was declared and areas along the Murray River were reserved as Murray Kulkyne Park (50,360 Ha: Young undated). The area was declared a Biosphere Reserve in 1981, under the UNESCO Man and the Biosphere Program. In 1982, the Hattah Lakes were designated under the Ramsar Convention on Wetlands of International Importance.

From 1919 to 1980 the area was managed by the Soil Conservation Authority and the Forestry Commission. The area excluding the mallee was cleared for pasture and supplied firewood for fencing, irrigation pumps, sleepers and vineyard trellis posts until around 1950 when most of the timber had been cut (Chandler 1938; Young undated). From 1980, management was for conservation by the government agency of the time, beginning with the National Parks Service and currently Parks Victoria.

There are 11 vegetation communities within HKNP (Table 2.2). Two of these, Pine Buloke Woodland and Savannah Woodland / Savannah Mallee / Plains Grassland Mosaic, are considered semi-arid woodland and total 9,320 Ha or 18.5% of the park; these woodlands occur in the eastern portion of the park. The mallee amounts to 43.2% of HKNP and dominates the western half of the

- 28 - Chapter 2 Study area and species park, and nearly all of the remaining area (28.5%) is Riverine Plain vegetation, except a small amount of shrubland (8.8%).

Table 2.2 Vegetation communities within HKNP. Broad Vegetation Community Area of Percent classification Park (Ha) of Park

Mallee Lowan Sands Mallee 21,766 43.2

Woorinen Sands Mallee

Shrubland Chenopod 4,443 8.8

Saline Shrubland

Riverine Drainage Line Grassy Woodland plains Riverine Grassy Forest

Plains Grassland / Drainage Line Grassy Forest 12,958 28.5 Mosaic

Black Box Chenopod Woodland

Lake Bed Herbland

Semi-arid Savannah Woodland / Savannah Mallee / Plains 6,312 12.5 Woodland Grassland Mosaic

Pine Buloke Woodland 3,008 6.0

Other Unknown / Unclassified 1,873 3.7

Bare rock / ground

Total 50,360 100

Source: LCC (1987)

2.4.2 Murray Sunset National Park

Pink Lakes State Park (507 km2) was established in 1979 following the recommendations of the LCC (LCC 1977). Further, extensive surveys by the LCC (LCC 1987) led to additional recommendations (LCC 1989), including the incorporation of the Sunset Country and other public land into MSNP in 1991. MSNP comprises an area of 646,140 Ha and is Victoria’s largest, continuous national park. The land was previously used for grazing and timber extraction.

It contains 15 vegetation communities, five of which are considered semi-arid woodland and total 9.9% of the park: Pine Buloke Woodland, Belah Woodland, Sandplain Grassland, Gypseous Plain

- 29 - Chapter 2 Study area and species

Grassland and Savannah Woodland / Savannah Mallee / Plains Grassland Mosaic, (Table 2.3). The mallee amounts to 80.2 % of the park, the shrublands 5.9%, and the remaining area (4.0%) is floodplain vegetation. My research focused on the semi-arid woodlands in the northern portion of MSNP, including woodlands around the Raak Plains, Rocket Lake, Taparoo, Berribee and Mopoke Hut.

Table 2.3 Vegetation communities within MSNP. Broad Vegetation Community Area of Percent classification Park (Ha) of Park

Mallee Broombush Mallee

Lowan Sands Mallee 518,125 80.2

Red-swale Mallee

Woorinen Sands Mallee

Shrubland Chenopod 46,109 5.9

Saline Shrubland

Floodplain Alluvial Plain Shrubland

Black Box Chenopod Woodland 25,866 4.0

Lake Bed Herbland

Riverine Grassy Forest

Semi-arid Belah Woodland 1,274 0.2 Woodland Gypseous Plain Woodland 5,438 0.8

Pine-Buloke Woodland 4,036 0.6

Sandplain Grassland 2,414 0.4

Savannah Woodland / Savannah Mallee / Plains 50,730 7.9 Grassland Mosaic

Total 646,140 100

Source: LCC (1987)

- 30 - Chapter 2 Study area and species

2.5 Management of national parks 2.5.1 Management objectives

Management objectives in the Mallee Parks Management Plan for Vegetation and Grazers include “actively rehabilitate degraded communities” through hand-planting and direct seeding of trees and shrubs, and “reduce total grazing pressure to a level which allows the native, perennial species to regenerate” (DNRE 1996b). Kangaroo populations are identified as a key component of total grazing and the management objective is to “maintain kangaroo populations within limits which will allow the progressive long-term recovery of key woody perennials while still preserving viable kangaroo populations” (DNRE 1996b). Scalded areas should be revegetated by controlling grazing pressure, increasing surface roughness and direct seeding. Eroded sand dunes should be reshaped and revegetated initially with a non-regenerating cover crop and then with native vegetation. Adequate ground cover should be maintained to prevent further erosion. The objective is to allow the progressive long-term recovery of key, woody, perennial vegetation (DNRE 1996b).

These statements are reasonable examples of well-defined, value-based goals for restoration ecology. The goals for restoration should include valued attributes, dynamics of the system being restored, the factors leading to degradation and the action required to achieve restoration must be included (Hobbs and Norton 1996; Hobbs and Harris 2001). The valued attributes of the landscape are identified as native, perennial plant species, particularly woody perennials, as well as viable populations of kangaroos. However, more detailed description of the desired species composition would make success more measurable. The vegetation dynamics in semi-arid woodlands involve infrequent recruitment due to unreliable rainfall, exacerbated by grazing pressure. This is acknowledged in the loose requirement of “progressive long-term recovery”, although the timescale is not defined. Grazing is identified as a key factor leading to degradation, and introducing propagules and managing grazing pressure are specified as the actions necessary to achieve restoration.

2.5.2 Grazing pressure

The combined influence (total grazing pressure) of stock, rabbits, goats and kangaroos has a detrimental effect on the vegetation in north-west Victoria (LCC 1989). Stock were introduced progressively to the mallee from around 1847 (Kenyon 1915a). Grazing by stock, particularly sheep in semi-arid areas is largely responsible for eliminating seedlings of long-lived native perennials (Lange and Purdie 1976; Lange and Willcocks 1980; Cheal 1993; Tiver and Andrew 1997). In HKNP, sheep were phased out by 1963 and cattle grazing commenced in 1959 and was phased out by 1977 (Cheal 1986). In MSNP, the phase out of stock was complete by 1996 (Sandell et al. 2001). Rabbits, kangaroos and goats now comprise the herbivores in HKNP and

- 31 - Chapter 2 Study area and species

MSNP. Natural recruitment of perennial plant species relies on significant management input to reduce grazing pressure (DNRE 1996b).

2.5.2.1 Rabbits The rabbit is widespread throughout southern Australia and is our most damaging vertebrate pest (Williams et al. 1995). Semi-arid woodland and grassland communities typically support high rabbit numbers (Cheal et al. 1992; Sandell et al. 2003).

Rabbits are selective feeders, but during droughts will eat plants, shrubs and grasses indiscriminately, including stripping the bark from shrubs, effectively ringbarking them. The rabbit is primarily responsible for an observed lack of recruitment in Australian rangelands (Cochrane and McDonald 1966; Crisp and Lange 1976; Auld 1995b; Cohn and Bradstock 2000); exceeding even the impact of stock (Ireland 1997). Even low densities of rabbits can eliminate sparsely distributed seedlings and root suckers (Crisp 1978; Lange and Graham 1983; Cooke 1987; Lord 2002). The greatest grazing pressure from rabbits is within 50 m of the warren and there are few palatable species within 100 m of warrens (Foran 1986; Leigh et al. 1989). Rabbits also affect the soil nutrient and water status and increase soil disturbance and erosion, especially around warrens where excavated soil is mounded up. The germination of weeds over native species is favoured in this area (Eldridge and Simpson 2002).

Rabbits arrived in the mallee in around 1866 and by the 1970s it was noted “The Hattah Kulkyne area has had the worst rabbit infestation in Victoria” (LCC 1977). As early as 1914, efforts to poison rabbits with sulphur were noted in the HKNP area (O'Donoghue 1915). The Lands Department attempted to control rabbits in the Kulkyne area from the 1960s (Anonymous 1984) and the European Rabbit Flea was released in the late 1970s to act as a vector for the Myxoma virus spread via inoculated rabbits (Williams 1980). In 1980, HKNP was divided into management areas which were prioritised for treatment. The earliest control operations focused on the management area known as the Mournpall Block. Reinvasion of treated areas and immigration of rabbits from the mallee areas was prevented by major fencing works. Methods included aerially baiting with 1080 carrot pieces and follow-up ripping or fumigation of warrens (Williams 1980). More recently warrens are being imploded using a Rodenator Pro which injects and ignites a mixture of propane and oxygen (P. Murdoch, Parks Victoria – Hattah, pers. comm. 2004). Rabbit control via ripping, fumigation and burrow implosion commenced in MSNP in 1991.

Rabbit abundance is surveyed through spotlight counts along permanent transects and monitoring the density of active warrens. Management targets are 5 rabbits per km spotlight transects or 25 active entrances per km2 (DNRE 1996b). Works have generally maintained the abundance of rabbits at relatively low levels (2-8 per km). When RHD arrived in the area in 1996, abundance

- 32 - Chapter 2 Study area and species was further reduced to 0.5 per km (Sandell 2002). Rabbit abundance fluctuates, however spotlight transects indicate that at HKNP numbers have steadily increased since 2003 (Figure 2.5).

5

4.5

4

3.5

3

2.5

2

1.5 Rabbits per transect km transect per Rabbits 1

0.5

0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Year

Figure 2.5 The abundance of rabbits at HKNP. Data are means of (generally) ten spotlight transects of 5 – 10 km in length. Error bars are ± one standard error. Data: Parks Victoria.

2.5.2.2 Kangaroos Kangaroos have increased in abundance throughout Australia as a result of clearing to promote grassland, the introduction of artificial watering points and virtual elimination of the dingo (Calaby and Grigg 1989; McAlpine et al. 1999). Established relationships between body size and diet suggest that the large macropods are almost pure grazers (Norbury 1989). Kangaroos consume items other than grass in proportions depending on their relative availability, quality and palatability (Beetham 1987; Dawson 1989). When grass is limited, believed to be a threshold less than 400 kg per ha biomass of grass, kangaroos may switch to poorer-quality browse. Similarly where high-quality forbs or seedlings of woody species are abundant, these may be preferred over grass, especially if the grass is dry (Coulson et al. 1990).

The effects of kangaroos on natural recruitment is variable but can be important (Gardiner 1986), especially in north-west Victorian national parks (Cheal 1986; Hayes 1993; Sluiter et al. 1997a; Coulson 2001). Kangaroos have also been shown to move into areas which have been destocked or revegetated via direct seeding, and the intense browsing is sufficient to prevent establishment of seedlings (Norbury and Norbury 1993). Three species of kangaroo are found in both HKNP and MSNP, Eastern Grey Macropus giganteous, Red M. rufus and Western Grey M. fuliginosus.

- 33 - Chapter 2 Study area and species

In the 1980s, browsing by kangaroos was identified as a major impediment to recruitment of the vegetation in HKNP (Cheal 1986; Mueck 1987). In 1984 a rabbit-proof fence around the 5,780 Ha Mournpall Block was upgraded to kangaroo proof. 250 kangaroos were herded out of the Block but mustering was abandoned due to stress and injuries to the kangaroos. A month later a cull of 787 animals was conducted but the culling program was abandoned due to public outcry (Coulson 2001). In HKNP, the abundance of kangaroos peaked in 1990 at approximately 60 kangaroos km-2 (Sluiter et al. 1997a). Culling resumed inside the Mournpall Block following the release of a kangaroo management plan in 1990 (DCE 1990). Culling was conducted by professional kangaroo shooters accredited to standards (CONCOM 1985) and under the advice of experts and veterinarians who formed the Kangaroo Technical Advisory Committee. All shooters are accompanied by an authorised officer employed by what is now Parks Victoria. Shooters were instructed to work at night, with a spotlight, and to take animals regardless of sex or size in all accessible habitats (DCE 1990). A target density of five kangaroos km-2 was proposed and is still current, despite indications that fewer than three kangaroos km-2 in dunefields and a “negligible density” of kangaroos on floodplains is necessary to allow recruitment (Sluiter et al. 1997a). The culling program was extended to the entire park in 1996 following observed favourable vegetation response inside the Mournpall Block (Sluiter et al. 1997a). This was preceded by a severe drought and substantial kangaroo mortality in 1994-95 (Sluiter et al. 1997a). The culling program in MSNP commenced in March 2001.

Various techniques have been used to monitor kangaroo populations in north-west Victoria, including aerial census (Short and Grigg 1982) and more recently, helicopter transects. The most complete dataset is from permanent walking transects (Morgan 1986) in HKNP and MSNP (Figure 2.6 and Figure 2.7).

- 34 - Chapter 2 Study area and species

70

60

50

40

30

20

Estimated density (kangaraoos / sq. km) sq. / (kangaraoos density Estimated 10

0 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 Year of census

Figure 2.6 Density of kangaroos outside the Mournpall Block of HKNP from 1983 to 2001 estimated from walking transects (Morgan 1986). The black line with diamonds is Western Grey Kangaroos and the grey line with squares is Red Kangaroos. Error bars indicate ± one standard error. Data: Parks Victoria.

30

25

20

15

10

5 Estimated density (kangaroos / sq. km) sq. / (kangaroos density Estimated

0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year of census

Figure 2.7 Density of kangaroos in the northern area (Taparoo) of MSNP from 1992 to 2001, estimated using walking transects (Morgan 1986). The Black line with diamonds is Western Grey Kangaroos and the grey line with squares is Red Kangaroos. Error bars indicate ± one standard error. Data: Parks Victoria.

- 35 - Chapter 2 Study area and species

2.5.2.3 Goats Goats formed feral populations in the mallee in the 1930s (Tiver and Andrew 1997). Goats are primarily browsers rather than grazers and consume seedlings, bark and foliage of woody perennial species (Wilson 1977) but can switch to grass and forbs when these are green (Wilson et al. 1975). They can suppress the recruitment of woody perennials and severely damage or destroy mature trees and shrubs (Wilson et al. 1976; Harrington 1979; Auld 1993; Greene et al. 1998; Anonymous 1999; Cheal 2005b).

In HKNP, goats are culled opportunistically and by the Judas goat method, and attempts are made to trap the goats at watering points (P. Murdoch, Parks Victoria – Hattah, pers. comm. 2004). The “Judas Goat” technique involves attaching a radio collar to a feral goat and releasing it back into the wild to join up with other goats. The herd which the collared animal has joined is then located by radio-tracking the Judas goat and the goats are destroyed (Anonymous 1999). In MSNP, goats are the subject of an aerial census and a program of ground-shooting utilising volunteer shooters (R. Manning, Parks Victoria – Mildura, pers. comm. 2004). An aerial estimate of goat abundance was conducted by helicopter in MSNP in 2004 otherwise there is little formal assessment of the abundance of this species

2.6 Study species - Casuarinaceae 2.6.1 Taxonomy

The family Casuarinaceae includes four genera and 95 species from south-east Asia, Australia and Oceania. Three genera, Allocasuarina, Casuarina and Gymnostoma, occur in Australia and are mostly endemic (Johnson and Wilson 1993) although there is still some debate over the generic divisions (Hwang 1992). Gymnostoma is believed to be the most primitive genus, and Allocasuarina the most derived (Johnson and Wilson 1989; Dilcher et al. 1990). The section itself is moderately advanced (Johnson and Wilson 1989). Allocasuarina is divided into fourteen sections, the largest, sect. Cylindropitys (29 spp.), includes closely related taxa believed to represent recent evolutionary radiation (Wilson and Johnson 1989). A. luehmannii is the only species in sect. Platypitys. The taxonomic subdivisions are supported by morphological (Johnson and Wilson 1989; Dilcher et al. 1990; Hwang and Conran 2000), cytological (Barlow 1983) and genetic studies (Ho et al. 2002; Steane et al. 2003).

2.6.2 Description

The Casuarinaceae are easily recognisable trees or shrubs with slender, jointed, ribbed, grey-green branchlets and leaves reduced to whorls of minute teeth, 5-20 teeth per whorl in Casuarina and 4- 14 per whorl in Allocasuarina (Johnson and Wilson 1989; Wilson and Johnson 1989). The

- 36 - Chapter 2 Study area and species general morphology is xeromorphic, with stomata in grooves along the sclerophyllous branchlets, except in Gymnostoma (Ladd 1989).

All species are wind pollinated, with the reduced, unisexual flowers borne on individual plants (dioecious) or less commonly, both sexes on the same plant (monoecious) (Johnson and Wilson 1989; Cunningham et al. 1992). The ratio of male to female trees in dioecious populations is close to 50:50 or a slight male bias (Boland et al. 1996). Monoecious individuals are generally predominantly male with low percentages of female flowers (Boland et al. 1996).

The evolution of unisexual flowers has occurred in fewer than 10% of plants and maximises outcrossing because male and female flowers are spatially separated, although it necessitates a mating partner (Barrett 2002). Monoecious individuals may suffer from reduced fitness due to the transfer of pollen between flowers on the same plant (geitonogamy: Eckert 2000; Reusch 2001). However the seeds from monoecious C. cristata and C. cunninghamiana are viable, possibly due to temporal separation of flowering (Boland et al. 1996). The occurrence of both dioecy and monoecy in one species is uncommon and spatial reproductive isolation is believed to allow the persistence of both sexual systems (Dorken et al. 2002). Such reproductive isolation is unknown in casuarinas and I have observed single populations with both dioecious and monoecious individuals.

The inflorescence is a whorl of tooth-like bracts, each containing two bracteoles and a single unisexual flower. The male inflorescence is a catkin-like spike and the female is a globular or ovoid head. The infructescence is a woody cone containing winged nuts termed a “samaras” possessing a single seed. The samara (hereafter termed a seed) is grey or yellow-brown in Casuarina and red-brown to black in Allocasuarina. The cones open at maturity and the seeds are short-lived except in some species of Allocasuarina (not including A. luehmannii) where the cones remain unopened for several years and seeds are long-lived (Johnson and Wilson 1989). The seed has a thin outer exocarp and a mesocarp of inflated cells with spirally thickened walls. These cells swell and absorb water upon wetting forming a mucilage, especially in Allocasuarina. The mucilage, a hygroscopic gel, is possibly an adaption to improve seed soil contact, assist anchorage and root penetration or retain water and increase opportunities for germination in dry conditions (Torrey 1983; Turnbull and Martensz 1983; Ladd 1989; Raymond 1990). In A. luehmannii the percentage and rate of germination was higher in seeds where the mucilage was removed suggesting that mucilage may not be to assist germination as thought previously, and may be to prevent post-dispersal seed harvesting (Raymond 1990; Macaulay and Westbrooke in prep.).

Casuarina pauper was originally described as a subspecies of C. cristata; they are now separated based on the lower height, poor form and shorter bracteoles of C. pauper (Wilson and Johnson

- 37 - Chapter 2 Study area and species

1989). C. pauper has finely corrugated bark, smooth branches, and long cones which distinguishes it from A. luehmannii which has deeply fissured bark and squat cones (Figure 2.8).

IMAGE REMOVED DUE TO COPYRIGHT

Figure 2.8 A cone and branchlet of Casuarina pauper (left) and Allocasuarina luehmannii (right). The scale bar represents 10 mm for the cone and 2 mm for the branchlet (from Figure 47 and 49 Wilson and Johnson 1989). Most Casuarina spp. and Allocasuarina spp. occupy temperate environments with C. pauper, A. luehmannii, C. cristata, C. inophloia and C. decaisneana being the only arid-dwelling species (Ladd 1989). There are six species of Casuarina in Australia with three species in Victoria. C. cunninghamiana has been introduced. C. obesa grows in saline swampy area near lakes or rivers. C. pauper occurs through western New South Wales, north-western Victoria and central South Australia to southern inland Western Australia. The southern limit of its distribution is Swan Hill in Victoria. The distribution is contiguous with the closely related C. cristata which occurs in the northern areas whilst C. pauper occupies the southern areas (Figure 2.9). The species occurs on red-brown soils with light textured topsoil and calcareous subsoil; in Victoria it is restricted to the sandy rises in the north-west of the state (Wilson and Johnson 1989).

Allocasuarina spp. are trees or shrubs growing generally in soils deficient in nutrients. There are 59 species in Australia with thirteen species in Victoria. A. luehmannii is distributed through central Queensland and New South Wales to north-western Victoria and adjoining South Australia. The north-western limit of its distribution is Red Cliffs in north-west Victoria (Figure 2.9). It occurs primarily on non-calcareous soils (Wilson and Johnson 1989).

IMAGE REMOVED DUE TO COPYRIGHT

Figure 2.9 The distribution of Casuarina pauper (left), Casuarina cristata (centre) and Allocasuarina luehmannii (right, from Wilson and Johnson 1989) 2.6.3 Recruitment

Casuarinaceae produce many thousands of viable seeds each year per tree (Torrey 1983; Turnbull and Martensz 1983; Boland et al. 1996) however the seedbank is relatively short lived and the species must produce annual seed-crops for germination in both C. pauper (Auld 1995c) and A. luehmannii (Raymond 1990; Hayes 1993). Post-dispersal removal of seeds by ants can be up to 90% in C. pauper (Chesterfield and Parsons 1985) with similar results for A. luehmannii (Raymond 1990). Seeds are wind-dispersed and occasional long-distance dispersal of seeds may occur via ants, emus or transport by water (Auld 1995c). Seeds germinate readily in the laboratory, mostly in 15-30 days at 20-25oC (Turnbull and Martensz 1983; Ladd 1989; Hwang and Conran 2000).

Seeds are released during summer and germinate following the first rains, generally autumn; persistence of seedlings requires low grazing pressure and favourable follow-up rains

- 38 - Chapter 2 Study area and species

(Chesterfield and Parsons 1985; Auld 1995c; Macaulay and Westbrooke in prep.), or floods (Westbrooke and Florentine 2005). No A. luehmannii emergents and only suckers of C. pauper were reported from repeat-surveyed belt transects in MSNP and HKNP from 1991-1998, despite a reduction in, or in some cases zero, total grazing pressure (Sandell et al. 2001; Sandell 2002). Historical records from the Wimmera in the early 1870s (Morcom and Westbrooke 1998) and the Hattah area (O'Donoghue 1916) suggest that stands of young A. luehmannii were not common.

Clonal growth via root suckering is present in many species of Casuarinaceae including C. glauca, C. junghuhniana, C. cristata, C. cunninghamiana, C. pauper and A. luehmannii (Boland et al. 1996). Root suckers are believed to arise from shallow lateral roots following disturbance (Torrey 1983; Auld 1995c; Morison and Harvey 2001). It is not known whether root suckers are an important mode of recruitment in these species, or whether root suckers are an unusual occurrence related to human-mediated disturbance (e.g. Batty and Parsons 1992). Apomixis has been reported in the Allocasuarina Sect. Cylindropitys, specifically A. distyla, A. monilifera and A. striata although its occurrence in A. luehmannii or C. pauper is unknown (Burbidge 1973).

2.6.3.1 Ploidy level There are numerous instances of polyploidy within Allocasuarina, including A. luehmannii which is tetraploid (2n=4x=56) based on counts from up to five seedlings from one or two plants at each of three localities: Trangie, NSW; near Dubbo, NSW, and Elmore, Vic (Barlow 1959a). Species of Casuarina appear to be consistently diploid (2n=2x=18), although no data exist for C. pauper. Ploidy levels can vary within some species of Allocasuarina, especially the section Cylindropitys. For example A. littoralis has diploid and tetraploid populations with occasional triploid individuals (Barlow 1958) and A. distyla, A. striata, A. gymnanthera, A. paludosa, A. grampiana and A. pusilla also have multi-ploidy populations (Barlow 1959b; Wilson and Johnson 1989). Where triploid populations occur, they tend to be exclusively female (Barlow 1959a). A. nana has monoecious tetraploids and dioecious diploids (Barlow 1959b).

2.7 Data collection

I used a hand-held Garmin 12XL Global Positioning System (GPS) to record locations in the field using World Geodetic System (WGS84), AMG Eastings and Northings, Zone 54. Geographic Information System (GIS) software (“MapInfo Professional”) was used to compile, store, retrieve, analyse and display spatial (mapped) data. I obtained the map layers used in this thesis from Parks Victoria via the Department of Sustainability and Environment (1999 dataset).

All photographs were taken by myself with an Olympus C750 Ultrazoom digital camera, unless otherwise credited. In 2005 I used a handheld Personal Digital Assitant (palmOne Tungsten C) to collect my data. I used the freeware CyberTracker (CyberTracker Conservation 2004) to design

- 39 - Chapter 2 Study area and species databases for the collection and analysis of data. Data were stored in Microsoft Access databases and statistical analyses were performed using programs described in the Methods section of each chapter.

The use of a shotgun to collect samples from very high trees, as described in Chapters 4 and 6, was carried out by Mr. P. Murdoch, who is a licensed shooter and as the Ranger in Charge of HKNP, is accredited to shoot in a national park.

- 40 - Chapter 3 State-and-transition modelling

Chapter 3 Development of a state-and-transition model for semi-arid woodlands of north-west Victoria using qualitative data

3.1 Scope and objectives

To restore an ecosystem, it is necessary first to understand the system components and interactions and how management or restoration may affect them (Hobbs and Norton 1996; Yates and Hobbs 1997; Jackson and Bartolome 2002). Conceptual models are used to organise this knowledge in a logical, accessible manner and in a way that encourages the incorporation of new information and experimental testing of long-held beliefs. The most widely applied model for rangeland vegetation in Australia is the “state-and-transition” (S-T) model (Westoby et al. 1989; Yates and Hobbs 1997). This model describes vegetation communities (states) and how changes (transitions) occur between states. The S-T model has been applied to many ecosystems and despite the variety of vegetation assessed, there are clear similarities in the models produced. There is also substantial theoretical knowledge, for example disturbance ecology and succession theory (Hobbs and Norton 1996), which can be applied to rangeland vegetation to create the framework of a widely applicable S-T model. I reviewed the literature to construct such a framework.

Information on vegetation change and restoration often appears as reports, notes, photographs and knowledge held by individuals. It is rarely compiled despite its usefulness for evaluation of past and future restoration activities. I reviewed historical information available for north-west Victoria and applied the S-T framework to construct a S-T model of Pine and Pine-Buloke Woodlands of Hattah Kulkyne National Park (HKNP). The historical information provided examples of vegetation communities and the conditions and treatments which resulted in favourable, unfavourable or no vegetation change. Restoration failures and gaps in the knowledge base were highlighted because these are important for learning (Redford and Taber 2000; Adams et al. 2002).

My objectives were to:

1. Present a widely applicable framework for a state-and-transition model for arid areas,

2. Review historical information on the management of HKNP, north west Victoria and use the S-T framework and historical information to construct a conceptual model of vegetation change in the Pine and Pine Buloke woodlands of HKNP, and

- 41 - Chapter 3 State-and-transition modelling

3. Use the model to highlight opportunities for management and identify gaps in the knowledge base.

3.2 Development of models to describe semi-arid vegetation 3.2.1 Range succession models

The range succession model (Clements 1916; Dyksterhuis 1949) is the earliest model of rangeland vegetation and is dominated by deterministic, Clementsian succession (Figure 3.1). It describes how a community, in the absence of disturbance, travels along a single continuum to a climax of perennial species. In rangelands, heavy grazing and drought lead to retrogression along the continuum to annual or short-lived species, whilst above-average rainfall or release from grazing accelerate successional tendency. There is a theoretical balance where grazing and successional tendencies hold the vegetation at a desirable equilibrium.

Autogenic succession, good rainfall

Annuals Perennials (Poor Vegetation (Good condition) condition)

Disturbance (drought, grazing)

Figure 3.1 Clementsian succession showing slow autogenic succession to perennials and rapid retrogression to annuals (based on Walker 1997). This model was criticised because it underplayed the importance of infrequent, unpredictable climatic events and could not account for all the observed vegetation dynamics of the rangelands, including persistent weeds and severe soil erosion (Briske et al. 2003, 2005). Vegetation change in rangelands was found to be more discontinuous, irreversible and unpredictable than Clementsian succession allowed, with decline and recovery following different paths (Westoby et al. 1989; Friedel 1991).

3.2.2 State-and-transition models

The state-and-transition model (Westoby et al. 1989) de-emphasised the orderly process of community development (Figure 3.2). The S-T model describes multiple, stable vegetation communities of varying species composition (“states”), separated by abrupt “transitions”. Transient states are included as seral or successional communities which do not persist indefinitely. Transitions between any state could occur according to mechanisms such as demographic inertia, overgrazing, altered physico-chemical soil conditions, fire regimes, stochastic effects, or priority in competition (Westoby et al. 1989). The model describes states and transitions and summarises the current position of knowledge for each, including how climate and/or management could produce desirable or undesirable change.

- 42 - Chapter 3 State-and-transition modelling

Grassland, Grassland, many Dense shrub cover, scattered woody shrub seedlings little grass shrubs

Natural events, management or both Recently burnt, many shrub seedlings or resprouts

Figure 3.2 State-and-transition model which describes the vegetation as persistent states (solid boxes) or transient states (broken boxes) separated by abrupt transitions (arrows). Transitions occur following natural events, management practices or both (based on Westoby et al. 1989). The concept of thresholds was introduced to emphasise the importance of transitions which were not reversible on a practical time scale (Friedel 1991) or through ecological resilience (Briske et al. 2005). They were used initially to explain why some vegetation did not return to its former state when grazing was removed (Laycock 1991; van de Koppel et al. 2002). Crossing a threshold involves the extensive alteration of an ecological process which can only be restored with active management (Stringham et al. 2003). Thresholds were attributed to feedback loops which reinforced the altered state. For example, high grazing pressure kills grasses, allowing for invasion of less palatable shrubs and soil erosion. Shrubs feedback through the hydrological cycle preventing the establishment of grasses and maintain the vegetation in the degraded state even when grazing pressure is lowered (Noy-Meir 1975; van de Koppel et al. 2002). Feedback loops are most difficult to overcome when it is connected to altered nutrient and water cycles and energy flow pathways (Ludwig et al. 1997; Scheffer and Carpenter 2003; Suding et al. 2004).

Many S-T models include seral stages as vegetation states and describe all shifts between states as transitions (e.g. Westoby et al. 1989; Yates and Hobbs 1997; Allen-Diaz and Bartolome 1998; West 1999). This has led to considerable confusion regarding the reversibility or otherwise of transitions because reversible shifts driven by successional tendencies are not distinguished from the crossing of thresholds. Recently, proponents of the S-T model have focused on clarifying definitions of states, transitions and thresholds (Bestelmeyer et al. 2003; Stringham et al. 2003). I have adapted the following definitions from Stringham et al. (2003):

• States are stable vegetation and soil domains always separated by thresholds. They are defined by the ecological process which has altered, e.g. loss of component species. States encompass a certain amount of variation manifest as phases within states. • Thresholds are boundaries in space or time between states indicating that one or more primary ecological processes have been extensively altered. Thresholds are also defined by the process that has altered.

- 43 - Chapter 3 State-and-transition modelling

• Transitions describe the causal factors involved in crossing a threshold. Transitions involve a decline in vegetation condition whilst reverse transitions are an improvement driven by active restoration such as erosion control or seeding. • Phases represent natural plant community dynamics within states; also termed seral stages. • Phase shifts occur between phases within a state. They do not involve thresholds and are relatively easily reversible through modified grazing pressure or favourable climate. These clearer definitions of states, transitions and thresholds represent a fusion of the range succession model of phase shifts within states, and the non-equilibrium model of alternative stable states separated by thresholds (Fernandez-Gimenez and Allen-Diaz 1999; Briske et al. 2003).

3.2.2.1 Managing-within-states model and phases One particular drawback of the early S-T model was its emphasis on infrequent and unpredictable events such as episodic recruitment of perennial species in rangelands, devaluing the importance of continuous management (Watson et al. 1996; Watson et al. 1997a, 1997b; Briske et al. 2003; Wiegand et al. 2004). For managers, the apparent “dependence” of rangeland species on above- average rainfall for recruitment appeared to suggest a “do nothing” approach in average rainfall years, or even that stocking rates were irrelevant to management in systems with extreme climatic variability from year-to-year (Fernandez-Gimenez and Allen-Diaz 1999; Illius and O’Connor 1999). It is now recognised that phase shifts within vegetation states influence the development of impending thresholds and observation and experience are important to generate these important small changes (Scheffer and Carpenter 2003; Briske et al. 2005). This was clearly expressed in an earlier model, the managing-within-states model (Figure 3.3) developed by Watson et al. (1996).

Continuous, good management (“conditioning”: Watson et al. 1996) maintains vegetation within a functional, resilient state (Briske et al. 2005). Resilience is the ability to return to a former state after being negatively impacted upon by an outside disturbance (Connell and Slatyer 1977). This can prevent an event outside the manager’s control (e.g. drought), from forcing vegetation across a threshold into a degraded state. A practical example of this is conservative stocking rates and a sophisticated rotational grazing system practiced on the edge of the Namib Desert. Paddocks are rested frequently and destocking is undertaken quickly following poor rainfall to prevent irreversible, grazing-induced loss of vegetation. This process ensured the viability of the farm over a 200 year simulation when all other strategies would have resulted in ecologically and economically degraded land (Jeltsch et al. 2001).

Watson et al. (1996) also demonstrate how “conditioning” could facilitate or accelerate a reverse transition across a threshold from a degraded to a desirable state. The simplest example of this is restoration of vegetation communities (Whisenant 1999). If reserves of seeds or individual plants are greatly depleted, a production pulse will be small, regardless of the size of the rainfall event

- 44 - Chapter 3 State-and-transition modelling

(Milton et al. 1994; Hodgkinson and Freudenberger 1997; Pickup et al. 1998; Chesson et al. 2004). However, a manager can increase the plant reserves by seeding an area or by reducing grazing pressure during plant growth or reproduction periods and achieve a better response following rainfall.

High rainfall event

Good rainfall Good management

Degraded condition Transitional state Good condition

Poor management Low rainfall

Drought

Figure 3.3 Author’s interpretation of managing within states, an early, more incremental view of the state-and-transition model. Extreme climatic conditions can result in rapid transitions between alternative states (large arrows). However, management can lead to incremental phase-shifts within a state (small, broken arrows) and, combined with climatic conditions, can facilitate the transition of the vegetation to alternative state (small solid arrows) following smaller climatic events. Management-induced phase-shifts make the transition between states less abrupt and more likely to occur.

3.3 Developing the framework of a state-and-transition model 3.3.1 The framework

State-and-transition models have been developed for many vegetation types and purposes (e.g. Archer et al. 1988; George et al. 1992; Huntsinger and Bartolome 1992; Ash et al. 1994; Milton et al. 1994; Yates and Hobbs 1997; Allen-Diaz and Bartolome 1998; Vayssieres and Plant 1998; Hill 2002). Whilst there is considerable variation in the definitions used and vegetation assessed by each modeller, there are important similarities amongst models which suggest a single framework could be used for model development. I re-evaluated the transitions described in published S-T models (listed above) and the review of Iglesias and Kothmann (1997) to determine in each case the threshold being crossed. It became clear that within the rangelands there are very few thresholds and consequently few, true vegetation states. The first step of degradation is often the loss of component species, followed by dominance of inappropriate plant species if degradation continues. Both of these are biotic thresholds. Further degradation sees impacts on the soil (abiotic thresholds) such as loss of soil structure, altered soil nutrition or altered soil hydrology (Whisenant 1999). Similarly, there are a limited variety of possible transitions and

- 45 - Chapter 3 State-and-transition modelling reverse transitions for rangeland vegetation (the causal factors involved in crossing a threshold). The reviews of Iglesias and Kothmann (1997) and Westoby et al. (1989), supplemented by restoration texts such as Whisenant (1999) and Vallejo et al.(2006) provide a useful summary of the common transitions and available tools to encourage reverse transitions (Table 3.1). The specific transition involved, like the vegetation phases within each state, can only be determined with reference to the target vegetation.

The states and thresholds were easily structured into a logical framework (Figure 3.4). The physical organisation of the framework also conveys important information. The vegetation condition declines overall from top to bottom, and within each state the position of the vegetation phase reflects its condition. The phases lower down are degraded (high grazing and / or drought) and at risk of making a transition to a more degraded state. Reverse transitions are only likely to be successful from phases higher up: conditioned phases (low grazing and / or favourable rainfall). For example some palatable species should be present, indicating that grazing pressure is sufficiently low for reintroduction of component species to succeed.

3.3.1.1 Phases and phase shifts Phases are seral vegetation communities within a defined state. Phase shifts are easily reversible changes to vegetation brought about by small climatic fluctuations or altered grazing pressure.

Phases are not described in the framework because they can only be determined with reference to the target vegetation. Within each vegetation state, there are as many phases as are necessary to describe the reversible changes which occur as a result of grazing and climatic fluctuations. Whilst small climatic fluctuations result in phase shifts, extreme rainfall events, seasons or years can induce change sufficient to cross a threshold.

- 46 -

Chapter 3 State-and-transition modelling

Table 3.1 The key thresholds for rangeland vegetation and the common transitions and reverse transitions across abiotic and biotic thresholds, based on (Westoby et al. 1989; Iglesias and Kothmann 1997; Whisenant 1999; Vallejo et al. 2006). Threshold Transition Reverse transition

Loss of component 1. Grazing pressure, absence of rare event required for 2. Direct seeding, hand planting, especially during a species recruitment, absence of propagules due to loss of La Nina cycle (Holmgren and Scheffer 2001). Use landscape connectivity, altered fire regimes, long nurse plants or guards to protect from browsing. term change, e.g. climate change, increase in atmospheric nitrogen or rising water tables.

Dominance of 3. Overgrazing favoring less palatable species which are 4. Herbicides, fire, mechanical removal. inappropriate better competitors, altered fire regimes, soil species disturbance favoring early colonisers. Competitive exclusion through use of soil moisture

Changes to soil 5. Trampling and loss of vegetation and soil fauna cause 6. Address erosion: ripping, dune reshaping, rilling, structure, compaction, erosion, loss of soil microbes and erosion bunds, erosion matting, cover crop or hydrology or symbiotic relationships, decrease infiltration rate and carefully designed wind-breaks. Increase soil chemistry water storing ability of soils and reduce obstructions moisture: shading with mulch, water harvesting, (patches of vegetation) necessary to capture rainfall, supplementary watering. Increase nutrients: add nutrients and seeds There may be allelopathic effects fertiliser, soil symbionts, top soil. Increase of inappropriate species or rising water tables leading patchiness using piles of branches to trap leaf litter to salinity. creating fertile patches.

- 47 -

Chapter 3 State-and-transition modelling

Alternative Historic Near historic Restored DESIRABLE community community community community STATES

1 1 1 Loss of component species 2 2 2

Dominance of Successful Slightly degraded Successful inappropriate species Phases of Phases of reintroduction Moderate reintroduction of tolerant of desirable BIOTIC species species STATES Minor Phases of 4 degradation Phases of Poor Poor dominance of 3 inappropriate Severe species Major

5 6 5 Loss of soil Figure 3.4 S-T framework of vegetation change 6 functioning from desirable states through loss of component species and invasion of inappropriate species to loss of soil functioning. States (double lined Structure Chemistry Hydrology boxes), thresholds (double line), phases (boxed text) state state state and phase shifts (broken arrows) are represented. ABIOTIC Transitions (red arrows) and reverse transitions STATES (green arrows) are numbered according to Table 3.1. - 48 - Chapter 3 State-and-transition modelling

3.3.1.2 Desirable vegetation states Desirable vegetation types are defined by the goals of the restoration activity.

Desirable vegetation states include the historical vegetation type, and because restoration to this state is likely to be impossible (Hobbs and Harris 2001; Oliver et al. 2002; Choi 2004), the desirable, restored state. Historical ecology (Wasson and Clark 1985) involves primarily qualitative records (e.g. diaries, photographs, stocking rates, survey maps) which can be used to describe the historical vegetation state (Swetnam et al. 1999; Foster 2000). The restored state should aim to meet desires and needs in the future (Davis and Slobodkin 2004) and have sufficient biotic and abiotic resources for development to continue without further assistance (SER 2002). Where permanent damage is done to ecological functioning, such as species extinctions or dryland salinity, only certain vegetation outcomes are possible; these are termed alternative states (Hobbs et al. 2003). For example a woodland overcome by dryland salinity cannot ever return to being the historical woodland but a good cover of salt-tolerant perennials may be a desirable, alternative vegetation state. Similarly, whilst restoration of some areas may be technically possible, the costs may be prohibitive (Hobbs et al. 2003; Choi 2004). Desirable vegetation states should be self-sustainable, resistant to invasion of inappropriate species, demonstrate adequate cycling of nutrient, water and biotic reserves and possess optimal composition and abundance of species.

3.3.1.3 Biotic vegetation states Biotic vegetation states are defined by (1) the absence of key, component species, or (2) the dominance of inappropriate plant species.

“Biotic” vegetation states result from the “desirable” state crossing a biotic threshold into states defined by (1) missing component species or (2) dominance of inappropriate plant species. Inappropriate plants species are often weeds but may be native species which do not normally dominate the target vegetation community, e.g. grasslands where woodlands once were. Often these altered habitats have intrinsic conservation and habitat values and the desire and resources required to return vegetation to the woodland state must be carefully weighed against preserving this new vegetation state (Oliver et al. 2002). Controlling dominance by inappropriate species does not return the vegetation to a desirable state but rather to the state defined by missing component species. These species must be reintroduced before the desirable state is reached.

3.3.1.4 Abiotic vegetation states Abiotic vegetation states are defined by (1) altered soil structure, (2) altered soil hydrology , or (3) altered soil chemistry.

- 49 - Chapter 3 State-and-transition modelling

Abiotic vegetation states result from biotic states crossing one of the three thresholds defined by (1) altered soil structure preventing the roots of desirable seedlings from penetrating, (2) altered soil hydrology (reduced soil moisture) preventing the seedlings of desirable species from establishing, or (3) altered soil chemistry preventing the seedlings of desirable species from surviving. These states are less clearly delineated because it is difficult to determine which particular threshold has been crossed and is preventing restoration. Often more than one threshold is crossed simultaneously and it may be convenient to treat the altered states simultaneously. Transitions will only occur from abiotic states to biotic states; a transition from an abiotic state to a desirable state is not possible (Aronson et al. 1993; Tongway 2003). This is illustrated in Figure 3.5 which is an alternative representation of vegetation transitions between states (Hobbs and Harris 2001).

Biotic transition Abiotic transition threshold threshold Fully functional Recovery requires active, 1 2 abiotic restoration, 3 e.g. soil 4 remediation Level of Recovery ecosystem requires Recovery function passive biotic requires active restoration, biotic 5 e.g. reduce restoration, 6 grazing e.g. replanting Non- pressure desirable functional species

Intact Ecosystem state Degraded

Figure 3.5 Conceptual model of vegetation change from intact to degraded states after crossing biotic and abiotic thresholds (from Hobbs and Harris 2001). 3.3.2 From framework to model

The S-T framework can be customised to most rangeland ecosystems by describing vegetation communities and fitting them as phases in the predetermined states. Vegetation communities can be determined subjectively using the knowledge of land managers (Yates and Hobbs 1997; Bestelmeyer et al. 2003) or field surveys can be used for a more data-driven approach. Field surveys use spatial variation at one point in time to represent all possible variation in vegetation communities (Vayssieres and Plant 1998). The vegetation is grouped into ecological sites with similar abiotic factors (underlying geology, climate, soil type, slope, topographic position and aspect). This is necessary to distinguish whether different vegetation communities are the result of a transition or due to differing environmental conditions (Friedel et al. 1993; Bestelmeyer et al. 2003). Some form of clustering algorithm based on species composition is then used (e.g. supervised clustering) to define vegetation communities, according to, for example, the relative

- 50 - Chapter 3 State-and-transition modelling proportions of trees, shrubs, and ground cover (e.g. Friedel 1991; Allen-Diaz and Bartolome 1998; Vayssieres and Plant 1998; Jackson and Bartolome 2002).

Phase shifts, transitions and reverse transitions should be defined using long-term datasets or photopoints showing vegetation change, consulting experts in the field or through experimental manipulation of vegetation, e.g. exclosures, introduction of propagules, soil remediation works (Yates and Hobbs 1997; Vayssieres and Plant 1998; Bestelmeyer et al. 2003).

3.3.3 Defining vegetation communities

3.3.3.1 Indicators Explicit definitions of communities are critical for the functioning and utility of a S-T model (Briske et al. 2005), to allow recognition of approaching thresholds (Bestelmeyer et al. 2003) and provide measures of when “successful restoration” has been achieved (Hobbs and Norton 1996). Indicators such as the composition and structure of plant communities can be used to define communities efficiently (Landsberg and Crowley 2004). Indicators are easily acquired measures or attributes that relate to basic processes of landscape condition and functioning (Ludwig et al. 2004). Plants directly reflect the level of damaging processes such as grazing, clearing, weed invasion, particularly in rangelands (Landsberg and Crowley 2004). Plant community composition and structure can act as a surrogate for elements which are more difficult to measure. For example the landscape functional integrity (the ability of the landscape to retain and not leak water and nutrient resources) can be defined by vegetation patch quantity and quality (Ludwig et al. 2004). There is also increasing evidence that vegetation structure or “habitat complexity” is an important determinant of the impact of introduced predators on native fauna (e.g. Arthur et al. 2003).

3.3.3.2 Community structure Indicators of plant community structure include density of large trees, tree canopy cover, understorey structure, weediness, recruitment, organic litter and abundance of logs. These attributes are part of the Victorian Habitat Hectares Assessment (Parkes et al. 2003) and are broadly supported as measures of vegetation condition (McCarthy et al. 2004). These attributes can be used to describe vegetation communities either quantitatively using the Habitat Hectares scoring method (DSE 2004), or by qualitative descriptions of each attribute.

The Habitat Hectares Assessment compares native vegetation with a benchmark representing the average characteristics of mature and apparently long-undisturbed remnants of the type of native vegetation predicted to have occupied the site before agricultural development. Whilst the benchmark values of these components should not be interpreted as a ‘prescribed ideal’ or a ‘climax’ state (DSE 2004), the benchmark values are useful indicators of vegetation in a desirable state.

- 51 - Chapter 3 State-and-transition modelling

3.3.3.3 Community composition Whilst constructing complete species lists of vegetation communities may not always be possible, it is important to recognise the characteristics of the species present. Species composition will depend on the degree of degradation the community underwent prior to restoration. Some species are easily reintroduced to an area via direct seeding or hand-planting, some cannot be reintroduced and require natural dispersal into the area and some species are particularly sensitive to abiotic conditions (e.g. salt or poor soil structure) and are unlikely to establish in areas which were previously highly degraded.

Successful restoration occurs when sufficient biotic and abiotic resources are present for the community to continue its development without further assistance (SER 2002). The presence of species with a low reintroduction potential indicates natural dispersal and colonisation is occurring and restoration is approaching completion. Dominance of species which are easily reintroduced indicates restoration is in the early stages. Occurrence of only hardy colonising species may indicate that abiotic functioning is not yet restored. Key texts to determine such indicator species for north-west Victoria include Cunningham et al. (1992), Ralph (1999), Bonney (2003) and Grant (2004).

Species composition also depends on the current pressures on the vegetation, namely grazing pressure. Historically, grazing pressure in Australia was fairly light. As a consequence, much of the flora is poorly equipped to survive the comparatively high levels of continuous grazing present under current stocking rates (Landsberg et al. 2003). The relative proportions of decreasers, increasers and invaders are useful descriptors of community composition. Decreaser species decline where grazing impacts are concentrated and tend to be palatable, drought-hardy perennial species. Increasers and invaders tend to be annuals and are often exotic and generally unpalatable (Illius and O’Connor 1999; Friedel et al. 2003; Landsberg and Crowley 2004), however, predicting grazing response from simple traits is difficult or impossible (Landsberg et al. 1999; Vesk et al. 2004a). As an approximate guide, rangelands in excellent condition contain (cover) 50-75% decreasers, 0-25% increasers and no invaders, good condition is 25-50% decreasers, 25-50% increasers and 0-25% invaders, fair condition is 0-25% decreasers, 25-50% increasers and 25-50% invaders and poor condition is no decreasers, 0-25% increasers and 50- 100% invaders (Allen-Diaz and Bartolome 1998).

Examples of increasers and decreasers can be obtained from the reviews of Vesk and Westoby (2001) and Westbrooke et al. (2001) and more recently published papers (Landsberg et al. 2002; Landsberg et al. 2003; Landsberg and Crowley 2004; Vesk et al. 2004a). Ground layer species generally exhibit the clearest trends (Landsberg et al. 2003). Caution must be exercised when extrapolating these studies which are predominantly from other regions, to north-west Victoria. Species may have dissimilar responses in different areas, depending upon the palatability of co-

- 52 - Chapter 3 State-and-transition modelling occurring species (Landsberg et al. 1999; Vesk et al. 2004a). Indeed 41% of 324 species were classified as inconsistent responders (Vesk and Westoby 2001).

3.4 Developing a model for semi-arid, north-west Victoria 3.4.1 Overview

The vegetation change in the semi-arid woodlands of HKNP has been described as “a remarkable story of achievement in habitat rehabilitation – perhaps the most dramatic turn around in semi- arid habitats in Australia” (Cheal 2005a). I applied the S-T framework described above to the Pine and Pine Buloke Woodlands of HKNP and constructed a conceptual model of vegetation change (Figure 3.6). This community was chosen because considerable historical information was available. The states, phases, transitions and phase shifts between each community are described with reference to actual sites within HKNP from historical information detailed in photographs and reports describing vegetation, management actions, climatic conditions, grazing pressure and restoration attempts. I also consulted with persons involved in restoration programs and revisited sites, sometimes tens of years after works were conducted (Murdoch and Murdoch 2006). Restoration failures and gaps in the knowledge base were highlighted because these are important for learning (Redford and Taber 2000; Adams et al. 2002). This preliminary model is intended to promote experimental tests of hypothesised transitions and thresholds, rather than provide definitive guidance (Yates and Hobbs 1997).

The desirable states include the historical community, the benchmark community (sensu White et al. 2003) and the restored community. In addition, I considered three altered states to be present at HKNP (Figure 3.6) defined by the thresholds referred to in the framework (section 3.3.1). The first threshold is defined by missing component species and the vegetation state includes four phases, the second threshold is defined by dominance of inappropriate species, the threshold appears twice to distinguish less persistent species from invasive, often perennial weeds. In total there are four phases in this state. The third threshold is defined by the loss of abiotic functioning and has one phase. Not all phases are described because this would be cumbersome and confusing. Rather, I described the most important and obvious vegetation phases to make the model most useful for managers. A model should be a simple system of generalisations or assumptions about ecosystem function (Westoby et al. 1989) and as such is merely an imperfect representation of reality (Scheffer and Carpenter 2003).

- 53 -

Chapter 3 State-and-transition modelling

Historic community Restored woodland Benchmark woodland

Missing component species 8 1 Regenerating woodlands

Dominance of inappropriate species II II

Partially Low diversity regenerating woodland 6 or 7 woodland Perennial III grassland I Degraded woodland Hopbush IV 2 III Annual grassland Figure 3.6 State-and-transition model for semi-arid woodlands in HKNP. The double lines represent thresholds and enclose vegetation states. Boxes represent vegetation communities which are phases within vegetation states. Numbered arrows are transitions explained in the text (green: improvement, 5 2 3 red: decline). Broken arrows are reversible phase shifts driven by rainfall and grazing pressure. Persistent 4 weeds

Loss of soil abiotic functioning

- 54 - Eroded dune Chapter 3 State-and-transition modelling

3.4.2 Transitions

The full progression of vegetation change and European impacts since settlement was described in section 2.3 and is not repeated here. The specific transitions between the vegetation states are summarised below from that information.

1. Extensive thinning of the overstorey through logging, ring-barking, wind-throw or senescence converts woodlands to savannah woodland. Heavy grazing by domestic stock (early), rabbits and kangaroos (later) and extended drought destroys the majority of palatable understorey species, leaving primarily unpalatable species. Recruitment of desirable species is prevented by grazing, even following rainfall.

2. Continued heavy grazing removes all understorey cover including the cryptogam layer and combined with drought, results in extensive bare areas. Soil in bare areas is further disturbed by rabbit burrows or scratches (Eldridge and Simpson 2002), kangaroo scrapes (Eldridge and Rath 2002) or previously by domestic stock (Lange 1969; Landsberg et al. 2002) and vehicles. Bare and disturbed areas are invaded by annual and perennial weed species.

3. The replacement of native perennial species with annual exotics means that during dry summers or following heavy grazing, vegetation cover is removed completely. Bare, sandy areas are exposed to erosive winds and metres of soil are removed.

Fire is not included as a transition although a high intensity fire kills Casuarina pauper, Allocasuarina luehmannii and Callitris gracilis woodlands (LCC 1987). In 1981, a large wildfire converted a Buloke Woodland in Wyperfeld National Park to an open woodland / grassland (Raymond 1990). However fire occurs rarely in these woodlands because there is generally insufficient understorey to carry it. There have been two recorded fires in Buloke woodlands in HKNP, the first in 1980 (Kulkyne Tk, 631,223E 6,164,822N), the second in summer 1997-8 (Chalka Ck Tk 630,000E 6,157,300N). The 1980 fire destroyed a number of trees but the 1997-8 fire destroyed very few trees because despite a good growth of grasses, it was of insufficient intensity (pers. obs.). In both fires, many trees which were burnt did not die and in both cases the fire did not alter the overall classification of the vegetation to a grassland dominated state. Consequently I have not included fire as a transition in HKNP.

3.4.3 Reverse transitions

Reverse transitions or “opportunities” lead to an improved vegetation state and occur as a result of extensive management input, or occasionally an exceptional sequence of climatic events or extended periods (tens of years) of low grazing pressure. Failure of any of these techniques leads to a reversion to the previous state, as indicated by double headed arrows in Figure 3.6. The

- 55 - Chapter 3 State-and-transition modelling reverse transitions are summarised below and use of these techniques at HKNP is described. More detailed information, including the location of many of these sites are described in the following sections and in Murdoch and Murdoch (2006).

4. Eroded dunes may be restored to annual grassland firstly by extensive dune reshaping with a tractor or bulldozer to alter erosive wind flow patterns. Dune reshaping has been used in the 1970s and 1980s on high priority sites, such as halting sand drift across main vehicle tracks. In the 1970s, this was followed by permanent fencing to reduce grazing pressure, fertilising and establishing annual, sterile ryecorn for two years, followed by hand-planting tubestock into the stubble (Transition 6). More recently, the sites have been left to regenerate naturally after the establishment of ryecorn. In the 1990s, some sites were stabilised with erosion matting or brush piles. The effectiveness of using brush piles for capture of resources such as litter and seeds, and for protection from browsing has been demonstrated (Ludwig and Tongway 1996). Brush piles of Dodonaea viscosa cut during track clearing were used to protect Glycine canescens from browsing, to close tracks and repair unstable dune areas. Some restoration texts recommend the reintroduction of microsymbionts although this has not been trialled at HKNP.

5. Very invasive weeds can be controlled by application of herbicides using a boomspray for large areas or a knapsack spray for smaller infestations. Fire may be useful over large areas to destroy the plants and seeds of certain species e.g. Horehound Marrubium vulgare. Few broadscale techniques other than fire are available for the control of Ward’s Weed Carrichtera annua (Cooke 2003). Hand-pulling may be appropriate for localised infestations without persistent root systems, such as the recent, isolated infestations of Crownbeard Verbesina encelioides in HKNP. Slashing Saffron Thistle Carthamus lanatus prior to flowering, whilst not used at HKNP, has been recommended (Cunningham et al. 1992) and may be useful for other species. Improved slasher design can substantially reduce the spread of weeds (Erakovic et al. 2004).

6. Low diversity woodlands can be created by hand-planting tubestock of desirable species into highly degraded areas. A small number of species are used that are good colonisers and establishment is assisted by protection from browsing (either individual tree guards or fences) and in one case, supplementary watering from Lake Mournpall. This technique is far more labour intensive than direct seeding (Transition 7) and has only been used extensively in the 1970s and more recently by volunteer groups on sites of less than 1 Ha.

7. Direct seeding has been used since 1989 in HKNP to establish desirable species without supplementary watering and often without protection from browsing. Adequate site preparation includes ripping and use of a knock-down herbicide to facilitate establishment

- 56 - Chapter 3 State-and-transition modelling

and control competition from annual species. The seed mix is generally comprised of large shrubs and trees Allocasuarina luehmannii, Callitris gracilis, Acacia ligulata, A. brachybotrya, Senna artemisioides, Hakea tephrosperma and more recently A. rigens

8. The transition from regenerating woodlands to the desirable state (restored woodlands) is unknown at HKNP. It may occur naturally over a long time period or it may require further supplementation of desirable species. Artificially enhancing the diversity of understorey species requires assessment of the site to determine missing species and follow-up direct seeding. Hand-planting may be useful for some species that are not able to be direct seeded. This transition is likely to require repeated treatment before the site is restored to a self-sustaining restored woodland.

3.4.4 Desirable states

3.4.4.1 Historical vegetation state Some idea of the structure of pre-European Australian semi-arid woodlands can be gleaned from the maps of early surveyors and from notes made by naturalists (Lunt 2002). The original survey and parish plans dating back to 1860 and contain useful information on soil types, vegetation, timber resources and flood frequencies, however plant species and vegetation structure are often difficult to interpret because the terms used varied between surveyors (White et al. 2003; Callister 2004). As other researchers have found, descriptions of the vegetation which existed soon after European settlement are lacking in detail regarding the understorey elements and quantitative analysis of the composition. In north-west Victoria, survey maps and the records of early explorers and settlers are largely qualitative, especially between 1830 and 1920, and their interpretation is subjective (White et al. 2003).

My review of the early literature (presented in section 2.3) succeeded in adding little more information than is present in existing summaries of the historical vegetation type (e.g. Callister and Westbrooke 2002; Macaulay and Westbrooke 2003; White et al. 2003; Callister 2004). These authors have provided lists of key species likely to be have been present in the historical vegetation community. Whilst the composition of the historical overstorey are adequately described for semi-arid woodlands, the understorey is less so:

“Disturbance has been ubiquitous and notions of what this EVC [vegetation community] may have looked like in an “unmodified” state are unreliable … Relatively intact remnants are quite shrubby with a suite of tall shrubs to 3 m and an open low shrub stratum above a relatively sparse field layer … The composition of the ground flora is poorly known due to the degradation of this EVC across its range in Victoria” (White et al. 2003).

- 57 - Chapter 3 State-and-transition modelling

3.4.4.2 Benchmark Semi-arid Woodland The “benchmark” community (sensu White et al. 2003) is described from some small reserves, roadsides and occasional remnants on private land. Typically, these areas are small and generally in higher rainfall areas to the south of the study site, e.g. Barrett Flora and Fauna Reserve (100 Ha SW of Warracknabeal), Glenlee Flora and Fauna Reserve (120 Ha near Nhill), and private remnants in Murtoa, Culgoa, Dadswell Bridge and Quambatook (Cheal and Lucas 2005). Consequently, the benchmark community is not described as a restoration goal for HKNP, although it is described here because it forms a useful reference for describing other vegetation states and phases.

Density of overstorey trees varies with a benchmark condition of 20 large trees per hectare (DBH > 40cm). Annotations from survey maps indicated that historically open woodlands (<50 tree per hectare) were far more common than dense woodlands (>300 trees per hectare:Callister 2004). Trees from a variety of size classes are present.

Below the canopy are tall shrubs to 3 m (benchmark: five species and 15% cover) and an open low shrub stratum (benchmark: five species and 20% cover) above a relatively sparse field layer (benchmark: 14 species and 25% cover). Benchmark conditions of 20% cover of organic litter and 20 m of logs per 0.1 hectare are present and there should be evidence of recruitment of all life forms (DSE 2004).

Species with low reintroduction potential should be present and include shrubs such as Exocarpus aphyllus and Alectryon oleifolius which are difficult to grow from seed and appear to be bird- dispersed (Ralph 1999) and Acacia oswaldii which is extremely slow growing from seed (pers. obs.). These shrubs can be quite long-lived, persisting even in degraded communities however young individuals indicate that natural dispersal and colonisation are operating. Smaller shrubs and herbs with low reintroduction potential include Ophioglossum spp., Actinobole uliginosum (following favourable winter rains), Wahlenbergia spp. and annual species such as Brachyscome perpusilla. Grasses such as Enneapogon avenaceus and Austrodanthonia caespitosa also have low reintroduction potential because they are not usually part of reintroduction programs in semi- arid woodlands of north-west Victoria, despite being relatively simple to sow (Ralph 1999; Bonney 2003). Rangelands in “excellent condition”, such as Benchmark semi-arid woodland, should have at least 50% cover of ground layer species which indicate low grazing pressure (decreasers) such as those in Table 3.2 (Allen-Diaz and Bartolome 1998).

- 58 - Chapter 3 State-and-transition modelling

Table 3.2 Decreaser species which occur in Victorian Pine-Buloke woodlands and are indicative of the benchmark community. Scientific name Common name Family Comments References

Atriplex Old man saltbush Chenopodiaceae Declines under heavy 1, 4 nummularia grazing, recovers well

Austrostipa Elegant Poaceae Grows unprotected in 1, 4 elegantissima speargrass ungrazed situations

Calandrinia Small purslane Portulaceae Winter-spring 1, 5 eremaea growing, palatable

Convolvulus Blushing Convolvulaceae 5 erubescens bindweed

Dactyloctenium Finger grass Poaceae Ephemeral, responds 1, 5 radulans to summer rain

Daucus Austral carrot Apiaceae Winter-spring 1, 3, 5 glochidiatus growing, readily eaten

Einadia nutans Climbing Chenopodaceae 5, 6 saltbush

Eremophila Tar bush Myoporaceae Presence indicates fair 4 glabra to good condition

Erodium Blue heron’s bill Geraniaceae Abundant following 3, 4, 5 crinitum winter rainfall and relished by stock

Maireana Pearl bluebush Chenopodiaceae Hardy plant but 4, 6 sedifolia absence indicates poor condition

Marsdenia Bush banana Asclepiadaceae Never common 5 australis

Rhyncharrhena Purple pentatrope Asclepiadaceae Relatively uncommon 5 linearis

Triraphis mollis Purple plume Poaceae 5 grass

Wahlenbergia Annual bluebell Campanulaceae Grows during cooler 1, 5 gracilenta months, good forage Sources 1: Cunningham et al. (1992), 2: Landsberg et al. (2002), 3: Landsberg et al. (2003), 4: Mitchell and Wilcox (1994), 5: Vesk and Westoby (2001), 6: Westbrooke et al. (2001).

- 59 - Chapter 3 State-and-transition modelling

3.4.4.3 Restored woodland The restored woodland is the goal of restoration activities in HKNP. Surprisingly given the resources focused on restoration, there is very little specific information describing the composition of desirable vegetation communities for the Pine and Pine Buloke Woodlands. HKNP is managed under the provisions of the National Parks Act 1975. The Mallee Parks Management Plan (DNRE 1996b) includes the resource conservation management aim:

“Maintain, or where possible enhance, wilderness values, and allow natural environmental processes to continue with the minimum of disturbance, and maintain biodiversity … Actively rehabilitate degraded communities to a level approaching pre-European settlement conditions” (DNRE 1996b).

These aims are very general and do not specify the desirable state of a restored semi-arid woodlands or how to determine when restoration success has been achieved. Restoring pre- European conditions is an unachievable goal (Oliver et al. 2002) because the vegetation composition is poorly known (White et al. 2003) and there have been many species extinctions, especially mammals (Wakefield 1966). The valued attributes are listed as wilderness values, natural environmental processes and biodiversity.

Presumably the restored woodland aims to be similar to the benchmark woodland described above and meets all the benchmark conditions, as well as the composition of species according to the range succession model (50-75% decreasers, 0-25% increasers and no invaders). However, overall species diversity is likely to be less and dominated more by easily reintroduced species (described in section 3.4.5.2).

3.4.5 Missing component species

3.4.5.1 Degraded woodland Much of the semi-arid woodland in HKNP is currently considered degraded (Gowans and Westbrooke 2002). These woodlands have been thinned although there are generally at least 20 trees per hectare. These tend to be from the mature size classes with few if any from smaller size classes. Stumps are numerous, especially of Callitris gracilis (Figure 3.7).

The woodland does not have the benchmark understorey structure, having a lower species diversity and lacking the smaller life forms. Palatable shrub species such as Maireana sedifolia and M. turbinata (Westbrooke et al. 2001) and perennial herbaceous species are heavily grazed or absent. Annual species and unpalatable shrub species such as Dodonaea viscosa and Senna artemisioides are dominant. The more palatable species such as Pittosporum phylliraeoides, Acacia oswaldii and Enchylaena tomentosa exist only as browsed, old individuals whilst Alectryon oleifolius, if present, shows a clear browse line. The ground layer contains

- 60 - Chapter 3 State-and-transition modelling predominantly native species of lower palatability including Aristida holathera, Austrostipa scabra, Sclerolaena obliquicuspis, Vittadinia spp. and Ajuga australis (Sluiter et al. 1997a), increasers (Table 3.3) or species which grow in bare areas of bare areas such as Rhodanthe moschata, Calotis erinacea, Salsola kali and Polycalymma stuartii (Cunningham et al. 1992). There is little if any recruitment of desirable species. Some annual or perennial exotic species are present, such as Brassica tournefortii and Silene spp. although weeds do not dominate (<50% cover). The quantity of organic litter is low resulting in increased bare areas susceptible to wind erosion, water (sheet) erosion and invasion by weeds. The benchmark quantity of logs is unlikely to be met, with most dead wood occurring as stumps.

Table 3.3 Increaser species indicative of the degraded vegetation state: Pine-Buloke woodland. Scientific name Common name Family Comments References

Brachyscome Hard headed Asteraceae Cool season annual 4 lineariloba daisy

Chenopodium Crested Chenopodiaceae Annual forb 3, 4 cristatum goosefoot

Enteropogon Windmill grass Poaceae Perennial 4 acicularis

Goodenia Cut leaf Goodeniaceae Cool season annual 2, 4 pinnatifida goodenia

Sclerolaena Grey copperburr Chenopodiaceae Not grazed by 1, 2, 4 diacantha kangaroos

Solanum esuriale Quenna Solanaceae Perennial 1, 4

Triptilodiscus Common sunray Asteraceae Cool season annual 4 pygmaeus

Sources 1: Cunningham et al. (1992), 2: Landsberg et al. (2002), 3: Landsberg et al. (2003), 4: Vesk and Westoby (2001).

- 61 - Chapter 3 State-and-transition modelling

Figure 3.7 An example of a degraded woodland - off Mournpall Tk (626,130E 6,166,180N), autumn 2002. Allocasuarina luehmannii trees are in the background with dead Callitris gracilis (right). There are few shrubs or understorey species present, except grasses including Aristida holathera. In the foreground are unvegetated bare areas exposed to wind erosion.

3.4.5.2 Partially regenerating woodland The partially regenerating woodland phase is similar to the degraded woodland in that it has at least 20 mature trees per hectare, although the size classes are only small or large, lacking the medium sizes. The benchmark understorey structure is not met, having a lower species diversity of primarily unpalatable species and some weeds (0-25% decreasers, <50% increasers, <50% invaders). Some of the rarer, palatable herbaceous species such as Swainsona phacoides, Glycine canescens, Marsdenia australis, Corynotheca licrota and Sida ammophila may be present. The most significant feature of the partially regenerating woodland is recruitment in some desirable, woody species and the loss of the browse line on mature shrubs. Recruitment is usually of the more common and resilient shrubby species, such as Acacia rigens, A. ligulata and A. brachybotrya which have a long-lived soil seed bank (Ralph 1999). Species such as Pimelea microcephala recover well when grazing is reduced (Westbrooke et al. 2001). Sometimes recruitment of Callitris gracilis and Allocasuarina luehmannii and root suckers of Hakea tephrosperma and Alectryon oleifolius are present where parent trees are nearby. The quantity of organic litter is still low in this woodland phase, although greater than in the degraded woodland. The benchmark quantity of logs is unlikely to be met with most dead wood occurring as stumps.

Phase shift I Partially regenerating woodland is a recent phase, being present in HKNP only in the last five to ten years, for example Figure 3.8. Vegetation studies between 1992 and 1996 (Sluiter et al. 1997a) recorded recruitment of H. tephrosperma and A. oleifolius primarily inside the Mournpall Block only (where culling of kangaroos commenced in 1990). Recruitment of Allocasuarina luehmannii and C. gracilis was not recorded although in some areas it occurs on

- 62 - Chapter 3 State-and-transition modelling studied vegetation transects, suggesting a post 1996 origin (section 4.5.2). Rainfall records (section 2.2.2) indicate that there have been few years since 1996 with extended periods of favourable soil moisture, however total grazing pressure has been reduced substantially and is probably the cause of this phase shift.

In 1986, exclosure plots were established to protect Sida ammophila and Swainsona phacoides from local extinction due to high grazing pressure (D. Cheal, Arthur Rylah Institute for Environmental Research, 1986, letter to Ranger, HKNP). These species are now observed frequently in partially regenerating woodlands in HKNP. Palatable (e.g. Triraphis mollis, Eragrostis setifolia, Enneapogon avenaceus) and less palatable grasses (e.g. Aristida holathera var. holathera and Austrostipa spp.) are also common. Aristida holathera now dominates many areas, despite not being recorded on the 1970 plant list for HKNP, or the 1983 revision and was only recorded at two location between 1992 and 1996 (Sluiter et al. 1997a). Similarly, T. mollis was recorded only once although it is now observed frequently. Clearly, revisiting the Sluiter et al. survey work to provide quantitative descriptions of change in ground flora would be beneficial (Cheal 2005a).

Areas of recruitment of woody, perennial species occur on a small spatial scale with substantial variation in recruitment occurring often over areas of less than 100 square metres. A. luehmannii particularly, often occurs as patches of recruitment surrounding a single mature tree. The incidence and spatial distribution, timing and factors involved in this recruitment are investigated in Chapter 4.

Figure 3.8 An example of a partially regenerating woodland – Kulkyne Tk (631,220E 6,165.650N), spring 2005. Some mature Allocasuarina luehmannii and Callitris gracilis and juveniles of both species as well as Acacia spp. are present. The native ground flora includes Aristida holathera, Calotis cymbacantha, Hibbertia virgata and Thysanotus baueri.

- 63 - Chapter 3 State-and-transition modelling

3.4.5.3 Regenerating woodlands There are few, if any woodlands in HKNP which are sufficiently diverse and self-perpetuating to meet this category. The regenerating woodland is similar to the restored woodland described above although the size classes of overstorey trees are still likely to be biased towards smaller classes indicating recent reintroduction or recruitment of these species. The woodland meets the benchmark for recruitment and litter and may meet the benchmark for logs. The species present are dominated more by easily reintroduced species than species with a low reintroduction potential. There are fewer decreasers and more invaders than the restored woodland (0-25% decreasers, <50% increasers, <50% invaders).

Phase shift II One example of a regenerating woodland in HKNP is the 1948 plot established by Forest Commission Officer Alf McLeod (Figure 3.9). This rabbit-proof exclosure was one of the first established by the Forest Commission of Victoria and is considered the oldest plot still in good condition (Moncrieff 1991). The plot was established to protect regenerating Callitris gracilis but also includes self-sown Allocasuarina luehmannii seedlings (Moncrieff 1991). The woodland is dominated by C. gracilis, and is best described as a Pine Woodland (Sluiter et al. 1997b). The understorey includes sparse Acacia spp. and native grasses; there are few weeds present. This phase shift is likely to have occurred in response to protection from browsing for 57 years, favourable rainfall periods within this time (notably 1954-1956 and 1973-1975) and a seed source from adjacent trees and shrubs.

Figure 3.9 An example of a regenerating woodland – the 1948 plot (621,639E 6,154,942N), spring 2005. The exclosure shows dense Callitris gracilis and a single Allocasuarina luehmannii seedling marked with an arrow. The sparse understorey typical of Pine woodlands is apparent.

- 64 - Chapter 3 State-and-transition modelling

3.4.5.4 Low diversity woodland The low diversity woodlands in HKNP are the result of numerous restoration efforts dating back to work by the Soil Conservation Authority in the 1960s. There is a diversity of forms (Pine woodlands, handplanted woodlands, direct seeded woodlands and root suckers), each of which occurs over a limited area (less than 5 Ha). Each woodland could be a separate phase however I have grouped them together for clarity in Figure 3.6.

There are few if any large trees >40 cm DBH (<20 per Ha), those that are present are of the smaller size classes indicating recent reintroduction. The woodland does not meet the understorey structure benchmark being dominated by easily reintroduced large shrubs and trees (examples below). There may be recruitment evident of these species, although not of other lifeforms. There are few if any logs reflecting the highly degraded origins of this phase. Organic litter is minimal but developing and may meet benchmark in Acacia shrublands. The ground flora tends to be predominantly increasers and invaders, although weeds do not dominate (25-50% decreasers, 25- 50% increasers and 0-25% invaders).

Phase shift II from this phase to a regenerating woodland has not occurred at HKNP and is likely to require extended periods with low rainfall and grazing pressure and may be dependent on an adequate seed source from adjacent areas. Limited areas, such as Pine plantations are showing some recruitment of a few species (namely Callitris gracilis). The following sections describe the characteristics of the low diversity woodlands at HKNP.

Pine woodlands were established in 1962-1980 by the Soil Conservation Authority (SCA) and are a handplanted monoculture of Callitris gracilis. At that time the open sandhills were covered wholly with annual herbaceous species without annual or perennial grass in any quantity (Moncrieff 1991). In some areas the trees established well (Figure 3.10), in other areas the trees developed poorly and remained stunted. In these areas it appears that competition from the C. gracilis and continual wind erosion prevents the establishment of shrubs or understorey.

- 65 - Chapter 3 State-and-transition modelling

Figure 3.10 Low diversity woodland (Callitris gracilis), north of Raak Tk, c. 1970s. The foreground is dominated by the weedy, native annual Polycalymma stuartii. There is no understorey under the handplanted C. gracilis and erosion is continuing under the trees. The C. gracilis are heavily browsed on the lower branches. Photograph: Parks Victoria. Handplanted woodland – The Friend’s Plantation was established in 1990 by the Friends of HKNP using Allocasuarina luehmannii, Callitris gracilis, C. verrucosa, Acacia ligulata, A. brachybotrya, Senna artemisioides and Hakea tephrosperma (Figure 3.11). Prior to planting the site was a grassland dominated by annual weeds. The site preparation involved extensive earthworks and dripper lines supplying water from Lake Mournpall. Hand-planted tubestock were protected with tree guards which were later removed, circa 1998, by which time the trees were established. Again the use of overstorey and large shrub species has meant that little understorey has been able to establish. However there are some signs of the plantation producing seed and seedlings of some species (mostly Acacia ligulata and A. brachybotrya) suggesting that the phase shift to regenerating woodlands may be possible.

The importance of protection from browsing is demonstrated by a 1987 trial. Hand-planting was conducted inside and outside a kangaroo proof exclosure, without supplementary watering (Johnson 1989; Moncrieff 1991). By 1989 there were no seedlings alive outside the exclosure, with a 40-60% survival rate of species inside. There appears to have been no monitoring since 1989 and when I revisited the site in 2005 the plants inside the exclosure were well-established. Other hand-planting has occurred in scattered locations, including a trial of long-rooted tubestock (Carrol Tubes: Anonymous undated) at the day visitor area at Lake Hattah in 2004 (Murdoch and Murdoch 2006).

- 66 - Chapter 3 State-and-transition modelling

Figure 3.11 Low diversity, handplanted woodland, spring 2005. Trees and shrubs (Callitris gracilis, Allocasuarina luehmannii, Acacia ligulata, Acacia brachybotrya, Dodonaea viscosa) established in 1990 in the Friends of Hattah plantation (623,200E 6,158,000N). Dripper lines are visible between the trees. Direct seeded woodland – There are many examples of direct seeding in HKNP since 1987 and usually the seed mix was dominated by the same overstorey trees and large shrub species used in hand-planting, with the recent addition of Acacia rigens. Direct seeding is usually conducted into prepared grasslands in autumn. Direct seeding into established Dodonaea viscosa has been suggested (Moncrieff 1991) but this may result in low success rates due to competition from established plants. The most successful direct seeded plants are generally Acacia spp. and Senna artemisioides with some areas dominated by A. brachbotrya (Figure 3.12) or A. ligulata. Direct seeding of Allocasuarina luehmannii often fails (Thomas 2003).

Successful direct seeding was conducted in 1987, with lower success in 1988 and 1989 within a kangaroo-proof trial area in the north of the Mournpall block (Johnson 1989; Moncrieff 1991). Further direct seeding occurred in 1991 on Boolungal Ridge (625,800E 6,160,400N) with successful establishment of a variety of species, especially S. artemisioides and Acacia spp. and the north-west aspect ’84 Dune (623,600E 6,157,600N) with successful establishment of C. gracilis and Acacia spp. Direct seeding in 1992 was south of Lake Mournpall, south of the Mournpall track (623,200E 6,157,700N). Monitoring during 1993 suggested low success rates although by 2000 it became obvious some recruitment of C. gracilis, Acacia spp. and Allocasuarina luehmannii had occurred. In 1993, a large area to the east of the Mournpall Tk and west of Lake Lockie was sown. Initial germination was very high with approximately 100 seedlings per metre of seed trail, although this was heavily browsed by kangaroos and final recruitment was limited. Direct seeding in 1994 was conducted west of Mournpall Tk near Lake

- 67 - Chapter 3 State-and-transition modelling

Konardin (approx. 622,500E 6,159,600N) with no recruitment observed. Direct seeding ceased due to dry years, until 2000 (Stockyard Tk – 628,700E 6,156,900N), 2001 (Kramen Tk 629, 000E 6,154,000N), and 2002 (south of Raak Tk 626,500E 6,167,900N). Success in 2000 was good, especially of Acacia brachybotrya with some C. gracilis, although subsequent years essentially failed (Thomas 2003). In 2005 (Florence Annie Tk 633,000E 6,161,600N) no germinants were observed from direct seeding, believed to be due to inadequate sowing depth and exposure of seed.

The success of direct seeding is highly dependent on adequate rainfall resulting in sufficient soil moisture (Chapter 2). In years considered to have good success (1987, 1991, 1992, 2000) the maximum duration of consecutive days of 0% soil moisture in the summer / autumn period following direct seeding was 31, 37, 31 and 33 days respectively. In years with poor success (1988, 1989, 1993, 1994, 2001, 2002) the dry periods were longer: 59, 48, 70, 52, 85 and 62 days. Total grazing pressure is also important. Kangaroos have been shown to move into areas which have been destocked or revegetated via direct seeding, and the intense browsing is sufficient to prevent establishment of seedlings (Norbury and Norbury 1993). Similar effects have been noted at HKNP (P. Murdoch, Parks Victoria – Hattah, pers. comm., 2005). In 1987, direct seeding was also conducted outside the kangaroo-proof exclosure but success was limited (by 1989, the last recorded assessment, 212 seedlings were noted inside versus 24 outside, Johnson 1989) . A further limitation is heavy frost. In 1994, -10oC temperatures destroyed many three year old plants which had established following the 1991 direct seeding (P. Murdoch, Parks Victoria – Hattah, pers. comm., 2004).

Figure 3.12 Low diversity woodland, direct seeded year 2000, Stockyard Tk (628,700E 6,156,900N), spring 2004. Although a mix of species was used, Acacia brachybotrya was the main species to establish.

- 68 - Chapter 3 State-and-transition modelling

Root suckers – The 1987 trial (Johnson 1989) also involved ripping around mature trees of A. luehmannii, Callitris gracilis and Alectryon oleifolius (Figure 3.13). In 1989, only suckers of A. oleifolius were noted (six suckers approximately 1 m high). This trial does not appear to have been considered again. In 2005, I found 53 A. oleifolius suckers with an average height of 1.5 m high up to 15 m from the parent trees. Four A. luehmannii suckers with an average height of 3 m were found within 10 m of the female parent.

Figure 3.13 Low diversity woodland, root suckers of Alectryon oleifolius, kangaroo proof plot (624,760E 6,161,150N), spring 2005. The suckers were established in 1987 by ripping around mature trees. 3.4.6 Dominance of inappropriate species

3.4.6.1 Annual grasslands The annual grasslands have fewer than 20 trees per hectare, or in some cases only stumps remain. The understorey benchmark structure is not met because there are very few if any shrubs present and little if any recruitment of desirable species. The ground cover is dominated by introduced annual species (no decreasers, 0-25% increasers and 50-100% invaders: Table 3.4) such as Mediterranean Turnip Brassica tournefortii, Little Medic Medicago minima, Onion Weed Asphodelus fistulosus, Skeleton Weed Chondrilla juncea and the grasses Brome Bromus spp., Fescue Vulpia spp. and Barley Grass Critesion murinum ssp. glaucum (Figure 3.14). Some “weedy” or grazing tolerant native species such as Vittadinia spp., Salsola kali, Polycalymma stuartii, Enchylaena tomentosa, and Ajuga australis may also be present (Sluiter et al. 1997a). In good years or seasons the ground flora may achieve complete coverage, however as soil moisture declines, the annual species die off leaving large areas with no litter or protection from erosion. The benchmark value for logs is not met.

- 69 - Chapter 3 State-and-transition modelling

Annual grasses dominated the ground flora in the 1970s, 1980s and into the 1990s (Earl 1984; LCC 1987; Mueck 1987) and high grazing pressure was clearly implicated e.g. Asphodelus fistulosus is positively associated with abundance of kangaroos (Sluiter et al. 1997a).

Table 3.4 Invader species indicative of the degraded vegetation state: Pine-Buloke woodland. Scientific name Common name Family Comments References *Carthamus Saffron Thistle Asteraceae Cool season annual, 2, 3 lanatus completely inedible at flowering *Centaurea Malta Thistle Asteraceae Cool season annual, 2, 3 melitensis invades weak pastures *Paronychia Nailwort Caryophyllaceae Perennial 2 brasiliana *Medicago Small Woolly- Fabaceae Cool season annual 2 minima burr Medic *Critesion Barley Grass Poaceae Cool season annual 2 murinum *Vulpia Brome Fescue Poaceae Cool season annual 2 bromoides *Carrichtera Ward’s Weed Brassicaceae Cool season annual 1, 2 annua *Schismus Arabian Grass Poaceae Cool season annual 1, 2 barbatus Sources 1: Landsberg et al. (2003), 2: Vesk and Westoby (2001), 3: Westbrooke et al. (2001).

Figure 3.14 Good coverage of ground flora as a result of reduced grazing pressure (right) following culling of kangaroos in HKNP, circa early 1990s. Outside the kangaroo-proof fence (left) the heavy grazing pressure has virtually eliminated the cover of annual weedy species. The dry conditions at the end of summer and high grazing pressure circa 1980s have exposed the soil to wind erosion. Photograph: Parks Victoria

- 70 - Chapter 3 State-and-transition modelling

3.4.6.2 Hopbush shrublands Hopbush shrublands meet the conditions described in section 3.4.6.1, except that they are dominated by Hopbush Dodonaea viscosa.

Phase shift III D. viscosa is considered an invasive shrub linked to high grazing pressure (Hodgkinson and Harrington 1985; Cunningham et al. 1992). Seedling survival during summer is critical for establishment of D. viscosa and is inversely related to the amount of herbaceous growth. Overgrazing reduces the herbaceous vegetation and this increases the available summer moisture and survival of D. viscosa (Harrington 1991). However, high grazing pressure can eliminate seedlings. Whilst generally considered unpalatable, D. viscosa seedlings were more common inside exclusion plots (Mueck 1987). This supports observations that D. viscosa started to dominate areas only after relaxation in extreme grazing pressure in HKNP (P. Murdoch, Parks Victoria – Hattah, pers.comm. 2003). The phase shift to Hopbush shrublands from a grassland state is not easily reversible, except perhaps using fire (Hodgkinson and Harrington 1985).

There is a dual image of shrubs in the literature, highlighted as good or bad depending on the researcher’s perspective (Holmgren and Scheffer 2001). Pastoralists particularly see D. viscosa as a problem species (e.g. Hodgkinson and Harrington 1985) however restoration ecologists may view thickets of D. viscosa as nurse plants for palatable seedlings such as Callitris gracilis (Moncrieff 1991). This is evident in HKNP where C. gracilis are just beginning to overtop the D. viscosa (Figure 3.15). D. viscosa also bind the soil and can foster good species diversity of less common species (M. Westbrooke, University of Ballarat, pers. comm. 2005).

Figure 3.15 Hopbush shrublands being overtopped by Callitris gracilis, Mournpall Tk (622,760E 6,159,980N), spring 2005.

- 71 - Chapter 3 State-and-transition modelling

3.4.6.3 Native perennial grasslands Native perennial grasslands meet the conditions described in section 3.4.6.1, except that they are dominated by native perennial species which are usually only minor components of the woodland, such as Kerosene Grass Aristida holathera var. holathera and in some years Spear Grass Austrostipa spp. These grasslands only become evident after the mid 1990s in HKNP (section 3.4.5.2).

Phase shift IV Reduced grazing pressure and favourable rainfall can lead to a phase shift from annual to native grasslands (Sluiter et al. 1997a). Grazing sensitive native species such as Small- leaf Swainson Pea Swainsona microphylla, Silky Swainson Pea S. phacoides, Cushion Knawel Scleranthus minusculus and Austrostipa spp. may appear after favourable rains, such as in 1981 and 1983 (Mueck 1981, 1983). The cover of Austrostipa spp. is greatly influenced (negative association) by abundance of kangaroos and to a lesser extent, a positive association with rainfall in the previous 12 months (Sluiter et al. 1997a).

Grasslands in HKNP are generally considered to be a degraded example of Pine Buloke Woodlands (LCC 1987; Gowans and Westbrooke 2002). Although perennial, native grasslands often show considerable diversity, include many rare flora species and may provide habitat for threatened fauna, I did not consider them as being a desirable vegetation state for HKNP. I acknowledge the strong case for preservation of altered communities with intrinsic values (Oliver et al. 2002), however the values must be weighed against their undesirability.

Native grasslands are not part of the management aim for semi-arid woodlands (section 2.5.1) because they are not believed to represent a historical community. Grassy plains appear to have been a natural feature of the southern mallee (e.g. J. W. Beilby’s account of the Pinnaroo area published in 1849 Kenyon 1914b). Other natural, grassy plains are recognised as palaeo-lakebeds from at least 7,000 years BP when groundwater tables were much higher than today, e.g. between Nyah and Manangatang, and around Pink Lakes and the Raak Plains (LCC 1987). However, the open park-like nature of HKNP described in 1941 was attributed to clearing (Jones 1942).

In addition, retention of grasslands may pose a threat to wooded areas by maintaining an artificially high population of kangaroos which places stress on woody perennial species in dry times, particularly when water is available in the Hattah Lakes system (Sluiter et al. 1997a). Kangaroos are favoured by open grasslands because they are preferentially grazers (Calaby and Grigg 1989). Kangaroos have increased in abundance throughout Australia partially as a result of clearing to promote grassland (Calaby and Grigg 1989; McAlpine et al. 1999).

3.4.6.4 Invasive weeds The invasive weeds phase meets the conditions described in section 3.4.6.1, except that they are dominated by a variety of highly invasive species which are characterised by the need for targeted

- 72 - Chapter 3 State-and-transition modelling control. Invasive weeds of semi-arid woodlands in HKNP include Horehound Marrubium vulgare, Paterson’s Curse Echium plantagineum, Bridal Creeper Asparagus asparagoides (Riverine areas), St John’s Wort Hypericum perforatum (isolated infestation in 2004), Olive Olea europea and isolated infestations of Crownbeard Verbesina encelioides. Ward’s Weed Carrichtera annua has not yet been recorded in HKNP although it dominates large areas of MSNP (pers. obs. 2004); its distribution is correlated with calcareous soils (Cooke 2003). Other new and emerging weed threats for the Victorian Mallee include Chilean Needle Grass Nassella neesiana, Mexican Feather-grass N. tenuissima, and the Cape Tulips Moraea flaccida and M. miniata.

3.4.7 Alteration of abiotic (soil) factors

3.4.7.1 Eroded dunes There are no trees or understorey shrubs present on the eroded dunes. There is also little if any ground cover and those species which are present are introduced annual species (no decreasers, 0- 25% increasers and 50-100% invaders) as described for annual grasslands. Some “weedy” or grazing tolerant native species such as Vittadinia spp., Salsola kali, Polycalymma stuartii, and Sclerolaena diacantha (Cunningham et al. 1992) may be present, however the sand tends to be mobile, and bare of vegetation. There is no organic litter layer nor are there logs present.

Actively eroding dunes were a feature of HKNP from at least the early 1900s (Kenyon 1906; Williamson 1913; Chandler 1938). In the 1970s and 1980s concerns were expressed about sand- drift from dunes blocking tracks in HKNP and in the 1980s the alignment of the Mournpall Tk was shifted four times to accommodate a mobile dune (D. Cheal, Arthur Rylah Institute for Environmental Research, pers. comm. 2003). This dune is now stable (Figure 3.16). At that time “The Razorback” was a popular tourist attraction due to the knife-edge sculpting of the sand. Such severe erosion is no longer apparent, although “blow-outs” are still common.

- 73 - Chapter 3 State-and-transition modelling

Figure 3.16 A dune in HKNP which was mobile in the 1980s and crossed the Mournpall Tk. The dune was stabilised and replanted and is now stable. Species present are Dodonaea viscosa, Callitris gracilis and Acacia ligulata.

3.5 Using the model to better manage semi-arid woodlands

One emphasis of the state-and-transition model is to highlight the need for experimental testing of how transitions might occur between states (Westoby et al. 1989; Laycock 1991; Yates and Hobbs 1997). It is clear that further experimental work is required to test transitions and phase shifts and refine descriptions of vegetation communities in HKNP. However, the model provides managers with a summary of existing knowledge to provide options for developing management plans with realistic species composition objectives and appropriate tools for reaching objectives (George et al. 1992).

Another advantage of developing S-T models is that they summarise assumptions about ecosystem function (Westoby et al. 1989). This process highlights opportunities, failures and gaps in the knowledge base which are important for learning (Redford and Taber 2000; Adams et al. 2002). The opportunities and knowledge gaps highlighted by the conceptual model of vegetation change in Pine and Pine Buloke Woodlands of HKNP (Figure 3.6) include:

1. Determining the cause of phase shift I and phase shift II, and whether it is possible to increase the frequency of this vegetation change. The recruitment of Allocasuarina luehmannii, Callitris gracilis and a variety of Acacia spp is on a limited spatial scale and some areas, often in close proximity, are showing no such recruitment. Identifying why recruitment is occurring in some areas but not others may assist in providing management options to enhance natural recruitment or increase the success of restoration activities. Chapter 4 describes the incidence and spatial distribution, timing and factors involved in recruitment of A. luehmannii.

2. Improving the success of the transition from phases defined by dominance of inappropriate species to low diversity woodlands. This is currently achieved by direct seeding and hand- planting which often fails, especially for A. luehmannii and C. pauper (Thomas 2003). Both species can produce root suckers and the feasibility of this as a new tool for restoration is investigated in Chapters 5 and 6.

3. Determining the techniques necessary for a transition to occur to a restored woodland desirable community, and

4. Considering the implication of new and emerging weed threats such as Carrichtera annua.

- 74 - Chapter 4 Natural recruitment

Chapter 4 Describing the occurrence of natural recruitment

4.1 Scope and objectives

The state-and-transition model presented in Chapter 3 indicated that in Hattah Kulkyne National Park (HKNP), some areas are undergoing a change from degraded to partially regenerating woodlands, characterised by the recruitment of Allocasuarina luehmannii, Callitris gracilis and a variety of Acacia spp. Other areas, often in close proximity, are showing no such recruitment. The model assisted to highlight this gap in the knowledge base – “why is recruitment occurring in some areas but not others?”. From a management point of view this key question can be interpreted as “Can we mimic the factors which are leading to recruitment in these areas, to enhance the success of restoration activities?”. To answer to these questions requires the incidence and spatial distribution, timing and factors involved in recruitment to be determined.

The recent recruitment of Allocasuarina luehmannii is particularly unexpected because the literature (Castle 1989; Raymond 1990; Morcom and Westbrooke 1998) conforms to beliefs for many semi-arid species: “Regeneration of this species [A. luehmannii] is thought to occur only as the result of a combination of exceptional rainfall conditions and low grazing pressure” (Sandell 2002). Similar sentiments are expressed regarding the recruitment of Casuarina pauper (Chesterfield and Parsons 1985; Auld 1995c). Chapter 2 indicates that exceptional rainfall has not occurred since the 1970s. Therefore I investigated further the occurrence of seedlings in HKNP and searched areas of Murray Sunset National Park (MSNP) to determine whether similar recruitment of C. pauper seedlings was occurring. I also attempted to determine when recruitment had occurred to compare with recorded rainfall since 1964.

I reviewed the factors reported to positively or negatively affect seedling establishment and from this identified techniques which could be implemented by land managers to enhance restoration of a specific area. The observed patchy nature of recruitment, occurring around some female trees but not others, suggested maternal effects were the most likely candidates. Maternal effects influence recruitment due to genetically based differences in seed production, seedling emergence and seedling survivorship between parent plants (Roach and Wulff 1987; Herrera 2000). Different microsites of the parent plant, including soil type (Lauenroth et al. 1994; Herrera 2002) or microtopography (Tongway and Hindley 1995) may also affect recruitment.

- 75 - Chapter 4 Natural recruitment

My objectives were to

1. Assess of level of seedling recruitment of A. luehmannii and search for seedling recruitment of C. pauper,

2. Identify the timeframe over which recruitment may have occurred,

3. Explore a number of maternal effects which may correlate with enhanced seedling recruitment, and

4. Provide options for managers to use this knowledge to enhance restoration success

4.2 Introduction 4.2.1 Background

The state-and-transition model presented in Chapter 3 describes a partially regenerating woodland of varying composition which may include juveniles of Acacia spp., Callitris gracilis, Alectryon oleifolius, Hakea tephrosperma and Allocasuarina luehmannii. I was particularly interested in the recruitment of A. luehmannii because this species is difficult to recruit from direct seeding (P. Murdoch, Parks Victoria – Hattah, pers. comm. 2002) and natural seedling recruitment is generally attributed to exceptional rainfall conditions (Castle 1989; Raymond 1990; Morcom and Westbrooke 1998; Sandell 2002). Similar sentiments are expressed regarding the recruitment of Casuarina pauper (Chesterfield and Parsons 1985; Auld 1995c). Recruitment between rainfall events is generally assumed to be from root suckers (e.g. Hall et al. 1964; Beadle 1981; Chesterfield and Parsons 1985; Sandell 2002; Macaulay and Westbrooke in prep.). However, my initial observations indicated that juvenile A. luehmannii between 10 cm and 150 cm high were present at HKNP, indicating multiple recruitment events. Preliminary excavations suggested the majority of recruits were seedlings.

4.2.2 Where and when is seedling recruitment occurring?

Determining where, when and why recruitment is occurring can assist to focus and enhance the success of future restoration activities. Recruitment may be patchy due to:

• soil type (e.g. Lauenroth et al. 1994; Herrera 2002), • favourable rainfall limited either spatially or temporally (e.g. Ogle and Reynolds 2004; Schwinning and Sala 2004; Westbrooke and Florentine 2005), or • the production of highly germinable seed by a single individual plant (e.g.Helenurm and Schaal 1996; Chacón and Bustamante 2001). My initial observations indicated that recruitment of A. luehmannii had occurred at a range of locations across HKNP but on a small spatial scale, often patches of recruitment surrounding a single tree. These observations were, like most flora surveys (e.g. Gowans and Westbrooke 2002)

- 76 - Chapter 4 Natural recruitment biased towards areas close to tracks. To verify preliminary observations of where recruitment was occurring, I planned an unbiased, quantitative survey (section 4.3.1) of the occurrence of A. luehmannii juveniles across HKNP. The survey was repeated in MSNP to determine whether similar recruitment of C. pauper was occurring.

The timing of recruitment is very important from a management perspective because it suggests whether it was the result of rainfall which is outside a manager’s influence, or whether other actions, such as reduced grazing pressure, are important. Estimating the age of an individual from its size is an imprecise science (Watson et al. 1997b) based on the “vague principle” that large individuals are likely to be old and young individuals small (Harper 1977). However, combined with revisiting sites previously surveyed for recruitment, the maximum likely age of recruits can be estimated.

4.2.3 Why has seedling recruitment occurred?

4.2.3.1 Overview Seedling recruitment involves germination of seeds and survival of seedlings (Harper 1977) and is believed to be limited by either the availability of seeds or the availability of microsites (Eriksson and Ehrlén 1992). Germination requires sufficient seed of high germinability (Hall et al. 1964) and suitable conditions for germination, particularly adequate soil moisture and microsite conditions (Bell 1999). Overall, herbivory and drought are the most frequent cause of seedling death (Moles and Westoby 2004).

The following review focuses on seed quantity and quality, soil moisture, microsite availability and herbivory as the main influences on seedling recruitment. My preliminary observations (Chapter 3) indicated a pattern of recruitment of A. luehmannii with patches of seedlings surrounding a single tree, at locations throughout HKNP. Recruitment does not appear restricted to a particular population of trees. Small scale variation in habitat or genetics of individual trees may be responsible. These were taken into account when choosing appropriate measures for the described variables.

4.2.3.2 Seed quality and quantity Genetic and environmental effects influence both the amount of seed produced per tree (e.g. Haig and Westoby 1988) and the seed quality (e.g. Roach and Wulff 1987). Genetic effects include simple genetic variation between individuals as well as more complex factors such as the ploidy level of the individual (section 2.6.3.1). Environmental effects are related to resource availability in the maternal environment, including the availability of pollen. The quantity of seed produced per tree can be estimated simply by qualitative or quantitative estimation.

- 77 - Chapter 4 Natural recruitment

Quality of seed is usually measured as percentage germination (Simons and Johnston 2000). Percentage germination often shows considerable intraspecific heterogeneity (e.g. Pérez-García 1997; Chacón and Bustamante 2001). Germination rate also influences recruitment with earlier germination of seeds linked to enhanced growth or survival of seedlings (Jones et al. 1997b; Kelly and Levin 1997), but the correlation is not universal (Verdú and Traveset 2005). Individual seed size is also correlated positively with offspring fitness (Castro 1999; Leishman 2001). Heavier seeds have a greater rate and percentage germination, seedling growth rate, and better seedling ability to survive competition, herbivory, shading, drought or nutrient limitation or adult plant size (Roach and Wulff 1987; Gomez 2004). These effects appear to hold for intraspecific comparisons also (Bonfil 1998; Castro 1999; Seiwa 2000; Simons and Johnston 2000; Chacón and Bustamante 2001; Lopez et al. 2003). Within seed collected from a single tree, variation in seed color may also be a useful predictor of germinability, indicating such things as variation in maturity (e.g. Cuthbertson 1997) or pathogen resistance (e.g. Grazywacz and Rosochacka 1980). In summary, the quality of seed may be measured by percentage and rate of germination, mass of individual seeds or color of individual seeds.

4.2.3.3 Soil moisture Soil moisture has a dominant role in recruitment in arid environments (Noy-Meir 1973; Chesson et al. 2004) and rainfall is the primary contributor to soil moisture (Ogle and Reynolds 2004). Although small-scale variation in rainfall can be substantial in semi-arid areas (Hodgkinson and Freudenberger 1997), it is unlikely to vary on the scale observed here. However, aspects of site topography affect the redistribution of rainfall and soil factors affect the availability of soil moisture at small spatial scale (Maestre et al. 2003).

Runoff redistributes rainfall so that runon zones such as drainage lines, roadsides, or depressions, receive more water than runoff zones such as ridges (Tongway and Hindley 1995; Hodgkinson and Freudenberger 1997; Chesson et al. 2004). Recruitment may occur predominantly in runon zones (Chesterfield and Parsons 1985; Ireland 1997; Tiver and Andrew 1997). Microtopography can be measured directly in the field by calculating slope etc. (Tongway and Hindley 1995).

Soil hydraulic properties such as water retention and the availability of soil moisture reflect soil physical properties including porosity, infiltration rate, soil matrix potential and saturated hydraulic conductivity. Variation in these properties can be adequately and simply described from soil texture (Cosby et al. 1984). Sands have the greatest infiltration rate due to large pore spaces; silty to heavy clays have the lowest. In arid areas, sandy soils allow deep penetration of rainfall and losses through evaporation are less than for clay soils which have shallow penetration (Tongway 2003). Sands also have lower matrix potentials than clays so more water is available to plants for uptake in sand versus clay before the wilting point is reached (Tongway et al. 1989).

- 78 - Chapter 4 Natural recruitment

However, the storage of available water is greater for the finer textured soils, being essentially linearly related to silt content (Lauenroth et al. 1994). Soil hydraulic properties are often estimated from soil textural classes (Saxton et al. 1986; Rawls and Pachepsky 2002; Nemes and Rawls 2004) which can be easily measured in the field (McDonald et al. 1984).

4.2.3.4 Microsite availability The seed microsite is a broad term covering the immediate site surrounding the seed and containing all the requirements necessary for germination and establishment (Eriksson and Ehrlén 1992). The success of seedling establishment can be increased by utilising favourable microsites (Maestre et al. 2003; Vallejo et al. 2006). Of particular interest are the effects of competition and facilitation by other plants on the establishment of seedlings. Recruitment may therefore be limited by the availability of either vegetated or unvegetated microsites. Vegetated microsites include the presence of annual or perennial understory vegetation, cryptogamic layer and surrounding trees. Soil disturbance and fire may also influence the recruitment of seedlings (Callister 2004).

4.2.3.5 Herbivory The degree of herbivory on seedlings depends on grazing pressure and palatability. Whilst grazing pressure by mammalian herbivores is a key influence on rangeland recruitment (Dyksterhuis 1949), substantial reduction in grazing pressure would not be expected to occur on the observed small scale but rather over broad areas of HKNP. Whilst the reduction in grazing pressure from rabbits and kangaroos since 1980 and especially since 1996 (Chapter 2) is likely to be important, it is unlikely to be the sole cause.

Intraspecific variation of palatability (between ecotypes, provenances or clones) of a species has been observed (Laitinen et al. 2004; Norman et al. 2004; Walsh et al. 2005) but is not always demonstrated (Ireland 1997; Buschmann et al. 2005). There are few if any experimental tests of maternal differences in palatability of seedlings, probably due to methodological difficulties. Whilst variation in palatability may affect the recruitment of A. luehmannii seedlings at HKNP, it was not considered here.

4.3 Methods 4.3.1 Where is recruitment occurring?

4.3.1.1 Design of survey The shape, size, number and location of quadrats for sampling the abundance of a plant species is determined by the study objectives, vegetation type and site characteristics (Kent and Coker 1992). The primary objective of this study was to broadly and efficiently survey Belah and Buloke woodlands to identify areas where and how much recruitment was occurring with more

- 79 - Chapter 4 Natural recruitment detailed investigation at a later date. To achieve this I used long, thin quadrats (belt transects) because in a heterogenous habitat they cross more habitat types than circular or square quadrats of the same area (Kenkel and Podani 1991; Krebs 1999). Belt transects are commonly used in many monitoring programs for rangelands (e.g. Western Australian Rangeland Monitoring System).

Covering the largest area possible is necessary to detect non-episodic recruitment of semi-arid, long-lived species because recruits are sparse (Kenkel and Podani 1991; Watson et al. 1997b). The length of the belt transect for my study was limited by the size of the patches of Buloke and Belah woodlands, generally not larger than 1 km in width. Therefore, I increased the length of the transect whilst remining in the woodland by joining together a number of independent, parallel transects, or “legs”, in a “folded transect” (Figure 4.1).

To detect juveniles, often less than 20cm high, the minimum amount of ground flora biomass was required. The surveys were conducted between June and July 2004 (early winter) following 112 consecutive days of 0% soil moisture, the longest dry period since records began in 1963 (Chapter 2). This resulted in extremely low ground flora biomass and maximised the chances of detecting juveniles during surveys. I surveyed the transects from a four-wheel motorbike (Figure 5.8). Field trials suggested that juveniles could be observed from the motorbike at a maximum distance of 40 m, equalling a transect width of 80 m so the legs were spaced 200 m apart so each was independent. The “legs” and the connecting segment between each leg yielded a folded transect with a total length of 7.0 km and approximate area of 0.56 km2. I oriented the legs north-south to avoid surveying into a rising or setting sun (Figure 4.1).

Figure 4.1 A “folded transect” maximises the transect length within a small area of vegetation. The legs of the transect run north-south and are 1.0 km long; the length of the connecting segment between each leg is 0.2 km. The total length of this transect is 7.0 km. I surveyed the non-mallee (eastern) half of HKNP and the north-western corner (Shearers’ Quarters) of MSNP. In MSNP, I only surveyed areas of Belah where a management program for kangaroos was in place because the literature suggested that kangaroo grazing pressure would eliminate recruits in other areas. The extent of the “Pine-Buloke” and “Belah” vegetation communities mapped for both HKNP and MSNP from 1999 information was used to determine the number of quadrats necessary to cover the area.

In HKNP, the Pine-Buloke community covers 31.6 km2, but excluding areas containing only pines (approximately 10.4 km2) or not containing A. luehmannii (approximately 3.4 km2) reduced the total area of Buloke woodland to 17.8 km2. In contrast, the Belah woodland around the Shearers’

- 80 - Chapter 4 Natural recruitment

Quarters appeared to be underestimated, covering only 5.3 km2. My field surveys indicated an additional four areas of Belah woodland covering an area of 4.2 km2, bringing the total area of Belah to 9.5 km2. The sampling intensity was arbitrarily determined to be one folded transect per km2 of target vegetation resulting in 18 transects for HKNP and 10 for MSNP with approximately 50% coverage of the woodlands.

Random points within the polygons which delineate the vegetation communities were selected using tools within the GIS program MapInfo. The transect was manipulated manually around that point so that the maximum area was within the Pine-Buloke or Belah communities. Each point was buffered by 2 km so transects could not overlap. The corner points (twelve) of the folded transect were uploaded to a hand-held Garmin 12XL Global Positioning System (GPS) and used to navigate the transect on the ground.

The four-wheel motorbike enabled the 7km transects to be traversed at a constant speed of 10 km / hour when mobile. It also enabled transects to be placed anywhere within the region, preventing any bias associated with restricting transects to within a set distance from a road or track. The limitations of using the motorbike were difficulties in thick mallee vegetation or Melaleuca lanceolata thickets. On occasions, these thickets had to be skirted however walking transects did not locate any juveniles in these vegetation types.

4.3.1.2 Information recorded on transects I used the GPS to record a track log for each transect, automatically marking locations at 30 second intervals. This enabled the exact location and length of the transect to be determined by downloading the track log to MapInfo. Whilst traversing the transect, I marked additional waypoints at the interface of distinct vegetation types. Where C. pauper or A. luehmannii was present, the density of the trees was recorded as scattered (>100m between trees), open (50-100m between trees), woodland (20-50m between trees) or dense (<20m between trees).

I marked the location of every patch of juveniles (up to 6 m in height) using the GPS. The distribution of juveniles was clearly clumped, with a small number of patches containing many individuals. Seedlings often surrounded a female tree and I considered this the “parent”. For root suckers, I termed the nearest tree of either sex as its parent. Individual juveniles within a patch were numbered sequentially with increasing distance from the nominated parent tree. I excavated a proportion of juveniles in each patch to determine seedling or root sucker origin. If the patch contained fewer than five juveniles, all were excavated. Every second juvenile was excavated in patches that contained six to twenty juveniles, otherwise between ten and thirty juveniles were excavated, the number determined proportionally according to the size of the patch. Juveniles that were clearly root suckers on exposed roots were recorded separately. The numbering was

- 81 - Chapter 4 Natural recruitment approximate where there were numerous juveniles but served to ensure that juveniles from a range of distances were randomly selected for excavation.

I excavated juveniles as they were encountered to a depth of 20-40 cm to establish whether they were root suckers or seedlings. Juvenile root suckers are clearly distinguished from seedlings by the presence of a “T-junction” (Maconochie 1982) where the stem of the sucker arises from the parent root. Seedlings have a long, tapering, single tap-root. My earlier exploratory excavations of approximately 30 juveniles to a depth of 1.5 m indicated that 20-40cm is sufficient to determine whether a juvenile is a root sucker. Ambiguous cases were excluded and another juvenile selected. For excavated seedlings, I recorded the diameter of the base of the stem 2 cm below ground level (above this height the stem was often grossly swollen due to browsing), the distance to the parent tree, the height and the growth form (single stem, few stems, bushy).

In September 2005, I revisited trees with more than 20 seedlings and mapped the spatial distribution of seedlings around the parent tree, and whether the seedlings occurred on bare soil, vegetated patches or in leaf litter. This visit occurred following rainfall and enabled the distribution of annual species to be considered in relation to seedlings.

4.3.2 When has seedling recruitment occurred?

The majority of recruitment detected in MSNP was root suckers, so the work described below was only carried out for A. luehmannii in HKNP.

I attempted to group seedlings into age classes based on size. Although I recorded height, I did not consider it indicative of age except in the broadest sense, for example a 2 m tall juvenile is likely to be older than a 10 cm high juvenile. Instead I used the basal diameter of the stem to group A. luehmannii seedlings into size classes. I made broad comparisons of the basal diameter of these seedlings with seedlings derived from direct seeding from known years.

In some areas of HKNP, extensive surveys have been conducted which included searching for juvenile woody perennials (Sluiter et al. 1997a; Sandell 1999; Gowans and Westbrooke 2002; Sandell 2002). Vegetation monitoring in HKNP conducted between 1992 and 1996 (Sluiter et al. 1997a) included in its aims: “Comment on the incidence of and prospects for the regeneration of woody perennial vegetation within the park”. During the research work, no A. luehmannii recruitment was recorded on the permanent transects (33.7 transect kilometres in total), however root suckers of Hakea tephrosperma and Alectryon oleifolius were noted, as were Acacia oswaldii seedlings. I located the original vegetation transects again with the assistance of marker pegs which were originally placed every 100 m along the transect.

A Vegetation Condition Assessment of HKNP was conducted in 2001 (Gowans and Westbrooke 2002). This study recorded, amongst other things, the Diameter at Breast Height (DBH) of

- 82 - Chapter 4 Natural recruitment dominant overstorey species, including juveniles, within quadrats measuring 50 m x 20 m. The methods also state that “Observations either within or in the vicinity of the quadrat relating to vegetation condition, disturbance history, or other management issues were noted” (p. 17 Gowans and Westbrooke 2002), suggesting that recruitment adjacent to the quadrat would be noted. Thirteen of 60 quadrats contained adult A. luehmannii. I relocated all of these quadrats and searched the quadrat area for juveniles in September 2004.

A number of exclusion plots have been established as well as direct seeding of A. luehmannii. I revisited the locations of these surveys and trials, where they occurred in semi-arid woodland vegetation, to determine whether any juvenile A. luehmanni were present in the vicinity. If juveniles were present they were excavated to determine seedling or sucker origin. The time since the last survey gave a maximum age of juvenile, assuming that all juveniles were detected at the time of surveying. I compared the basal diameter of the known age seedlings with the seedlings of unknown age.

4.3.3 Why is seedling recruitment occurring?

4.3.3.1 Selecting parent versus non-parent trees The field surveys (section 4.3.1) demonstrated that patches of seedlings were generally closely associated with a female tree, the parent female. There were however, numerous female trees with no seedlings. I located 21 female trees which had at least 20 seedlings surrounding the tree. Many trees were associated with fewer than 20 seedlings, however I chose trees with many seedlings because these were likely to demonstrate conditions most suitable for recruitment. These females represented the range of tree and site conditions for the subset of regenerating females and were termed “parent females”. Fifty-two females were selected at random from within the known distribution of A. luehmannii at HKNP (based on 1999 information supplemented with personal observation). Random points were selected using MapInfo tools from polygons surrounding areas containing A. luehmannii within 500 m of a track. I determined the nearest, female tree to the sampling point. These females represented the range of tree and site conditions for females in the population as a whole and were termed “non-parent females”. I measured factors which influence the occurrence of seedlings, with the aim to identify factors that distinguish parent from non- parent females. The sample sizes were not equal because there were many non-parents but only limited parents available.

4.3.3.2 Seed quantity and quality I used surrogates for seed quantity (current and previous cone crop and seeds per cone) and seed quality (seed mass, and germination rate and percentage germination).

Each tree was visited between 23rd and 27th January 2005 to assess the current seed crop. At this time all cones were brown and some had opened. I conducted a visual assessment of the cone crop

- 83 - Chapter 4 Natural recruitment and classed it as heavy, moderate, light, poor or absent, and whether the cones on the tree were all open, all closed or a mixture. Production of cones in previous years was measured as the average number of cones from four, 20 cm x 20 cm quadrats placed at a distance of 2 m north, south, east and west of the trunk of the female trees (within the drip line).

I used a 4-m-long pole pruner to remove a sample of opened and unopened cones (at least 40 cones, depending on availability). In some cases it was necessary to use a shotgun to sample very high, cone-bearing branches. The cones were stored with naphthalene flakes at ambient temperature (20oC to 35oC) for one month, by which time the valves had opened. The seeds were either naturally released from the open cones or could be removed easily with forceps.

In addition, up to fifteen unopened cones per tree (depending on availability) were stored individually in paper envelopes. After drying, the number of open valves and the number of seeds produced per cone was counted (574 cones from 53 trees) to estimate the relationship between the number of open valves and the number of seeds produced per cone. A strong relationship existed between open valves and seeds per cone (R2 = 0.9743) and I used the relationship to determine the mean seeds per cone for 40 cones per tree from the 2005 season (where available).

Seeds = 0.9829*valves -0.0709

A sample of 100 seeds (where available) was weighed using Mettler Toledo PG503-S scales with 0.001 g accuracy and the mean individual seed mass determined for each tree.

The germination characteristics of a sample of 100 seeds from each tree (where available) were tested in August – September 2005 (Mortlock and Lloyd 2001). I thoroughly mixed the available seed from each tree and counted 25 seeds into a plastic, 5.5 cm petri dish. The petri dishes were filled to a depth of 0.5cm with vermiculite and lined with Whatman No. 1 filter paper, dampened with distilled water. Soaking the seed for 10 minutes in a 1% sodium hypochlorite solution (Mortlock and Lloyd 2001) was sufficient to overcome problems with mould experienced in earlier trials. Seeds were germinated in a single germination chamber using 12:12 hour alternating dark (15oC) and natural light (25oC) cycles. This incubation temperature was chosen based on germination requirements of a number of species of Casuarinaceae (Turnbull and Martensz 1982; Clemens et al. 1983; Turnbull and Martensz 1983; Kuo 1984; Hwang 1991; Moncur et al. 1997). The filter paper was moistened as required, generally every five days and germinants were counted and removed daily. The criterion of germination was when the radicle emerged from the testa. I repeated this four times so each tree had four replicates of 25 seeds. I recorded the Germination Index (Maguire 1962) and total percentage germination after 20 days, both averaged over the four replicates.

- 84 - Chapter 4 Natural recruitment

The Germination Index was calculated as:

GI = ∑ (Gt / Tt) where Gt = number of germinants on tth day, Tt = day of germination test. Larger values indicate a faster germination rate.

The seeds collected from each tree were thoroughly mixed and the overall colour was recorded. I was interested to determine, regardless of parent status, whether there is an association between germination percentage and rate, seed color and trees with open, half open or closed cones. Both seed color and the degree to which cones are open may indicate maturity, with darker seeds and a greater percentage of open cones hypothesised to be more mature and therefore have better germination. This has important implications for the future collection of viable seeds for revegetation. I used a one-factor Analysis of Variance (ANOVA) to compare the percentage germination and the Germination Index (GI) of dark, medium and light-colored seeds, and open, half open, or closed cone crops at the time of collection. Where a significant difference was observed in germination, the Student-Newman-Keuls (SNK) procedure was used for a posteriori comparison of seed color and cone crop (Underwood 1997).

4.3.3.3 Soil moisture Landform pattern is the relief and slope of the area within a 300 m radius, and landform element is the runoff / runon situation at a finer scale (within a 20 m radius). I visually assessed the hydrology, landform pattern and landform element around each tree in September 2005 (Tongway and Hindley 1995).

• Hydrology –runoff, neutral, runon. • Landform pattern –slope category: <1%, 1-3%, 3-10% • Landform element - Ridge, mid slope, lower slope, flat. Soil type was determined from the field texture of a moist, hand-manipulated bolus (McDonald and Isbell 1990). Relevant categories were sandy loam and loamy sand.

4.3.3.4 Microsite availability In September 2005, I mapped the area within a radius of 30 m surrounding the female trees that was bare ground because patches of more than 20 seedlings tended to be associated with a bare area. The effect of competition with adult trees was also considered using the density of trees surrounding the female tree. This was estimated as the number of trees within a 30 m and 100 m radius (few 0-5, some 5-20, lots >20).

- 85 - Chapter 4 Natural recruitment

4.3.3.5 Measured variables In summary, the measured variables were seed quantity, seed quality, soil moisture, microsite availability, and density of surrounding trees (Table 4.1). The variables were limited to those that could be measured in the field, with the exception of germination percentage and rate. This variable can be measured quite easily with minimal equipment (Mortlock and Lloyd 2001) however I also tried to find a simple field correlate using mass and color of the seeds.

Table 4.1 List of factors believed to influence the recruitment of seedlings and how and when they were measured. Factor Measure Data Values Data collected Identification Parent Binary Yes, no June-July 2004 Seed quantity Current cone crop Ordinal Absent, poor, light, Jan. 2005 moderate, heavy Previous cone crop Continuous Average of cones from four Jan. 2005 quadrats Seeds per cone Continuous Average of seeds from 40 Feb. 2005 cones Seed quality Seed mass Continuous Average individual seed Feb. 2005 mass Seed color Binary Light (orange, pale orange, Feb. 2005 yellow), dark (dark brown, brown, light brown) Germination rate Continuous Average Germination Index Aug. – from four trials Sept. 2005 Percentage Continuous Average from four trials Aug. – germination Sept. 2005 Soil moisture Hydrology Nominal Runoff, runon, neutral Sept. 2005 Slope Ordinal <1%, 1-3%, >3% Sept. 2005 Landform Nominal Flat, lower slope, mid slope, Sept. 2005 ridge Soil type Nominal Sandy loam, loamy sand Sept. 2005 Microsite Bare area Continuous Square metres of “bare” Sept. 2005 availability ground Trees within 30 m Continuous Number of trees within 30 Sept. 2005 m Trees within 100 m Ordinal Few: <5, some: 6-30, lots: Sept. 2005 >30

- 86 - Chapter 4 Natural recruitment

4.3.4 Tree-based modelling of key factors influencing recruitment

Classical methods of statistical analysis tend not to suit restoration ecology where environmental variation is high and sample sizes are small (Holl et al. 2003). I was restricted by the small number of female trees showing extensive recruitment and consequential unequal sample sizes. Standard multiple regression procedures used to detect ecological relationships are limited by nonlinear relationships, heteroscedasticity of variances and interactions among variables; all often present in ecological data (Underwood 1997; Rejwan et al. 1999). In addition, identifying predictor variables is often difficult because, as was the case with my data, a mixture of continuous and categorical, ordered and unordered variables are recorded.

Tree-based models (Brieman et al. 1984) are an exploratory technique based on uncovering structure in data. They can be used to summarise a multivariate data set comprised of a mixture of variables, to find the most important predictor variable(s) for the dependent variable. The relationship is presented in an easily interpretable graphical display (e.g. Merler et al. 1996; De'ath and Fabricius 2000). There are a variety of programs available including QUEST (Loh and Shih 1997) and CART (Brieman et al. 1984); I used DTREG (Sherrod 2003). Tree-based models outperform traditional approaches when the data are complex and non-linear, such as ecological data (Olden and Jackson 2002). This method was particularly useful here because it is not sensitive to unbalanced data, i.e. more females without seedlings than with seedlings. Analysing unbalanced data with traditional statistical analyses causes immense difficulties (Underwood 1997). Tree-based modelling is also attractive because it is not affected by potentially correlated predictor variables (Jackson and Bartolome 2002).

Classification trees are used for binary and categorical response variables and regression trees for continuous response variables. Classification trees were used here to uncover variables important in the initiation of root suckers and also in Chapter 5. Splits are based on the proportion of presences and absences in each group and the nodes are defined by the dominant category (present or absent). Where the response variable is not binary, the two most dominant categories define the split. Misclassification rates are used to summarise the error rate of the tree.

The DTREG tree is built by an exhaustive search of predictor categories to find the one which gives the best improvement in the purity of the node. Various splitting criteria may be used including the information index, the twoing index and the Gini index (De'ath and Fabricius 2000). The latter is used by DTREG because it maximises the heterogeneity of the categories of the target variable (Sherrod 2003). For classification trees, the entire data set undergoes a binary split; the first node is defined as the explanatory variable that minimises the misclassification rate. This continues until no new split provides a reduction in the misclassification rate. The tree does not split nodes which contain fewer than ten records and the maximum number of levels is set to ten.

- 87 - Chapter 4 Natural recruitment

Pruning trees is necessary to avoid overfitting and an overly large tree (Jackson and Bartolome 2002). V-fold cross validation chooses the best trade-off in between minimising classification error and large tree size without requiring an independent dataset (Sherrod 2003).

Missing data occurred where insufficient seed or cones were able to be collected to calculate seed mass, seed color, germination rate or percentage germination. Missing data are handled well by classification trees. DTREG uses the “surrogate splitter” technique to enable cases with some missing values to be used in the tree-building. Surrogate splitters are predictor variables that are not as good at splitting a group as the primary splitter but which yield similar splitting results. Surrogate splitters, along with primary splitters, are also used to rank the overall importance of the predictor values.

The tree produced by DTREG shows the most important predictor variable(s) to separate parent from non-parent trees. I also considered the overall importance of the variables as generated by a Decision Tree Forest. Clearly a variable used as a primary splitter is important, however other variables may also substantially reduce the misclassification rate but may be masked by the primary splitter. Only variables with an importance >0 are shown. The Decision Tree Forest grows a number of unpruned trees (in this case 200) using independent, bootstrapped samples from the dataset. Each tree uses only a random selection of the predictor variables for each node. The predicted versus actual category for each observation at each node of each tree is then used as “votes” for the best category. The technique has the advantage of greater accuracy than the single tree model without overfitting, but it is recommended the two be considered together for complete understanding of the data. DTREG uses the “Random forest” algorithm (Brieman 2001).

4.4 Results 4.4.1 Where is recruitment occurring?

4.4.1.1 Recruitment of Casuarina pauper In MSNP, I traversed ten folded transects; a total of 70.94 transect km. The vegetation types encountered were in descending order representation (Figure 4.2):

• Savannah woodland excluding Belah, 19.1 km (28%), • Pure Belah, 18.6 km (26%), • Mallee, 12.2 km (17%), • Savannah with scattered Belah, 8.0 km (11%), • Grassland, 7.1 km (10%), and • Saline Shrubland, 6.0 km (8%).

- 88 - Chapter 4 Natural recruitment

C. pauper was present over 26.6 transect km (37%). The most common vegetation types containing Belah was Belah woodland (50%), followed by Savannah with scattered Belah (30%) (Figure 4.3).

8%

10% 28% Savannah woodland

Belah

Mallee

11% Savannah with scattered Belah

Grassland

Saline shrubland

17% 26%

Figure 4.2 Vegetation types traversed in the 70.9 km of folded transects in Murray Sunset National Park. The slices of pie are labelled with the percentage of the total distance in each vegetation type.

50%

30% Belah woodland

Belah open

Belah dense

Belah scattered

Savannah with scattered Belah

3%

4%

13%

Figure 4.3 The vegetation types encountered in the 26.6 km of folded transects traversed in Murray Sunset National Park which contained Casuarina pauper. The vegetation type refers to the tree spacing: scattered (trees >100 m apart), open (50-100 m apart), woodland (25-50 m apart) and dense (<25 m apart).

- 89 - Chapter 4 Natural recruitment

I recorded 41 patches of C. pauper recruitment in MSNP all within vegetation types containing C. pauper. Most (58%) of these patches occurred within, or on the immediate edge of Belah woodland which was also the most frequently encountered vegetation type. When the number of patches was expressed per transect kilometre, the greatest density of patches occurred in the scattered Belah vegetation type (5.0 patches per transect kilometre). The average density over all vegetation types containing Belah was 1.5 patches per transect km (Table 4.2).

Table 4.2 The number and percentage of patches of Casuarina pauper recruitment occurring in each vegetation type, the kilometres traversed in each vegetation type and the patches per kilometre in each vegetation type. Vegetation type No. of patches % of patches Km traversed Patches / km

Belah dense 0 0 1.1 0

Belah open 2 5 3.4 0.6

Belah scattered 4 10 0.8 5.0

Savannah with 11 27 8.0 1.4 scattered Belah

Belah woodland 24 58 13.3 1.8

Total (Average) 41 100 26.6 (1.5)

A total of 260 juveniles in the 41 patches was encountered. Most patches (22%) contained only one juvenile (Figure 4.4), however patch size ranged up to 28 juveniles in a single patch (mean of 6.3 juveniles, median result of 3 juveniles). I excavated 161 juveniles and found nearly all (93%) were root suckers. 95 (63%) of the root suckers were on exposed roots and for the remainder the root junction was between 2 cm and 8 cm below the surface (further details presented in section 5.4.1.1). Of the 11 seedlings which I excavated, they occurred singly (one case) or as a group of seedlings greatly outnumbered by root suckers (three cases). The distance of the seedlings to the nearest female C. pauper ranged from 17 m to 30 m (mean 22.5 ± SD 3.6 m).

- 90 - Chapter 4 Natural recruitment

10

9

8

7

6

5

4 Number of patches of Number 3

2

1

0 1 2 3 4 5 6 to 10 11 to 15 16 to 20 21 to 25 26 to 30 Number of juveniles in a patch

Figure 4.4 Frequency distribution of the number of Casuarina pauper juveniles in each patch of recruitment.

4.4.1.2 Recruitment of Allocasuarina luehmannii In HKNP, I traversed a total of 135.1 transect km in eighteen folded transects. The vegetation types encountered were in descending order representation (Figure 4.5)

• Riverine, 42.6 km (32%), • Pure Buloke, 28.6 km (21%), • Grassland, 17.4 km (13%), • Pine woodland, 14.8 km (11%), • Mallee, 13.1 km (10%), • Hopbush, 9.5 km (7%), • Pine with scattered Buloke, 4.5 km (3%), and • Riverine with open Buloke, 4.5 km (3%). A. luehmannii was present over 37.6 transect km (29%). The most common vegetation types containing A. luehmannii was the Buloke scattered (43%), followed by Buloke woodland (21%) (Figure 4.6).

- 91 - Chapter 4 Natural recruitment

21%

32% Buloke Pine Buloke scattered Riverine Buloke open 3% Grassland

3% Hopbush Mallee Pine Riverine 13% 11%

7% 10%

Figure 4.5 Vegetation types traversed in the 135.1 km of folded transects in Hattah Kulkyne National Park. The slices of pie are labelled with the percentage of the total distance in each vegetation type.

5% 2% 11%

13%

Buloke dense Buloke open Buloke scattered Buloke wland Pine Buloke scattered 22% Riverine Buloke open

47%

Figure 4.6 The vegetation types encountered in the 37.6 km of folded transects traversed in Hattah Kulkyne National Park which contained Allocasuarina luehmannii. The vegetation type refers to the tree spacing: scattered (trees >100 m apart), open (50-100 m apart), woodland (25-50 m apart) and dense (<25 m apart). I recorded 69 patches of A. luehmannii recruitment in HKNP; all of these patches occurred within vegetation types containing A. luehmannii. Most (35%) of these patches were observed in areas of scattered A. luehmannii which was also the most frequently encountered Buloke vegetation type. When the number of patches was expressed per transect kilometre, the greatest density of patches

- 92 - Chapter 4 Natural recruitment occurred in the Riverine with open Buloke vegetation type (3.1 patches per transect kilometre). The average density of patches over the 37.6 km of vegetation types traversed containing Buloke was 1.8 patches per transect km (Table 4.3).

Table 4.3 The number and percentage of patches of recruitment occurring in each vegetation type, the kilometres traversed in each vegetation type and the patches per kilometre in each vegetation type. Vegetation type No. of patches % of patches Km traversed Patches / km

Buloke dense 1 1 0.6 1.7

Buloke open 11 16 4.0 2.8

Buloke scattered 24 35 16.2 1.5

Buloke woodland 18 26 7.8 2.3

Pine with scattered 1 1 4.5 0.2 Buloke

Riverine with open 14 20 4.5 3.1 Buloke

Total (Average) 69 100 37.6 (1.8)

A total of 644 juveniles in the 69 patches was encountered. Most patches (42%) contained only one juvenile (Figure 4.7) however patch size ranged up to 146 juveniles in a single patch (9.3 ± 22.1, median = 2). I excavated 334 of these and found that there were twice as many seedlings as root suckers (219 seedlings versus 115 root suckers). 35 of the root suckers were on exposed roots and for the remainder the root junction was found between 1 cm and 17 cm below the soil surface (more details presented in section 5.4.1.2). Within the 69 patches, seedlings occurred singly (18 cases), in a patch of all seedlings (24 cases), or in a mixed patch of seedlings and root suckers (7 cases). The remaining 20 patches were comprised of root suckers. The distance of seedlings to the parent tree ranged from 5 m to 75 m (22.5 ± 14.0 m).

- 93 - Chapter 4 Natural recruitment

35

30

25

20

15 Number of patches of Number 10

5

0

1 2 3 4 5 0 0 10 2 3 00 o o o 1 200 t t 6 t to 11 to 15 16 21 to 25 26 31 to 50 1 51 to 101 to 150 15 Number of individuals in a patch

Figure 4.7 Frequency distribution of the number of Allocasuarina luehmannii juveniles in each patch of recruitment. Data on the shape of the seed shadow and the preferred microsite of seedlings were collected for 21 female trees with more than 20 seedlings around them. The distance from the parent tree was summarised as the number of seedlings in 10 m width doughnuts, arranged in concentric circles around the parent tree. The seedlings occurred between 5 m and 63 m from the parent tree showing a unimodal peak between 10 m and 20 m where 42% of the seedlings occurred (Figure 4.8). Each of the 21 patches of seedlings (1113 seedlings in all) occurred in association with an area almost totally devoid of vegetation. For three trees, perennial grasses (primarily Aristida holathera) were present but fewer than 3% of seedlings (32) occurred in association with annual grasses or herbs, shrubby vegetation (primarily Enchylaena tomentosa) or leaf litter. The seedlings in these habitats occurred at the interface with a bare patch.

- 94 - Chapter 4 Natural recruitment

45

40

35

30

25

20

15 Percentage ofPercentage seedlings (%) 10

5

0 <10 m 10-20 m 20-30 m 30-40 m 40-50 m >50 m Distance from parent tree

Figure 4.8 Frequency distribution of the number of Allocasuarina luehmannii seedlings at increasing distances from the parent tree. 4.4.2 When did recruitment occur?

4.4.2.1 Casuarina pauper The height of the 11 seedlings which I excavated ranged between 18 cm and 73 cm (43 ± 18 cm). All seedlings were multi-stemmed indicating browsing at some stage of their development. The diameter of the base of the stem ranged from 9 mm to 37 mm (21 ± 9 mm). It was not useful to group such a small number of seedlings into age classes.

4.4.2.2 Allocasuarina luehmannii The Ranger-in-Charge of HKNP believes that recruitment of A. luehmannii became noticeable in early 2000 (P. Murdoch, Parks Victoria – Hattah, pers. comm. 2002). I sought to verify this observation by comparing the sizes of known and unknown age juveniles and by revisiting vegetation survey sites.

The height of the 219 excavated seedlings ranged between 14 cm and 450 cm (106 ± 72 cm). Most were multi-stemmed indicating browsing at some stage of their development. The diameter of the base of the stem below ground level ranged from 2 mm to 67 mm (23.7 ± 13.2 mm: Figure 4.9). Most seedlings were between 16 mm and 25 mm basal diameter although a range of size classes were present.

- 95 - Chapter 4 Natural recruitment

25

20

15

10 Frequency of (%) occurrence of Frequency 5

0 1 to 5 6 to 10 11 to 15 16 to 20 21 to 25 26 to 30 31 to 35 36 to 40 41 to 45 46 to 50 51 to 55 56 to 60 61 to 65 65 to 70 Diameter of the base of the seedling stem (mm)

Figure 4.9 Size classes of Allocasuarina luehmannii seedlings based on the diameter of the stem measured 2 cm below ground level. In 1991, seven A. luehmannii seedlings were recorded in an exclosure established in 1948 (Moncrieff 1991). Although the juveniles were not tagged, nor was the basal diameter recorded in 1991, in 2005 seven juveniles were located which were between 80 and 295 mm in diameter (mean 180 mm). I found seedlings recruited from direct seeding in 1993 near Lake Lockie which had basal diameters of 75 mm, 77 mm, 86 mm and 130 mm. All of these recruits are larger than the seedlings recorded, suggesting a post 1993 origin of unknown age seedlings.

One of the marker pegs for transect 4 of vegetation monitoring in HKNP conducted between 1992 and 1996 (Sluiter et al. 1997a) occurs within a large patch of seedling recruitment on Florence Annie Track. The 146 seedlings in this patch ranged in basal diameter from 2 mm to 64 mm. It is unlikely that this large patch of seedlings would have been overlooked in the original survey by Sluiter et al., thus the oldest seedlings in this patch germinated some time after 1996.

Of the quadrats surveyed in 2001 containing A. luehmannii (Gowans and Westbrooke 2002), PBW 07 was the only quadrat to contain juvenile A. luehmannii. In 2001, it was reported as containing one individual A. luehmannii with a DBH of 20 mm. No additional juveniles were recorded adjacent to the quadrat. In 2004, I located four A. luehmannii with a DBH of 12, 19, 21 and 24 mm indicating that some recruitment has occurred since 2001.

These observations suggest that recruitment has occurred since 1996 on a number of occasions and corresponds with the observations of field staff that recruitment became noticeable in 2000.

- 96 - Chapter 4 Natural recruitment

4.4.3 Why is seedling recruitment occurring?

4.4.3.1 Classification tree A very simple classification tree was produced indicating that the amount of bare ground within a 30 m radius of the female tree was most useful in separating parent from non-parent females (Figure 4.10). The split was based on parent trees having at least 264 m2 of bare ground within a 30 m radius of the tree. The misclassification rate was very low, indicating that this split was a robust predictor. Only one parent was misclassified as a non-parent and four non-parents misclassified as parents.

Non-parent tree n=73 MC = 28.8%

Non-parent tree Parent tree Bare area ≤ 264 m2 Bare area > 264 m2 n=49 MC = 2.0% n=24 MC = 16.7%

Figure 4.10 The classification tree for recruitment of Allocasuarina luehmannii indicating that >264 m2 of bare ground within a 30 m radius of a female tree is a robust predictor of the presence of seedlings. Each box (node) contains the predictor variable on which the split was based and the binary outcome (bold), the sample size (n) and the misclassification rate (MC).

4.4.3.2 Other important factors The decision tree forest indicated the overall importance of other factors (Table 4.4) calculated from the both primary splitter variables and other variables which substantially reduce the misclassification rate (Brieman 2001):

• More parent trees occurred in runon zones (67%) than did non-parent trees (21%). • Parent trees tended to have a heavier 2005 seed crop than the non-parent trees (Figure 4.11). • Parent trees tended to have a heavier cumulative previous years’ cone crop (37.2 ± 19.2 cones per 20 cm quadrat) than the non-parent trees (16.4 ± 13.5 cones per 20 cm quadrat). • Parent trees tended to have fewer trees within 30 m (2.9 ± 2.6 trees) than the non-parent trees (7.7 ± 7.7 trees). • More parent trees occurred on the flat and fewer on the midslope landforms than did non- parent trees (Figure 4.12). • More parent trees had few (<5) trees within 100 m of the target tree (19%) than did non- parent trees (6%).

- 97 - Chapter 4 Natural recruitment

• Slightly more parent trees occurred on sandy loam (43%) than did non-parent trees (35%). Conversely fewer parent trees occurred on loamy sand (57% versus 65%). • Slightly more parent trees occurred in areas of <1% slope (67%) than non-parent (60%). • Parents tended to have darker seeds (83%) than non-parents (56%).

Table 4.4 The overall importance of variables in predicting the presence of seedlings of Allocasuarina luehmannii. Variable Overall importance

Amount of bare area within 30 m radius of tree 100.0

Hydrology (runoff / runon zone) 71.8

Current (2005) cone crop 69.8

Previous years’ cone crop 50.0

Number of trees within 30 m of target tree 47.5

Landform (ridge, mid-slope, lower slope etc) 46.7

Abundance of trees within 100 m of target tree 39.2

Soil type 32.3

Slope 30.9

Color of seeds 12.4

35

30

25

20

15

10 Percentage of trees of Percentage 5

0 Absent Poor Light Good Heavy 2005 cone crop

Figure 4.11 The percentage of Allocasuarina luehmannii trees which had >20 seedlings (shaded bars) or no seedlings (open bars) categorised by the size of the 2005 cone crop.

- 98 - Chapter 4 Natural recruitment

45

40

35

30 25

20

15

10

Percentage of trees (%) trees of Percentage 5

0 Flat Lower slope Mid slope Ridge Landform

Figure 4.12 The percentage of Allocasuarina luehmannii trees which had >20 seedlings (shaded bars) or no seedlings (open bars) occurring in various landforms.

4.4.3.3 Germination and the maturity of seed I was able to collect 100 seeds from 45 of the 55 trees which produced a cone crop in 2005 and use these for the germination trials. Seeds were from both parent trees (n=15) and non-parent trees (n=30). Germination was rapid with first germinants appearing after three days, 50% germination occurring by the fifth day and all viable seed germinated by 20 days (Figure 4.13). The mean percentage germination for all 45 trees was 45.7%.

Rate of germination (Parents: GI=2.3; Non-parents: GI=2.2) and percentage germination (Parents: 47%; Non-parents: 45%) did not differ between parent and non-parent trees. However, the germination trials were probably confounded by differing maturity of seed at the time of collection.

The seed colour was generally consistent within seed collected from an individual tree but varied between trees. The range of colors observed were dark brown and brown (together classified as “dark”, n=16), light brown (classified as medium, n=14) and orange, pale orange and yellow (together classified as “light”, n=15). There was a clear effect of seed color on both the

Germination Index (F2,44=5.21 p=0.009) and percentage germination (F2,44=5.27 p=0.009). SNK comparisons showed that light colored seeds (mean GI=1.4) had a significantly lower GI than dark (mean GI=2.7) and medium colored (mean GI=2.5) seeds, indicating that they germinate more slowly. SNK comparisons also showed that light colored seeds (mean germination=33%) had a significantly lower percentage germination than dark (mean germination=52%) and medium colored (mean germination=52%) seeds.

- 99 - Chapter 4 Natural recruitment

Of the 55 trees which produced a cone crop in 2005, the degree to which the cones were open varied: 38% were closed, 36% were half-open and 26% were open at the time of collection. I was able to collect enough seed to determine color from 54 trees; in one tree there were cones on the tree but all the seeds had been released from the cones. I compared dark seeds (combining the dark and medium seeds) with light seeds in terms of the condition of the cones at the time of seed collection. Trees where the cones were predominantly open at the time of collection tended to have darker seeds than cones which were closed (Figure 4.14) although the difference was not significant (χ=5.1, df=2, p=0.08).

30

25

20

15

10

5 Percentage germination (%) germination Percentage 0 1234567891011121314151617181920 Days

Figure 4.13 The percentage of viable Allocasuarina luehmannii seeds which germinated each day during the germination trial. Data based on four replicates of 25 seeds from each of 45 trees.

14

12

10

8

6

4 Number of trees of Number

2

0 Closed Half open Open Valves in cone

Figure 4.14 The numbers of Allocasuarina luehmannii trees where the seeds were dark (shaded bars) or light (open bars) in color, classified by how open the cones were when collected.

- 100 - Chapter 4 Natural recruitment

4.5 Discussion 4.5.1 Recruitment of Casuarina pauper

Savannah woodland comprises 80% of the woodland vegetation in MSNP and is considered a degraded example of other woodland communities (Westbrooke et al. 2001). This was the most frequently encountered vegetation type during the 71 km field survey of C. pauper in MSNP. C. pauper trees were present in just over one third of this woodland, most frequently spaced on average 20-50 m apart. The highest encounter rates of seedlings occurred in where C. pauper trees were more than 100 m apart. This corresponds with the suggestion that recruitment mostly occurs in “erosion halos” (Chesterfield and Parsons 1985). Erosion halos are bare areas surrounding isolated trees or on the edge of stands where the trees presumably alter wind patterns, causing erosion of top soil. Erosion halos are not present within stands.

Recruitment of C. pauper was primarily root suckers (93%) which agrees with the observations of others (Chesterfield and Parsons 1985; Cunningham et al. 1992). Most (63%) of the root suckers occurred on exposed roots making them obvious to a casual observer. Few seedlings were detected in MSNP and would be easily overlooked without excavation. The seedlings occurred within 30 m of a female tree. All seedlings were heavily browsed and were a variety of sizes from 9 mm to 37 mm basal diameter. Little further could be deduced from the limited seedling recruitment of this species and it is not considered further.

4.5.2 Recruitment of Allocasuarina luehmannii

4.5.2.1 Where is recruitment occurring? Outside of the mallee vegetation types, Riverine woodland is the most common in HKNP, making up 29% of the area (LCC 1987). This was the most frequently encountered vegetation type during the 135 km field survey of A. luehmannii in HKNP. It is not surprising that a large amount of this vegetation was surveyed whilst actually targeting A. luehmannii because the species is climatically marginal in HKNP and is often restricted to areas adjacent to riverine woodlands (Cheal and Lucas 2005).

Thirty-eight km (29%) contained A. luehmannii trees, most frequently as scattered Buloke woodland where the A. luehmannii trees were spaced on average more than 100 m apart. This broad spacing of trees is probably an artefact of anthropogenic modification (Gowans and Westbrooke 2002). Patches of recruitment occurred in habitats containing A. luehmannii at rates of 1.8 patches per km. The highest encounter rates occurred in riverine woodland with open Buloke (3.1 patches / km) where the A. luehmannii trees were 50-100 m apart. This suggests a greater occurrence of recruitment within floodplains, a possibility which concurs with suggestions that recruitment requires high soil moisture which may be provided in areas where flooding

- 101 - Chapter 4 Natural recruitment saturates the lower soil profile and / or rainfall runon saturates the upper soil profile (Cheal and Lucas 2005). The effects of soil type and hydrology are discussed in section 4.5.2.3.

4.5.2.2 Broad-scale influences on recruitment Recruitment of A. luehmannii was primarily through seedlings (66%). The excavation of over 220 A. luehmannii seedlings of various sizes was particularly unexpected because the literature (e.g. Castle 1989; Raymond 1990; Morcom and Westbrooke 1998; Macaulay and Westbrooke in prep.) conforms to beliefs for many semi-arid species: “Regeneration of this species [A. luehmannii] is thought to occur only as the result of a combination of exceptional rainfall conditions and low grazing pressure” (Sandell 2002).

The observation that recruitment of perennial species within the rangelands occurs more frequently than previously believed has also been reported from in-depth studies of Acacia papyrocarpa (Ireland 1997) and two species of Eremophila (Watson et al. 1997a). Small survey area combined with the high spatial variability of recruitment may contribute to studies “overlooking” recruitment (Watson et al. 1997a, 1997b). In addition, 30% of the root suckers occurred on exposed roots which, as for C. pauper, would make root suckers more obvious than seedlings and could lead to the number of seedlings being underestimated if excavation were not undertaken. Regardless of these considerations, seedling recruitment is clearly evident at HKNP with up to 146 seedlings present in a single patch. Seedlings would be unlikely to be overlooked if they occurred in similar numbers in other areas. This suggests the possibility of unique or unusual factors operating at HKNP that are not present elsewhere.

The broad range of size classes present, comparison of size with known-age seedlings, revisiting field sites and consultation with Parks Victoria staff all point to recruitment occurring on a number of occasions since 1996. Section 2.2.2 outlines rainfall and soil moisture for the period 1963-2004 and shows that 1971 and 1973 were the only years to clearly meet both the wet and dry soil moisture criteria for recruitment of woody seedlings, as defined by Harrington (1991). The years 1980 and 1992 partially met the criteria, and 1974, 1986, 1989 and 1990 only met the summer / autumn wet criterion and were eliminated by a dry period(s) the following summer / autumn. The period from 1996 contained three of the five longest recorded periods of 0% soil moisture: 112 days in 2004, 85 days in 2001/02 and 72 days in 1997. The years 1999 and 2000 were the only years to receive rainfall above the median, however neither of these years experienced the critical period of >50% soil moisture during summer / autumn, and both had periods of >30 days of 0% soil moisture the following summer. Clearly, at no time during 1996- 2004 could the rainfall be considered “exceptional”, or even particularly favourable. Although small-scale variation in rainfall can be substantial in semi-arid areas (Hodgkinson and Freudenberger 1997), a single, heavy, localised fall of rain is unlikely to be the cause of

- 102 - Chapter 4 Natural recruitment recruitment because each patch of seedlings contains a variety of size classes suggesting recruitment over a number of years. In addition, recruitment was not restricted to a specific area of HKNP. Trees in low-lying areas may benefit from run-on from moderate rainfall events and this is discussed in section 4.5.2.4.

The importance of low grazing pressure is highlighted for recruitment of rangeland species (Section 1.2.1) and circumstantial evidence appears to support this factor for recruitment of A. luehmannii at HKNP. Historically, browsing pressure from mammalian herbivores in HKNP was extremely high. The Hattah Kulkyne area was considered to have had the worst rabbit infestation in Victoria (LCC 1977) and kangaroos experienced an exponential increase in numbers following rabbit control in HKNP in the 1980s (Cheal 1986; Mueck 1987). However, in 1996 Rabbit Haemorrhagic Disease arrived in the mallee, reducing the abundance of rabbits to fewer than 0.5 per km (Figure 4.15, Sandell 2002). This was substantially lower than at any other time since rabbit control works began in HKNP in earnest in 1980. Culling of kangaroos was also extended to the whole of HKNP in 1996 and prior to that a severe drought in 1994-5 substantially reduced the abundance of kangaroos (Chapter 2, Sluiter et al. 1997a). Consequently, the abundance of mammalian grazers has been lower since 1996 than it has been at any time since HKNP was declared.

4.5.2.3 Bare areas affect recruitment The presence of a large area almost devoid of vegetation within the vicinity of a female tree was found to be the most important factor predicting the occurrence of A. luehmannii seedlings; 97% of seedlings occurred in these bare areas. Perennial grasses were present occasionally but not annual grasses or herbs, shrubs or leaf litter.

Seedling recruitment may be limited by the availability of suitable microsites close to a tree with an abundant supply of seed (Eriksson and Ehrlén 1992). Dispersal of seeds is governed by wind velocity and direction, height of the seed source and seed morphology (Nathan et al. 2000). Most wind-dispersed seeds fall close to the parent tree in a right-skewed, leptokurtic pattern (Handel 1985) and this pattern was present for A. luehmannii seedlings. Most seedlings occurred between 10 m and 20 m from the parent female, indicating that the availability of microsites within this range is important. The occasional long-distance dispersal of Casuarinaceae seeds via ants, emus or transport by water (Auld 1995c) was also supported with one seedling 75 m from the nearest female tree.

Plant cover in water limited ecosystems is usually less than 60% and is commonly arranged as patches of high vegetation cover in a low-cover matrix (reviewed in Aguiar and Sala 1999). Recruitment varies within and between patches according to seed availability and survival of seedlings. Seeds are retained at a lower density on areas of bare soil because wind speeds are high

- 103 - Chapter 4 Natural recruitment whereas vegetated patches trap seeds (Aguiar and Sala 1997) are associated with high soil organic matter, mineral nutrients and water infiltration (van de Koppel et al. 2002) and low evaporation rates and a favourable microclimate (Tongway and Ludwig 1994; Ludwig and Tongway 1995; Mandujano et al. 1998; Facelli and Brock 2000; Castro et al. 2002) which facilitate establishment of seedlings. Vegetation can also protect seedlings from browsing, acting as “nurse plants” (Chesterfield and Parsons 1985; Auld 1990, 1995b; Maestre et al. 2001; Castro et al. 2002; Gomez-Aparicio et al. 2004). However, seedlings growing in vegetated patches may be adversely affected by competition for light, nutrients or water (Holmgren et al. 1997; Aguiar and Sala 1999; Maestre et al. 2005). Seedlings may also be suppressed by physical, chemical or biological interactions with leaf litter (Xiong and Nilsson 1999; Barritt and Facelli 2001).

Whether bare areas are a cause or effect of seedling establishment is unknown but association of A. luehmannii seedlings at HKNP with bare areas suggests that the effects of competition are more detrimental than the advantages of facilitation from existing vegetation (Aguiar and Sala 1997; Maestre et al. 2005). The widely-held belief that facilitation is more important than competition in stressful environments such as arid areas (Bertness and Callaway 1994) has recently been shown not to hold. In strongly water-limited environments, facilitation should only occur only when neighbours increase soil moisture beyond their own water uptake requirements (Maestre et al. 2005).

Seedlings of both C. pauper and A. luehmannii establish following the ripening of seed in summer (Chesterfield and Parsons 1985; Raymond 1990; Auld 1995c) and adequate soil moisture during both the first and second summer / autumn period is critical for the survival of woody seedlings (Harrington 1991; Davis et al. 1998; Davis et al. 1999; Rey Benayas et al. 2001; Maestre et al. 2003; Schwinning et al. 2005). Competition from annual weedy species during this time may be particularly important. As grazing pressure increases on rangelands, perennial grasses are progressively replaced with annual species (Illius and O’Connor 1999; Friedel et al. 2003; Landsberg and Crowley 2004). In arid areas, summer-growing annuals deplete soil moisture more rapidly than perennial grasses (Davis et al. 1998). In addition, the loss of perennial grasses and associated organisms reduces the percolation of rainfall into the soil profile (Maestre et al. 2001). This is associated with a shortened period of water supply to woody vegetation, causing drought stress during low rainfall periods (Anderson and Hodgkinson 1997). These alterations to seasonal water relations have been shown to inhibit establishment of seedlings of woody, perennial species (Davis et al. 1998; Gordan and Rice 2000; Rey Benayas et al. 2001).

The implications for woody seedlings of competition with annual weeds need to be investigated further. There are two factors to be considered. Firstly, since ground flora surveys were completed in HKNP in 1996 (Sluiter et al. 1997a), there has been a substantial increase in native perennial grasses, particularly Aristida holathera (Chapter 3). Woody seedlings are believed to establish

- 104 - Chapter 4 Natural recruitment more readily in the interstitial spaces between perennial grass tussocks, than the dense annual grasses (Gordan and Rice 2000; Maestre et al. 2001). Is it possible that a shift from annual to perennial grasses has contributed to higher soil moisture availability for woody seedlings? Or could restoration works be facilitated by the presence of perennial grass tussocks (Maestre et al. 2001)? Secondly, HKNP still has substantial infestation of weedy annuals. These are controlled by a single application of knockdown herbicide prior to direct seeding, or local removal at the time of hand-planting. No follow-up treatment is applied. Could better control of weeds enhance the success of restoration works (Rey Benayas et al. 2001)?

4.5.2.4 Other factors affecting recruitment The second most important variable in predicting the occurrence of A. luehmannii seedlings around female trees is hydrology; seedlings are associated positively with runon zones. Runon zones tend to have higher levels of soil moisture for a given rainfall event than surrounding areas (Hodgkinson and Freudenberger 1997; Chesson et al. 2004) and observations suggest that seedling recruitment occurs more readily in these areas (Chesterfield and Parsons 1985; Ludwig 1987; Woodell 1990; Ireland 1997; Tiver and Andrew 1997). The occurrence of seedlings in flat areas rather than sloping ground (sixth most important variable) is probably related to flat areas receiving more runon than sloping land. The inferred importance of soil moisture suggest further investigations may be warranted into the possibilities of water harvesting, which utilises artificially created runon zones, to enhance restoration success (Whisenant 1999; Vallejo et al. 2006).

The third and fourth most important variables are enhanced production of cones (current and previous season’s crop), and therefore seeds, by a female tree. Variation in seed production can be attributed to maternal genetics, microhabitat (Haig and Westoby 1988) and pollination. The latter factor may be relevant because pollen is wind-dispersed in the Casuarinaceae and both C. pauper and A. luehmannii are primarily dioecious. The production of seed therefore requires pollen from neighbouring male trees to be transported by wind to a female tree. Habitat fragmentation and isolation of dioecious trees can reduce pollen availability and therefore seed production (Knapp et al. 2001; Sork et al. 2002a; Sork et al. 2002b). Pollination tends to be enhanced if males are nearby (Wilcock and Neiland 2002) or numerous (Knapp et al. 2001). Intervening vegetative structure can also affect pollination, with closed forests tending to have shorter average pollen dispersal distances (Smouse et al. 2001) and reduced seed production (Arista and Talavera 1996). Parent trees had fewer trees within 30 m and 100 m than did non-parent trees (fifth and seventh most important variables respectively). The weak association of seedlings with trees which are isolated from other trees may suggest that competition with adult trees (e.g. Gordan and Rice 2000) is more important than pollen limitation. It has been demonstrated that even small amounts of pollen can produce good seed set (Sorenson and Webber 1997).

- 105 - Chapter 4 Natural recruitment

Maternal genetics and microhabitat are worthy of further investigation. Seed production in Allocasuarina spp. varies between trees in terms of the cone crop, individual seed mass and number of seeds per cone (Clout 1989; Pepper et al. 2000; Crowley and Garnett 2001; Maron and Lill 2004). Seed production parameters of individual trees of A. luehmannii can vary from year to year (Maron and Lill 2004), although other species of Allocasuarina are reported to be consistent on a year-to-year basis (Crowley and Garnett 2001). This factor may be worth investigating further as a tree which produces reliably good quantities of seed is an asset for collecting seed for revegetation.

Soil type was of lesser importance. Small scale variation in soil can be important for recruitment of seedlings (Lauenroth et al. 1994; Maestre et al. 2003; Tongway et al. 2003) however the field techniques used in this study identified little variation between sites with only sandy loam and loamy sand being recorded. Laboratory techniques to assess soil texture such as using the hydrometer method to determine particle size distribution may be useful. Higher rate or percentage germination of seed was not associated with parent trees, however as noted previously the germination trials were probably confounded by differing maturity of seed at the time of collection.

4.5.3 Seed collection for revegetation

Generally, casuarinas produce many thousands of viable seeds each year per tree (Torrey 1983; Turnbull and Martensz 1983; Boland et al. 1996); this includes both C. pauper (Auld 1995c; Callister 2004) and A. luehmannii (Castle 1989; Hayes 1993; Macaulay and Westbrooke in prep.). Seed is readily collected and stored (Turnbull and Martensz 1982) but cones and seeds of both A. luehmannii and C. pauper are not held on the tree (pers. obs.). Seed collection can therefore be limited to a narrow window after the seed is ripened but before it is released from the cones. Determining when cones are ripe can be difficult and the temptation may be to collect seed too early to avoid seed losses. However, I found that the maturity of A. luehmannii seed had a substantial effect on germination. Dark-colored seed (dark brown, brown and light brown) had a higher percentage germination and germination was more rapid than for light-colored seed (orange, pale orange and yellow). Dark colored seed most often was derived from trees where some or the majority of cones had opened. Therefore gains in the amount of seed collected early are likely to be negated by the reduced germinability of unripe seed. There are two recommendations to maximise the germinability of collected seed: collect only from trees where at least some of the cones have opened, and check that the seed inside the cones is of a dark color. The relationship between seed color and germinability may not hold for species other than A. luehmannii, e.g. Casuarina spp. have light colored seeds (Grant 2004).

- 106 - Chapter 5 Root suckers

Chapter 5 The use of root suckers to enhance recruitment

5.1 Scope and objectives

The state-and-transition model presented in Chapter 3 indicated that in HKNP the transition activated by enhancing recruitment of seedlings via direct seeding or hand planting often fails due to low rainfall during the establishment phase or high grazing pressure (Thomas 2003). The model assisted to highlight a restoration opportunity – the possibility of artificially initiating root suckers to enhance restoration success in arid areas.

Clonal growth occurs in both Casuarina pauper and Allocasuarina luehmannii via root suckers, believed to arise from shallow lateral roots following disturbance (Torrey 1983; Auld 1995a, 1995c; Morison and Harvey 2001). Clonal growth in other species offers improved establishment of juveniles compared to recruitment of seedlings (Callaghan 1984; De Steven 1989; Barsoum 2001), through recovery or escape from browsing (Koop 1987; Chazdon 1991; Peterson and Jones 1997; Canadell and Lopez-Soria 1998), or enhanced survival under stressful conditions (Ashmun et al. 1982; Salzman and Parker 1985; Tissue and Nobel 1988; Pennings and Calloway 2000). The suggested enhanced establishment and survival of root suckers over seedlings makes them an attractive option for establishing juveniles in harsh environments. I was interested in determining whether roots suckers could be initiated as required from the roots of existing trees of C. pauper and A. luehmannii and whether this technique could be used for restoration of degraded woodlands.

To resolve this, I characterised root suckers which had established naturally in the field to compare the natural conditions under which suckers grow with the results from artificial treatments. I also treated roots from male and female trees in both autumn and spring and measured the response of roots to damage and exposure, particularly whether they produced root suckers. My objectives were to:

1. Describe the root system and parent tree (gender, distance) of root suckers which have established naturally, and

2. Determine experimentally which diameter root, gender of parent tree, season of treatment and damage and exposure combination produced the greatest number of persistent root suckers.

-107- Chapter 5 Root suckers

5.2 Introduction 5.2.1 What is clonal growth?

True clonal growth allows for the spatial spread and multiplication of genetically identical individuals (ramets) that are, or have the capacity to be, independent (Peterson and Jones 1997; Del Tredici 2001). Many plant species are able to recruit individuals into a population both sexually and asexually. Asexual reproduction may involve clonal growth such as the production of bulbs, rhizomes, tillers or root suckers (Klimes et al. 1997), or agamospermy which is the production of seed without fertilisation. Resprouting, the replacement of a damaged trunk after the loss of above-ground biomass (for example Bellingham and Sparrow 2000; Yamada et al. 2001) allows perpetuation of a single individual, however it is not true clonal growth. Plants which grow from lignotubers following disturbance are generally considered resprouters rather than clonal although some lignotubers do break up and form independent ramets (Lacey and Johnston 1990; Tyson et al. 1998).

Clonal growth is common in many communities and may be particularly important in harsh conditions such as nutrient-poor, cold, shaded or aquatic conditions (Koop 1987; Peterson and Jones 1997), and especially on the edge of a species’ range (Mitton and Grant 1996; Wesche et al. 2005). Estimates of the generality of clonal growth suggest that 67% of the flora of central Europe and 70% of temperate zone species are capable of clonal growth (Klimes et al. 1997).

5.2.2 Advantages and disadvantages of clonal growth

Clonal growth has many potential benefits, most of which are due to physiological integration (PI) between the parent and juvenile. PI allows the movement of photoassimilates, inorganic nutrients and water between connected ramets, especially from mother to daughter ramet (Ashmun et al. 1982; Salzman and Parker 1985; Tissue and Nobel 1988; Caraco and Kelly 1991), although bidirectional transport occurs (van Kleunen and Stuefer 1999). PI is linked to increased fitness of a genet. It allows plants to exploit patchy resources such as light, nutrients or water (Reinartz and Popp 1987; De Kroon et al. 1996; van Kleunen and Stuefer 1999) and recover after damage such as herbivory (Chazdon 1991; Canadell and Lopez-Soria 1998). PI leads to high initial growth rates that allow ramets to escape quickly from size classes that are vulnerable to browsing (Koop 1987; Chazdon 1991; Peterson and Jones 1997). The ability to produce root suckers allows some trees to establish despite high grazing pressure, through risk spreading where a small percentage of abundant ramets is likely to escape damage (Piqueras 1999). The translocation of resources between connected individuals tends to increase in stressful conditions (Ashmun et al. 1982; Tissue and Nobel 1988) and can ameliorate environmental stresses such as salinity (Salzman and Parker 1985; Pennings and Calloway 2000). Clonal plants are often better adapted than sexually reproducing species, allowing survival and establishment of juveniles in adverse sites (Auld

-108- Chapter 5 Root suckers

1995a; Peterson and Jones 1997; Bond and Midgley 2001). Clonal growth enables species to persist where seedling establishment is not possible or limited due to grazing, short growing season, low temperature, shade, little soil moisture or low nutrient levels (Callaghan 1984; Koop 1987; Jelinski and Cheliak 1992; Klimes et al. 1997; Peterson and Jones 1997; Jones and Gliddon 1999; Kitamura et al. 2000; Dorken and Eckert 2001; Young et al. 2002).

However, PI does not always mean better adaptation (references in Schmid and Bazzaz 1987). Transfer of nutrients is not universal (Klimes and Kilimesova 1999), nor is improved competitive ability (Pennings and Calloway 2000; Peltzer 2002). The benefits of PI tend to be correlated positively with the degree of integration. Integration is greatest where the ramets are closely spaced and the genet is compact (“phalanx” growth form), and least in the where the ramets are widely spaced and spreading (“guerrilla” growth form, Schmid and Bazzaz 1987; van Kleunen and Stuefer 1999).

Other advantages of clonal growth include release from the requirement for a mate when population size is small or where pollinators are limited. Anchorage and stability may be increased in erosion prone sites, especially for woody plants (Yamada et al. 2001). Competitive ability may be enhanced where clonal thickets allow competitive exclusion of other species and a decreased risk from fire for ramets in the centre of a clump (Peterson and Jones 1997).

An important disadvantage of clonal growth is that clonal reproduction may restrict the resources available for sexual reproduction (Klimes et al. 1997), indeed modelling of rates of increase suggest that this tradeoff is so great that there is no absolute fitness advantage of clonal growth (Silvertown et al. 1993). Clonal growth also does not preclude losses due to self-thinning for clonal trees and shrubs (e.g. Jones and Raynal 1987; Shepperd and Smith 1993; Peterson and Jones 1997) and physiological integration may increase the transmission of diseases (Klimes et al. 1997). Another disadvantage is the potential reduction in genetic variation as a result of clonal growth which may limit the ability of the population to survive environmental challenges (Hamrick and Godt 1996; Young et al. 2002).

There are physical aspects of clonality which have important implications for conservation, especially for developing appropriate management and recovery strategies for populations of threatened plants. The following disadvantages of clonal growth have been suggested (Handel 1985; McClintock and Waterway 1993; Stewart and Porter 1995; Escaravage et al. 1998; Sydes and Peakall 1998; Hogbin and Peakall 1999; Jones and Gliddon 1999; Young et al. 2002):

• Genetic individuals in a threatened population may not be identified correctly (impacting on reserve design, reintroductions, sampling strategies for ex situ collections and identifying the conservation value or status of a population), (Hogbin and Peakall 1999)

-109- Chapter 5 Root suckers

• Clonality in self-compatible or dioecious species may limit fruit set because pollen flow is restricted within clones, • Extensive clonal mats or stands may limit sites available for seedling establishment, and • Clonality in self-compatible plants may increase the level of inbreeding depression.

5.2.3 Root suckers

5.2.3.1 Anatomy of roots and root suckers The root of a dicotyledon (Figure 5.1) includes the following tissues (Esau 1977):

• On the outside is the periderm which is the protective tissue formed during secondary growth and includes the phellogen (cork cambium) which forms phellum (cork) towards the outside and phelloderm towards the centre. • The pericycle is the site of origin of lateral roots. • The middle layer is secondary phloem and inside that, vascular cambium which forms the secondary tissues, phloem towards the outside and xylem towards the centre. • The secondary xylem is in the centre and forms the “wood”.

Figure 5.1 The anatomy of a dicotyledonous tree root showing the major tissues. In a number of plants, clonal growth occurs from buds on roots which, when released from inhibition, develop into root suckers (

-110- Chapter 5 Root suckers

Table 5.1). Callus tissue is an important precursor to the growth of suckers in some species. Following injury to a root, the living cells beneath the periderm are exposed and die. Callus tissue arises from phloem and xylem in the vascular region and a new periderm forms beneath. Callus tissue may form on damaged roots or at the ends of severed roots, or even at the base or bend of lateral roots as a natural consequence of their growth (Esau 1977; Bosela and Ewers 1997).

-111- Chapter 5 Root suckers

Table 5.1 The traits and origin of additional (develop spontaneously during early growth) and reparative (develop in response to injury) root buds leading to root suckers. Additional buds Reparative buds

Traits Form during the early Form de novo in response to senescence, injuries, or growth of roots; typically other disturbances, presumably at any time during endogenous, most secondary growth, typically exogenous, bud traces subtended by bud traces absent or not to the centre of a root. to the centre of the root.

Origin Pericycle or secondary Phellogen Phellogen or Callus tissue phloem secondary phloem

(unusual)

Species Liquidamber styraciflua1, Populus Araucaria Fagus Alnus incana2, Sassafras tremuloides4 cunninghamii5 grandifolia6 albidum3

Table modified from Bosela and Ewers (1997). Sources 1: Kormanik and Brown (1967), 2: Paukkonen et al. (1992), 3: Bosela and Ewers (1997), 4: Schier (1981), 5: Burrows (1990), 6: Jones and Raynal (1986).

5.2.3.2 Initiation and growth of root suckers Root suckers develop from shallow lateral roots (Kormanik and Brown 1967; Schier 1981; Jones and Raynal 1986, 1988; Shepperd and Smith 1993; Des Rochers and Lieffers 2001). There are many physiological and environmental factors which influence the initiation of root suckers (Figure 5.2). Influences are related to the parent tree, severity of disturbance and season of disturbance. The depth and diameter of the root and the distance to the parent tree also influence the production of root suckers.

-112- Chapter 5 Root suckers

Season of damage * • Carbohydrate/nutrient/hormone levels in parent tree • Growing season (soil moisture, temperature)

Severity of Parent tree disturbance ROOT SUCKER • Gender * • Wounding * • Health • Severing * • Age (size) • Exposure * Root • Genetics • Loss of above • Depth * ground parts • Diameter * • Distance to parent tree *

Figure 5.2 Factors affecting initiation of root suckers divided into severity of disturbance, season of disturbance, factors related to the parent and factors related to the particular root (Schier 1981; Frey et al. 2003). An asterisk indicates factors investigated in this experiment. The initiation of root suckers requires some form of disturbance to the plant (Bosela and Ewers 1997; Fraser et al. 2004). Disturbance may remove above ground parts (e.g. fire, clearfelling, disease), damage the roots themselves (e.g. erosion exposing the root, or mechanical damage) or be external (e.g. pulses of nutrients or soil water) (Table 5.2). Disturbance is believed to disrupt apical dominance which is mediated, probably indirectly, by the plant hormones auxin and cytokinin. The major auxin in higher plants, indole acetic acid (IAA), is synthesised in shoot apical tissue and actively transported distally in the phloem to the roots where it inhibits the initiation of sucker buds (Frey et al. 2003). Loss of above ground parts decreases the production of auxin releasing buds from inhibition. Alternatively the flow of auxins can be interrupted by root injuries to the phloem or xylem tissue (Schier 1981; Frey et al. 2003; Fraser et al. 2004). Exposure of roots to heat and sunlight may cause auxin to degrade or translocate (Kormanik and Brown 1967) although the role of temperature has been disregarded for Trembling Aspen Populus tremuloides (Fraser et al. 2004). Cytokinins are believed to promote suckering; these are transported proximally to the base of the tree (Schier 1981). Recently, the role of hormones has been questioned, and substances like ethylene produced during the activation of wound-induced genes suggested as being important as a stimulus of root suckering (Fraser et al. 2004).

The season of disturbance influences a number of disparate factors. Auxin concentration is lowest in autumn following leaf senescence and highest in the spring associated with the flush of new leaves in deciduous species (Schier 1973b, 1981). Carbohydrate reserves also fluctuate seasonally in deciduous species particularly (Bonicel et al. 1987), tending to be lowest in spring and highest in autumn. Whilst some studies have found no relationship between carbohydrate reserves and numbers of root suckers produced (Tew 1970; Schier 1973b; Fraser et al. 2002), growth of

-113- Chapter 5 Root suckers suckers in coppiced trees was greatest following autumn treatment which corresponded with greatest carbohydrate reserves (P. tremuloides,Landhausser and Lieffers 2002). Root suckers from excised P. tremuloides roots grew equally well in spring and autumn (Tew 1970), or peaked in autumn (Schier 1973b). Damaged roots of American Beech Fagus grandifolia produced the most root suckers in spring and survival was also highest at this time (Jones and Raynal 1988) suggesting that factors other than carbohydrate reserves may be involved. Nutrient reserves (Canadell and Lopez-Soria 1998; Frey et al. 2003) and the availability of soil nutrients and water (Fraser et al. 2002; Cruz et al. 2003a) are important for the growth rate of suckers and could be expected to vary seasonally. Very dry or saturated soil inhibits the initiation of suckers, and higher temperatures stimulate early initiation of suckers (but not total numbers), improving growth and survival (reviewed in Fraser et al. 2002; Frey et al. 2003). However C. pauper has been observed to sucker prolifically following flooding (M. Westbrooke, University of Ballarat, pers. comm. 2005).

Individual variation in clonal growth has been reported which could be attributed to the health, size and genetics of the parent tree (e.g. McLellan et al. 1997). Gender of the parent plant also appears important. Female, woody, dioecious plants are believed to incur greater reproductive costs and have fewer resources available for maintenance and growth compared to males. Males tend to be larger, grow faster, are able to survive harsher environments than females and are expected to be able to contribute more resources to clonal growth (Freeman et al. 1976; Obeso et al. 1998). Enhanced survival and clonal growth should lead to male biased sex-ratios in clonal species although supporting evidence (Obeso et al. 1998) is overwhelmed by dissenting views (Sakai and Burris 1985; Reinartz and Popp 1987; Sakai and Sharik 1988). Nevertheless, dense swards of males produced by root suckering have been noted in Casuarina glauca (Boland et al. 1996).

Shallow, small diameter roots are universally recognised as producing root suckers. Root suckers are reported as developing from roots shallower than 15 cm (Jones and Raynal 1986; Wisniewski and Parsons 1986), 30 cm (Schier 1981) or 60 cm (Kormanik and Brown 1967). Suckers are reported as being produced from roots with diameters between 5-40 mm (Sweetgum Liquidamber styraciflua: Kormanik and Brown 1967), 4-20 mm (P. tremuloides: Shepperd and Smith 1993) and 5-150 mm (F. grandifolia: Jones and Raynal 1988). The average root diameter for existing suckers of F. grandifolia was 29 mm (Jones and Raynal 1986) and 8.6 mm for P. tremuloides (Des Rochers and Lieffers 2001).

-114- Chapter 5 Root suckers

Table 5.2 Various forms of disturbance believed to initiate the production of roots suckers in a variety of species. Disturbance Species References

Fire Populus tremuloides (Schier 1981)

Alectryon oleifolius (Chesterfield and Parsons 1985; Wisniewski and Parsons 1986)

sclerophyllous species (Lacey and Johnston 1990)

rainforest trees (Williams 2000)

Clearfelling or coppicing Populus tremuloides (Schier 1981)

Alnus incana (Paukkonen et al. 1992)

Disease or senescence of the Fagus grandifolia (Jones and Raynal 1986) parent tree

Damage or severing of roots Araucaria cunninghamii (Burrows 1990)

Alectryon oleifolius (Wisniewski and Parsons 1986)

Allocasuarina luehmannii (Morison and Harvey 2001)

Populus tremuloides (Fraser et al. 2004)

Exposure of roots Liquidamber styraciflua (Kormanik and Brown 1967) (experimentally or as a result Fagus grandifolia (Jones and Raynal 1986) of erosion) Alectryon oleifolius (Wisniewski and Parsons 1986)

Casuarina pauper (Chesterfield and Parsons 1985)

Branching of lateral roots Xanthoxylum americanum (Reinartz and Popp 1987)

Pulses in soil resources, Rumex acetosella (Klimes and Kilimesova 1999) including seasonal production Acacia carneorum (Auld 1990)

5.2.3.3 Root suckers and browsing Browsing by mammalian herbivores is a particular problem in semi-arid woodlands (section 1.2.1). The literature suggests that although root suckers are susceptible to heavy browsing, they are less so than seedlings. Following browsing, carbohydrate reserves in roots, stem or lignotubers are mobilised (van der Heyden and Stock 1996; Tolvanen and Laine 1997; Sakai 1998). It is exhaustion of carbohydrate reserves following multiple defoliation events that results in the death of the plant (Canadell and Lopez-Soria 1998); plants with artificially reduced carbohydrate

-115- Chapter 5 Root suckers reserves have limited resprouting ability (Cruz et al. 2003b). Physiological integration of root suckers with the parent plant increases access to carbohydrates and suggests that suckers should be better able to resprout following browsing than seedlings. Artificially defoliated ramets can access communal carbohydrate reserves for recovery, however growth and survival is reduced if all ramets of a genet are defoliated (Chazdon 1991). The ability to recover after damage depends on the intensity, frequency and timing of damage (Tolvanen 1994). Repeated defoliation by herbivores can eliminate suckers (Chesterfield and Parsons 1985; Auld 1990, 1993; Baker et al. 1997; Hessl and Graumlich 2002; Lord 2002). Repeated browsing of C. pauper suckers caused substantial reductions in growth rates and some mortality (Auld 1995a; Denham and Auld 2004).

5.3 Methods 5.3.1 Field characteristics of suckers

During field surveys described in section 4.3.1 a number of juvenile root suckers were located. The root suckers were described and the results are presented in this chapter. Chapter 4 is primarily concerned with seedling recruitment, whereas this chapter deals with the characteristics and initiation of root suckers. These measures are summarised:

• Proximal and distal root diameter to estimate the range of diameters of parent roots which produce root suckers, • Depth of the connection to the parent root to estimate the depth from which root suckers can form, • Distance to parent tree to estimate the average and maximum distance from the parent tree that root suckers can form, • Gender of the parent tree, • Any evidence of the root sucker developing its own roots, and • Height of longest branchlet and number of branches to estimate impacts of browsing. No measure was possible to group root suckers into age classes because the diameter of the parent root does not reflect the age of the sucker and no basal stem as such was present.

5.3.2 Experiment 1: Initiating callus and suckers

I supplemented the distribution of the vegetation communities described in 1999 with personal observation to map the occurrence of A. luehmannii and C. pauper in Hattah Kulkyne National Park (HKNP) and Murray Sunset National Park (MSNP). Tools within MapInfo were used to construct polygons around areas containing trees which were within 350 m of a track and to select random points, buffered by 10 m, within the polygons. The nearest, isolated tree (not within 10 m of a tree of any species) to the sampling point was located and a single root excavated. I recorded

-116- Chapter 5 Root suckers the gender, Diameter at Breast Height (DBH) and condition of the parent tree, and the surrounding soil and vegetation type.

One root between 20 mm and 55 mm in diameter was located per tree by digging with a shovel around the tree (less than 40 cm deep). A 120 cm long section was exposed between 3.0 m and 4.2 m from the parent tree. The section was marked using numbered aluminium tags attached at both ends using light wire twitched to allow growth in the diameter of the root (Figure 5.3). The root was photographed and the diameter was recorded at the end closest to the tree (proximal end), mid-point and end farthest from the tree (distal end) and averaged for the overall root diameter. The location of any old damage, branching or adventitious roots was noted. Sixty-four roots of each species were excavated in autumn 2003 (April – May) and spring 2003 (September – October), totalling 256 roots. Half the roots were from female trees and half were from male trees. All roots were protected from mammalian herbivores by a cage constructed of wire netting; insects were not excluded (Figure 5.4). The root was treated with a combination of treatments: exposure (exposed, buried) and damage (undamaged; four x 5 cm long scrapes spaced evenly along the root; one 20 cm scrape in the midsection; severed completely in the midsection). The scrapes were to an even depth on each root: patches of periderm were removed from the top of the root to a depth that damaged the cambium and exposed the xylem, or secondary wood. Damaging the cambium is important to stimulate sucker growth in other species (Schier and Campbell 1976). Cut roots were completely severed at the mid-point and I endeavoured to keep the cut ends physically separated.

To determine which treatment to apply to each root, I first grouped the roots according to the gender of the parent tree within each species and in each season. The roots were ordered according to the average diameter of the root and divided into four size classes. One root from each size class was randomly allocated to each of the eight treatment-exposure combinations. This ensured that each damage-exposure combination included an even number male and female roots and a range of root diameters.

I monitored the roots every two to three months following treatment to ensure the cages were still secure, and to check for visible changes in the exposed roots. The location of any root suckers, new roots or root buds was noted, and a photograph was taken of each exposed root. Roots were scored as to whether they produced callus on the inflicted damage, either from the periderm around the scrapes or from the cut faces. Callus was present (good and excellent callus) or absent (no or poor callus). The exposed roots were assessed as dead or alive by scratching the root with a sharp knife; the scratch showed red and wet if the root was alive. For each sucker that developed, the time of eruption, location, number of branchlets, and length of the longest branchlet was recorded.

-117- Chapter 5 Root suckers

At eighteen months after treatment, the buried roots were excavated and the total number of roots which had produced suckers, callus tissue or both was scored. For the roots which had no living suckers, I cut the root 20 cm proximal and distal to the target section, at the proximal and distal ends and at the midpoint of the target section to determine which parts of the roots were alive. “Alive” indicated that more than 80% of the root was alive, “mostly dead” roots had <50% of the root alive. “Dead” roots showed no signs of living vascular tissue when sampled destructively at 18 months post-treatment. Roots which produced suckers were monitored for 29 months post- treatment for the autumn treatment and 25 months post-treatment for the spring treatment.

Figure 5.3 The method used of twitching the wire to securely attach tags to the section of root, whilst still allowing growth of the root.

Figure 5.4 A cage used to cover the root and prevent browsing by mammalian herbivores. 5.3.3 Experiment 2: Casuarina pauper trial

The literature suggests that root suckering is more prevalent in C. pauper than in A. luehmannii (Cunningham et al. 1992). I conducted a further trial to initiate root suckers from C. pauper only. The trial used a larger sample size and aimed to determine whether either gender of the parent tree or diameter of the target root had an impact on the survival of the root or the production of callus or suckers.

-118- Chapter 5 Root suckers

Four areas containing numerous, well spaced trees of C. pauper were chosen. In October 2003, I selected twenty male and twenty female trees within each area, each >20 m apart. One root between 16 mm and 36 mm in diameter per tree was located and an 80 cm length of root was exposed between 3.0 m and 3.8 m from the parent tree. A combination of scraping and exposure was believed to be most likely to initiate sucker growth in this species (pers. obs. and Jones and Raynal 1987). I removed a single 10 cm long scrape of periderm from the midpoint of the root. All roots remained exposed and were tagged, caged and described as in section 5.3.2.

The roots were monitored every two to three months for 18 months following treatment as described in section 5.3.2. At eighteen months after treatment, the number of roots which had suckers, callus tissue or both was scored. For roots which had grown suckers by 18 months post- treatment, I continued to monitor the number, growth and form of suckers for a further six months, totalling 24 months post-treatment.

5.3.4 Analysis of Experiments 1 and 2

5.3.4.1 Descriptive results The responses of roots to each damage-treatment combination were described and presented as a tree diagram. Tree diagrams are a graphic organiser used to stratify results into progressively greater detail. The tree progressively branches to terminal nodes which show all recorded outcomes of an experiment. The tree diagram was used to show the interaction between root health, the production of callus and the production of suckers. Within each tree diagram, the results from male and female parent trees and the autumn and spring treatment were combined because, looking ahead to the results, the classification trees indicated that these factors were less important.

5.3.4.2 Classification trees The program DTREG (Sherrod 2003) was used to create classification trees to determine the most important predictor variables for the response variables: root health (alive, mostly dead or dead), callus tissue (yes or no) and sucker (yes or no), all measured at 18 months post-treatment. The predictor variables for Experiment 1 were gender of parent tree (male, female), season (autumn, spring), treatment (1-scrape, 4-scrape, cut, none) and exposure (exposed, buried). The predictor variables for Experiment 2 were gender of parent tree (male, female) and the continuous variable: diameter of the target root. The criteria for building and pruning the tree described in section 4.3.4 were used. Classification trees were used to guide the use of predictor variables for the log-linear models and assist in the interpretation of the data.

-119- Chapter 5 Root suckers

5.3.4.3 Log-linear models Categorical data with multiple variables and categories for each variable are generally presented as a contingency table. Log-linear models, a form of generalised linear model (GLM) designed for the analysis of count data, are the best method of analysing complex contingency tables (Quinn and Keough 2002). They use a complex iterative algorithm (Newton-Raphson) to model the expected cell frequencies against the variables to maximise the likelihood of observing the dataset in question (Agresti 1996). Log-linear models use log-link to connect the response variable and its probability distribution to the categorical predictor variables.

The log-linear model tests the basic null hypothesis of no association (interaction) between any of the variables. The backward elimination process for log-linear modelling is to fit a full (saturated) model for the expected frequency in each cell which exactly fits the observed data. Next a series of reduced models are fitted where the interaction terms are progressively removed. The fit of each reduced model is contrasted with the saturated model in a hierarchical manner by comparing the observed and expected frequencies and calculating the log-likelihood of each model. The G2 statistic compares the fit of two models.

G2 = -2(log-likelihood reduced model - log-likelihood saturated model)

The G2 statistic for the saturated model is zero so the G2 statistic for any model is the difference in fit between that model and the saturated model, also called the deviance. A deviance close to zero indicates that the reduced model fits as well as the saturated model. Deviance is compared to a χ2 distribution and a non-significant result indicates that omitting the interaction term makes little difference to the fit of the data, consequently the null hypothesis of no interaction is accepted. This continues until the fit of the model to the data becomes unacceptable and the most parsimonious model is found. The best model minimises the Akaike Information Criteria (AIC). The AIC estimates the amount of information a model provides and penalizes the model for any excess parameters needed to explain the data, i.e. it accounts for both explanatory power and parsimony (Sloane and Morgan 1996; Quinn and Keough 2002; Simonoff 2003).

2 AIC = G – (dfsaturated model – dfreduced model)

2 AIC = G – 2dftest of model

Log-linear models do not distinguish response and predictor variables, they only demonstrate association between variables. When a response variable is present, the number of models to be fitted is reduced (Quinn and Keough 2002). Models of interactions between predictor variables alone can be ignored because they are not of interest.

The predictor and response variables used in the log-linear analyses were as described in section 5.3.4.2 with the following alterations: Firstly, gender was omitted as a predictor variable from

-120- Chapter 5 Root suckers

Experiment 1 because it was not an important predictor variable in any classification trees, it was not a significant variable of the log-linear analysis of Experiment 2 and it did not appear to influence the occurrence of suckers of either species in the field. Secondly, root diameter in Experiment 2 was arbitrarily split into small (<23.5 mm) and large (>23.5 mm) categories containing 80 roots each to produce a result more applicable in the field. Finally the response variable “root health” was reduced to alive (alive and mostly dead) and dead to maintain adequate cell frequencies. I tested three hypotheses of no association between the predictor variables for each of the three response variables:

H0a: There is no association between gender of the parent tree, diameter of the root and

1. root health,

2. production of callus tissue,

3. production of root suckers.

H0b: There is no association between season of treatment, method of treatment, exposure of the root and

1. root health,

2. production of callus tissue,

3. production of root suckers.

The recommended sample size for contingency tables is five times the total number of cells in the table (Agresti 1996). Experiment 1 produced a four-way table with 2x4x2x2 cells and a required sample size is 160. Although my sample size of 128 is less than this, severely inflated or depressed empirical levels only occurred when testing a four-way interaction with only 2.5 times the number of cells (Stelzl 2000). Nevertheless a type I error may result, and relationships identified as significant are interpreted cautiously and classification trees were used also to further clarify the main predictor variables. The required sample size of 40 was exceeded for Experiment 2 (2x2x2 cells with n= 128).

All log-linear analyses were conducted using the program R (R Development Core Team 2005).

5.4 Results 5.4.1 Field characteristics of suckers

All suckers showed signs of browsing and most were multi-stemmed; there was little association between height and stem diameter. Suckers taller than 150 cm tended to have a single dominant stem (Figure 5.8), although often with other less-dominant stems.

-121- Chapter 5 Root suckers

5.4.1.1 Casuarina pauper Of the 161 excavated juveniles of C. pauper (Chapter 4) I found nearly all (150 or 93%) were root suckers. The characteristics of these root suckers are summarised as:

• The diameter of the proximal section ranged from 12 to 70 mm (mean 36 ± SD 12 mm) and the distal section ranged from 11 to 75 mm (39 ± 13 mm). In 136 cases (91%), the distal section was thicker than the proximal section; the average difference was 4 mm (± 3 mm). • 95 (63%) of the root suckers were on exposed roots (Figure 5.5) and the remaining 55 had a root junction that was between 2 and 8 cm below the surface (4 ± 2 cm).This suggests the depth from which root suckers can form. • The root suckers formed from 3 to 30 m from the parent tree (13 ± 6 m) • Of the 40 patches which contained root suckers, there were 17 patches containing a total of 89 root suckers where the parent was identified as female and 17 patches containing 51 root suckers where the parent was identified as male. For six patches (ten individuals) the gender of the parent tree could not be identified with certainty. • I found one sucker which when excavated, revealed that the distal portion of the root was broken and a new root had developed. I could not determine whether this damage initiated the sucker growth or occurred after the sucker grew. I also observed during the course of other work, three suckers which had developed their own root system. In each case the sucker was single stemmed, with the stem of the sucker 42, 50 and 69 mm in diameter. There was substantial distal thickening of the root, with the proximal section varying between 31 and 57 mm whilst the distal section was between 44 and 83 mm. In each case the root was an unbranched root extending straight down below the sucker.

Figure 5.5 Typical occurrence of Casuarina pauper root suckers on exposed roots.

-122- Chapter 5 Root suckers

5.4.1.2 Allocasuarina luehmannii Of the 334 excavated juveniles of A. luehmannii (Chapter 4) I found one third of these (115 or 34%) were root suckers. The characteristics of these root suckers are summarised as:

• The diameter of the proximal section ranged from 8 to 52 mm (28 ± 11 mm) and the distal section ranged from 2 to 83 mm (26 ± 14 mm). In 107 cases (93%), the distal section was thicker than the proximal section; the average difference was 8 mm (± 8 mm). • Thirty-five (30%) of the root suckers were on exposed roots and the remaining 80 had a root junction that was between 1 and 17 cm below the surface (6 ± 4 cm). • The root suckers formed from 2 to 19 m from the parent tree (9 ± 4 m) • Of the 18 patches which contained root suckers, there were eight patches containing a total of 43 excavated root suckers where the parent was identified as female. There were ten patches containing 47 root suckers where the parent was identified as male. • I found three suckers which when excavated, revealed that the portion of the root on the side closest to the parent was broken and in one case a new root had developed (Figure 5.6). I could not determine whether this damage initiated the sucker growth or occurred after the sucker grew. I also observed five suckers which had developed their own root system whilst both portions of the root proximal and distal to the parent tree remained intact (Figure 5.7). In each case the sucker was large and single stemmed, with the stem of the sucker between 24 and 62 mm in diameter. There was substantial distal thickening of the root, with the distal section between 6 and 38 mm larger than the proximal section. In each case the root was an unbranched root, three roots were on the root portion proximal to the parent tree and two were on the portion distal to the parent tree (Table 5.3).

Table 5.3 A list of the characteristics of all Allocasuarina luehmannii suckers which were not connected to the parent tree and / or had produced their own roots. Prox. diam Dist. diam Sucker collar Own root Location of (mm) (mm) (mm) diam (mm) own root

24 25 18 4 8cm distal

11 9 4 NA

13 20 8 NA

45 83 62 37 2 cm proximal

49 70 53 24 14 cm proximal

22 45 43 19 14 cm distal

27 43 37 14 11 cm distal

22 28 24 4 4 cm proximal

-123- Chapter 5 Root suckers

Figure 5.6 The section of Allocasuarina luehmannii root proximal to the parent tree has broken and this root sucker has developed its own root.

Figure 5.7 This Allocasuarina luehmannii sucker has formed its own root, in addition to the intact connection with the parent tree.

Figure 5.8 Allocasuarina luehmannii sucker showing single stemmed form.

-124- Chapter 5 Root suckers

5.4.2 Observations from Experiments 1 and 2

All roots were treated successfully. Both the scraping and severing treatment clearly showed the anatomy of the root for both C. pauper (Figure 5.9) and A. luehmannii.

The partial death of exposed roots was inevitably the upper portion of the root section which was most exposed. The partial death of buried roots only occurred following the “cut” treatment and was always a result of the portion distal to the cut dying. Only exposed roots were found to be completely dead. Some roots were classified as dead even though the buried portion of the root proximal or distal to the target section was alive. Lightly scratching the roots during two to three monthly assessments indicated that dead roots had died approximately 10 months following treatment.

Roots receiving no treatment could not produce callus on damage and were excluded from the analyses of production of callus tissue. Callus was also produced under flaking or cracked bark and preceded sucker growth from cracks. This callus was not included in the analyses because it could not be detected until a sucker erupted, meaning that callus tissue in cracks which did not produce a sucker was probably overlooked. The growth of callus tissue was not associated with old damage, branching or adventitious roots noted prior to treatment.

The growth of root suckers was not associated with old damage, branching or adventitious roots which were recorded prior to treatment. Suckers which died before 18 months were recorded but only included in the descriptive results. Occasionally suckers were produced on roots adjacent to the target root; these were noted and monitored but again only included in the descriptive results. Suckers were sometimes produced at more than one location along the root. I considered suckers separately where they were clearly distinguishable at 18 months post-treatment. Suckers sometimes started out as individuals but later merged as more sucker branchlets were produced along the root.

Periderm

Pericycle Secondary phloem

Secondary xylem

Figure 5.9 Root anatomy of Casuarina pauper, left: scraped and right: cut, spring 2003.

-125- Chapter 5 Root suckers

5.4.3 Experiment 1: Casuarina pauper trial

5.4.3.1 Descriptive results Sixty-one (38%) of the 160 exposed roots were confirmed dead at 18 months post-treatment and 18 (11%) roots were mostly dead (Figure 5.10). The dead and mostly dead roots produced no callus. Of the remaining 81 (51%) healthy roots, 69 (85%) produced callus. One dead root and one mostly dead root produced a sucker each from callus tissue in the cracks of the root but only the sucker from the mostly dead root was alive at 18 months. Of the callused, alive roots, suckers were produced from callus tissue at the distal end of the scrape in nine cases and six of these suckers were alive at 18 months post-treatment. Suckers were produced in the cracks of six roots and five of these suckers were alive 18 months post-treatment. Four adjacent roots produced suckers, only one of which was alive at 18 months.

Exposed 1Scrape 160

A 81 MD 18 D 61

C 69 NC 12 NC 18 NC 61

N 54 D 9(6) Cr 6(5) N 12 N 17 Cr 1(1) N 60 Cr 1(0)

Figure 5.10 Outcome tree for treatment of 160 roots in Experiment 2 (does not include suckers produced on roots adjacent to the target root). The first row indicates the health of the roots: A=alive, MD=mostly dead and D=dead. The second row indicates the growth of callus on the damage: NC=no callus and C=callus, and the third row indicates the growth of suckers: N=no sucker, D=sucker from damage on the root and Cr=sucker from crack in root. The numbers following indicate the number of roots in each class and numbers in parenthesis indicate the number of suckers alive at 18 months post-treatment. The classification trees and results of the log-linear analyses suggest that the diameter of the treated root is important. Accordingly the 160 roots were separated into small (<23.5 mm) and large (>23.5 mm) and the outcome trees are presented below (Figure 5.11).

a Exposed, small 1Scrape 80

A 36 MD 4 D 40

C 26 NC 10 NC 4 NC 40

N 20 D 4(3) Cr 2(1) N 10 N 4 N 39 Cr 1(0)

-126- Chapter 5 Root suckers

Exposed, large 1Scrape 80 b

A 45 MD 14 D 21

C 43 NC 2 NC 14 NC 21

N 34 D 5(3) Cr 4(4) N 2 N 13 Cr 1(1) N 21

Figure 5.11 Outcome tree for (a) small (<23.5 mm) and (b) large (>23.5 mm) roots. Other details as in Figure 5.10.

5.4.3.2 Production of suckers Twenty-one of the 160 trees (13.1%) produced root suckers. Three of these trees produced suckers from roots adjacent to the target root, one tree produced suckers from a root growing off the target root and two trees produced suckers from both the target and an adjacent root. Statistical analysis of so few trees would be uninformative; analysis of the results is descriptive only.

The diameter of the 17 target roots which produced suckers was between 19 and 33 mm (26 ± 6 mm) which was within the range of exposed, treated roots: between 15 and 36 mm (24 ± 5 mm). The six roots adjacent or growing from the target roots that produced suckers had diameters between 4 and 23 mm (13 ± 7 mm).

On eight roots the suckers died at or prior to 18 months post-treatment. These suckers tended to be very small with poor growth rates before death.

Six trees produced suckers at more than one location along the root, totalling 31 suckers from 21 trees. Eight trees produced single suckers from cracks in the target root; all but two of these were from the distal portion of the root. Four trees produced single suckers from callus tissue on the distal portion of the scrape. Two trees produced single suckers from the distal portion of broken roots adjacent to the target root. One tree produced a single sucker from a root off the target root. Three trees produced two suckers, one each from callus tissue and cracks. One tree produced a sucker from the scrape and another from an adjacent root. One tree produced one sucker from the scrape, two suckers from cracks and one from an adjacent root. Finally, one tree produced four suckers from cracks in a root adjacent to the target root. Half (15) of the suckers had a bushy growth form. Single stemmed suckers grew from cracks (six), scrapes (seven) and adjacent roots (two).

Eleven suckers from eleven roots died. Ten of the suckers were single suckers and on one root one of the two suckers died. Deaths occurred between ten and 20 months post-treatment and was followed by, or occurred simultaneously with the death of the root.

-127- Chapter 5 Root suckers

The timing of emergence varied. The first root produced suckers at four months post-treatment, followed by seven roots at seven months, four roots at ten months, eight roots at 13 months, and one root at 18 months post-treatment. Often multiple suckers were produced at different times on the same root. The relative growth rate per month of the 31 suckers was highly variable, between 0 and 15 cm / month, averaging 3 cm cm-1 m-1. Growth rates appeared higher in the warmer months although there was substantial variation (Figure 5.12). Statistical analysis of this result would be invalid because suckers were not independent with multiple suckers on the same root, and unbalanced and small sample sizes. The height of the suckers was between 17 cm and 60 cm at the final assessment of the treated roots (25 months post-treatment: average 27 ± 11 cm).

9

8

7

6

20 5

4 18

3

21 2 1 18 18

Relative growth rate (cm cm-1 m-1) cm-1 (cm rate growth Relative 1 12 0 autumn 2004 winter 2004 spring 2004 early summer late summer autumn 2005 winter 2005 2005 2005 Season of growth

Figure 5.12 The growth rate of Casuarina pauper suckers from the spring treatment of roots. The number of suckers in each period appears above the bar. The error bars represent one standard deviation.

5.4.3.3 Classification tree The classification trees produced using DTREG showed that diameter was the most important predictor variable (Figure 5.13). Small roots were more likely to be dead, produce no callus and no suckers. The split was based on roots with a diameter <30 mm, <23 mm and <27 mm respectively. The misclassification (error) rates were high indicating that the response of the tree roots cannot be very accurately predicted from diameter of the root. Gender of the parent tree did not lead to a reduction in the misclassification rates for any response variable and was not included in the trees. This supports the results of the field trial suggesting that gender is unimportant for the production of root suckers.

-128- Chapter 5 Root suckers

Alive No callus a b n=160 MC 49% n=160 MC 43%

Diameter <30mm Diameter >30mm Diameter <23mm Diameter >23mm Dead Mostly dead No callus Callus n=137 MC 58% n=23 MC 61% n=75 MC 31% n=85 MC 46%

No sucker c n=160 MC 8%

Diameter <27mm Diameter >27mm No sucker Sucker n=117 MC 5% n=43 MC 86%

Figure 5.13 Classification trees produced for Experiment 2 showing root diameter as the most important predictor variable 18 months post-treatment for (a) the health of the root, (b) the production of callus tissue, and (c) the production of root suckers. Each box (node) contains the predictor variable on which the split was based and the binary outcome (bold), the sample size (n) and the misclassification rate (MC).

5.4.3.4 Log-linear regression Log-linear regression of gender of the parent tree (male, female) and diameter of the root (small:<23.5 mm, large: >23.5 mm) against each of the three the response variables indicated that a simple model was most appropriate. For each of the response variables (health of the root, production of callus and production of suckers) the three-way interactions were not significant (

-129- Chapter 5 Root suckers

Table 5.4). For health of the root, both gender and diameter of the root were significant main effects, however the model including gender minimised the AIC. For the production of callus tissue, diameter was the only significant main effect. For the production of root suckers, no terms were significant although diameter reduced the AIC to the greatest extent (Table 5.4). With the exception of the production of root suckers, the models support the classification trees with gender being relatively unimportant and diameter being an important predictor variable.

-130- Chapter 5 Root suckers

Table 5.4 Analysis of deviance table for the association between gender of parent tree, root diameter and (a) root health, (b) production of callus, and (c) production of suckers at 18 months post-treatment. The most appropriate model is in bold. (a) Model DF Deviance AIC Pr (Chi)

Gender x Health 1 10.3 58.4 0.03

Diameter x Health 1 5.1 53.2 <0.01

Gender x Diameter x Health 2 0.0 54.1 0.74

(b) Model DF Deviance AIC Pr (Chi)

Gender x Callus 1 9.0 57.4 0.26

Diameter x Callus 1 2.8 51.3 <0.01

Gender x Diameter x Callus 2 0.0 54.4 0.45

(c) Model DF Deviance AIC Pr (Chi)

Gender x Sucker 1 5.7 49.0 1.00

Diameter x Sucker 1 4.3 47.5 0.22

Gender x Diameter x Sucker 2 0 49.2 0.12

5.4.4 Experiment 2: Initiating callus and suckers – Casuarina pauper

The following two sections deal descriptively firstly with C. pauper and then with A.luehmannii. The roots from male and female trees and autumn and spring treatment are pooled and only the main effects of treatment and exposure are considered. Gender and season were not found to be important in the classification trees or log-linear analyses.

5.4.4.1 Roots with no damage Seven of the 16 exposed, undamaged roots were identified as dead at 18 months post-treatment (Figure 5.14). The dead roots had a significantly smaller average diameter (27 ± 4 mm) than the roots still alive at 18 months post-treatment (41 ± 7 mm) (t=4.26, df=12, p=0.00). Two roots were mostly dead. In three of the nine roots, although the exposed portion was completely or mostly dead, the buried portion of the root proximal or distal to the target section was alive. Seven roots were entirely alive after 18 months. Because the roots were undamaged, they could not be scored as producing callus tissue, however two roots produced suckers from callus tissue formed in cracks (Figure 5.15). A sucker was also produced from callus tissue on an exposed and accidentally damaged root adjacent to a (dead) target root.

-131- Chapter 5 Root suckers

No callus was produced on the buried, undamaged roots and all roots except one were alive at 18 months post-treatment (Figure 5.14). The only root which died was from a tree which was wind- thrown during the experiment.

a Exposed None 16 b Buried none 16

A 7 MD 2 D 7 A 15 D 1

NC 7 NC 2 NC 7 NC 15 NC 1

N 5 Cr 2(2) N 2 N 7 N 15 N 1

Figure 5.14 Outcome tree for (a) exposed and (b) buried roots which received no treatment. Other details as in Figure 5.10.

Figure 5.15 An undamaged Casuarina pauper root exposed in spring 2003. Above: Callus tissue forming under peeling bark 12 months post-exposure. Middle: Developing suckers 15 months post-exposure Below: Suckers 18 months post-exposure.

5.4.4.2 Root with scrapes The responses of exposed roots to one long scrape or four short scrapes were similar and are presented here together (Figure 5.16).

-132- Chapter 5 Root suckers

Ten of the 32 exposed roots were confirmed dead at 18 months post-treatment. The dead roots tended to be those with a smaller average diameter (29 ± 7 mm) than the roots still alive at 18 months post-treatment (35 ± 9 mm) although the difference was not significant (t=1.72, df=22, p=0.10). Eight roots were mostly dead. In 14 of the 18 cases, although the exposed portion was completely or mostly dead, the buried portion of the root proximal to the target section was alive. The dead and mostly dead roots produced no callus (Figure 5.17). Of the remaining 14 healthy roots, 12 produced callus and suckers were produced from callus tissue at the distal end of the scrape in three cases (Figure 5.18) and from cracks in another two roots. One of these roots also produced a sucker on a root adjacent to the target root.

All of the 32 scraped, buried roots were healthy 18 months post-treatment. Eighteen of the roots produced callus, but no suckers were produced from any of the buried, scraped roots (Figure 5.19).

a Exposed 1 or 4scrape 32 Buried 1 or 4Scrape 32

b A 14 MD 8 D 10 A 32

C 12 NC 2 NC 8 NC 10 C 26 NC 6

N 7 D 4(4) Cr 1(1) N 2 N 8 N 10 N 26 N 6

Figure 5.16 Outcome tree for (a) exposed and (b) buried roots receiving either the 1-scrape or 4-scrape treatment. Other details as in Figure 5.10.

Figure 5.17 Casuarina pauper root scraped in spring 2003. Scrape shown is 18 months post- treatment and has no development of callus tissue.

-133- Chapter 5 Root suckers

Figure 5.18 Casuarina pauper root scraped and exposed in spring 2003. Above: Development of callus tissue and beginnings of root suckers 15 months post-treatment. Below: Suckers 18 months post-treatment. Suckers have formed at the distal end of the fourth scrape.

Figure 5.19 Casuarina pauper root scraped and re-buried in spring 2003. The scrape shown is 18 months post-treatment and shows typical development of thick black callus tissue from the damaged cambium but no suckers.

-134- Chapter 5 Root suckers

5.4.4.3 Cut roots In six of the 16 cut and exposed roots the entire root died (Figure 5.20) and the root shrank away from the bark. Ten roots were mostly dead with the portion distal to the cut dying. No callus was produced on the cut faces and no suckers were produced from the roots.

No cut, buried roots died, however in ten roots the portion distal to the cut died. Six alive roots and seven mostly dead roots produced callus. For eight of these roots, callus was produced on the proximal, cut face only (Figure 5.21) and in six roots the callus was produced on both cut faces. This allowed four of the roots to “re-join” where the cut faces were in sufficient proximity, presumably re-initiating a vascular connection between the two halves (Figure 5.22). In these cases both the proximal and distal portions were still alive 18 months post-treatment. No suckers were produced from the cut and re-buried roots.

a Exposed Cut 16 b Buried Cut 16

MD 10 D 6 A 6 MD 10

NC 10 NC 6 C 6 C 7 NC 3

N 10 N 6 N 6 N 7 N 3

Figure 5.20 Outcome tree for (a) exposed and (b) buried roots receiving the cut treatment. Other details as in Figure 5.10.

Figure 5.21 Casuarina pauper root cut and re-buried in spring 2003. Eighteen months post- treatment a proliferation of callus tissue has formed on the proximal cut face. The distal face (not illustrated) did not produce callus.

-135- Chapter 5 Root suckers

Figure 5.22 Casuarina pauper root cut and re-buried in spring 2003. Eighteen months post- treatment the proximal and distal cut faces have “re-joined” via callus tissue forming from both the proximal and distal faces.

5.4.4.4 Production of root suckers In total, eight of the 128 trees (6.3%) produced root suckers. One tree produced suckers from both the target and an adjacent root and one tree produced suckers only from an adjacent root. Statistical analysis of so few samples would be uninformative; analysis of the results is descriptive only.

The suckers were only formed on exposed roots. The suckers grew from roots with diameters from 26 to 47 mm (35 ± 3 mm), which was within the diameter range of roots treated: 21 to 57 mm (34 ± 9 mm). The two roots adjacent to the target roots which produced suckers had diameters of 9 mm and 7 mm.

Four of the eight trees produced suckers at more than one location along the root. Suckers were fairly evenly distributed amongst the treatments of exposed roots. Suckers were produced on the following roots

• The 1-scrape treatment produced two suckers from two trees: a single sucker from a scrape, and a single sucker in a crack, • The 4-scrape treatment produced six suckers from three trees: a single sucker from a scrape, two suckers in cracks, and the third root produced three suckers, one in the distal portion of scrape 4, one in a crack, and one from an adjacent root. • No treatment produced five suckers from three trees: a single sucker in a crack, two suckers in cracks, and two suckers from an adjacent root. This totals 13 suckers from eight trees. The suckers were invariable produced in the distal portion of the roots and suckers from callus tissue were always from the distal edge of the scrape. Most

-136- Chapter 5 Root suckers

(nine) of the suckers had a bushy form; the four which were single stemmed came from cracks (two) and scrapes (two).

The first root to produce suckers did so at ten months post-treatment, followed by three roots at 13 months, two roots at 16 months and two roots at 18 months post-treatment. Only one root had suckers which died, this was a single sucker which died 26 months post-treatment.

The relative growth rate of the 13 suckers was highly variable, between 0 cm and 6 cm / month, averaging 2 cm per month. No seasonal trends in growth rate were discernable with so few suckers. The height of the suckers was between 6 and 38 cm at the final assessment, 25 months post-treatment (22 ± 13 cm).

5.4.5 Experiment 2: Initiating callus and suckers – Allocasuarina luehmannii

5.4.5.1 Roots with no damage Three of the 16 exposed, undamaged roots were confirmed dead at 18 months post-treatment (Figure 5.23). Four roots were mostly dead. In five roots, although the exposed portion was completely or mostly dead, the buried portion of the root proximal (three roots) or proximal and distal to the target section was alive. Nine roots were alive at 18 months and two of these roots produced suckers from cracks in the peeling bark (Figure 5.24). An additional two roots produced suckers from roots adjacent to the target root, one from the distal portion of a broken root and one from callus tissue in cracks in the adjacent root.

No callus or suckers were produced on the buried, undamaged roots and all roots were alive at 18 months post-treatment (Figure 5.23).

a Exposed None 16 Buried None 16

b A 9 MD 4 D 3 A 16

NC 9 NC 4 NC 3 NC 16

N 7 Cr 2(2) N 4 N 3 N 16

Figure 5.23 Outcome tree for (a) exposed and (b) buried undamaged roots. Other details as in Figure 5.10.

-137- Chapter 5 Root suckers

Figure 5.24 Allocasuarina luehmannii root undamaged but exposed in spring 2003. Above: Suckers forming in a crack 10 months post-exposure. Middle: The entire root at 15 months and below: 18 months post-exposure.

5.4.5.2 Roots with scrapes The responses of exposed and buried roots to one long scrape or four short scrapes were similar and are presented here together (Figure 5.25).

Two of the 32 exposed roots were confirmed dead at 18 months post-treatment. Eleven roots were mostly dead. In all cases the buried portions of the root proximal and distal to the target section were alive. The 13 dead roots and mostly dead roots did not produce callus (Figure 5.26). However, four of the mostly dead roots produced suckers from cracks in the roots, one of which was dead by 18 months post-treatment. Of the remaining 19 alive roots, 16 produced callus and three of these produced suckers from the callus on the damage (Figure 5.27). In each case the

-138- Chapter 5 Root suckers suckers formed at the distal end of the scrape. Another callused root produced suckers from cracks in the bark, but the sucker was dead by 18 months. An additional two roots produced suckers from roots adjacent to the target root. In both cases the adjacent root had been broken and the suckers formed on the distal side of the break. Finally, one root produced suckers from callus tissue growing in the cracks of a root branching from the target root.

All of the 32 scraped, buried roots were healthy 18 months post-treatment. Twenty-six of the roots produced thick, red or orange, often bubbly callus (Figure 5.28). Six roots did not produce any callus tissue. No suckers were produced from any of the buried, scraped roots, however two roots produced a protrusion which I considered to be a root bud (Figure 5.31).

a Exposed 1 or 4scrape 32

A 19 MD 11 D 2

C 16 NC 3 NC 11 NC 2

N 12 D 3(3) Cr 1(0) N 3 N 7 Cr 4(3) N 2

Buried 1 or 4Scrape 32

b A 32

C 26 NC 6

N 26 N 6

Figure 5.25 Outcome tree for (a) exposed and (b) buried scraped roots. Other details as in Figure 5.10.

Figure 5.26 Allocasuarina luehmannii root scraped and exposed in autumn 2003. Scrape shown is 18 months post-treatment and has no development of callus tissue.

-139- Chapter 5 Root suckers

Figure 5.27 Allocasuarina luehmannii root scraped and exposed in spring 2003. Above: development of callus tissue and beginnings of root suckers at 13 months and below: 18 months post-treatment. Suckers have formed in the fourth (most distal) scrape.

Figure 5.28 Allocasuarina luehmannii root scraped and re-buried in spring 2003. The scrape shown is 18 months post-treatment and shows typical development of bubbly red callus tissue from the damaged cambium.

-140- Chapter 5 Root suckers

5.4.5.3 Cut roots Three of the cut, exposed roots were completely dead and 13 were mostly dead by 18 months post-treatment (Figure 5.29). The roots did not produce any callus on the cut faces. In four of the mostly dead roots, suckers were produced from cracks in the portion distal to the break. One of these roots also produced a sucker from a root off the target root, and another of these roots also produced a sucker from a root adjacent to the target root. In the latter case the adjacent root was broken and sucker grew from the portion distal to the break.

Five of the cut, re-buried roots were mostly dead (the distal portion being dead), whilst the 11 remaining roots were healthy at 18 months post-treatment (Figure 5.29). Two mostly dead roots produced callus on the proximal cut face. Ten alive roots produced callus on the distal face only (six roots) or on both faces (four roots). In two of the roots with callus on both faces, the cut faces were able to “re-join”, presumably re-establishing a vascular connection. Suckers formed on the distal face of one of the mostly dead roots and seven of the alive roots; none of these suckers were alive 18 months post-treatment (Figure 5.30). The suckers formed in a circular pattern from just under the bark and grew to a maximum of 11 cm. One root produced a root bud from the proximal cut face (Figure 5.31).

a Exposed Cut 16 b Buried Cut 16

MD 13 D 3 A 11 MD 5

NC 13 NC 3 C 10 NC 1 C 2 NC 3

N 9 Cr 4(4) N 3 N 4 D 6(0) D 1(0) N 2 N 2 D 1(0)

Figure 5.29 Outcome tree for (a) exposed and (b) buried cut roots. Other details as in Figure 5.10.

Figure 5.30 Allocasuarina luehmannii root cut and re-buried in autumn 2003. Numerous suckers grew from under the bark on the face distal of the cut. The longest sucker was 8 cm.

-141- Chapter 5 Root suckers

Figure 5.31 Allocasuarina luehmannii root cut and re-buried in spring 2003. The protrusion is from the portion of the root proximal to the cut and is 20 mm long x 4 mm wide.

5.4.5.4 Production of root suckers In total, 27 of the 128 trees (21.1%) produced root suckers. Four of these trees produced suckers from roots adjacent to the target root and one tree produced suckers from both the target and an adjacent root. Statistical analysis of so few trees would be uninformative; analysis of the results is descriptive only.

Eight trees produced suckers on the distal portion of cut, buried roots however these suckers were all dead when excavated at 18 months post-treatment. These suckers grew to a maximum length of 11 cm (5 ± 4 cm). The shallowest of these roots was 15 cm deep indicating that suckers have little ability to emerge from depth. The diameter of the buried roots that produced suckers varied between 19 and 50 mm (36 ± 10 mm) which was within the range of all buried roots of 18 to 56 mm (34 ± 9 mm). The suckers from buried roots are not considered further.

The suckers which persisted to at least 18 months post-treatment were only formed on exposed roots. The diameter of the 14 exposed roots which produced suckers was between 21 and 52 mm (36 ± 9 mm) which was also within the range of exposed, treated roots between 16 and 56 mm (34 ± 9 mm). The seven roots adjacent or growing from the target roots and produced suckers had diameters between 9 and 18 mm (13 ± 3 mm).

Twelve of the 19 trees produced suckers at more than one location along the root. Suckers were fairly evenly distributed amongst the treatments of exposed roots. Suckers were produced on the following roots

• The 1-scrape, exposed treatment produced six suckers from five trees: two trees produced single suckers from cracks, one tree produced two suckers from a root off the target root,

-142- Chapter 5 Root suckers

and two trees produced a single sucker from the distal portion of a broken root adjacent to the target root. • The 4-scrape, exposed treatment produced 11 suckers from six trees: three trees produced two suckers from cracks in the target roots, two trees produced single suckers from callus on the target root and one tree produced three suckers from callus on the target root. • The cut, exposed treatment produced ten suckers from four trees: one tree produced a single sucker from a root growing from the adjacent root and a sucker from a crack in the target root, one tree produced a single sucker from both the target root and an adjacent root and two trees produced two suckers each from cracks in the target root. • No treatment, exposed roots produced 12 suckers from four trees: a single sucker from the distal portion of a broken root adjacent to the target root, one tree produced two suckers from cracks in an intact root adjacent to the target root, one tree produced two suckers from cracks in the target root and one tree produced seven suckers from cracks in the target root. This totals 39 suckers from 19 trees. Twenty-eight of the 30 suckers from the target roots were produced in the distal portion of the roots and suckers from callus tissue were always from the distal edge of the scrape. The four suckers on broken adjacent roots were all on the distal side of the break. Most of the suckers had a bushy growth form; the four of the five suckers which were single stemmed came from callus tissue in scrapes.

The timing of emergence of suckers varied between trees and treatments. The first root to produce suckers did so at four months post-treatment, followed by one root at eight months, three roots at ten months, five roots at 13 months, four at 15 months and five at 18 months post-treatment. Often multiple suckers were produced at different times on the same root. Considering the eruption of each of the 39 suckers individually, it appeared that suckers from the autumn treated roots were slower to develop (Figure 5.32). Suckers did not erupt until January 2004 regardless of whether the roots were treated in autumn 2004, or spring 2004.

Twelve suckers from eight roots died. All of the suckers died on seven roots and one of two suckers died on an eighth root. Deaths occurred 16 and 27 months post-treatment and was associated with the death of the root.

The per month growth rate of the 39 suckers was highly variable, between 0 cm and 9 cm / month, averaging 3 cm per month. Growth rates appeared higher in the warmer months (Figure 5.33). Statistical analysis of this result would be invalid because suckers were not independent with multiple suckers on the same root, and unbalanced and small sample sizes. The height of the suckers was between 17 and 60 cm at the final assessment of the autumn treated roots (29 months post-treatment: average 37 ± 16 cm) and 11 and 51 cm at the final assessment of the spring treated roots (25 months post-treatment: average 31 ± 10 cm).

-143- Chapter 5 Root suckers

12

11

10

9

8

7

6

5

4

Number of erupting suckers Number of 3

2

1

0 4 8 10 13 15 16 18 18 20 Months post treatment

Figure 5.32 The time of eruption of suckers in months following treatment for spring (grey bars) and autumn (black bars) treated roots. The eruption of suckers peaked sooner after treatment for the roots treated in spring.

9.0

8.0

7.0

6.0 16 12

5.0 19

4.0 1 12

3.0 4 2 2.0 5 18 Relative growth rate ( cm cm-1 month-1) cm-1 ( cm rate growth Relative 18 9 1.0 9

0.0 autumn 2004 winter 2004 spring 2004 summer 2005 autumn 2005 winter 2005 Season of growth

Figure 5.33 The growth rate of Allocasuarina luehmannii suckers from the spring (grey bars) and autumn (black bars) treatment of roots. The number of suckers in each period appears above the bar. The error bars represent one standard deviation. Growth rates appear higher during the warmer months.

-144- Chapter 5 Root suckers

5.4.6 Statistical analyses

5.4.6.1 Classification tree analysis – Casuarina pauper Simple trees were produced by the classification tree analysis for C. pauper (Figure 5.34). The most important predictor variable in determining the health of the root, production of callus tissue and production of suckers 18 months post-treatment was whether the root was buried or exposed. Exposed roots were more likely to be dead, fewer exposed roots produced callus, but only exposed roots produced suckers. The treatment of the buried roots was also important with cut roots likely to be classified “mostly dead”. The treatment of exposed roots was also important in the production of suckers, with exposed, cut roots not producing suckers.

a Alive n=128 MC 42%

Buried Alive Exposed Dead n=64 MC 17% n=64 MC 64%

Cut Mostly dead 1-, 4-scrape, none Alive n=16 MC 37% n=48 MC 2%

Callus b n=96 MC 47%

Buried Callus Exposed No callus n=48 MC 19% n=48 MC 25%

c No sucker n=128 MC 6%

Buried No sucker Exposed Sucker n=64 MC 0% n=64 MC 89%

Cut No sucker 1-, 4-scrape, none Sucker n=16 MC 0% n=48 MC 85%

Figure 5.34 Classification trees showing the important predictor variables 18 months post- treatment for (a) the heath of the root, (b) the production of callus tissue on treated roots, excluding the “none” treatment, and (c) the production of root suckers in Casuarina pauper. Other details as in Figure 5.13.

-145- Chapter 5 Root suckers

5.4.6.2 Classification tree analysis – Allocasuarina luehmannii Simple trees were produced by the classification tree analysis for A. luehmannii (Figure 5.35). The most important predictor variable in determining the health of the root, production of callus tissue and production of suckers 18 months post-treatment was whether the root was buried or exposed. As with C. pauper, exposed roots were more likely to be dead, fewer exposed roots produced callus, but only exposed roots produced suckers. The treatment of the buried roots was also important with cut roots likely to be classified “mostly dead”. The season and treatment of exposed roots also reduced the misclassification rate; autumn roots, other than the 4-scrape treatment, were more likely to be dead.

a Alive n=128 MC 32%

Buried Alive Exposed Dead n=64 MC 8% n=64 MC 87%

Cut Mostly dead 1-, 4-scrape, none Alive spring Mostly dead autumn Dead n=16 MC 69% n=48 MC 0% n=32 MC 53% n=32 MC 81%

1-scrape, cut Mostly dead 4-scrape, none Dead 4-scrape Mostly dead 1-scrape, cut, none Dead n=16 MC 31% n=16 MC 87% n=8 MC 37% n=24 MC 75%

Callus No sucker b c n=96 MC 44% n=128 MC 9%

Buried Callus Exposed No callus Buried No sucker Exposed Sucker n=48 MC 21% n=48 MC 33% n=64 MC 0% n=64 MC 81%

Figure 5.35 Classification trees showing the important predictor variables 18 months post- treatment for (a) the heath of the root, (b) the production of callus tissue on treated roots, excluding the “none” treatment, and (c) the production of root suckers in Allocasuarina luehmannii. Other details as in Figure 5.13.

5.4.6.3 Log-linear regression – Casuarina pauper Log-linear regression for C. pauper was conducted of season of treatment (autumn, spring), treatment (1-scrape, 4-scrape, cut, none) and exposure of the root (exposed, buried) against each of the three the response variables. For each of the response variables (health of the root, production of callus and production of suckers) the four-way interactions were not significant. For health of the root, exposure was a significant main effect, and the model including exposure

-146- Chapter 5 Root suckers minimised the AIC. For the production of callus tissue, there was a significant interaction between treatment, exposure and the production of callus tissue. This means that the callus outcome depends on the interaction of treatment and exposure, the outcome is different for different levels of treatment depending on the exposure of the root. For the production of root suckers, exposure, treatment and season were all significant main effects although exposure reduced the AIC to the greatest extent (Table 5.5). With the exception of the production of root suckers, the models support the classification trees with exposure being the most important predictor variable.

Table 5.5 Analysis of deviance table for Casuarina pauper showing the association between season of treatment, treatment, exposure and (a) root health, (b) production of callus, and (c) production of suckers at 18 months post-treatment. The most appropriate model is in bold. (a) Model DF Deviance AIC Pr (Chi) Treatment x Health 3 34.5 139.4 0.7 Exposure x Health 1 6.6 107.5 0.00 Season x Health 1 35.6 136.5 0.68 Treatment x Exposure x Health 6 2.7 123.6 0.87 Exposure x Season x Health 2 3.9 116.8 0.53 Treatment x Season x Health 6 3.6 124.4 0.95 Treatment x Season x Exposure x Health 6 0.0 148.9 1.00

(b) Model DF Deviance AIC Pr (Chi) Treatment x Callus 2 50.0 133.8 0.19 Exposure x Callus 1 20.9 102.7 0.00 Season x Callus 1 51.2 133.0 0.15 Treatment x Exposure x Callus 4 4.2 100.0 0.02 Exposure x Season x Callus 2 13.6 105.3 0.37 Treatment x Season x Callus 4 14.9 110.7 0.96 Treatment x Season x Exposure x Callus 4 0.00 115.8 0.8

(c) Model DF Deviance AIC Pr (Chi) Treatment x Suckers 3 48.8 141.7 0.01 Exposure x Suckers 1 33.3 122.3 0.00 Season x Suckers 1 50.7 139.6 0.00 Treatment x Exposure x Suckers 6 12.6 121.6 0.98 Exposure x Season x Suckers 2 12.2 113.2 0.47 Treatment x Season x Suckers 6 5.0 113.9 0.19 Treatment x Season x Exposure x Suckers 6 0.00 136.9 0.90

-147- Chapter 5 Root suckers

5.4.6.4 Log-linear regression – Allocasuarina luehmannii Log-linear regression for A. luehmannii was also conducted using the same predictor and response variables. The most appropriate models were the same as those selected for C. pauper with the exception that exposure was the only significant main effect for the production of root suckers (Table 5.6).

Table 5.6 Analysis of deviance table for Allocasuarina luehmannii showing the association between season of treatment, treatment, exposure and (a) root health, (b) production of callus, and (c) production of suckers at 18 months post-treatment. The most appropriate model is in bold. (a) Model DF Deviance AIC Pr (Chi) Treatment x Health 3 19.9 113.4 0.53 Exposure x Health 1 10.6 100.0 0.00 Season x Health 1 19.9 109.4 0.14 Treatment x Exposure x Health 6 5.9 115.4 0.99 Exposure x Season x Health 2 5.9 107.4 0.92 Treatment x Season x Health 6 0.7 110.1 0.49 Treatment x Season x Exposure x Health 6 0.00 137.5 0.99

(b) Model DF Deviance AIC Pr (Chi) Treatment x Callus 2 44.6 127.5 0.03 Exposure x Callus 1 30.1 111.0 0.00 Season x Callus 1 45.3 126.2 0.01 Treatment x Exposure x Callus 4 7.0 101.8 0.04 Exposure x Season x Callus 2 12.7 103.5 0.11 Treatment x Season x Callus 4 15.6 110.4 0.83 Treatment x Season x Exposure x Callus 4 0.00 114.9 0.86

(c) Model DF Deviance AIC Pr (Chi) Treatment : Suckers 3 27.0 123.0 0.27 Exposure : Suckers 1 13.0 105.1 0.00 Season : Suckers 1 29.4 121.5 0.22 Treatment : Exposure : Suckers 6 7.2 119.2 0.99 Exposure : Season : Suckers 2 7.4 111.5 0.92 Treatment : Season : Suckers 6 1.3 113.4 0.39 Treatment : Season : Exposure : Suckers 6 0.0 140.1 0.99

-148- Chapter 5 Root suckers

5.4.7 Effects of exposure

The classification trees and log-linear models indicated that exposure was an important predictor variable for the health of the root, and the production of callus and suckers.

5.4.7.1 Effect of exposure of Casuarina pauper roots Exposure decreased the health of the root, with 67% being dead or mostly dead 18 months post- treatment, compared with only 17% of buried roots (Figure 5.36). Callus was more likely to form on buried roots (81% of treated, buried roots produced callus versus 25% for exposed roots). The production of callus differed according to the health of the treated root with callus more likely to be produced on alive roots (85% of alive, damaged roots); only 25% of mostly dead roots and no dead roots produced callus. Thus, greater production of callus on buried roots is likely due to the greater proportion of alive, buried roots than alive, exposed roots. Indeed, if the dead and mostly dead roots are excluded there is little difference between the production of callus on exposed roots (57%) and buried roots (60%). The seven suckers alive 18 months post-treatment were only formed on exposed roots (11% of exposed roots produced suckers).

a All exposed 64

A 21 MD 20 D 23

C 12 NC 9 NC 20 NC 23

N 7 D 4(4) Cr 1(1) N 7 Cr 2(2) N 20 N 23

b All buried 64

A 53 MD 10 D 1

C 32 NC 21 C 7 NC 3 NC 1

N 32 N 21 N 7 N 3 N 1

Figure 5.36 Outcome tree for (a) exposed and (b) buried roots for Casuarina pauper. Other details as in Figure 5.10.

-149- Chapter 5 Root suckers

5.4.7.2 Effect of exposure – Allocasuarina luehmannii roots Similar effects of exposure were also observed for A. luehmannii. Exposure decreased the health of the root, with 56% being dead or mostly dead 18 months post-treatment, compared with only 8% of buried roots (Figure 5.37). Callus was more likely to form on damaged, buried roots (79% buried, damaged roots produced callus versus 33% for exposed, damaged roots). The production of callus differed according to the health of the root with callus more likely to be produced on alive roots (84% of all alive, damaged roots); only 7% of mostly dead roots and no dead roots produced callus. Therefore the apparent greater production of callus on buried roots is probably due to the greater proportion of alive, buried roots than alive, exposed roots. Indeed, if the dead and mostly dead roots are excluded there is little difference between the production of callus on exposed roots (57%) and buried roots (61%). The 12 suckers alive 18 months post-treatment were only formed on exposed roots (19% of exposed roots produced suckers).

a All exposed 64

A 28 MD 28 D 8

C 16 NC 12 NC 28 NC 8

N 12 D 3(3) Cr 1(0) N 10 Cr 2(2) N 20 Cr 8(7) N 8

b All buried 64

A 59 MD 5

C 36 NC 23 C 2 NC 3

N 30 D 6(0) N 22 D 1(0) N 2 N 2 D 1(0)

Figure 5.37 Outcome tree for (a) exposed and (b) buried roots. Other details as in Figure 5.10.

-150- Chapter 5 Root suckers

5.5 Discussion 5.5.1 The growth and development of root suckers

5.5.1.1 The relationship between callus and suckers Following injury to a root, the living cells beneath the periderm are exposed and die. Callus tissue arises from phloem and xylem in the vascular region and a new periderm forms beneath (Esau 1977). Callus tissue is an important precursor to the growth of suckers in some species such as Fagus grandifolia (Jones and Raynal 1986) and always preceded the growth of root suckers of C. pauper and A. luehmannii. This suggests that the root buds and suckers on these species are reparative and exogenous in origin. Additive buds form on uninjured roots early in the development of the root and are identified by bud traces Dissections of damaged roots were not carried out so the presence of bud traces could not be established, however the growth of suckers from callus tissue makes it unlikely that root buds are additive. Additive buds have the disadvantage that suckering is limited by the number of buds formed early in the development of the root.

Not all damaged roots produce callus and not all callus produces suckers. Callus tissue formed on undamaged, exposed roots in cracks in the bark of the root. Exposure is believed to initiate root suckers by increasing heat and light (Kormanik and Brown 1967) or freezing and thawing of roots (Jones and Raynal 1986). In this case my observations indicated heat was killing the uppermost portion of the root, stimulating reparative callus tissue, from which a new periderm develops. This callus tissue was often not visible until a sucker erupted from it, so the frequency of occurrence on undamaged, exposed roots may have been underestimated. Callus tissue also formed in response to scraping the root and damaging the cambium. Callus formed on the distal portion of severed roots although it was not associated with the cut face, and could have been merely a result of exposure. Undamaged, buried roots did not produce callus.

Callus tissue did not occur on all damaged and / or exposed roots. It was only produced on roots which were alive 18 months following treatment. This is not surprising because a root under sufficient stress to cause tissue death is unlikely to commit resources to reparative tissue. This accounted for apparently greater production of callus on damaged, buried roots which tended to be healthier than exposed roots. When the health of the root was considered, the production of callus tissue on damage occurred in similar numbers on both exposed and buried roots. Fifteen percent of damaged roots of both species were alive at 18 months post-treatment but did not produce callus. The cause of this was not established.

The callus produced on buried roots does not produce long-lived suckers. Damage to the roots produced callus tissue that was often quite thick and appeared to be reparative. This suggests that light may play a role in triggering the development of buds on damaged roots. Suckers did form

-151- Chapter 5 Root suckers on some buried, severed A. luehmannii roots (minimum depth of 15 cm) but they died prior to 18 months post-treatment before reaching the soil surface, indicating that suckers have little ability to emerge from depth. Root suckers are reliant on root carbohydrate reserves until they reach the soil surface (Schier and Zasada 1973). The premature death of suckers suggests that carbohydrate reserves were insufficient (Frey et al. 2003). Light to produce carbon from photosynthesis, is required for the persistence of root suckers (van der Heyden and Stock 1996). The importance of photosynthesis for the growth of suckers is also supported by the poor performance of suckers in low light conditions (Frey et al. 2003). It is unlikely that burial forms a physical impediment to the emergence of root suckers because in most instances the suckers were growing in loose sand.

Not all callus on exposed roots produced suckers alive at 18 months post-treatment. The production of suckers from callus tissue on the scrapes of exposed roots only occurred in 33% and 9% of C. pauper trees in Experiments 1and 2 respectively, and 19% of A. luehmannii trees. Aside from the observation that the sucker-producing tissue appeared thicker, I could not establish the reason for this.

In summary, not all roots produce callus, only a proportion of those scraped and / or exposed. Roots must survive the damage / exposure to produce callus and 15% of damaged / exposed roots do not produce callus, perhaps because they were damaged insufficiently. Not all callus produces suckers. Suckers require light but only between 9% and 33% of exposed, damaged roots produce suckers.

5.5.1.2 Plant hormones The production of root suckers from individual, damaged roots is believed to be associated with an interruption to the flow of the plant hormones auxin and cytokinin, which inhibit and promote the initiation of suckers respectively (Schier 1981; Frey et al. 2003). However, the role of other substances produced during the wound response may also be important (Fraser et al. 2004). I found supporting evidence for a linkage between production of suckers and disruption to the transport of hormones. Suckers were produced on the distal portion of cut or broken roots which could either be attributed to accumulation of cytokinins (transported proximally) or reduction in auxins (transported distally). Suckers were also generally produced on the distal edge of scrapes which formed callus tissue. This has been observed also by other researchers (Kormanik and Brown 1967; Wisniewski and Parsons 1986; Jones and Raynal 1988). However, there was no proximal or distal bias in the production of suckers in P. tremuloides (Fraser et al. 2004) so the role of disruption of the transport of hormones may not be ubiquitous.

5.5.1.3 The form of root suckers A single-stemmed form of juveniles is desirable as it resembles the adult tree. Root suckers from treated roots were most often bushy and rarely had a single stemmed form. Those that were single

-152- Chapter 5 Root suckers stemmed tended to arise from callus tissue in localised damage (scrapes) rather than callus in cracks in the bark, although this was not always the case. It is possible that localised damage and exposure on a short root segment would be more successful in producing single stemmed suckers than exposing root segments of 0.8 to 1.2 m, as I did. Long cracks were produced in exposed roots which often produced long lines of suckers along the root.

Established suckers in the field were also generally bushy (had a large number of branchlets), however larger root suckers often had a single dominant stem. This suggests that whilst root suckers may initially form a clump, a single stem becomes dominant as the suckers age; this has been observed in other species (Jones and Raynal 1986; Peterson and Jones 1997; Bond and Midgley 2001; Fraser et al. 2004). Even young suckers produce auxin and exert apical dominance over other suckers as well as competing with them (Frey et al. 2003).

There were insufficient suckers initiated on treated roots to draw any conclusions about the growth rate of single versus multiple suckers. It is possible a single sucker would grow more quickly without the competing influences of other suckers on the same root.

5.5.1.4 Persistence of root suckers Many root suckers initiated on treated roots, died within two years of treatment, often less than this. The death of a root was always associated with the death of all the root suckers growing from it. The health of roots 18 months post-treatment was clearly related to exposure which killed many roots, especially of C. pauper. Presumably this was related to exposed roots being subject to extremes of temperature without the insulating effects of soil. The effect was most pronounced in C. pauper which has thinner bark on the root than does A. luehmannii (pers. obs.). Sometimes on living roots, only some suckers died. The cause of this was unknown but could be attributed to thinning of smaller, subordinate suckers due to competition for nutrients or carbohydrates from the common parent root (Frey et al. 2003).

5.5.1.5 Independence from the parent tree Reparative root buds are connected to the vasculature of the parent root via basipetal differentiation of procambial strands (Schier 1973a). Root suckers of trees are initially heavily reliant upon the parent root. Gradually, the cambium and secondary vascular system of the growing root sucker align with the distal portion of the root causing characteristic “distal thickening” (Kormanik and Brown 1967; Schier 1973a) and new roots form at the base of the sucker becoming the major root system (Reinartz and Popp 1987). Pruning the distal portion of the root or the new roots reduced growth far more than pruning the proximal portion, especially in older suckers, suggesting transport from the parent ramet decreases over time (Zahner and De Byle 1965). Eventually the proximal portion decays naturally and the sucker is independent (De Byle 1964; Zahner and De Byle 1965).

-153- Chapter 5 Root suckers

This pattern of development was evident for C. pauper and A. luehmannii. Most root suckers showed signs of distal thickening, however only a few established root suckers in the field had produced their own roots. Those that did tended to have a large stem diameter (around 40 mm) and substantial distal thickening suggesting that root suckers of these species are reliant on the parent tree for many years. This is not unusual for suckers of tree species. Connections remain for 25 to 50 years in P. grandidentata (De Byle 1964; Zahner and De Byle 1965) and at least ten years in F. grandifolia (Jones and Raynal 1986). No new roots are present in L. styraciflua root suckers even when the root collar is 150 mm in diameter (Kormanik and Brown 1967). The formation of a new root system on root suckers is a critical requirement for suckers to be a useful restoration technique. In some genera such as Acacia, suckers never form their own root system and die when the parent plant dies (Lacey and Johnston 1990).

Very few suckers were found in the field that were not connected to the parent tree. This may be interpreted in two ways: suckers on the distal portion of severed roots do not persist, or damage that severs roots is uncommon. Three of four A. luehmannii suckers initiated on the distal portion of severed roots persisted at least two years and were in good health, suggesting that the latter scenario may be more likely. However suckers not connected to the parent tree would not benefit from photoassimilates from the parent, although they would still have a more substantial root system than a newly established sucker. Severing the connection to the parent tree reduces growth but not survival in P. tremuloides suckers (Peltzer 2002). The callus tissue on severed roots was sometimes sufficient to allow the severed portions to re-join but whether this could happen on roots with suckers is unclear. So, despite the possibility that disconnected suckers can persist, it seems prudent for restoration purposes to attempt to damage but not sever roots when attempting to initiate sucker growth.

5.5.2 What influences the initiation of root suckers?

Figure 5.2 summarised factors affecting the initiation of root suckers in other species: factors related to the particular root, factors related to the parent tree, season of disturbance and severity of disturbance. My analysis was limited by the small number of suckers produced. Other studies which have artificially initiated root suckers have had a much higher success rate (Jones and Raynal 1988; Fraser et al. 2004) allowing sucker height, number of suckers and sucker biomass to be measured as quantitative response variables. Nevertheless, observations were noted on the importance for the initiation of root suckers of the root, the parent tree and the season and severity of disturbance.

5.5.2.1 The root Field surveys of existing suckers suggested the root zone generally had a radius of around 9 m from A. luehmannii trees (up to 19 m) and around 13 m from C. pauper trees (up to 30 m). This is

-154- Chapter 5 Root suckers similar to another semi-arid, Australian species, Alectryon oleifolius, where root suckers can form up to 8.2 m from the parent stem (Wisniewski and Parsons 1986). It is likely that suckers can occur further from larger trees due to the greater spread of the root system (Jones and Raynal 1986). As a guide, the root system of Allocasuarina luehmannii has a radial spread roughly the same as the height of the parent tree (Story 1967).

Root suckers of other species develop from shallow, lateral roots (Kormanik and Brown 1967; Schier 1981; Jones and Raynal 1986, 1988; Shepperd and Smith 1993; Des Rochers and Lieffers 2001) and this appears to be the case also for C. pauper (Torrey 1983; Auld 1995a, 1995c) and A. luehmannii (Morison and Harvey 2001). My research indicated that roots less than 10 cm deep or exposed roots are required to produce suckers. A substantial number of suckers in the field occurred on exposed roots (63% of C. pauper and 30% of A. luehmannii). Those where the origin was buried averaged 4 cm and 6cm deep for C. pauper and A. luehmannii respectively. It is possible that these suckers arose from exposed roots which were later reburied. This is supported by Experiments 1 and 2; no suckers emerged from buried roots. Some A. luehmannii suckers grew (a maximum of 11 cm in length) but later died although one root was only 15 cm deep. Exposure to light soon after sucker initiation appears essential for the persistence of root suckers. This is presumably due to a requirement for carbohydrates from photosynthesis to maintain existing tissues (Kozlowski 1992; Des Rochers and Lieffers 2001). Suckers growing from excised roots grow rapidly until the carbohydrate reserves are exhausted, at which point some suckers die unless there is sufficient leaf biomass for photosynthesis (Tew 1970). In A. luehmannii most of the lateral roots are less than 30 cm deep (Story 1967) so shallow roots should not be limiting for this species. Techniques such as ground-penetrating radar could be utilised to increase search efficiency (Hruska et al. 1999; Butnor et al. 2001).

Suckers grew from roots of varying diameters in the field trials and the examination of suckers in the field. Experiment 2 suggested that for C. pauper smaller roots (<23.5 mm) are less able to survive exposure and produced fewer suckers. However, non-target roots of small diameter produced suckers when damaged incidentally. The diameter of these roots was between 7 mm and 18 mm. These roots tended to have only short lengths exposed, rather than the 80 cm or 120 cm length of the target roots. Two thirds of roots where 1.2 m was exposed but only half of roots with 0.8 m exposed were dead or mostly dead. Root carbohydrate reserves (starch) are greatest in the smaller roots (Reichenbacker et al. 1996) so exposing small sections of small diameter roots may prove to be as or more successful than exposing greater lengths of larger roots. Cutting roots also caused some root death with the distal portion of approximately half of the buried, cut roots dying. Death of tree roots, whilst undesirable, is unlikely to influence the health of the tree unless a large proportion of roots die. Treating roots needs to balance the amount of damage necessary to initiate suckering and minimising root death.

-155- Chapter 5 Root suckers

5.5.2.2 The parent tree The health, age and genetics of the parent tree were not investigated although it may be important (e.g. McLellan et al. 1997). The gender of the parent tree had little effect on the suckering response for either A. luehmannii or C. pauper, or on the occurrence of root suckers in field surveys. The lower reproductive costs for males of dioecious species are hypothesised to avail more resources for vegetative growth, including growth of root suckers (Obeso et al. 1998). However, a lack of sex-related bias in the production of root suckers has been found in many studies (Sakai and Burris 1985; Reinartz and Popp 1987; Sakai and Sharik 1988) and was the case here.

5.5.2.3 The season of disturbance I damaged roots in spring and autumn but could not clearly demonstrate seasonal effects on the growth of suckers. The seasonal effects on the production of root suckers may have been masked by the small number of roots which produced suckers. Nevertheless, my observations and those of others suggest that spring may be a better period to attempt to initiate roots suckers.

Seasonal variation in carbohydrate / nutrient / hormone levels in the parent tree relates primarily to deciduous species (Tolvanen and Laine 1997). Seasonal variation in growth according to temperature and soil moisture is probably more important in Mediterranean sclerophylls. These species undergo two periods of little growth, during the cold winter and during the late summer / autumn dry period, and most growth is concentrated in the spring (Körner 2003). An early spring treatment would be followed by a good period of growth before the onset of the dry period, whereas an autumn treatment may experience limited rainfall and growth before the onset of winter. Accordingly, I found A. luehmannii suckers from roots treated in autumn were slower to develop than suckers from roots treated in spring, and the growth rate of suckers for A. luehmannii and C. pauper appeared to be greatest during spring and summer. Morison and Harvey (2001) also advise that light grading or scarifying in early spring, well out from the dripline of A. luehmannii can produce root suckers. Spring treatments of F. grandifolia generally produced the most callus, buds and suckers and root mortality was lowest (Jones and Raynal 1987).

5.5.2.4 The severity of disturbance The disturbance to a root system can range from complete loss of above ground parts, to severing, scraping or exposing single roots. When the above ground parts of A. luehmannii or C. pauper are removed (e.g. wind-throw or chaining) the trees do not resprout from the trunk or produce long- lived root suckers (pers. obs., M. Westbrooke pers. comm. 2004, Chesterfield and Parsons 1985), although the loss of large branches can result in sprouting or suckering (pers. obs). Maintenance respiration, which is the energy required to maintain existing tissues, such as a root system,

-156- Chapter 5 Root suckers requires substantial inputs of carbohydrates from photosynthesis (Kozlowski 1992). Presumably, without the carbohydrate input from the parent tree the root system and therefore root suckers cannot survive. In P. tremuloides clonal stands with high root / shoot ratios can die due to insufficient carbohydrates produced by the suckers to support the root system (Des Rochers and Lieffers 2001).

For root suckers to be produced, the root needs to be damaged sufficiently to induce the production callus tissue which always precedes sucker growth. However not all damaged roots produce callus and not all callus produces suckers. Exposure, or a combination of damage and exposure, is necessary to initiate the growth of sucker-producing callus tissue. Even then, suckers only occur at low rates. When exposed target roots and adjacent roots are considered only 13% and 8% of C. pauper trees in Experiments 1 and 2 respectively, and 27% of A. luehmannii trees produced suckers that were alive 18 months post-treatment.

5.5.3 Are root suckers a viable tool for restoration?

5.5.3.1 Why is initiating suckers a tool worth investigating? Little is known about the ecology of root suckers in Australian species (Lacey and Johnston 1990) despite numerous woody perennials from arid areas of Australia having the ability to produce root suckers (Chapter 1, Maconochie 1982). Overseas studies have shown juvenile root suckers are connected to the parent plant via the root system, termed physiological integration which allows improved establishment of juvenile ramets compared to seedlings (Callaghan 1984; De Steven 1989; Barsoum 2001). Such an advantage is particularly important for long-lived woody plants where juveniles are the most vulnerable age class (Peterson and Jones 1997). Physiological integration is also a disadvantage for restoration because root suckers can only be initiated close to the parent trees, not in degraded communities where no trees are present. However supplementing recruitment in existing woodlands may be more prudent where resources are limited than attempting to recreate a woodland from degraded grassland (Bradley 1988).

In Australia rangelands, the recruitment of seedlings of woody, perennial species is limited by lack of soil moisture and high grazing pressure since European settlement (Hall et al. 1964; Chesterfield and Parsons 1985; Auld 1990, 1993; Cheal 1993; Auld 1995c). Restoration attempts by hand-planting or direct seeding also fail, often for the same reasons (Thomas 2003). The importance of identifying alternative tools for restoration was highlighted in the state-and- transition model presented in Chapter 3. Using root suckers to enhance recruitment may have merit where there is no recruitment of seedlings of existing trees or shrubs. The advantages of root suckers as a tool for restoration are the same as those suggested for an Australian Acacia (Auld 1990):

• maintenance of an existing population,

-157- Chapter 5 Root suckers

• increased area occupied by a genetic individual and, • the increased number of individuals may lead to increased seed production for subsequent sexual recruitment. Juvenile root suckers are suggested to have a lower reliance on soil moisture than seedlings (Maconochie 1982). Here, suckers of A. luehmannii and C. pauper were initiated, albeit in low numbers, during a period of very low rainfall. During this time no new seedlings were detected in woodlands surrounding treated trees. This is not conclusive evidence of a lower reliance on soil moisture, but does demonstrate that suckers can be initiated in low rainfall years. The transport of water from parent to daughter ramet is an advantage of sucker growth over seedlings (Caraco and Kelly 1991).

Interspecific comparisons of C. pauper suckers with seedlings of other species have demonstrated the survival of seedlings is very limited due to desiccation and browsing, whereas suckers can persist under these conditions (Chesterfield and Parsons 1985; Auld 1995a; Denham and Auld 2004). The response of root suckers to browsing was not tested here but the multistemmed nature of suckers in the field suggests they can resprout following damage.

Artificially initiating the growth of root suckers was shown to be possible for A. luehmannii and C. pauper and the applicability of this technique is discussed below.

5.5.3.2 “Best-bet” initiation of root suckers Few studies have attempted to artificially initiate the growth of root buds or suckers (but see Kormanik and Brown 1967; Jones and Raynal 1987; Bosela and Ewers 1997; Fraser et al. 2004). No studies that I could find have considered root suckers as a possible tool for restoration. Based on the results presented previously, personal observation, and the literature I can make the following “best-bet” recommendations for artificially stimulating root suckers from A. luehmannii and C. pauper trees.

• The initiation of suckers requires the production of callus tissue. Callus tissue can be stimulating by damaging the vascular cambium of the root, e.g. by scraping, or by exposure which causes the bark of the root to crack • Suckers cannot emerge from depth and the treated roots must either be exposed or less than 5 cm deep. • It is possible that suckers on the distal portion of severed roots may persist, however they will lack many of the advantages of physiological integration with the parent. Attempts should be made to avoid severing roots during treatment. The connection of suckers to the parent tree is likely to remain for tens of years. • Suckers do not persist after the root dies so attempts should be made to minimise exposure which kills roots.

-158- Chapter 5 Root suckers

• The choice of root diameter is a trade-off between smaller roots which may produce more suckers, and larger roots which are more resilient to damage and exposure. If damage and exposure can be minimised, roots with diameters between 20 and 30 mm should be targeted. • Small amounts of localised damage on short segments of exposed roots may improve the proportion of suckers having a single-stemmed form. • Roots should be treated at approximately 10 m from the parent tree, depending on the size of the parent. • There is no indication that the gender of the parent tree influences the survival of treated roots or the production of callus or root suckers. • Season of treatment was not demonstrated to be important although an early spring treatment may allow a longer growth period following treatment and faster initiation of suckers. • In summary, a suitable technique would treat a number of roots per tree to allow for low success rates. The damage would be to very shallow roots and may involve some excavation of roots. The treatment would aim to scrape but not cut roots to encourage the production of suckers on callus from damage rather than from cracks. A light, inclined disc with a spring to “bounce over” rather than cut through roots may be appropriate. This could be pulled behind a 4WD motorbike.

5.5.3.3 Suckering of different species There is no clear dichotomy between species in their ability to resprout, rather a continuum of responses from very likely to very unlikely (Vesk et al. 2004b). A similar continuum is likely to be present for the production of root suckers.

Root suckering in A. luehmannii is believed to be less frequent than in C. pauper (Cunningham et al. 1992). I did find a higher proportion of suckers than seedlings of C. pauper in the field but could not attribute this to an enhanced intrinsic ability to produce suckers. Rather, I achieved a higher level of success in producing root suckers from treated A. luehmannii roots. The “bark” of C. pauper was substantially thinner than A. luehmannii thus allowing the former roots to be damaged more easily, which could lead to a greater field occurrence of suckers. Also, the grazing pressure in HKNP has been substantially lower than MSNP for a number of years and it is possible that grazing eliminates seedlings but not root suckers of C. pauper in MSNP, whereas both seedlings and suckers of A. luehmannii can persist.

The techniques recommended above may also be suitable for other species which produce root suckers in response to root damage (Appendix 1). In HKNP, Alectryon oleifolius appears to produce suckers readily (pers. obs and Wisniewski and Parsons 1986) and because it rarely

-159- Chapter 5 Root suckers produces seedlings, is a good candidate for artificially initiating root suckers. Other species could include Acacia hakeoides, Acacia melvillei, Acacia salicina, Acacia victoriae,Eremophila longifolia, Eremophila sturtii, Grevillea hugelii, Hakea leucoptera, Hakea tephrosperma, Leptospermum coriaceum, Lycium australe, Pittosporum phylliraeoides, Santalum acuminatum and Senna artemisioides.

-160- Chapter 6 Conservation genetics

Chapter 6 Conservation genetics

6.1 Scope and objectives

Restoration ecology is a diverse, applied field drawing on many tools to preserve biodiversity, one being conservation genetics. Conservation genetics provides information on otherwise unobservable demographic processes, such as current genetic diversity, population structure, the mating system, and the likely impact of restoration activities on the genetic structure of populations (Holsinger and Vitt 1997). Whilst conservation genetics tends to focus on very rare or endangered species, (e.g. Young et al. 2002; Neel and Ellstrand 2003; Tero et al. 2003) the genetics of dominant species which are the focus of habitat restoration efforts are also important (e.g. Gustafson et al. 1999; Thomas et al. 1999; Cengel et al. 2000; Gustafson et al. 2002, 2004). Given the lack of practical conservation outcomes from genetic studies of rare plants (Hogbin et al. 2000), restoration ecology may indeed be better served by directing resources to the conservation genetics of dominant plant species.

Casuarina pauper and Allocasuarina luehmannii are known to be able to produce both juvenile root suckers and seedlings although the relative importance of each in the mature population is unknown. There are three scenarios regarding the recruitment ecology of these species:

1. Seedlings could be the dominant mode of reproduction with root suckers being an unusual result of damage to the roots of existing trees,

2. Root suckers could be an important mode of reproduction in average rainfall years with substantial sexual recruitment occurring only in years with favourable climate (e.g. Eriksson and Froborg 1996; Mitton and Grant 1996), or

3. Root suckers could be the dominant mode of reproduction.

Determining the relative importance of sexual reproduction versus clonal growth for C. pauper and A. luehmannii has important management implications. Chapter 5 has demonstrated that root suckers are a viable tool for restoration because they can be initiated artificially. However, whilst the technique can be used freely in the two latter scenarios because extensive clonal recruitment is present, it can only be used in the third scenario if there are negligible impacts on the conservation genetics of the species.

I used Amplified Fragment Length Polymorphisms to determine the frequency of identical genotypes (indicating clonal growth) in populations of mature trees of C. pauper and A. luehmannii. I also estimated the genetic diversity of these stands to comment on the likely impact of introducing root suckers on the genetic structure of populations of these species. My objectives were:

-161- Chapter 6 Conservation genetics

1. Determine whether clonal individuals (root suckers) exist in populations of mature A. luehmannii or C. pauper,

2. Estimate the genetic diversity and population structure of these species in north-west Victoria, and

3. Use this information to predict the likely impact of artificially introducing root suckers to existing populations as a restoration tool.

6.2 Introduction to conservation genetics 6.2.1 Genetic diversity and population structure

Determining the population genetic structure of target plant species suggests appropriate conservation strategies (Stewart and Porter 1995; Hogbin and Peakall 1999; Jones and Gliddon 1999; Hogbin et al. 2000). The genetic diversity of a population is often measured as heterozygosity (HW): the average percentage of genotypes in a population that are heterozygous for a particular gene locus. A species divided into isolated subpopulations will generally have less genetic diversity overall than a widespread species because fragmentation interrupts the flow of genes between individuals. Genetic drift (loss of alleles through random processes) and inbreeding (non-random mating) may result in a higher than normal proportion of homozygotes which increases the chance of recessive, deleterious mutations being expressed (inbreeding depression), loss of distinctive variants and impact on the long-term persistence of a population or species (Schaal et al. 1991; Hamrick and Godt 1996; Young et al. 2002).

When genetic data are collected from multiple populations of a species, heterozygosity can be partitioned into the gene diversity within populations (HW), and the average gene diversity

between populations in excess of that observed within populations (HB):B this yields the total gene

diversity (HT = HW + HB).B A high degree of genetic differentiation between subpopulations (HB B is high) means the loss of a single population would lead to a dramatic loss of genetic diversity (e.g Sydes and Peakall 1998; Hogbin and Peakall 1999; Warburton et al. 2000). In contrast, if most of the genetic diversity is held within populations (gene flow is high), the conservation of a few large populations will capture most of the genetic diversity of the species. The mating system of the plant species primarily determines where genetic diversity is held (Krauss 2000b). In outcrossing populations most (>80%) is contained within populations; however for selfing species this may be as low as 50% (Hamrick et al. 1992).

The mating system is determined genetically by the extraction and electrophoretic separation of proteins (allozymes) or fragments of DNA. Principles of population genetics then make it possible to ascertain whether or not the frequency distribution of genotypes in natural populations is consistent with random mating as described by Hardy-Weinberg Equilibrium (HWE). HWE states

-162- Chapter 6 Conservation genetics if alleles A and a occur with frequencies p and q, the probabilities of obtaining the following genotypes will be AA = p2, Aa = 2pq and aa = q2. Strong departures from HWE indicate non- random mating such as selfing or clonality.

6.2.1.1 Clonality Clonality is indicated by a lack of segregation within loci, a lack of recombination between loci or both. For diploid organisms, fixed heterozygosity, absence of segregation genotypes, and deviation from HWE indicate that segregation is lacking. Independent of ploidy, overrepresentation of identical genotypes, absence of recombinant genotypes, linkage disequilibrium, correlation between two independent sets of genetic markers indicate that recombination is lacking (Tibayrenc et al. 1991). The more criteria providing evidence for a lack of segregation or recombination, the more likely that clonal growth is the predominant mode of reproduction (reviewed in Tibayrenc et al. 1990; Tibayrenc et al. 1991).

Clonality reduces phenotypic and genotypic variation which may reduce the ability of the population to survive environmental challenges (Hamrick and Godt 1996; Young et al. 2002). Numerous studies of populations of clonal plants have focused on determining their level of genetic variation. Reviews of such studies have shown that on average, populations exhibiting clonal growth are not genetically less variable than non-clonal populations (Ellstrand and Roose 1987; Hamrick and Godt 1990; Widen et al. 1994; McLellan et al. 1997). The considerable genetic variability found in many populations has been attributed to low levels of seedling recruitment, long-lived clones, environmental heterogeneity leading to selection for locally adapted genotypes, frequency-dependent selection involving pathogens, somatic mutation, retention of genotypic variation from a sexual establishment event, or gene flow (Jelinski and Cheliak 1992; McClintock and Waterway 1993; Watkinson and Powell 1993; Eriksson and Froborg 1996; Escaravage et al. 1998; Holderegger et al. 1998).

The recruitment of seedlings into a clonal population appears particularly important for maintaining genetic diversity. The frequency varies between plant species (e.g. Eriksson 1989; Eriksson 1992) and may also vary spatially or temporally within a species (e.g. Eckert and Barrett 1993; Yeh et al. 1995; Gugerli et al. 1999; Stehlik and Holderegger 2000; Kjølner et al. 2004; Clark-Tapio et al. 2005). Models of genetic variation have indicated that three sexually derived individuals per generation are sufficient to produce patterns of allelic variation similar to sexually reproducing populations. Strong allelic divergence between gene copies only appears where there is less than one sexual event every third generation (Bengtsson 2003).

In some very small or extensively clonal plant populations, the capacity for sexual reproduction may be lost due to self-incompatibilty, inbreeding depression, pollen sterility, pollen-pistil incompatibility or pistil dysfunction (Eriksson and Bremer 1993; Bushakra et al. 1999; Warburton

-163- Chapter 6 Conservation genetics et al. 2000). The persistence of Casuarina obesa in Victoria has been attributed to its ability to produce root suckers. One 15 Ha stand is entirely female, seeds are not viable and there are no seedlings (Lacey and Johnston 1990). In other populations, clonal growth may predominate despite high annual seed production because the environment is unsuitable for the establishment of seedlings (Eriksson 1989; Lacey and Johnston 1990; Jelinski and Cheliak 1992). Long-lived clones or constant clonal growth may allow the genet to persist until “windows of opportunity” provide conditions that are appropriate for sexual reproduction (Jelinski and Cheliak 1992; Eriksson and Froborg 1996; Mitton and Grant 1996; Jones and Gliddon 1999). However in approximately 40% of clonal plants, repeated seedling recruitment makes an important contribution to the population dynamics (Eriksson 1989).

Generally sexual reproduction is enhanced under favourable physiological conditions and resource availability (Harada et al. 1997; Mandujano et al. 2001; Mejias et al. 2002). A striking example of the variation in the importance of clonal growth within a species is the dioecious Populus tremuloides (reviewed in Mitton and Grant 1996). P. tremuloides is an abundant and widespread tree, occurring from arid to alpine areas of North America. In semi-arid areas, despite annual production of numerous viable seeds, recruitment of seedlings is rare due to desiccation and browsing, and the populations consist of few, large clones produced from root suckering. The largest, which may be 1 million years old, extends over 43 Ha and contains about 47,000 ramets. Occasional pulses of seedling recruitment occur following above-average rainfall. In moist, humid areas where the climate allows regular establishment of seedlings the clones are smaller and sexual reproduction is more common.

6.2.2 Amplified Fragment Length Polymorphisms

Markers (allozymes or fragments of DNA ) vary in their cost, difficulty and laboratory facilities required (O'Hanlon et al. 2000). The choice of a marker depends on the aims of the study. This study was designed primarily to detect clonal individuals, therefore variability was extremely important. Protein markers (isozymes or allozymes) were the earliest molecular markers and remain today one of the simplest and cheapest techniques (Hamrick and Godt 1990). They do not however, provide the variation offered by DNA markers to unambiguously detect clonal individuals (reviewed by O'Hanlon et al. 2000).

DNA markers that require prior sequence knowledge for the study species, such as microsatellites, are expensive and time-consuming to develop. Amplified Fragment Length Polymorphisms (AFLPs) and Random Amplified Polymorphic DNA (RAPD) are from the appealing group of DNA markers which can be generated from species for which there is no prior sequence knowledge. RAPD has a short developmental time, is simple and costs are reasonable, although reproducibility is poor. In contrast, AFLPs are highly reliable, and are more informative than

-164- Chapter 6 Conservation genetics

RAPDs. However it is a more difficult technique and more expensive (Jones et al. 1997a; O'Hanlon et al. 2000; Caetano-Anolles 2001). It requires high-resolution electrophoresis and radioactive or fluorescent labels to visualise fragments and higher quality DNA because incomplete digestion of the DNA template can lead to false polymorphisms. The technique is often used to determine the presence of clonal growth (e.g. Escaravage et al. 1998; Pornon et al. 2000; van der Hulst et al. 2000; Douhovnikoff et al. 2004; Kjølner et al. 2004). AFLPs are also used frequently for studies of the conservation genetics of rare plants (e.g. Allnutt et al. 1999; Gaudeul et al. 2000; Krauss 2000b; Roniker 2002; Tero et al. 2003).

AFLPs examine polymorphisms in the DNA sequence. The DNA fragments are generated using Polymerase Chain Reaction (PCR) to amplify sections of DNA between primer sites. Polymorphisms in the AFLP products can result from alterations in the DNA sequence, including mutations at the restriction site, mutations in the sequences adjacent to the restriction site or insertions or deletions in the amplified fragments (Vos et al. 1995). Like RAPD, the scoring of the AFLP products can be ambiguous, particularly for autoradiographs but less so for the automated sequencing system. Also, there is no a priori reason to accept that co-migrating bands are homologous and non-homologous bands would significantly bias population genetics (Robinson and Harris 1999). The probability of this appears slight (Yan et al. 1999; Nybom 2004). Both AFLPs and RAPDs are dominant markers: the heterozygote cannot be distinguished from the dominant homozygote.

For dominant markers, estimates of genetic variation and population structure are complicated (Lynch and Milligan 1994; Zhivotovsky 1999; Krauss 2000a; Holsinger et al. 2002; Hill and Weir 2004). Within population genetic diversity cannot be determined directly so an alternative measure of genetic variation is used that can be obtained easily from dominant molecular markers, e.g. the percentage of polymorphic loci. Using dominant markers to determine the degree of genetic differentiation between populations requires assumptions about HWE, e.g. homozygosity may be assumed for mainly selfing populations, or HWE assumed for outcrossing species (Krauss 2000b). The frequency of the null allele at a given locus can then be estimated by taking the square root of the frequency of the null homozygote (individuals that lack the marker: Lynch and Milligan 1994). Early modifications to reduce bias in this method excluded loci where observed frequency is less than 1 – (3/N) (Lynch and Milligan 1994). However this incurs a loss of information and may yield its own bias (Hill and Weir 2004). Bayesian methods (Zhivotovsky 1999; Holsinger et al. 2002) or the moment estimate method (Hill and Weir 2004) are recommended.

Dealing with polyploids is also difficult using dominant markers. Casuarina spp. appear to be consistently diploid (2n=2x=18), although no data exist for C. pauper (Barlow 1959a). However, A. luehmannii is a tetraploid (2n=4x=56) based on seedlings from three populations (Barlow

-165- Chapter 6 Conservation genetics

1959a). The frequency of alleles can be greatly underestimated for polyploids. A diploid individual with a band present may be either ++ (homozygous) or +- (heterozygous), however a tetraploid individual with a band present may be ++++, +++-, ++--, or +---. However, the bias that occurs at single loci should be greatly reduced when multiple loci are assayed (Eckert et al. 2003).

There is currently no software available that deals explicitly with polyploidy and dominant markers. Alternative approaches analyse band phenotypes, for example using AMOVA (Excoffier et al. 1992) or Shannon’s Index. Modifications to AMOVA have been proposed for a triploid species (Palacios and Gonzales-Candelas 1997) however this technique still requires assumption of random mating or self-fertilisation. Unlike AMOVA, Shannon’s Index (S) does not assume Hardy-Weinberg Equilibrium, is not greatly affected by the bias associated with dominant markers and is comparable between studies. It can be nested into heirarchical levels and the variation partitioned within and between groups (Bussell 1999). Shannon’s Index and AMOVA are highly correlated (Landergott et al. 2001).

AFLPs can, like RAPDs, detect the over-representation of identical genotypes as an indicator of clonal growth, with high resolution and independent of ploidy level (Weising et al. 1995). Measures of similarity between pairs, such as Jaccard’s (1908) coefficient are also based on band phenotypes. Jaccard’s coefficient considers the absence of a band as uninformative. A value of one would indicate a multilocus, identical phenotype, suggesting a clonal individual (then tested using the approach of Stenberg et al. 2003). Shannon’s Index and Jaccard’s coefficient have been used in other studies of polyploid taxa using dominant markers (Allnutt et al. 1999; Gustafson et al. 1999; Landergott et al. 2001; Massa et al. 2001; Gustafson et al. 2004) and were selected for use in this study.

6.2.3 The mating system of Casuarina pauper and Allocasuarina luehmannii

Clonal growth via root suckering and apomixis has been reported in both C. pauper and A. luehmannii (Chapter 2) but it is unknown how important this process is to population maintenance and expansion. Root suckering is believed to be not as common in A. luehmannii as it is in C. pauper (Cunningham et al. 1992), however even the importance of root suckers in C. pauper is contentious. Some suggest that root suckers rather than seedlings are the future for recruitment (Hall et al. 1964) and that clonal groves are present (Beadle 1981). Indeed the survival of seedlings is very limited due to desiccation and browsing, whereas suckers can persist under these conditions (Auld 1995a; Denham and Auld 2004). Others note that suckering is variable and is restricted to “erosion halos” around C. pauper clumps and is absent from single trees or dense stands (Chesterfield and Parsons 1985).

-166- Chapter 6 Conservation genetics

Juvenile root suckers of both C. pauper and A. luehmannii are present in the field (Chapter 4) and root suckers of both species can be stimulated from exposed and / or damaged roots (Chapter 5). Determining the relative importance of sexual versus asexual recruitment is a priority for management to decide whether seedling recruitment or clonal growth should be the focus for assessing recruitment and devising restoration techniques. The possible management outcomes from this genetic study were identified a priori (Table 6.1) as recommended by Hogbin et al.(2000).

Table 6.1 The outcomes, conservation implications and management options provided by genetic studies of Casuarina pauper and Allocasuarina luehmannii. Incorporate* indicates that artificially stimulated root suckers should only be used if genetic studies indicate high levels of genetic diversity within populations. Outcome of Predominantly Both sexual and Predominantly sexual clonal clonal genetic study

Conservation None Identify spatial and Consider Implications temporal variation in consequences of clonal growth encouraging sexual reproduction

Management options

1. Hand planting Continue Abandon Abandon

(low priority)

2. Direct seeding Continue Continue Abandon

(high priority) (low priority)

3. Root suckers Incorporate* Incorporate Incorporate

(low priority) (high priority) (high priority)

6.3 Methods 6.3.1 Collection of plants

Leaf tissue was collected between 28-10-03 and 9-11-03 from each of three wild populations of C. pauper (C1, C2, C3) and three wild populations of A. luehmannii (A1, A2, A3) in north-west Victoria (Figure 6.1). Samples from thirty-two trees were collected from each population, totalling 192 samples. Each population was geographically distinct, being separated by at least 2.5 km. I considered the populations chosen as most likely to contain clonal trees because the trees were often closely associated or clumped. The individuals sampled from a population were

-167- Chapter 6 Conservation genetics selected as the thirty-two nearest neighbours to a large tree (diameter at breast height) in the population. All individuals sampled were mature trees and no samples were taken from obvious root suckers. This sampling protocol was used to maximise the possibility of detecting mature, clonal trees within populations of C. pauper and A. luehmannii.

Branchlets were collected into zip-lock plastic bags and stored on ice for a maximum of four hours. Samples were transported to the laboratory where the soft growing tips were removed from the branchlets and immediately snap-frozen in liquid nitrogen and stored at -80oC until use.

HATTAH-KULKYNE NATIONAL PARK & C3 C2 A3 A2 01530 A1 kilometres MURRAY-SUNSET C1 NATIONAL PARK

Figure 6.1 The site locations for the collection of genetic material of Casuarina pauper (C1, C2 and C3) and Allocasuarina luehmannii (A1, A2, and A3) in Murray Sunset and Hattah Kulkyne National Park. 6.3.2 Extraction of DNA

DNA was extracted from leaf tips using a modified CTAB protocol (Doyle and Doyle 1990) which produces DNA of sufficiently high quality for AFLP analysis of tree species (Cervera et al. 2000). I weighed and ground approximately 0.5 g of leaf tips under liquid nitrogen and transferred 0.25 g of ground tissue into each of two 2 mL eppendorf tubes. All equipment including mortar and pestles, eppendorf tubes and racks were frozen in liquid nitrogen before use to prevent thawing of tissues which I found resulted in the rapid degradation. 0.5 mL of CTAB extraction buffer (0.02 M EDTA, 1.4 M NaCl, 0.1 M Tris pH 8.0, 2% CTAB, 2% soluble PVP, 0.3% ß- mercaptoethanol) was added to each tube and vortexed thoroughly. The slurry was incubated at 65oC for 30 min.

Phases were separated by centrifugation for 10 min at 20 000 x g. I extracted the supernatant to a new tube and mixed it with an equal volume of chloroform : isoamyl alcohol (24:1). The phases were again separated by centrifugation and the chloroform : isoamyl alcohol extraction repeated. Following the second extraction, the supernatant from the two tubes was combined into a single

-168- Chapter 6 Conservation genetics

1.5 mL eppendorf tube. The double extraction reduced substantially the levels of protein contamination of the samples. I added 0.6 mL of ice-cold isopropanol to the aqueous phase and incubated at -20oC overnight to precipitate the DNA.

The DNA was pelleted at 20 000 x g for 10 min. I washed the pellet briefly in 0.5 mL of wash buffer (76% ethanol, 0.01M sodium acetate) then added a further 0.5 mL of wash buffer and held the pellet at room temperature for 30 min. The pellet was re-centrifuged and 1.0mL of 95% ethanol added and maintained at room temperature for a further 30 min, then air-dried for two hours or dried in a hot block at 37oC for 10 min. The DNA was resuspended in 75 µL of TE (10 mM Tris, pH 8.0, 1mM EDTA) containing 3U RNAse One (Promega) and incubated at 37oC for 1 hr or incubated at -4oC overnight to ensure the DNA was completely resuspended. DNA was stored at -20oC.

Between 1 µL and 5 µL of DNA was electrophoresed on a 0.8% agarose gel and the intensity of the DNA band was compared with a known quantity of Lambda DNA cut with Hind III (Promega). I found 0.5 g of fresh tissue yielded approximately 0.2 µg of high quality DNA per µL ± 0.1 µg\µL.

6.3.3 AFLP analyses

Briefly, the steps involved are:

1. Digestion of a sample DNA with restriction endonucleases (EcoR I and Mse I or Pst I and Mse I )

2. Ligation of adapters to the ends of the DNA fragments to provide template DNA for amplification.

3. Preselective amplification with primers containing one selective nucleotide

4. Selective amplification using primers which include a sequence complementary to the ligated adapter plus two to four arbitrary nucleotides that extend into the restriction fragment are used to amplify a smaller subset of the fragments. The primers are labelled, either radioactively or fluorescently, at the 5’ end.

5. Analysis of the amplified fragments by denaturing polyacrylamide gel electrophoresis. The polymorphic fragments are visualised by labelling, either fluorescently or radioactively. The bands are interpreted and scored as present (1) or absent (0).

I conducted an initial trial run of the AFLP procedure to confirm DNA of sufficient quality. Following this, all AFLP reactions up to the production of autoradiographs were carried out under contract by Dr. Joe Barrins, School of Botany, University of Melbourne. The reproducibility of

-169- Chapter 6 Conservation genetics

AFLP analyses is extremely high with increasing familiarity with the method (Jones et al. 1997a). Dr. Barrins has extensive experience with the technique and this ensured a reproducible result.

AFLP products were generated using the AFLP Analysis System I Kit following the protocol recommended by the manufacturers (Invitrogen Life Technologies). The two exceptions were: half-reactions were used throughout (approximately 100ng DNA) and the concentration of Mg in the 10x PCR buffer for the preamplification reaction was doubled, using 2.5µL of 200mM Tris-

HCl, 500mM KCl and 30mM MgCl2. Incubations used standard water baths, incubation or drying ovens. The preamplification and selective amplification were carried out using a GeneAmp® PCR System 2700 (Applied Biosystems).

Ten primer combinations (Sigma Aldrich) were screened:

• E-ACG with M-TA, M-GA, M-CTA, or M-TGCA and • E-ACGG with M-GCAG, M-GGTA, M-ATCC, M-AGAT, M-CGAG, or M-CTAG The primers chosen for the selective amplification contained four selective nucleotides and produced the clearest profiles. The primer combination E-ACGG and M-ATCC was used initially and where identical AFLP multilocus phenotypes (hereafter referred to as genotypes) were observed, the samples were rerun using E-ACGG and M-GCAG to verify clonal genotypes. The EcoRI was end labelled with T4 kinase and [γ33P]-ATP (Amersham, UK).

The AFLP products were separated on 5% or 5.5% denaturing (sequencing) polyacrylamide gels. Gels were prepared from Long Ranger Singel™ (Proscience) according to the manufacturer's instructions. The gel was pre-warmed and run at 70 – 80 W in TBE 1.0x buffer on a BioRad vertical gel electrophoresis system for approximately two hours. After electrophoreses, the gel was transferred to chromatography paper, dried on a vacuum gel dryer (BioRad) at 80oC and exposed to Kodak BioMax Biological Imaging Film for 24 hours. The autoradiographs were posted to Mildura where I scored the bands as present (1) or absent (0) using a lightbox. Autoradiographs were scored twice, on separate occasions and only unambiguous bands which were scored the same on both occasions were retained.

6.3.4 Data analysis

Jaccard’s similarity matrix was derived from the binary data set using NTSYSpc v2.02i (Rohlf 1998). A coefficient of similarity equalling one would suggest an identical (clonal) genotype pair. The Jaccard similarity coefficient is

J = a \ (n – d)

Where a is the number of positive matches (1,1), n is the total sample size and d is the number of negative matches. Negative matches are not considered informative.

-170- Chapter 6 Conservation genetics

Overall genetic diversity was calculated as the percentage of polymorphic bands. The diversity in the frequency of AFLP band profiles was calculated using Shannon’s Index (S) using the program PopGene v1.31 (Yeh et al. 1997). As suggested in Allnutt et al. (1999), Shannon’s Index is denoted as S, not H as in other studies, to distinguish it from other measures of diversity such as heterozygosity (H), against which it is not comparable.

Three analogous measures (ΦST GST FST) (Nybom 2004) are commonly used to quantify and test the degree of differentiation between populations (termed population structure). ΦST is derived from Analysis of Molecular Variation (AMOVA: Excoffier et al. 1992) and uses the distances between observations. GST and FST use allelic frequencies as a measure of the difference between the mean heterozygosity between populations, and the potential heterozygosity if all individuals of the species mixed freely and non-assortively (Hartl and Clark 1997). GST is equivalent to

SBETWEEN determined using the following equations (based on Bussell 1999):

S = - ∑ pi log2 pi

Where pi is the frequency of the presence or absence of the band.

The average diversity over all populations (SPOP) was for each locus was calculated as

SPOP = 1/n ∑ S

Where n is the number of populations and the diversity of the combined populations (SToT) was calculated as

SToT = - ∑ ps log2 ps

Where ps is the frequency of presence or absence of the band in the whole sample. The proportion of diversity within the populations (SWITHIN ) was:

SWITHIN = SPOP \ STOT

The proportion of diversity between the populations (SBETWEEN or GST) was the species diversity minus the average diversity among all populations, divided by the species diversity.

SBETWEEN = STOT - SPOP \ STOT

These measures are based on phenotypic band diversity and are therefore comparable between studies, regardless of ploidy.

6.4 Results 6.4.1 Clonal genotypes

The primer combination of E-ACGG and M-ATCC produced 37 scorable bands for A. luehmannii from population A1, 34 bands for population A2, and 33 bands for population A3. Of the A. luehmannii samples collected, six plants were excluded from each of populations A1 and A2, and

-171- Chapter 6 Conservation genetics two plants were excluded from population A3 because of poor amplification. Pair-wise genetic similarity (Jaccard’s coefficient) for A1 ranged from 0.08 to 0.83 (average of 0.50), A2 from 0.13 to 0.81 (average of 0.46) and A3 from 0.06 to 0.85 (average of 0.49). No identical genotypes were detected in the AFLP products obtained for 82 samples from the three populations.

The primer combination of E-ACGG and M-ATCC produced 24 scorable bands for C. pauper from population C1, 21 bands for population C2, and 26 bands for population C3. Jaccard’s coefficient indicated five pairs of identical, multilocus genotypes in population C1, and two genotype pairs and one group of three identical genotypes in population C2; all genotypes in population C3 were unique. For C3, Jaccard’s coefficient ranged from 0.23 to 0.94 (average of 0.48) however, only eight of the 32 samples collected amplified satisfactorily.

A second primer pair, E-ACGG and M-GCAG, was used to verify clonal pairs detected from populations C1 and C2. The second primer pair increased the total number of scorable bands to 44 for C1 and 38 for C2 and separated all previously recognised genotype pairs. Jaccard’s coefficient ranged from 0.52 to 0.91 for population C1 (average of 0.72) and from 0.18 to 0.96 for population C2 (average of 0.57). Twenty-two samples from C1 amplified clearly for both primer pairs and a further seven samples amplified for one of the primer pairs. Twenty-six samples from C2 amplified clearly for both primer pairs and a further four samples amplified for one of the primer pairs. All samples which amplified for only one primer pair were able to be distinguished as genetically unique. No identical genotypes were detected in the AFLP products obtained for 65 samples from three populations.

6.4.2 Genetic variation

The percentage of polymorphic loci averaged over the three populations of A. luehmannii and C. pauper was 97% and 86% respectively (Table 6.2). The diversity in the frequency of AFLP band profiles (Shannon’s Index) within each population was calculated for populations A1, A2, A3 and C3 using primer pair E-ACGG and M-ATCC. Shannon’s Index for populations C1 and C2 was calculated using the data matrix from two primer pairs E-ACGG and M-ATCC and E-ACGG and M-GCAG (Table 6.2).

Intraspecies comparisons to determine population genetic structure were made for populations A1 and A2 (A. luehmannii) and C1 and C2 (C. pauper) (Table 6.2). Poor amplification of the controls relating population A3 to A1 and A2 and population C3 to C1 and C2 prevented inclusion of populations A3 and C3 in the calculation of total genetic diversity (STOT). The controls did not allow unambiguous matching of bands between autoradiographs, thus violating the first assumption for analysis of population genetic structure (Lynch and Milligan 1994). To determine the population structure of C.pauper, only samples which amplified well for both primer pairs were included.

-172- Chapter 6 Conservation genetics

The genetic diversity calculated from Shannon’s Index (S) ranged from 0.28 to 0.33 (SPOP = 0.29) for A. luehmannii and from 0.30 to 0.42 (SPOP = 0.34) for C. pauper. The total genetic diversity of

A. luehmannii was STOT = 0.28, and for C. pauper STOT = 0.35. The proportion of genetic diversity held within the A. luehmannii populations was 94.4% (GST = 0.056) and C. pauper populations was 87.7% (GST = 0.123) between populations (Table 6.2).

Table 6.2 The primer combinations, number and percentage of polymorphic bands, and Shannon’s Index (SPOP and STOT) for three populations of Allocasuarina luehmannii (A1, A2, A3) and Casuarina pauper (C1, C2, C3). Primer Population polymorphic \ Shannon’s Shannon’s Shannon’s combination (n) total bands Index S Index STOT Index GST E-ACGG + (%)

M-ATCC A1 (26) 36 \ 37 = 97% 0.28 (0.24)

M-ATCC A2 (26) 33 \ 34 = 97% 0.25 (0.24) 0.28 (0.22) 0.056

M-ATCC A3 (30) 32 \ 33 = 97% 0.33 (0.26)

M-ATCC \ C1 (22) 33 \ 44 = 75% 0.30 (0.25) M-GCAG 0.35 (0.25) 0.123 M-ATCC \ C2 (26) 36 \ 38 = 95% 0.31 (0.29) M-GCAG

M-ATCC C3 (8) 23 \ 26 = 89% 0.42 (0.24)

Numbers in parenthesis are SD.

6.5 Discussion 6.5.1 Absence of clonal genotypes

There has been much debate in the literature as to the extent to which root suckers contribute to the population maintenance and expansion of A. luehmannii and C. pauper. Most of the debate stems from an apparent paucity of seedlings despite abundant production of seed, and the presumption that any juveniles are root suckers which have established following root disturbance. In the Wimmera region of Victoria, most young stands of A. luehmannii are believed to be the result of suckering (Macaulay and Westbrooke in prep.). Root suckers of C. pauper are believed to be common (Hall et al. 1964; Beadle 1981), although mainly in erosion “halos” (Chesterfield and Parsons 1985). This is the first study which uses genetics to show that mature, clonal trees are not present in existing woodlands.

The sampling of individuals within populations was biased to maximise the chances of detecting clonal individuals. The absence of clonal genotypes recorded here from mature stands of A. luehmannii and C. pauper in north-west Victoria, indicates that clonal growth was not extensive

-173- Chapter 6 Conservation genetics in this area in the past. This finding emphasises that field observations of a plant’s ability to reproduce clonally does not always correspond with extensive clonal growth detected using genetic markers.

This study does not preclude the presence of extensive clonal growth in other populations however other populations in more favourable environmental conditions may be even less likely to exhibit clonal growth. The absence of clonal growth in the sampled populations is despite both C. pauper and A. luehmannii being at the edge of their respective distributions which is believed to favour clonality over sexual reproduction (Eckert and Barrett 1993; Peck et al. 1998; Dorken and Eckert 2001). Also, clonal growth is often enhanced as precipitation decreases and environmental conditions become less favourable (Mitton and Grant 1996; Mandujano et al. 2001; Mejias et al. 2002).

The absence of clonal genotypes from these populations affirms the importance of seedling recruitment in these populations and concurs with Chapter 4 which found a higher than expected ratio of seedlings to root suckers of A. luehmannii in these populations. In the past, root suckers may have had little, if any importance, in the natural recruitment dynamics of either A. luehmannii or C. pauper. There are three possibilities why the present day occurrence of young root suckers, does not translate to the presence of clonal genotypes in mature woodlands:

1. Seedlings have been misidentified as root suckers,

2. Root suckers do not persist to maturity, or

3. The disturbance events which lead to suckering are more frequent now than they were historically.

Clonal individuals of these species cannot be identified by above-ground morphology and must be excavated (pers. obs.). Excavation is labour intensive, damaging to the plant and may be inconclusive where the roots connecting sister ramets have decayed (Harada et al. 1997; Sydes and Peakall 1998; Bushakra et al. 1999). However where excavation occurs, such as in Chapter 4, root suckers are clearly identifiable by the persistent connection to the parent tree. It is unlikely that excavated juveniles (<2m tall) would be misidentified. However, some authors (e.g. Sandell 2002) assume without excavation that juveniles are root suckers because they have established outside a recruitment event, or because some individuals are clearly suckers on exposed roots. Such assumptions would lead to an over estimation of the proportion of root suckers in populations.

High ramet mortality would manifest as an absence of root suckers in stands of mature trees. High mortality of young ramets has been observed in other species (e.g. Williams 2000; Auge et al. 2001) and in some species the root suckers do not persist following the death of the parent plant

-174- Chapter 6 Conservation genetics

(Lacey and Johnston 1990). The formation of an independent root system is a prerequisite for independent growth but in clonal trees it often taken decades (De Byle 1964; Zahner and De Byle 1965; Kormanik and Brown 1967). Large root suckers of C. pauper and A. luehmannii are susceptible to wind-throw when they do not develop their own root system (pers. obs.). Although this does not appear to kill the sucker, presumably it precludes them from becoming a mature tree. However, root suckers of both species appear able to develop their own root system and / or produced seed or pollen indicating they are reproductively mature (Chapter 4). So, whilst some mortality of root suckers is probable, I believe that root suckers could persist to maturity.

It is apparent that some form of root disturbance is a pre-requisite for the formation of root suckers in these species (Chapter 5). Such disturbance would generally be human-mediated, e.g. machinery, trampling by domestic stock, or land-use practices which have lead to increased erosion by wind and water (Stanley 1983). Root disturbance prior to European settlement may have resulted from burrowing by native marsupials (e.g. Eldridge and Rath 2002) and these species were far more common in the past (Wakefield 1966). Natural erosion could also have damaged roots although rates prior to European settlement probably did not exceed a few millimetres per thousand years (Galloway 1983). Human-mediated causes of root damage appear more prevalent and severe and it is probable that disturbance leading to roots sucker is more common now than historically. This is the most likely cause of young root suckers without clonal, mature trees (Batty and Parsons 1992).

6.5.2 Genetic structuring of the population

Allocasuarina luehmannii and C. pauper are primarily dioecious trees with wind-dispersed pollen and seeds. Dioecious species are outcrossing and wind-pollinated species generally have high levels of gene flow so that most genetic variation is contained within, rather than between, populations (Huff et al. 1993; Nybom 2004).

The proportion of polymorphic bands was very high but this is not a comparable measure across studies because primers are chosen for their ability to produce polymorphic bands; the more primers trialled the higher the percentage of polymorphic loci (Nybom and Bartish 2000). However, Shannon’s Index is a robust measure comparable across studies (Bussell 1999). Genetic diversity of A. luehmannii and C. pauper (SPOP = 0.29 and 0.34 respectively) was moderate but is comparable to other studies (range 0.17 to 1.47) and the genetic differentiation between populations (GST or SBETWEEN = 0.056 and 0.123 respectively) is also comparable (range 0.16 – 0.61) with outcrossing tree species (Table 6.3). Insect-pollinated and / or animal dispersed outcrossing species tend to have less genetic variation and larger population genetic structuring than wind-dispersed and wind-pollinated species. Similarly, population genetic structuring in inbreeding species tends to be much higher (>0.50 Bussell 1999). Genetic diversity is negatively

-175- Chapter 6 Conservation genetics correlated with population differentiation (Nybom and Bartish 2000; Nybom 2004), so the low values of differentiation suggest a higher than observed genetic variation.

Two other studies have used Shannon’s Index to determine the diversity of clonal or potentially clonal, wind-pollinated dioecious trees: Trembling Aspen (Populus tremuloides,Yeh et al. 1995) and the Monkey Puzzle Tree (Araucaria araucana, Bekessy et al. 2002). P. tremuloides is extensively clonal via root suckering and root suckering has been reported in A. araucana. As in the current study, high diversity (SPOP = 0.65 for both studies) and little differentiation between populations (GST = 0.026 and 0.128 respectively) were reported (Table 6.3).

Table 6.3 A summary of other studies with have used RAPD or AFLPs and Shannon’s Diversity Index to estimate genetic diversity. All species listed are trees and a variety of pollination and dispersal mechanisms are involved. SPOP is the mean value of S (Shannon’s Index) over all populations studied and GST (also known as SBETWEEN) is the proportion of genetic variation held between populations; low values indicate a high level of gene flow.

Species Pollination / dispersal SPOP GST Reference Araucaria araucana wind / wind 0.65 (Bekessy et al. 2002) Allocasuarina luehmannii wind / wind 0.29 0.056 This study Casuarina pauper wind / wind 0.34 0.123 Chamaecyparis formosensis wind / wind 0.329 (Hwang et al. 2001) Chamaecyparis taiwanensis 0.429 Fitzroya cupressoides wind / wind 0.55 (Allnutt et al. 1999) Picea glauca wind / wind 0.416 (Rajora 1999) Picea mariana wind / wind 0.234 (Isabel et al. 1995) Populus tremuloides wind / wind 0.65 (Yeh et al. 1995) Plathymenia reticulata insect / wind 0.332 0.160 (Lacerda et al. 2001) Swietenia macrophylla insect / gravity and 0.36 0.20 (Gillies et al. 1999) wind

Caesalpinia echinata insect / explosive 0.373 (Cardoso et al. 1998) dehiscence Gliricidia sepium insect / explosive 0.599 (Chalmers et al. 1992) dehiscence Colubrina oppositifolia insect / explosive 1.468 0.375 (Kwon and Morden 2002) dehiscence

insect / animal Alphitonia ponderosa 1.152 0.381 dispersed Prunus mahaleb insect / animal 0.172 (Jordano and Godoy 2000) Chaenomeles cathayensis 0.337 0.606 (Bartish et al. 2000) Chaenomeles japonica 0.620 0.278 Chaenomeles speciosa 0.706 0.176 Cedrela odorata 1.460 0.55 (Gillies et al. 1997)

-176- Chapter 6 Conservation genetics

High genetic diversity and low population genetic structure are correlated with longevity, large population size, outbreeding, wind dispersal of pollen and seeds, wide geographic distribution and occurrence in late successional habitats (Loveless and Hamrick 1984; Hamrick and Godt 1990; Nybom and Bartish 2000; Nybom 2004). Both A. luehmannii and C. pauper meet all of these criteria so low population genetic structure is not unexpected. Occasional, very long-distance dispersal of seeds or pollen would contribute to high levels of gene flow (Yeh et al. 1995), such as the long-distance dispersal of C. pauper and A. luehmannii seeds in the droppings of Emus or in flood-waters (Auld 1995c). The moderate values of Shannon’s Index may have been influenced by the sampling regime which was biased to detect clonal genotypes.

6.5.3 Impact of introducing clonal genotypes for restoration

The restoration of dominant species in plant populations via direct seeding or hand-planting has been a cause for some concern regarding the genetic integrity of both fragments and restored patches (Lesica and Allendorf 1999; Thomas et al. 1999; Gustafson et al. 2004). A major concern when restoring minimally disturbed sites such as national parks using non-local seeds is possible intraspecific genetic variation in response to the climatic, edaphic, or biotic environment. This may result in poor establishment or persistence of the restored population (Lesica and Allendorf 1999). For example, intraspecific genetic differentiation in mycorrhizae dependency has been identified (Schultz et al. 2001). This may be particularly important for the Casuarinaceae where variation in the occurrence of filamentous soil bacteria of the genus Frankia (Actinomycetales) has been observed (Lawrie 1982; Reddell and Bowen 1985) and may limit the establishment of non-local seedlings. Genetic contamination (e.g. outbreeding depression), the loss of locally adapted genotypes, selection for traits related to propagation, or differences in population phenology between the introduced and remnant population (e.g. flowering time), may make some restoration efforts harmful, rather than beneficial, to remnant vegetation (Millar and Libby 1989; Lesica and Allendorf 1999).

Root suckers as a tool for restoration have none of the potential problems which may result from reseeding with non-local seeds. Root suckers are derived from existing genotypes, allowing for and retaining local adaptation, variation and phenology, even on a very fine scale within a site. However, the genetic and physical implications on remnant populations of the deliberate, artificial introduction of clonal genotypes to a sexual population have not been considered previously.

Restoration should preserve the genetic balances present in remnant vegetation (Lesica and Allendorf 1999) and clonality may reduce genetic variation (Hamrick and Godt 1996; Young et al. 2002). However, clonal species tend not to be genetically depauperate where the species is long-lived, there are low levels of seedling recruitment, environmental heterogeneity, or frequency-dependent selection involving pathogens (McClintock and Waterway 1993; Watkinson

-177- Chapter 6 Conservation genetics and Powell 1993; McLellan et al. 1997; Holderegger et al. 1998; Eckert 1999). In addition polyploid species such as A. luehmannii are believed to retain genetic diversity for longer than diploid species (Glendinning 1989). Seedling recruitment is especially important; three sexually derived individuals per generation are sufficient to produce patterns of allelic variation similar to sexually reproducing populations (Bengtsson 2003). Therefore introduction of clonal genotypes into sexual populations of A. luehmannii and C. pauper would be unlikely to impact negatively on population genetic structure because subsequent seedling recruitment would create new genotypes through recombination. The introduction of clonal genotypes to supplement recruitment may actually increase diversity. Stimulating root suckers allows genotypes to persist in the population for longer, thus increasing the probability that it is exposed to conditions conducive to the establishment of seedlings (Auld 1990).

Caution must be exercised as to the number of clonal genotypes introduced to a population. Too many clones of a single genotype may have a disproportionate influence on the genotypes of sexual propagules. This could be considered a form of the “founder effect” where the restoration of a remnant population using a small number of individuals results in altered allele frequencies and loss of genetic variation (Lesica and Allendorf 1999). High density stands of root suckers are also undesirable because they may limit the sites available for recruitment of seedlings of this and other species (e.g. Young et al. 2002), especially in semi-arid woodlands which tend to be quite open. Pollen flow may also be obstructed if the sex ratio of dioecious species differs substantially from parity (e.g. McClintock and Waterway 1993; Reusch 2001). It is also possible that initiating extensively clonal stands of monoecious individuals would result in fitness costs associated with geitonogamy (Kennington and James 1997; Eckert 2000; Reusch 2001; Dorken et al. 2002).

All of these limitations would likely be overcome by using a conservative approach, such as aiming for little more than a one to one replacement of existing trees rather than promoting extensive clonal stands. However, a final caution: although root suckers were observed to produce seeds and pollen, there is no definitive evidence that they can persist to become mature, independent plants with their own root system and a desirable adult form. Therefore, the artificial introduction of root suckers should not be viewed as the final solution for restoring semi-arid woodlands, but rather an emergency measure for degraded woodlands with little natural recruitment.

-178- Chapter 7 Conclusions

Chapter 7 Summary of results The semi-arid woodlands of north-west Victoria, like many arid and semi-arid landscapes worldwide, are degraded (Westbrooke et al. 2001; Vallejo et al. 2006). This study was undertaken in response to concerns about the lack of regeneration of overstorey dominants in semi-arid woodlands of north-west Victoria. The theory, models and tools of restoration ecology in arid areas are explored and developed, focusing on semi-arid woodlands and the recruitment of two dominant woodland species: Casuarina pauper and Allocasuarina luehmannii. The structure of this research is outlined in Figure 7.1. Each box is dealt with in more detail in the following sections of this chapter.

7.1 Background information sources 7.1.1 Restoration ecology in arid areas

To restore an ecosystem, it is necessary first to understand its components, their interactions and the effects of management (Hobbs and Norton 1996; Yates and Hobbs 1997). Restoration ecology draws on disturbance ecology, succession theory and assembly rules to assist the recovery of an ecosystem that has been degraded (SER 2002) and restore valued processes or attributes of a landscape (Davis and Slobodkin 2004). There are two key drivers of vegetation change in arid areas: rainfall and grazing (Ludwig et al. 1997). Rainfall is the trigger for most events including growth, recruitment and regeneration (Chesson et al. 2004; Ogle and Reynolds 2004) and is a key influence on restoration success (Whisenant 1999). Further, whilst the amount of rainfall received is outside the influence of land managers, how the landscape uses rainfall reflects the degree of degradation. For example, smaller, shorter and less frequent production pulses follow rainfall and indicate that resources are being lost from the landscape rather than being converted into biomass (Tongway and Ludwig 1997).

Grazing initially leads to altered vegetation communities with more undesirable and fewer desirable species. Where minimal degradation has occurred, restoration success may be achieved by reducing grazing pressure, reintroducing component species and reducing competition from inappropriate species. However, in severely degraded landscapes vegetation cover is reduced overall and soil properties are altered. In these circumstances, simply reducing grazing pressure will be unlikely to result in positive vegetation change (Friedel 1991). In particular, restoration must address water and nutrient cycling by restoring soil function and increasing availability of soil moisture before any recovery of the vegetation can be expected. Australian (e.g. Ludwig et al. 1997; Brown et al. 2003) and overseas texts (e.g. Whisenant 1999; Perrow and Davy 2002; Falk et al. 2005; van Andel and Aronson 2006) deal extensively with the tools to allow restoration to be achieved. This thesis focuses on techniques to enhance the recruitment of desirable species.

- 179 - Chapter 7 Conclusions

7.1.1 Restoration ecology in arid 7.1.2 Semi-arid woodlands north-west 7.1 areas Victoria BACKGROUND Provides theory, models and tools Focus of research, degraded and for restoring valued processes. requires restoration.

7.2 State-and-transition model 7.2 ORGANISATION S-T framework developed and applied to HKNP. Gaps OF KNOWLEDGE identified for research, results feedback to model.

7.3 7.3.1 Natural recruitment 7.3.2 Root suckers are a viable tool RESEARCH: HKNP: A. luehmannii seedlings for restoration Casuarina recruited post 1996 following Produced in low numbers by damaging pauper and reduction of grazing pressure. and/or exposing shallow roots of male Allocasuarina luehmannii in Seedlings positively associated or female trees. Treat in spring. Hattah Kulkyne with bare areas. Little negative impact on population and Murray genetics. Sunset National Parks

FEEDBACK TO S-T MODEL

Enhancing recruitment of Practical initiation of root suckers seedlings Technique for broad scale, mechanised Role of competition with ground initiation of suckers to achieve one-for- flora, spatial variation in soil one replacement of existing trees. moisture and variation in the quantity of seed produced. 7.4 FUTURE RESEARCH Other gaps/opportunities: Final transition to a desirable, restored woodland community; Success of direct seeding; Emerging weeds.

Figure 7.1 The structure, organisation and key findings of research into the restoration ecology of semi-arid woodlands in north-west Victoria.

- 180 - Chapter 7 Conclusions

7.1.2 The semi-arid woodlands of north-west Victoria

The semi-arid woodlands of Australia have a non-eucalypt overstory and include Belah woodlands (dominated by C. pauper) and Pine-Buloke woodlands (dominated by A. luehmannii and Callitris gracilis). In north-west Victoria, the woodlands originally comprised only a small proportion of the total vegetation cover but are now the most degraded vegetation community in the state. This is a result of extensive clearing, overgrazing from domestic, introduced and native mammals, fire and competition from weeds since European settlement (LCC 1987). For this reason these woodlands, and the dominants Casuarina pauper and A. luehmannii, were the focus of this research. Hattah Kulkyne (HKNP) and Murray Sunset National Park (MSNP) in north- west Victoria contain much of the remaining woodlands. In these parks grazing pressure has been progressively reduced and direct seeding and hand-planting are used in an effort to encourage the progressive long-term recovery of key, woody, perennial vegetation (DNRE 1996b). This research explored further the options for the restoration of semi-arid woodlands, specifically novel techniques to enhance recruitment of key species.

Rainfall is typically erratic and soil moisture is affected by high evaporation. Rainfall records from 1964 in HKNP indicated that over 70% of rainy days receive < 5 mm which would stimulate little plant growth and recruitment. However, higher rainfall during periods critical for the establishment of perennial seedlings (summer and autumn) will result in greater recruitment success (Harrington 1991). I developed a simple model of soil moisture based on daily rainfall and evaporation at HKNP to indicate the frequency of conditions conducive to recruitment success. Conditions required for recruitment include a wet summer or autumn (>30 days of soil moisture greater than 50%) not followed by a dry summer (>30 days of 0% soil moisture: Harrington 1991). This eliminated all years except 1971 and 1973; although 1980 and 1992 met the wet and dry criteria, there was a dry period in the following autumn. The wet criterion only was met in 1974, 1986, 1989 and 1990. The period post 1993 was particularly poor with long summer and autumn dry periods in every year except 1996 and above median rainfall only in 1995, 1999 and 2000.

7.2 Organisation of knowledge: State-and-transition modelling

State-and-transition (S-T) models provide an approach to organise knowledge of vegetation change in a logical manner, draw attention to gaps in knowledge and encourage the incorporation of new information (Westoby et al. 1989). They are favoured in arid areas (e.g. George et al. 1992; Ash et al. 1994) where change is often discontinuous rather than successional, affected by unpredictable climatic events and often not reversible on a practical time scale (Friedel 1991) or through ecological resilience (Briske et al. 2005). I used the principles of S-T modelling to

- 181 - Chapter 7 Conclusions organise general knowledge on restoration ecology and semi-arid woodlands into a framework for vegetation change and developed a S-T model specifically for vegetation change and recovery in HKNP. The gaps identified in the HKNP model formed the basis for further research.

S-T models have been constructed for many vegetation communities but this is the first time a general framework has been developed for semi-arid habitats. The framework outlined the causes and possible solutions for shifts between degraded and desirable vegetation communities and clearly defined two types of vegetation change: phase shifts within vegetation states and transitions between vegetation states. Phase shifts follow small climatic fluctuations or altered grazing pressure and are easily reversible, but transitions between vegetation states are not. Vegetation states are separated by thresholds which indicate primarily ecological processes have been altered and must be actively restored. Thresholds and states in arid areas are limited to loss of component species, dominance of inappropriate species and finally, loss of soil functioning (including soil structure, hydrology or chemistry). Vegetation becomes progressively more degraded as these thresholds are crossed and restoration must also proceed in a stepwise manner. The framework also emphasised the importance of defining the desirable vegetation community (the goal), considering the intended land use, e.g. productive, aesthetic or conservation. The degree of degradation also affects goal setting; if ecological functioning has been permanently damaged some desirable communities may be impossible to achieve.

Qualitative accounts of vegetation management in HKNP were fitted to this general framework to create a model which specifically addressed vegetation change and recovery in HKNP. This represents the first time that the experiences of staff in the park, historical photographs and reports have been drawn together to examine vegetation composition, management of grazing pressure and restoration attempts. Photographs and indicators of the composition (overstorey and understory structure and recruitment) and structure of plant communities (reintroduction potential, increasers, decreasers) were used to define vegetation states and phases. The resultant model does not provide definitive guidance but is a summary of existing knowledge and highlights restoration failures and gaps in the knowledge base. Key opportunities identified for further research were learning from natural recruitment and exploring root suckers as a tool for restoration.

7.3 Research 7.3.1 Lessons from natural recruitment

The state-and-transition model highlighted the occurrence of spatially limited, natural recruitment of a variety of woody, perennial species in HKNP. I hoped to provide management options to enhance the success of future restoration activities by exploring the factors associated with recruitment. I surveyed large areas for evidence of recruitment of A. luehmannii in HKNP and C. pauper in MSNP. These species were priorities because both are the target of restoration activities

- 182 - Chapter 7 Conclusions however the success of direct seeding and hand-planting is limited (P. Murdoch, Parks Victoria – Hattah, pers. comm. 2002). Extensive survey coverage was important because a small survey area combined with the high spatial variability of recruitment may contribute to studies “overlooking” recruitment (Watson et al. 1997a, 1997b). Accordingly, only 1.5 (C. pauper) and 1.8 (A. luehmannii) patches of recruitment were noted per transect km (80 m width).

In MSNP recruitment of C. pauper over 71 transect km was limited and predominantly root suckers (93%). Root suckers were confirmed via excavations to 20-40 cm depth and were found to be associated with erosion halos around isolated trees or on the edges of stands. This was not unexpected and has been recorded as the dominant mode of recruitment for this species (Chesterfield and Parsons 1985; Cunningham et al. 1992).

The lack of seedling recruits of C. pauper in MSNP is possibly due to higher grazing pressure compared to HKNP. The culling of kangaroos did not commence in MSNP until 2001, five years later than in HKNP and abundance of kangaroos is still greater. Extended periods of low-grazing pressure, dependent on the amount of rainfall received, may be necessary before there is sufficient grass for grazers to not consume seedlings of perennial trees and shrubs (Coulson et al. 1990). Once the cover of grasses has been reduced to low levels, even low grazing pressure suppresses regrowth and maintains the vegetation in a degraded state, especially if rainfall is low (Noy-Meir 1975). The greater susceptibility of seedlings than root suckers to grazing pressure may also explain the predominance of root suckers over seedlings.

The recruitment of A. luehmannii observed over 135 transect km in HKNP was predominantly seedlings (66%). I compared the basal stem diameter of known age seedlings with unknown age seedlings and the size differences supported a post 1995 establishment date. In addition, vegetation surveys completed in 1996 (Sluiter et al. 1997a) failed to record seedlings which are now present in abundance in the surveyed areas and I noted additional seedlings present inside quadrats first surveyed in 2001 (Gowans and Westbrooke 2002). This suggests multiple recruitment events post-1996.

The recent recruitment of seedlings was unexpected and contrary to the literature which suggests natural seedling recruitment is rare and associated with exceptional rainfall conditions as for many other semi-arid, woody perennials (e.g. Hall et al. 1964; Crisp 1978; Chesterfield and Parsons 1985). The years 1996, 1999 and 2000 have been the only years post-1995 to receive rainfall above the median, however none of these years experienced the critical wet period during summer or autumn (Harrington 1991), and all had dry periods in the following summer.

Reduced grazing pressure is correlated with the seedling recruitment of A. luehmannii at HKNP. In 1996, Rabbit Haemmorhagic Disease decimated rabbit numbers. High mortality of kangaroos during the drought years of 1994-5 and the extension of the culling program throughout HKNP in

- 183 - Chapter 7 Conclusions

1996 greatly reduced the abundance of kanagaroos. The apparent reliance on a high rainfall event for recruitment of semi-arid species may be an artefact of high grazing pressure (Wiegand et al. 2004). There is increasing evidence that some germination of semi-arid species occurs in most years, however grazing eliminates the seedlings and prevents recruitment (Watson et al. 1997a). Under grazing, only high rainfall years would result in sufficient seedlings and alternative food to allow some seedlings to escape browsing (Milton and Wiegand 2001). The recruitment of A. luehmannii seedlings suggests that under low grazing pressure seedlings can persist.

Seedling recruits of A. luehmannii were found within 40 m of a female tree assumed to be the parent (mostly within 10-20 m). Twenty of 21 parent trees examined with more than 20 seedlings had a large area (>264 m2) within 30 m of the tree which was devoid of vegetation, or occasionally with only perennial grasses. Nearly all (97%) of the 1113 seedlings surrounding these trees occurred in these bare areas. Bare areas were present in only 4 of 52 non-parent trees examined. The cause of bare areas is unknown but the establishment of A. luehmannii seedlings in them suggests that competition of seedlings with annual species, but not perennial grasses, is detrimental. The increase in annual weeds in overgrazed rangelands alters seasonal water relations in the critical summer / autumn period (Anderson and Hodgkinson 1997; Maestre et al. 2001) and has been shown to inhibit establishment of seedling, woody, perennial species (Davis et al. 1998; Gordan and Rice 2000; Rey Benayas et al. 2001). Woody seedlings establish more readily in the interstitial spaces between perennial grass tussocks than within dense annual grasses (Gordan and Rice 2000). Annual weeds are currently controlled by a single application of knockdown herbicide prior to direct seeding, or local removal at the time of hand-planting. No follow-up treatment is applied. Better control of weeds (e.g. Rey Benayas et al. 2001), or introducing propagules to areas dominated by perennial grasses (e.g. Maestre et al. 2001) may enhance the success of restoration activities.

The occurrence of seedlings also showed some association with hydrology: 67% of parents occurring in areas where moisture accumulates following rainfall, compared with only 21% of non-parents. This association is supported by the highest density recruitment during field surveys observed in Riverine woodland with A. luehmannii trees 50-100 m apart. Riverine woodlands occur in low-lying areas subject to occasional flooding from the Murray River. The inferred importance of soil moisture suggest further investigations may be warranted into water harvesting which utilises artificially created runon zones to enhance restoration success (Whisenant 1999). Enhanced recruitment in runon zones has been observed by others (Chesterfield and Parsons 1985; Ludwig 1987; Woodell 1990; Ireland 1997; Tiver and Andrew 1997).

Seedlings were also associated with trees which produced a heavy cone crop. Although recruitment of A. luehmannii in the stand as a whole is not believed to be limited by seed production, seed limitation may occur in areas with trees which are poor producers of seeds.

- 184 - Chapter 7 Conclusions

Maternal variation in seed production and whether good seed producers do so consistently or vary annually is worth investigating further, as a tree which produces reliably good quantities of seed is an asset for collecting seed for revegetation. Greater production of seed means a higher likelihood that seeds will find a suitable microsite (Eriksson and Ehrlén 1992).

Variation in seed quality between trees was not correlated with the occurrence of seedlings but the germination trials were probably confounded by differing maturity of seed at the time of collection. Dark-colored, mature seed had a higher percentage germination and germination was more rapid than for light-colored seed. This sends a clear message that care should be taken to only collect mature seed for restoration activities.

In summary, the lessons learnt from natural recruitment in HKNP are – continue restoration efforts even in low rainfall years, maintain very low grazing pressure, consider better control of weeds or focus on areas dominated by perennial grasses, consider increasing soil moisture by utilising water harvesting techniques, and revisit good seed producing trees and ensure only mature seed is collected for restoration.

7.3.2 Root suckers as a tool for restoration

7.3.2.1 Initiation of root suckers The state-and-transition model highlighted the frequent failure of direct seeding and hand-planting due to low rainfall and high grazing pressure (Thomas 2003). Root suckers were explored as an alternative tool because there is evidence that they are better able to establish and persist under these conditions than are seedlings (Callaghan 1984; De Steven 1989; Peterson and Jones 1997; Barsoum 2001; Bond and Midgley 2001). I surveyed existing root suckers and initiated new root suckers from existing C. pauper and A. luehmannii trees to develop “best-bet” recommendations for producing root suckers for restoration.

Damage and / or exposure which produced callus tissue always preceded the growth of root suckers of C. pauper and A. luehmannii (see also Jones and Raynal 1986), however not all roots produced callus and not all callus resulted in suckering. The percentage of roots producing callus tissue could only be counted accurately on roots where the treatement involved scraping bark from the root. Approximately 50% of exposed, scraped roots died or were mostly dead, probably due to heat, and these did not produce callus. Excluding these, callus formed in response to damage to the vascular cambium in approximately 85% of exposed and 81% of buried roots. The calli on buried, scraped roots did not produce root suckers and only 13% of the calli on exposed roots produced suckers. Suckers were also produced from callus tissue from cracks in the bark in 9% of exposed, alive or mostly dead roots. The absence of long-lived suckers on buried roots suggests that light is an important trigger for sucker growth (van der Heyden and Stock 1996) and

- 185 - Chapter 7 Conclusions root carbohydrate reserves (Schier and Zasada 1973; Frey et al. 2003) are insufficient to sustain growth of suckers on roots deeper than approximately 5cm.

Approximately 40% of root suckers initiated on treated roots died with two years of treatment, often less than this. Generally all suckers died following the death of an exposed root indicating the root needs to be damaged sufficiently to produce suckers but not to kill the root. Treated roots ranged from 16 to 55 mm however the smaller roots appeared less able to survive exposure. The choice of root diameter is therefore a trade-off between smaller roots which may produce more suckers because root carbohydrate reserves are greater (Reichenbacker et al. 1996), and larger roots which are more resilient to damage and exposure. If damage and exposure can be minimised, targeting roots with diameters between 20 and 30 mm seems appropriate.

Root suckers from treated roots were most often bushy and rarely had the desirable, single stemmed form. Whilst self-thinning may result in single-stemmed suckers (Jones and Raynal 1986; Peterson and Jones 1997; Bond and Midgley 2001; Frey et al. 2003; Fraser et al. 2004), it is possible that minimising the amount of damage and exposure would be more successful in producing single stemmed suckers.

Field surveys indicated most suckers were within 10 m of the parent tree. The reason for this was not examined but this observation suggests this is the optimal distance to stimulate suckers for restoration purposes; this may be increased for large trees which tend to have a broader spread of roots. The gender of the parent tree had little effect on the suckering response or health of treated roots for either A. luehmannii or C. pauper, or on the occurrence of root suckers in field surveys. The season of treatment was not demonstrated to be important although an early spring treatment may allow a longer growth period following treatment and faster initiation of suckers.

Physiological integration with the parent tree provides important sustenance for the root sucker (Chapter 5). Although some disconnected suckers did persist for over two years, it seems prudent for restoration purposes to attempt to damage but not sever roots when attempting to initiate sucker growth. The ability to eventually become independent of the parent tree is a critical requirement for suckers to be a useful restoration technique. Root suckers appear to be reliant on the parent tree for many years; only very large suckers (around 40 mm basal diameter) with substantial distal thickening had produced their own roots. It is not unusual for suckers of tree species to remain connected to the parent tree for decades (De Byle 1964; Zahner and De Byle 1965; Kormanik and Brown 1967; Jones and Raynal 1986).

Although not previously considered in restoration ecology texts, the deliberate initiation of root suckers in the field was found to be a viable tool for restoration. Suckers of A. luehmannii and C. pauper were initiated, albeit in low numbers, during a period of very low rainfall. During this time no new seedlings were detected in woodlands surrounding treated trees. This is not conclusive

- 186 - Chapter 7 Conclusions evidence of a lower reliance on soil moisture, but does demonstrate that suckers can be initiated regardless of poor rainfall.

A suitable technique needs to be developed for the broad-scale, mechanised initiation of root suckers based on the findings above. It should treat a number of roots per tree to allow for low success rates. The damage should be to very shallow roots and may involve some excavation of roots. The treatment should aim to scrape but not cut roots to encourage the production of suckers on callus from damage rather than from cracks.

7.3.2.2 Genetic implications of initiating establishment of root suckers There has been much debate in the literature as to the extent to which root suckers contribute to the population maintenance and expansion of A. luehmannii and C. pauper. Most of the debate stems from an apparent paucity of seedlings despite abundant production of seed, and the presumption that any juveniles are root suckers (Hall et al. 1964; Beadle 1981; Chesterfield and Parsons 1985; Cunningham et al. 1992; Macaulay and Westbrooke in prep.). It is important to understand the mating system of these species to determine whether current restoration focusing on seedlings is appropriate and also to predict the impact of artificially initiating root suckers in existing populations. This is the first study which uses genetics to determine whether mature, clonal trees are present in existing woodlands of A. luehmannii and C. pauper.

No identical genotypes were detected in the genetic fingerprint (Amplified Fragment Length Polymorphisms) obtained for 82 individuals of A. luehmannii from three populations or from 65 individuals of C. pauper from three populations. This does not exclude the presence of clonal growth in other populations however it implies that in the past, root suckers have played little role in the recruitment dynamics of these species. The use of seed and tubestock for restoration is validated. The present day occurrence of young root suckers in these populations (Chapter 4) is most likely because the frequency of disturbance resulting in root suckers is more common now than historically (Batty and Parsons 1992).

The proportion of genetic diversity held within the A. luehmannii and C. pauper populations was very high, 94.4% and 87.7% respectively. This indicates a high level of gene flow and suggests that any negative impact on population genetics from introducing small numbers of root suckers would be minimal. Even small amounts of natural seedling recruitment following reduced grazing pressure (as discussed in Chapter 4) would be sufficient to restore patterns of genetic variation (Bengtsson 2003). In addition, root suckers are derived from existing genotypes, allowing for and retaining local adaptation, variation and phenology, even on a very fine scale within a site. This often not the case when tubestock or seed is introduced from remote populations (Lesica and Allendorf 1999; Thomas et al. 1999; Gustafson et al. 2004). Finally, stimulating root suckers allows a genotype to persist in the population for longer, thus increasing the probability that it is

- 187 - Chapter 7 Conclusions exposed to conditions conducive to the establishment of seedlings (Auld 1990). However, caution should be exercised as to the number of root suckers introduced to a population. Too many root suckers of a single genotype may have a disproportionate influence on the genotypes of sexual propagules. Generating clonal stands may obstruct the flow of pollen in these dioecious species or may limit the sites available for recruitment of seedlings of this and other species. There is also no genetic evidence that root suckers can persist to become mature, independent plants with their own root system and a desirable adult form, although flowering and fruiting suckers were observed. Initiation of root suckers should aim for little more than a one to one replacement of existing trees and the artificial introduction of root suckers should be viewed only as an emergency measure to halt the loss of woodlands through senescence without recruitment.

7.4 Future research

This research took a structured approach, using state-and-transition modelling, to identify two key areas to enhance recruitment of A. luehmannii and C. pauper: natural seedling recruitment and the initiation of root suckers. Contrary to the literature, seedling recruitment of A. luehmannii was detected in HKNP. This was broadly attributed to reduced grazing pressure and was correlated positively with the presence of bare areas, runon zones and trees with a high production of seeds. Experimental manipulation is required to elucidate causation in these relationships. Determining whether trees are consistently good producers of seeds should have immediate practical benefits for seed collection. Little seedling recruitment of C. pauper was observed in MSNP. Whether this is attributable to grazing pressure or a greater dependence on soil moisture than A. luehmannii will have implications for restoration work involving this and other species in MSNP.

The research demonstrated that initiating root suckers of A. luehmannii and C. pauper was a viable tool for enhancing recruitment of these species, however a suitable tool for broadscale, mechanised initiation of suckers needs to be developed. Although young root suckers were detected in the field, genetic studies did not identify any clonal, mature trees and the persistence of root suckers to maturity requires further investigation. Root suckers are likely to be resistant to browsing although this area requires further work. Finally, investigating use of similar techniques to initiate root suckers of other species would enhance the applicability of this research.

The state-and-transition model also highlighted further opportunities for research not investigated in this thesis: An understanding of, and techniques for, the final transition to a desirable, restored woodland community; techniques to improve the success of direct seeding, and monitoring and management of emerging weeds. Both the general S-T model and that specific to the Pine-Buloke woodlands of HKNP would also benefit from the contributions of other researchers, particularly in defining key vegetation communities. Revisiting the locations of restoration activities in HKNP and recording size and numbers of established plants would be useful.

- 188 -

References

Adams, J., Brugger, C., Ding, Y. L., and Flores, M. (2002). Conserveonline and Fortaleza: sharing conservation success and failure on the internet. Conservation Ecology (online) 6, 7. Aerts, R., and van der Peijl, M. J. (1993). A simple model to explain the dominance of low- productive perennials in nutrient-poor habitats. Oikos 66, 144-147. Agresti, A. (1996). "An introduction to categorical data analysis." (Wiley: New York.) Aguiar, M. R., and Sala, O. E. (1997). Seed distribution constrains the dynamics of the Patagonian Steppe. Ecology 78, 93-100. Aguiar, M. R., and Sala, O. E. (1999). Patch structure, dynamics and implications for the functioning of arid ecosystems. Trends in Ecology and Evolution 14, 273-277. Allen, R. G., Pereira, L. S., Raes, D., and Smith, M. (1998). Crop evapotranspiration: guidelines for computing crop water requirements.FAO irrigation and drainage paper 56. FAO: Rome, Italy. Allen-Diaz, B., and Bartolome, J. W. (1998). Sagebrush-grass vegetation dynamics: comparing classical and state-transition models. Ecological Applications 8, 795-804. Allison, S. K. (2002). When is a restoration successful? Results from a 45 year-old tallgrass prairie restoration. Ecological Restoration 20, 10-17. Allnutt, T. R., Newton, A. C., Lara, A., Premoli, A. C., Armesto, J. J., Vergara, R., and Gardner, M. (1999). Genetic variation in Fitzroya cupressoides (alerce), a threatened South American conifer. Molecular Ecology 8, 975-987. Anand, M., and Desrochers, R. E. (2004). Quantification of restoration success using complex systems concepts and models. Restoration Ecology 12, 117-123. Anderson, V. J., and Hodgkinson, K. C. (1997). Grass-mediated capture of resource flows and the maintenance of banded mulga in a semi-arid woodland. Australian Journal of Botany 45, 331-342. Anonymous (1916). Report on Hattah sanctuary. Victorian Naturalist 32, 138-139. Anonymous (1984). Hattah-Kulkyne National Park and Murray-Kulkyne Park Management Plan National Parks Service, Department of Conservation, Forests and Lands: Mildura. August 1984. Anonymous (1999). Threat abatement plan for competition and land degradation by feral goats Biodiversity Group, Environment Australia: Canberra. Anonymous (2004a). Australian Plant Name Index - What's its name?, Australian National Herbarium, Australian National Botanic Gardens, Australian Biological Resources Study. http://www.anbg.gov.au/win/index.html, 2004. Anonymous (2004b). Long term containment facility Environment Effects Statement - Climate Study URS: Melbourne. December 2004. Anonymous (undated). The Carroll tree tube. Giving your trees a head start Westralian Sands Limited: Capel, Western Australia. ANZECC, and ARMCANZ (1999). National principles and guidelines for rangeland management Australian and New Zealand Environment and Conservation Council and Agriculture and Resource Management Council of Australia and New Zealand, Commonwealth of Australia: Canberra. Archer, S. R., Scifres, C., Bassham, C. R., and Maggio, R. (1988). Autogenic succession in subtropical savannas: conversion of grassland to thorn woodland. Ecological Monographs 58, 111-127. Arista, M., and Talavera, S. (1996). Density effect on the fruit-set, seed crop viability and seedling vigour of Abies pinsapo. Annals of Botany 77, 187-192. Aronson, J., Floret, C., Le Floc’h, E., Ovalle, C., and Pontainer, R. (1993). Restoration and rehabilitation of degraded ecosystems in arid and semi-arid environments. Restoration Ecology 1, 8-17.

- 189 -

Arthur, A. D., Pech, R. P., Drew, A., Gifford, E., Henry, S., and McKeown, A. (2003). The effect of increased ground-level habitat complexity on mouse population dynamics. Wildlife Research 30, 565-572. Ash, A. J., Bellamy, J. A., and Stockwell, T. G. H. (1994). State and transition models for rangelands. 4. Application of state and transition models in northern Australia. Tropical Grasslands 28, 223-228. Ashmun, J. W., Thomas, R. J., and Pitelka, L. F. (1982). Translocation of photoassimilate between sister ramets in two rhizomatous forest herbs. Annals of Botany 49, 403-415. Attiwill, P. M., and Leeper, G. W. (1987). "Forest soils and nutrient cycles." (Melbourne University Press: Melbourne.) Auge, H., Neuffer, B., Erlinghagen, F., Grupe, R., and Brandl, R. (2001). Demographic and random amplified polymorphic DNA analyses reveal high levels of genetic diversity in a clonal violet. Molecular Ecology 10, 1811-1819. Auld, T. D. (1990). Regeneration in populations of the arid zone plants Acacia carnei and A. oswaldii. Proceedings of the Ecological Society of Australia 16, 267-272. Auld, T. D. (1993). The impact of grazing on regeneration of the shrub Acacia carnei in arid Australia. Biological Conservation 65, 165-176. Auld, T. D. (1995a). The impact of herbivores on regeneration in four trees from arid Australia. Rangeland Journal 17, 213-227. Auld, T. D. (1995b). Seedling survival under grazing in the arid perennial Acacia oswaldii. Biological Conservation 72, 27-32. Auld, T. D. (1995c). Soil seedbank patterns of four trees and shrubs from arid Australia. Journal of Arid Environments 29, 33-45. Austin, A. T., Yahdjian, L., Stark, J. M., Belnap, J., Porporato, A., Norton, U., Ravetta, D. A., and Schaeffer, S. M. (2004). Water pulses and biogeochemical cycles in arid and semiarid ecosystems. Oecologia 141, 221-235. Baker, W. L., Munroe, J. A., and Hessl, A. E. (1997). The effects of elk on aspen in the winter range in Rocky Mountain National Park. Ecography 20, 155-165. Barlow, B. A. (1958). Diploid and tetraploid Casuarina littoralis Salisb. (Syn. C. suberosa Ott. and Dietr.). Australian Journal of Botany 6, 38-46. Barlow, B. A. (1959a). Chromosome numbers in the Casuarinaceae. Australian Journal of Botany 7, 230-237. Barlow, B. A. (1959b). Polyploidy and apomixis in the Casuarina distyla species group. Australian Journal of Botany 7, 238-251. Barlow, B. A. (1983). The casuarinas - a taxonomic and biogeographic review. In "Casuarina. Ecology, management and utilisation". (Ed. S. J. Midgley, J. W. Turnbull and R. D. Johnston) pp. 10-18 (CSIRO: Melbourne). Barrett, S. C. H. (2002). The evolution of plant sexual diversity. Nature Reviews Genetics 3, 274- 284. Barritt, A. R., and Facelli, J. M. (2001). Effects of Casuarina pauper litter and grove soil on emergence and growth of understorey species in arid lands of South Australia. Journal of Arid Environments 49, 569-579. Barsoum, N. (2001). Relative contributions of sexual and asexual regeneration strategies in Populus nigra and Salix alba during the first years of establishment on a braided gravel bed river. Evolutionary Ecology 15, 4-6. Bartish, I. V., Garkava, L. P., Rumpunen, K., and Nybom, H. (2000). Phylogenetic relationships and differentiation among and within populations of Chaenomeles Lindl. (Rosaceae) estimated with RAPDs and isozymes. Theoretical and Applied Genetics 101, 554-563. Bastin, G. N., Pickup, G., Chewings, V. H., and Pearce, G. (1993). Land degradation assessment in central Australia using a grazing gradient assessment. The Rangelands Journal 15, 217- 233. Batty, A. I., and Parsons, R. F. (1992). Regeneration of Acacia melvillei in part of semi- arid south-eastern Australia. Proceedings of the Royal Society of Victoria 104, 89-97. Beadle, N. C. W. (1981). "The vegetation of Australia." (Cambridge University Press: Cambridge.)

- 190 -

Beatley, J. C. (1974). Phenological events and their environmental triggers in Mojave-desert ecosystems. Ecology 55, 856-863. Beetham, P. (1987). Diet analysis of the western grey kangaroo (Macropus fuliginosus) in Wyperfeld National Park by microscopic faecal analysis. Victorian Naturalist 104, 32-39. Bekessy, S. A., Allnutt, T. R., Premoli, A. C., Lara, A., Ennos, R. A., Burgman, M. A., Cortes, M., and Newton, A. C. (2002). Genetic variation in the vulnerable and endemic Monkey Puzzle tree, detected using RAPDs. Heredity 88, 243-249. Bell, D. T. (1999). Turner Review No. 1 The process of germination in Australian species. Australian Journal of Botany 47, 475-517. Bellingham, P. J., and Sparrow, A. D. (2000). Resprouting as a life history strategy in woody plant communities. Oikos 89, 409-416. Bengtsson, B. O. (2003). Genetic variation in organisms with sexual and asexual reproduction. Journal of Evolutionary Biology 16, 189-199. Bertness, M. D., and Callaway, R. M. (1994). Positive interactions in communities. Trends in Ecology and Evolution 9, 191-193. Bestelmeyer, B. T., Brown, J. R., Havstad, K. M., Alexander, R., Chavez, G., and Herrick, J. E. (2003). Development and use of state-and-transition models for rangelands. Journal of Range Management 56, 114-126. Boland, D. J., Moncur, M. W., and Pinyopusarerk, K. (1996). Review of some floral and vegetative aspects to consider when domesticating Casuarina. In "Recent Casuarina research and development. Proceedings of the third international Casuarina workshop" Da Nang, Vietnam, (Ed. K. Pinyopusarerk, T. W. Turnball and J. J. Midgley) pp. (CSIRO Forestry and Forest Products: Canberra). Bond, W. J., and Midgley, J. J. (2001). Ecology of sprouting in woody plants: the persistence niche. Trends in Ecology and Evolution 16, 45-51. Bonfil, C. (1998). The effect of seed size, cotyledon reserves, and herbivory of seedling survival and growth in Quercus rugosa and Q. laurina (Fagaceae). American Journal of Botany 85, 79-87. Bonicel, A., Haddad, G., and Gagnaire, J. (1987). Seasonal variations in the starch and major soluble sugars in the different organs of young poplars. Plant Physiology and Biochemistry 25, 451-459. Bonney, N. (2003). "What seed is that? A field guide to the identification, collection and germination of native seed in South Australia." (N. Bonney: Adelaide.) Bosela, M. J., and Ewers, F. W. (1997). The mode of origin of root buds and root sprouts in the clonal tree Sassafras albidum (Lauraceae). American Journal of Botany 84, 1466-1481. Boucher, D. H. (1997). General pattern of age-by-stage distributions. Journal of Ecology 85, 235- 240. Bowers, J. E. (1996). Perennial seedling emergence on Sonoran Desert dunes. Journal of Arid Environments 33, 63-72. Bowers, J. E., and Turner, R. M. (2002). The influence of climatic variability on local population dynamics of Cercidium microphyllum (foothill paloverde). Oecologia 130, 105-113. Bradley, J. (1988). "Bringing back the bush: The Bradley method of bush regeneration." (Lansdowne-Rigby: Sydney.) Brieman, L. (2001). Decision tree forests. Machine Learning 45, 5-32. Brieman, L., Friedman, J. H., Olshen, R. A., and Stone, C. J. (1984). "Classification and regression trees." (Wadsworth: Belmont.) Briske, D. D., Fuhlendorf, S. D., and Smeins, F. E. (2003). Vegetation dynamics on rangelands: a critique of the current paradigms. Journal of Applied Ecology 40, 601-614. Briske, D. D., Fuhlendorf, S. D., and Smeins, F. E. (2005). State-and-transition models, thresholds, and rangeland health: A synthesis of ecological concepts and perspectives. Rangeland Ecology and Management 58, 1-10. Brown, C. L., Hall, F., and Mill, J. (2003). "Plant conservation: approaches and techniques from an Australian perspective." (ANPC: Canberra.) Burbidge, A. H. (1973). "Heteroploidy in the Casuarina distyla group". B.Sc. (Hons.) Thesis. School of Biological Sciences, Flinders University.

- 191 -

Bureau of Meteorology (2004). Bureau of Meteorology climate education. http://www.bom.gov.au/lam/climate.htm, 2005. Burrows, G. E. (1990). Anatomical aspects of root bud development in hoop pine (Araucaria cunninghamii). Australian Journal of Botany 38, 73-78. Buschmann, H., Edwards, P. J., and Dietz, H. (2005). Variation in growth pattern and response to slug damage among native and invasive provenances of four perennial Brassicaceae species. Journal of Ecology 93, 322-334. Bushakra, J. M., Hodges, S. A., Cooper, J. B., and Kaska, D. D. (1999). The extent of clonality and genetic diversity in the Santa Cruz Island ironwood, Lyonothamnus floribundus. Molecular Ecology 8, 471-475. Bussell, J. D. (1999). The distribution of random amplified polymorphic DNA (RAPD) diversity amongst populations of Isotoma petraea (Lobeliaceae). Molecular Ecology 8, 775-789. Butnor, J. R., Doolittle, J. A., Kress, L., Cohen, S., and Johnson, K. H. (2001). Use of ground penetrating radar to study tree roots in the southeastern United States. Tree Physiology 21, 1269-1278. Caetano-Anolles, G. (2001). Plant genotyping using arbitrarily amplified DNA. In "Plant genotyping: the DNA fingerprint". (Ed. R. J. Henry) pp. 29-46 (CAB International: New York). Calaby, J. H., and Grigg, G. C. (1989). Changes in macropodid communities and populations in the past 200 years. In "Kangaroos, wallabies and rat kangaroos". (Ed. G. C. Grigg, P. J. Jarman and I. Hume) Volume 2: pp. 813-820 (Surrey Beatty and Sons: Chipping Norton, NSW). Callaghan, T. V. (1984). Growth and translocation in a clonal southern hemisphere sedge, Unicinia meridensis. Journal of Ecology 72, 529-546. Callister, K. (2004). "Casuarina pauper (Belah) woodlands of north-west Victoria: monitoring and regeneration". PhD Thesis. School of Science and Engineering, University of Ballarat. Callister, K., and Westbrooke, M. E. (2002). Guidelines for the management and rehabilitation of Belah and Pine-Buloke woodland remnants in north western Victoria Centre for Environmental Management, University of Ballarat: Ballarat, Victoria. July, 2002. Canadell, J., and Lopez-Soria, L. (1998). Lignotuber reserves support regrowth following clipping of two mediterranean shrubs. Functional Ecology 12, 31-38. Caraco, T., and Kelly, C. K. (1991). On the adaptive value of physiological integration in clonal plants. Ecology 72, 81-93. Cardoso, M. A., Provan, J., Powell, W., Ferreira, P. C. G., and De Oliviera, D. E. (1998). High genetic differentiation among remnant populations of the endangered Caesalpinia echinata Lam. (Leguminosae - Caesalpinioideae). Molecular Ecology 7, 601-608. Castle, L. D. (1989). "The regeneration of Allocasuarina luehmannii (R.T. Bak.) L. Johnson (Buloke) at Wyperfeld National Park". B. Sc. (Hons) Thesis. Monash University. Castro, J. (1999). Seed mass versus seedling performance in Scots pine: a maternally dependent trait. New Phytologist 144, 153-161. Castro, J., Zamora, R., Hodar, J. A., and Gomez, J. M. (2002). Use of shrubs as nurse plants: a new technique for reforestation in Mediterranean mountains. Restoration Ecology 10, 297-305. Cengel, B., Velioglu, E., Tolun, A. A., and Kaya, Z. (2000). Pattern and magnitude of genetic diversity in Pinus nigra Arnold subspecies pallasiana populations from Kazdagi: implications for in situ conservation. Silvae genetica 49, 249-256. Cervera, M. T., Remington, D., Frigerio, J. M., Storme, V., Ivens, B., Boerjan, W., and Plomion, C. (2000). Improved AFLP analysis of tree species. Canadian Journal of Forest Research 30, 1608-1616. Chacón, P., and Bustamante, R. O. (2001). The effects of seed size and pericarp on seedling recruitment and biomass in Cryptocarya alba (Lauraceae) under two contrasting moisture regimes. Plant Ecology 152, 137-144.

- 192 -

Chalmers, K. J., Waugh, R., Sprent, J. I., Simons, A. J., and Powell, W. (1992). Detection of genetic variation between and within populations of Gliricidium sepium and G. maculata. Heredity 69, 465-472. Chandler, L. G. (1938). In the land of pine and belar. Victorian Naturalist 55, 115-119. Chazdon, R. L. (1991). Effects of leaf and ramet removal on growth and reproduction of Geonoma congesta, a clonal understorey palm. Journal of Ecology 79, 1137-1146. Cheal, D. C. (1986). A park with a kangaroo problem. Oryx 20, 95-99. Cheal, D. C. (1993). Effects of stock grazing on the plants of semi-arid woodlands and grasslands. Proceedings of the Royal Society of Victoria 105, 57-95. Cheal, D. C. (2005a). Changes in vegetation condition at Hattah Kulkyne National Park. A database scoping study Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment: Melbourne. Cheal, D. C. (2005b). Goat damage in mallee in Murray-Sunset National Park. Victorian Naturalist 120, 230-233. Cheal, D. C., and Lucas, A. (2005). DRAFT Recovery Plan for Buloke woodlands of the Riverina and Murray-Darling depression bioregions 2005-2009 Arthur Rylah Institute for Environmental Research: Heildelberg, Melbourne. Cheal, D. C., Parkes, D., Parsons, R. F., and Sluiter, I. R. K. (1992). "Vascular plants and communities." (Australian National Parks and Wildlife Service and the Murray-Darling Basin Commision: Canberra.) Chesson, P., Gebauer, R. L., Schwinning, S., Huntly, N., Wiegand, K., Ernest, M. S., Sher, A., Novoplansky, A., and Weltzin, J. F. (2004). Resource pulses, species interactions, and diversity maintenance in arid and semi-arid environments. Oecologia 141, 236-253. Chesterfield, C. J., and Parsons, R. F. (1985). Regeneration of three tree species in arid south- eastern Australia. Australian Journal of Botany 33, 715-732. Chiew, F. H. S., and McMahon, T. A. (1992). An Australian comparison of Penman potential evapotranspiration estimates and Class-a evaporation-pan data. Australian Journal of Soil Research 30, 101-112. Choi, Y. D. (2004). Theories for ecological restoration in changing environment: toward ‘futuristic’ restoration. Ecological Research 19, 75-81. Clark-Tapio, R., Alfonso-Corrado, C., Eguiarte, L. E., and Molina-Freaner, F. (2005). Clonal diversity and distribution in Stenocereus eruca (Cactaceae), a narrow endemic cactus of the Sonoran Desert. American Journal of Botany 92, 272-278. Clemens, J., Campbell, L. C., and Nurisjah, S. (1983). Germination, growth and mineral ion concentrations of Casuarina species under saline conditions. Australian Journal of Botany 31, 1-9. Clements, F. E. (1916). "Plant succession." (Carnegie Institute: Washington, USA.) Clout, M. N. (1989). Foraging behaviour of the glossy black cockatoos. Australian Wildlife Research 16, 467-473. Cochrane, G. R., and McDonald, N. H. E. (1966). A regeneration study in the Victorian mallee. Victorian Naturalist 83, 220-226. Cohn, J. S., and Bradstock, R. A. (2000). Factors affecting post-fire seedling establishment of selected mallee understorey species. Australian Journal of Botany 48, 59-70. CONCOM (1985). Code of practice for the humane shooting of kangaroos Australian National Parks and Wildlife Service: Canberra. Connell, J. H., and Slatyer, R. O. (1977). Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist 111, 1119-1144. Cooke, B. D. (1987). The effects of rabbit grazing on the regeneration of she oaks, Allocasuarina verticillata and saltwater ti-trees, Melaleuca halmaturorum, in the Coorong National Park, South Australia. Australian Journal of Ecology 13, 11-20. Cooke, J. (2003). "Ecology of Carrichtera annua (L.) DC. in semi-arid Australia". B.Sc. (Hons.) Thesis. School of Botany and Zoology, Australian National University. Cosby, B. J., Hornberger, G. M., Clapp, R. B., and Ginn, T. R. (1984). A statistical exploration of the relationships of soil moisture characteristics to the physical properties of soils. Water Resources Research 20, 682-690.

- 193 -

Coulson, G. M. (1988). The history of kangaroo populations in Hattah-Kulkyne National Park, north-western Victoria. Victorian Naturalist 105, 68-73. Coulson, G. M. (2001). Overabundant kangaroo populations in southeastern Australia. In "Wildlife, land and people: priorities for the 21st century. Proceedings of the second international wildlife management congress" Betheseda, Maryland, USA, (Ed. R. Field, R. J. Warren, H. Okarma and P. R. Sievert) pp. 238-242 (The Wildlife Society.). Coulson, G. M., and Norbury, G. L. (1988). Ecology and management of western grey kangaroos (Macropus fuliginosus), at Hattah-Kulkyne National Park.72. Arther Rylah Institute for Environmental Research: Melbourne. August 1988. Coulson, G. M., Norbury, G. L., and Walters, B. (1990). Forage biomass and kangaroo populations (Marsupialia: Macropodidae) in summer and autumn at Hattah-Kulkyne National Park, Victoria. Australian Mammalogy 1, 219-221. Crisp, M. D. (1978). Demography and survival under grazing of three Australian semi-desert shrubs. Oikos 30, 520-528. Crisp, M. D., and Lange, R. T. (1976). Age structure, distribution and survival under grazing of the arid-zone shrub Acacia burkittii. Oikos 27, 86-92. Critchley, W., and Siegert, K. (1991). A manual for the design and construction of water harvesting schemes for plant production Food and Agriculture Organization of the United Nations: Rome. Crowley, G. M., and Garnett, S. T. (2001). Food value and tree selection by glossy black cockatoos Calyptorhyncus lathami. Austral Ecology 26, 116-126. Cruz, A., Pérez, B., and Moreno, J. M. (2003a). Plant stored reserves do not drive resprouting of the lignotuberous shrub Erica australis. New Phytologist 157, 251-262. Cruz, A., Pérez, B., and Moreno, J. M. (2003b). Resprouting of the Mediterranean-type shrub Erica australis with modified lignotuber carbohydrate content. Journal of Ecology 91, 348-356. Cunningham, G. M., Mulham, W. E., Milthorpe, P. L., and Leigh, J. H. (1992). "Plants of western New South Wales." (New South Wales Government Printing Office and Soil Conservation Service New South Wales: Sydney.) Cuthbertson, E. G. (1997). Chondrilla juncea in Australia. 3. Seed maturity and other factors affecting germination and establishment. Australian Journal of Experimental Agriculture 10, 62-66. CyberTracker Conservation (2004). CyberTracker. http://www.cybertracker.co.za Davis, M. A., and Slobodkin, L. B. (2004). The science and values of restoration ecology. Restoration Ecology 12, 1-3. Davis, M. A., Wrage, K. J., and Reich, P. B. (1998). Competition between tree seedlings and herbaceous vegetation: support for a theory of resource supply and demand. Journal of Ecology 86, 652-661. Davis, M. A., Wrage, K. J., Reich, P. B., Tjoelker, M. G., Schaeffer, T., and Muermann, C. (1999). Survival, growth, and photosynthesis of tree seedlings competing with herbaceous vegetation along a water-light-nitrogen gradient. Plant Ecology 145, 341-350. Dawson, T. J. (1989). Diets of macropodid marsupials: general patterns and environmental influences. In "Kangaroos, wallabies and rat-kangaroos". (Ed. G. Grigg, P. Jarman and I. Hume) Volume 1: pp. 129-142 (Surrey Beatty & Sons Pty Limited: New South Wales). DCE (1990). Restoring the balance. A kangaroo control program for Hattah-Kulkyne National Park Department of Conservation and Environment: Melbourne. De Byle, N. V. (1964). Detection of functional intraclonal aspen root connections by tracers and excavation. Forest Science 34, 19-40. De Kroon, H., Fransen, B., van Rheene, J. W. A., van Dijk, A., and Kreulen, R. (1996). High levels of inter-ramet water translocation in two rhizomatous Carex species, as quantified by deuterium labelling. Oecologia 106, 73-84. De Steven, D. (1989). Genet and ramet demography of Oenocarpus mapora spp. mapora, a clonal palm of Panamanian tropical moist forest. Journal of Ecology 77, 579-596. De'ath, G., and Fabricius, K. E. (2000). Classification and regression trees: A powerful yet simple technique for ecological data analysis. Ecology 81, 3178-3192.

- 194 -

Del Tredici, P. (2001). Sprouting in temperate trees: A morphological and ecological review. The Botanical Review 67, 121-140. Denham, A. J., and Auld, T. D. (2004). Survival and recruitment of seedlings and suckers of trees and shrubs of the Australian arid zone following habitat management and the outbreak of Rabbit Calicivirus Disease (RCD). Austral Ecology 29, 585-599. Des Rochers, A., and Lieffers, V. J. (2001). Root biomass of regenerating aspen (Populus tremuloides) stands of different densities in Alberta. Canadian Journal of Forest Research 31, 1012-1018. Dilcher, D. L., Christophel, D. C., Bhagwandin, H. O., and Scriven, L. J. (1990). Evolution of the Casuarinaceae: morphological comparisons with some extant species. American Journal of Botany 77, 338-355. DNRE (1996a). East Gippsland regional comprehensive assessment: national estate report Commonwealth of Australia and Natural Resources & Environment: DNRE (1996b). Mallee parks management plan National Parks Service, Department of Natural Resources and Environment: East Melbourne. Doran, J. C., and Turnbull, J. W. (1997). Australian trees and shrubs: species for land rehabilitation and farm planting in the tropics Australian Centre for International Agricultural Research, ACIAR: Monographs MN024: Canberra. Dorken, M. E., and Eckert, C. G. (2001). Severely reduced sexual reproduction in northern populations of a clonal plant, Decodon verticillatus (Lythraceae). Journal of Ecology 89, 339-350. Dorken, M. E., Friedman, J. E., and Barrett, S. C. H. (2002). The evolution and maintenance of monoecy and dioecy in Sagittaria latifolia. Evolution 56, 31-41. Douhovnikoff, V., Cheng, A. M., and Dodd, R. S. (2004). Incidence, size and spatial structure of clones in second-growth stands of coast redwood, Sequoia sempervirens (Cupressaceae). Ecology 91, 1140-1146. Doyle, J. J., and Doyle, J. L. (1990). Isolation of plant DNA from fresh tissue. Focus 12, 13-15. DSE (2004). Vegetation Quality Assessment Manual–Guidelines for applying the habitat hectares scoring method. Version 1.3 Victorian Government, Department of Sustainability and Environment: Melbourne. Dyksterhuis, E. J. (1949). Condition and management of rangeland based on quantitative ecology. Journal of Range Management 2, 104-115. Earl, G. E. (1984). Vegetation in the Mournpall Block, June 1984 National Parks Service: June 1984. Eckert, C. G. (1999). Clonal plant research: Proliferation, integration but not much evolution. American Journal of Botany 86, 1649-1654. Eckert, C. G. (2000). Contributions of autogamy and geitonogamy to self-fertilisation in a mass- flowering, clonal plant. Ecology 81, 532-542. Eckert, C. G., and Barrett, S. C. H. (1993). Clonal reproduction and patterns of genotypic diversity in Decodon verticillatus (Lythraceae). American Journal of Botany 80, 1175- 1182. Eckert, C. G., Lui, K., Bronson, K., Corradini, P., and Bruneau, A. (2003). Population genetic consequences of extreme variation in sexual and clonal reproduction in an aquatic plant. Molecular Ecology 12, 331-344. Edwards, F., Bainbridge, D. A., Zink, T. A., and Allen, M. F. (2000). Rainfall catchments improve survival of container transplants at Mojave Desert site. Restoration Ecology 18, 100-103. Eldridge, D. J., and Rath, D. (2002). Hip holes: kangaroo (Macropus spp.) resting sites modify the physical and chemical environment of woodland soils. Austral Ecology 27, 527-536. Eldridge, D. J., and Robson, A. D. (1997). Bladeploughing and exclosure influence soil properties in a semi-arid Australian woodland. Journal of Range Management 50, 191-198. Eldridge, D. J., and Simpson, R. (2002). Rabbit (Oryctolagus cuniculus L.) impacts on vegetation and soils, and implications for management of wooded rangelands. Basic and Applied Ecology 3, 19-29.

- 195 -

Ellstrand, N. C., and Roose, M. L. (1987). Patterns of genotypic diversity in clonal plant species. American Journal of Botany 74, 123-131. Entwisle, T. J. (1989). Casuarinaceae. In "Flora of Australia". (Ed. A. S. George) Volume 3 Hamamelidales to Casuarinales: pp. (Bureau of Flora and Fauna, Australian Government Publishing Services: Canberra). Erakovic, L. J., Subic, A., and Tu, J. (2004). CFD modelling of slasher with single and double counter rotating blades using sliding mesh technique. In "Computational Techniques and Applications Conference" Melbourne, (Ed. pp.). Eriksson, O. (1989). Seedling dynamics and life histories in clonal plants. Oikos 55, 231-238. Eriksson, O. (1992). Evolution of seed dispersal and recruitment in clonal plants. Oikos 63, 439- 448. Eriksson, O., and Bremer, B. (1993). Genet dynamics of the clonal plant Rubus saxatilis. Journal of Ecology 81, 533-542. Eriksson, O., and Ehrlén, J. (1992). Seed and microsite limitation of recruitment in plant populations. Oecologia 91, 360-364. Eriksson, O., and Froborg, H. (1996). "Windows of opportunity" for recruitment in long lived clonal plants - experimental studies of seedling establishment in Vaccinium shrubs. Canadian Journal of Botany 74, 1369-1374. Esau, K. (1977). "Anatomy of seed plants." (John Wiley & Sons: New York.) Escaravage, N., Questiau, S., Pornon, A., Doche, B., and Taberlet, P. (1998). Clonal diversity in a Rhododendron ferrugineum L. (Ericaceae) population inferred from AFLP markers. Molecular Ecology 7, 975-982. Excoffier, L., Smouse, P. E., and Quattro, J. M. (1992). Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131, 479-491. Facelli, J. M., and Brock, D. J. (2000). Patch dynamics in arid lands: localized effects of Acacia papycarpa on soils and vegetation of open woodlands of South Australia. Ecography 23, 479-491. Falk, D., Palmer, M., and Zedler, J. (2005). "Foundations of Restoration Ecology." (Island Press: Covelo, CA.) Fernandez-Gimenez, M. E., and Allen-Diaz, B. (1999). Testing a non-equilibrium model of rangeland vegetation dynamics in Mongolia. Journal of Applied Ecology 36, 871-885. Florentine, S. K., and Westbrooke, M. E. (2005). Invasion of the noxious weed Nicotiana glauca R. Graham after an episodic flooding event in the arid zone of Australia. Journal of Arid Environments 60, 531-545. Foran, B. D. (1986). The impact of rabbits and cattle on an arid calcareous shrubby grassland in central Australia. Vegetatio 66, 49-59. Foster, D. R. (2000). From bobolinks to bears: interjecting geographical history into ecological studies, environmental interpretation, and conservation planning. Journal of Biogeography 27, 27-30. Fraser, E. C., Lieffers, V. J., and Landhausser, S. M. (2004). Wounding of aspen roots promotes suckering. Canadian Journal of Botany 82, 310-315. Fraser, E. C., Lieffers, V. J., Landhausser, S. M., and Frey, B. R. (2002). Soil nutrition and temperature as drivers of root suckering in trembling aspen. Canadian Journal of Forest Research 32, 1685-1691. Freeman, D. C., Kilikoff, L. C., and Harper, K. T. (1976). Differential resource utilization by the sexes of dioecious plants. Science 193, 597-599. Freudenberger, D. O., Hodgkinson, K. C., and Noble, J. C. (1997). Causes and consequences of landscape dysfunction in rangelands. In "Landscape ecology, function and management". (Ed. J. A. Ludwig, D. J. Tongway, D. O. Freudenberger, J. Noble and K. C. Hodgkinson) pp. 63-92 (CSIRO Publishing: Collingwood). Frey, B. R., Lieffers, V. J., Landhausser, S. M., Comeau, P. G., and Greenway, K. J. (2003). An analysis of sucker regeneration of trembling aspen. Canadian Journal of Forest Research 33, 1169-1179.

- 196 -

Friedel, M. H. (1991). Range condition assessment and the concept of thresholds: a viewpoint. Journal of Range Management 44, 422-426. Friedel, M. H., Pickup, G., and Nelson, D. J. (1993). The interpretation of vegetation change in a spatially and temporally diverse arid Australian landscape. Journal of Arid Environments 24, 241-260. Friedel, M. H., Sparrow, A. D., Kinloch, J. E., and Tongway, D. J. (2003). Degradation and recovery processes in arid grazing lands of central Australia. 2: Vegetation. Journal of Arid Environments 55, 327-348. Galloway, R. W. (1983). Natural and anthropogenic erosion and their relative impact in the arid zone. In "What future for Australia's arid lands?" (Ed. J. Messer and G. Mosley) pp. 45-46 (Australian Conservation Foundation: Melbourne). Gao, Q., and Reynolds, J. F. (2003). Historical shrubgrass transitions in the northern Chihuahuan Desert: modeling the effects of shifting rainfall seasonality and event size over a landscape gradient. Global Change Biology 9, 1475-1493. Gardiner, H. G. (1986). Dynamics of perennial plants in the mulga (Acacia aneura F. Muell.) zone of Western Australia. II: Survival of perennial shrubs and grasses. Australian Rangeland Journal 8, 28-35. Gaudeul, M., Taberlet, P., and Till-Bottraud, I. (2000). Genetic diversity in an endangered alpine plant, Eryngium alpinum L. (Apiaceae), inferred from amplified fragment length polymorphism markers. Molecular Ecology 9, 1625-1637. George, M. R., Brown, J. R., and Clawson, W. J. (1992). Application of non-equilibrium ecology to management of Mediterranean grassland. Journal of Range Management 5, 436-440. Gillies, A. C. M., Cornelius, J. P., Newton, A. C., Navarro, C., Hernandez, M., and Wilson, J. (1997). Genetic variation in Costa Rican populations of the tropical timber species Cedrela odorata L., assessed using RAPDs. Molecular Ecology 6, 1133-1145. Gillies, A. C. M., Navarro, C., Lowe, J. L., Newton, A. C., Hernandez, M., Wilson, J., and Cornelius, J. P. (1999). Genetic diversity in Mesoamerican populations of mahogany (Swietenia macrophylla), assessed using RAPDs. Heredity 83, 722-732. Glendinning, D. (1989). Some aspects of autotetraploid population dynamics. Theoretical and Applied Genetics 78, 233-242. Golluscio, R. A., Sala, O. E., and Lauenroth, W. K. (1998). Differential use of large summer rainfall events by shrubs and grasses: a manipulative experiment in the Patagonian steppe. Oecologia 115, 17-25. Gomez, J. M. (2004). Bigger is not always better: conflicting selective pressures on seed size in Quercus ilex. Evolution 58, 71-80. Gomez-Aparicio, L., Zamora, R., Gomez, J. M., Hodar, J. A., Castro, J., and Baraza, E. (2004). Applying plant facilitation to forest restoration: a meta-analysis of the use of shrubs as nurse plants. Ecological Applications 14, 1128–1138. Gordan, D. R., and Rice, K. J. (2000). Competitive suppression of Quercus douglasii (Fagaceae) seedling emergence and growth. American Journal of Botany 87, 986-994. Gowans, S., and Westbrooke, M. E. (2002). Hattah-Kulkyne National Park Vegetation Condition Assessment Centre for Environmental Management, University of Ballarat: Ballarat. Grant, H. (2004). "The Australian Seedsman." (Dubbo Printing Works: Dubbo, New South Wales.) Grazywacz, A. P., and Rosochacka, J. (1980). The colour of Pinus sylvestris L. seed and their susceptibility to damping-off. I. The colour and quality of seeds and fatty acids content of the seed coat. European Journal of Forest Pathology 10, 138-144. Greene, R. S. B., Nettleton, W. D., Chartres, C. J., Leys, J. F., and Cunningham, R. B. (1998). Runoff and micromorphological properties of a grazed haplargid, near Cobar, NSW, Australia. Australian Journal of Soil Research 36, 87-108. Grice, A. C., and Barchia, I. (1992). Does grazing reduce survival of indigenous perennial grasses of the semi-arid woodlands of western New South Wales? Australian Journal of Ecology 17, 195-205.

- 197 -

Gugerli, F., Eichenberger, K., and Schneller, J. J. (1999). Promiscuity in populations of the cushion plant Saxifraga oppositifolia in the Swiss Alps as inferred from random amplified polymorphic DNA (RAPD). Molecular Ecology 8, 453-461. Gustafson, D. J., Gibson, D. J., and Nickrent, D. L. (1999). Random amplified polymorphic DNA variation among remnant big bluestem (Andropogon gerardii Vitman) populations from Arkansas' Grand Prairie. Molecular Ecology 8, 1693-1701. Gustafson, D. J., Gibson, D. J., and Nickrent, D. L. (2002). Genetic diversity and competitive abilities of Dalea purpurea (Fabaceae) from remnant and restored grasslands. International Journal of Plant Science 163, 979-990. Gustafson, D. J., Gibson, D. J., and Nickrent, D. L. (2004). Conservation genetics of two co- dominant grass species in an endangered grassland ecosystem. Journal of Applied Ecology 41, 389-397. Haig, D., and Westoby, M. (1988). On limits to seed production. American Naturalist 131, 757- 759. Hall, E. A. A., Specht, R. l., and Eardley, C. M. (1964). Regeneration of the vegetation on Koonamore Vegetation Reserve, 1926-1962. Australian Journal of Botany 12, 205-264. Hamrick, J. L., and Godt, M. W. (1990). Allozyme diversity in plant species. In "Plant population genetics, breeding and genetic resources". (Ed. A. H. D. Brown, M. T. Clegg, A. L. Kahler and B. S. Weir) pp. 43-63 (Sinauer: Sunderland, MA). Hamrick, J. L., and Godt, M. W. (1996). Conservation genetics of endemic plant species. In "Conservation genetics: case histories from nature". (Ed. J. C. Avise and J. L. Hamrick) pp. 281-304 (Chapman & Hall: New York). Hamrick, J. L., Godt, M. W., and Sherman-Broyles, S. L. (1992). Factors influencing levels of genetic diversity in woody plant species. New Forests 5, 95-124. Handel, S. N. (1985). The intrusion of clonal growth patterns on plant breeding systems. American Naturalist 125, 367-384. Harada, Y., Kawano, S., and Iwasa, Y. (1997). Probability of clonal identity: inferring the relative success of sexual vs. clonal reproduction from spatial genetic patterns. Journal of Ecology 85, 591-600. Harper, J. L. (1977). "Population biology of plants." (Academic Press: New York.) Harrington, G. N. (1979). The effects of feral goats and sheep on the shrub populations in a semi- arid woodland. Australian Rangeland Journal 1, 334-345. Harrington, G. N. (1991). Effects of soil moisture on shrub seedling survival in a semi-arid grassland. Ecology 72, 1138-1149. Hartl, D. L., and Clark, A. G. (1997). "Principles of population genetics." (Sinauer Associates: Sunderland, MA.) Hayes, B. (1993). "Factors effecting the regeneration of woody perennial species within Hattah- Kulkyne National Park". B.Sc.(Hons.) Thesis. Department of Zoology, University of Melbourne. Helenurm, K., and Schaal, B. A. (1996). Genetic and maternal effects of offspring fitness in Lupinus texensis (Fabaceae). American Journal of Botany 83, 1596-1608. Herrera, C. M. (2000). Individual differences in progeny viability in Lavandula latifolia: a long- term field study. Ecology 81, 3036-3047. Herrera, C. M. (2002). Topsoil properties and seedling recruitment in Lavandula latifolia: stage- dependence and spatial decoupling of influential parameters. Oikos 97, 260-270. Hessl, A. E., and Graumlich, L. J. (2002). Interactive effects of human activities, herbivory and fire on quaking aspen (Populus tremuloides) age structures in western Wyoming. Journal of Biogeography 29, 889-902. Hill, M. (2002). RangeASSESS: A spatial framework for analysis of potential for carbon sequestration in the rangelands Cooperative Research Centre for Greenhouse Accounting: Canberra. Hill, W. G., and Weir, B. S. (2004). Moment estimation of population diversity and genetic distance from data on recessive markers. Molecular Ecology 13, 895-908.

- 198 -

Ho, K. Y., Ou, C. H., Yang, J. C., and Hsiao, J. Y. (2002). An assessment of DNA polymorphisms and genetic relationships of Casuarina equisetifolia using RAPD markers. Botanical Bulletin of Academia Sinica 43, 93-98. Hobbs, R. J., Cramer, V. A., and Krisjanson, L. J. (2003). What happens if we cannot fix it? Triage, palliative care and setting priorities in salinising landscapes. Australian Journal of Botany 51, 647-653. Hobbs, R. J., and Harris, J. A. (2001). Restoration ecology: repairing the earth's ecosystems in the new millenium. Restoration Ecology 9, 239-246. Hobbs, R. J., and Norton, D. A. (1996). Towards a conceptual framework for restoration ecology. Restoration Ecology 4, 93-110. Hobbs, R. J., Sparrow, A. D., and Landsberg, J. (1994). A model of soil moisture balance and herbage growth in the arid rangelands of central Australia. Journal of Arid Environments 28, 281-298. Hodgkinson, K. C., and Freudenberger, D. O. (1997). Production pulses and flow-ons in rangeland landscapes. In "Landscape ecology, function and management". (Ed. J. A. Ludwig, D. J. Tongway, D. O. Freudenberger, J. Noble and K. C. Hodgkinson) pp. 13-24 (CSIRO Publishing: Collingwood). Hodgkinson, K. C., and Harrington, G. N. (1985). The case for prescribed burning to control shrubs in eastern semi-arid woodlands. Australian Rangeland Journal 7, 64-74. Hogbin, P. M., and Peakall, R. (1999). Evaluation of the contribution of genetic research to the management of the endangered plant Zieria prostrata. Conservation Biology 13, 514-522. Hogbin, P. M., Peakall, R., and Sydes, M. A. (2000). Achieving practical outcomes from genetic studies of rare Australian plants. Australian Journal of Botany 48, 375-382. Holderegger, R., Stehlik, I., and Schneller, J. J. (1998). Estimation of the relative importance of sexual and vegetative reproduction in the clonal woodland herb Anemone nemorosa. Oecologia 117, 105-107. Holl, K. D., Crone, E. E., and Schultz, C. B. (2003). Landscape restoration: moving from generalities to methodologies. Bioscience 53, 491-502. Holm, A. M., Watson, I. W., Loneragan, W. A., and Adams, M. A. (2003). Loss of patch-scale heterogeneity on primary productivity and rainfall-use efficiency in Western Australia. Basic and Applied Ecology 4, 569-578. Holman, J. E., and Playford, J. (2000). Molecular and morphological variation in the Senna artemisioides complex. Australian Journal of Botany 48, 569-579. Holmgren, M., and Scheffer, M. (2001). El Nino as a window of opportunity for the restoration of degraded arid ecosystems. Ecosystems 4, 151-159. Holmgren, M., Scheffer, M., and Huston, M. A. (1997). The interplay of facilitation and competition in plant communities. Ecology 78, 1966-1975. Holsinger, K. E., Lewis, P. O., and Dey, D. K. (2002). A Bayesian approach to inferring population structure from dominant markers. Molecular Ecology 11, 1157-1164. Holsinger, K. E., and Vitt, P. (1997). The future of conservation biology: what's a geneticist to do? In "The ecological basis of conservation: heterogeneity, ecosystems and biodiversity". (Ed. S. T. A. Pickett, R. S. Ostfeld, M. Shachak and G. E. Likens) pp. 202- 216 (Chapman & Hall: New York). Hruska, J., Cermak, J., and Sustek, S. (1999). Mapping tree root systems with ground penetrating radar. Tree Physiology 19, 125-130. Huff, D. R., Peakall, R., and Smouse, P. E. (1993). RAPD variation within and among natural populations of outcrossing buffalo grass (Buchloe dactyloides (Nutt.) Engelm.). Theoretical and Applied Genetics 86, 927-934. Huntsinger, L., and Bartolome, J. W. (1992). Ecological dynamics of Quercus dominated woodlands in California and southern Spain: a state and transition model. Vegetatio 99, 299-305. Huxman, T. E., Wilcox, B. P., Breshears, D. D., Scott, R. L., Snyder, K. A., Small, E. E., Hultine, K., Pockman, W. T., and Jackson, R. B. (2005). Ecohydrological implications of woody plant encroachment. Ecology 86, 308-319.

- 199 -

Hwang, R. H. (1991). Seedlings of Australian Casuarinas: I. Germination. Western Australian Naturalist 18, 188-197. Hwang, R. H., and Conran, J. G. (2000). Seedling characteristics in the Casuarinaceae. Telopea 8, 429-439. Hwang, S., Lin, H., Kuo, Y., and Lin, T. (2001). RAPD variation in relation to population differentiation of Chamaecyparis formosensis and Chamaecyparis taiwanensis. Botanical Bulletin of Academia Sinica 42, 173-179. Hwang, Y. H. (1992). The Casuarinaceae: Allocasuarina is unsupported. Australian Systematic Botany Newsletter 70, 16-19. Iglesias, R. M. R., and Kothmann, M. M. (1997). Structure and causes of vegetation change in state and transition model applications. Journal of Range Management 50, 399-408. Illius, A. W., and O’Connor, T. G. (1999). On the relevance of nonequilibrium concepts to arid and semiarid grazing systems. Ecological Applications 9, 798-813. Ireland, C. (1997). "Sustaining the western myall woodlands : ecology and management". PhD Thesis. Department of Environmental Science and Rangeland Management, University of Adelaide. Isabel, N., Beaulieu, J., and Bousquet, J. (1995). Complete congruence between gene diversity estimates derived from genotypic data at enzyme and random amplified polymorphic DNA loci in black spruce. Proceedings of the National Academy of Sciences of the USA 92, 6369-6373. IUCN (1999). Biological diversity of dryland, mediterranean, arid, semi-arid savannah and grassland ecosystems. In "Fourth meeting of the subsidiary body on scientific, technical and technological advice" Montreal, Canada, (Ed. I. Thiaw and M. Chouchena-Rojas) pp.). Jackson, R. D., and Bartolome, J. W. (2002). A state-transition approach to understanding nonequilibrium plant community dynamics of California grasslands. Plant Ecology 162, 49-65. Jelinski, D. E., and Cheliak, W. M. (1992). Genetic diversity and spatial subdivision of Populus tremuloides (Salicaceae) in an heterogeneous landscape. American Journal of Botany 79, 728-736. Jeltsch, F., Stephan, T., Wiegand, T., and Gerhard, E. W. (2001). Arid rangeland management supported by dynamic spatially explicit simulation models. In "Sustainable land use in deserts". (Ed. S. W. Breckle, M. Veste and W. Wucherer) pp. 229-240 (Springer Verlag: New York). Jenik, J. (1994). Clonal growth in woody plants: a review. Folia Geobotanica et Phytotaxonomica 29, 291-306. Jensen, M. E., Burman, R. D., and Allen, R. G. (1990). Evapotranspiration and irrigation water requirements American Society of Civil Engineers: New York. Johnson, G. (1989). Hattah Kulkyne National Park Report on the revegetation program (1987- 1989) Conservation, Forests and Lands: Mildura. August 1989. Johnson, L. A. S., and Wilson, K. L. (1989). Casuarinaceae: a synopsis. In "Evolution, systematics, and fossil history of the Hamamelidae: higher Hamamelidae". (Ed. P. R. Crane and S. Blackmore) Volume 2: pp. 167-188 (Clarendon Press: Oxford). Johnson, L. A. S., and Wilson, K. L. (1993). Casuarinaceae. In "The families and genera of vascular plants". (Ed. K. Kubitzki, J. G. Rohwer and V. Bittrich) Volume 2: Flowering plants - Dicotyledons. Magnoliid, Hamamelid and Caryophyllid Families: pp. 237-242 (Springer-Verlag: Berlin). Jones, B., and Gliddon, C. (1999). Reproductive biology and genetic structure in Lloydia serotina. Plant Ecology 141, 151-161. Jones, C., Edwards, K. J., and Castaglione, S. (1997a). Reproducability testing of RAPD, AFLP and SSR markers in plants by a network of European laboratories. Molecular Breeding 3, 381-390. Jones, J. (1942). The Kulkyne National Forest. Emu 42, 1-12.

- 200 -

Jones, R. H., Allen, B. P., and Sharitz, R. R. (1997b). Why do early-emerging tree seedlings have survival advantages? A test using Acer rubrum (Aceraceae). American Journal of Botany 84, 1714-1718. Jones, R. H., and Raynal, D. J. (1986). Spatial distribution and development of root sprouts in Fagus grandifolia (Fagaceae). American Journal of Botany 73, 1723-1731. Jones, R. H., and Raynal, D. J. (1987). Root sprouting in American beech: production, survival and the effects of parent tree vigour. Canadian Journal of Forest Research 17, 539-544. Jones, R. H., and Raynal, D. J. (1988). Root sprouting in American beech (Fagus grandifolia): effects of root injury, root exposure and season. Forest Ecology and Management 25, 79- 90. Jordano, P., and Godoy, J. A. (2000). RAPD variation and population genetic structure in Prunus mahaleb (Rosaceae), an animal-dispersed tree. Molecular Ecology 9, 1293-1305. Keig, G., and McAlpine, J. R. (1974). WATBAL: A computer system for the estimation and analysis of soil moisture regimes from simple climatic data CSIRO Division of Water Research: Canberra. Kelly, M. G., and Levin, D. A. (1997). Fitness consequences and heritability aspects of emergence date in Phlox drummondii. Journal of Ecology 85, 755-766. Kemp, P. R., Reynolds, J. F., Pachepsky, Y. A., and Chen, J. L. (1997). A comparative modelling study of soil water dynamics in a desert ecosystem. Water Resources Research 33, 73-90. Kenkel, N. C., and Podani, J. (1991). Plot size and estimation efficiency in plant community studies. Journal of Vegetation Science 2, 539-544. Kennington, W. J., and James, S. H. (1997). The effect of small population size on the mating system of a rare clonal mallee, Eucalyptus argutifolia (Myrtaceae). Heredity 78, 252-260. Kent, M., and Coker, P. (1992). "Vegetation description and analysis: a practical approach." (John Wiley & Sons: Brisbane.) Kenyon, A. S. (1906). The mallee frontage of the Murray River - An undeveloped province of Victoria. Journal of Agriculture 4, 257-266. Kenyon, A. S. (1914a). The story of the mallee I: Discovery and exploration. Victorian Historical Magazine 4, 57-74. Kenyon, A. S. (1914b). The story of the mallee: Introduction. Victorian Historical Magazine 4, 23-56. Kenyon, A. S. (1915a). The story of the mallee II: Settlement - the squatters. Victorian Historical Magazine 4, 121-152. Kenyon, A. S. (1915b). The story of the mallee III: The agricultural occupation. Victorian Historical Magazine 4, 175-200. Kitamura, K., Takasu, H., Hayashi, K., Ohara, M., Ohkawa, T., Utech, F. H., and Kowano, S. (2000). Demographic genetic analysis of the American beech (Fagus grandifolia Ehrh.) I. Genetic substructurings of northern populations with root suckers in Quebec and Pennsylvania. Plant Species Biology 15, 43-58. Kjølner, S., Såstad, S. M., Taberlet, P., and Brochmann, C. (2004). Amplified fragment length polymorphism versus random amplified polymorphic DNA markers: clonal diversity in Saxifraga cernua. Molecular Ecology 13, 81-86. Klimes, L., and Kilimesova, J. (1999). Root sprouting in Rumex acetosella under different nutrient levels. Plant Ecology 141, 33-39. Klimes, L., Klimesova, J., Hendriks, R. J. J., and van Groenendael, J. (1997). Clonal plant architecture: A comparative analysis of form and function. In "The ecology and evolution of clonal plants". (Ed. H. de Kroon and J. van Groenendael) pp. 1-29 (Backhuys Publishers: Leiden, The Netherlands). Knapp, E. E., Goedde, M. A., and Rice, K. J. (2001). Pollen-limited reproduction in blue oak: implications for wind pollination in fragmented populations. Oecologia 128, 48-55. Kodela, P. G. (2001). Acacia atrox (Fabaceae: Mimosoideae), a new rare species from the North Western Slopes, New South Wales. Telopea 9, 415-419. Koop, H. (1987). Vegetative reproduction of trees in some European natural forests. Vegetatio 72, 103-110.

- 201 -

Kormanik, P. P., and Brown, C. l. (1967). Root buds and the development of root suckers in sweetgum. Forest Science 13, 338-345. Körner, C. (2003). Carbon limitation in trees. Journal of Ecology 91, 4-17. Kozlowski, T. T. (1992). Carbohydrate sources and sinks in woody plants. The Botanical Review 58, 107-222. Krauss, S. L. (2000a). Accurate gene diversity estimates from amplified fragment length polymorphism (AFLP) markers. Molecular Ecology 9, 1241-1246. Krauss, S. L. (2000b). Patterns of mating in Persoonia mollis (Proteaceae) revealed by an analysis of paternity using AFLP: implications for conservation. Australian Journal of Botany 48, 349-356. Krebs, C. J. (1999). "Ecological methodology." (Addison-Wesley Educational Publishers: Sydney.) Kuo, S. R. (1984). Studies on the seed germination and seedling growth of Casuarina National Taiwan University: Taiwan. Kwon, J. A., and Morden, C. W. (2002). Population genetic structure of two rare tree species (Colubrina oppositifolia and Alphitonia ponderosa, Rhamnaceae) from Hawaiian dry and mesic forests using random amplified polymorphic DNA markers. Molecular Ecology 11, 991-1002. Lacerda, D. R., Acedo, M. D. P., Lemos Filho, J. P., and Lovato, M. B. (2001). Genetic diversity and structure of natural populations of Plathymenia reticulata (Mimosoideae), a tropical tree from the Brazilian Cerrado. Molecular Ecology 10, 1143-1152. Lacey, C. J., and Johnston, R. D. (1990). Woody clumps and clumpwoods. Australian Journal of Botany 38, 299-334. Ladd, P. G. (1989). The status of the Casuarinaceae in Australian forests. In "Australia's ever changing forests". (Ed. K. J. Frawley and N. Semple) pp. 63-85 (Department of Geography and Oceanography, Australian Defence Forces Academy: Canberra). Laitinen, M. L., Julkunen-Tiitto, R., Yamaji, K., Heinonen, J., and Rousi, M. (2004). Variation in birch bark secondary chemistry between and within clones: implications for herbivory by hares. Oikos 104, 316. Landergott, U., Holderegger, R., Kozlowski, G., and Schneller, J. J. (2001). Historical bottlenecks decrease genetic diversity in natural populations of Dryopteris cristata. Heredity 87, 344- 355. Landhausser, S. M., and Lieffers, V. J. (2002). Leaf area renewal, root retention and carbohydrate reserves in a clonal tree species following above ground disturbance. Journal of Ecology 90, 658-665. Landsberg, J., and Crowley, G. M. (2004). Monitoring rangeland biodiversity: plants as indicators. Austral Ecology 29, 59-77. Landsberg, J., James, C. D., Maconochie, J. R., Nicholls, A. O., Stol, J., and Tynan, R. (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., James, C. D., Morton, S. R., Muller, W. J., and Stol, J. (2003). Abundance and composition of plant species along grazing gradients in Australian rangelands. Journal of Applied Ecology 40, 1008-1024. Landsberg, J., Lavorel, S., and Stol, J. (1999). Grazing response groups among understorey plants in arid rangelands. Journal of Vegetation Science 10, 693-696. Lange, R. T. (1969). The piosphere, sheep track and dung patterns. Journal of Rangeland Management 22, 396-400. Lange, R. T., and Graham, C. G. (1983). Rabbits and the failure of regeneration in Australian arid zone Acacia. Australian Journal of Ecology 8, 377-381. Lange, R. T., and Purdie, R. (1976). Western myall (Acacia sowdenii) and its survival prospects and management needs. Australian Rangeland Journal 1, 64-69. Lange, R. T., and Willcocks, M. C. (1980). Experiments on the capacity of present sheep flocks to extinguish some tree populations of the South Australian arid zone. Journal of Arid Environments 3, 223-229.

- 202 -

Lauenroth, W. K., Sala, O. E., Coffin, D. P., and Kirchner, T. B. (1994). The importance of soil water in the recruitment of Bouteloua gracilis in the shortgrass steppe. Ecological Applications 4, 741-749. Lawrie, A. C. (1982). Field nodulation in nine species of Casuarina in Victoria. Australian Journal of Botany 30, 447-460. Laycock, W. A. (1991). Stable states and thresholds of range condition on North American rangelands: a viewpoint. Journal of Range Management 44, 427-433. LCC (1977). Final recommendations: Mallee study area Land Conservation Council, Victorian Government Printer: Melbourne. LCC (1987). Mallee Area Review - Study Area Report Land Conservation Council, Victorian Government Printer: Melbourne. LCC (1989). Mallee Area Review Final Recommendations Land Conservation Council, Victorian Government Printer: Melbourne. Leigh, J. H., Wood, D. H., Holgate, M. D., Slee, A., and Stranger, M. G. (1989). Effects of rabbit and kangaroo grazing on two semi-arid grassland communities in central-western New South Wales. Australian Journal of Botany 37, 375-96. Leishman, M. R. (2001). Does the seed size/number trade-off model determine plant community structure? An assessment of the model mechanisms and their generality. Oikos 93, 294- 302. Lesica, P., and Allendorf, F. W. (1999). Ecological genetics and the restoration of plant communities: mix or match? Restoration Ecology 7, 42-55. Loh, W. Y., and Shih, Y. S. (1997). Split selection methods for classification trees. Statistica Sinica 7, 815-840. Lopez, G. A., Potts, B. M., Vaillancourt, R. E., and Apiolaza, L. A. (2003). Maternal and carryover effects on early growth of Eucalyptus globulus. Canadian Journal of Forest Research 33, 2108-2115. Lord, D. (2002). The impact of rabbit grazing on the sucker recruitment of the threatened species purple-wood wattle (Acacia carnei). Range Management Newsletter March, 11-14. Loveless, M. D., and Hamrick, J. L. (1984). Ecological determinants of genetic structure in plant populations. Annual Review of Ecology and Systematics 15, 65-95. Lovich, J., and Bainbridge, D. A. (1999). Anthropogenic degradation of the Southern California desert ecosystem and prospects for natural recovery and restoration--a review. Environmental Management 24, 309-326. Ludwig, J. A. (1987). Primary productivity in arid lands: myths and realities. Journal of Arid Environments 13, 1-7. Ludwig, J. A., and Tongway, D. J. (1995). Spatial organisation of landscapes and its function in semi-arid woodlands, Australia. Landscape Ecology 10, 51-63. Ludwig, J. A., and Tongway, D. J. (1996). Rehabilitation of semiarid landscapes in Australia. 2. Restoring vegetation patches. Restoration Ecology 4, 398-406. Ludwig, J. A., Tongway, D. J., Bastin, G. N., and James, C. D. (2004). Monitoring ecological indicators of rangeland functional integrity and their relationship to biodiversity at local to regional scales. Austral Ecology 29, 108-120. Ludwig, J. A., Tongway, D. J., Freudenberger, D. O., Noble, J. C., and Hodgkinson, K. C. (1997). "Landscape ecology, function and management." (CSIRO Publishing: Collingwood.) Lunt, I. D. (2002). Grazed, burnt and cleared: how ecologists have studied century-scale vegetation changes in Australia. Australian Journal of Botany 50, 391-407. Lynch, M., and Milligan, B. G. (1994). Analysis of population genetic structure with RAPD markers. Molecular Ecology 3, 91-99. Macaulay, L. A., and Westbrooke, M. E. (2003). Guidelines for the management and rehabilitation of the Wimmera Buloke Woodlands of western Victoria University of Ballarat: Ballarat. Macaulay, L. A., and Westbrooke, M. E. (in prep.). Are heavy autumn storms the key to seedling regeneration of Buloke (Allocasuarina luehmannii) in the Wimmera?

- 203 -

Maconochie, J. R. (1982). Regeneration of arid zone plants: a floristic survey. In "Evolution of flora and fauna of arid Australia". (Ed. W. R. Barker and P. J. Greenslade) pp. 141-144 (Peacock Publications: Adelaide). Maestre, F. T., Bautista, S., Cortina, J., and Bellot, J. (2001). Potential for using facilitation by grasses to establish shrubs on a semi-arid degraded steppe. Ecological Applications 11, 1641-1655. Maestre, F. T., Cortina, J., Bautista, S., Bellot, J., and Vallejo, R. (2003). Small-scale environmental heterogeneity and spatiotemporal dynamics of seedling establishment in a semiarid degraded ecosystem. Ecosystems 6, 630-643. Maestre, F. T., Valladares, F., and Reynolds, J. F. (2005). Is the change of plant-plant interactions with abiotic stress predictable? A meta-analysis of field results in arid environments. Journal of Ecology 93, 748-757. Maguire, J. D. (1962). Speed of germination aid in selection and evaluation for seedling emergence and vigour. Crop Science 2, 176-177. Maidment, D. R. (1993). "Handbook of Hydrology." (McGraw Hill: New York.) Mandujano, M. C., Franco, M., and Golubov, J. (2001). Integration of demographic annual variability in a clonal desert cactus. Ecology 82, 344-359. Mandujano, M. C., Montana, C., Mendez, I., and Golubov, J. (1998). The relative contributions of sexual reproduction and clonal propagation in Opuntia rastrera from two habitats in the Chihuahuan Desert. Journal of Ecology 86, 911-921. Marcar, N. E., and Crawford, D. S. (2004). "Trees for saline landscapes." (CSIRO: Melbourne.) Maron, M., and Lill, A. (2004). Discrimination among potential buloke (Allocasuarina luehmannii) feeding trees by the endangered south-eastern red-tailed black-cockatoo (Calyptorhynchus banksii graptogyne). Wildlife Research 31, 311-317. Martin, L. M., Moloney, K. A., and Wilsey, B. J. (2005). An assessment of grassland restoration success using species diversity components. Journal of Applied Ecology 42, 327-336. Massa, A. N., Larson, S. R., Jenson, K. B., and Hole, D. J. (2001). AFLP variation in Bromus Section Ceratochloa germplasm of Patagonia. Crop Science 41, 1609-1616. McAlpine, C. A., Grigg, G. C., Mott, J. J., and Sharma, P. (1999). Influence of landscape structure on kangaroo abundance in a disturbed semi-arid woodland of Queensland. Rangeland Journal 21, 104-134. McCarthy, M. A., Parris, K. M., van der Ree, R., McDonnell, M. J., Burgman, M. A., Williams, N. S. G., McLean, N., Harper, M. J., Meyer, R., Hahs, A., and Coates, T. (2004). The habitat hectares approach to vegetation assessment: an evaluation and suggestions for improvement. Ecological Management and Restoration 5, 24-27. McClintock, K. A., and Waterway, M. J. (1993). Patterns of allozyme variation and clonal diversity in Carex lasiocarpa and C. pellita (Cyperaceae). American Journal of Botany 80, 1251-1263. McDonald, R. C., and Isbell, R. F. (1990). Soil profile. In "Australian soil and land survey field handbook". (Ed. R. C. McDonald, R. F. Isbell, J. G. Speight, J. Walker and M. S. Hopkins) pp. (Inkata Press: Melbourne). McDonald, R. C., Isbell, R. F., Speight, J. G., Walker, J., and Hopkins, M. S. (1984). "Australian soil and land survey field handbook." (Inkata Press: Melbourne.) McLellan, A. J., Pratt, D., Kaltz, O., and Schmid, B. (1997). Structure and analysis of phenotypic and genetic variation in clonal plants. In "The ecology and evolution of clonal plants". (Ed. H. de Kroon and J. van Groenendael) pp. 185-210 (Backhuys Publishing: Leiden). Mejias, J. A., Arroyo, J., and Ojeda, F. (2002). Reproductive ecology of Rhododendron ponticum (Ericaceae) in relict Meditteranean populations. Botanical Journal of the Linnean Society 140, 297-311. Merler, S., Furlanello, C., Chemini, C., and Nicolini, G. (1996). Classification tree methods for analysis of mesoscale distribution of Ixodes ricinus (Acari: Ixodidae) in Trentino, Italian Alps. Journal of Medical Entomology 33, 888-893. Millar, C. I., and Libby, W. J. (1989). Disneyland or native ecosystem: genetics and the restorationist. Restoration and Management Notes 7, 18-24.

- 204 -

Milton, S. J., Dean, W. R. J., Du Plessis, M. A., and Siegfried, W. R. (1994). A conceptual model of arid rangeland degradation: the escalating cost of declining productivity. Bioscience 44, 70-76. Milton, S. J., and Wiegand, T. (2001). How grazing turns rare seedling recruitment events to non- events in arid environments. In "Sustainable land use in deserts". (Ed. S. W. Breckle, M. Veste and W. Wucherer) pp. 197-207 (Springer Verlag: New York). Mitchell, A. A., and Wilcox, D. G. (1994). "Arid shrubland plants of Western Australia." (University of Western Australia: Nedlands.) Mitchell, P. B. (1991). Historical perspectives on some vegetation and soil changes in semi-arid New South Wales. Vegetatio 91, 169-182. Mitchell, T. L. (1839). Three expeditions into the interior of eastern Australia: with descriptions of the recently explored region of Australia Felix and the present Colony of New South Wales T & W Boone: London. July, 2004. Mitton, J. B., and Grant, M. C. (1996). Genetic variation and the natural history of quaking aspen. Bioscience 46, 25-31. Moles, A. T., and Westoby, M. (2004). What do seedlings die from and what are the implications for evolution of seed size? Oikos 106, 193-199. Moncrieff, D. (1991). Hattah-Kulkyne National Park monitoring and revegetation strategy Department of Conservation and Environment: Mildura. March 1991. Moncur, M. W., Boland, D. J., and Harbard, J. L. (1997). Aspects of the floral biology of Allocasuarina verticillata (Casuarinaceae). Australian Journal of Botany 45, 857-869. Morcom, L. A., and Westbrooke, M. E. (1998). The pre-settlement vegetation of the western and central Wimmera Plains of Victoria, Australia. Australian Geographical Studies 36, 273- 288. Morgan, D. G. (1986). Estimating vertebrate population densities by line transect methods Melbourne College of Advanced Education: Melbourne. March 1986. Morison, P., and Harvey, G. (2001). Ants, floods and why buloke don't regenerate. Land for Wildlife News 3, 13. Morrison, P. C. (1941). Our newest National Forest - 120,000 acres of glory. Wild Life 3, 431- 433. Mortlock, W., and Lloyd, M. V. (2001). Guideline 8: Basic germination and viability tests for native plant seed. In "Floradata". (Ed. W. Mortlock and L. M. V.) pp. (Australian Centre for Mining Environmental Research, Australian National Botanic Gardens, CSIRO Forestry and Forest Products, Greening Australia: Canberra). Mueck, S. G. (1981). Vegetation monitoring survey 1981 Hattah-Kulkyne National Park National Parks and Wildlife Service: Mueck, S. G. (1983). Vegetation monitoring survey October 1983 Hattah-Kulkyne National Park National Parks and Wildlife Service: Mueck, S. G. (1987). Vegetation monitoring survey October 1986 Hattah-Kulkyne National Park National Parks and Wildlife Service: July 1987. Murdoch, P. J., and Murdoch, F. A. (2006). Vegetation monitoring plots and restoration efforts at Hattah Kulkyne National Park, 1970-2005 Parks Victoria: Mildura. Murphy, S. R., and Leys, J. F. (2004). Risk of deep drainage in the mallee. Journal of the Australian Association of Natural Resource Management 7, 14-28. Nathan, R., Safriel, U. N., Noy-Meir, I., and Schiller, G. (2000). Spatiotemporal variation in seed dispersal and recruitment near and far from Pinus halepensis trees. Ecology 81, 2156- 2169. Neel, M. C., and Ellstrand, N. C. (2003). Conservation of genetic diversity in the endangered plant Eriogonum ovalifolium var. vineum (Polygonaceae). Conservation Genetics 4, 337- 352. Nemes, A., and Rawls, W. J. (2004). Chapter 4: Soil texture and particle distribution as estimation of soil hydrology. In "Elsevier developments in soil sciences". (Ed. A. Nemes and W. J. Rawls) Volume 30: pp. 47-64 (Elsevier: Amsterdam). Noble, J. C. (1986). Plant population ecology and clonal growth in arid rangeland ecosystems. In "Rangelands : a resource under siege Proceedings of the 2nd International Rangeland

- 205 -

Congress" Adelaide, 1984, (Ed. P. J. Joss, P. W. Lynch and O. B. Williams) pp. (Cambridge University Press). Norbury, G. L. (1989). "Diet selection by western grey kangaroos in relation to declining food availability". PhD Thesis. Monash University. Norbury, G. L., and Norbury, D. C. (1993). The distribution of red kangaroos in relation to range regeneration. Rangeland Journal 15, 3-11. Norman, H. C., Freind, C., Masters, D. G., Ritoul, A. J., Dynes, R. A., and Williams, I. H. (2004). Variation within and between two saltbush species in plant composition and subsequent selection by sheep. Australian Journal of Agricultural Research 55, 999-1007. Noy-Meir, I. (1973). Desert ecosystems: environment and producers. Annual Review of Ecology and Systematics 4, 25-41. Noy-Meir, I. (1975). Stability of grazing systems: an application of predator-prey graphs. Journal of Ecology 63, 459-481. Nybom, H. (2004). Comparison of different nuclear DNA markers for estimating intraspecific genetic diversity in plants. Molecular Ecology 13, 1143-1155. Nybom, H., and Bartish, I. V. (2000). Effects of life history traits and sampling strategies on genetic diversity estimates obtained with RAPD markers in plants. Perspectives in Plant Ecology, Evolution and Systematics 3, 93-114. Obeid, M., and Seif El Din, A. (1970). Ecological studies of the vegetation of the Sudan 1. Acacia senegal (L.) Willd. and its natural regeneration. Journal of Applied Ecology 7, 507-518. Obeso, J. R., Alvarez-Santullanoa, M., and Retuertob, R. (1998). Sex ratios, size distributions, and sexual dimorphism in the dioecious tree Ilex aquifolium (Aquifoliaceae). American Journal of Botany 85, 1602-1608. O'Donoghue, J. G. (1915). Wanderings on the Murray flood-plain. Victorian Naturalist 32, 7-20 and 26-35. O'Donoghue, J. G. (1916). Rambles in Raak. Victorian Naturalist 33, 7-15 and 19-28. Ogle, K., and Reynolds, J. F. (2004). Plant responses to precipitation in desert ecosystems: integrating functional types, pulses, thresholds, and delays. Oecologia 141, 282-294. O'Hanlon, P. C., Peakall, R., and Briese, D. T. (2000). A review of new PCR-based genetic markers and their utility to weed ecology. Weed Research 40, 239-254. Olden, J. D., and Jackson, D. A. (2002). A comparison of statistical approaches for modelling fish species distributions. Freshwater Biology 47, 1976-1995. Oliver, I., Smith, P. L., Lunt, I. D., and Parkes, D. (2002). Pre-1750 vegetation, naturalness and vegetation condition: what are the implications for biodiversity conservation? Ecological Management and Restoration 3, 176-178. Palacios, C., and Gonzales-Candelas, F. (1997). Analysis of population genetic structure and variability using RAPD markers in the endemic and endangered Limonium dufourii (Plumbaginaceae). Molecular Ecology 6, 1107-1121. Parkes, D., Newell, G., and Cheal, D. C. (2003). Assessing the quality of native vegetation: The "habitat hectares" approach. Ecological Management and Restoration 4, S29-S38. Paruelo, J. M., and Sala, O. E. (1995). Water losses in the Patagonian steppe: a modelling approach. Ecology 76, 510-520. Paruelo, J. M., Sala, O. E., and Beltrán, A. B. (2000). Long term dynamics of water and carbon in semi-arid ecosystems: a gradient analysis in the Patagonian Steppe. Plant Ecology 150, 133-143. Paukkonen, K. A., Kauppi, A., and Ferm, A. (1992). Root and stump buds as structural faculties for reinvigoration in Alnus incana. Flora 187, 353-367. Peck, J. R., Yearsley, J. M., and Waxman, D. (1998). Explaining the geographic distributions of sexual and asexual populations. Nature 391, 889-892. Peltzer, D. A. (2002). Does clonal integration improve competitive ability? A test using aspen (Populus tremuloides [Salicaceae]) invasion into prairies. American Journal of Botany 89, 494-499. Pennings, S. C., and Calloway, R. M. (2000). The advantages of clonal integration under different ecological conditions: a community wide test. Ecology 81, 709-716.

- 206 -

Pepper, J. W., Male, T. D., and Robert, G. E. (2000). Foraging ecology of the South Australian glossy black cockatoo (Calyptorhynchus lathami halmaturinus). Austral Ecology 25, 16- 24. Pérez-García, F. (1997). Germination of Cistus ladanifer seeds in relation to parent material. Plant Ecology 133, 57-62. Perrow, M. R., and Davy, A. J. (2002). "Handbook of ecological restoration." (Cambridge University Press: Cambridge.) Peterson, C. J., and Jones, R. H. (1997). Clonality in woody plants: A review and comparison with clonal herbs. In "The ecology and evolution of clonal plants". (Ed. H. de Kroon and J. van Groenendael) pp. 263-289 (Backhuys Publishing: Leiden). Pickup, G., Bastin, G. N., and Chewings, V. H. (1998). Identifying trends in land degradation in non-equilibrium rangelands. Journal of Applied Ecology 35, 365-377. Piqueras, J. (1999). Herbivory and ramet performance in the clonal herb Trientalis europaea L. Journal of Ecology 87, 450-460. Pornon, A., Escaravage, N., Thomas, P., and Taberlet, P. (2000). Dynamic of genotypic structure in clonal Rhododendron ferrugineum (Ericaceae) population. Molecular Ecology 9, 1099- 1111. Quinn, G. P., and Keough, M. (2002). "Experimental design and data analysis for biologists." (Cambridge University Press: Melbourne.) R Development Core Team (2005). R: A language and environment for statistical computing, ISBN 3-900051-07-0, R Foundation for Statistical Computing. http://www.R-project.org. Rajora, O. P. (1999). Genetic biodiversity impacts of silvicultural practices and phenotypic selection in white spruce. Theoretical and Applied Genetics 99, 954-961. Ralph, M. (1999). "Seed collection of Australian native plants for revegetation, tree planting and direct seeding." (Bushland Horticulture: Fitzroy, Victoria.) Rawls, W. J., and Pachepsky, Y. A. (2002). Soil consistence and structure as predictors of water retention. Soil Science Society of America Journal 66, 115-1126. Raymond, K. L. (1990). "The regeneration biology of Allocasuarina luehmannii (R.T. Bak.) L. Johnson at Wyperfeld National Park". B.Sc.(Hons.) Thesis. Department of Ecology and Evolution, Monash University. Read, J. (1995). Recruitment characteristics of the white cypress pine (Callitris glaucophylla) in arid South Australia. Rangeland Journal 17, 228-240. Reddell, P., and Bowen, G. D. (1985). Frankia source affects growth, nodulation and nitrogen fixation in Casuarina species. New Phytologist 100, 115-122. Redford, K. H., and Taber, A. (2000). Writing the wrongs: Developing a safe-fail culture in conservation. Conservation Biology 14, 1567-1568. Reichenbacker, R. R., Schultz, R. C., and Hart, E. R. (1996). Artificial defoliation effect on Populus growth, biomass production and total nonstructural carbohydrate concentration. Environmental Entomology 25, 632-642. Reinartz, J. A., and Popp, J. W. (1987). Structure of clones of northern prickly ash (Xanthoxylum americanum). American Journal of Botany 74, 415-428. Rejwan, C., Collins, N. C., Brunner, J., Shuter, B. J., and Ridgway, M. S. (1999). Tree regression analysis on the nesting habitat of smallmouth bass. Ecology 80, 341-348. Reusch, T. B. H. (2001). Fitness-consequences of geitonogamous selfing in a clonal marine angiosperm (Zostera marina). Journal of Evolutionary Biology 14, 129-138. Rey Benayas, J. M., Espigares, T., and Castro-Diez, P. (2001). Simulated effects of herb competition on planted Quercus faginea seedlings in Mediterranean abandoned cropland. Applied Vegetation Science 6, 213-222. Reynolds, J. F., Kemp, P. R., Ogle, K., and Fernández, R. J. (2004). Modifying the pulse–reserve paradigm for deserts of North America: precipitation pulses, soil water, and plant responses. Oecologia 141, 194-210. Reynolds, J. F., Kemp, P. R., and Tenhunen, J. D. (2000). Effects of long-term rainfall variability on evapotranspiration and soil water distribution in the Chihuahuan Desert: a modeling analysis. Plant Ecology 150, 145-159.

- 207 -

Ringrose, A. J., Rhodes, D. W., and Hall, G. F. (1989). Reclamation of a scalded, red duplex soil by waterponding. Australian Journal of Soil Research 27, 779-795. Roach, D. A., and Wulff, R. D. (1987). Maternal effects in plants. Annual Review of Ecology and Systematics 18, 209-235. Robinson, J. P., and Harris, S. A. (1999). Amplified Fragment Length Polymorphisms and microsatellites: a phylogenetic perspective. In "Which DNA marker for which purpose?" (Ed. E. M. Gillet) pp. (Institut für Forstgenetik und Forstpflanzenzüchtung: Universität Göttingen). Rohlf, F. J. (1998). NTSYSpc: Numerical Taxonomy System, ver. 2.02i Exeter Publishing, Ltd: Setauket, New York. Roniker, M. (2002). The use of AFLP markers in conservation genetics - a case study on Pulsatilla vernalis in the Polish lowlands. Cellular and Molecular Biology Letters 7, 677- 684. Ross, J. (2004). Census of vascular plants of Victoria, National Herbarium of Victoria. http://www.rbg.vic.gov.au/research_and_conservation/plant_information/viclist, 2004. Sadras, V. O., and Baldock, J. (2003). Influence of size of rainfall events on water-driven processes I. Water budget of wheat crops in south-eastern Australia. Australian Journal of Agricultural Research 54, 341-351. Sadras, V. O., Baldock, J., Roget, D. K., and Rodriguez, D. (2003). Measuring and modelling yield and water budget components of wheat crops in coarse-textured soils with chemical constraints. Field Crops Research 84, 241-260. Sakai, A. K., and Burris, T. A. (1985). Growth in male and female aspen clones: a twenty-five year longitudinal study. Ecology 66, 1921-1927. Sakai, A. K., and Sharik, T. L. (1988). Clonal growth of male and female bigtooth aspen, Populus grandidentata. Ecology 69, 2031-2033. Sakai, S. (1998). A test for the resource remobilization hypothesis: tree sprouting using carbohydrates from above-ground parts. Annals of Botany 82, 213-216. Sala, O. E., and Lauenroth, W. K. (1982). Small rainfall events: an ecological role in semiarid regions. Oecologia 53, 301-304. Sala, O. E., Lauenroth, W. K., Parton, W. J., and Trlica, M. J. (1981). Water status of soil and vegetation in a shortgrass steppe. Oecologia 48, 327-331. Salzman, A. G., and Parker, M. A. (1985). Neighbours ameliorate local salinity stress for a rhizomatous plant in a heterogenous environment. Oecologia 65, 273-277. Sandell, P. R. (1999). Assessing the implications of RCD for the management of large conservation reserves, grazing leasehold and dryland agriculture in northwest Victoria Parks Victoria and Natural Resources and Environment: Kew, Victoria. October 1999. Sandell, P. R. (2002). Implications of rabbit haemorrhagic disease for the short-term recovery of semi-arid woodland communities in north-west Victoria. Wildlife Resarch 29, 591-598. Sandell, P. R., Ballentine, M., and Horner, G. (2001). Vegetation recovery in the Victorian mallee parks 1991-1998 Parks Victoria: Melbourne. December 2001. Sandell, P. R., Bone, D., Dennis, A., and Stevens, M. (2003). Pilot application of protocols for monitoring rabbits in two national parks in Western Victoria Parks Victoria: Mildura. Saxton, K. E., Rawls, W. J., Romberger, J. S., and Papendick, R. I. (1986). Estimating generalized soil-water characteristics from texture. Soil Science Society of America Journal 50, 1031- 1036. Schaal, B. A., O'Kane, S. L., and Rogstad, S. H. (1991). DNA variation in plant populations. Trends in Ecology and Evolution 6, 329-333. Scheffer, M., and Carpenter, S. R. (2003). Catastrophic regime shifts in ecosystems: linking theory to observation. Trends in Ecology and Evolution 18, 648-656. Schier, G. A. (1973a). Origin and development of aspen root suckers. Canadian Journal of Forest Research 3, 45-53. Schier, G. A. (1973b). Seasonal variation in sucker production from excised roots of Populus tremuloides and the role of endogenous auxin. Canadian Journal of Forest Research 3, 59-61.

- 208 -

Schier, G. A. (1981). Physiological research on adventitious shoot development in aspen roots.INT-107. USDA Forest Service: Ogden, UT. Schier, G. A., and Campbell, R. G. (1976). Differences among Populus species in the ability to form adventitious shoots and roots. Canadian Journal of Forest Research 6, 253-261. Schier, G. A., and Zasada, J. C. (1973). Role of carbohydrate reserves in the development of root suckers in Populus tremuloides. Canadian Journal of Forest Research 3, 243-250. Schirmer, J., and Field, J. (2000). The cost of revegetation ANU Forestry, Greening Australia, Environment Australia: Canberra. Schmid, B., and Bazzaz, F. A. (1987). Clonal integration and population structure in perennials: effects of severing rhizome connections. Ecology 68, 2016-2022. Schultz, P. A., Millar, R. M., Jastrow, J. D., Rivetta, C. V., and Bever, J. D. (2001). Evidence of mycorrhizal mechanism for the adaption of Andropogon gerardii (Poaceae) to high and low nutrient prairies. American Journal of Botany 88, 1650-1656. Schwinning, S., and Sala, O. E. (2004). Hierarchy of responses to resource pulses in arid and semi-arid ecosystems. Oecologia 141, 211-220. Schwinning, S., Starr, B. I., and Ehleringer, J. R. (2003). Dominant cold desert plants do not partition warm season precipitation by event size. Oecologia 136, 252-260. Schwinning, S., Starr, B. I., and Ehleringer, J. R. (2005). Summer and winter drought in a cold desert ecosystem (Colorado Plateau) part I: effects on soil water and plant water uptake. Journal of Arid Environments 60, 547-566. Seiwa, K. (2000). Effects of seed size and emergence time on tree seedling establishment: importance of developmental constraints. Oecologia 123, 208-215. SER (2002). The SER primer on ecological restoration Society for Ecological Restoration Science and Policy Working Group: www.ser.org. Shepperd, W. D., and Smith, F. W. (1993). The role of near-surface lateral roots in the life cycle of aspen in the central Rocky Mountains. Forest Ecology and Management 61, 157-170. Sherrod, P. H. (2003). DTREG - Classification and regression trees for data mining and modelling. http://www.dtreg.com/ Short, J., and Grigg, G. C. (1982). The abundance of kangaroos in suboptimal habitats: wheat, intensive pastoral, and mallee. Australian Wildlife Research 9, 221-227. Shuttleworth, J. W. (1993). Evaporation. In "Handbook of Hydrology". (Ed. D. R. Maidment) pp. (McGraw Hill). Silvertown, J., Franco, M., Pisanty, I., and Mendoza, A. (1993). Comparative plant demography - relative importance of life-cycle components to the finite rate of increase in woody and herbaceous perennials. Journal of Ecology 81, 465-476. Simonoff, J. S. (2003). "Analyzing categorical data." (Springer-Verlag: New York.) Simons, A. J., and Johnston, M. O. (2000). Variation in seed traits of Lobelia inflata (Campanulaceae): sources and fitness consequences. American Journal of Botany 87, 124-132. Sloane, D., and Morgan, S. P. (1996). An introduction to categorical data analysis. Annual Review of Sociology 22, 351-375. Sluiter, I. R. K., Allen, G. G., Morgan, D. G., and Walker, I. S. (1997a). Vegetation responses to stratified kangaroo grazing pressure at Hattah-Kulkyne National Park, 1992-96 Department of Natural Resources and Environment: East Melbourne. August 1997. Sluiter, I. R. K., Minchin, P. R., and Jaensch, S. C. (1997b). The Buloke and Pine woodlands of semi-arid and dry sub-humid Victoria and nearby areas Ogyris Ecological Research: Mildura. Smouse, P. E., Dyer, R. J., Westfall, R. D., and Sork, V. L. (2001). Two-generation analysis of pollen flow across a landscape. I. Male gamete heterogeneity among females. Evolution 55, 260-271. Smyth, A., and James, C. D. (2004). Characteristics of Australia's rangelands and key design issues for monitoring biodiversity. Austral Ecology 29, 3-15. Snyder, K. A., Donovan, L. A., James, J. J., Tiller, R. L., and Richards, J. H. (2004). Extensive summer water pulses do not necessarily lead to canopy growth of Great Basin and northern Mojave Desert shrubs. Oecologia 141, 325-334.

- 209 -

Sorenson, F. C., and Webber, J. E. (1997). On the relationship between pollen capture and seed set in conifers. Canadian Journal of Forest Research 27, 63-68. Sork, V. L., Davis, F. W., Dyer, R. J., and Smouse, P. E. (2002a). Mating patterns in a savanna population of Valley Oak (Quercus lobata Neé) USDA Forest Service: Sork, V. L., Davis, F. W., Smouse, P. E., Apsit, V. J., Dyer, R. J., Fernandez, J. M., and Kuhk, B. (2002b). Pollen movement in declining populations of California Valley oak, Quercus lobata: where have all the fathers gone? Molecular Ecology 11, 1657-1668. Sparrow, A. D., Friedel, M. H., and Tongway, D. J. (2003). Degradation and recovery processes in arid grazing lands of central Australia. 3 Implications at landscape scale. Journal of Arid Environments 55, 349-360. Stafford-Smith, D. M., and Morton, S. R. (1990). A framework for the ecology of arid Australia. Journal of Arid Environments 18, 255-278. Stanley, R. J. (1983). Soils and vegetation: an assessment of current status. In "What future for Australia's arid lands?" (Ed. J. Messer and G. Mosley) pp. 8-18 (Australian Conservation Foundation: Melbourne). Steane, D. A., Wilson, K. L., and Hill, R. S. (2003). Using matK sequence data to unravel the phylogeny of Casuarinaceae. Molecular Phylogenetics and Evolution 28, 47-59. Stehlik, I., and Holderegger, R. (2000). Spatial genetic structure and clonal diversity of Anemone nemorosa in late successional deciduous woodlands of Central Europe. Journal of Ecology 88, 424-435. Stelzl, I. (2000). What sample sizes are needed to get correct significance levels for log-linear models? - A Monte Carlo Study using the SPSS-procedure "Hiloglinear". Methods of Psychological Research 5, online. Stenberg, P., Lundmark, M., and Saura, A. (2003). MLGsim: a program for detecting clones using a simulation approach. Molecular Ecology Notes 3, 329-331. Stewart, C. N., and Porter, D. M. (1995). RAPD profiling in biological conservation: an application to estimating clonal variation in rare and endangered Iliamna in Virginia. Biological Conservation 74, 135-142. Story, R. (1967). Pasture patterns and associated soil water in partially cleared woodland. Australian Journal of Botany 15, 175-187. Stringham, T. K., Krueger, W. C., and Shaver, P. L. (2003). State and transition modeling: an ecological processes approach. Journal of Range Management 56, 106-113. Suding, K. N., Gross, K. L., and Houseman, G. (2004). Alternative states and positive feedbacks in restoration ecology. Trends in Ecology and Evolution 19, 46-53. Swetnam, T. W., Allen, C. D., and Betancourt, J. L. (1999). Applied historical ecology: Using the past to manage for the future. Ecological Applications 9, 1189-1206. Sydes, M. A., and Peakall, R. (1998). Extensive clonality in the endangered shrub Haloragodendron lucasii (Haloragaceae) revealed by allozymes and RAPDs. Molecular Ecology 7, 87-93. Tero, N., Aspi, J., Sjikamaki, P., Jakalaniemi, A., and Tuomi, J. (2003). Genetic structure and gene flow in a metapopulation of an endangered plant species, Silene tatarica. Molecular Ecology 12, 2073-2085. Tew, R. K. (1970). Root carbohydrate reserves in vegetative reproduction of aspen. Forest Science 16, 318-320. Thomas, B. (2003). Hattah Kulkyne revegetation program (2000-2003) Greening Australia: Mildura. March 2003. Thomas, B. R., MacDonald, S. E., Hicks, M., Adams, D. L., and Hodgetts, R. B. (1999). Effects of reforestation methods on genetic diversity of lodgepole pine: an assessment using microsatellite and randomly amplified polymorphic DNA markers. Theoretical and Applied Genetics 98, 793-801. Thrall, P. H., Millsom, D. A., Jeavons, A. C., Waayers, M., Harvey, G. R., Bagnall, D. J., and Brockwell, J. (2005). Seed inoculation with effective root-nodule bacteria enhances revegetation success. Journal of Applied Ecology 42, 740-751.

- 210 -

Tibayrenc, M., Kjellberg, F., Arnaud, J., Oury, B., Breniere, S. F., Darde, M., and Ayala, F. J. (1991). Are eukaryotic microorganisms clonal or sexual? A population genetics vantage. Proceedings of the National Academy of Sciences of the USA 88, 5129-5133. Tibayrenc, M., Kjellberg, F., and Ayala, F. J. (1990). A clonal theory of parasitic protozoa: the population structures of Entamoeba, Giardia, Leishmania, Naegleria, Plasmodium, Trichomonas, and Trypanosoma and their medical and taxonomical consequences. Proceedings of the National Academy of Sciences of the USA 87, 2414-2418. Tissue, D. T., and Nobel, P. S. (1988). Parent-ramet connections in Agave deserti: influences of carbohydrates on growth. Oecologia 75, 266-271. Tiver, F. (1994). "Vegetation patterns of eastern South Australia : edaphic control and effects of herbivory". PhD Thesis. Department of Botany, University of Adelaide. Tiver, F., and Andrew, M. H. (1997). Relative effects of herbivory by sheep, rabbits, goats and kangaroos on recruitment and regeneration of shrubs and trees in eastern South Australia. Journal of Applied Ecology 34, 903-914. Tolvanen, A. (1994). Differences in recovery between a deciduous and evergreen ericaceous clonal dwarf shrub after simulated aboveground herbivory and below ground damage. Canadian Journal of Botany 72, 853-859. Tolvanen, A., and Laine, K. (1997). Effects of reproduction and artificial herbivory on vegetative growth and resource levels in deciduous and evergreen dwarf shrubs. Canadian Journal of Botany 75, 656-666. Tongway, D. J. (2003). Reading the landscape. A training course in monitoring rangelands by landscape function analysis CSIRO Sustainable Ecosystems: Canberra. Tongway, D. J., and Hindley, N. L. (1995). Manual for soil condition assessment of tropical grasslands CSIRO Sustainable Ecosystems: Canberra. Tongway, D. J., and Ludwig, J. A. (1990). Vegetation and soil patterning in semi-arid mulga lands of Eastern Australia. Australian Journal of Ecology 15, 23-34. Tongway, D. J., and Ludwig, J. A. (1994). Small-scale resource heterogeneity in semi-arid landscapes. Pacific Conservation Biology 1, 201-208. Tongway, D. J., and Ludwig, J. A. (1997). The nature of landscape dysfunction in rangelands. In "Landscape ecology, function and management". (Ed. J. A. Ludwig, D. J. Tongway, D. O. Freudenberger, J. Noble and K. C. Hodgkinson) pp. 49-61 (CSIRO Publishing: Collingwood). Tongway, D. J., Ludwig, J. A., and Whitford, W. G. (1989). Mulga log mounds: fertile patches in semi-arid woodlands of eastern Australia. Australian Journal of Ecology 14, 263-268. Tongway, D. J., Sparrow, A. D., and Friedel, M. H. (2003). Degradation and recovery processes in arid grazing lands of central Australia. 1 Soil and land resources. Journal of Arid Environments 55, 301-326. Torrey, J. G. (1983). Root development and root nodulation in Casuarina. Casuarina ecology management and utilization. S. J. Midgley, J. W. Turnbull and R. D. Johnston. Canberra, CSIRO: 180-192. Turnbull, J. W., and Martensz, P. N. (1982). Aspects of seed collection, storage and germination in Casuarinaceae. Australian Forest Research 12, 281-294. Turnbull, J. W., and Martensz, P. N. (1983). Seed production, collection and germination in Casuarinaceae. In "Casuarina ecology management and utilization". (Ed. S. J. Midgley, J. W. Turnbull and R. D. Johnston) pp. 126-132 (CSIRO: Canberra). Tyson, M., Vaillancourt, R. E., and Reid, J. B. (1998). Determining the clone size and age in a mallee eucalypt using RAPDs. Australian Journal of Botany 46, 161-172. Underwood, A. J. (1997). "Experiments in ecology: their logical design and interpretation using analysis of variance." (Cambridge University Press: Melbourne.) Urban, A. (1996). "Wildflowers and plants of inland Australia." (Portside Editions, Pty Ltd: Fisherman's Bend, Victoria.) Vallejo, R., Aronson, J., Pausas, J. G., and Cortina, J. (2006). Restoration of Mediterranean woodlands. In "Restoration ecology: the new frontier". (Ed. J. van Andel and J. Aronson) pp. 193-207 (Blackwell Publishing: Malden MA).

- 211 -

van Andel, J., and Aronson, J. (2006). "Restoration ecology: the new frontier." (Blackwell Publishing: Malden MA.) van de Koppel, J., Rietkerk, M., van Langevelde, F., Kumar, L., Klausmeier, C. A., Fryxell, J. M., Hearne, J. W., van Andel, J., De Ridder, N., Skidmore, A., Stroosnijder, L., and Prins, H. H. T. (2002). Spatial heterogeneity and irreversible vegetation change in semiarid grazing systems. The American Naturalist 159, 209-218. van den Berg, L., and Kellner, K. (2005). Restoring degraded patches in a semi-arid rangeland of South Africa. Journal of Arid Environments 61, 497-511. van der Heyden, F., and Stock, W. D. (1996). Regrowth of a semiarid shrub following simulated browsing: the role of reserve carbon. Functional Ecology 10, 647-653. van der Hulst, R. G. M., Mes, T. H. M., Den Jins, J. C. M., and Bachmann, K. (2000). AFLP markers reveal that population structure of triploid dandelions (Taraxacum officianale) exhibits both clonality and recombination. Molecular Ecology 9, 1-8. van Kleunen, M., and Stuefer, J. F. (1999). Quantifying the effects of reciprocal assimilate and water translocation in a clonal plant by the use of steam-girdling. Oikos 85, 135-145. Vayssieres, M. P., and Plant, R. E. (1998). Identification of vegetation state and transition domains in California's hardwood rangelands California Department of Forestry and Fire Protection: Davis, California. Verdú, M., and Traveset, A. (2005). Early emergence enhances plant fitness: a phylogenetically controlled meta-analysis. Ecology 86, 1385-1394. Vesk, P. A., Leishman, M. R., and Westoby, M. (2004a). Simple traits do not predict grazing response in Australian dry shrublands and woodlands. Journal of Applied Ecology 41, 22- 31. Vesk, P. A., Warton, D. I., and Westoby, M. (2004b). Sprouting by semi-arid plants: testing a dichotomy and predictive traits. Oikos 107, 72-89. Vesk, P. A., and Westoby, M. (2001). Predicting plant species' responses to grazing. Journal of Applied Ecology 38, 897-909. Vilagrosa, A., Cortina, J., Gil-Pelegrin, E., and Bellot, J. (2003). Suitability of drought- preconditioning techniques in mediterranean climate. Restoration Ecology 11, 208-216. VNPA (1959). Request to the Government of Victoria asking that all of the Kulkyne State Forest be made National Park. Victorian National Parks Association Newsletter 27, 1-25. Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot, J., Peleman, J., and Kuiper, M. (1995). AFLP: a new technique for DNA fingerprinting. Nucleic Acids Research 23, 4407-4414. Wakefield, N. A. (1966). Mammals of the Blandowski expedition to north-western Victoria. Proceedings of the Royal Society of Victoria 79, 371-391. Walker, B., and Langridge, J. (1996). Modelling plant and soil water dynamics in semi-arid ecosystems with limited site data. Ecological Modelling 87, 153-167. Walker, B. H., Stone, L., Henderson, L., and Vernede, M. (1986). Size structure analysis of the dominant trees in a South African savanna. South African Journal of Botany 52, 397-402. Walker, S. (1997). Models of vegetation dynamics in semi-arid vegetation: application to lowland central Otago, New Zealand. New Zealand Journal of Ecology 21, 129-140. Walsh, D., Andrew, M. H., Sinclair, R., and Coleman, D. (2005). Evidence for palatability variation between the sex phenotypes of bladder saltbush from small-plot grazing trials. The Rangeland Journal 27, 89-96. Warburton, C. L., James, E. A., Fripp, Y. J., Trueman, S. J., and Wallace, H. M. (2000). Clonality and sexual reproductive failure in remnant populations of Santalum lanceolatum (Santalaceae). Biological Conservation 96, 45-54. Wasson, R. J., and Clark, R. L. (1985). Environmental history for explanation and prediction. Search 16, 258-263. Watkinson, A. R., and Powell, J. C. (1993). Seedling recruitment and the maintenance of clonal diversity in plant populations - a computer simulation of Ranunculus repens. Journal of Ecology 81, 707-717. Watson, I. W., Burnside, D. G. B., and Holm, A. M. (1996). Event-driven or continuous; which is the better model for managers? Rangeland Journal 18, 351-369.

- 212 -

Watson, I. W., Westoby, M., and Holm, A. M. (1997a). Continuous and episodic components of demographic change in arid zone shrubs: models of two Eremophila species from Western Australia compared with published data on other species. Journal of Ecology 85, 833-846. Watson, I. W., Westoby, M., and Holm, A. M. (1997b). Demography of two shrub species from an arid grazed ecosystem in Western Australia. Journal of Ecology 85, 815-832. Weising, K., Nybom, H., Wolff, K., and Meyer, W. (1995). "DNA fingerprinting in plants and fungi." (CRC Press: Boca Raton.) Wellington, A. B., and Noble, I. R. (1985). Seed dynamics and factors limiting recruitment of mallee Eucalyptus incrassata in semi-arid, south-eastern Australia. Journal of Ecology 73, 657-666. Wesche, K., Ronnenberg, K., and Hensen, I. (2005). Lack of sexual reproduction within mountain steppe populations of the clonal shrub Juniperus sabina L. in semi-arid southern Mongolia. Journal of Arid Environments 63, 390-405. West, N. E. (1999). Managing for biodiversity of rangelands. In "Biodiversity in agroecosystems". (Ed. W. Collins and C. Qualset) pp. 101-119 (CRC Press: Washington, D.C.). West, N. E., Rea, K. H., and Harniss, R. O. (1979). Plant demographic studies in sagebrush-grass communities in southeastern Idaho. Ecology 60, 376-388. Westbrooke, M. E. (1998). "The ecology and regeneration status of Casuarina pauper / Alectryon oleifolius ssp. canescens (Belah) woodlands in south-eastern Australia". PhD. Thesis. Department of Botany, School of Life Sciences, Faculty of Science, Technology and Engineering, La Trobe University. Westbrooke, M. E., and Florentine, S. K. (2005). Rainfall-driven episodic flood events: are they a major factor in moulding New South Wales arid land vegetation patterns. Australian Geographer 36, 171-181. Westbrooke, M. E., Florentine, S. K., and Milberg, P. (2005). Arid land vegetation dynamics after a rare flooding event: influence of fire and grazing. Journal of Arid Environments 61, 249-260. Westbrooke, M. E., Gibson, M., Gowans, S., and Callister, K. (2001). Vegetation condition assessment of the semi-arid woodlands of Murray-Sunset National Park Centre for Environmental Management, University of Ballarat: Ballarat. April 2001. Westbrooke, M. E., Humphries, R., and Morcom, L. A. (1988). The ecology and conservation status of Myoporum platycarpum woodlands in Victoria Dept. Applied Biology, Ballarat C.A.E.: Ballarat. April 1988. Westoby, M., Walker, B. H., and Noy-Meir, I. (1989). Opportunistic management for rangelands not at equilibrium. Journal of Range Management 42, 266-274. Whisenant, S. G. (1999). "Repairing damaged wildlands: a process-orientated, landscape-scale approach." (Cambridge University Press: Cambridge.) White, M., Oates, A., Barlow, T., Pelikan, M., Brown, J., Rosengren, N., Cheal, D. C., Sinclair, S., and Sutter, G. (2003). The vegetation of north-west Victoria: a report to the Wimmera, North Central and Mallee Catchment Management Authorities Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment: Melbourne. Widen, B., Cronberg, N., and Widen, M. (1994). Genotypic diversity, molecular markers and spatial distribution of genets in clonal plants, a literature survey. Folia Geobotanica et Phytotaxonomica 29, 245-263. Wiegand, K., Jeltsch, F., and Ward, D. (1999). Analysis of the population dynamics of Acacia trees in the Negev desert, Israel, with a spatially-explicit computer simulation model. Ecological Modelling 117, 203-224. Wiegand, K., Jeltsch, F., and Ward, D. (2004). Minimum recruitment frequency in plants with episodic recruitment. Oecologia 141, 363-372. Wiegand, K., Ward, D., Thulke, H. H., and Jeltsch, F. (2000). From snap-shot information to long-term population dynamics of Acacias by a simulation model. Plant Ecology 150, 97- 114. Wilcock, C., and Neiland, R. (2002). Pollination failure in plants: why it happens and when it matters. Trends in Ecology and Evolution 7, 270-277.

- 213 -

Williams, B. G., and Semple, W. S. (2001). Implications of the chemical composition of saline seepage scalds. In "Wanted Sustainable Futures for Saline Lands, Seventh National PURSL Conference" Launceston, (Ed. C. Grose, L. Bond and T. Pinkard) pp.). Williams, J. (1980). Plan for rabbit control works in Hattah-Kulkyne National Park and Murray- Kulkyne Park 1980-85 National Parks Service: Mildura. Williams, K., Parer, I., Coman, B., Burley, J., and Braysher, M. (1995). "Managing vertebrate pests: rabbits." (Australian Government Publishing Service: Canberra.) Williams, P. R. (2000). Fire-stimulated rainforest seedling recruitment and vegetative regeneration in a densely grassed wet sclerophyll forest of north-eastern Australia. Australian Journal of Botany 48, 651-658. Williamson, H. B. (1913). Botanical notes of a trip to Mildura. Victorian Naturalist 30, 101-110. Wilson, A. D. (1977). The digestibility and voluntary intake of the leaves of trees and shrubs by sheep and goats. Australian Journal of Agricultural Research 28, 501-508. Wilson, A. D., Leigh, J. H., Hindley, N. L., and Mulham, W. E. (1975). Comparison of the diets of goats and sheep on a Casuarina cristata - Heterodendrum oleifolium woodland community in western New South Wales. Australian Journal of Experimental Agriculture and Animal Husbandry 15, 45-53. Wilson, A. D., Mulham, W. E., and Leigh, J. H. (1976). A note on the effects of browsing by feral goats on a belah (Casuarina cristata) - rosewood (Heterodendrum oleifolium) woodland. Australian Rangeland Journal 1, 7-12. Wilson, K. L., and Johnson, L. A. S. (1989). Casuarinaceae, Australian Biological Resources Study. http://www.deh.gov.au/biodiversity/abrs/online-resources/flora/main/index.html, Viewed 15 February, 2005. Wisniewski, and Parsons, R. F. (1986). Regeneration of Heterodendrum in arid south-eastern Australia. Proceedings of the Royal Society of Victoria 98, 87-95. Woodell, S. R. J. (1990). Regeneration in the shrub Acacia burkittii FvM. ex Benth. in the arid zone of South Australia. Biological Conservation 51, 39-48. Xiong, S., and Nilsson, C. (1999). The effects of plant litter on vegetation: a meta-analysis. Journal of Ecology 87, 984-994. Yamada, T., Kumagawa, Y., and Suzuki, E. (2001). Adaptive significance of vegetative sprouting for a tropical canopy tree, Scaphium longiflorum (Sterculiaceae), in a peat swamp forest in Central Kalimantan. Ecological Research 16, 641-647. Yan, G., Romero-Severson, J., Walton, M., Chadee, D. D., and Severson, D. W. (1999). Population genetics of the yellow fever mosquito in Trinidad: comparisons of AFLP and RFLP markers. Molecular Ecology 8, 951-963. Yates, C. J., and Hobbs, R. J. (1997). Woodland restoration in the Western Australian wheatbelt: a conceptual framework using a state and transition model. Restoration Ecology 5, 28. Yates, C. J., Norton, D. A., and Hobbs, R. J. (2000). Grazing effects on plant cover, soil and microclimate in fragmented woodlands in south-western Australia: implications for restoration. Austral Ecology 25, 36-47. Yeh, F. C., Chong, D. K. X., and Yang, R. C. (1995). RAPD variation within and among natural populations of trembling aspen (Populus tremuloides Michx.) from Alberta. Journal of Heredity 86, 454-460. Yeh, F. C., Yang, R. C., Boyle, T., Ye, Z. H., and Mao, J. X. (1997). POPGENE, the user-friendly shareware for population genetic analysis Molecular Biology and Biotechnology Centre, University of Alberta, Canada: Young, A. G., Hill, J. H., Murray, B. G., and Peakall, R. (2002). Breeding system, genetic diversity and clonal structure in the subalpine forb Rutidosis leiolepis F. Muell. (Asteraceae). Biological Conservation 106, 71-78. Young, J. (undated). Biosphere Reserves - Murray Valley Reserve. UNESCO Review, 12-16. Zahner, R., and De Byle, N. V. (1965). Effect of pruning the parent root on growth of aspen suckers. Ecology 46, 373-375. Zhang, L., Dawes, W. R., Hatton, T. J., Hume, I. H., O'Connell, M. G., Mitchell, D. C., Milthorpe, P. L., and Yee, M. (1999). Estimating episodic recharge under different crop/pasture

- 214 -

rotations in the Mallee region Part 2 Recharge control by agronomic practices. Agricultural Water Management 42, 237-249. Zhang, L., Walker, G. R., and Dawes, W. R. (2002). Water balance modelling: concepts and applications. In "Regional water and soil assessment for managing sustainable agriculture in China and Australia". (Ed. T. R. McVicar, L. Rui, J. Walker, R. W. Fitzpatrick and C. Liu) Volume 84: pp. 31-47 (ACIAR). Zhivotovsky, L. A. (1999). Estimating population structure in diploids with multilocus dominant DNA markers. Molecular Ecology 8, 907-913. Zimmer, W. J. (1944). Notes on the regeneration of Murray Pine (Callitris spp.). Transactions of the Royal Society of South Australia 68, 183-190.

- 215 -

Appendix 1: Australian trees and shrubs from the rangelands known to produce root suckers Scientific name Common Family Form Habitat States Comments Source name Acacia atrox Myall Creek Mimosaceae Shrub Deep clay, N Spreading by suckering 14 Wattle basalt woodland Acacia Purple Wood Mimosaceae Shrub or Mulga N S Propagates itself mainly by suckers 1, 14 carneorum Wattle small communities tree Acacia coriacea Wiry Wattle Mimosaceae Shrub or Sandy rises N Q S W Adventitious buds on lateral roots 6 small NT tree Acacia dealbata Silver Wattle Mimosaceae Tree Riverine N V T S Produces root suckers freely 1 forest Acacia Hakea Wattle Mimosaceae Shrub Mallee N Q V S W In higher rainfall areas it suckers freely 1 hakeoides communities Acacia Brigalow Mimosaceae Tree Clay scrub N Q Communities that arise from remnant suckers will 1 harpophylla community soon be indistinguishable from uncleared communities Acacia implexa Hickory Mimosaceae Tall Rocky soils N Q V In higher rainfall zones it suckers from the roots 1 Wattle shrub to or red gum small communities tree Acacia melvillei Yarran Mimosaceae Tree to Heavy loam N V Q Tends to form small stands due to its suckering 1 15m grassland and habit, especially on disturbed sites woodland

- 216 -

Scientific name Common Family Form Habitat States Comments Source name Acacia Sandplain Mimosaceae Shrub or Sand N Q V S W Root suckering is common and thickets are formed 1 murrayana Wattle small communities with young trees often a considerable distance from tree the parent Acacia pickardii None Mimosaceae Bushy Gibber- S NT Spreads by suckering 8 shrub or covered tree 3–5 sandplains m high over clay rises Acacia salicina Cooba Mimosaceae Tree Along A Dense clumps with numerous suckers 1 drainage lines Acacia victoriae Prickly Wattle Mimosaceae Tall Variable Q V S W Can regenerate readily from suckers and sometimes 1 shrub NT forms thickets Alectryon Cattlebush Sapindaceae Small to Widespread, A Suckers readily from the roots, particularly if the 1 oleifolius medium sandy soils roots are exposed or damaged. Reproduction tree appears entirely from root suckers Allocasuarina Buloke Casuarinaceae Tree Woodland N Q V S Root suckers occur but they are less frequent than 1 luehmannii those of Belah Alstonia Quinine Bush Apocynaceae Small to Sandy rises N Q Falling trees results in suckering from the roots 1 constricta medium tree Atalaya Whitewood Sapindaceae Small to Open plains N Q S W Small isolated clumps, often with numerous root 1, 3 hemiglauca medium and flats, NT suckers, suckers may be toxic. tree dunes Canthium Wild Lemon Rubiaceae Shrub or Sandy N Suckers freely from the roots and forms 1 oleifolium small woodland characteristic thickets tree communities

- 217 -

Scientific name Common Family Form Habitat States Comments Source name Casuarina Swamp Oak Casuarinaceae Tree Brackish N Q Frequently producing root suckers 9 glauca situations Casuarina obesa Swamp Casuarinaceae Tree Swampy N V S W Clump like growth form 1 Sheoak areas Casuarina Belah Casuarinaceae Tree Woodland N Q V S W Propagation is chiefly from root suckers which 1 pauper initiate readily from exposed roots. Duboisia Pituri Solanaceae Shrub Sandy soils N Q S W Disturbance associated with road making has lead 1 hopwoodii NT to an increase in density and growth, probably due to the stimulation of root suckers Eremophila Berrigan Myoporaceae Shrub or Various A Often surrounded by a circle of smaller plants 1 longifolia Tree except heavy which have established from seed or from root clay suckers Eremophila Budda Myoporaceae Shrub or Various N Q Capable of regenerating from the roots after 1 mitchellii tree mechanical disturbance Eremophila Turpentine Myoporaceae Shrub Various A Propagation also takes place from root suckers 1 sturtii which form readily if the roots are cut or disturbed Eremophila Sandhill Myoporaceae Small Sand NT Reproduces from suckers and seedlings 3 willsii Native shrub Fuschia Grevillea White Spider Proteaceae Shrub or Sand plains NT Shoots from the roots after fires 3 albiflora Flower small and dunes tree Grevillea Honey Proteaceae Shrub Mallee W NT S Shoots from the roots after drought or fire 3 eriostachya Grevillea communities Grevillea Comb Proteaceae Shrub Mallee V S W Occasionally forms thickets 1, 7 huegelii Grevillea communities - 218 -

Scientific name Common Family Form Habitat States Comments Source name Grevillea Desert Proteceae Shrub Sand plains W N Q S Shoots again after fire 3 juncifolia Grevillea and dunes NT Grevillea Rattlepod Proteaceae Shrub Sandy rises W N Q S Can shoot from the roots 3 stenobotrya Grevillea NT Hakea ivoryi Corkbark Proteaceae Tree Sandplain N Q S Capable of suckering from the roots 1 Tree woodlands Hakea Needlewood Proteaceae Small Woodland A Large parent plants surrounded by a thicket of 1, 3 leucoptera tree smaller offspring, some or all of which may have resulted from root suckering Hakea Hooked Proteaceae Small Woodland N Q V S Large parent plants surrounded by a thicket of 1 tephrosperma Needlewood tree smaller offspring, some or all of which may have resulted from root suckering Leptospermum Mallee Myrtaceae Shrub Mallee dunes V S Adults killed by frost produced suckers from lateral 10 coriaceum Teatree roots Lycium australe Australian Solanaceae Shrub Open V S W Often in rather dense colonies 1, 7 Boxthorn bluebush communities Myoporum Boobyalla Myoporaceae Shrub Sand N Q S T W 7 insulare communities Owenia acidula Colane Meliaceae Small Sand N Q Y S The trees, particularly those on sandy soils, are 1 tree communities often surrounded by suckers which have developed from exposed roots Pittosporum Weeping Pittosporaceae Shrub or Woodland A Parent plant surrounded by a large number of 1 phylliraeoides Pittosporum small juveniles tree

- 219 -

Scientific name Common Family Form Habitat States Comments Source name Santalum Sweet Santaleaceae Shrub or Woodland A Suckers following some root disturbance 1 acuminatum Quandong tree Santalum Northern Santalaceae Shrub or Woodland A Occasionally appearing as small closely spaced 5 lanceolatum Sandlewood tree clumps. Young individuals are likely to be root suckers from the oldest specimens Senna Silver Cassia Caesalpinaceae Shrub Various N Q V S W Sometimes suckering and forming dense colonies. 3, 4 artemisioides NT Genetic evidence of clonal growth. Reshoots from and other Senna the roots. Senna nemophila shoots very quickly spp. following rain, both from seed and the perennial roots Solanum Velvet potato Solanaceae Forb Range of Q S W NT 6 ellipticum bush soils, rarely clay Solanum Desert Solanaceae Shrub Clay N Q S May have spreading underground stems 1 oligacanthum nightshade depressions in dunefields Ventilago Supplejack Rhamnaceae Small to Various N Q NT W Suckers freely if the roots are exposed or injured. 1 viminalis medium The suckers regrow well tree Sources: 1: Cunningham et al. (1992), 2: Doran and Turnbull (1997), 3: Urban (1996), 4: Holman and Playford (2000), 5: Warburton et al. (2000), 6: Maconochie (1982), 7: Tiver (1994), 8: Kodela (2001), 9: Entwisle (1989), 10: LCC (1987).

- 220 -