The University of Notre Dame Australia ResearchOnline@ND

Theses

2020

Causes of Variable Jarrah (Eucalyptus Marginata) and Marri (Corymbia Calophylla) Seedling Density at Establishment Following Bauxite Mining in the Northern Jarrah Forest of Western Australia

Tai White-Toney The University of Notre Dame Australia

Follow this and additional works at: https://researchonline.nd.edu.au/theses

Part of the Arts and Humanities Commons

COMMONWEALTH OF AUSTRALIA Copyright Regulations 1969

WARNING The material in this communication may be subject to copyright under the Act. Any further copying or communication of this material by you may be the subject of copyright protection under the Act. Do not remove this notice.

Publication Details White-Toney, T. (2020). Causes of Variable Jarrah (Eucalyptus Marginata) and Marri (Corymbia Calophylla) Seedling Density at Establishment Following Bauxite Mining in the Northern Jarrah Forest of Western Australia (Doctor of Philosophy (College of Arts and Science)). University of Notre Dame Australia. https://researchonline.nd.edu.au/theses/262

This dissertation/thesis is brought to you by ResearchOnline@ND. It has been accepted for inclusion in Theses by an authorized administrator of ResearchOnline@ND. For more information, please contact [email protected]. Running head: JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING

Causes of variable jarrah (Eucalyptus marginata) and marri (Corymbia calophylla) seedling density at establishment following bauxite mining in the Northern Jarrah Forest of Western Australia

Tai White-Toney

Submitted in fulfilment of the requirements for a Doctor of Philosophy

School of Arts and Sciences, Fremantle Campus

January 2020

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING ii

Declaration of Authorship

To the best of the candidate’s knowledge, this thesis contains no material previously published by another person, except where due acknowledgement has been made.

This thesis is the candidate’s own work and contains no material which has been accepted for the award of any other degree or diploma in any institution.

Tai White-Toney

DATE

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING iii

Abstract

Seedling densities at establishment of jarrah (Eucalyptus marginata D. Don ex Sm.) and marri (Corymbia calophylla (Lindl). K. D. Hill & L. A. S. Johnson) following bauxite mining in the Northern Jarrah Forest of south-western Australia are variable and few factors explored during 40 years of field studies have explained this variability. This study explored establishment within the framework of limitation due to amount of seed supply and number of suitable microsites for establishment. First, relationships were identified between seedling establishment densities on 654 mine sites restored between 1998 and 2017 and restoration practices (seed and fertiliser application rates), climate conditions (rainfall evenness and seasonal air temperatures) and seed supply (measured as proportion of forest perimeter). Then, canopy-stored seed as a seed source on restoration sites, and its influence on seedling densities, was investigated. Mean (± SE) jarrah seedfall in the first 12.7 m of the forest edges was 10,000 ± 2,624 seeds/ha/year. The variable, but potentially high, jarrah seedfall within 12.7 m of forest edge is an important seed source influencing jarrah seedling densities on restoration sites, but there was very low presence of marri seed in both forest and restoration sites. Last, three complementary field trials were implemented to investigate the importance of microsite characteristics on jarrah and marri seedling establishment in the restoration system. Seedling survival to establishment was high, 83.9–89.2% for jarrah and

93.8–96.8% marri, and survival was not related to microtopography generated by soil ripping or substrate type. However, seedling density was highest in furrows for both species, most likely because of seed movement during surface water run-off. Also, marri density at establishment increased with sulphur and organic carbon and decreased with soil conductivity. Net seed supply of jarrah and marri over a restoration site (including both canopy-stored and sown seed) is variable, and supply, rather than microsite characteristics, is the most likely limitation to seedling establishment. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING iv

Acknowledgements

In some ways, this acknowledgements section has been the hardest part of this thesis to write. I find myself at the conclusion of one extraordinary journey and simultaneously on the precipice of a new one. I have thought numerous times about the confluence of events that have brought me to this place, and indeed, gratitude is the primary sentiment I feel and most wish to convey. Very first, thank you to Dylan Korczynskyj for taking a Skype call from a young and eager American girl, for taking that first gamble on me and for betting on me every day since. What I’ve learned from you goes far beyond, but importantly has included, good ecology. Thank you for sharing with me your family and the University of

Notre Dame Australia academic community. Andrew Grigg, you and Alcoa welcomed me warmly from the start, provided unconditional support for this project through funding, industry knowledge and help from the very best people (‘rock’ stars, in my view). Your knowledge has been indispensable, and you have taught me that a good research project is never really finished (not even this one). Max Bulsara (Notre Dame), thank you for your genuine, thoughtful and continuous support of my learning, and also for providing me with well-timed and crucial boosts to morale. I can truly say running Generalised Linear Models is my new favourite pastime, thanks to you. Importantly, as I was privileged to spend such a large portion of my research time in the Northern Jarrah Forest, I would like to acknowledge the Traditional Owners and custodians of the region—past, present and emerging:

The Pindjarup people of the Noongar nation. Thank you for sharing your land with me. My long hours in the field were mostly accompanied by true-crime podcasts and march flies, but

I must absolutely thank my many vastly under-appreciated field helpers; most notably ‘the

Cams’ (Cam Richardson and Cam Blackburn) of Alcoa. Cams, thanks for the long hours that you’ve both endured with me. Thanks for continuing to help me even after I stole your markers, ignited small fires and mistook door grease for hummus. Special thanks also to JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING v

Justine Barker, Rowan Beals, Jenna Blackwell, Randall Heaysman, Chris Johnstone, Greg

Mullins and Kylie Walmsley (all Alcoa). Josi Rosa (University of São Paolo), your presence on site and during my commute brought me incredible joy; you constantly make me ‘better’.

Tharcila Peixoto (Alcoa), Natasha Mikich, Beatrice Clifford and Eloise Ashton (Notre

Dame), thank you for your help with the seed traps. Your enthusiasm helped get me through.

Lachlan Warner (Notre Dame), thanks for taking on the germination project, for sharing the long drive with me and for giving me my very own marri seedlings (which did die, but that is not important). Joe Kinal (Department of Biodiversity, Conservation and Attractions), thank you for access to the seed traps. To my friends and colleagues at Murdoch University,

University of Western Australia and Kings Park (special mention to Laura Skates and Lauren

Svejcar), thank you for taking me under your collective wings and reminding me that what I was experiencing was ‘normal’. Rachel Standish and Joe Fontaine (Murdoch University), thanks for welcoming me to the Terrestrial Ecology Research Group community. Ryan

Tangney (Kings Park Science), thank you for proposing the use of X-rays for my jarrah seed

(they weren’t small rocks after all). Kate O’Neill (Notre Dame), thank you for sewing hundreds of seed bags for me. To the Notre Dame and University of Portland (UP) study abroad and residences staff (especially Eddie Contreras, Peta Sanderson, Chelsae Currie,

Debbie Guy, Oz Taylor and Mez Badsha), each of you has been an incredible cheerleader— thank you for support and patience. To my Cleos kids (UP), all 108 of you, part of this story is yours too. You helped me make field equipment, studied with me into the early hours of the morning and mistakenly diagnosed me with liver failure when I developed a few bruises from carrying seed traps. Being your mum-friend and doing this research full-time was an incredible challenge, but an even more incredible reward. To all of my dear friends in both

America and Australia (too many of you to list), you all are so important to me. Thanks for listening, thanks for pushing me, thanks for catching me, thanks for your patience when I JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING vi didn’t call you back, and thank you for only sometimes making fun of me for petting plants or walking around with seed capsules in my hand. Thanks to my mum cheer squad, Teresa

Newell and Gayleen Webb— this last year has been made possible through your friendship and support. To my fellow Notre Dame HDR friends (special mention to Jane Minson, Toni

Church and Alex Wallis), you all have inspired me every day and have been incredible travel companions on this journey. Tara Holt (UP), thanks for the crab lab. So much of who I am now is a direct outgrowth of my time with you. I have worn your ring everyday of this process and carry you and your fire with me always. To my Australian families (the Bakers,

Dyces and Gannons) thank you for feeding, housing and caring for me. I am so grateful to be your ‘American daughter’ and would not have made it without your incredible hospitality.

Thank you for giving me a place to relax and laugh, and even sometimes to do some writing.

A very, very special ‘thank you’ to the Dyce family—Dennis, Kay, Emma and Sarah—for opening your home to me for this last year; I am forever grateful to be your D3. Knowing I always had a gorgeous family cheering me on and a beautiful home to rest in has meant so much to me. Last, thank you to my family. Mum, Dad, Dan and Shea, thank you for letting me have this adventure, even when two years abroad became three, and three became four, and four became… (?) Thanks for telling me that I could do it, and that nothing worth doing is ever easy. Thanks for visiting me at crucial moments and giving me the biggest hugs, laughs and space for tears. I know you all sacrificed, and it was all worth it.

Thank you to the University of Notre Dame Australia for support through the Vice

Chancellor’s International Fee Remission Research Scholarship. This thesis has been copyedited by Elite Editing, and editorial intervention was restricted to Standards D and E of the Australian Standards for Editing Practice. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING vii

(From top left corner) Cameron Richardson, Cameron Blackburn, Greg Mullins, Josi Rosa, Eloise Ashton, Tharcila Peixoto, Beatrice Clifford, Lachlan Warner, Natasha Mikich, Andrew Grigg and Dylan Korczynskyj, Photos T. White-Toney.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING viii

List of Publications

White-Toney, T., Korczynskyj, D., Grigg, A. and Bulsara, M. (2019). Variable tree

establishment in bauxite mine restoration in south-west Australia is linked to rainfall

distribution, seasonal temperatures, and seed rain. Ecological Management &

Restoration 20(3), 266-270. doi:10.1111/emr.12389.

Conference Abstracts

White-Toney et al. (2016). ‘What are the drivers of establishment success for the overstorey

species following bauxite mining in the northern jarrah forest?’. Paper presented at

the Ecological Society of Australia meeting in Fremantle, WA.

White-Toney et al. (2017). ‘When things don’t add up: Unexpected establishment patterns of

dominant tress in jarrah forest rehabilitation’. Paper presented at the Ecological

Society of Australia meeting in Hunter Valley, NSW.

White-Toney et al. (2017). ‘When things don’t add up: Unexpected establishment patterns of

dominant tress in jarrah forest rehabilitation’. Paper presented at the University of

Notre Dame Higher Degree by Research Conference meeting in Fremantle, WA.

White-Toney et al. (2017). ‘When A+B = C sometimes and A + B = D other times: Variable

establishment of jarrah and marri following bauxite mining in the Northern Jarrah

Forest’. Paper presented at the Three-Minute Thesis competition in Fremantle, WA.

White-Toney et al. (2018). ‘Variable establishment of jarrah (Eucalyptus marginata) and

marri (Corymbia calophylla) following bauxite mining’. Paper presented at Kings

Park Science in Perth, WA.

White-Toney et al. (2018). ‘PhD student research project overview: Establishment of jarrah

and marri after bauxite mining’. Paper presented at the Goldfield Environmental

Management Group meeting in Kalgoorlie, WA. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING ix

White-Toney et al. (2018). ‘Variable densities of jarrah (Eucalyptus marginata) and marri

(Corymbia calophylla) following bauxite mining’. Paper presented at the 2018

Revegetation Industry Association of WA Seminar in Mandurah, WA.

Other Media

White-Toney, T. (7 September 2018). The establishment of jarrah and marri following

bauxite mining. Botanical Bulletin 7, 3.

White-Toney, T. (2018). The land of giants: Giant mines and giant trees. High & Cliff—

Podcast Collection 1(6). Retrieved from

https://researchonline.nd.edu.au/highandcliff/6.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING x

Table of Contents

Declaration of Authorship ...... ii Abstract ...... iii Acknowledgements ...... iv List of Publications ...... viii List of Tables ...... xii List of Figures ...... xiv Chapter 1: General Introduction ...... 1 1.1 Scope ...... 2 1.2 Introduction to Mine Site Restoration ...... 3 1.3 Ecology of the Northern Jarrah Forest ...... 5 1.3.1 Location ...... 5 1.3.2 Vegetation ...... 6 1.3.3 Climate and soil ...... 6 1.3.4 Fire in the Northern Jarrah Forest ...... 9 1.4 Land Uses and Management of the Northern Jarrah Forest ...... 9 1.4.1 History and impact of the timber industry ...... 9 1.4.2 Phytophthora cinnamomi and dieback disease ...... 10 1.4.3 Bauxite mining and mine site restoration ...... 11 1.5 Seedling Recruitment ...... 13 1.6 The Study Species: Jarrah (Eucalyptus marginata (D. Don Ex Sm.)) and Marri (Corymbia calophylla (Lindl). K.D. Hill & L.A.S. Johnson)) ...... 15 1.6.1 Species introduction ...... 15 1.6.2 Flowering and seed production ...... 17 1.6.3 Natural seedling recruitment ...... 18 1.6.4 Recruitment of seedlings on Alcoa’s jarrah forest restoration sites ...... 19 1.7 Research Structure ...... 21 Chapter 2: Long-term Seedling Density Monitoring ...... 24 2.1 Chapter Introduction ...... 25 2.2 Variable Tree Establishment in Bauxite Mine Restoration in South-west Australia is Linked to Rainfall Distribution, Seasonal Temperatures and Seed Rain (publication manuscript) ...... 26 2.2.1 Summary ...... 26 2.2.2 Introduction ...... 27 2.2.3 Methods ...... 28 2.2.4 Results and discussion ...... 32 2.3 Supplementary Material ...... 36 Chapter 3: Seed Sources ...... 42 3.1 Chapter Overview and General Introduction (Eucalypt Seed Sources) ...... 43 3.2 Tree Seedfall into Jarrah Forest Restoration Sites (draft manuscript) ...... 45 3.2.1 Abstract ...... 45 3.2.2 Introduction ...... 46 3.2.3 Methods ...... 49 3.3 Results ...... 54 3.4 Discussion ...... 58 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING xi

Chapter 4: Suitability of Establishment Microsites ...... 64 4.1 Introduction ...... 65 4.1.1 Scope ...... 65 4.1.2 Background ...... 66 4.2 Methods ...... 71 4.2.1 Study site ...... 71 4.2.2 Experiment A: Effect of ripline microtopography, soil chemistry and debris cover on seedling emergence and establishment ...... 73 4.2.3 Experiment B: Difference in seedling emergence and establishment by ripline microtopography ...... 75 4.2.4 Experiment C: Difference in seedling emergence by substrate type ...... 76 4.2.5 Data analysis ...... 77 4.3 Results ...... 77 4.3.1 Climate ...... 77 4.3.2 Experiment A: Effect of ripline microtopography, soil chemistry and debris cover on seedling emergence and establishment ...... 78 4.3.3 Experiment B: Difference in seedling emergence and establishment by ripline microtopography ...... 80 4.3.4 Experiment C: Difference in seedling emergence by substrate type ...... 84 4.4 Discussion ...... 86 Chapter 5: Synthesis ...... 93 5.1 Rationale ...... 94 5.1.1 Flowering to seed release ...... 97 5.1.2 Restoration practices ...... 98 5.1.3 Fauna ...... 100 5.1.4 Climate ...... 100 5.1.5 Seed to seedlings ...... 102 5.2 Management Applications ...... 103 References ...... 106 Appendices ...... 129

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING xii

List of Tables

Table 2.1 Explanatory variables included in the saturated models for jarrah and marri establishment density (seedlings/ha). Final models included only significant variables (p<0.05); variables denoted “NS” and all interaction terms (not listed) did not significantly improve the model. The β coefficients (standard error in parentheses) for the final models indicate the magnitude and direction (+ or –) of the effect of an increase in one unit of each explanatory variable on seedling density. The observed range of each variable is provided to assist in assessing the potential magnitude of effect on seedling density. WS wet season, DS dry season – see text for definitions. . 29 Table 2.2 Overall effect size of linear regressions for jarrah and marri models, and the ranked component effect sizes for each explanatory variable. Upper and lower 95% confidence intervals (CI), and the direction of effect on density, from GLMs are also shown. WS wet season, DS dry season – see text for definitions and model selection...... 36 Table 3.1 Mean and standard error (SE) of jarrah seedfall (seeds/ha/year) in forest and restoration sites transects (n=28) illustrating variability in adjacent forest traps...... 55 Table 3.2 Explanatory variables included in the saturated models for jarrah seedfall in restoration and forest sites (in seeds/ha/year). Final models only included significant variables (p<0.05). ‘NS’ indicates non-significant variables that were included in the saturated model, and ‘NA’ marks variable that were not included. The β coefficients (standard error in parentheses) for the final models indicate the magnitude and direction (+ or –) of the effect of an increase in one unit of each explanatory variable on the amount of seedfall. The observed range of each variable is provided to assist an assessment of the potential magnitude of the effect on seedfall...... 57 Table 3.3 Restoration site geometry (area and perimeter) and its impact on the area and amount of seed supplied from canopy-stored seed in jarrah forest stands adjacent to three restoration sites...... 60 Table 4.1. Summary of climate data comparing temperature averages and rainfall evenness from 1998 to 2015 with records for 2016 (Experiment A) and 2017 (Experiment B) restoration seasons. The wet season was differentiated from the dry season as the period in each year when rainfall was greater than twice the mean temperature (see Chapter 2). Rainfall evenness from 0 (uneven rainfall, all on one day) to 1 (even rainfall across all days) was calculated using a modified measure of species evenness JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING xiii

(Simpson’s Diversity index; Bronikowski & Webb, 1996; Standish et al., 2015; Chapter 2). Evenness was calculated for periods between 0 and 30 days, related to germination and emergence (column 2), and over the duration of the total wet season (column 3)...... 73 Table 4.2 A complete list of explanatory variables (left) investigated for potential use in the Generalised Linear Models to account for cumulative jarrah and marri seedling density (seedlings/ha). The coefficients (β) and standard errors for the final models (right) indicate the extent and direction of the effect of each explanatory variable on seedling density, i.e. the increase or decrease in seedling density (seedlings/ha) as a consequence of an increase in one unit of an explanatory variable. Variables that do not show a coefficient are those that did not significantly (p>0.05) improve the model (‘NS’). The significant explanatory variables explained 32% of the variability in marri seedling density at establishment (pseudo R2 = 0.316)...... 80 Table 4.3 Chemical composition of the four typical restoration site soil types, which were assessed for their effect on the emergence of jarrah and marri seedlings. One sample from each substrate type was submitted for analysis, therefore values are totals...... 86

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING xiv

List of Figures

Figure 1.1 The Northern Jarrah Forest (dark green vegetation, a) is located south of Perth, Western Australia (a, b). Pink areas illustrate Alcoa’s past and present mining and restoration sites near Dwellingup, WA (a), where temperature data was accessed for this study. The basemap was retrieved from GoogleMaps in QGIS ...... 5 Figure 1.2 Mean monthly rainfall (mm) and air temperatures for Dwellingup, WA (32°72oS 116°07oE) between 1935 and 2018 (Bureau of Meteorology, 2018)...... 7 Figure 1.3 Recently emerged jarrah (left) and marri (right) seedlings from surface-sown seed on a restoration site. The cotyledons (green) have protruded from the seed (black). Scales provided in white near the seed. Photo T. White-Toney...... 14 Figure 1.4 Key physical characteristics used to distinguish between jarrah and marri trees including bark (a; jarrah, left), underside of leaves (b; jarrah, left), seeds (c; jarrah, top) and capsules (d; jarrah, left). Note: marri capsules (d) are immature (green) in this image (and brown at maturity). Photographs T. White-Toney...... 16 Figure 2.1 The study area is located in south west Australia (a) and comprised of restoration sites (orange areas, b & c) associated with Alcoa’s Huntly and Willowdale mines located within the Northern Jarrah Forest (dark green areas, b). The position of the Dwellingup climate station used to obtain daily air temperature, and rain gauge locations (c; blue squares) are shown...... 28 Figure 2.2 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given seeding rates, when all significant model parameters (Table 2.1) are held at their mean value. Y-axis values differ because of lower seeding rates for marri compared to jarrah...... 37 Figure 2.3 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given forest perimeter, when all significant model parameters (Table 2.1) are held at their mean value...... 38 Figure 2.4. Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given fertiliser application rates, when all significant model parameters (Table 2.1) are held at their mean value...... 38 Figure 2.5 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given rain evenness at 30 days, when all significant model parameters (Table 2.1) are held at their mean value...... 39 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING xv

Figure 2.6 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given full wet season (WS) evenness, when all significant model parameters (Table 2.1) are held at their mean value...... 39 Figure 2.7. Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given dry season (DS) average maximum temperatures, when all significant model parameters (Table 2.1) are held at their mean value...... 40 Figure 2.8 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given wet season (WS) lowest temperatures, when all significant model parameters (Table 2.1) are held at their mean value...... 40 Figure 3.1 A seed trap transect comprised five traps within the restoration site (shown) and one trap in the forest (not shown). Seed traps were metal funnels (1 m2) supported by fence droppers (1 m height) with a bag for seed collection. Photo T. White-Toney. .. 52 Figure 3.2 Jarrah seedfall (seeds/ha/year) for jarrah forest (dark grey; n=28) and restoration site (within 12.7 m of the forest edge; light grey; n=28) seed traps. Median and lower (Q1) and upper (Q3) quartiles represent observations inside the nine to 91 percentile range and one outlier (black dot) with 540,000 seeds/ha/year seedfall ...... 54 Figure 3.3 Total monthly jarrah seedfall (seeds/ha) in forest (dark grey) and restoration site (n=28; light grey) traps (0 – 12.7 m of forest edge; n=28)...... 55 Figure 3.4 Total jarrah seedfall (seeds/ha/year) from forest edge into all restoration sites (n=140)...... 56 Figure 3.5 Three jarrah forest restoration sites (names for identification to the left) illustrating total perimeter (orange) and forested perimeter (light green) The estimate of seeded area for each site was calculated from the forested perimeter and the maximum seedfall distance of 12.7 m. The basemap was retrieved from GoogleMaps in QGIS...... 60 Figure 4.1 A post-mining restoration site, demonstrating the typical surface characteristics following ripping. The direction of the riplines is perpendicular to the downslope of the landscape. Clay accumulates in the furrows (e.g. centre middle) and banks and crests (e.g. bottom right) are dominated by variable amounts of other substrates including topsoil, overburden and occasionally pitfloor. The surface can also include cover by woody debris, leaf litter and caprock...... 70 Figure 4.2 Representation of a 25 m2 plot with four riplines. The percentage of the total area that was furrow (yellow) and bank-crests (white) was calculated from the mean ripline width (black dotted lines) and length (dark blue line)...... 75 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING xvi

Figure 4.3 Percentage mortality of jarrah (n=174) and marri (n=124) seedlings over 196 days prior to establishment, in the three microtopographic locations and across all 60 plots. Mortality is presented as Kaplan–Meier failure curves. Microtopography locations for seedling establishment were furrow (orange), bank (blue) and crest (green). There was no effect of establishment location on the mortality of tree seedlings (p=0.19), irrespective of species ...... 79 Figure 4.4 Mean (± SD) jarrah (n=174) and marri (n=124) seedling density (per m2) at establishment in the furrow and bank-crest microtopography locations across all 60 plots. Seedlings per m2 was calculated from the proportion of the 60 25-m2 area restoration sites that were furrow...... 79 Figure 4.5 Cumulative seedling emergence for jarrah (n=121) and marri (n=32), across five plots, between April 2017 and February 2018. Seed was applied on 20 April 2017 (autumn) and first emergence was in week five (autumn)...... 81 Figure 4.6 Proportion (in brackets) of jarrah (n=121) and marri (n= 32) seedlings that emerged in a location different (blue) from the location of seed application (yellow) across five plots. Bank (mid-slope) locations were considered anywhere between discrete crest and furrow locations, differentiated by substrate type and texture...... 82 Figure 4.7 Percentage mortality of jarrah (n=121) and marri (n=32) seedlings over 377 days prior to establishment, in the three microtopographic locations across five plots. Mortality is presented as Kaplan–Meier failure curves. Microtopography locations for seedling emergence were furrow (orange), bank (blue) and crest (green). There was no effect of emergence location on the mortality of tree seedlings (p=0.37), irrespective of species...... 83 Figure 4.8 Percentage mortality of jarrah (n=121) and marri (n=32) seedlings over 377 days prior to establishment and five plots. Mortality is presented as Kaplan–Meier failure curves. There was no effect of establishment location on the mortality of tree seedlings (p=0.92), irrespective of species...... 84 Figure 4.9 Mean total (± SE) emergence of jarrah and marri seedlings on four substrate types (n=4). No values were significantly different (i.e. p>0.05)...... 85 Figure 4.10 Cumulative mean (± SE) number of emergent jarrah and marri seedlings on four substrate types over time (topsoil, green; overburden, yellow; gravel, orange; and pit floor, blue; n = 4 four per substrate type)...... 85 Figure 5.1 Summary of factors effecting jarrah and marri seedling establishment and their contribution to establishment limitation (A – E). Single-headed arrows indicate one JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING xvii

way- interactions (causation) and double-headed arrows represent a two-way relationship between factors. Relationships identified in this study (blue) and in others (grey) are explained in the writing adjacent to arrows. The broad relationship between climate and fauna is indicated with an arrow and dotted line without description...... 96

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING xviii

Man, his stock and his fire stick are the only serious dangers that jarrah forests have to face … Its extraordinary capacity of surviving has enabled this species to persist when a less hardy tree would have long ago been wiped out

———Lane-Poole (n.d.; in Abbott & Loneragan 1986)

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING xix

Jarrah leaves, flowers and bark. Photo T. White-Toney. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 1

Chapter 1: General

Introduction

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 2

1.1 Scope

This study investigated ecological factors, including climate and restoration practices, that affect jarrah (Eucalyptus marginata) and marri (Corymbia calophylla) seedling densities at establishment, following bauxite mining in the Northern Jarrah Forest. Variability in seedling densities at establishment has been observed in these co-dominant, overstorey species as part of post-mining restoration monitoring (Koch, 2007a; Norman & Koch, 2009).

Target stand density (total, maximum and minimum) for restored mine sites has been continually reduced over the last 40 years to better match the typical stand density of surrounding forest areas (Koch & Samsa, 2007). Achieving restoration target requires understanding the causes of variability in seedling establishment, especially as jarrah emergence from seed in restoration sites is highly variable, at 2–38% (McChesney, Koch &

Bell, 1995). This variability requires a better understanding of the factors that influence seedling density in order to achieve stand density targets for restoration of the Northern

Jarrah Forest.

Some of this variability was explained in a study on restoration sites, which cited as responsible ambient maximum and minimum temperatures (Norman & Koch, 2009) and, related to seedling recruitment for increased timber production, amount of seedfall (Cargill, van Etten & Stock, 2019). The current study investigated the role of seed supply and microsite conditions in variable seedling establishment. These two broad categories of factors—amount of seed supply and number of preferred microsites—limit seedling establishment (Eriksson & Ehrlén, 1992), and form the framework selected for this study.

Most previous work on tree seedling establishment in jarrah forest has studied jarrah alone, or has compared responses of the two species as adult trees, and there is a need for analysis of ecological factors affecting marri separately from those affecting jarrah (e.g. Szota, Farrell,

Koch, Lambers, & Venekelaas, 2011). Co-dominant species such as jarrah and marri can JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 3 cohabit based on separate niches because the species are differentiated based on characteristics of suitable establishment sites, susceptibility to parasites or predators and aspects of germination and dormancy (Harper, Clatworthy, McNaughton & Sagar, 1961). The variability in jarrah and marri seedling densities highlights the need to better understand the separate responses of these species to ecological cues. Based on the framework of establishment limitation and previous work on jarrah forest restoration sites, the three questions informing this study of jarrah and marri seedlings were:

1. Which ecological factors, including restoration practices, are most important for

explaining the variability in seedling density at establishment observed in long-term

monitoring data?

2. To what extent is seed entering the mine restoration site from each of broadcast

sowing and seedfall from the forest canopy? And what is the effect of these seed

sources on seedling density at establishment?

3. What physical and chemical microsite characteristics best explain seedling survival,

mortality and ultimately density at establishment?

The ecology, current and historic land use and management, seedling recruitment process and biology of the focal species (jarrah and marri) is presented and discussed here to provide a better understanding of this study’s three objectives and methodologies.

1.2 Introduction to Mine Site Restoration

Ecological restoration is defined as the return of native biota and ecosystems that were present before disturbance; here, mining. The best approach is to return vegetation following disturbance, and is used when soils and environmental conditions remain stable and there are reference (undisturbed) areas of natural vegetation available (Lamb, Erskine, &

Fletcher, 2015). Common biophysical challenges of mine site restoration (according to Lamb et al., 2015) include prologued topsoil storage and minimal availability (e.g. Rokich, Dixon, JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 4

Sivasithamparam, & Meney, 2000), patchy or failed vegetation return due to poor soil condition (e.g. Cummings, Reid, Davies, & Grant, 2005; Morrison, Lamb, & Hundloe, 2005), climate conditions and water stress (Morrison et al., 2005), invasive exotic species (Erskine

& Fletcher, 2013), unstable landforms (Williams, 2001) and toxic materials remaining present in the soil surfaces (Gore, Preston, & Fryirs, 2007); in addition, wildlife recovery requires specific slow-forming characteristics (Craig et al., 2010). Governments and industries involved in restoration generate criteria that indicate the success of restoration and determine any potential liabilities related to the return of the land to pre-disturbance purposes

(Elliot, Gardner, Allen, & Butcher, 1996). These ‘completion criteria’ represent biophysical milestones to assess when the restored area has reached the desired sustainable state (Grant,

2006; Grant, Duggin, Meek, & Lord, 2001) and has the capacity to withstand normal disturbances to the area (e.g. grazing or fire; Environmental Protection Authority, 1995).

Restoration approaches that are successful in other ecosystems do not always apply to

Mediterranean-type ecosystems. Globally, these ecosystems are threatened by habitat loss and fragmentation, changes in land management (e.g. forestry and prescribed burning), hydrological changes, and the introduction of pests, pesticides and weeds (Close et al., 2009;

Linder et al., 2010; Ozkan et al., 2009). These challenges therefore require the mimicking of natural disturbance processes (i.e. fire) on restoration sites prior to the application of seeds or seedlings to increase seedling survival (Ruthrof et al., 2010). The landscapes of

Mediterranean-type ecosystems have been, and continue to be, altered, and hence rely on successful restoration practices to conserve endemic plant species. This thesis explores the mine site restoration in the Northern Jarrah Forest of south-western Australia, which includes these government regulated completion criteria (stand density; Section 1.6.4), the specific challenges of restoration relevant to Mediterranean-type ecosystems (Section 1.4.3) and echoes the global challenges to forest restoration. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 5

1.3 Ecology of the Northern Jarrah Forest

1.3.1 Location

The Northern Jarrah Forest occurs on the Darling Plateau in south-western Australia.

Jarrah dominates the forest from 300 km north of the state’s capital, Perth (Figure 1.1) to the south coast near Esperance (Churchill, 1968). This physiographic region is overlain by a metamorphic rock layer, including granite, and has an average elevation of 300 m (Williams

& Mitchell, 2001). The jarrah forest is open, dry and sclerophyllous (Abbott & Loneragan,

1986), and occurs in areas with more than 600 mm of annual rainfall and total summer evaporation of 500–770 mm (Gentilli, 1989; Koch & Samsa, 2007).

a) b)

Figure 1.1 The Northern Jarrah Forest (dark green vegetation, a) is located south of Perth, Western Australia (a, b). Pink areas illustrate past and present mining areas and restoration sites near Dwellingup, WA (a), where temperature data was accessed for this study. The basemap was retrieved from GoogleMaps in QGIS.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 6

1.3.2 Vegetation

In fertile, low, swampy depressions, jarrah’s dominance is replaced by flooded gum

(E. rudis), bullich (E. megacarpa) and Swan River blackbutt (E. patens). Jarrah forest is also replaced by karri forest (E. diversicolor) to the south and wandoo forest (E. wandoo, E. accendens) to the east (Dell & Havel, 1989). Marri is the most common co-dominant tree species, with a typical stand ratio of three-to-one, jarrah-to-marri (Koch & Samsa, 2007). The high biodiversity and differing presence of mid and upper-storey species underlies the classification of jarrah forest into 21 unique forest types based on key indicator species (Bell

& Heddle, 1989; Havel 1975), highlighting the high spatial heterogeneity of jarrah forest

(Calver & Wardell-Johnson, 2004). The Northern Jarrah Forest region also contains over

0.5% of the world’s plant richness and has lost at least 70% of its primary vegetation, classifying it as a world biodiversity hotspot (Myers, Mittermeier, da Fonseca, Kent, &

Mittermeier, 2000). Geomorphologic and relative climatic stability, a lack of geological rejuvenation and isolation of the region has favoured diversification and endemism, leading to floral biodiversity (Hopper & Gioia, 2004). The majority of this floral biodiversity is within the understorey, and includes species in the Anthericaceae, Apiaceae, Asteraceae,

Cyperaceae, Dasypogonaceae, Epacridaceae, Fabaceae, Orchidaceae, Proteaceae and

Restionaceae families (Gardner & Bell, 2007). This region is, therefore, of importance for conservation and highlights the significance of the restoration-focused applications identified in this study.

1.3.3 Climate and soil

The soil of the Northern Jarrah Forest is nutrient poor, which has forced endemic flora to develop strategies to either acquire these nutrients or do without them (e.g. Lambers et al.,

2003). This type of response to soil characteristics, and associated adaptations, contributes to the high floral diversity in Mediterranean-type ecosystems such as the jarrah forest (e.g. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 7

Wassen, Venterink, Lapshina, & Tanneberger, 2005). The Northern Jarrah Forest soils comprise acidic, iron-stone gravels and sands with an underlying clay layer (Gardner & Bell,

2007; Williams & Mitchell, 2001). Weathering of parent rock during the Cretaceous and

Oligocene periods led to the formation of the duricrust and lateritic profiles of alumina, quartz and iron oxides underneath clay (Gilkes, Scholtz, & Dimmock, 1973). The infertile soil presents challenges to seedling survival in this ecosystem, along with pronounced dry and wet seasons (Figure 1.2) where seedlings are affected seasonally by limited water availability (e.g. Szota et al., 2011). The Northern Jarrah Forest receives between 900 and

1,300 mm of rainfall annually, 60% of which falls between June and August (Gentilli, 1989).

The dry season can last for 4–6 months and annual mean air temperatures range from 13 to

25°C (Gardner & Bell, 2007).

250 35

30 C)

200 o 25

150 20

100 15 10 Average (mm) rainfall Average 50 5 ( air temperature Average

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

Figure 1.2 Mean monthly rainfall (mm) and air temperatures for Dwellingup, WA (32°72oS 116°07oE) between 1935 and 2018 (Bureau of Meteorology [BOM], 2018).

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 8

1.3.3.1 Changing climate

South-western Australia has experienced increasing drought conditions since the

1970s (Bates, Hope, Ryan, Smith & Charles, 2008), which is expected to produce a shift in the jarrah forest stand favouring more drought-tolerant plants and having more variable forest regeneration (Batini, 2007). Mean temperatures in the south-west of Australia, including the

Northern Jarrah Forest, have increased by 0.15°C per decade in all seasons except summer since 1975, and summer temperatures have increased by 0.1°C per decade (Bates et al.,

2008). In the same period (1975–2018), wet season rainfall has decreased by 20% (BOM,

2018). This drying and warming trend is expected to continue, with up to 40% rainfall reduction and up to a five-degree Celsius increase in mean annual temperature estimated by

2070 (BOM, 2007). In Mediterranean-type forests, endemic tree species have declined in health over the last 30 years in response to extreme weather events and land fragmentation

(e.g. Brouwers, Matusick, Ruthrof, Lyons, & Hardy, 2013). During the hottest drought on record (summer 2010), jarrah trees were found to be more susceptible to drought than marri trees. Jarrah and marri crowns were noted to experience discolouration and death within two weeks, and there was a higher proportion of jarrah crowns dying than marri crowns

(Matusick, Ruthrof, Brouwers, Dell, & Hardy, 2013), a lower probability of survival of jarrah than marri trees at 16 months post-drought (Ruthrof, Matusick, & Hardy, 2015) and higher mean mortality of jarrah trees in severe drought-affected forest zones (19%) compared with marri (seven percent; Steel, Fontaine, Ruthrof, Burgess & Hardy, 2019). The higher tolerance of marri to drought may in part be attributed to its osmotic adjustment and earlier stomatal closure, which also enables marri to better survive in low-quality (i.e. lower density and basal areas) forest stands (Szota et al., 2011). Knowledge of the measured effect of drought conditions on mature jarrah forest trees supports the inclusion of the effect of climate in all ecological studies in jarrah forest and has important implications for forest management. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 9

1.3.4 Fire in the Northern Jarrah Forest

The natural occurrence of fire throughout the region has driven endemic species to evolve strategies to cope. Physiological and morphological strategies exhibited include survival mechanisms along a continuum from obligate seeding to resprouting, post-fire flowering and geophytes avoiding the direct impact of fire (Burrows & Wardell-Johnson,

2003; Gill, Groves, & Noble, 1981). Jarrah persists in this landscape because of its ability to resprout following fire and recruit new individuals during the inter-fire period. Like jarrah, marri shows adaptations for surviving fire, including the development of protected lignotuberous growth for energy storage and the ability to repsrout from epicormic buds beneath the thick fibrous bark that covers the trunk and stems (Abbott & Lonergan, 1986;

Burrows & Wardell-Johnson, 2003). Firedistributes ash to the forest floor, which improves the seedbed and consequently seedling establishment (Cargill et al., 2019; Florence, 2004;

Ruthrof et al., 2010). Specifically, this ‘ashbed effect’ on seedling establishment includes increases in nitrogen and phosphorus availability for seedlings, increases in water availability

(Loneragan & Loneragan, 1964), displacement of pathogens (Renbuss, Chilvers, &

Prior,1973) and reduced competition from surrounding vegetation (Bond & van Wilgen,

1996). As in other Mediterranean-type ecosystems, the combination of these conditions provides a window of opportunity for enhanced recruitment during the first winter post-fire

(Abbott & Loneragan, 1986). The interaction between fire and recruitment of jarrah and marri seedlings is explained in Section 1.6.2.

1.4 Land Uses and Management of the Northern Jarrah Forest

1.4.1 History and impact of the timber industry

Jarrah has been logged for timber since 1833 when the first sawmill began operation.

By 1920, 50% of the forest canopy had been removed (Wallace, 1965; Wardell-Johnson,

Calver, Burrows & Di Virgilio, 2015), permanently changing the jarrah forest landscape JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 10

(Calver & Wardell-Johnson, 2004). Timber extraction for harvesting currently occurs and is part of mining practices in the jarrah forest (Koch, 2007a); future timber production continues to be a land use factored into post-mining restoration outcomes (Gardner & Bell, 2007). With the thorough and intensive removal of old-growth jarrah, there are few untouched areas of forest that represent pre-colonial forest structure, richness and function (Mills, 1989). The removal of large jarrah trees from the forest opened up the canopy, allowing non-targeted marri and mid-storey tree species to increase in number (Calver & Wardell-Johnson, 2004).

Logging also affects stand density and basal area because of coppicing from the large jarrah stumps left behind (Abbott & Loneragan, 1986). These changes to jarrah forest from logging are apparent in forested areas near mining activity and are reflected in forest characteristics observed in this study. Importantly, also, work to improve jarrah silviculture has provided important information on jarrah seedling recruitment that has informed and advanced aspects of this study (e.g. Cargill, 2014).

1.4.2 Phytophthora cinnamomi and dieback disease

Plant dieback and death caused by the soil pathogen Phytophthora cinnamomi was first observed in the Northern Jarrah Forest during wet seasons in the mid-1940s and 1960s affecting jarrah trees growing in low-stocking, poor-quality forest stands (Davison, 2014).

The pathogen was likely to have been first introduced to WA at or soon after European settlement of the area in 1828. The soil pathogen affects 40% of jarrah forest species, including those in the families/subfamilies Proteaceae, Myrtaceae, Ericaceae,

Xanthorrhoeoideae and Dilleniaceae (Podger, 1972; Shearer & Tippett, 1989), and has a higher presence in forest areas that have been marked by significant disturbance (e.g. logging and fire; Calver & Wardell-Johnson, 2004; Shearer & Tippett, 1989). P. cinnamomi infects hosts as motile zoospores produced from sporangia in the mycelium of previously infected roots growing in moist, warm and aerobic soil (Hardham & Blackman, 2018). The zoospores JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 11 can swim short distances in soil water and rely on the movement of infected soil for dispersal over long distances (e.g. by earth-moving equipment or large animals; Shearer & Tippett,

1989). The identification and understanding of the pathogen’s lifecycle has informed management actions that minimise spread from infected areas into uninfected areas (Koch,

2007a). Restricting movement of the soil pathogen has been a focus of forest management since the 1970s, including post-mining restoration practices (Davison, 2018).

1.4.3 Bauxite mining and mine site restoration

The commercial potential for mineral extraction in the Northern Jarrah Forest was first noted in 1957, with Alcoa World Alumina (hereafter, ‘Alcoa’) beginning its mining operations in 1963 and presently mining and restoring approximately 550 ha/year (Bell &

Hobbs, 2007). Worsley Alumina also mines in the area, with operations commencing in 1974

(Gardner & Stoneman, 2003). Bauxite deposits in the Northern Jarrah Forest occur in one to

100 hectare pockets (typically 10–20 ha); therefore, mining regions of the Northern Jarrah

Forest are characterised by a mosaic of unmined forest, active mine sites and sites undergoing restoration (Koch, 2007a). Alcoa’s mining takes place in jarrah forest managed by the

Western Australian Government, which stipulates forest characteristics and land uses

(Conservation Commission, 2004; Koch & Samsa, 2007). There has been continual development of restoration practices overtime (i.e. proposed, trialled, monitored and updated) especially related to the acceptable stand density and species richness in a restored jarrah forest stand (Grant, 2006; Grant & Koch, 2007). Specifically, post-mine restoration goals have transitioning from the return of fast-growing exotics resistant to P. cinnamomi, to forest containing overly dense jarrah stands for timber production, to forest with a canopy dominated by both jarrah and marri (Grant & Koch, 2007). The current goal is to produce a restored forest that includes all native species, retains the ecosystem values of the unmined JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 12 jarrah forest and can be incorporated into management of adjacent forest. (Bell & Hobbs,

2007; Gardner & Bell, 2007).

The post-mining restoration approach taken influences the density, abundance and spatial distribution of seedlings, and thus the structure and composition of the resultant mature forest (Grant, 2006). The broad sequential stages of mining and restoration include pre-mining site surveys, mining, clearing and restoration (full details in Koch, 2007a). Pre- mining site surveys identify the grade of bauxite deposits (Koch, 2007a) and the presence of

P. cinnamomi to minimise the spread of the soil pathogen (Colquhoun & Kerp, 2007).

Surveys also describe species richness and abundance in the pre-mined forest, to match vegetation requirements for restoration (Koch, 2007a). These surveys then inform the selection of sites for mining, which later dictate the proximity of each restoration site to adjacent unmined forest. Following timber harvesting and clearing, forest topsoil is collected to be spread on a nearby site undergoing restoration. Of 168 established non-weed species on restoration sites, up to 70% originates from forest topsoil (Koch & Ward, 1994); however, in untreated forest topsoil, jarrah seed has been recorded at a low density of one seed/m2 (Koch,

Ward, Grant, & Ainsworth, 1996). Therefore, the replacement of forest topsoil is important for returning understorey plant species richness during restoration, although a seed mix is also sown to return species that do not regenerate from forest topsoil (Koch et al., 1996;

Koch, 2007a, Koch 2007b). Jarrah and marri seeds do not show dormancy (i.e. they are soft seeded; Baskin & Baskin, 2007) and have not been found accumulated in the soil seed bank

(Koch et al., 1996; Tacey & Glossop, 1989). Optimal restoration practices also include deep ripping of the mine pit floor to relieve compaction and increase water penetration to aid root establishment and prevent erosion (e.g. Gardner & Bell, 2007). The species richness of understorey species in restoration sites varies from 0.75 to 4.95 plants/m2 and 16% of this variability has been explained by topsoil handling and a further four percent by rainfall JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 13 evenness at 30 days and total pre-establishment rainfall (Standish et al., 2015). This suggests that standardised restoration practices contain some inherent variability in application, including the timing of specific restoration steps, quality of topsoil and sown seed, and availability of natural resources in the seedling microsite (Koch, 2007a; Standish et al.,

2015).

1.5 Seedling Recruitment

This study investigated seedling establishment by considering the developmental stages of a tree from seed to seedling. Here, the context of those stages, and other key terms, is defined for use throughout this study.

Seedling recruitment represents the interface between the sessile and dispersal phases of seed plants and requires seed to arrive at a suitable microsite, germinate and survive

(Eriksson & Ehrlén, 1992). A seed is a mature ovule containing embryo and nutrients inside a seed coat, and is the product of sexual reproduction in vascular plants (Vander Wall, Forget,

Lambert, & Hulme, 2004). Germination involves uptake of water by the seed (imbibition), embryo expansion, cell elongation and then radicle protrusion (Bewley 1997; Finch-Savage

& Leubner-Metzger, 2006). Successful germination is determined by the timing of seed maturity (including the seed’s physiological, morphological and physical states) and the environmental conditions experienced by the seed (Baskin & Baskin, 1998). Seeds germinate when triggered by seasonal cues or can remain in the seedbank until an event breaks dormancy, stimulating germination (e.g. fire or a change in light levels; Vander Wall et al.,

2004). There is often a discrepancy between seed germination success under laboratory conditions and in field studies, because of higher variability in field environmental conditions

(Hegarty, 1977). While germination marks seed viability and is the standard assessment point in laboratory investigations (International Seed Testing Association, 2019), field studies often report emergence, marked by the appearance of the cotyledon(s) (Figure 1.3). Emergence is JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 14 the first developmental stage after germination and is an important pre-condition for establishment and future survival (e.g. Huang, Han, Wu, & Wu, 2008). After emergence, the seedling produces its first true leaves and a root system to survive through the first summer drought, marking establishment (e.g. Kollman, 2008). Seedling establishment in

Mediterranean climate systems requires developing seedlings to cope with low soil water availability and high evaporative demand (Close, Ruthrof, Turner, Rokich, & Dixon, 2009).

Recruitment success is marked by seedling establishment and is constrained by limitation in the amount of seed supply and number of suitable microsites (Aicher, Larios, & Suding,

2011; Duncan, Diez, Sullivan, Wangen, & Miller, 2009; Eriksson & Ehrlén, 1992).

Ultimately, seedling establishment confirms the distribution and presence of persistent vegetation in mature forests (Tilman, 1993). Our understanding of the life phases of jarrah and marri, with a particular focus on reproduction to establishment is explored in Section 1.6.

4mm

1cm

Figure 1.3 Recently emerged jarrah (left) and marri (right) seedlings from surface-sown seed on a restoration site. The cotyledons (green) have protruded from the seed (black). Scales provided in white near the seed. Photo T. White-Toney.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 15

1.6 The Study Species: Jarrah (Eucalyptus marginata (D. Don Ex Sm.)) and Marri

(Corymbia calophylla (Lindl). K.D. Hill & L.A.S. Johnson))

1.6.1 Species introduction

Jarrah and marri are the co-dominant tree canopy species throughout the Northern

Jarrah Forest (see Section 1.3.2). The majority of studies on trees in the Northern Jarrah

Forest have focused on jarrah because of its importance as a timber resource (e.g. Bradshaw,

1999) and because marri constitutes a lower proportion of the forest stand (usually 20–40%;

Szota et al., 2011). Recent work has highlighted differences in the responses of adult jarrah and marri trees to drought conditions in the region, related to differences in physiology of the species (Section 1.2.1.2; Szota et al. 2011). The two species grow to a similar height, usually not exceeding 40 m (Abbott & Loneragan, 1986) and physical differences between them are illustrated in Figure 1.4.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 16

marri

a jarrah

b

c

d

Figure 1.4 Key physical characteristics used to distinguish between jarrah and marri trees including bark (a; jarrah, left), underside of leaves (b; jarrah, left), seeds (c; jarrah, top) and capsules (d; jarrah, left). Note: marri capsules (d) are immature (green) in this image (and brown at maturity). Photos T. White-Toney.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 17

1.6.2 Flowering and seed production

Jarrah reproduces sporadically over both space and time, with an individual tree flowering once every four to seven years (Abbott & Loneragan, 1986). Jarrah flower buds initiate as early as December (early summer) and then open between September and

December (spring–summer) in the following year (Johnstone & Kirkby, 1992). Relatively little is known about the pollination ecology of jarrah (Abbott & Lonergan, 1986); however, observations have indicated a wide variety of invertebrate and vertebrate visitors to jarrah flowers. Likely invertebrate pollinators of jarrah include 83 species across five orders (Yates,

Hooper, & Taplin, 2005) and five vertebrate visitors: western spinebill (Acanthorhynchus superciliosus), brown honeyeater (Lichmera indistincta), New Holland honeyeater

(Phylidonyris novaehollandiae), purple-crowned lorikeet (Glossopsitta porphyrocephala) and honey possum (Tarsipes rostratus; Brown et al., 1997). After pollination, jarrah capsules form and mature, and can be retained in the canopy for several years prior to seed shed

(Abbott & Loneragan, 1986). Typically, seeds are shed between December and March

(summer–autumn), 24–27 months after the initiation of flower buds (Johnstone & Kirkby,

1992). This seed shedding occurs due to the natural drying of capsules with high ambient temperatures over summer; shedding may be induced sooner (late spring) via heating from fire (Abbott & Loneragan, 1986; Christensen, 1971; Cremer, 1965).

Sufficient jarrah seed supply to recruit jarrah seedlings is further complicated by the issue of seed viability. Jarrah capsules, like those of other Eucalyptus species, produces seed and ‘chaff’, which consists of infertile particles, ovulodes, aborted seed and sterile ovules

(Bohte & Drinnan, 2005; Carr & Carr, 1962). The proportion of viable seed per gram of commercially available seed mix is most often determined via destructive germination trials

(Boland, Brooker, & Turnbull, 1980), but criteria used for seed purity assessment are not always reported. Most often for jarrah, seed is distinguished from chaff on the basis of size JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 18

(>4 mm) and colour (black; Abbott & Loneragan, 1986; Cargill, van Etten, Whitford,

McCaw, & Stock, 2016). Variability in jarrah viability has been noted, but has not been considered a significant limiting factor in natural jarrah recruitment (e.g. Boland et al., 1980;

Cargill, 2014; Florence, 2004).

No published studies have investigated seedling recruitment in marri, but previous findings for jarrah have been extrapolated to marri based on the logic that other jarrah forest species germinate under similar conditions to jarrah (i.e. warm and wet; McChesney et al.,

1995; Standish et al., 2015). Marri trees usually flower once every five years and the reproductive cycle, from flower bud to seed shed, occurs over 17 months. Marri initiates flower buds in August (late winter), flowers in January (mid-summer), develops capsules between March and December, and releases seed the following January (summer; Johnstone

& Kirkby, 1992). While marri shows variability in wet mass, length and diameter of its fruit, seed number and dry seed mass (Cooper et al., 2003), there is no published work on marri seed viability. Some marri trees produce capsules containing no seed, classified as ‘male fruits’ (Carr, Carr, & Ross, 1971). These male fruits are produced both by single trees in conjunction with developed ovaries considered ‘normal’ for the species (Carr et al., 1971), and as the only type of fruit produced by an individual marri tree (Cooper et al., 2003). The cause of the phenomenon and its evolutionary significance are not well understood, but it is often attributed to ovule abortion late in development (Carr et al. 1971; Cooper et al., 2003).

1.6.3 Natural seedling recruitment

Following seed supply to the seedbed, germination of jarrah seed occurs under moist, warm conditions; usually during the wet winter months before the start of the summer drought (Abbott, 1984; McChesney et al., 1995). Jarrah seed germination has been reported to lie between 1.6% (Gill, Brooker, & Moore, 1992) and 90% (Abbott & Loneragan, 1986), based on factors such as timing of seed collection, seed storage conditions, seed treatment JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 19 and light and temperature. In forest conditions, jarrah germination is influenced by the soil water potential, atmospheric humidity, soil temperature, soil surface characteristics, shade, pathogens and allelopathic inhibition (Battaglia & Reid, 1993; Cargill et al., 2019). However, though they have some influence, these characteristics of a suitable seedbed for jarrah are generally not to be important for seedling establishment compared with the effect of the amount of seed supplied (Cargill et al., 2019), as is explored further in this study (Chapter 4).

Most mortality of jarrah seedlings occurs within the first year (prior to establishment; Abbott

& Loneragan, 1986). After establishment and under favourable conditions, jarrah seedlings typically grow one metre in height and one centimetre in diameter each year (Ward & Koch,

1995), dependent on environmental conditions including site quality, stocking density and fertiliser application (Koch & Samsa, 2007).

1.6.4 Recruitment of seedlings on Alcoa’s jarrah forest restoration sites

Alcoa has been mining and restoring the jarrah forest since the 1960s (Gardner &

Bell, 2007) and because of the importance of jarrah in the overstorey, jarrah seedling establishment has been the focus of much research (Koch & Samsa, 2007; Norman & Koch,

2009). One restoration benchmark for jarrah forest is the density of jarrah and marri seedlings, although the specific target densities have changed overtime (Koch, 2007a;

Norman & Koch, 2009). Original overstorey restoration practices included fast-growing, exotics and softwoods species (e.g. pine, cypress and eastern Australian eucalypts) resistant to the soil pathogen Phytophthora cinnamomi (Section 1.4.2). Later, with the change in thinking towards returning a native functional ecosystem, restoration practices were adapted to include jarrah and marri, the native dominant tree species (Gardner & Bell, 2007). The stand densities of jarrah and marri in unmined jarrah forest for jarrah and marri have been altered by timber harvesting (e.g. Calver & Wardell-Johnson, 2004), with an average combined density of jarrah and marri of 593 stems/ha (Koch & Samsa, 2007). In contrast, JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 20 jarrah and marri densities on restoration sites tend to lie between 300 and 5,000 stems/ha, with the highest historical upper target density of 2,500 stems/ha introduced in 2005 (Norman

& Koch, 2009). Current target density stipulates a combined density range of both overstorey species between 600 and 1,400 stems/ha (Alcoa, 2016).

The current target range was chosen based on the current understanding of the effect of tree density on forest characteristics, such as tree growth. High density forests increase competition, resulting in slow growth and high water use (Croton & Reed, 2007). Previous work on restored jarrah forest sites found the percentage of seedling establishment was negatively affected: 47% establishment at 10,000 stems/ha and 75% establishment at 625 stems/ha; Koch & Ward, 2005). The implementation of the upper threshold for target density

(1,400 stems/ha; Alcoa, 2016) increases seedling survival and better mimics densities observed in unmined forest. However, restoration practices that provide accurate control of seedling recruitment to this novel system is essential to achieve an acceptable target density.

Jarrah seedling emergence on post-mining restoration sites is highly variable, between 2% and 38% (Koch & Ward, 2005; McChesney et al., 1995; Norman & Koch, 2009), and is a critical development phase that affects future jarrah forest structure (Koch & Samsa, 2007).

Jarrah establishment, therefore, has been the focus of many studies (published and unpublished) that have guided restoration practices and informed the current restoration target (Alcoa, 2016; Norman & Koch, 2009). Alcoa’s routine tree seedling density monitoring takes place in autumn, when jarrah and marri seedling establishment is quantified

(Koch, 2007a). Variability in seedling densities continues to be observed despite the aforementioned changes to practices, and this observation led to the central objective of this study: to explain variability in jarrah and marri seedling establishment on jarrah forest restoration sites. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 21

1.7 Research Structure

The following chapters sequentially address the three objectives of this thesis:

Chapter 2 is a published paper, Chapter 3 is a drafted paper for publication, and Chapter 4 is a conventional thesis chapter.

CHAPTER 2 reports sources of variability in jarrah and marri seedling density at establishment using data collected from restoration sites between 1998 and 2017. This chapter achieves the first objective of this study—to use long-term monitoring data to identify the most important factors explaining seedling density at establishment (Section 1.1).

This investigation included factors related to jarrah forest restoration processes (Section

1.4.3) and the climate experienced over the study period (Section 1.3.3). Results of this research guided the development of field studies, including the contribution of canopy-stored seed entering restoration sites (Chapter 3) and suitable microsite characteristics important for establishment (Chapter 4). This chapter also considers the importance of hand-sown jarrah and marri seed for determining seedling densities, and helps to address objective 2 (Section

1.1).

CHAPTER 3 achieves the second objective by investigating the contribution of canopy-stored seed to observed seedling densities at establishment (Section 1.1). In this study, seed supply by sowing was explored (Chapter 2) and used as a seed source for the study of establishment microsites (Chapter 4). Then, seedfall from forest canopies was explored as a the only other seed source influencing seedling establishment on restoration sites.Results of this study also explain the relationship between the characteristics of unmined forest stands and potential seedfall in forest and restoration sites (Section 3.). Viability of jarrah seed has not been investigated in previous relevant work (e.g. Cargill et al., 2019).

While not a planned focus for this study, an exploration of the typical definition of viable JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 22 jarrah seed (Section 1.6.2)—which has implications for seed mix purity and consequently germination rates per gram—was undertaken (Appendix 2)

CHAPTER 4 is the culmination of three separate field trials (identified as A, B and

C) that aimed to highlight the importance of seedling microsite for supporting seedling establishment in restoration sites. To achieve the third objective (Section 1.1) this part of the study focused on ecological factors acting on jarrah and marri from seed application to establishment. Experiment A investigated the effect of soil chemistry, debris cover and microtopography on seedling establishment. Experiment B explored one of the same phenomena as Experiment A, microtopography, and its effect on seedling distribution and mortality. Finally, Experiment C investigated the effect of substrate type on seedling emergence. The results of this experiment include identification of the ecological characteristics of the restoration sites that most influence jarrah and marri densities, with potential to inform the restoration practices employed. It also is the first ecological study to investigate marri seedlings along with jarrah, to compare their responses to the same factors.

CHAPTER 5 provides a synthesis of the findings of the entire thesis in relation to the three objectives, explaining the variable establishment densities of jarrah and marri.

Specifically, this chapter suggests that the most important factors affecting jarrah and marri seedling densities at establishment are the process of flowering to seed release, restoration practices, biotic factors, climate and survival from seed to seedling. This chapter also informs management practices related to reducing and explaining the variability in jarrah and marri densities at establishment on jarrah forest restoration sites.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 23

Jarrah seedling near forest edge in a restoration site. Photo T. White-Toney.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 24

Chapter 2: Long-term Seedling

Density Monitoring

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 25

2.1 Chapter Introduction

This chapter addresses the first question posed in this thesis: Which ecological factors, including restoration practices, are most important for explaining long-term monitoring data of seedling densities at establishment? The chapter consists of two sections:

2.2) the manuscript of a published paper, and 2.3) supplementary analysis; because of this structure, the chapter headings deviate from the other chapters. In combination, this chapter explore the effect of restoration practices, seed sources and climate conditions on seedling establishment.

Section 2.2. is the penultimate draft of a manuscript now published in the journal

Ecological Management and Restoration. The only changes to the document have been to match referencing and table/figure numbers with the structure of the overall thesis. The statistical outputs from the models used in this paper can be found in Appendix 1. The full citation for the publication can be found at:

White-Toney, T., Korczynskyj, D., Grigg, A. and Bulsara, M. (2019). Variable tree

establishment in bauxite mine restoration in south-west Australia is linked to rainfall

distribution, seasonal temperatures and seed rain. Ecological Management &

Restoration 20(3), 266-270 doi:10.1111/emr.12389.

Section 2.3 offers a supplementary analysis in support of the published paper, which was omitted because of the journal’s word count/size restrictions. This section also includes predictive margins for the significant explanatory variables identified in the results and discussion section of the published paper, provided as Figures 2.3–2.9. These predictive margins were generated from the model outputs using the β value within the observed range for each significant explanatory variable.

The combined analysis provided in Sections 2.2 and 2.3 relies on historical records of jarrah (Eucalyptus marginata) and marri (Corymbia calophylla) seedling densities obtained JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 26 from Alcoa, who collected the data during routine monitoring These historical records also included the practices applied during restoration of each site: seeding rate and timing, fertiliser rate, and soil pathogen presence (all unpublished). Topsoil handling scores used in the analysis within this chapter were adapted from Alcoa records and from Standish et al.,

(2015).

2.2 Variable Tree Establishment in Bauxite Mine Restoration in South-west Australia is

Linked to Rainfall Distribution, Seasonal Temperatures and Seed Rain (publication

manuscript)

2.2.1 Summary

Reasons for variable establishment of Jarrah (Eucalyptus marginata D. Don ex Sm.) and Marri (Corymbia calophylla (Lindl). K. D. Hill & L. A. S. Johnson) on restored forest sites after bauxite mining in south west Australia are not well understood. To refine restoration outcomes, we compiled tree seedling density establishment data from surveys of

654 previously mined sites restored between 1998 and 2017, and applied Generalised Linear

Models to discriminate the effects of 24 climatic and restoration practice variables. Final models explained 50% and 31% of the variation in Jarrah and Marri density, respectively.

Broadcast seeding and fertiliser rates were positively related to seedling density. A more even rainfall distribution in the early wet season increased seedling density. However, persistent rain later in the wet season decreased density, possibly as a result of ripline soil saturation or ponding. Higher average daily maximum temperatures in the dry season decreased seedling density probably due to drought stress, but warmer daily temperature minima in both wet and dry seasons increased density. Seed rain from surrounding unmined forest was implicated as a significant, but highly variable, source of additional seed to restored sites. Restoration practices that influence soil moisture relations (tillage, depth and texture of returned soil), JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 27 shallow burial of applied seed and timing of fertiliser application are likely to be important in refining restoration outcomes.

2.2.2 Introduction

Variability in seedling establishment of Jarrah (Eucalyptus marginata D. Don ex Sm.) in the Northern Jarrah Forest of south west Australia, including after bauxite mining, has been the subject of several investigations (Cargill et al., 2019; McChesney et al., 1995;

Stoneman & Dell, 1994). Seed germination occurs during the winter wet season in this

Mediterranean-type ecosystem (Abbott, 1984). However, the interaction of seed supply and seedbed conditions leads to complex outcomes (Cargill et al., 2019). Alcoa of Australia

(Alcoa) has been mining bauxite and undertaking rehabilitation in the Northern Jarrah Forest for more than 50 years. Since the late 1980s, restoration sites have been directly seeded with

Jarrah and Marri (Corymbia calophylla (Lindl). K.D.Hill & L.A.S. Johnson), following methods described in detail by Koch (2007a). Routine assessment of tree densities indicated variable establishment, but to date investigations into underlying reasons have been inconclusive (Norman & Koch, 2009). Tree establishment densities falling outside acceptable ranges specified in mine rehabilitation standards require remedial treatment such as infill planting. Therefore, a better understanding of variable establishment could lead to improved cost efficiency. A more recent study into variability in understorey species return in the same restoration sites (Standish et al., 2015) identified the importance of both restoration practices and climatic drivers, suggesting that tree establishment may be affected by similar variables, along with seed supply. Here, we use a 20-year record of restoration monitoring in an effort to better understand the factors driving tree establishment variability in this environment, with the ultimate aim of improving restoration outcomes. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 28

2.2.3 Methods

We compiled records of Jarrah and Marri seedling densities from surveys of restoration sites at Alcoa’s Huntly (32°42’S 116°03’E) and Willowdale (35°09’S 116°05’E;

Figure 2.1) mines that had been restored over the period 1998 to 2017 inclusive (n = 654).

Surveys were carried out usually in late March, approximately one year after the completion of restoration and seeding activities, using a 2 m-wide transect of sufficient length to randomly cover 3-5% of each site. We then obtained or calculated, for each restoration site, a range of climatic and restoration practice variables (Table 2.1) that were considered likely to be important to seedling establishment (McChesney et al., 1995; Standish et al., 2015).

Figure 2.1 The study area is located in south-west Australia (a) and consists of restoration sites (orange areas, b & c) associated with Alcoa’s Huntly and Willowdale mines located within the Northern Jarrah Forest (dark green areas, b). The position of the Dwellingup climate station used to obtain daily air temperature, and rain gauge locations (c; blue squares) are shown.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 29

Table 2.1 Explanatory variables included in the saturated models for jarrah and marri establishment density (seedlings/ha). Final models included only significant variables (p < 0.05); variables denoted ‘NS’ and all interaction terms (not listed) did not significantly improve the model. The β coefficients (standard error in parentheses) for the final models indicate the magnitude and direction (+ or –) of the effect on seedling density of an increase in 1 unit of each explanatory variable. The observed range of each variable is provided to assist in assessing the potential magnitude of effect on seedling density. WS, wet season; DS, dry season—see text for definitions.

Explanatory Variable description; observed range β coefficient variable Jarrah Marri Restoration practice Jarrah seeding Jarrah seeding rate (g/ha): 1,360 (1998– +1.2 (0.2) NS rate 99), 700 (2000–02), 550 (2003–12), 490 (2013–15), 285 (2016–17) Marri seeding Marri seeding rate (g/ha): 500–540 NS +0.9 (0.2) rate (1998–99), 250 (2000–10), 200 (2011– 12), 180 (2013–15), 237 (2016–17) Forest perimeter Proportion of the perimeter of a mine pit +428.1 +139.3 that is forested: 0–0.99 (82.3) (35.9) Perimeter-to-area Perimeter of a restoration site divided by +0.6 (0.2) NS ratio the area of restoration site: 91.5– 928.7 m/ha Fertiliser rate Rate of fertiliser applied before +1.3 (0.3) +0.4 (0.1) monitoring date (kg/ha): 500 (1998– 2003), 280 (2004–15), 140 (2009), 0 (2016–17) Topsoil handling Relative measure of topsoil storage, NS NS score movement and spreading: 1–7 ('worst' to 'best'; Standish et al., 2015) Soil pathogen Detection of Phytophthora cinnamomi in NS NS presence the topsoil: present or absent Wet start Number of days between seeding and the NS –1.0 (0.2) start of the wet season: −47–186 Climate Annual rainfall 2 Total rainfall (January–December) 2 +0.3(0.01) +0.1 (0.05) years prior years before the year of monitoring: 454.7–1442.9 mm

Rain evenness Rainfall evenness over wet season: 0.13– –5839.2 –2755.7 full WS 0.28 (1118.0) (520.0) Rain evenness at As above for first 30 days of wet season: +2887.5 +807.0 30 days 0.10–0.34 (472.9) (190.6) JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 30

Explanatory Variable description; observed range β coefficient variable Jarrah Marri Rain evenness at As above for first 60 days of wet season: NS NS 60 days 0.10–0.39 Rain evenness at As above for first 90 days of wet season: +1235.0 NS 90 days 0.01–0.33 (477.5) Rain evenness at As above for first 120 days of the wet NS +1561.8 120 days season: 0.14–0.35 (346.0) WS rainfall Total rainfall from the start of the wet NS NS season to the start of the dry season; 339.4–1400.3 mm False start Wet (Rain > 2T) followed by dry (R < NS NS 2T) conditions before the onset of the sustained wet season: yes or no WS average Average daily maximum temperature NS NS max. temp. recorded during the wet season: 16.5– 20.5°C WS highest Highest daily maximum temperature NS NS temp. recorded during the wet season: 24.3– 38.0°C WS lowest temp. Lowest daily minimum temperature +80.9 +56.7 recorded during the wet season: −2.0– (33.6) (14.2) 1.0°C DS average max. Average daily maximum temperature –158.7 –98.8 temp. recorded during the dry season; 27.1– (38.5) (15.7) 31.2°C DS highest temp. Highest daily maximum temperature +54.2 NS recorded during the dry season: 35.0– (19.0) 42.0°C DS lowest temp. Lowest daily minimum temperature +50.6 +38.7 (7.6) recorded during the dry season: 1.5–8.0°C (19.2) Days over 40°C Number of days in the dry season with a NS NS maximum temperature >40.1°C: 0–4 DS rainfall Total rainfall from the start of the dry –1.3 (0.6) NS season to the time of monitoring (31 March): 4.8–265.0 mm

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 31

Daily rainfall records were obtained from one of six rain gauges nearest to each restoration site (mean distance 7.8 km; Figure 2.1). Daily air temperature data for all sites were sourced from the nearest Climate Station at Dwellingup (32°72’S 116°07’E; Bureau of

Meteorology http://www.bom.gov.au; Figure 2.1). The wet season was defined as the period in each year when rainfall (R) was greater than twice the mean temperature (2T), following

Standish et al. (2015), based on running 30-day rainfall totals and 30-day average maximum temperatures (i.e. R/T > 2). ‘False starts’ to a wet season (Standish et al., 2015) were also identified. In our study, the dry season was defined as the period from the end of the wet season to the last day of March, coincident with the approximate timing of tree monitoring.

Wet season rainfall evenness during the first 30, 60, 90 and 120 days and for the full season was calculated using a modified measure of species evenness (Simpson’s Diversity index;

Bronikowski & Webb, 1996). Evenness values vary from 0 (all rain recorded for the period falls on one day) to 1 (rain recorded for the period falls in equal amounts each day). Seed of

Jarrah and Marri was typically broadcast onto the freshly prepared soil surface in the dry season prior to the first winter wet season. Bulking of seedlots and seed viability checks as standard practice minimised the potential for variability in seed quality across restoration sites and years. Reductions in seeding rate over the study period (described in Table 2.1) were a management response to changes in mine rehabilitation standards for acceptable tree establishment densities. A standard fertiliser blend of di-ammonium phosphate plus potassium and micronutrients was typically applied in the first spring after completion of restoration activities. From 2016 onwards, application was delayed until the second spring after establishment, resulting in zero fertiliser at the time of monitoring. Topsoil management for each restored site was scored from ‘best’ (7) to ‘worst’ (1), following Standish et al.

(2015). Seed rain was not measured as part of this study, however three surrogate variables were included: the proportion of each restoration site’s perimeter that was composed of intact JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 32 mature forest, the perimeter to area ratio of the restoration site, and total annual rainfall two years prior to monitoring, which is correlated with seed supply (Johnstone & Kirkby, 1999).

We used Generalised Linear Models (GLMs) with normal distributions and identity link functions within the statistical package Stata Ver. 15 (StataCorp, 2017) to identify the explanatory variables, and their interactions, important for predicting Jarrah and Marri seedling densities at establishment. The analysis omits collinear variables prior to model fit, and because no collinear variables were identified, all potential variables were included in the model selection process. The final model for each species was obtained by backward step elimination of variables from the saturated model until only significant variables remained (p

< 0.05). The goodness of fit of each model was assessed using Pseudo R2, and effect size was calculated to rank the relative importance of each explanatory variable. Model output includes the expected change in tree seedling density for a unit change in a given explanatory variable (β value), holding all other model variables at their mean value (Vittinghoff,

Glidden, Shiboski, & McCulloch, 2012). Each β value must be interpreted within the context of the observed range for the explanatory variable.

2.2.4 Results and discussion

A total of 13 climate and restoration practice variables were included in the final model for Jarrah, and 11 for Marri (Table 2.1), explaining 50.1% and 30.6% of the variation in seedling density, respectively. No one factor had an effect size (ES) greater than 8% (Table

2.2), which may help to explain why previous short-term attempts to isolate single important explanatory variables were met with limited success (Norman & Koch, 2009). As expected, seed application rate was highly significant in both models, and ranked first and second by effect size for Jarrah and Marri, respectively (Table 2.2). Observed trends in annual average density for Jarrah, from 2368 seedlings/ha in 1998 to 247 seedlings/ha in 2017, and for Marri JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 33 from 852 seedlings/ha in 1999 to 229 seedlings/ha in 2013, are consistent with reductions in seeding rates for each of the two species over the study period (Table 2.1).

Additional seed supplied from the forest surrounding restored sites is strongly implicated in this study and, due to high spatial heterogeneity of canopy-borne seed in the forest (Cargill et al., 2019), is likely to be an important contributor to seedling density variability among restoration sites within a single year. Furthermore, temporal variation in

Jarrah seed production has previously been reported (Abbott & Loneragan, 1986) which may contribute to variability across years. Restoration sites with an increasing proportion of forest perimeter (Jarrah ES, 0.041; Marri ES, 0.023), and to a lesser extent those with a higher ratio of perimeter to area (Jarrah ES, 0.014), were estimated to have increased seedling establishment: more than 400 additional seedlings/ha for a fully forested perimeter for Jarrah and nearly 140 seedlings/ha for Marri (Table 2.1) when compared to sites with no forest edge. Restoration site geometry (here measured as perimeter to area ratio) becomes increasingly important with increasing seed dispersal distance since this enhances overall site coverage by seed rain. The inclusion of this variable in the model for Jarrah but not Marri may be because Marri seed is much larger than Jarrah (average seed mass 0.113 g for Marri vs. 0.020 g for Jarrah; Abbott, 1984) and hence has a smaller dispersal distance. Abbott &

Loneragan (1986) report dispersal distances for Jarrah seed of up to 1.5 times tree height; dispersal distances for Marri seed are unknown. The importance of seed rain is underscored by the inclusion in the final models of both species of annual rainfall two years prior to restoration activities (Jarrah ES, 0.007; Marri ES, 0.006). Years with high wet season rainfall are associated with abundant flowering in the following spring (Johnstone & Kirkby, 1999), potentially resulting in increased capsule production and seed supply in the summer dry JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 34 season approximately 12 months later. Quantification of the potential contribution of seedfall from the forest on the perimeter of restoration sites is the focus of a separate study.

The positive relationship between fertiliser rate and both Jarrah (ES, 0.030) and Marri

(ES, 0.046) density (Table 2.1) was unexpected, as this had not been detected in previous fertiliser field trials (e.g. Lockley & Koch, 1996). Presumably, increased nutrient availability from fertiliser application promotes root growth of the young seedlings, thus improving survival over the ensuing dry season. While higher fertiliser application rates may increase tree seedling density, lower application rates promote greater understorey species richness and similarity of restored sites to unmined forest (Daws et al., 2015), presenting competing objectives. Since current practice is to apply fertiliser in the second year after establishment, earlier application of fertiliser warrants further investigation as a possible solution.

Jarrah and Marri densities were both positively and negatively related to the evenness of rainfall distribution during the wet season depending on the period of measurement, while the total amount of wet season rainfall received was not important (Table 2.1). Rainfall received more evenly through the early part of the wet season (typically May–July for Jarrah,

May–August for Marri) had a positive relationship with density (effect size of rainfall evenness over the first 30 days was 0.055 for Jarrah and 0.027 for Marri). However, continued wet conditions later into the wet season had a negative relationship (effect size of rainfall evenness over the full wet season was 0.041 for Jarrah and 0.042 for Marri). Early and more even rainfall is likely to provide a consistently moist soil environment at the surface, promoting successful seedling emergence and survival of the small seedling, which is at heightened risk of desiccation. Persistent rainfall may result in saturated soil or even ponding within the contour riplines, leading to anoxic soil conditions and seedling death.

Jarrah seedlings maintain open leaf stomata even after waterlogging, leaving this species JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 35 particularly susceptible to anoxic soil conditions (Davison & Tay, 1985). Brief but heavy rainfall events in the dry season may also have a similar effect (Table 2.1).

Given the importance of rainfall evenness, soil management practices that influence moisture holding capacity (tillage depth, and depth and texture of returned soil layers over the mine pit floor) could be important in determining tree seedling density among and within restoration sites. Stoneman, Dell and Turner (1994) also implicated seedbed conditions in bauxite mine pits, as they related to water deficits, in successful Jarrah seedling survival.

Topsoil quality as determined by storage, processing and return, and by the presence of a soil pathogen, were not included in either model (Table 2.1). This contrasts with understorey species which have been shown to be strongly influenced by soil management practices

(Standish et al., 2015). This difference most likely reflects differing sources of seed, being dominated by the soil seedbank in the case of understorey but not for Jarrah or Marri (Koch et al., 1996). The results of this study are consistent with the conclusions of Cargill et al.

(2019) in that the overwhelming driver of Jarrah seedling establishment is seed supply.

Temperature conditions were most important during the dry season. Higher average daily maximum temperatures were associated with reduced densities of both species (Table

2.1; Jarrah ES, 0.026 and Marri ES, 0.058), probably linked to drought stress. A positive relationship between density and higher dry season minimum temperatures (Jarrah ES, 0.011 and Marri ES, 0.039), and possibly also brief hot spells for Jarrah (ES, 0.013), may be related to enhanced capsule ripening and seed fall leading to increased contribution of seed from peripheral forest under these conditions, although this requires further investigation. A positive relationship with the only significant wet season temperature variable, the lowest daily minimum (Jarrah ES, 0.009 and Marri ES, 0.024), possibly relates to the damaging effects of frost (Matusick, Ruthrof, Brouwers, & Hardy, 2014). Alternatively, lower wet JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 36 season overnight temperatures are often associated with dry periods with clear night skies, suggesting an indirect effect of soil moisture availability.

The only variable in this study that was significant for Marri but not Jarrah was the delay between seeding and the onset of the wet season (Table 2.1). A longer delay between seed broadcasting and commencement of the wet season was associated with lower Marri establishment (ES, 0.045). Marri seed is much larger than Jarrah, as indicated above, and it may be that these seeds are more likely to remain on the soil surface and be subject to greater predation and/or temperature and moisture extremes than Jarrah seed. Seed delivery approaches that result in shallow (5–15 mm) burial may therefore lead to improved Marri establishment success.

2.3 Supplementary Material

Table 2.2 Overall effect size (ES) of linear regressions for jarrah and marri models, and the ranked component ESs for each explanatory variable. Upper and lower 95% confidence intervals (CI), and the direction of effect on density, from GLMs are also shown. WS, wet season; DS, dry season—see text for definitions and model selection.

Variable ES Lower CI Upper CI Direction of effect Jarrah Model 0.502 0.441 0.537 Jarrah seeding rate 0.076 0.042 0.118 + Rain evenness at 30 days 0.055 0.026 0.093 + Rain evenness full WS 0.041 0.016 0.075 – Forest perimeter 0.041 0.016 0.074 + Fertiliser 0.030 0.010 0.061 + DS average max. temp. 0.026 0.007 0.055 – Perimeter-to-area ratio 0.014 0.002 0.037 + DS highest temp. 0.013 0.001 0.035 + DS lowest temp. 0.011 0.001 0.032 + Rain evenness at 90 days 0.010 0.000 0.031 + WS lowest temp. 0.009 0.000 0.029 + DS rainfall 0.009 0.000 0.029 – Annual rainfall two years prior 0.007 0.000 0.026 +

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 37

Variable ES Lower CI Upper CI Direction of effect Marri Model 0.306 0.239 0.349 DS average max. temp. 0.058 0.028 0.010 – Marri seeding rate 0.050 0.023 0.087 + Fertiliser 0.046 0.020 0.082 + Wet start 0.045 0.019 0.080 – Rain evenness full WS 0.042 0.017 0.076 – DS lowest temp. 0.039 0.015 0.072 + Rain evenness at 120 days 0.031 0.010 0.061 + Rain evenness at 30 days 0.027 0.008 0.056 + WS lowest temp. 0.024 0.006 0.053 + Forest perimeter 0.023 0.006 0.050 + Annual rainfall two years prior 0.006 0.000 0.024 +

Figure 2.2 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given seeding rates, when all significant model parameters (Table 2.1) are held at their mean value. Y-axis values differ because of lower seeding rates for marri than for jarrah. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 38

Figure 2.3 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given forest perimeter, when all significant model parameters (Table 2.1) are held at their mean value.

Figure 2.4. Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given fertiliser application rates, when all significant model parameters (Table 2.1) are held at their mean value. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 39

Figure 2.5 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given rain evenness at 30 days, when all significant model parameters (Table 2.1) are held at their mean value.

Figure 2.6 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given full wet season (WS) evenness, when all significant model parameters (Table 2.1) are held at their mean value. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 40

Figure 2.7. Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given dry season (DS) average maximum temperatures, when all significant model parameters (Table 2.1) are held at their mean value.

Figure 2.8 Predictive margins (95% CI) of jarrah (black) and marri (grey) seedling densities for given wet season (WS) lowest temperatures, when all significant model parameters (Table 2.1) are held at their mean value. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 41

Restoration site seed trap (front) and forest seed trap (back; from Section 3.2). Photo T.

White-Toney.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 42

Chapter 3: Seed Sources

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 43

3.1 Chapter Overview and General Introduction (Eucalypt Seed Sources)

This chapter investigates the sources of jarrah (Eucalyptus marginata) and marri

(Corymbia calophylla) seed on jarrah forest restoration sites, in line with question 2 of this study: To what extent is seed entering the mine restoration site from each of broadcast sowing and seedfall from the forest canopy? And what is the effect of these seed sources on variability in seedling density at establishment? Broadcast seed sowing was a management variable included in Chapter 2. In this chapter, sown seed was used as a comparison with seedfall from the forest canopy, but only seedfall was directly measured and explored. This chapter is structured as a draft paper for publication, hence the inclusion of an abstract, introduction, etc. However, to support the reader references to other sections of the thesis have been included and the manuscript is prefaced by this “General Introduction” which offers a more complete review of the alternative sources of seed supporting eucalypt recruitment.

Under natural conditions, eucalypt recruitment is reliant upon seed sourced from tree canopies (Abbott & Loneragan, 1986). Dispersal of this seed is dependent on seed size

(Janzen, 1970), tree height (Thomson, Moles, Auid, & Kingsford, 2011), the individual tree relative to its last year of flowering and the environmental conditions experienced during its growth (e.g. soil nutrients and climate; Florence, 2004).

When the forest canopy has been removed, such as by tree clearing for agriculture or mining, seed for recruitment must be supplied by other means. Forest topsoil is an important seed source for many jarrah forest species, but not jarrah and marri (e.g. Tacey & Glossop,

1980). Following the clearing of jarrah trees prior to mining, Koch et al. (1996) found 5.3 jarrah seeds/m2 in forest topsoil. This was much lower than the level measured for a cohabitating tree species, Allocasuarina fraseriana (15.8 seeds/m2), but higher than for marri and E. patens (both 1.0 seed/m2). In topsoil harvested from forest sites that did not undergo JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 44 clearing, there was 1.0 jarrah seed/m2 (Koch et al.,1996). Further, the persistence of viable seed of understorey species in the forest topsoil means that topsoil handling practices are important for the return (and richness) of understorey species to jarrah forest restoration sites

(Standish et al., 2015), but in this study was not of apparent value to jarrah and marri seedling establishment densities (Chapter 2). Other work on jarrah seedling recruitment has not included the seedbank as an important source of jarrah seed (e.g. Cargill et al., 2019). Based on this previous work illustrating low presence of jarrah and marri seed in the seedbank (e.g.

Koch et al., 1996), jarrah and marri seed is currently sown from seed mix applied to restoration sites following topsoil return (Koch, 2007a). Forest topsoil, therefore, has not been considered a significant source of jarrah and marri seed (Koch et al., 1996; 2009; Cargill et al., 2019) and was not investigated in this study.

Despite consistent seeding rates within single years (Table 2.1) and across two methods of seed sowing (hand and helicopter; Koch, 2007a), variability in establishment densities of jarrah and marri is observed (e.g. Chapter 2; Norman & Koch, 2009). Seed source as a cause of this variability is an important consideration for understanding seedling establishment (e.g. Harper, Williams, & Sagar, 1965). The number of seeds, and their viability, has an important and obvious effect on emergence from seed and later establishment of seedlings. Viable jarrah seeds have been defined previously as those between four mm (Cargill et al., 2019) and 4.5 mm (Abbott & Loneragan, 1986) but this size cut-off has not been validated, despite reported variability in jarrah seed viability (e.g.

Florence, 2004). There has also been no published work reporting the viability of marri seed.

Germination trials are the standard practice for identifying seed viability, but new, non- destructive methods for assessing viability are also used (e.g. X-ray imaging; Elias,

Copeland, McDonald, & Baalbaki, 2012). The use of X-ray imaging was investigated to determine fullness of jarrah and marri seed as an alternative to the established size-based JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 45 criteria for distinguishing viable seed from associated particles. This investigation was an aside to focus of the study and those results are included in Appendix 2.

The amount of unviable seed and chaff included in a seed mix is another important consideration for emergence and subsequent establishment. Like other Eucalyptus species, jarrah produces seed and chaff that are not easily visually distinguished from each other

(Boland et al., 1980). Consequently, the number of jarrah seedlings resulting from supplied seed depends on the purity (i.e. proportion of seed to chaff) of the seed mix (Baker, 2019).

The amount of unviable seed and chaff included in a seed mix is another important consideration for emergence and subsequent establishment. Marri seeds are large (12.9 mm;

Abbott, 1984) and easily distinguished from chaff (Boland et al., 1980); therefore, the purity of marri seed does not impact seedling density.

The phenomenon of seed supplied from jarrah forest canopies increasing seedling establishment in nearby cleared areas undergoing restoration has been observed, but never quantified (Koch, 2007b; Ward, Koch, & Nichols, 1990). This is another important aspect of seed supply to restoration sites explored in this study. The influence of seed from forest edges was also suggested by the positive relationship between forest perimeter and seedling establishment (Chapter 2) and was investigated in the field study reported here, in Section

3.2.

3.2 Tree Seedfall into Jarrah Forest Restoration Sites (draft manuscript)

3.2.1 Abstract

Recruitment of tree seedlings from forest seedfall into adjacent areas is a well- established phenomenon and assists native forest regeneration. The recruitment of jarrah in this way is preferred after selective logging in the Northern Jarrah Forest of south-west

Australia for timber production. However, successful recruitment of jarrah from forest seedfall is variable because seed production fluctuates between years and individual trees. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 46

This study investigated tree canopies near edges of post-mining restoration sites as a seed source influencing observed variable seedling densities on restoration sites, including identifying seed dispersal distances and any forest characteristics related to seedfall. Seedfall was measured using 168 seed traps along 28 transects from forest edges up to 75 m into restoration sites. Only four marri seeds were captured, with maximum dispersal distance of

6.0 m. The maximum dispersal distance of jarrah seed was 12.7 m from the forest edge, with an average of 10,000 ± 2,624 (SE) seeds/ha/year. Distance from forest was the only factor that predicted the amount of seedfall into restoration sites, explaining 14% of the variability.

Canopy capsule density predicted 22% of seedfall variability in forest. Tree density, height, basal area and canopy cover, and ground elevation were not significant predictors. Seed supply within 12.7 m of forest edge can be up to four times more than the amount of seed sown by management during restoration. However, because restoration sites differ markedly in the amount of forested perimeter and their perimeter-to-area ratio the influence of seed fall on tree seedling recruitment will vary accordingly. In addition, the coupling of this relationship with the natural spatial and temporal variability in seedfall further affects the amount and location of seed suppled to restoration sites, and so is a contributing factor to the observed variability in seedling densities.

3.2.2 Introduction

Natural recruitment of soft-seeded tree species relies on a sufficient supply of canopy seed (Cremer, 1965) and during inter-fire periods for Mediterranean environments or other fire-prone habitats (Abbot & Burrows, 2003). In addition to seed supply, seedling recruitment requires a suitable seedbed for emergence (e.g. Harper et al., 1965). Previous work in eucalypt forests has found that seedbed preparation is less important for determining recruitment success than is seed supply (Cargill et al., 2019; Vesk, Davidson, & Chee, 2010).

Consequently, silvicultural practices in south-western Australia focus on seed supply to JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 47 increase seedling regeneration and therefore timber production (Abbott & Williams, 2011;

McCaw, 2011). ‘Seed trees’ are selected based on expected seed production and are left during selective logging in a spatial pattern that encourages seed supply (Abbott & Williams,

2011; Burrows, 2000).

The number of seeds produced by an individual tree (‘seed production’), and timing, are important determinants of seedling recruitment and for explaining seedling densities and distribution patterns (Yates, Hobbs, & Bell, 1994). Tree seed production depends on the characteristics of individual trees, the surrounding vegetation (i.e. forest stand characteristics) and the tree species. Individual trees have cyclical seed production, which is dependent on seasonal rainfall and temperature (Abbott & Loneragan, 1986; Florence, 2004). Canopy seedfall for Eucalyptus species occurs when capsules dry because of heat from fire or warming weather, and open to release seed (Boland et al., 1980; Florence, 2004). Therefore, the majority of seedfall in eucalypt forests occurs in the warmer, drier summer and autumn months. Seed production of individual trees also depends on the number of years since the last large seed production cycle, and age (e.g. Abbott & Loneragan, 1986).

Seed dispersal distance is dependent on seed size and tree height. Small-seeded species have increased probability of survival compared with large-seeded species, because their seeds disperse further from the parent plant (Hyatt et al., 2003; Janzen, 1970). However, a recent cross-species analysis of dispersal distance by seed mass and plant height revealed plant height to have greater explanatory power than seed mass for describing seed dispersal distance (Thomson et al., 2011). Seed from eastern Australian rainforest tree and shrub species has not been found more than 10 m into adjacent cleared areas (Charles, Dwyer, &

Mayfield, 2017). Similarly, dispersal of Eucalyptus and Corymbia species from a plantation in south-eastern Brazil was not beyond 10 m of the forest edge in a study by Miolaro et al.

(2017). These two factors—timing and dispersal distance—affect seedling recruitment JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 48 patterns in forest, and the likelihood of tree recolonisation in cleared areas. In these cleared areas undergoing restoration, soft-seeded species often require hand sowing to increase seedling recruitment. However, there is the potential for seed entering restored areas from nearby forest to be an additional seed source on restoration sites, increasing the number of seedlings near the forest (e.g. Charles et al., 2017; Miolaro et al., 2017). As a result, cleared land adjacent to remnant patches of vegetation provides an important and novel study system for seedling recruitment.

In the Northern Jarrah Forest of south-western Australia, mining for bauxite occurs in lateritic deposits coincident with the highest quality jarrah stands (Bradshaw, 1999). The areas for mining are intermixed with forest, leading to a unique matrix of forest, restoration sites and active mining areas (Craig, Stokes, Hardy, & Hobbs, 2015; Chapter 2). Variable densities of jarrah seedlings at establishment have been the source of a number of investigations, but with limited success at explaining the observed variability (Cargill et al.,

2019; Norman & Koch, 2009; Stoneman & Dell, 1994). Marri is co-dominant with jarrah, and marri seedling densities at establishment are also variable, but have not been the topic of any targeted studies. In this system, jarrah and marri seed is sown (by hand or by helicopter) on restoration sites (Koch, 2007a) because of its low persistence and presence in the forest topsoil returned to mine sites (Koch et al., 1996; 2009; Tacey & Glossop, 1980). These sites are suitable for jarrah and marri establishment due to the absence of a canopy, no competition

(Bond & van Wilgen; 1996), increased water availability (Hatch, 1960; Loneragan &

Loneragan, 1964), and soil nutrients (Loneragan & Loneragan, 1964; Cummings et al., 2007) and the very low probability of fire due to the absence of fuel.

The varied proportion of forest edges surrounding restoration sites is produced by the mining process unique to the Northern Jarrah Forest, and the quantity and timing of seed supply from forest edges to restoration sites is so far unquantified. Two studies have JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 49 indicated the potential of canopy-stored seed for tree recruitment in restoration sites. Ward et al. (1990) suggested that overhanging tall trees (including jarrah) along forest edges may be contributing a large number of seeds, and Koch (2007b) estimated that jarrah seed dispersal occurs up to 20 m into restoration sites. Additionally, one of the primary factors influencing both jarrah and marri seedling establishment in jarrah forest restoration sites was the percentage of perimeter that was forested, and this was likely to be related to seed supply

(Chapter 2). This study therefore aimed to further explore seed production in the jarrah forest and seed dispersal as a possibly important factor related to seedling establishment in adjacent cleared sites. To do so, this study identified the dispersal distance of seed from forest edges into restoration sites and highlighted forest characteristics that could predict the amount of seedfall observed in forest stands and restoration sites.

3.2.3 Methods

3.2.3.1 Study site

The Northern Jarrah Forest of south-western Australia is a tall, dry sclerophyll forest with 30–70% canopy cover (Specht, Roe, & Boughton, 1974), that experiences a

Mediterranean-type climate, with cool, wet winters and warm, dry summers (Gentilli, 1989).

Across its range, jarrah forest is dominated by two upper canopy species, jarrah and marri.

Jarrah’s dominance is replaced by E. megacarpa and E. patens in more fertile, moist areas

(Dell & Havel, 1989), and these swamp regions are included in the forest regions where mining occurs. Approximately 550 ha of jarrah forest per year has been mined for bauxite and restored since the 1960s (e.g. Bradshaw, 1999; Koch, 2007a). Each restored mine site has a distinctly different shape, size and proportion of forest perimeter (Craig et al., 2015), and an increased proportion of forest perimeter has been found to increase seedling densities

(Chapter 2), therefore, these aspects of restoration site geometry were explored related to the variable opportunity for canopy-stored seed supply (Chapter 2). JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 50

3.2.3.2 Study species

Jarrah and marri both grow to a maximum height of 50 m and occur along the Darling

Scarp of south-west Australia (Abbott & Loneragan, 1986). Like other eucalypt species, jarrah and marri trees are generally animal pollinated, including by birds, mammals and insects (Vesk et al., 2010). Jarrah flower buds initiate as early as December (early summer), flowering occurs between September and the following December (spring to summer), and seed is shed between the third December and March (summer to autumn). The whole jarrah reproductive cycle usually takes 24–27 months (Johnstone & Kirby, 1992). Natural jarrah recruitment occurs when seed is released from dried capsules in the upper canopy

(Christensen, 1971; Cremer, 1965). Individual trees experience years of ‘bumper crop’ seed production once every four to seven years (Abbott & Loneragan, 1986; Florence, 2004).

Jarrah seed and chaff are morphologically similar, but are usually separated by size (smaller than four mm) and colour (black; Abbott & Loneragan, 1986; Cargill et al., 2019) as they were in this study. Marri flowers initiate in August (late winter), flower in January (mid- summer), develop capsules between March and December, and release seed the following

January (summer; Johnstone & Kirkby, 1992). Marri seed is usually 12.7 mm long and easily distinguished visually from chaff (Boland et al., 1980; Johnstone & Kirby, 1992).

3.2.3.3 Sampling design

A total of 168 steel seed traps (1-m2 opening) were placed 1 m above the ground along 28 transects perpendicular to the forest edge of 10 restoration sites (Table 3.1). Site selection ensured that plant regrowth height did not impede seedfall into the traps (Figure

3.1). Each transect included six traps and extended 75 m into the restoration site. This range was chosen based on the maximum predicted seedfall of a 50 m tree of 75 m (i.e. seed dispersal of 1.5 times the tree height; Abbott & Loneragan, 1986; Cargill et al., 2019). One seed trap was placed 20 m within the forest edge, with five additional traps in the restoration JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 51 site. Restoration site traps were randomly located along transects in 15-m increments; that is,

0–15 m, 15–30 m, 30–45 m, 45–60 m and 60–75 m (Figure 3.1). Data were analysed as forest traps (n = 28), restoration site transects (n = 28), restoration sites (n = 10) and individual restoration site traps (n = 140). Traps were set between November and December 2017, and seed collection occurred monthly between January 2018 and January 2019.

Custom seed bags (15 cm2) made from CoolarooTM weed control mat were attached with hook-and-loop fasteners and zip ties to the funnel opening (approximately 78 cm2) at the base of each trap (Figure 3.1). The funnel-bag formed a ‘pit-trap’ like arrangement that deterred invertebrate-thieves leaving the bag. To further control for seed theft by invertebrates (Majer, 1980), seed bags were affixed above the ground, and holes repaired during each monthly collection. Seed bag contents were frozen for three days to kill invertebrates and bags collected in wet conditions were dried at 30 oC for three days prior to sorting. Contents were sieved and then jarrah seeds separated from chaff (Boland et al., 1980) based on their size (smaller than 4 mm) and colour (Abbott & Loneragan, 1986; Cargill et al.,

2019). The seed collected per trap (1 m2) was used to calculate total seedfall per hectare per year. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 52

Figure 3.1 A seed trap transect comprised five traps within the restoration site (shown) and one trap in the forest (not shown). Seed traps were metal funnels (1 m2) supported by fence droppers (1 m height) with a bag for seed collection. Photo T. White-Toney.

3.2.3.4 Forest stand characteristics related to seedfall

Forest stand surveys were conducted within a 20-m radius of each forest trap from

June to July 2018, with the diameter of sampling area stretching from the edge of the restoration site, 40 m into the forest. The total sampling area per forest stand, therefore, was

1,257 m2. The number of jarrah and marri trees in 10 categories of diameter at breast height

(DBH) were counted. The smallest DBH category was for trees less than 5 cm and upto 10 cm, and categories increased by 5 cm up to a DBH of 50 cm. All trees with a DBH >50 cm pooled into the largest size category. DBH was used to calculate density (stems/ha) and basal area (m2/ha) of each stand. Jarrah capsule number was conservatively estimated (canopy capsules/1,257 m2) from the lowest DBH value of each category range, capsule clump density (CDEN; a visual estimate) and capsule clump distribution (CDIS; a visual estimate), following the method validated by Cargill et al. (2016). Orientation of the forest in relation to JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 53 the restoration site was also recorded because of the predicted effect of wind on seed dispersal.

Canopy height was determined using a canopy height model (CHM) created from light detection and ranging (LiDAR; e.g. Simard, Pinto, Fisher & Baccini, 2011) point cloud data for the jarrah forest collected in 2017 and accessed by Alcoa. The height values from the

CHM were used to determine the 90th percentile and maximum tree height in the sampling area (1,256 m2). Canopy cover was measured from the LiDAR data as the proportion of returns (i.e. pulses emitted from the LiDAR) that encountered a tree canopy (more than two metres tall) rather than the forest floor (0 metres). A digital elevation model was also generated from LiDAR returns off the forest floor to calculate elevation change between the traps in the forest and restoration site.

3.2.3.5 Data analysis

GLMs with normal distributions and identity link functions within the statistical package Stata Ver. 15 (StataCorp, 2017) were used to identify the explanatory variables important for predicting seedfall in the restoration and forest sites, separately. The analysis omits collinear variables prior to model fit, but because no collinear variables were identified, all potential variables (n = 12) were included in the saturated model for model selection. The final models were obtained by backward step elimination of variables from the saturated models until only significant variables remained (p < 0.05). The goodness of fit of each model was assessed using pseudo R2. The model outputs include the expected change in seeds per hectare per year for a unit change in a given explanatory variable (β value), holding all other model variables at their mean value (Vittinghoff et al., 2012). Each β value must be interpreted within the context of the observed range for the explanatory variable (Table 3.2). JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 54

3.3 Results

Jarrah seed was present in 97% (27/28) of forest traps and average seedfall was

150,000 ± 23,511 (SE) seeds/ha/year, varying from 0 to 540,000 seeds/ha/year between forest traps (Figure 3.2). Marri seed was captured in three forest traps and one restoration site trap.

The 28 forest traps were present across 10 restoration sites and seedfall also varied between adjacent forest traps (Table 3.1). Forest seedfall had an annual peak in the warm months of late summer to autumn, with the most seedfall occurring between January and April.

However, seed was present in forest seed traps throughout the year (Figure 3.3).

Restoration site seedfall (0–12.7 m) Forest seedfall

Figure 3.2 Jarrah seedfall (seeds/ha/year) for jarrah forest (dark grey; n=28) and restoration site (within 12.7 m of the forest edge; light grey; n=28) seed traps. Median and lower (Q1) and upper (Q3) quartiles represent observations inside the nine to 91 percentile range and one outlier (black dot) with 540,000 seeds/ha/year seedfall.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 55

Table 3.1 Mean and standard error (SE) of jarrah seedfall (seeds/ha/year) in forest and restoration site’s transects (10 locations and 28 transects) illustrating variability in adjacent forest traps.

Number Forest Restoration Sites Restoration site of Transects Mean (SE) Mean (SE) McCarthy 2 2 65,000 17,678 0 0 Karri 14 2 160,000 84,853 0 0 Karri 17 2 145,000 24,749 0 0 Roberts 14 2 115,000 3,536 0 0 Roberts 11 2 210,000 7,071 10,000 7,071 Downes 9W 4 100,000 29,368 10,000 5,000 DeMarte 2 5 252,000 73,506 12,000 5,215 Rhodes 1 3 28,000 9,813 14,142 8,164 Karri 10 3 290,000 68,069 20,000 15,275 Downes 2 3 76,667 30,671 23,333 7,200

140,000 120,000 100,000 80,000 60,000

Total seeds/ha Total 40,000 20,000 0 JanJAN FebFEB MarMAR AprAPR MayMAY JUNEJun JULYJul AUGAug SEPT Sep OctOCT NovNOV DecDEC Restoration site seedfall Forest seedfall

Figure 3.3 Total monthly jarrah seedfall (seeds/ha) in forest (n=28; dark grey) and restoration site (n=28; light grey) traps (0–12.7 m of forest edge; n=28).

Jarrah seed was present in 43% (12/28) of restoration site trap transects in the restoration site and maximum seedfall was 12.7 m from the forest edge (Figure 3.4). Average seedfall within 12.7 m of forest edge was 10,000 ± 2,624 (SE) seeds/ha/year (Figures 3.2 &

3.4). The timing of seedfall in the restoration sites mirrored forest seedfall, with an annual peak in the warm months of late summer–autumn (January–April). March had the highest JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 56 seedfall for both forest and restoration sites and no seedfall was recorded in restoration site traps between August and November (Figure 3.3).

60,000

50,000

40,000

30,000

20,000

10,000 Jarrah seedfall Jarrahseedfall (seeds/ha/year) 0 0 10 20 30 40 50 60 70 80 Distance from forest edge (m)

Figure 3.4 Total jarrah seedfall (seeds/ha/year) from forest edge into all restoration sites (n = 140).

The only forest parameter that had a significant relationship with forest seedfall was the estimated number of capsules (Cargill et al., 2016), which explained 22.2% of the variability in forest seedfall (Table 3.2). Of the 10 explanatory variables related to forest stand characteristics, only distance from the forest edge had a significant relationship with restoration site seedfall, explaining 14.4% of the variability (Table 3.2). The GLMs predicted reduced jarrah seedfall by 124 seeds/ha/year with every metre from forest edge. There were no relationships between forest characteristics and seedfall patterns in the restoration site; however, these forest surveys provided a useful assessment of the variability in jarrah forest stand characteristics (Table 3.2 for the range of each variable).

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 57

Table 3.2 Explanatory variables included in the saturated models for jarrah seedfall in restoration and forest sites (in seeds/ha/year). Final models only included significant variables (p < 0.05). ‘NS’ indicates non-significant variables that were included in the saturated model, and ‘NA’ marks variables that were not included. The β coefficients (standard error in parentheses) for the final models indicate the magnitude and direction (+ or –) of the effect on the amount of seedfall of an increase in one unit of each explanatory variable. The observed range of each variable is provided to assist an assessment of the potential magnitude of the effect on seedfall.

Explanatory Variable description; observed range β coefficient variable Restoration Forest sites sites Forest seedfall Number of seeds: 0–540,000 seeds/ha/year NS NA Distance from Distance between each restoration site trap –124.0 (25.7) NA forest and the nearest forest edge: 0–81 m Jarrah density Number of stems in 20-m radius of each trap: NS NS 127–1,433 stems/ha Orientation Orientation of restoration site trap to the NS NS forest trap: 0–330 degrees Capsule Estimated number of capsules (from Cargill NS +143.6 estimate 2019; model 6) in the 20-m radius forest (52.7) sampling area: 83–1,728 Jarrah basal area Total basal area of jarrah per hectare (ΣDBH2 NS NS * π): 2–26 m2/ha Number of Number of jarrah trees over 50 m: 0–6 NS NS jarrah >50 m Maximum tree Maximum height of canopy in sampling area NS NS height (from LiDAR point cloud): 20–30 m 90th percentile Minimum height of canopy in the top 10% for NS NS for height height in 20-m radius of each trap (from LiDAR point cloud): 16–24 m Proportion of Proportion of canopy cover: 0.4–0.9 NS NS cover Forest trap Elevation (above sea level) of forest trap; NS NS elevation 258–335 m Difference in Difference in elevation between the forest trap NS NS elevation and the restoration site trap within the 0–15-m range: −5–3 m Transect Difference in elevation between the forest NS NS elevation trap and the restoration site trap within the difference 60–75-m range: −13–13 m

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 58

3.4 Discussion

Jarrah seedling density at establishment in jarrah forest restoration sites is variable

(Norman & Koch, 2009; Stoneman & Dell, 1994); the potential for sites near remnant patches of jarrah forest to receive seedfall has been observed previously (Koch, 2007b; Ward et al.,

1990). This study has quantified this contribution of seedfall to these sites. Seedfall in jarrah forest stands varied by both forest stand (n = 28) and restoration site (n = 10; Figure 3.2;

Table 3.1 and there were 0–50,000 seeds present in restoration sites within 12.7 m of forest edge. This finding is aligned with the positive relationships between seedling densities and both proportion of forested perimeter and perimeter-to-area ratio (Chapter 2; Table 2.1). This study confirms the opportunity for restoration sites with increased perimeter-to-area ratios and proportions of forest perimeter to receive large amounts of seed within the 12.7 m seedfall zone (up to 50,000 seeds/ha) and this opportunity is likely to increase with increasing perimeter-to-area ratio and proportion of forest perimeter. However, the reliability of seedfall on site edges remains questionable because of annual fluctuations in seed production from jarrah forest stands and the unique shape and proximity to the forest of each restoration site.

The high heterogeneity of jarrah forest seedfall was strongly evident in this (0–50,000 seeds/ha/year; Figure 3.2; Table 3.1) and in other studies: 108,300–582,500 seeds/ha/year

(Cargill et al., 2016) and 20,000–850,000 seeds/ha/year (Abbott & Loneragan, 1986). The jarrah forest does not offer a spatially and temporally consistent seed supply (Figure 3.2;

Figure 3.3), and this has implications for both jarrah silviculture and recruitment of jarrah to restoration sites. The spatial separation of the canopy from the seedbed is a barrier to seedling establishment in restoration sites, although this is less relevant for studies of jarrah recruitment where the site for establishment is directly beneath the canopy (e.g. Cargill et al.,

2019). JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 59

This seed dispersal distance of 12.7 m confirmed in thi study is consistent with other studies of eucalypt species, which reported seed travelling up to 10 m from the forest

(Charles et al., 2017; Miolaro et al., 2017). Seed from the jarrah canopy adjacent to restoration sites is an important source of seed, but limited by distance to the forest edge, the only factor that helped explain the amount of seedfall (Table 3.1).The estimated average contribution of jarrah seed to restoration sites from canopy seedfall was 312 g/ha and a maximum restoration site seedfall of 50,000 seeds/ha/year was equivalent to the application of seed at 1,562 g/ha. In comparison, the jarrah seed application rate in 2016–17 was

285 g/ha (Table 2.1), illustrating that seedfall from adjacent forest could be as much as quadruple the amount of seed currently sown. Seedfall therefore offers a potentially important source of additional seed that can help to explain variability in observed seedling densities (Norman & Koch, 2009; Chapter 2).

The potential impact of canopy-stored seed on forest edges of restoration sites was tested by estimating the amount of seed supply to three sites, close in proximity but representing different jarrah forest types (Havel, 1975): Karri 17, Downes 2 and Rhodes

(Figure 3.5). These estimates were calculated from parameters measured in this study: the maximum seedfall distance (12.7 m), average seedfall (10,000 seeds/ha), maximum seedfall

(50,000 seeds/ha) and calculations related to the perimeters and areas of the restoration sites

(Table 3.3). These estimates relied on the presumption that seed production in all jarrah forest stands were consistent with the amount of seed production measured during this study, noting that the spatial and temporal variability in seed production can be high. Despite these three restoration sites having a high proportion of forest on their perimeter (52%–82%), they were estimated to receive seed across 10% to 23% of the total area (Table 3.3; Figure 3.5). The high heterogeneity in timing and amount of seedfall from forest canopy to restoration sites

(Figure 3.2; Figure 3.3) may not suggest canopy-stored seed as a reliable source of seed for JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 60 management use. However, this does provide an explanation for areas with especially high seedling densities that are present within 12.7 m of forest edge.

Table 3.3 Restoration site geometry (area and perimeter) and its impact on the area and amount of seed supplied from canopy-stored seed in jarrah forest stands adjacent to three restoration sites.

Total Total Forest Forested Area Estimated Estimated area perimeter perimeter perimeter receiving average max. seed (ha) (m) (m) (%) seed (%) number number of seeds of seeds Downes 2 8.24 1,266 652 52 10 8,280 41,402 Karri 17 8.93 1,967 1,620 82 23 20,574 102,870 Rhodes 1 16.47 2,773 1,991 72 15 25,286 126,429

Downes 2

Karri 17 Rhodes 1

Figure 3.5 Three jarrah forest restoration sites (names for identification to the left) illustrating total perimeter (orange) and forested perimeter (light green). The estimate of seeded area for each site was calculated from the forested perimeter and maximum seedfall distances of 12.7 m. The basemap was retrieved from GoogleMaps in QGIS.

Previously, the best indicator of jarrah seed dispersal was thought to be tree height

(Abbott & Loneragan, 1986). Based on the maximum tree height observed in this study

(30 m; Table 3.2), expected seedfall should have been up to 45 m into the restoration site. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 61

This is a significant over-estimate from the seed dispersal distance measured (12.7 m; Figure

3.4). The 45-m dispersal distance, based on maximum tree height, assumes the presence of a

30-m tree on the edge of the forest, but the LiDAR assessment of tree height did not include the proximity or distribution of these trees to forest edge, as reflected in the lower-than- expected dispersal distance. Another factor limiting seed dispersal distance was seed size

(e.g. Janzen, 1970). While marri seedfall amount was unexpectedly low (n = 4 seeds), maximum dispersal distance in this study was 6 m, reflecting the difference in seed size

(Johnstone & Kirkby, 1992) and therefore dispersal distance. Additionally, only jarrah seed

>4 mm was used in this study (consistent with Cargill et al., 2019); however, there is potential for smaller-sized viable jarrah seed (explored in Appendix 2). Smaller viable jarrah seeds would disperse farther into restoration sites than a 4-mm seed, but were not captured in this study. This relationship between minimum viable seed size and maximum dispersal distance of viable jarrah seed from forest edge will be important to explore in future work.

Restoration practices in Mediterranean-type ecosystems include the return of endemic species through the soil seedbank or from hand-sown seed supply and planting, when seeds are not present or do not persist in seedbanks (e.g. Koch et al., 1996; Rokich et al., 2000).

Jarrah and marri restoration in the Northern Jarrah Forest relies on hand sowing because these species are not represented in the seedbank (Koch, 2007b; Koch et al., 1996). The results of this study show jarrah seed entering 43% of restoration sites within 12.7 m of the forest edge

(Table 3.3), contributing a seeding amount in that area higher than the amount supplied by hand (contrast 312 g/ha of seed rain with a hand seeding rate of 285 g/ha). For example, a restoration site in this study (Rhodes 1; Table 3.3), with a total area of 16.47 hectares and forest covering 72% of its perimeter, could expect 15% of its total area to receive jarrah seed: a contribution of 126,429 seeds within that area (Table 3.3). There is a potentially large impact of seedfall on seedling densities, however there was high variability in the amount JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 62

(Figure 3.2), dispersal distance (Figure 3.4) and timing (Figure 3.3) of jarrah seedfall. For canopy-stored seed supply to exert a large impact on the seedling density of any given restoration site, an alignment of seed production by the adjacent forest stand and a high proportion of forested perimeter would be required. Therefore, these results offer a way for practitioners, despite otherwise uniform seed application during restoration practices, to explain high seedling densities in sites with a high proportion of forested perimeters, which has been an important long-term question for jarrah forest restoration. Future work should measure seedfall with more sampling within 12.7 m of the forest edge, explore the potential for viable jarrah seed smaller than four millimetres and measure seedfall over multiple years to inform predictions despite the irregular flowering cycle of both tree species.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 63

Flags (orange) marking emerged seedlings in a restoration site (from Section 4.3.2). Photo T.

White-Toney. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 64

Chapter 4: Suitability of

Establishment Microsites

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 65

4.1 Introduction

4.1.1 Scope

Understanding the factors that lead to variable tree seedling densities at establishment

(Norman & Koch, 2009; Chapter 2) in the Northern Jarrah (Eucalyptus marginata) Forest is a priority for improving post-mining restoration outcomes (Alcoa, 2016; Koch, 2007a).

Seedling recruitment is broadly limited by the amount of seed supplied and availability of suitable microsites (Duncan et al., 2009; Gomez-Aparicio, 2008; Harper et al., 1965), and the latter is the remaining perspective to be investigated in this study. Chapter 2 addressed broad- scale effects of management practices, seed supply and climate factors on seedling establishment at the scale of individual restoration sites. Chapter 3 further investigated the amount, viability and variability of seed sources to jarrah forest restoration sites. This chapter addresses the third objective of the overarching study: What physical and chemical microsite characteristics best explain seedling survival, mortality and ultimately density at establishment?

The main previous work investigating jarrah seedling establishment on post-mine sites in jarrah forest examined single restoration practices in isolation, and not the interaction between factors or the combined effect (Norman & Koch, 2009). Factors related to emergence, rather than establishment, were suggested to the cause of variability in seedling establishment densities based on limited significant relationships between ecological factors and seedling densities (Norman & Koch, 2009). Based on the limitation of previous field studies to examine interactions between ecological factors through fieldwork and the importance of microsite suitability for seedling establishment, three separate field trials were designed. These three experiments (A, B and C) considered the effect of restoration practices on suitability of jarrah and marri (Corymbia calophylla) seedling microsites, and the results are contextualised within the potential effect of climate (temperature and rainfall evenness), JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 66 as indicated to be important in Chapter 2. The restoration practices and microsite factors specifically investigated were substrate type, cover by debris (e.g. wood and rock fragments), soil chemistry and the soil ripping process (and the resultant microtopography). Importantly, this is the first field study to include a comparison of the ecological factors affecting establishment of both dominant tree species (jarrah and marri) on restored mine sites in the

Northern Jarrah Forest.

4.1.2 Background

Restoration practices are established based on broad criteria to achieve restoration goals, and variability within these criteria cause some of the variability in observed vegetation patterns on forest restoration sites (Grant, 2006; Macdonald et al., 2015; Standish et al., 2015). Mine site restoration in the Northern Jarrah Forest expects restored vegetation to match the composition, ecosystem values and management demands of the surrounding forest

(Gardner & Bell, 2007; Grant, 2006; Koch, 2007a). To better achieve this goal, restoration practices have evolved and been refined with technological advancement and ecological understanding of the region (Grant, 2006; Grant & Koch, 2007; Section 1.3.3). In an analysis of these broad-scale changes to jarrah forest restoration, topsoil handling practices (classified based on the thickness of topsoil spread and topsoil storage) and the timing of seed application were shown to influence understorey species richness (Standish et al., 2015).

However, no relationships were found between jarrah and marri establishment densities and restoration practices related to soil return (topsoil handling or the presence of the soil pathogen P. cinnamomi: Section 1.3.2; Chapter 2). The effect of changes in restoration practices were investigated in Chapter 2, but that study was unable to capture the potential importance of microsite suitably. Therefore, this study focuses on a fine-scale investigation of microsite suitability for jarrah and marri seedlings on restoration sites. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 67

Restoration practices play an important role in encouraging seedling recruitment and ameliorating the degradation and modification of natural landscapes (Jackson & Hobbs,

2009; Macdonald et al., 2015). Increasing seedling survival is particularly important in the context of Mediterranean-type ecosystems, which are threatened by worsening drought

(reduced rainfall and increasing temperatures; Bates et al., 2008) and intensive disturbances related to land use (de Dios, Fischer, & Colinas, 2007; Ruthrof et al., 2010). Typically, altered landscape states in Mediterranean ecosystems are the result of land conversion (e.g. mining, agriculture, grazing; Klausmeyer & Shaw, 2009; Le Houreou, 2000), weed invasion, fire management approaches and changes to hydrology (Ruthrof et al., 2013; Yates & Hobbs,

1997). Restoration practices, in particular deep soil ripping (Bocio, Navarro, Ripoll, Jiménez,

& de Simón, 2004) and soil nutrient amendments (e.g. fertiliser application, Rokich & Dixon,

2007; or topsoil and debris return, Macdonald et al., 2015) have been used with effect to increase the likelihood of seedling establishment after land clearing, such as mining. These practices increase seedling establishment by facilitating the seedling’s access to sufficient water to produce roots and leaves during the growing season and enable independence from maternal resources (Kollman, 2008; Section 1.4).

Native topsoils provide important nutrients and facilitate nutrient cycling, delivering organic matter and microorganisms to restoration sites, but supply of native topsoils to restoration sites can be limited by the timing and amount of forest cleared for mining (Grant,

2006; MacDonald et al., 2015). Nutrient supply to restoration sites is important in the

Northern Jarrah Forest where other (non-topsoil) natural substrates are acidic, iron-stone gravels, sands and clay (Gardner & Bell, 2007). Soil nutrients are added to the jarrah forest restoration system through return of forest topsoil (and associated debris) and the addition of nitrogen and phosphorous fertiliser (Koch, 2007a). The spreading of jarrah forest topsoil onto restoration sites varies both within and between restoration sites (Standish et al., 2015). This JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 68 small-scale difference in topsoil thickness and distribution, and the associated provision of carbon and other nutrients by soil and debris cover, influences the location and success of seedling establishment (Macdonald et al., 2015; Pinno, Landhaüsser, MacKenzie, Quideau &

Chow, 2012; Wolken, Landhaüsser, Lieffers, & Dyck, 2010). Therefore, when topsoil from forests is spread thinly and unevenly on jarrah forest restoration sites (e.g. Standish et al.,

2015), other substrate types, including overburden, gravel, pit floor and clay dominate in patches (Norman & Koch, 2009). The difference in chemical and nutrient composition of these soil types in the jarrah forest restoration system has not been explored. If substrates differ in the composition of soil nutrients, their heterogeneous distribution on jarrah forest restoration sites may be reflected in patchiness of jarrah and marri seedling establishment.

Soil nutrient amendments increase seedling establishment in Mediterranean ecosystem restoration sites, where soil nutrients are limited (Ruthrof et al., 2010; 2013;

Yates, Norton, & Hobbs, 2000). The optimal amount and timing of fertiliser application for jarrah forest restoration has been investigated in a number of studies, with conflicting results for overstorey and understorey species (Chapter 2; Daws et al., 2015). Reducing fertiliser application in jarrah forest restoration sites increases understorey species richness, which has prompted a delay in the timing of fertiliser application under current restoration practices in the Northern Jarrah Forest (Daws et al., 2015). However, additional fertiliser reduces mortality in jarrah seedlings (Stoneman et al., 1994) and increases jarrah seedling densities

(Koch & Ward, 2005) and heights (Lockley & Koch, 1996). This is reflected also in Chapter

2, in the positive relationship between fertiliser application and densities at establishment for both species. The restoration sites used for this fieldwork did not receive fertiliser during the timing of this study, as per current restoration practices (Table 2.1; Daws et al., 2015); therefore this work assessed the effect of unamended soil nutrients on establishment densities. If seedling establishment densities are increased in restoration sites with soil JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 69 containing higher levels of nitrogen and phosphorus (soil compounds known to be limited in jarrah forest soils; e.g. Hingston et al., 1982), those results would inform the need for fertiliser application to enhance jarrah and marri seedling establishment.

The heavy machinery involved in post-mining landscape contouring and soil return compacts the soil (Koch, 2007a), necessitating the ripping of the ground surface in new restoration sites to increase rainwater filtration, and consequently soil moisture and nutrient availability for seedlings (Bocio et al., 2004). Soil ripping increases survival, height and crown health of Eucalyptus species seedlings (Li, Duggin, Grant, & Loneragan, 2003;

Ruthrof et al., 2016). In jarrah forest restoration sites, the desired state following ripping is riplines that are ‘on contour’ (i.e. perpendicular to the downslope of the landscape), deeper than 1.2 m, contain no unblasted caprock and have greater than 90% topsoil cover (Grant,

2006; Koch, 2007a). However, restoration site preparation can deviate from this target because of the slope of the surrounding landscape and variability introduced by machine operators (Grant, 2006). During soil ripping, the tine separates the compacted soil surface, creating a furrow with tilled soil in banks and crests on either side of the ripline. Heavier, dense soil materials (e.g. clay) accumulate at the bottom of the furrows (Figure 4.1), creating differences in the physical characteristics of the resultant microtopographic locations.

Although there are three distinct locations on each rip line—furrow, bank and crest—the furrows are distinctly different from the banks and crests based on soil substrate. For this study, ripline microsites were defined by either location (three types: furrow, bank and crest) or substrate (two types: furrow and ‘bank-crests’) depending on the research question. These categories are both broadly referred to as ‘microtopography’. Previous work has found improved Eucalyptus species seedling establishment in non-compacted, deep sandy soils with reduced bulk density and increased water filtration similar to what occurs on clay-dominated JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 70 substrates (Ruthrof et al., 2013). Therefore, this study predicted seedlings would be more likely to establish in the clay-dominated furrows.

Figure 4.1 A post-mining restoration site, demonstrating the typical surface characteristics following ripping. The direction of the riplines is perpendicular to the downslope of the landscape. Clay accumulates in the furrows (e.g. centre middle), and banks and crests (e.g. bottom right) are dominated by variable amounts of other substrates including topsoil, overburden and occasionally pit floor. The surface can also be covered by woody debris, leaf litter and caprock.

The hydraulic properties of a substrate, and consequently the amount of soil water near the seed coat surface, affects the rate of water uptake by the seed, and therefore germination (Hadas, 1970). The gravimetric separation of substrate components between furrows and bank-crests (Figure 4.1) is likely to also influence the soil water-holding capacity of the two microtopography types. Although jarrah germination is stimulated by warm, wet conditions (McChesney et al., 1995; Section 1.5.3), waterlogging reduces jarrah seedling survival (Davison & Tay, 1985). Waterlogging is more likely in fine sands (Norman & Koch,

2009). Therefore, as furrows accumulate denser soil materials (i.e. clay; Figure 4.1) with higher water-holding capacity (e.g. Yates et al., 2000), it was predicted that there may be preferential survival and growth of seedlings in response to higher water availability in furrows. The positive relationship between seedling density and rainfall early in the wet JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 71 season (30-day rainfall evenness) was explained by rainfall stimulating jarrah and marri germination and emergence (Chapter 2; McChesney et al., 1995; Standish et al., 2015).

However, a negative relationship was identified between total wet season evenness and seedling densities, suggesting an adverse effect of water pooling with extended rainfall on tree seedling establishment densities (Chapter 2). This interaction between soil type, microtopography and water storage is also likely to be important for understanding jarrah and marri seedling establishment densities.

This study explored the relationship between restoration practices (related to ripping, topsoil return and soil nutrients), establishment microsite suitability and seedling development from seed to established seedling. The objectives of the experiments overlap and offered a comprehensive approach to reveal the importance of the targeted microsite factors on jarrah and marri seedling establishment. Experiment A examined the effect of ripline microtopography, soil chemistry and debris cover on seedling establishment.

Microtopography was then assessed in relation to differences in seedling emergence and establishment in Experiment B. Finally, Experiment C characterised differences in seedling emergence by substrate type.

4.2 Methods

4.2.1 Study site

These three experiments took place on two active mine sites in the Northern Jarrah

Forest: Huntly (32°42’S 116°03’E) and Willowdale (35°09’S 116°5’E; Figure 2.1).

Experiment A was performed at Huntly and Willowdale and Experiment B at Huntly and the substrates for Experiment C were collected from Huntly. These studies occurred between

September 2016 and October 2018. Descriptions are provided of the surrounding forest in

Sections 1.2 and 3.3.3.1, and restoration practices in Section 1.3.3. Experiment A was undertaken on restoration sites and received seed following Alcoa’s routine restoration JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 72 practices (e.g. Koch, 2007b), Experiment B occurred on restoration sites and the application of seed is described in Section 1.4.3, and Experiment C occurred in a glasshouse at Alcoa’s

Marrinup Nursery (10kms from the Huntly Mine) and the amount and location of seed was also set by experimental design. Therefore, the classification from A to C was from the least

(A) to most (C) different from standard restoration practices. These studies occurred between

September 2016 and October 2018 and supporting descriptions are provided elsewhere in the thesis for the surrounding forest (Sections 1.2 and 3.3.3.1) and restoration practices (Section

1.3.3).

Based on the importance of average temperatures and rainfall evenness (calculation described in Chapter 2; Bronikowski & Webb, 1996; Standish et al., 2015) on jarrah and marri seedling establishment densities (Chapter 2), rainfall and temperature experienced over the study period is presented in Table 4.1. Rainfall data for each mine site were collected from the closest rain gauge, and air temperature from the nearest climate station, in

Dwellingup (BOM, 2018; Figure 2.1). Average temperature and rainfall evenness for 1998–

2015 were compared with those for the 2016–17 (Experiment A) and 2017–2018 restoration seasons (Experiment B; Table 4.1).

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 73

Table 4.1. Summary of climate data comparing temperature averages and rainfall evenness from 1998 to 2015 with records for the 2016 (Experiment A) and 2017 (Experiment B) restoration seasons. The wet season was differentiated from the dry season as the period in each year when rainfall was greater than twice the mean temperature (see Chapter 2). Rainfall evenness from 0 (uneven rainfall, all on one day) to 1 (even rainfall across all days) was calculated using a modified measure of species evenness (Simpson’s Diversity Index; Bronikowski & Webb, 1996; Standish et al., 2015; Chapter 2). Evenness was calculated for periods of 0–30 days, related to germination and emergence (column 2), and over the duration of the total wet season (column 3).

Dry season mean maximum 30-day rainfall Total wet season

temperature (ºC) evenness rainfall evenness WILLOWDALE 1998–2015 29.0 0.22 0.21 2016 28.0 0.30 0.25 2017 28.2 0.13 0.20 HUNTLY 1998–2015 29.0 0.23 0.20 2016 28.0 0.27 0.23 2017 28.2 0.16 0.22

4.2.2 Experiment A: Effect of ripline microtopography, soil chemistry and debris cover on seedling emergence and establishment

The effect of soil chemistry, debris cover and ripline microtopography on seedling establishment was studied on five restoration sites established prior to May 2016 (4 months before the start of monitoring) at each active mine site. Sites were selected to ensure application of similar restoration practices; specifically, 10 mm of topsoil and sown with

285 g/ha of jarrah and 237 g/ha of marri seed. Topsoil used on all restoration sites in this study was classified as free of the soil pathogen P. cinnamomi based on jarrah forest dieback mapping (Koch, 2007b). Jarrah and marri seedlings were monitored in a total of 60 randomly located, 25 m2, north-facing plots for survival analysis (i.e. six plots per restoration site, and five restoration sites at each of the two mines). All seedlings present in a plot at the start of monitoring in spring (September 2016) were tracked until autumn (April 2017). Monitoring occurred seven times over the study period, with 7- 19 days between monitoring visits, JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 74 depending on the duration of previous monitoring. For each seedling, the emergence location was marked and mapped, and the species and position within the microsite topography were recorded. The total number of observation days in this study was 196.

The total number of jarrah and marri seedlings that survived were the response variables for separate GLMs for each species (Section 4.3.4). The 10 candidate factors used in the models were selected based on current understanding of restoration practice, and specifically related to substrate characteristics and ripping practices. The length of each ripline within each 25-m2 plot varied depending on their orientation with respect to the north- facing plot. Furrows were distinguished by the characteristic change in substrate texture: denser material consisting of finer particles (i.e. silt and clay; Figure 4.1). The width of each furrow was measured one to three times depending on the length of the ripline. The width was averaged per ripline and multiplied by the length to determine the area of the 25-m2 plot that was furrow (Figure 4.2). Slope was measured in 1-m increments upslope across the five m length of each, and averaged per plot. A visual estimate of percentage cover by leaf litter, rocks and woody debris (wider than 2 cm; Cargill, 2014) was made from four photos taken of

1 m2 quadrats from each corner of each 25-m2 plot. Estimates from the four samples were averaged. One soil sample was collected from each of three microsite locations—crest, bank and furrow—and bulked for analysis for each plot. Soil samples were dried at 30oC for three to seven days and sieved at one millimetre prior to a standard soil analysis (CSBP Soil and

Plant Analysis Laboratory, 2018). Soil analysis included determination of ammonium nitrogen (mg/kg), phosphorous (mg/kg), potassium (mg/kg), sulphur (mg/kg), organic carbon

(%), conductivity (dS/m) and water pH. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 75

1 m

Figure 4.2 Representation of a 25-m2 plot with four riplines. The percentage of the total area that was furrow (yellow) and bank-crests (white) was calculated from the mean ripline width (black dotted lines) and length (dark blue line).

4.2.3 Experiment B: Difference in seedling emergence and establishment by ripline microtopography

To study the effect of ripline microtopography on seedling emergence and establishment, five 625-m2 plots were established on a single restoration site and seed applied according to the design described below. Each plot was six metres wide and 10 riplines long with a south-west aspect and surrounded by a 5-m buffer zone to control for seed entering the sampling area during routine hand sowing to the remainder of the restoration site. Seed was applied to three riplines per plot, alternating with un-seeded riplines so that seed movement up and downslope could be identified. Each seeded treatment received four grams of jarrah seed mix and three grams of marri seed by hand to each location in the ripline: crest, furrow and bank per plot. Emergence density was estimated for each species from the average width of ripline furrows (method described for Experiment A; Figure 4.2) and the 6-m length of the JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 76 ripline. Seed was applied in April (autumn) 2017 and plots were monitored weekly until July

(winter) 2017 and then monthly to May (13 months later). The frequency of monitoring changed from weekly to monthly because of mine haul road closures in response to extreme rainfall events. Survival was tracked from seed supply (day 0) to a maximum of 377 days and analysed using survival analysis (Section 4.3.4).

4.2.4 Experiment C: Difference in seedling emergence by substrate type

The effect of four substrate types on seedling emergence was studied between May and June (winter) 2018. These substrate types included overburden, forest topsoil, gravel and mined pit floor; 40 L of each type was collected from the top 20 cm and sieved at two centimetres to remove large rock fragments and debris. Four replicates per substrate per species were added to nursery trays (34 × 28 × 5 cm), lined with perforated cloth and three to four cm of substrate added. There were also four unseeded replications from each substrate type to exclude seed storage in the substrate. Each seeded tray received five grams of jarrah seed and 10 grams of marri seed. Seed was applied on 2 August 2018 (winter), with emergence monitoring and watering occurring twice per week from 9 August to

18 October 2018 (spring). Trays were located within the glasshouse, following a randomised block design to control for edge effects (Figure 4.11; Appendix 5). All emerged seedlings were removed from the tray at the time of monitoring and recorded. Emergence per species was analysed using one-way ANOVA (p < 0.05) and a Bonferonni post-hoc analysis

(StataCorp, 2017). One sample of each substrate type was submitted for soil chemical analysis (CSBP, 2018) to understand difference in soil nutrients between these four substrate types and the potential effect on establishment densities and distributions. Collinearity between soil compounds limited further data analysis. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 77

4.2.5 Data analysis

GLMs with normal distributions and identity link functions within the statistical package Stata Ver. 15 (StataCorp, 2017) were selected to identify the explanatory variables in

Experiment A and their interactions, important for predicting jarrah and marri seedling densities per plot at establishment (Section 4.3.1). The saturated models included 10 candidate factors, and the least significant factors were sequentially removed until only significant factors remained in the final models. Only plots further than 13 m from forest edge were used for this analysis (n = 46) to control for the potential effect of seed rain on seedling density (Section 3.3).

Seedling survival, in both Experiments A and B, was analysed using Cox proportional hazards models (based on ecological applications of these methods, e.g. Tenhumber, Keller &

Possingham, 2001). These analyses determined the probability of seedling survival as a consequence of species and location within ripline microtopography (Figure 4.1). Seedlings were monitored between the start of the field trial (Experiment A, September 2016) or the timing of seed supply (Experiment B, April 2017) and the point of mortality or establishment in autumn (late April 2017 in A and early May 2018 in B), the typical timing of tree seedling establishment (Grant, 2006; Koch, 2007a; see Section 1.4). Each seedling was then monitored from its emergence to death or at a point when seedling establishment was presumed.

Kaplan–Meier failure curves were constructed to represent survival (mortality %).

4.3 Results

4.3.1 Climate

Dry season average maximum temperatures over both study years (28.0oC in 2016;

29.2oC in 2017) were consistent with historical averages (28.3oC). Thirty-day rainfall evenness was lower in 2017 at the location of the field study (Experiment B; Huntly: 0.16) JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 78 than the historical average (0.23), but wet season evenness across both mines and years was similar to historical averages (Table 4.1).

4.3.2 Experiment A: Effect of ripline microtopography, soil chemistry and debris cover on seedling emergence and establishment

The majority of seedling emergence (79.9%; n = 246) occurred prior to the start of monitoring in spring, with the latest emergence (1.2%; n = 3) noted after summer. Survival through the first summer (a critical period of establishment) was high for both species (Figure

4.3; 83.9% for jarrah; 96.8% for marri) and surviving seedlings of both species were present in the furrow and on the bank-crest locations. More seedlings of both species established in the furrow than in the bank-crest (Figure 4.4), but there were no differences in seedling survival according to microtopography (Cox analysis; p = 0.19; Figure 4.3). Average (± SE) emergence was 0.16 (± 0.03) jarrah seedlings/m2 and 0.17 (± 0.03) marri seedlings/m2 in furrows; and 0.09 (± 0.01) jarrah seedlings/m2 and 0.06 (± 0.01) marri seedlings/m2 on bank- crests (Figure 4.4). Seedling establishment densities varied from 0 to 3,200 jarrah seedlings/ha and 0 to 2,400 marri seedlings/ha. No significant relationships were identified between any of the explanatory variables and jarrah seedling density, but three factors explained 31.6% of the variability in marri density (pseudo R2 = 0.316; Table 4.2). Organic carbon (+1084.9 ± 261.2) and sulphur (+67.3 ± 25.4) had a positive relationship with marri density and conductivity had a negative relationship (–76,986.6 ± 28,035.4; Table 4.2). JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 79

Figure 4.3 Percentage mortality of jarrah (n = 174) and marri (n = 124) seedlings over 196 days prior to establishment, in the three microtopographic locations and across all 60 plots. Mortality is presented as Kaplan–Meier failure curves. Microtopography locations for seedling establishment were furrow (orange), bank (blue) and crest (green). There was no effect of establishment location on the mortality of tree seedlings (p = 0.19), irrespective of species.

0.3 Jarrah 0.3 Marri

)

)

2 2 0.2 0.2

0.1 0.1

Seedling density Seedling density (m Seedling density Seedling density (m

0 0 Furrow Bank-Crest Furrow Bank-Crest Establishment location Establishmnet location

Figure 4.4 Mean (± SD) jarrah (n = 174) and marri (n = 124) seedling density (per m2) at establishment in the furrow and bank-crest microtopography locations across all 60 plots. Seedlings per m2 was calculated from the proportion of the 60 25-m2 area restoration sites that were furrow. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 80

Table 4.2 Explanatory variables (left) investigated for potential use in the generalised linear models to account for cumulative jarrah and marri seedling density (seedlings/ha). The coefficients (β) and standard errors for the final models (right) indicate the extent and direction of the effect of each explanatory variable on seedling density; that is, the increase or decrease in seedling density (seedlings/ha) as a consequence of an increase by one unit in an explanatory variable. Variables with no coefficient are those that did not significantly (p > 0.05) improve the model (‘NS’). The significant explanatory variables explained around 32% of the variability in marri seedling density at establishment (pseudo R2 = 0.316).

Explanatory Description; range Jarrah Marri variable β coefficient (standard error) Average slope of Slope of the plot: 5.6–38.0o NS NS plot Total debris cover Visual estimate of total cover by all NS NS estimate debris: 0.03–0.2% Ammonium 1.0–4.0 mg/kg NS NS nitrogen Phosphorous 1.0–3.0 mg/kg NS NS Potassium 14.0–86.0 mg/kg NS NS Sulphur 2.6–22.2 mg/kg NS +67.3 (25.4) Organic carbon 1.1–3.3 % NS +1,084.9 (261.2) Conductivity 0–0.03 dS/m NS –76,986.6 (28,035.4) pH (H2O) 6.0–6.6 NS NS Percentage furrow Total furrow area/furrow area: 0.1– NS NS 0.5%

4.3.3 Experiment B: Difference in seedling emergence and establishment by ripline microtopography

Emergence began five weeks after seed was sown (late autumn) and even occurred in the following summer (i.e. after 38 weeks; Figure 4.5). Despite equal seed application across furrows, banks and crests, more seedlings emerged in furrows than on bank-crests: with 10.4 jarrah per m2 and 3.6 marri per m2 emerging in furrows compared to 2.4 jarrah per m2 and 0.5 marri per m2 emerging on crests (calculated from average furrow sizes; Section 4.3.3). JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 81 average emergence from seed was 6.3 ± 1.8 (SE) jarrah seedlings per gram and 1.9 ± 0.3 marri seedlings per gram.

Survival did not differ by emergence location (Cox analysis; p = 0.37) or species (Cox analysis; p = 0.92), but seed movement was observed (Figure 4.6). Kaplan–Meier survival estimates show high survival across planting and emergence location and species (Figures 4.7

& 4.8). Seeds moved between ripline locations, with 36.9% (n = 41) of jarrah and 15.4% (n =

4) of marri seedlings emerging in different ripline microtopography locations to those in which the seed was sown. Seed movement was classified as partial (Figure 4.6a) or complete

(Figure 4.6b) movement downslope or upslope (Figure 4.6c). Complete downslope movement was observed for marri seedlings (3.8%; n = 1; Figure 4.6b), but partial marri seed movement downslope was more common (11.5%; n = 3; Figure 4.6a). Higher occurrence of partial (16.2%; n = 18; Figure 4.6a) and complete (12.6%; n = 14; Figure 4.6b) downslope movement was observed for jarrah seeds than for marri. Additionally, partial upslope movement was observed for jarrah seeds (7.2%; n = 8; Figure 4.6c). Survival to establishment was high: 89.2% for jarrah and 93.8% for marri.

140 120 100 80 60

40 (number (number seedlings) of

Cummultive emergence emergence Cummultive 20 0 1 2 3 4 5 6 7 8 10 12 18 19 21 23 26 31 38 41 43 45 47 49 51 54 Number of weeks post seed application

Jarrah Marri

Figure 4.5 Cumulative seedling emergence for jarrah (n = 121) and marri (n = 32), across five plots, between April 2017 and February 2018. Seed was applied on 20 April 2017 (autumn) and first emergence was in week five (autumn). JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 82

Figure 4.6 Proportion (in brackets) of jarrah (n = 121) and marri (n = 32) seedlings that emerged in a location different (blue) from the location of seed application (yellow) across five plots. Bank (mid-slope) locations were considered anywhere between discrete crest and furrow locations, differentiated by substrate type and texture. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 83

Figure 4.7 Percentage mortality of jarrah (n = 121) and marri (n = 32) seedlings over 377 days prior to establishment, in the three microtopographic locations across five plots. Mortality is presented as Kaplan–Meier failure curves. Microtopography locations for seedling emergence were furrow (orange), bank (blue) and crest (green). There was no effect of emergence location on the mortality of tree seedlings (p = 0.37), irrespective of species. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 84

Figure 4.8 Percentage mortality of jarrah (n = 121) and marri (n = 32) seedlings over 377 days prior to establishment and five plots. Mortality is presented as Kaplan–Meier failure curves. There was no effect of establishment location on the mortality of tree seedlings (p = 0.92), irrespective of species.

4.3.4 Experiment C: Difference in seedling emergence by substrate type

The average (± SE) number of seedlings to emerge per gram of seed across all substrate types was 9.4 ± 0.5 for jarrah and 8.7 ± 0.5 for marri. Rate of emergence did not differ significantly across substrate type for marri (ANOVA, 15 df, p = 0.27) or jarrah

(ANOVA, 15 df, p = 0.12; Figure 4.9). Both species showed a similar pattern of emergence over time (Figure 4.10). The highest level of emergence for marri occurred in the pit floor, which was the substrate with lowest emergence for jarrah (p = 0.27; Figure 4.9). Of all substrate types, topsoil had the highest concentration of all compounds of all substrate types, except for sulphur, which was highest in the pit floor (Table 4.3).There was no emergence of JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 85 jarrah or marri seedlings in unseeded trays (n = 16, 4 per substrate type) and mean (± SE) emergence in potting mix was 54 ± 2.8 (SE) jarrah seedlings and 85 ± 3.5 marri seedlings.

140 120 100 80

60 SD) emergents emergents SD)

± 40 ( a 20

Mean Mean 0

Gravel Gravel

Topsoil Pitfloor Topsoil Pitfloor

Overburden Overburden Jarrah Marri

Figure 4.9 Mean total (± SE) emergence of jarrah and marri seedlings on four substrate types (n = 4). No values were significantly different (i.e. p > 0.05).

120 Jarrah 120 Marri

100 100

80 80

60 60

40 40

20 20

Cummulative mean number of emergents of number mean Cummulative Cummulative mean number of emergents of number mean Cummulative 0 0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 Weeks post seed application Weeks post seed application

Figure 4.10 Cumulative mean (± SE) number of emergent jarrah and marri seedlings on four substrate types over time (topsoil, green; overburden, yellow; gravel, orange; and pit floor, blue; n = four per substrate type).

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 86

Table 4.3 Chemical composition of the four typical restoration site soil types, which were assessed for their effect on the emergence of jarrah and marri seedlings. One sample from each substrate type was submitted for analysis; therefore, values are totals.

Unit Compound Topsoil Overburden Gravel Pit floor mg/kg ammonium 4.75 1.33 2.00 2.00 nitrate mg/kg nitrate oxygen 2.00 2.00 0.00 0.00 mg/kg phosphorous 2.75 2.50 0.00 0.00 mg/kg potassium 69.50 0.00 0.00 22.00 mg/kg sulphur 3.73 4.08 14.73 114.33 % organic carbon 2.99 0.94 0.37 0.13 dS/m conductivity 0.04 0.01 0.01 0.02 pH pH level (H2O) 6.53 6.45 6.60 6.00 mg/kg copper 1.08 0.21 0.27 0.10 mg/kg iron 49.04 19.62 8.90 12.10 mg/kg manganese 7.78 0.73 0.34 0.07 mg/kg zinc 0.33 0.18 0.28 0.10 mg/kg aluminium 0.10 0.17 0.07 0.07 mEq/100 g calcium 5.69 1.13 0.49 0.17 mEq/100 g magnesium 1.25 0.29 0.24 0.22 mEq/100 g potassium 0.15 0.02 0.03 0.04 mEq/100 g sodium 0.09 0.04 0.03 0.07 % total nitrogen 0.10 0.04 0.02 0.00

4.4 Discussion

The chemical and physical characteristics of microsites determine their suitability for seedling establishment (e.g. Harper et al., 1965) and this study furthers understanding of microsite suitability for jarrah and marri seedling establishment on jarrah forest restoration sites. Both the soil chemical properties and physical characteristics of the microsite, as a consequence of soil ripping, contribute to patterns of seedling densities at establishment on restoration sites. There were no differences in seedling establishment (survival) by microtopography location (Experiment A, Figure 4.3; and B, Figure 4.7), species JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 87

(Experiment B, Figure 4.8) or substrate type (Experiment C; Figure 4.9), and no physical or chemical characteristics had relationships with jarrah seedling establishment (Experiment A,

Table 4.2). Marri establishment densities had a positive relationship with carbon and sulphur, and a negative relationship with conductivity (Experiment A, Table 4.2).

Improved water penetration and loosening of the soil to encourage seedling root growth are well-established benefits of soil ripping on post-mining restoration in

Mediterranean ecosystems (e.g. E. blakelyi, Li et al., 2003; E. gomphocephala, Ruthrof et al.,

2016). Within ripline, differences in microtopography were predicted to increase seedling establishment in furrows where clay accumulates, as it has higher water retention than other substrate types, promoting seedling survival through summer drought (Yates et al., 2000).

This relationship was suggested also by the results of Chapter 2, in the relationship between early rainfall evenness and seedling density, and knowledge of jarrah’s requirement for warm, wet conditions for germination (McChesney et al., 1995) and differences in water storage of different substrates in the mining environment (Norman & Koch, 2009). However, no evidence to support this conclusion was found in this study, as survival did not differ based on microtopographic location (Figure 4.3).

From broadcast seeding (unknown distribution; Figure 4.4) and equal seed supply to each microtopographic location, higher seedling densities were found in the furrow than the bank-crest. Rather than confirming differential survival between ripline locations (e.g. no differences, Figure 4.3), this difference in distribution is most likely caused by seed movement downslope because of rain (Figure 4.6). This phenomenon is more likely to affect the smaller jarrah seed than it is for marri seed (Abbott & Loneragan, 1989; Johnstone &

Kirkby, 1992), as was observed in the pattern of seed movement: 83% of all seedlings emerging in different locations were jarrah (Figure 4.7). Most of this movement was downslope (Figure 4.6), explaining higher concentrations of jarrah and marri seedlings in JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 88 furrows—creating competition in such microtopographic locations—and distribution patterns. The upslope movement of jarrah seed (7.2% of seedlings; Figure 4.6) probably results from water pooling. High rainfall in the early wet season may cause water pooling in the furrows, causing jarrah seeds to float and be deposited further in the ripline bank, but not causing mortality. Therefore, the relationship with early rainfall and seedling establishment is positive (Table 2.1), especially with rainfall in the early wet season stimulating jarrah germination and emergence (McChesney et al., 1995). However, continued high rainfall may cause saturation of the ripline clay soils, which may decrease seedling emergence rates through waterlogging, known to influence jarrah seedling survival (Davison & Tay, 1985).

The total wet season rainfall observed during this field trial (Experiment B; Table 4.1) did not differ from the long-term average, meaning a negative effect of ongoing, heavy rainfall on seedling establishment as suggested in Chapter 2 may not have been experienced during the time of this field study. Thus, further exploration is required of the relationship between seedling survival and soil type, ripline microphotography and rainfall to explain seedling distribution, emergence and establishment on mine sites undergoing restoration.

Jarrah forest topsoil returned to restoration sites contains the nutrients required for seedling establishment (e.g. Koch et al., 2009), but the spread of topsoil is not always consistent in thickness or distribution on jarrah forest restoration sites (Grant, 2006; Standish et al., 2015). Therefore, other substrates can dominate in small pockets across the heterogeneous restoration site surface. Differences in soil substrate characteristics including soil texture and chemistry were expected to cause differences in emergence because of differences in soil microbes in forest topsoil compared with other soil types (e.g. MacDonald et al., 2015). Although not significantly different, the highest number of marri seedlings emerged in the pit floor substrate (Figure 4.9). Pit floor was also the soil substrate with the highest sulphur levels (Table 4.3), whereas topsoil had the highest concentration of nearly all JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 89 other soil compounds (but not pH, which was highest in gravel; and aluminium, which was highest in overburden; Table 4.3). Marri seedling density at establishment had a positive relationship with sulphur (Table 4.2), explaining this increased emergence on the pit floor substrate. Sulphur is required for the biosynthesis of proteins, co-enzymes, vitamins and amino acids (Gigolashvili & Kopriva, 2014), and its uptake by plants is controlled in response to changes in nutrient demand induced by seedling development or environmental conditions (e.g. Yoshimoto, Takahashi, Smith, Yamaya, & Saito, 2002). The relationship between soil sulphur and marri establishment is an important starting point for future work.

Marri establishment also increased with organic carbon (Table 4.2), which was in highest concentrations in topsoil (Table 4.3), and is often used as a proxy for soil quality and ecosystem recovery in post-mining soils (Frouz, Pizl, Cienciala, & Kalcik, 2009; MacDonald et al., 2015; Turcotte, Quideau, & Oh, 2009). The increased establishment of marri in substrates containing increased sulphur (pit floor) and organic carbon (topsoil), coupled with the heterogeneity of soil substrate deposition on restoration sites (e.g. Grant, 2006), may explain some variability in marri density and distribution based on the distribution of substrate types on the soil surface.

Increased soil conductivity has a negative relationship with other woody plant seedlings, as observed for marri (Table 4.2). The presence of more ions in the soil lowers the water potential, osmotic pressure and nutrient update, and therefore root function and elongation (Allen, Chambers, & Strine, 1994; Baligar, Fageria, & Elrashidi, 1998; Jacobs &

Timmer, 2003). Eucalyptus species have differential survival, height and health in altered saline environments, which is reflected in their natural distribution in environments with high salinity (Pepper & Craig, 1986). Waterlogging and salinity have similar effects on the survival of eucalypt trees; under salt stress, stomatal conductance and transpiration are reduced (Andrade et al., 2019; Morris, 1980; Parsons, 1968). In Western Australia, water JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 90 tables are expected to keep rising, and soil salinity and waterlogging are important threats to survival of vegetation (National Land and Water Resources Audit, 2001). Marri seedlings undertake osmotic adjustment (Szota et al., 2011) and therefore have higher tolerance to waterlogging and drought than does jarrah. However, marri’s susceptibility to soil salinity, indicated by the negative relationship between marri establishment and soil conductivity, requires further exploration. Differences in response to changes to the jarrah forest environment have important implications for restoration and management of jarrah forest and anticipating shifts in vegetation (e.g. Batini, 2007).

Factors related to emergence rather than establishment have been previously suggested as the primary cause of observed establishment densities of jarrah and marri on jarrah forest restoration sites (Norman & Koch, 2009), although variability in establishment remains largely unexplained by restoration practices (Chapter 2). This study identified relationships between ecological factors related to microsite characteristics resulting from restoration practices that influence seedling establishment in other Mediterranean regions: nutrient amendments (e.g. Ruthrof et al., 2013; Yates et al., 2000), topsoil supply (e.g.

Standish et al., 2015) and the soil ripping process (e.g. Bocio et al., 2004). Marri seedling establishment increased with sulphur and organic carbon, which were highest in pit floor and topsoil, respectively. The heterogeneous distribution of soil substrates on the restoration site, coupled with increased establishment of marri within pit floor and topsoil substrate types, may be partially responsible for variability in the number and location of marri seedlings both between and within restoration sites. For both species, higher establishment for both species in furrows than on bank-crests, created by the soil ripping process, is explained by downslope seed movement, rather than increased survival in furrows as a result of increased water accessibility as was predicted (e.g. Yates et al., 2000). While interactions between climate conditions and establishment were not directly measured in this study, as in other JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 91

Mediterranean systems rainfall and temperature are important aspects of jarrah forest seedling establishment (Chapter 2). Most relationships with the ecological factors explored here (e.g. soil nutrients and seed movement) had an enhancing effect on seedling establishment when considering the added effect of rainfall and temperature on the restoration system. Few factors related to microsite conditions explained variability in seedling establishment (Table 4.2), suggesting the need to consider establishment limitation according to seed supply amounts (Chapter 2; Cargill et al., 2019) and viability (Section 3.2) over microsite conditions for jarrah and marri recruitment. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 92

Jarrah tree marking separation between restoration site (left) and forest (right). Photo T.

White-Toney. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 93

Chapter 5: Synthesis JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 94

5.1 Rationale

This study investigated the ecological factors and restoration practices contributing to variable establishment of jarrah (Eucalyptus marginata) and marri (Corymbia calophylla) seedlings on bauxite mine restoration sites in the Northern Jarrah Forest of south-western

Australia. Seedling establishment was explored within the framework of the limitations set by seed supply and the number of suitable microsites (Aicher et al., 2011; Duncan et al., 2009;

Eriksson & Ehrlén, 1992). The questions addressed in the study stemmed from this framework, and were:

1. Which ecological factors, including restoration practices, are most important for

explaining long-term monitoring data of seedling densities at establishment? (Chapter

2)

2. To what extent is seed entering the mine restoration site from each of broadcast

sowing and seedfall from the forest canopy? And what is the effect of these seed

sources on variability in seedling density at establishment? (Chapter 3)

3. What physical and chemical microsite characteristics are most suitable for seedling

densities at establishment? (Chapter 4)

Establishment is marked by the survival of a seedling through the first summer drought, at which point the seedling has been recruited to the population (defined in Section

1.4.2). Recruitment requires the flowering, pollination, setting and dispersal of viable seed, and for the seed to evade predation and encounter conditions adequate for germination, emergence and seedling survival (Harper, 1977). All of these processes influence seedling establishment and can be separated into five categories in the jarrah forest restoration context:

A) natural process of flowering and seed release; B) conditions created by restoration practices; C) interactions with fauna; D) climate conditions; and then E) successful germination, emergence, survival and establishment. These categories of factors and their JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 95 interactions are presented in Figure 5.1 as a summary of current knowledge of jarrah and marri establishment, including that generated in this study. Figure 5.1 provides the structure for a review of this study and its significance.

Seed release occurs with heating and drying of the capsule (Chapter 3) A) Flowering to seed release Seed B) Restoration practices production Canopy- mediated Pollination stored seed (Yates et. al, 2005; Marri: positive effect by high Brown et al., 1997) as seed Physical and wet season source chemical of a longer delay between seedbed rainfall two (Chapters 2,3) properties of years prior the soil preparation and timing (Chapter 2) affect of the wet season microsite (Chapter 2) suitability Jarrah: higher seedfall (Chapter 4) into restoration sites during warm months (Chapter 3)

C) Fauna D) Climate

Establishment Predation of Seed and seedling Stimulates influenced by seed and presence germination maximum and seedlings encourages faunal (Abbott, 1984; McChesney et al., (Majer, 1980; minimum colonisation (Craig 1995) Stoneman & Dell, et al., 2015) temperatures and 1994) rainfall evenness (Chapter 2)

E) Survival from seed to seedling

Factors affecting seed viability (Chapter 3), germination (McChesney et al., 1995), emergence (Chapters 2, 4), and survival to establishment (Chapters 2, 4)

Figure 5.1 Summary of factors influencing jarrah and marri seedling establishment and their contribution to establishment limitation (A–E). Single-headed arrows indicate one-way interactions (causation) and double-headed arrows represent a two-way relationship between factors. Relationships identified in this study (blue) and in others (grey) are explained in the writing adjacent to arrows. The broad relationship between climate and fauna is indicated with an arrow and dotted line without description. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 97

5.1.1 Flowering to seed release

Flowering and seed release are clear determinants of natural seedling recruitment densities and distributions, and the effect of seed supply from canopy-stored seed was investigated as part of objective 2 in this study. Individual jarrah trees flower once every four to seven years (Abbott & Loneragan, 1986) and seed production amount and seed viability vary among individual trees and are dependent on available environmental resources (e.g. soil nutrients and water) and climate (Chapter 3; Florence, 2004). The important effects of seed supply on seedling densities at establishment has been found previously for jarrah: in this context of unmined forest, the single most significant factor influencing jarrah emergence and establishment was seed supply (Cargill et al., 2019). The importance of seed supply from jarrah trees adjacent to the cleared restoration sites has been inferred previously, but not measured (Koch, 2007b; Ward et al., 1990), and this study was the first to predict (Chapter 2) and measure the amount of seed entering restoration sites and indicate its effect on seedling densities at establishment (Chapter 3). Increasing the proportion of forested perimeter on a restoration site from 0 to 100% increases jarrah seedling density at establishment from

(approximately) 800 to 1,200 stems/ha and marri density from 300 to 440 stems/ha (Figure

2.3). Fieldwork following this estimation found 10,000 ± 2,624 (mean ± SE) jarrah seeds/ha/year within 12.7 m of the forest edge. Nearly half (43%) of seed trap transects received seed, confirming the spatial variability of the contribution of canopy-stored seed rain to seedling establishment densities, both within and between single restoration sites (Chapter

3). Jarrah seed production in forest sites (n = 26) was variable: 0–540,000 seeds/ha/year.

However, despite presence of marri in forest stands, there were too few marri seeds captured to perform any substantial analysis (n = 4). Despite the positive relationship between forest perimeter and marri density at establishment (Chapter 2), the lack of marri seedfall suggests that the larger more energy-costly seed of marri (Johnstone & Kirkby, 1992; Figure 1.4) is JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 98 likely to fall as seed rain in restoration sites less frequently than does jarrah seed. Therefore, marri seed sourced from the adjacent forest canopy is less likely to be an important influence on observed densities at establishment. These results relate to the limitation of establishment by seed source and highlight, along with Cargill et al. (2019), that seed supply is responsible for observed variability in seedling densities for jarrah.

5.1.2 Restoration practices

Long-term restoration monitoring data were used to identify positive relationships between both jarrah and marri seedling densities at establishment, and seeding rate, forest perimeter and fertiliser (Table 2.1). In fulfilment of objective 1, at a broad scale, these results confirm which restoration practices most influence seedling establishment through microsite suitability (objective 3) and seed supply (objective 2). Sown seed is considered the primary source of jarrah and marri seed during mine restoration (Koch, 2007a). Increased seed supply during restoration increased both jarrah and marri densities at establishment (Figure 2.2). The presence of seed in the seedbank was not considered in this study because previous work indicated limited storage of jarrah and marri seed in topsoil, especially compared with understorey species (e.g. Koch et al., 1996). This is consistent with other studies of jarrah recruitment, where seed sourced from the soil seedbank has been considered negligible (e.g.

Cargill, 2014). It is therefore not surprising that while ‘optimal’ topsoil handling practices significantly were important for understorey species with long-lived seed (Standish et al.,

2015), topsoil handling was not important for jarrah or marri (Table 2.1).

Microsite conditions for jarrah and marri establishment were explored via objective 3.

Similar to previous studies investigating soil bed suitability for jarrah (e.g. Cargill et al.,

2019; Norman & Koch, 2009), this study did not identify significant relationships with many physical or chemical properties of microsites for jarrah or marri establishment. The relationship between jarrah and marri establishment and soil nutrition is complex: this study JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 99 identified a positive relationship between fertiliser application and establishment of both species (Chapter 2) as well as relationships between marri establishment and sulphur (+), organic carbon (+) and conductivity (–; Table 4.2). However, there was no difference in the emergence of seedlings across the four typical substrate types found on Alcoa's restoration sites (Figure 4.10), despite differences in soil characteristics and chemistry (Table 4.3). The relationship between marri density and sulphur, and increased emergence of marri in pit floor

(NS, Figure 4.10) with high sulphur levels (Table 4.3) requires future investigation. The microtopographic locations created by ripping influence seedling distribution, with more established jarrah and marri seedlings in furrows than on banks and crests (Figure 4.4).

However, there was no relationship between seedling survival and microsite location (Figure

4.7), indicating that seed movement rather than differences in survival was the cause (Figure

4.6).

Coupled with the timing of natural recruitment of jarrah and marri, two factors were observed as important for restoration practices to increase seedling densities. For marri, a longer delay between seed supply to restoration sites and the start of the wet season (cueing germination) decreases marri establishment (Chapter 2). This may be because of higher levels of predation (Section 5.1.3) or temperature and moisture extremes on the soil surface, and should be considered with regard to timing of restoration practices and/or it may justify shallow seed burial. For jarrah, seedfall from adjacent forest to restoration sites occurs during warmer months (Figure 4.5), so early seedbed preparation during warm months may increase the amount of seed entering the restoration system as canopy-stored seed, which then increases seedling establishment densities. These results use knowledge of the natural recruitment process of jarrah and marri to inform restoration practice and explain observed variability due to the spatial variability in seed supply. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 100

5.1.3 Fauna

Interactions between both vertebrates and invertebrates and jarrah and marri occur at all life stages (seed, seedling and adult tree) and influence both seed supply (objective 2) and microsite suitability (objective 3). While some fauna species enhance seed supply by pollination (see Section 1.6.2), others decrease the likelihood of establishment success through foraging and seed theft from restoration sites. Ants are the most common predator of eucalypt seeds following seed dispersal (Yates et al., 1995). In a study of seed theft by ants in the Northern Jarrah Forest, 3–9% of jarrah seed was removed by ants in 24 hours, which was low compared with 48–66% of Acacia extensa seeds. Therefore, with spatially heterogonous dispersal of seeds in a restoration site, ants are likely to gather jarrah seed even less efficiently than in the forest (Majer, 1980). Vertebrates consume both jarrah and marri seed from capsules (e.g. Johnstone & Kirkby, 1992) but the effect of this on restoration site seed supply has not been described. Vertebrates also interact with seedling establishment by foraging: in one study, caged plots to exclude vertebrates had significantly more jarrah emergents (Stoneman & Dell, 1994). Therefore, interactions with biota have a complex relationship with jarrah and marri seedling establishment. The presence of both vertebrates and invertebrates is required for pollination but may contribute to reduced seed supply or establishment failure from foraging. These interactions were outside the scope of this study, but are important considerations for explaining seedling densities at establishment.

5.1.4 Climate

Every process in seedling recruitment from flowering to establishment is mediated specifically by rainfall and temperature. Therefore, climate was a consideration in all objectives. Rainfall, temperature and photoperiod are the main three climate factors influencing flowering (Rathcey & Lacey, 1985). Therefore, there is a direct relationship between climate factors and flower production, which then affects plant–pollinator JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 101 relationships (Stenseth & Mysterud, 2002) and seed production (del Cacho, Estiarte,

Peñuelas, & Lloret, 2012). This study identified a positive relationship between rainfall 2 years prior to seedling establishment (Table 2.1), where increased rainfall increases flower production and then seedfall (as suggested by Johnstone & Kirkby, 1992). Additionally, heat from fire or warm temperatures induces the opening of capsules to release seed (Boland et al.,

1980); therefore, the majority of seedfall for eucalypt species occurs in warm summer months

(Chapter 3). Similarly, the proportion of viable seed is dependent on the environmental conditions—especially climate and soil characteristics—experienced by the parent tree

(Boland et al., 1980; Penfield & MacGregor, 2017). Thus, seed supply from parent trees to the restoration site affects the later observed seedling densities at establishment, specifically in the amount, timing and viability of seedfall. This study observed seedfall of jarrah seed over one single restoration season, which provides valuable but limited ability to estimate seedfall, given high variability between and within jarrah forest sites (Cargill et al., 2016).

Water availability is considered a primary constraint on seedling survival in

Mediterranean ecosystems (Larcher, 2000) and rainfall evenness was an important factor for seedling density at establishment of both species (Table 2.1). Increasing wet season rainfall evenness over the wet season decreased density at establishment and was suggested to be caused by the over-saturation of water in soils (Chapter 2), yet there was no difference in mortality in relation to microtopography or in emergence in relation to substrate type, expected to be related to water storage capacity (Chapter 4). The effect of winter rainfall unevenness (contrast 0.16 in 2017 with 0.23 for 1998–2015) was a possible contributing factor to the lower-than-expected seedling emergence rates (2.0 ± 1.1 jarrah seedlings/g and

0.7 ± 0.2 marri seedlings/g; Chapter 4). However, the ability to assess the effect of climate conditions on germination, emergence and establishment was temporally limited in this study because the field trials all occurred within single field seasons. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 102

Long-term monitoring data on establishment identified more relationships with jarrah and marri seedling establishment and climate factors than restoration practices (Table 2.1).

There were relationships between seedling densities at establishment, illustrating threats to seeding establishment with both increasing (negative relationship with dry season average maximum temperature) and decreasing temperatures (positive relationships with lowest temperatures in both wet and dry seasons) for both species (Table 2.1). Previously, high and low temperature were the only factors significantly explaining jarrah density at establishment on mine restoration sites (Norman & Koch, 2009), consistent also with the results of this study. Increasing temperatures are expected in the jarrah forest region (e.g. Bates et al.,

2008), and the altered conditions at restoration sites expose seedlings to higher temperatures than seedlings establishing under covered, forested conditions (Stoneman & Dell, 1994).

There are implications for management of the restored forest based on the results of this study and previous work on the effect of both increasing (Matusick et al., 2013; Ruthrof et al., 2015; Section 1.2.1.1) and decreasing (e.g. frost; Matusick et al., 2014) temperatures on adult jarrah and marri trees.

5.1.5 Seed to seedlings

Following seed application, germination, emergence and survival to establishment must occur. Germination was not explored in this study; however, the use of X-rays to assess viability proved to have merit, particularly for jarrah (Chapter 3). This work highlights the importance of clear descriptions of seed purity and particle size for jarrah seed mixes supplied for restoration (as discussed in Section 3.2). In this study, mean seedling emergence in the field (6.3 ± 1.8 (SE) jarrah seedlings/g and 1.9 ± 0.3 marri seedlings/g; Chapter 4,

Experiment B) was lower than in the substrate trial (9.4 ± 0.5 jarrah seedlings/g and 8.7 ± 0.5 marri seedlings/g; Chapter 4, Experiment C). This low field emergence could be related to low rainfall evenness (Chapter 2), low seed viability in the seed mix (Chapter 3) or increased JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 103 variability in field environmental conditions (Hegarty, 1977). Survival between emergence and establishment, however, was high: 86.6% ± 1.9% for jarrah and 95.3% ± 1.1% for marri

(Chapter 4). Jarrah germination and emergence on post-mining restoration sites is variable

(2–38%; Koch & Ward, 2005; McChesney et al., 1995; Norman & Koch, 2009). Therefore, future investigations into jarrah and marri seedling establishment should continue to explore the ecological factors that stimulate germination and emergence, rather than acting on survival of the emerged seedling through summer to establishment.

5.2 Management Applications

The findings from this study have broadened our knowledge of the factors that influence seedling densities of jarrah and marri seedlings related to both microsite locations and seed supply. While they have application to restoration managements, it was not within the scope to re-evaluate post-mining restoration practices in the Northern Jarrah Forest, rather explain establishment variability in jarrah and marri Through the thesis the relevance to restoration has been highlighted and collectively gathered together and discussed here.

With continued clearing, mining, and restoration of the jarrah forest, responsibility lies with restoration practices to produce resilient jarrah forest stands (Bell & Hobbs, 2007), which specifically includes further understanding the dynamics of marri seedling recruitment.

This study measured low seed supply from forest edges to restoration sites for marri. The design for Chapter 3 was limited to only considering jarrah seed dispersal, therefore, the process of flowering to seed production, and its link to seed supply to the restoration site, requires further investigation for marri. The lack of marri seeds suggests the sampling period occurred during a year of low seed production for marri trees in the Northern Jarrah Forest, requiring sampling of marri seedfall over a longer time period and with altered seed collection methods. Importantly, in all chapters, this work presents differences in the response of jarrah and marri to the same ecological stimuli. Future work on tree JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 104 establishment in jarrah forest restoration sites should not focus solely on jarrah as in most previous studies, because differences in the two dominant species responses are evident as seedlings and adult trees. Recent studies in adult jarrah forest have found higher drought tolerance of marri trees compared with jarrah, related to physiological differences. Marri crowns completely collapse only on sites severely affected by drought, whereas jarrah trees more regularly respond to even less severe drought conditions with complete crown dieback and, if they survive, resprouting (Matusick et al., 2013). Given the expected continued increase in temperatures and decrease in rainfall for the Northern Jarrah Forest (Section 1.2.1;

Bates et al., 2008), the change to composition of forest vegetation is unknown, but more sustained marri tree crowns could lead to improved marri flower production, and subsequent increased pollination and seed production by marri trees compared with jarrah (Ruthrof et al.,

2015). Forest composition change like this would affect the amount of jarrah and marri seed supplied to restoration sites from forest edges, reflected then in seedling densities at establishment.

Marri seedlings showed a stronger response to available soil compounds (organic carbon, sulphur and soil ions measured as conductivity) than did jarrah seedlings (no significant relationships with soil compounds). Despite this, the relationship between soil nutrients, the need for soil amendments (topsoil return and fertiliser application) and jarrah and marri establishment remains unclear. The limitation of seedling establishment based on type and patchiness of soil substrates, and therefore nutrient supply, to restoration sites is another aspect highlighted in this study, specifically for marri seedling establishment.

Previous field studies have not clearly identified a relationship between tree seedling establishment and various fertiliser treatments (Section 4.2); however, increasing fertiliser increases both jarrah and marri seedling densities when applied prior to seedling establishment (Table 2.1). Later fertiliser application increases establishment of understorey JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 105 species (Daws et al., 2015; Standish et al., 2015). Therefore, the relationship between fertiliser application and seedling establishment requires further exploration to determine the optimal timing and amount of fertiliser application to maximise seedling establishment of all jarrah forest species seedlings. Finally, an important result was the ability for this work to explain variability in seedling establishment densities with variability in jarrah seed supply to the jarrah forest restoration site. Most previous work on jarrah seedling establishment on restoration sites focused on describing the suitability of microsites for seedling establishment.

However, this work has indicated the importance of seed supply. The mosaic of mine, restoration and forest sites in the Northern Jarrah Forest offers the opportunity for greatly increased seed supply to restoration sites along forest edges. On any one restoration site, there is variability in the amount of forest along the perimeter, the site’s total perimeter-to-area ratio and the dispersal distance of jarrah seed from forest edges. These three factors play an important role in the variable supply of jarrah seed, both within and between restoration sites, as reflected in variability in seedling densities, despite consistency in amount of seed sown by management.

Ultimately, the results of this study have furthered the understanding of primary ecological factors affecting the establishment of jarrah and marri on restoration sites in the

Northern Jarrah Forest. As with all research, new questions are generated and this study proposes that future work should 1) investigate both tree separately; 2) be related to specific interactions between restoration practices and climate conditions experienced by seedlings; 3) further explore the importance of variability in canopy-stored seed dispersal on observed patterns and densities of established seedling and 4) re-evaluate restoration management with the findings of this study in mind.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 106

References

Abbott, I. (1984). Emergence, early survival, and growth of seedlings of six tree species in

Mediterranean forest of Western Australia. Forest Ecology & Management, 9, 51–66.

Abbott, I., & Burrows, N. (eds) (2003). Fire in Ecosystems of South-West Western

Australia: Impacts and Management. Backhuys, Leiden.

Abbott, I., & Loneragan, O. W. (1986). Ecology of jarrah (Eucalyptus marginata) in the

northern jarrah forest of Western Australia (Department of Conservation & Land

Management Bulletin No. 1). Perth, WA.

Abbott, I., & Williams, M. R. (2011). Silvicultural impacts in jarrah forest of Western

Australia: Synthesis, evaluation, and policy implications of the FORESTCHECK

monitoring project of 2001–2006. Australian Forestry, 74, 350–369.

doi:10.1080/00049158.2011.10676378

Aicher, R. J., Larios, L., & Suding, K. N. (2011). Seed supply, recruitment and assembly:

Quantifying relative seed and establishment limitation in a plant community context.

American Naturalist, 178(4), 464–477.

Alcoa. (2016). Appendix A: Completion criteria for 2016 onwards—Current era

rehabilitation. Retrieved from https://www.alcoa.com/australia/en/pdf/mining-

operations-rehabilitation-program-completion-criteria-overview-area-certification-

process.pdf

Al-Hammond, B. A., & Al-Ammar, B. S. (2017). Seed viability of five wild Saudi Arabian

species by germination and X-ray tests. Saudi Journal of Biological Sciences, 24,

1424–1429. doi:10.1016/j.sjbs.2017.04.004

Allen, J. A., Chambers, J. L., & Strine, M. (1994). Prospects for increasing the salt tolerance

of forest trees: A review. Tree Physiology, 14, 843–853. doi:10.1093/treephys/14.7-8-

9.843 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 107

Andrade, J. R. de, Júnior, S. de O. M., da Silva Santos, A. F., da Silva, V. M., Bezerra, L. T.,

da Silva, J. R. R., … Endres, L. (2019). Photosynthetic performance in Eucalyptus

clones cultivated in saline soil. Emirates Journal of Food & Agriculture, 31(5), 368–

379. doi:10.9755/ejfa.2019.v31.i5.1955

Baligar, V. C., Fageria, N. K., & Elrashidi, M. A. (1998). Toxicity and nutrient constraints on

root growth. Horticulture Science, 33, 960–965.

Barker, J. (2019). Eucalyptus marginata: Seed purity, viability and germination.

(Environmental Research Note). Unpublished report. Pinjara, WA: Alcoa of

Australia.

Baskin, C. C., & Baskin, J. M. (1998). Seeds: Ecology, biogeography and evolution of

dormancy and germination. San Diego, CA: Academic Press.

Baskin, C. C., & Baskin J. M. (2007) A revision of Martin’s seed classification system, with

particular reference to his dwarf-seed type. Seed Science Research, 17, 11–20.

doi:10.1017/S0960258507383189

Bates, B. C., Hope, P., Ryan B., Smith I., & Charles S. (2008). Key findings from the Indian

Ocean Climate Initiative and their impact on policy development in Australia.

Climatic Change, 89, 339–354. doi:10.1007/s10584-007-9390-9

Batini, F. (2007). The response of Australian forests to a drying climate: A case study of the

jarrah forest. Australian Forest, 70(4), 213–214.

Battaglia, M., & Reid, J. B. (1993). The effect of microsite variation on seed germination and

seedling survival of Eucalyptus delegatensis. Australian Journal of Botany, 41, 169–

181. doi: 10.1071/BT9930169

Bell, D. T., & Heddle, E. M. (1989). Floristic, morphologic and vegetational diversity. In B.

Dell, J. J. Havel, & N. Malajczuk (Eds.), The jarrah forest: A complex Mediterranean

ecosystem (pp. 53–66). Dordecht, UK: Kluwer Academic. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 108

Bell, D. T., & Hobbs, R. J. (2007). Jarrah forest ecosystem restoration: A foreword.

Restoration Ecology, 15, S1–S2. doi:10.1111/j.1526-100X.2007.00286.x

Bewley, D. (1997). Seed germination and dormancy. Plant Cell, 9, 1055–1066. Retrieved

from https://www.jstor.org/stable/41433778

Bocio, I., Navarro, F. B., Ripoll, M. A., Jiménez, M. N., & De Simón, E. (2004). Holm oak

(Quercus rotundifolia Lam.) and Aleppo pine (Pinus halepensis Mill.) response to

different soil preparation techniques applied to forestation in abandoned farmland.

Annals of Forest Science, 60, 171–178. doi:10.1051/forest:2004009

Bohte, A., & Drinnan, A. (2005). Ontogeny, anatomy and systematic significance of ovular

structures in the ‘eucalypt group’ (Eucalypteae, Myrtaceae). Plant Systematics &

Evolution, 255, 17–39. doi:10.1007/s00606-004-0291-3

Boland, D. J., Brooker, M. I., & Turnbull J. W. (1980). Eucalyptus seed. Canberra: CSIRO

Division of Forest Research.

Bond, W. J., & van Wilgen, B. W. (1996). Fire and plants. London: Chapman & Hall.

Bureau of Meteorology. (2007). Annual climate summary 2007. Retrieved from

http://www.bom.gov.au/climate/annual_sum/2007/AnClimSum07.pdf

Bureau of Meteorology. (2018). Annual climate summary 2018. Retrieved from

http://www.bom.gov.au/climate/current/annual/aus/

Bradshaw, F. J. (1999). Trends in silvicultural practices in the native forests of Western

Australia. Australian Forestry, 62(3), 255–264.

doi:10.1080/00049158.1999.10674790

Bronikowski, A., & Webb, C. (1996). Appendix: A critical examination of rainfall variability

measures used in behavioral ecology studies. Behavioral Ecology & Sociobiology, 39,

27–30. doi:10.1007/s0026500502

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 109

Brouwers, N., Matusick, G., Ruthrof, K., Lyons, T., & Hardy, G. (2013). Landscape-scale

assessment of tree crown dieback following extreme drought and heat in a

Mediterranean eucalypt forest ecosystem. Landscape Ecology, 28, 69–80.

doi:10.1111/aec.12729

Brown, E. M., Burbidge, A. H., Dell, J., Edinger, D., Hopper, S. D., & Wills, R. T. (1997).

Pollination in Western Australia: A database of animals visiting flowers (Handbook

No. 15). Perth: WA Naturalists Club.

Burrows, G. E. (2000). Seed production in woodland and isolated trees of Eucalyptus

melliodora (yellow box, Myrtaceae) in the South Western Slopes of New South

Wales. Australian Journal of Botany, 48, 681–685. doi:10.1071/BT99058

Burrows, N., & Wardell-Johnson, G. (2003). Interactions of fire and plants in south-western

Australia’s forested ecosystems: A review. In I. Abbott & N. Burrows (Eds.), Fire in

South-Western Australian ecosystems: Impacts and management. Leiden, UK:

Backhuys.

Calver, M. C., & Wardell-Johnson, G. (2004). Sustained unsustainability? An evaluation of

evidence for a history of overcutting in the jarrah forests of Western Australia and its

consequences for fauna conservation. In D. Lunney (Ed.), Conservation of Australia’s

forest fauna (pp. 94–104). Sydney, NSW: Surrey Beatty.

Carr, S. G. M., & Carr, D. J. (1962). Convergence and progression in Eucalyptus and

Symphyomyrtus. Nature, 196(4858), 969–972. doi:10.1038/196969a0

Carr, S. G. M., Carr, D. T., & Ross, F. L. (1971) Male flowers in eucalypts. Australian

Journal of Botany, 19, 73–83.

Cargill, J. (2014). Fate of Eucalyptus marginata seed from canopy-store to emergence in the

northern jarrah forests of Western Australia: Research to help improve regeneration

following shelterwood treatment. Retrieved from https://ro.ecu.edu.au/theses/1415 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 110

Cargill, J., van Etten, E. J. B., & Stock W. D. (2019). The influence of seed supply and

seedbed on seedling recruitment in shelterwood-treated Jarrah (Eucalyptus marginata)

forest. Forest Ecology & Management, 432, 54–63. doi:10.1016/j.foreco.2018.09.008

Cargill, J., van Etten, E. J. B., Whitford, K. R., McCaw, L. W., & Stock, W. D. (2016). A

refined method for estimating capsule crops in individual jarrah (Eucalyptus

marginata) crowns. Australian Forestry, 79(3), 208–216.

doi:10.1080/00049158.2016.1218264

Charles, L. S., Dwyer, J. M., & Mayfield, M. M. (2017). Rainforest seed rain into abandoned

tropical Australian pasture is dependent on adjacent rainforest structure and extent.

Austral Ecology, 42, 238–249. doi:10.1111/aec.12426

Christensen, P. S. (1971). Stimulation of seedfall in karri. Australian Forestry, 35, 182–190.

Churchill, D. M. (1968). The distribution and prehistory of Eucalyptus diversicolor F. Muell.,

E. marginata Donn Ex Sm. and E. calophylla R. Br. in relation to rainfall. Australian

Journal of Botany, 16, 125–151.

Cicero, S. M., & Banzatto-Junior, H. L. (2003). Avaliação do relacionamento entre danos

mecânicos e vigor, em sementes de milho, por meio da análise de imagens. Revista

Brasileira de Sementes, 25(1), 29–36. Retrieved from

https://www.agrolink.com.br/downloads/100984.pdf

Close, D. C., Ruthrof, K. X., Turner, S., Rokich, D. P., & Dixon, K. W. (2009).

Ecophysiology of species with distinct leaf morphologies: Effects of plastic and

shadecloth tree guards. Restoration Ecology, 17, 33–41. doi:10.1111/j.1526-

100X.2007.00330.x

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 111

Close, D. C., Davidson, N. J., Johnson, D. W., Abrams, M. D., Hart, S. C., Lunt, I. D., …

Adams, M. A. (2009). Premature decline of Eucalyptus and altered ecosystem

processes in the absence of fire in some Australian forests. Botanical Review, 75,

191–202. doi:10.1007/s12229-009-9027-y

Colquhoun, I. J., & Kerp, N. L. (2007). Minimizing the spread of a soil‐borne plant pathogen

during a large‐scale mining operation. Restoration Ecology, 15, S85–S93.

doi:10.1111/j.1526-100X.2007.00296.x

Cooper, C. E., Withers, P. C., Mawson, P. R., Johnstone, R., Kirkby, T., Prince, J., …

Robertson, H. (2003). Characteristics of marri (Corymbia calophylla) fruits in relation

to the foraging behaviour of the forest red-tailed black cockatoo (Calyptorhynchus

banksii naso). Journal of the Royal Society of Western Australia, 86, 139–142.

Retrieved from

https://www.rswa.org.au/publications/Journal/86(4)/v86pt4cooperetal139-142.pdf

Craig, M. D., Hobbs, R. J., Grigg, A. H., Garkaklis, M. J., Grant, C. D., Fleming, P. A., &

Hardy, G. E. St. J. (2010). Do thinning and burning sites revegetated after bauxite

mining improve habitat for terrestrial vertebrates? Restoration Ecology, 18, 300–310.

doi:10.1111/j.1526-100X.2009.00526.x

Craig, M. D., Stokes, V. L., Hardy, G. E. St. J, & Hobbs, R. J. (2015). Edge effects across

boundaries between natural and restored jarrah (Eucalyptus marginata) forests in

south-western Australia. Austral Ecology, 40, 186–197. doi:10.1111/aec.12193

Cremer, K. W. (1965). How eucalypt fruits release their seed. Australian Journal of Botany,

13, 11–16. doi:10.1071/BT9650011

Cremer, K. W., Cromer, R. N., & Florence, R. G. (1978). Stand establishment. In W. E. Hillis

& A. G. Brown (Eds.), Eucalypts for wood production (pp. 81–135). Melbourne, Vic:

CSIRO Publishing. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 112

CSBP Soil and Plant Analysis Laboratory. (2018). CSBP Soil and Plant Analysis Laboratory.

Retrieved from https://www.csbp.com.au/CSBP-Lab

Cummings, J., Reid, N., Davies, I., & Grant C. (2005) Adaptive restoration of sand-mined

areas for biological conservation. Journal of Applied Ecology, 42, 160–170.

doi:10.1111/j.1365-2664.2005.01003.x

Cummings, J., Reid, N., Davies, I., & Grant, C. (2007) Experimental manipulation of

restoration barriers in abandoned eucalypt plantations. Restoration Ecology, 15, 156–

167. doi:10.1111/j.1526-100X.2006.00200.x

Davison, E. M. (2014). Resolving confusions about jarrah dieback—Don’t forget the plants.

Australasian Plant Pathology, 43(6), 691–701. doi:10.1007/s13313-014-0302-y

Davison, E. M. (2018). Relative importance of site, weather and Phytophthora cinnamomi in

the decline and death of Eucalyptus marginata—Jarrah dieback investigations in the

1970s to 1990s. Australasian Plant Pathology, 47, 245–257. doi:10.1007/s13313-018-

0558-8

Davison, E. M., & Tay, F. C. S. (1985). The effect of waterlogging on seedlings of

Eucalyptus marginata. New Phytologist Trust, 101, 743–753. doi:10.1111/j.1469-

8137.1985.tb02879.x

Daws, M. I., Standish, R. J., Koch, J. M., Morald, T. K., Tibbett, M., & Hobbs, R. J. (2015).

Phosphorus fertilisation and large legume species affect forest restoration after

bauxite mining. Forest Ecology & Management, 354, 10–17.

doi:10.1016/j.foreco.2015.07.003 de Dios, V. R., Fischer, C., & Colinas, C. (2007). Climate change effects on Mediterranean

forests and preventive measures. New Forests, 33, 29–40. doi:10.1007/s11056-006-

9011-x

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 113 del Cacho, M., Estiarte, M., Peñuelas, J., & Lloret, F. (2012). Inter-annual variability of seed

rain and seedling establishment of two woody Mediterranean species under field-

induced drought and warming. Population Ecology, 55, 277–289.

doi:10.1007/s10144-013-0365-6

Dell, B., & Havel, J. J. (1989). The jarrah forest, an introduction. In B. Dell, J. J. Havel, & N.

Malajczuk (Eds.), The jarrah forest: A complex Mediterranean ecosystem (pp. 1–10).

Dordrecht, UK: Kluwer Academic.

Duncan, R. P., Diez, J. M., Sullivan, J. J., Wangen, S., & Miller, A. L. (2009). Safe sites, seed

supply, and the recruitment function in plant populations. Ecology, 90(8), 2129–2138.

doi:10.1890/08-1436.1

Elias, S. G., Copeland, L. O., McDonald, M. B., & Baalbaki, R. Z. (2012). Seed testing:

Principles and practices. East Lansing, MI: Michigan State University Press.

Elliott, P., Gardner, J., Allen, D., & Butcher, G. (1996). Completion criteria for Alcoa of

Australia Limited’s bauxite mine rehabilitation. 3rd International and the 21st Annual

Minerals Council of Australia Environmental Workshop, Newcastle, 14–18 October

1995. Canberra, ACT: Minerals Council of Australia.

Environmental Protection Authority. (1995). Rehabilitation and revegetation. Canberra,

ACT: Commonwealth Publishing Service.

Eriksson, O., & Ehrlén, J. (1992). Seed and microsite limitation of recruitment in plant

populations. Oecologia, 91, 360–364. doi:10.1007/BF00317624

Erskine P. D., & Fletcher, A. T. (2013). Novel ecosystems created by coal mines in central

Queensland’s Bowen Basin. Ecological Processes, 2, 1–12. Retrieved from

https://link.springer.com/article/10.1186/2192-1709-2-33 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 114

Finch-Savage, W. E., & Leubner-Metzger, G. (2006). Seed dormancy and the control of

germination. New Phytologist, 171(3), 501–523. doi:10.1111/j.1469-

8137.2006.01787.x

Florence, R. G. (2004). Ecology and silviculture of eucalypt forests. Melbourne, Vic: CSIRO

Publishing.

Frouz, J., Pizl, V., Cienciala, E., & Kalcik, J. (2009). Carbon storage in post-mining forest

soil, the role of tree biomass and soil bioturbation. Biogeochemistry, 94, 111–121.

Gagliardi, B., & Marcos-Filho, J. (2011). Relationship between germination and bell pepper

seed structure assessed by the X-ray test. Scientia Agricola, 68, 411–416.

doi:10.1590/S0103-90162011000400004

Gardner, J. H., & Bell, D. T. (2007). Bauxite mining restoration by Alcoa World Alumina

Australia in Western Australia: Social, political, historical, and environmental

contexts. Restoration Ecology, 15(s4), S3–S10. doi:10.1111/j.1526-

100X.2007.00287.x

Gardner, J., & Stoneman, G. (2003). Bauxite mining and conservation of the jarrah forest in

south-west Australia. Rehabilitation of mining and resources projects as it relates to

Commonwealth responsibilities. Submission 2—Attachment 1. Retrieved from

https://www.aph.gov.au/DocumentStore.ashx?id=7556809e-1029-4162-8aaf-

eb5bb90a562f&subId=510096

Gentilli, J. (1989). Climate of the jarrah forest. In B. Dell, J. J. Havel, & N. Malajczuk (Eds.),

The jarrah forest: A complex Mediterranean ecosystem (pp. 23–40). Dordrecht, UK:

Kluwer Academic.

Gigolashvili, T., & Kopriva, S. (2014). Transporters in plant sulfur metabolism. Frontiers in

Plant Science, 5, 442. doi:10.3389/fpls.2014.00442 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 115

Gilkes, R. J., Scholtz, G., & Dimmock, G. M. (1973). Lateritic deep weathering of granite.

Journal of Soil Science, 24, 523–536. doi:.1111/j.1365-2389.1973.tb02319.x

Gill, A., Brooker, M., & Moore, P. (1992). Seed weights and numbers as a function of fruit

size and subgenus in some eucalyptus species from south-western Australia.

Australian Journal of Botany, 40(1), 103. doi:10.1071/BT9920103

Gill, A. M., Groves, R. H., & Noble, I. R. (1981). Fire and the Australian biota. Canberra,

ACT: Australian Academy of Science.

Gomez-Aparicio, L. (2008). Spatial patterns of recruitment in Mediterranean plant species:

Linking the fate of seeds, seedlings and saplings in heterogeneous landscapes at

different scales. Journal of Ecology, 96, 1128–1140. doi:10.1111/j.1365-

2745.2008.01431.x

Gore, D., Preston, N., & Fryirs, K. (2007) Post-rehabilitation environmental hazard of Cu,

Zn, As and Pb at the derelict Conrad Mine, eastern Australia. Environmental

Pollution, 148, 491–500. doi:10.1016/j.envpol.2006.12.016

Grant, C. D. (2006). State-and-transition successional model for bauxite mining rehabilitation

in the jarrah forest of Western Australia. Restoration Ecology, 14(1), 28–37.

doi:10.1111/j.1526-100X.2006.00102.x

Grant, C., & Koch, J. (2007). Decommissioning Western Australia’s first bauxite mine: Co-

evolving vegetation restoration techniques and targets. Ecological Management &

Restoration, 8(2) 92–105. doi:10.1111/j.1442-8903.2007.00346.x

Grant, C. D., Duggin, J., Meek, I., & Lord, M. (2001). End point criteria and successional

pathways for manganese mining rehabilitation on Groote Eylandt, Northern Territory.

26th Annual Minerals Council of Australia Environmental Workshop, Adelaide, 14–

18 October 2001. Canberra, Australia: Minerals Council of Australia. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 116

Hadas, A. (1970). Factors affecting seed germination under soil moisture stress. Israel

Journal of Agriculture Research, 20, 3–14.

Hardham, A. R., & Blackman, L. M. (2018). Pathogen profile update: Phytophthora

cinnamomi. Molecular Plant Pathology, 19(2), 260–285. doi:10.1111/mpp.12568

Harper, J. L. (1977). The population biology of plants. New York, NY: Academic Publishing.

Harper, J. L., Clatworthy, J. N., McNaughton, I. H., & Sagar, G. R. (1961). The evolution and

ecology of closely related species living in the same area. Evolution, 15(2), 209–227.

doi:10.2307/2406081

Harper, J. L., Williams, J. T., & Sagar, G. R. (1965). The behaviour of seeds in soil: I. The

heterogeneity of soil surfaces and its role in determining the establishment of plants

from seed. Journal of Ecology, 53(2), 273–286. doi:10.2307/2257975

Hatch, A. B. (1960). Ash bed effects in Western Australian forest soils. Bulletin of the

Forests Department Western Australia, 19, 1–19.

Havel, J. J. (1975). Site-vegetation mapping in the northern jarrah forest (Darling Range).

Perth, WA: Western Australian Forest Department Bulletin.

Hegarty, T. W. (1977). Seed vigour in field bean (Vicia faba L.) and its influence on plant

stand. Journal of Agricultural Science, 88, 169–173.

Hingston, F. J., Malajczuk, N., & Grove, T. S. (1982). Acetylene reduction (N2 fixation) by

jarrah forest legumes following fire and phosphate application. Journal of Applied

Ecology, 19, 631–645. doi:10.2307/2403495

Hopper, S. D., & Gioia, P. (2004). The Southwest Australian Floristic Region: Evolution and

conservation of a global hot spot of biodiversity. Annual Review of Ecology,

Evolution & Systematics, 35, 623–650.

doi:10.1146/ANNUREV.ECOLSYS.35.112202.130201 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 117

Huang, D., Han, J. G., Wu, W. L., & Wu, J. Y. (2008). Soil temperature effects on emergence

and survival of Iris Iactea seedlings. New Zealand Journal of Crop & Horticultural

Science, 36(3), 183–188. doi:10.1080/01140670809510234

Hyatt, L. A., Rosenberg, M. S., Howard, T. G., Bole, G., Fang, W., Anastasia, J., …

Gurevitch, J. (2003). The distance dependence predication of the Janzen–Connell

hypothesis: A meta-analysis. Oikos, 103, 590–602. doi:10.1034/j.1600-

0706.2003.12235.x

International Seed Testing Association. (2019). International rules for seed testing 1, i-44.

Retrieved from https://www.seedtest.org/en/international-rules-for-seed-testing-

_content---1--1083.html

Jacobs D. F., & Timmer V. R. (2003). Fertilizer-induced changes in rhizosphere electrical

conductivity: Relation to forest tree seedling root system growth and function. New

Forests, 30, 147–166. doi:10.1007/s11056-005-6572-z

Jackson, S. T., & Hobbs, R. J. (2009). Ecological restoration in the light of ecological history.

Science, 325, 567-569. doi:10.1126/science.1172977

Janzen, D. H. (1970). Herbivores and the number of tree species in tropical forests. American

Naturalist, 104, 501–528. Retrieved from www.jstor.org/stable/2459010

Johnstone, R. E., & Kirkby, T. (1999). Food of the forest red-tailed black cockatoo

Calyptorhynchus baksii naso in south-west Western Australia. Western Australian

Naturalist, 22(3), 167–177.

Klausmeyer, K., & Shaw, M. R. (2009). Climate change, habitat loss, protected areas and the

climate adaptation potential of species in Mediterranean ecosystems worldwide. PLoS

One, 4, 1–9. doi:10.1371/journal.pone.0006392 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 118

Koch, J. M. (2007a). Alcoa’s mining and restoration process in south western Australia.

Restoration Ecology, 15(Supplement), S11–S16. doi:10.1111/j.1526-

100X.2007.00288.x

Koch, J. M. (2007b). Restoring a jarrah forest understorey vegetation after bauxite mining in

Western Australia. Restoration Ecology, 15(4), S26–S39. doi:10.1111/j.1526-

100X.2007.00290.x

Koch,J. M., Ruschmann, A. M., & Morald, T. K. (2009). Effect of time since burn on soil

seedbanks in the jarrah forest of Western Australia. Australian Journal of

Botany, 57(8), 647-660. https://doi.org/10.1071/BT09101

Koch, J. M., & Samsa, G. P. (2007). Restoring jarrah forest trees after bauxite mining in

Western Australia. Restoration Ecology, 15 (4), S17–S25. doi:10.1111/j.1526-

100X.2007.00289.x

Koch, J., & Ward, S. (1994). Establishment of understory vegetation for rehabilitation of

bauxite- mined areas in the jarrah forest of western Australia. Journal of

Environmental Management, 4, 11–15.

Koch, J. M., & Ward, S. C. (2005). Thirteen-year growth of jarrah (Eucalyptus marginata) on

rehabilitated bauxite mines in south-western Australia. Australian Forestry, 68(3),

176–185. doi:10.1080/00049158.2005.10674963

Koch, J. M., Ward, S. C., Grant, C. D., & Ainsworth, G. L. (1996). Effects of bauxite mine

restoration operations on topsoil reserves in the Jarrah Forest of Western Australia.

Restoration Ecology, 4, 368–376. doi:10.1111/j.1526-100X.1996.tb00189.x

Kollman, J. (2008). Spatial variation in seedling emergence and establishment—Functional

groups among and within habitats? In M. A. Leck, V. T. Parker, & R. L. Simpson

(Eds.), Seedling ecology and evolution (pp. 274–292). Cambridge, United Kingdom:

Cambridge University Press. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 119

Lamb, D., Erskine, P. D., & Fletcher, A. (2015). Widening gap between expectations and

practice in Australian minesite rehabilitation. Ecological Management & Restoration,

16(3), 186–195. doi:10.1111/emr.12179

Lambers, H., Cramer, M. D., Shane, M. W., Wouterlood, M., Poot, P., & Veneklaas, E. J.

(2003). Introduction: Structure and functioning of cluster roots and plant responses to

phosphate deficiency. Plant Soil, 248, ix–xix. doi:10.1093/aob/mcl114

Larcher, W. (2000). Temperature stress and survival ability of Mediterranean sclerophyllous

plants. Plant Biosystems, 134, 279–295. doi:10.1080/11263500012331350455

Le Houreou, H. (2000). Restoration and rehabilitation of arid and semiarid Mediterranean

ecosystem in North Africa and West Asia: A review. Arid Soil Research &

Rehabilitation, 14, 3–14. doi:10.1080/27089030600263139

Li, J., Duggin, J. A., Grant, C. D., & Loneragan, W. A. (2003). Germination and early

survival of Eucalyptus blakelyi in grasslands of the New England Tablelands, NSW,

Australia. Forest Ecology & Management, 173, 319–334. doi:10.1016/S0378-

1127(02)00013-0

Lindner, M., Maroschek, M., Netherer, S., Kremer, A., Barbati, A., Garcia-Gonzaloa, J., …

Marchetii, M. (2010). Climate change impacts, adaptive capacity, and vulnerability of

European forest ecosystems. Forest Ecology and Management, 259, 698–709.

doi:10.1016/j.foreco.2009.09.023

Lockley, I. R., & Koch, J. M. (1996). Response of two eucalypt species to fertiliser

application on rehabilitated bauxite mines in Western Australia (Environmental

Research Note No. 27). Unpublished report. Pinjara, WA: Alcoa of Australia.

Loneragan, O. W., & Loneragan, J. F. (1964). Ashbed and nutrients in the growth of

seedlings of Karri (Eucalyptus diversicolor F.v.M.). Journal of the Royal Society of

Western Australia, 47, 75–80. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 120

Macdonald, S. E., Landhäusser, S. M., Skousen, J., Franklin, J., Frouz, J., Hall, S., …

Quideau, S. (2015). Forest restoration following surface mining disturbance:

Challenges and solutions. New Forests, 46, 703–732. doi: 10.1007/s11056-015-9506-

4mc

Majer, J. D. (1980). The influence of ants on broadcast naturally spread seeds in rehabilitated

bauxite mined areas. Reclamation Review, 3, 3–9. Retrieved from

https://www.researchgate.net/publication/45636110_The_influence_of_ants_on_broa

dcast_and_naturally_spread_seeds_in_rehabilitated_bauxite_mines

Martyn, A. J., Seed, U. L., Ooi, M. K. J., & Offard, C. A. (2011). Seed fill, viability and

germination of NSW species in the family Rutaceae. Cunninghamia, 11(2), 203–212.

Retrieved from

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.511.6378&rep=rep1&type=

pdf

Matusick, G., Ruthrof, K., Brouwers, N. C., Dell, B., & Hardy, G. St. J. (2013). Sudden forest

canopy collapse corresponding with extreme drought and heat in a Mediterranean-

type eucalypt forest in southwestern Australia. European Journal of Forest Research,

132, 497–510. doi:10.1007/S10342-013-0690-5

Matusick, G., Ruthrof, K. X., Brouwers, N. C., & Hardy, G. St. J. (2014). Topography

influences the distribution of autumn frost damage on trees in a Mediterranean-type

Eucalyptus forest. Trees, 28, 1449–1462. doi:10.1007/s00468-014-1048-4

McCaw, L. W. (2011). Characteristics of jarrah (Eucalyptus marginata) forest at

FORESTCHECK monitoring sites in south-west Western Australia: Stand structure,

litter, woody debris, soil and foliar nutrients. Australian Forestry, 74, 254–271.

doi:10.1080/00049158.2011.10676370 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 121

McChesney, C. J., Koch, J. M., & Bell, D. T. (1995). Jarrah forest restoration in Western

Australia: Canopy and topographic effects. Restoration Ecology, 3, 105–110.

doi:10.1111/j.1526-100X.1995.tb00083.x

Mills, J. (1989). The impact of man on the northern jarrah forest from settlement in 1829 to

the Forests Act 1918. In B. Dell, J. J. Havel, & N. Malajczuk (Eds.), The jarrah

forest: A complex Mediterranean ecosystem (pp. 229–279). Dordrecht, UK: Kluwer

Academic.

Miolaro, L. G., Gonclaves, A. N., Mendes, J. C. T., Moreira, R., Brune, A., & da Silva, P. H.

M. (2017). Spontaneous regeneration of eucalypts from seed production areas.

Biological Invasions, 91, 1733–1737. doi:10.1007/s10530-017-1397-1

Morris, J. D. (1980). Factors affecting salt tolerance of eucalypts. In K. M. Old, G. A. Kile, &

C. P. Ohmart (Eds.), Eucalypt dieback in forests and woodlands: Proceedings of

conference held at the CSIRO Division of Forest Research (pp. 190–204). Canberra,

ACT: CSIRO Publishing.

Morrison, B., Lamb, D., & Hundloe, T. (2005) Assessing the likelihood of mine site

revegetation success: A Queensland case study. Australasian Journal of

Environmental Management, 12, 165–182. doi:10.1080/14486563.2005.9725087

Myers, N., Mittermeier, R. A., da Fonseca, G. A. B., Kent, J., & Mittermeier, C. G. (2000).

Biodiversity hotspots for conservation priorities. Nature, 403(6772), 853–858.

doi:10.1038/35002501

National Land and Water Resources Audit. (2001). Australian Dryland Salinity Assessment

2000. Extent, impacts, processes, monitoring and management options. Canberra,

ACT: Land and Water Australia.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 122

Norman, M. A., & Koch, J. M. (2009). Variability in jarrah (Eucalyptus marginata) density

on rehabilitated bauxite mines relates to factors affecting emergence rather than

survival (Environmental Research Note No. 30). Unpublished report. Pinjara, WA:

Alcoa of Australia.

Ooi, M. K. J. (2007). Dormancy classification and potential dormancy-breaking cues for

shrub species from fire-prone south-eastern Australia. In S. Adkins, S. Ashmore, & S.

C. Navie (Eds.), Seeds: Biology, development and ecology. Trowbridge, UK: CAB

International.

Ozkan, K., Senol, H., Gulsoy, S., Mert, A., Suel, H., & Eser, Y. (2009) Vegetation-

environment relationships in mediterranean mountain forests on limeless bedrocks of

Southern Anatolia, Turkey. Journal of Environmental Engineering and Landscape

Management, 17, 154–163. doi:10.3846/1648-6897.2009.17.154-163

Panchal, K. P., Pandya, N. R., Albert, S., & Gandhi, D. J. (2014). A X-ray image analysis for

assessment of forage seed quality. International Journal of Plant, Animal &

Environmental Science, 4, 103–109.

Parsons, R. F. (1968). Effects of waterlogging and salinity on growth and distribution of three

mallee species of Eucalyptus. Australian Journal of Botany, 16, 101–108.

Penfield, S., & MacGregor, D. R. (2017). Effects of environmental variation during seed

production on seed dormancy and germination. Journal of Experimental Botany,

68(4), 819–825. doi:10.1093/jxb/erw436

Pepper, R. G., & Craig, G. F. (1986). Resistance of selected Eucalyptus species to soil

salinity in Western Australia. Journal of Applied Ecology, 23(3), 977–987.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 123

Pinno, B. D., Landhaüsser, S. M., MacKenzie, M. D., Quideau, S. A., & Chow, P. S. (2012).

Trembling aspen seedling establishment, growth, and response to fertilization on

contrasting soils. Canadian Journal of Soil Science, 92, 143–151.

doi:10.4141/cjss2011-004

Podger, F. D. (1972). Phytophthora cinnamomi, a cause of lethal disease in indigenous plant

communities in Western Australia. Phytopathology, 62, 972–981. Retrieved from

https://www.apsnet.org/publications/phytopathology/backissues/Documents/1972Arti

cles/Phyto62n09_972.PDF

Rathcke, B., & Lacey, E. P. (1985) Phenological patterns of terrestrial plants. Annual Review

of Ecology, Evolution, and Systematics, 16, 179–214.

Renbuss, M. A., Chilvers, G. A., & Pryor, L. D. (1973). Microbiology of an ashbed.

Proceedings of the Linnean Society of New South Wales, 97, 302–310.

Rokich, D. P., & Dixon, K. W. (2007). Recent advances in restoration ecology, with a focus

on the Banksia woodland and the smoke germination tool. Australian Journal of

Botany, 55, 375–389. doi:10.1071/BT06108

Ruthrof, K. X., Bader, M. K. F., Matusick, G., Jakob, S., & Hardy, G. E. St. J. (2016).

Promoting seedlings physiological performance and early establishment in degraded

Mediterranean-type ecosystems. New Forests, 47, 367–376. doi:10.1007/s11056-015-

9520-6

Ruthrof, K. X., Douglas, T. K., Calver, M. C., Barber, P. A., Dell, B., & Hardy, G. E. St. J.

(2010). Restoration treatments improve seedling establishment in a degraded

Mediterranean-type Eucalyptus ecosystem. Australian Journal of Botany, 58, 646–

655. doi: 10.1071/BT10211

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 124

Ruthrof, K. X., Fontaine, J. B., Buizer, M., Matusick, G., McHenry, M. P., & Hardy, G. E. St.

J. (2013). Linking restoration outcomes with mechanisms: The role of site

preparation, fertilisation and revegetation timing relative to soil density and water

content. Plant Ecology, 214(8), 987–998. Retrieved from

https://www.jstor.org/stable/23500412

Ruthrof, K. X., Matusick, G., & Hardy, G. E. St. J. (2015). Early differential responses of co-

dominant canopy species to sudden and severe drought in a Mediterranean-climate

type forest. Forests, 6, 2082–2091. doi:10.3390/f6062082

Sahlen, K., Bergsten, U., & Wiklund, K., (1995). Determination of viable and dead Scots

pine seeds of different anatomical maturity after freezing using IDX method. Seed

Science and Technology, 23, 405–414.

Shearer, B. L., & Tippett, J. T. (1989). Jarrah dieback: The dynamics and management of

Phytophthora cinnamomi in the jarrah (Eucalyptus marginata) forest of south-

Western Australia (Bulletin No. 3). Perth, WA: Department of Conservation and Land

Management.

Simard, M., Pinto, N., Fisher J. B., & Baccini A. (2011), Mapping forest canopy height

globally with spaceborne lidar, Journal of Geophysical Research, 116 (G04021),

doi:10.1029/2011JG001708.

Specht, R. L., Roe, M. E., & Boughton, V. H. (1974). Conservation of major plant

communities in Australia and Papua New Guinea. Australian Journal of Botany

Supplementary Series, 7, 1–667. doi:10.1071/BT7407001

Standish, R. J., Daws, M. I., Gove, A. D., Didham, R. K., Grigg, A. H., Koch, J. M., &

Gibson, D. (2015). Long-term data suggest jarrah-forest establishment at restored

mine sites is resistant to climate variability. Journal of Ecology, 103, 78–89.

doi:10.1111/1365-2745.12301 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 125

StataCorp. (2017). Stata statistical software: Release 15. College Station, TX: StataCorp.

Steel, E. J., Fontaine, J. B., Ruthrof, K. X., Burgess, T. I., & Hardy, G. E. St. J. (2019).

Changes in structure of over- and midstory tree species in a Mediterranean-type forest

after an extreme drought-associated heatwave. Austral Ecology, 44,1,438-1,450.

doi:10.1111/aec.12818.

Stenseth, N. C., & Mysterud, A. (2002). Climate, changing phenology, and other life history

and traits: Nonlinearity and match–mismatch to the environment. Proceedings of the

National Academy of Sciences of the United States of America, 99, 13379–13381.

Stoneman, G. L., & Dell, B. (1994). Emergence of Eucalyptus marginata (jarrah) from seed

in Mediterranean-climate forest in response to overstorey, site, seedbed and seed

harvesting. Australian Journal of Ecology, 19, 96–102. doi:10.1111/j.1442-

9993.1994.tb01548.xStoneman

Stoneman, G. L., Dell, D., & Turner, N. C. (1994). Mortality of Eucalyptus marginata

(jarrah) seedlings in Mediterranean-climate forest in response to overstorey, site,

seedbed, fertilizer application and grazing. Australian Journal of Ecology, 19, 103–

109. doi:10.1111/j.1442-9993.1994.tb01549.x

Szota, C., Farrell, C., Koch, J. M., Lambers, H., & Venekelaas, E. J. (2011). Contrasting

physiological responses of two co-occurring eucalypts to seasonal drought at restored

bauxite mine sites. Tree Physiology, 31, 1052–1056. doi:10.1093/treephys/tpr085

Tacey, W. H., & Glossop, B. L. (1980). Assessment of topsoil handling techniques for

rehabilitation of sites mined for bauxite within the jarrah forest of Western Australia.

Journal of Applied Ecology, 17(1), 195–201. doi:10.2307/2402974

Tangney, R., Merritt, D. J., Fontaine, J. B., & Miller, B. P. (2019). Seed moisture content as a

primary trait regulating the lethal temperature thresholds of seeds. Journal of Ecology,

107, 1093–1105. doi:10.1111/1365-2745.13095 JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 126

Tenhumber, B., Keller, M. A., & Possingham, H. P. (2001). Using Cox’s proportional hazard

models to implement optimal strategies: An example from behavioural ecology.

Mathematical & Computer Modelling, 33, 597–607. doi:10.1016/S0895-

7177(00)00264-8

Thomson, F. J., Moles, A. T., Auid, T. D., & Kingsford, R. T. (2011). Seed dispersal distance

is more strongly correlated with plant height than with seed mass. Journal of Ecology,

99, 1299–1307. doi:10.1111/j.1365-2745.2011.01867.x

Tilman, D. (1993). Species richness of experimental productivity gradients: How important is

colonization limitation? Ecology, 74, 2179–2191.

Turcotte, I., Quideau, S., & Oh, S. (2009). Organic matter quality in reclaimed boreal forest

soils following oil sands mining. Organic Geochemistry, 40, 510–519.

Vander Wall, S. B., Forget, P. M., Lambert, J. E., & Hulme, P. E. (2004). Seed fate

pathways: Filling the gap between parent and offspring. In P. Forget, J. Lambert, & P.

Hulme (Eds.), Seed fate: Predation, dispersal and seedling establishment. Retrieved

from https://ebookcentral.proquest.com

Vesk, P. A., Davidson, A., & Chee, Y. E. (2010). Spatial distribution and prediction of seed

production by Eucalyptus microcarpa in a fragmented landscape. Austral Ecology,

35, 60–71. doi:10.1111/j.1442-9993.2009.02012.x

Vittinghoff, E., Glidden, D. V., Shiboski, S. C., & McCulloch, C. E. (2012). Regression in

biostatistics: Linear, logistic, survival, and repeated measures models. New York,

NY: Springer Science + Business Media.

Wallace, W. R. (1965). Fire in the jarrah forest environment. Journal of the Royal Society of

Western Australia, 49, 33–44. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 127

Ward, S. C., & Koch, J. M. (1995). Early growth of jarrah (Eucalyptus marginata Donn ex

Smith) on rehabilitated bauxite mine sites in south-west Australia. Australian

Forestry, 58, 65–71. doi:10.1080/00049158.1995.10674646

Ward S. C., Koch, J. M., & Nichols, O. G. (1990). Bauxite mine rehabilitation in the Darling

Range, Western Australia. Proceedings Ecological Society of Australia, 16, 557–565.

Wardell-Johnson, G. W., Calver, M., Burrows, N., & Di Virgilio, G. (2015). Integrating

rehabilitation, restoration and conservation for a sustainable jarrah forest future during

climate disruption. Pacific Conservation Biology, 21, 175–185. doi:10.1071/PC15026

Wassen, M. J., Venterink, H. O., Lapshina, E. D., & Tanneberger, F. (2005). Endangered

plants persist under phosphorus limitation. Nature, 437, 547–550.

doi:10.1038/nature03950

White-Toney, T., Korczynskyj, D., Grigg, A., & Bulsara, M. (2019). Variable tree

establishment in bauxite mine restoration in south-west Australia is linked to rainfall

distribution, seasonal temperatures and seed rain. Ecological Management &

Restoration, 20(3), 266–270. doi:10.1111/emr.12389.

Williams, K., & Mitchell, D. (2001). Jarrah forest 1 (JFI–Northern Jarrah Forest

subregion). A biodiversity audit of Western Australia’s 53 biogeographical

subregions in 2002. Perth, WA: Department of Parks and Wildlife. Retrieved from

https://www.dpaw.wa.gov.au/images/documents/about/science/projects/waaudit/jarra

h_forest01_p369-381.pdf

Wolken, J. M., Landhaüsser, S. M., Lieffers, V. J., & Dyck, M. (2010) Differences in initial

root development and soil conditions affect establishment of trembling aspen and

balsam poplar seedlings. Botany, 88, 275–285. doi:10.1139/B10-016.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 128

Yates, C. J., & Hobbs, R. J. (1997). Temperate eucalypt woodlands: A review of their status,

processes threatening their persistence and techniques for restoration. Australian

Journal of Botany, 45, 949–973. Retrieved from

https://www.rswa.org.au/publications/Journal/88(4)/vol88pt4yatesetal147-153.pdf

Yates, C. J., Hobbs, R. J., & Bell, R. W. (1994). Factors limiting the recruitment of

Eucalyptus salmonophloia in remnant woodlands. 1. Pattern of flowering, seed

production and seed fall. Australian Journal of Botany, 42, 531–542.

doi:10.1071/BT9940531

Yates C. J., Hooper, S. D., & Taplin, R. H. (2005). Native insect flower visitor diversity and

feral honeybees on jarrah (Eucalyptus marginata) in Kings Park, an urban bushland

remnant. Journal of the Royal Society of Western Australia, 88, 147–153. Retrieved

from https://www.rswa.org.au/publications/Journal/88(4)/vol88pt4yatesetal147-

153.pdf

Yates, C. J., Norton, D. A., & 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. doi:10.1046/j.1442-9993.2000.01030.x

Yates, C. J., Taplin, R., Hobbs, R. J., & Bell, R. W. (1995). Factors limiting the recruitment

of Eucalyptus salmonophloia F. Muell. in remnant woodlands. II. Post‐dispersal seed

predation and soil seed reserves. Australian Journal of Botany, 43, 145–155.

doi:10.1071/BT9950145

Yoshimoto, N., Takahashi, H., Smith, F. W., Yamaya, T., & Saito, K. (2002). Two distinct

high-affinity sulfate transporters with different inducibilities mediate uptake of sulfate

in Arabidopsis roots. Plant Journal, 29, 465–473. doi:10.1046/j.0960-

7412.2001.01231.x

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 129

Appendices

Appendix 1: Generalised Linear Model Outputs of Best Models, Chapter 2

Below are the outputs of the best fit GLMs and pseudo R2 values from Stata (Ver. 15; 2017) used in Chapter 2, Table 2.1.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 130

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 131

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 132

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 133

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 134

Appendix 2: Jarrah and Marri Seed Viability, Section 3.1

This study was not designed specifically to investigate seed viability and size, as the literature on jarrah seed does not adequately present inconsistencies in classification of jarrah seed from chaff by size prior to data collection. Although there are some consequent limitations to this work, it offers important findings and suggestions related to the importance of more consistent viability and purity reports to improve recruitment of jarrah from seed.

The focus of this work, then, was to 1) assess the use of X-ray imaging to determine seed fullness for jarrah and marri seed; and 2) report jarrah viability of 3-mm capsule particles as a baseline for better understanding seed purity and viability in other work where ‘seed’ classification is not clear.

Introduction

Eucalyptus species capsules have differential placental expression (Boothe &

Drinnan, 2005), leading to the production of both sterile ovulodes and fertile ovules (or

‘seeds’; Boland et al., 1980; Carr & Carr, 1962). Chaff is composed of ovulodes along with unviable seed: underdeveloped or abnormal ovules; non-fertilised ovules; and post- zygotically aborted ovules (Boland et al., 1980; Boothe & Drinnan 2005). The proportion of chaff to seed (combined considered ‘particles’) expelled from a capsule depends on the species of eucalypt and on the environmental conditions experienced by the parent tree during reproduction (e.g. climate and soil characteristics; Boland et al., 1980; Penfield &

MacGregor 2017).

Fundamentally, a viable seed is ‘full’, with an intact endosperm and embryo—the tissues essential for successful germination (e.g. Martyn, Seed, Ooi, & Offard, 2011).

Viability is reported as the number of germinants per sample size (weight). Accuracy of this measure is dependent on the purity of the sample; that is, the proportion of chaff to seed. The seed collection process for eucalypt seeds involves three steps: 1) extracting seed from an JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 135 appropriate providence, suitable for the restored area; 2) drying and shaking fruit; and 3) separating seed from chaff, usually by sieving. Most problems in purity assessments occur in this last step for species such as jarrah, in which seed and chaff are a similar weight and size

(Boland et al., 1980).

Tests of viability that produce results quickly and are non-destructive to the seed are preferred. Historically, seed viability testing has relied on germination trials, and international standards (International Seed Testing Association, 2019) recommend tests of ‘pure seed particles’ (Boland et al., 1980). Other types of viability test are increasingly used to assess the presence of endosperm and embryo in a seed, although not all seeds with these criteria will germinate: a viable seed still relies on specific environmental conditions and, potentially, dormancy-breaking cues to stimulate germination (Baskin & Baskin, 1998). Quick, destructive viability tests that reveal the physical integrity of seeds include the ‘fast green’, indoxyl acetate, ferric chloride, sodium hypochlorite, tetrazolium and X-ray tests (Elias et al.,

2012). In most cases, destructive and non-destructive tests report similar proportions of viability in the same species. Tetrazolium and cut tests both accurately described the viability of three Leucopogan species (Ooi, 2007). Viability in five Saudi Arabian species (Moringa peregrina, Abruspreca torius, Arthrocnemum macrostachyum, Acacia ehrenbergiana and

Acacia tortilis) was accurately assessed by both germination trials and X-ray images (Al-

Hammond & Al-Ammar, 2017). Additionally, cut tests and X-ray radiology are often used interchangeably to assess viability in members of the Rutaceae (Martyn et al., 2011).

Germination trials, cut tests and tetrazolium testing are all destructive methods of viability testing, focused on identifying and reporting the presence of endosperm and embryo in seeds.

X-ray viability testing is based on seed tissues absorbing electromagnetic waves (X- rays) differently depending on their density, and this provides an image of dark and light contrast based on the internal structure of each seed (Elias et al., 2012). Images from X-rays JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 136 also illustrate internal seed structure, morphology, mechanical damage and seed fullness at a greater level of detail than other viability tests (Panchal, Pandya, Albert, & Gandhi, 2014).

The use of X-rays for viability testing has been successful in Pinus species (Salen, Bergsten,

& Wiklun, 1995), crops (e.g. corn; Cicero & Banzatto-Junior, 2003) and native WA species

(e.g. Tangney, Merritt, Fontaine, & Miller, 2019). X-ray radiology is simple, accurate, allows for high numbers of simultaneous viability assessments and is non-destructive to seeds

(Gagliardi & Marcos-Filho, 2011). For restoration purposes, where high volumes of seed are required, X-ray viability testing is a logical application for improving rigour of seed quality monitoring and therefore restoration outcomes (e.g. Tangney et al., 2019).

Jarrah and marri seed for the seed mix is collected for forest restoration following land clearing for mining in the region (Koch, 2007a) and so is an economically important resource (Koch & Samsa, 2007). Marri seed is easily distinguished from chaff, and therefore viability, but not purity, is an important consideration influencing seedling density at establishment. The viability of jarrah seed has been reported as between 1.6% (Gill, Brooker,

& Moore, 1992) and 90% (Abbott & Loneragan, 1986), which is to some extent related to the proportion of chaff to seed (or ‘purity’ of seed mix). Abbott and Loneragan (1986) described fertile jarrah seed size as 4.5 × 3.0 mm and Cargill et al. (2019) described jarrah seed as smaller than four mm and dark in colour, but no study has clearly identified the lower limit for viable jarrah seed. Based on the four mm seed classification size (e.g. Cargill et al., 2019) cut tests and germination trials found viability of 31 seed lots to be 41–79%, with germination rates of 42–72% (Barker, 2019). Cut and germination tests produced similar viability proportions for jarrah, confirming that the use of tests for seed fullness (e.g. cut tests and X-ray imaging) is a valid way to test jarrah seed viability. As a result, the use of X-ray imaging to also report seed fullness of jarrah seeds is likely to be accurate and achieve greater conservation of seed resources than cut or germination tests. JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 137

Recent published work with jarrah seed concluded there was a need for further studies on the variability in viability of jarrah seed by site and between years (Cargill et al., 2016).

The lack of consensus on the size of viable jarrah seed and proportions of viable seed expelled from jarrah capsules informed this portion of the study. The lack of consistency in reporting seed size (and therefore viability) may be a cause of the variability in reported germination rates across work with jarrah seed (1.6–90%; Abbott & Loneragan, 1986; Gill et al., 1992). Here, the viability of all three mm jarrah particles (including chaff and seed) captured in forest and restoration site seed traps was reported. This study used the three mm cut-off to identify the likelihood of viable jarrah seeds smaller than four mm. The purpose of this study was to 1) assess the use of X-ray testing to distinguish chaff from seed on the basis of fullness; and 2) report seed viability of three mm particles as a baseline for comparison with other jarrah viability reports that do not include seed size.

Methods

In this study, 198 seed traps were placed in forest (n = 32) and restoration sites

(n = 166) to capture jarrah seed and chaff. For more detail of the seed collection sites and the sampling method, refer to Section 3.3.3. Particles assessed in this study were smaller than three mm in size (smaller than four mm particles in the study reported in Section 3.3.3). From these traps, 2,837 particles were X-rayed using the Faxitron Specimen Radiography System

(MX-20 Cabinet X-ray Unit; Faxitron, Wheeling, IL, USA). Images were assessed to determine seed fullness. Viable seeds were determined only when the particle was completely full, based on the presence of a white-to-cream coloured solid fill, illustrating the presence of endosperm and embryo (Martyn et al., 2011; Tangney et al., 2019; Figure 3.1). Final results reported are jarrah seedfall (2,132 particles) from 26 of the 32 forest sites that did not have E. patens trees present in the stand, because of similarity in appearance in seeds of these two species (Boland et al., 1980). JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 138

Results

Distinguishing between unviable (‘empty’ or aborted) (Figure 3.1a), and viable

(‘full’) jarrah seed (Figure 3.1b) by ‘fullness’ for the 2,132 particles was straightforward.

Chaff (Figure 3.1c) was structurally very different from seed: uniformly dense, but not ‘full’.

Partially full seeds (Figure 3.1a) were easy to identify and seeds presenting this way were not considered viable. The proportion of all smaller than three mm particles that were viable jarrah seed was variable in the forest: 0.14 ± 0.17% (mean ± SD). Of seedfall in the restoration sites, viability of seed was 0.14 ± 0.31%. Four marri seeds were captured in

January (n = 2), March (n = 1) and November (n = 1); three from forest sites and one from a restoration site (0–15 m from forest edge). X-ray images of all marri seeds are presented in

Figure 3.2.

a) b)

c)

1 mm

Figure 3.6 X-ray of inviable jarrah seed (partial ‘fullness’; a), viable seed (b) and inviable chaff (c). X-ray image produced on Faxitron Specimen Radiography System (MX-20 Cabinet X-ray Unit; Faxitron, Wheeling, IL, USA) and cropped and simplified from larger X-ray image to include the three capsule particles.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 139

a) b)

1cm 1cm

c) d)

1cm 1cm

Figure 3.7 X-ray of all four marri seeds captured in this study and classified as ‘completely full’. Location and timing of marri seedfall was a) forest trap in January, b) forest trap in March, c) restoration site trap (six m from forest edge) in January, and d) forest trap in November. Other particles in images are jarrah. X-ray image produced on Faxitron Specimen Radiography System (MX-20 Cabinet X-ray Unit; Faxitron, Wheeling, IL, USA).

Applications and future work

X-ray seed imaging was a practical, non-destructive means of assessing jarrah and marri viability. Images included herein can help future studies determine viable jarrah seed from chaff particles and unviable seed using X-ray imaging (Figures 3.6 & 3.7). The proportion of viable three mm jarrah particles (14% in both forest and restoration sites) was low compared with that in seed mix of particles smaller than four mm (60% mean viability from cut and germination tests; Barker, 2019). These reported mean viabilities of three mm

(14%) and four mm particles (60%) provide a guide for estimating the viability of jarrah seed based on the particle size cut-off for jarrah seed mix.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 140

Appendix 3: Generalised Linear Model Outputs of Best Models, Section 3.3

Below are the outputs of the best fit GLMs and pseudo R2 values from Stata (Ver. 15; 2017) used in Section 3.3, Table 3.2.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 141

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 142

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 143

Appendix 4: Generalised Linear Models Outputs of Best Models, Chapter 4

Below are the outputs of the best fit GLMS and pseudo R2 values from Stata (Ver. 15; 2017) used in Chapter 4, Table 4.2. There were no response variables with a significant relationship with jarrah density, so no model is provided for jarrah.

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 144

JARRAH AND MARRI ESTABLISHMENT AFTER BAUXITE MINING 145

Appendix 5: Glasshouse germination trial set-up, Chapter 4

Jarrah None Marri Marri Jarrah Jarrah None Marri Marri Jarrah Jarrah None Marri Marri Jarrah Jarrah Marri None Marri Jarrah Jarrah Marri None Jarrah Marri Jarrah Marri None Jarrah Marri Marri Jarrah None Jarrah Marri Marri Jarrah Jarrah None Marri Marri Jarrah

Door Walkway

Marri None Jarrah Jarrah Marri Marri None Jarrah Jarrah Marri Marri None Jarrah Jarrah Marri Jarrah Jarrah Marri None Marri Jarrah Jarrah Marri Marri None Jarrah Jarrah Marri Marri Marri Jarrah None Jarrah Marri Marri Jarrah None Marri Jarrah Marri Jarrah None

Figure 4.11 Glasshouse layout from Alcoa’s Marrinup Nursery (10 km from the Huntly Mine; Figure 2.1) For experiment C (topsoil, green; overburden, yellow; gravel, orange; and pit floor, blue). Location of 42 trays were set by randomised block design, with four replicates from each substrate type including four jarrah, four marri, and four unseeded replicates (to control for potential seed storage in the substrate). Jarrah and marri seed was applied to 36 trays with potting mix to control for seed viability (grey).