In Tropical Iron Ore Systems Alan Levett B.Sc. H. Majoring in Geological

Development of ferruginous duricrusts (canga) in tropical iron ore systems

Alan Levett

B.Sc. H. Majoring in Geological Sciences (UQ) 2014

B.Sc. Extended Major in Biomedical Sciences (UQ) 2011

A thesis submitted for the degree of Doctor of Philosophy at

The University of Queensland – 2020

School of Earth and Environmental Sciences

Abstract Iron-rich duricrusts (canga) that cap iron ore deposits in Brazil evolve by the redox-based cycling of iron oxide minerals, continually forming new iron-rich cements that effectively prevent the physical erosion of these landforms. This thesis aimed to understand the biological influences on the geochemical cycling of iron within canga, to determine if these processes could be used to accelerate iron cement formation, for example, in a post-mining context. Re- cementing waste material (canga) from iron ore mining may provide an environmentally- friendly solution for iron ore mine remediation in Brazil and the restoration of a rare ecosystem associated with canga outcrops. Field emission scanning electron microscopy revealed an array of microorganisms fossilised within canga, highlighting the innate role of microorganisms in the evolution of canga horizons. Nanoscale elemental mapping of these microfossils revealed a role for aluminium in the long-term preservation of organic and inorganic biosignatures. A suite of high-resolution analytical techniques were correlated to understand the fossilisation processes that contribute to biosignature preservation in near-neutral iron-rich environments. Further investigation of the microfossils in canga, revealed the complex weathering patterns induced by microorganisms. Grains within canga could be completely weathered and texturally replaced by microfossils; only remnants of the grain’s shape and traces of most immobile elements (for example, thorium) were left behind. The minor and trace element proportions of the canga cements revealed the microscale redistribution of titanium, likely to be driven by microorganisms, and the long-range redistribution of phosphorus, both preserved within these iron-rich duricrusts. Aeolian phosphorus from Africa appears to provide an external phosphorus source for the canga-associated ecosystems.

As part of fieldwork in the montane regions of Northern Brazil, a lake-edge microbial ecosystem where canga cements appeared to form relatively rapidly was characterised. Examining these rocks, fossilised biofilms were found to play an intricate role in the formation of the meniscus-type cements within canga. Without the organic framework provided by microbial biofilms, and connecting grains together, the iron oxide minerals would simply have enlarged the existing grains without making cements. Armed with new information revealing the microbiome in canga plays an important role in the dissolution and the re-precipitation of iron oxides, these processes were applied to accelerate the reformation of iron-rich crusts in the laboratory. Microbially-mediated iron reduction was promoted within an open-air bioreactor and the iron-rich solutions were allowed to flow over crushed canga on a slope of approximately 10. Astonishingly, within 6 months the top 3 – 5 cm of the experiment had

been stabilised by fossilisation of the microbial biofilms that naturally aggregate materials. This experiment is a platform for field-scale trial aiming to progressively stabilise the slope of an expended mine site. Understanding and harnessing microbial processes that contribute to iron cement formation in canga offers an alternative remediation strategy for iron ore mines in Brazil.

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

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my research higher degree candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award. I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis.

iii

Publications during candidature Peer-reviewed Journal Articles: Levett, A., Gagen, E.J., Shuster, J., Rintoul, L., Tobin, M., Vongsvivut, J., Bambery, K., Vasconcelos, P., Southam, G. (2016) Evidence of biogeochemical processes in iron duricrust formation. Journal of South American Earth Sciences, 71, 131-142.

Gagen, E.J., Levett, A., Shuster, J., Fortin, D., Vasconcelos, P., Southam, G. (2018) Microbial diversity in actively forming iron oxides from weathered banded iron formation systems. Microbes and Environments, 33, 385-393.

Spier, C.A., Levett, A., Rosière, C.A. (2018) Geochemistry of canga (ferricrete) and evolution of the weathering profile developed within itabirite and iron ore in the Quadrilátero Ferrífero region, Minas Gerais, Brazil. Mineralium Deposita, 54, 983-1080.

Levett, A., Gagen, E.J., Diao, H., Guagliardo, P., Rintoul, L., Paz, A., Vasconcelos, P.M., Southam, G. (2019) The role of aluminium in the preservation of microbial biosignatures. Geoscience Frontiers, 10, 1125-1138.

Gagen, E.J., Levett, A., Paz, A., Cecilio, C.F., Southam, G., Oliveira, G., Siqueira, J., Bittencourt, J., Gastauer, M., Vasconcelos, P., Valadares, R., Alves, R. (2019) Biogeochemical processes in canga ecosystems: protection of iron ore and importance in iron duricrust restoration in Brazil. Ore Geology Reviews, 107, 573-586.

Levett, A., Gagen, E.J., Southam, G. (2019). Small but mighty: microorganisms offer inspiration for iron ore remediation and mine waste stabilisation. Microbiology Australia, 40, 190-194.

Paz, A., Gagen, E.J., Levett, A., Zhao, Y., Kopittke, P.M., Southam, G. (2020) Biogeochemical cycling of iron oxides in the rhizosphere of plants grown on ferruginous duricrust (canga). Science of the Total Environment, 713, 136637.

Levett, A., Gagen, E.J., Vasconcelos, P.M., Zhao, Y., Paz, A., Southam, G. (2020) Biogeochemical cycling of iron: implications for biocementation and slope stabilisation. Science of the Total Environment, 707, 136128.

Levett, A., Vasconcelos, P.M., Gagen, E.J., Rintoul, L., Spier, C.A., Guagliardo, P., Southam, G. (2020) Microbial weathering signatures in lateritic ferruginous duricrusts. Earth and Planetary Science Letters 538, 116209.

Conference Abstracts: Levett, A., Gagen, E.J., Southam, G. (2019) Microorganims promote hillslope stabilisation and mine waste stabilisation. Australian Microbial Ecology Conference (AusME), Western Australia, February 11-13.

Paz, A., Gagen, E.J., Levett, A., Southam, G. (2019) Evidence of primary surface colonizers on iron duricrust (canga) in Carajás, Brazil. Australian Microbial Ecology Conference (AusME), Western Australia, February 11-13.

Paz, A., Gagen, E.J., Levett, A., Southam, G. (2019) Biogeochemical cycling of iron oxide in the rhizosphere of plants grown on ferruginous duricrust (canga) in Carajás, Brazil. Goldschmidt, Barcelona, 18-23 August.

Levett, A., Paz, A., Gagen. E.J., Diao, H., Guagliardo, P., Rintoul, L., Vasconcelos, P., Southam, G. (2018) The role of aluminium in microbial fossilisation and preservation of organic biosignatures. Gordon Research Conference: Geobiology, Galveston, Texas, USA. January 21-26.

Levett, A., Paz, A., Gagen. E.J., Diao, H., Guagliardo, P., Rintoul, L., Vasconcelos, P., Southam, G. (2017) Preservation of organic and inorganic biosignatures in iron oxide mineralised microfossils. The 23rd International Symposium for Environmental Biogeochemistry, Cairns, Australia. September 25-29.

Levett, A., Gagen, E.J., Shuster, J., Paz, A., Southam, G. (2016) Iron oxide diagenesis: the role of iron-oxidising bacteria. The 26th Annual V.M. Goldschmidt Conference, Yokohama, Japan, June 26-July 1.

Southam, G., Gagen, E.J., Levett, A., Monteiro, H., Paz, A., Vaconcelos, P. (2016) The importance of iron duricrust formation to Brazil iron ore production. The 26th Annual V.M. Goldschmidt Conference, Yokohama, Japan, June 26-July 1.

Gagen, E.J., Levett, A., Shuster, J., Monteiro, H., Rintoul, L., Tyson, G., Vasconcelos, P., Southam, G. (2015) Biogeochemical cycling of iron in canga ecosystems. Gordon Research Conference: Applied and Environmental Microbiology. South Hadley, USA. July 12–17.

Accepted publications included in thesis Levett, A. Gagen, E.J., Diao, H., Guagliardo, P., Rintoul, L., Paz, A., Vasconcelos, P.M., Southam, G. (2019) The role of aluminium in the preservation of microbial biosignatures. Geoscience Frontiers, 10, 1125-1138.

This paper is incorporated as Chapter Three of this thesis. Author contributions were as follows:

A. Levett Experiment design (90%)

NanoSIMS data processing (100%)

Electron microscopy imaging (100%)

Manuscript preparation (80%)

Infrared data analysis (20%)

E.J. Gagen Manuscript preparation (10%)

Experiment design (5%)

H. Diao FIB-SEM sample preparation (100%)

P. Guagliardo NanoSIMS data collection (100%)

L. Rintoul Infrared data collection (100%)

Infrared data analysis (80%)

A. Paz Experiment design (5%)

P.M. Vasconcelos Manuscript preparation (5%)

G. Southam Manuscript preparation (5%)

This paper is incorporated as Chapter Five of this thesis. Author contributions were as follows:

A. Levett Experiment design (100%)

NanoSIMS data processing (100%)

Electron microscopy imaging (100%)

Manuscript preparation (80%)

Infrared data analysis (70%)

Data Interpretation (50%)

P.M. Vasconcelos Data interpretation (30%)

Manuscript preparation (5%)

E.J. Gagen Manuscript preparation (5%)

L. Rintoul Infrared data collection (100%)

Infrared data analysis (30%)

C.A. Spier Data interpretation (20%)

Manuscript Preparation (5%)

P. Guagliardo NanoSIMS data collection (100%)

G. Southam Manuscript preparation (5%)

This paper is incorporated as Chapter Seven of this thesis. Author contributions were as follows:

A. Levett Project design (50%)

DNA processing (50%)

Electron microscopy imaging (100%)

Manuscript preparation (70%)

E.J. Gagen DNA processing (50%)

Project design (10%)

vii

Manuscript preparation (20%)

P.M. Vasconcelos Manuscript preparation (5%)

Y. Zhou Nanoindentation collection (100%)

A. Paz Project design (5%)

G. Southam Project Design (35%)

Manuscript preparation (5%)

Levett, A., Gagen, E.J., Vasconcelos, P.M., Zhao, Y., Southam, G. (2020) Biocement stabilisation of an artificial slope and reformation of iron-rich crusts. Accepted with major revisions to Proceeding of the National Academy of Sciences of the United States of America.

This paper is incorporated as Chapter Eight of this thesis. Author contributions were as follows:

A. Levett Experimental design (80%)

Electron microscopy imaging (100%)

Manuscript preparation (70%)

Drop-cone experiment (100%)

DNA analysis (90%)

Water chemistry analysis (100%)

E.J. Gagen Manuscript preparation (15%)

DNA analysis (10%)

P.M. Vasconcelos Manuscript preparation (5%)

Y. Zhao Nanoindentation (100%)

G. Southam Manuscript preparation (5%)

Submitted publications Levett, A., Gagen, E.J., Rintoul, L., Guagliardo, P., Vasconcelos, P.M., Southam, G. (In review, 2020) Characterisation of iron oxide encrusted microbial fossils. Scientific Reports.

This paper is incorporated as Chapter Four of this thesis. Author contributions were as follows:

viii

A. Levett Experimental design (90%)

NanoSIMS data processing (100%)

Electron microscopy imaging (100%)

Manuscript preparation (80%)

Infrared data analysis (100%)

E.J. Gagen Manuscript preparation (5%)

Experimental design (10%)

L. Rintoul Infrared data collection (100%)

P. Guagliardo NanoSIMS data collection (100%)

P.M. Vasconcelos Manuscript preparation (5%)

G. Southam Manuscript preparation (10%)

Levett, A., Vasconcelos, P.M., Jones, M.M.W., Rintoul, L., Paz, A., Gagen, E.J., Southam, G. (In review, 2020) Titanium mobility and long-range phosphorus transport signals preserved in ferruginous duricrusts. Chemical Geology.

This paper is incorporated as Chapter Six of this thesis. Author contributions were as follows:

A. Levett Experimental design (90%)

NanoSIMS data processing (100%)

Electron microscopy imaging (100%)

Manuscript preparation (75%)

Infrared data analysis (100%)

XFM data processing (100%)

P.M. Vasconcelos Manuscript preparation (20%)

M.M.W. Jones XANES data processing (100%)

L. Rintoul Infrared data collection (100%)

A. Paz Experiment design (5%)

E.J. Gagen Manuscript preparation (5%)

G. Southam Manuscript preparation (5%)

Contributions by others to the thesis Contributions by others to the thesis are outlined in the Acknowledgements section of each chapter.

Statement of parts of the thesis submitted to qualify for the Award of another degree None.

Research involving human or animal subjects No animal or human subjects were involved in this research.

Acknowledgements Dad had a saying, ‘you will never regret the education you get.’ I suspect he will be right about this one as well. For instilling in me with a work ethic to last a life time, encouraging me to pursue my dreams and for the after-school algebra work sheets from age seven, my thesis is dedicated to my dad, Malcolm Paul Levett. To my mother, Linda, your continued love and support means the world to me. My siblings, Ian, Peter and Melissa, it has been so great to share this time in Brisbane together. Hopefully, by some miracle, we may all find ourselves living in close quarters again one day. To my fiancée, Emily, I would need a second thesis to thank you for your unbelievable strength and support. Coming home to you and Bronte has enriched my life beyond compare.

My supervisors, Gordon, Emma and Paulo; from an undergraduate you empowered me with a belief I often did not have in myself. Gordon, thank you for the endless support and fostering such an enjoyable and creative environment in which to work. Emma, your patience, guidance and leadership are inspiring. Paulo, thank you for pushing me to be a better a scientist, for which I surely am with your guidance. Thank you to everyone in the Geomicro laboratory, past and present, who made it such a fun place to learn. In particular, Anat, thank you for unwavering friendship and daily discussions about science, a rock no one has heard of and life.

To all the technical assistants at the CMM (UQ), CARF (QUT), CMCA (UWA), Rock preparation laboratory, Australian Synchrotron and Advanced Light Source: without you, this thesis would not have been possible. Llew Rintoul, thank you for our science discussions mixed with laughing to the point of tears; yours is an unlikely friendship I am very grateful to have.

Financial Support I acknowledge support the Vale S.A.-UQ Geomicrobiology initiative and the Australian Research Council Linkage Program (LP140100805) to G. Southam and P. Vasconcelos. This research was supported by an Australian Government Research Training Program Scholarship.

Keywords canga; microfossils; biocements; redox iron cycling; electron microscopy; infrared spectroscopy; NanoSIMS; bioremediation

Australian and New Zealand standard research classification (ANZSRC) ANZSRC code: 040202, Inorganic Geochemistry, 50%

ANZSRC code: 060599 Microbiology not elsewhere classified (Geomicrobiology), 40%

ANZSRC code: 040306 Mineralogy and Crystallography, 10%

Fields of research (for) classification FoR code: 0402, Geochemistry, 50%

FoR code: 0605, Microbiology, 30%

FoR code: 0403, Geology, 20%

xii

Table of contents

Abstract ...... i

Declaration by author ...... iii

Publications during candidature ...... iv

Peer-reviewed Journal Articles: ...... iv

Conference Abstracts: ...... v

Accepted publications included in thesis ...... vi

Submitted publications...... viii

Contributions by others to the thesis ...... x

Statement of parts of the thesis submitted to qualify for the Award of another degree ...... x

Research involving human or animal subjects...... x

Acknowledgements ...... xi

Financial Support ...... xii

Keywords ...... xii

Australian and New Zealand standard research classification (ANZSRC) ...... xii

Fields of research (for) classification ...... xii

Table of contents ...... xiii

List of Figures ...... xviii

List of Tables ...... xxvi

List of abbreviations ...... xxvii

Chapter 1 Introduction and thesis structure ...... 30

1.1. Introduction ...... 30

1.2. Thesis structure and research objectives ...... 31

References ...... 32

Chapter 2 Strategising the bioremediation of iron ore mines ...... 33

Abstract ...... 34

xiii

2.1. Introduction ...... 35

2.2. What is canga? ...... 37

2.3. Drivers for the remediation of iron ore mines ...... 37

2.4. Biologically-mediated iron oxide dissolution mechanisms in canga ...... 38

2.5. Iron oxidation, biomineralization and fossilisation processes in canga ...... 40

2.6. Conclusions ...... 42

Acknowledgements ...... 43

References ...... 43

Chapter 3 The role of aluminium in the preservation of microbial biosignatures ...... 47

Abstract ...... 48

3.1. Introduction ...... 49

3.2. Materials and methods ...... 50 3.2.1. Site description and sample collection ...... 50 3.2.2. Sample characterisation ...... 51 3.2.3. Electron Microscopy ...... 52 3.2.4. Nanoscale secondary ion mass spectrometry ...... 52

3.3. Results ...... 54 3.3.1. Permineralised microfossils ...... 54 3.3.2. Encrusted cell envelopes ...... 57

3.4. Discussion ...... 62 3.4.1. Permineralised microfossils ...... 66 3.4.2. Encrusted cell envelopes ...... 66 3.4.3. A model for the role of aluminium in microbial fossilisation ...... 68 3.4.4. The relative age of microfossils ...... 70

3.5. Conclusions ...... 70

Acknowledgements ...... 71

References ...... 71

Chapter 4 Characterisation of iron oxide encrusted microbial fossils ...... 75

xiv

Abstract ...... 76

4.1. Introduction ...... 77

4.2. Materials and methods ...... 78 4.2.1. Scanning electron microscopy ...... 78 4.2.2. Focused ion beam scanning electron microscopy ...... 78 4.2.3. Nanoscale secondary ion mass spectrometry (NanoSIMS) ...... 79 4.2.4. Infrared microspectroscopy ...... 79 4.2.5. Transmission electron microscopy ...... 80

4.3. Results ...... 80

4.4. Discussion ...... 90

4.5. Conclusions ...... 94

Acknowledgements ...... 95

References ...... 95

Chapter 5 Microbial weathering signatures in lateritic ferruginous duricrusts ...... 98

Abstract ...... 99

5.1. Introduction ...... 100

5.2. Geological and environmental setting ...... 101

5.3. Methods and materials ...... 104 5.3.1. Sample collection and bulk characterisation ...... 104 5.3.2. Microscale sample characterisation ...... 104 5.3.3. Raman spectroscopy ...... 105

5.4. Results ...... 106 5.4.1 Bulk sample characterisation ...... 106 5.4.2. Geochemical signatures of microbial activity in canga ...... 106

5.5. Discussion ...... 120 5.5.1. A model for microfossil textural replacement ...... 120 5.5.2. Microbially-accelerated weathering signatures ...... 122 5.5.3. Mineral nucleation: microbial influences and abiotic precipitates ...... 124 5.5.4. Canga: extremely stable but biogeochemically dynamic ...... 125

5.6. Conclusions ...... 126

Acknowledgements ...... 126

References ...... 127

Chapter 6 Microbiological and chemical influences on the evolution of iron-rich duricrusts in the Serra Sul de Carajás, Pará, Brazil ...... 131

Abstract ...... 132

6.1. Introduction ...... 133

6.2. Geological setting ...... 134

6.3. Materials and methods ...... 135 6.3.1. Sample characterisation ...... 135 6.3.2. Electron microscopy ...... 136 6.3.3. Synchrotron-based X-ray fluorescence microscopy ...... 136 6.3.4 Raman Spectroscopy ...... 137

6.4. Results ...... 138 6.4.1. Bulk sample characterisation ...... 138 6.4.2. Electron microscopy ...... 140 6.4.3. Chemical and mineralogical mapping of select canga regions ...... 143

6.5. Discussion ...... 146 6.5.1. Titanium and chromium mobility and redistribution ...... 146 6.5.2. Chemical stability of iron oxide minerals with canga ...... 147 6.5.3. Sources of phosphorus within canga...... 150

6.6. Conclusions ...... 153

Acknowledgements ...... 153

References ...... 153

Chapter 7 Biogeochemical cycling of iron: implications for biocementation and slope stabilisation ...... 159

Abstract ...... 160

7.1. Introduction ...... 161

xvi

7.2. Geological and environmental setting ...... 162

7.3. Methods and materials ...... 166 7.3.1. Locations for sample collection ...... 166 7.3.2. Scanning electron microscopy ...... 166 7.3.3. Nanoindentation ...... 167 7.3.4. Water chemistry ...... 167 7.3.5. DNA extraction, sequencing and sequence analysis ...... 168

7.4. Results ...... 169 7.4.1. Characterisation of canga rica from the edge of Lake Violão ...... 169 7.4.2. Water chemistry ...... 174 7.4.3. Microbial community structures ...... 176

7.5. Discussion ...... 181 7.5.1. Ferruginous biocement formation: industrial applications ...... 182 7.5.2. Limitations of natural iron reduction ...... 183 7.5.3. Contributors to biocement formation: iron oxidation and microbial fossilisation: ...... 183

7.6. Conclusions ...... 185

Acknowledgements ...... 185

References ...... 186

Chapter 8 Biocement stabilisation of an artificial slope and the reformation of iron-rich crusts ...... 189

Abstract ...... 190

8.1. Introduction ...... 191

8.2. Materials and methods ...... 192 8.2.1. Experimental Vessels ...... 192 8.2.2. Drop-cone test ...... 193 8.2.3. Scanning electron microscopy ...... 193 8.2.4. Water chemistry ...... 194 8.2.5. DNA extraction, sequencing and sequence analysis ...... 196 8.2.6. Nanoindentation ...... 198

xvii

8.3. Results ...... 199 8.3.1. Formation of iron biocements ...... 199 8.3.2. Iron fossilisation of microorganisms ...... 199 8.3.3. Open-air bioreactor ...... 206 8.3.4. Microorganisms in the bio- and control reactors ...... 206 8.3.5. Microorganisms fossilised to create biocements ...... 207

8.4. Discussion ...... 209

8.5. Conclusions ...... 213

Acknowledgements ...... 213

Reference ...... 213

Chapter 9 Discussion, implications and conclusions ...... 217

9.1. Distributions of biosignatures in canga...... 217

9.2. Canga: a reservoir for iron-associated microbial life ...... 218

9.3. Accelerated canga formation in natural environments ...... 221

9.4. Promoting iron-cycling to accelerate re-cementation ...... 222

9.5. Harnessing biogeochemical processes for iron ore bioremediation ...... 223

9.6. Final observations ...... 225

References ...... 226

Appendix 1 Supplementary information for Chapter Five ...... 228

Appendix 2 Supplementary information for Chapter Seven...... 241

Appendix 3 Supplementary information for Chapter Eight ...... 242

List of Figures Fig. 2.1. Digital elevation model of the Carajás region generated using the Shuttle Radar Topography Mission (SRTM) from the USGS Earth Explorer (https://earthexplorer.usgs.gov/). N1 highlights some canga regions in Serra Norte, S11 highlights canga regions in Serra Sul, Carajás Mineral Province, Pará...... 36 Fig. 2.2. Weathering profile from S11D region. Canga forms a highly resistant cap above the friable iron ore within the saprolite (A). Canga is highly heterogeneous, primarily composed of detrital fragments that have undergone several generations of recycling, cemented together by authigenic iron oxide precipitation (B and C). Many of the iron concretion textures have organic-like textures (D). Dissolution-depressions

xviii

within of ferruginous duricrusts form ephemeral pools that are often covered in iron sheens indicating the interaction of iron with microbial biofilms (E)...... 39 Fig. 2.3. Model to highlight the main mechanisms hypothesised to play a role in iron reduction and iron oxidation in canga. In bold are the mechanisms that could be used to promote the biogeochemical cycling of iron in circumneutral environments to accelerate the reformation of iron-rich concretions for iron ore remediation. Backscattered electron micrographs highlight iron mineralised biofilms that form meniscus- type concentrations, generated in an experiment in Chapter Eight...... 41 Fig. 3.1. Photograph of goethite cemented vein in the exposed saprolite from Serra do Gandarela, Quadrilátero Ferrífero (QF), State of Minas Gerais, Brazil. The white arrow indicates the region from which the hand sample was collected, approximately 15 m below the surface. (B) Photograph of goethite-cemented hand sample highlighting the presence of iron oxide coated roots...... 51 Fig. 3.2. Backscattered electron scanning electron micrographs highlighting rod-shaped microfossils (A) and large permineralised microbial biofilms (B) within an iron-rich duricrust capping iron ore deposits in the Carajás Mineral Province, State of Pará, Brazil. Permineralised microfossils formed around highly weathered kaolinite-rich clasts...... 58 Fig. 3.3. High magnification backscattered electron micrograph of rod-shaped permineralised microfossils, with evidence of colony formation and cell replication (arrows). The rectangle highlights the region analysed using NanoSIMS (see Fig. 3.4)...... 59 Fig. 3.4. NanoSIMS micrographs of permineralised microfossils presented as black and white intensity maps, with white areas having a higher relative elemental concentration compared with darker regions. NanoSIMS micrographs reveal that organic biosignatures are not preserved with permineralised microfossils. Aluminium is enriched around the microfossils, while iron is enriched within the intracellular regions of microfossils. Phosphorus colocalises with aluminium (r = 0.41). The composite micrograph highlights aluminium (green) enrichment around microfossils, with iron (blue) is enriched within intracellular regions. All micrographs are 8 × 8 µm...... 60 Fig. 3.5. Field Emission scanning electron micrographs of mineral encrusted cell envelope structures identified within a goethite cemented vein that cross-cut the saprolite of a weathered banded iron formation in the Serra do Gandarela, Quadrilátero Ferrífero in the State of Minas Gerais, Brazil. (A) Backscattered electron micrograph highlighting the preservation of encrusted cell envelopes within pore spaces. (B) Secondary electron micrograph of encrusted cell envelopes highlighting the three-dimensional preservation of encrusted cell envelopes that had formed in pore spaces...... 61 Fig. 3.6. Backscattered electron scanning electron micrograph of encrusted cell envelopes that had infilled pore spaces between gibbsite-rich clasts. The rectangle highlights the regions from which a lamella was extracted using a focused ion beam scanning electron microscope for NanoSIMS analysis (see Figs. 3.7. – 3.8.)...... 63 Fig. 3.7. High magnification backscattered electron Field Emission micrograph of encrusted cell envelope sample that had been extracted using a focused ion beam scanning electron microscope from the region highlighted in Fig. 3.6. A variety of fossilisation textures were associated with the encrusted cell envelopes. The rectangle represents the region analysed using NanoSIMS (see Fig. 3.8)...... 64 Fig. 3.8. NanoSIMS analysis of mineral encrusted cell envelopes are displayed as black and white intensity

xix

maps, with higher concentrations represented by white areas and lower concentrations displayed as darker regions. NanoSIMS micrographs highlight the preservation of carbon and nitrogen biosignatures associated with the microbial cell envelope. Phosphorus was distributed throughout the sample and only enhanced due to edge effects associated with the hollow encrusted cell envelopes. Aluminium and iron are present throughout the sample but aluminium appears to be slightly enriched where preserved organic biosignatures are preserved (see Fig. 3.9). In contrast, iron concentrations are reduced in regions where organic carbon and nitrogen are preserved. The composite micrographs highlights the poor correlation of iron (blue) with preserved organic nitrogen (green). All micrographs are 8 × 8 µm...... 65 Fig. 3.9. Intensity line plots of ion counts plotted against pixel distance cross-cutting microfossils for a partially infilled microfossil (Fig. 3.8; ROI 1) and a completely infilled microfossil (Fig. 3.8; ROI 2). For ROI 1, aluminium is enhanced in association with the organic biosignatures and was slightly enriched in regions where organic carbon and nitrogen are preserved. Iron poorly correlates with organic carbon and nitrogen and is slightly increased in intracellular lumen. Similarly, for the completely infilled cell (Fig. 3.8; ROI 2), aluminium is enhanced in regions with the preserved organic carbon and nitrogen. Note, iron is out of phase with the preserved organic biosignatures. Phosphorus is relatively consistent for both ROI 1 and 2, highlighting these cells were not affected by edge enhancements...... 69 Fig. 4.1. Backscattered electron micrographs highlight the preservation of the cell envelopes structures, possibly including extracellular polymeric substances...... 81 Fig. 4.2. (A) Backscattered electron micrograph of fossilised biofilm from which 3D reconstruction was created (see Video 4.1). (B) 3D orthogonal sections highlighting the preservation of mineralised cell envelopes during cell replication...... 82 Fig. 4.3. (A) Backscattered scanning electron micrograph of FIB lamella highlighting the region mapped using NanoSIMS (white square). (B) Composite image highlighting the relative distributions of carbon (red), nitrogen (green) and iron (blue). Carbon (C), nitrogen (D), aluminium oxide (E) and iron oxide (F) NanoSIMS elemental micrographs highlighting the preservation of carbon and nitrogen associated with the cell envelope structures...... 84 Fig. 4.4. (A) Backscattered scanning electron micrograph of FIB lamella highlighting the region mapped using synchrotron-based Fourier transform infrared spectroscopy (FT-IR). (B) Contoured heatmap generated by integrating all spectra between 2910 and 2940 cm-1, corresponding to the antisymmetric stretch of aliphatic moieties (CH2). Infrared spectrum of a region containing methylene moieties (blue spectrum), associated with the microfossil-rich regions of the lamella. Red spectrum highlights the minerals around microfossils are gibbsite-rich...... 85 Fig. 4.5. (A) Backscattered scanning electron micrograph highlighting region mapped using Raman spectroscopy, which revealed iron oxide minerals associated with microfossils are minor lepidocrocite (B) and goethite (C). Lepidocrocite is enriched within the extracellular regions between fossilised cell envelopes...... 87 Fig. 4.6. (A) Bright field transmission electron micrograph of a TEM lamella containing a mineralised cell envelope. Selected area electron diffraction patterns could not conclusively distinguish intracellular minerals (B), biologically-influenced mineralisation associated with the cell envelope structure (C) and minerals within the matrix between fossilised cell envelopes...... 89

Fig. 4.7. Bright field transmission electron micrograph of fossilised cells highlight that iron oxide minerals follow the direction of the cell envelope structure. In contrast, secondary mineralisation that infills microfossils to form permineralised fossils grow from the edge of the cell envelope structure towards to the centre of the intracellular void (arrows indicate directionality)...... 90 Fig. 5.1. Geographical context of the study site in the Serra Sul de Carajás Mineral Province, Pará, Brazil. (A) Digital elevation model of the Carajás region generated using the Shuttle Radar Topography Mission (SRTM) from the USGS Earth Explorer (https://earthexplorer.usgs.gov/). (B) Google Earth image of the Serra Sul canga plateau and the drill core location. (C) Remaining portion of the drill core sample after a billet (D) was prepared to make thin section samples...... 103 Fig. 5.2. Low magnification backscattered electron micrograph highlight microfossils forming around grains in canga (A). The white square represents the regions mapped by NanoSIMS (see Fig. 5.3.). (B) High magnification micrograph demonstrating the influence of microorganisms on the precipitation of iron oxide minerals from supersaturated solution throughout canga. Microfossils are commonly rod-shaped and fossilised via microbially-influenced biomineralisation...... 107 Fig. 5.3. Scanning electron micrograph (A) and NanoSIMS elemental micrographs (50 × 50 m) of the region highlighted by the white square in Fig. 5.2 for Al (B), Ti (C), Cr (D) and Fe (E). Aluminium is enriched with in the less weathered portions of the grains, highlighted by the green lines. Titanium and chromium are preserved in the highly weathered portion of the grain, highlighted by the blue line. Newly formed iron-rich cements (red line), including cements formed by microbial fossilisation, and are depleted in titanium and chromium. Raman spectroscopy (F) highlights the phase transformation of the titanium-rich minerals from rutile to anatase. Raman colour map: blue = goethite, red = hematite, white = anatase, yellow = rutile...... 108 Fig. 5.4. Low resolution BSE micrograph highlighting the continuous and relatively large areas containing fossilised microorganisms. The white square approximately represents the region mapped using NanoSIMS (see Fig. 5.5). (B) High-resolution insert showing the preservation of rod-shaped microfossils that form lamellae (black arrow) around a weathered grain (white oval)...... 110 Fig. 5.5. NanoSIMS elemental micrographs (50 × 50 µm) of the region highlight by the white square in Fig. 5.4. Aluminium (A), titanium (B), chromium (C), and thorium (E) are relatively enriched in the grain to the left. In comparison, the iron-enriched region around this grain shows a depletion in the titanium and chromium indicating that these are newly formed iron precipitates are not part of the weathered grain. Titanium, chromium, thorium and uranium signatures are preserved within the rounded regions where microfossils are preserved. These geochemical signatures demonstrate that the microfossils have actively weathered and texturally replaced the grain, with the chemical signatures of the immobile elements partially preserved. Raman colour map (F): blue = goethite, white = anatase...... 111 Fig. 5.6. A series of field emission BSE micrographs demonstrating the preservation of microfossils that are at various stages of weathering and texturally replacing rounded material. The weathered grains all appear to have microfossil lamellae banding, which may be a result of fossilisation at different time periods. (A) An overview of two, almost circular grains (white arrows) that have been completely weathered and replaced by microfossils. (B) High magnifcation BSE micrograph demonstrating the nature of microbial fossilisation. (C) Microfossils are initially preserved around a grain. Two lamellae bands of microfossils

xxi

can be seen. The outermost band may represent the first microfossils. As the grain is weathered, additional microfossil lamellae may replace the grains texture. The microfossils on the right have been permineralised, whereas the microfossils of the left are hollow mineralised cell envelopes. (D) Abundant rod-shaped microfossils forming around a highly weathered grain. Some microfossils have been preserved within the grain. (E) Microfossils forming around and within (high-resolution inset) a grain...... 113 Fig. 5.7. (A) Low-magnification BSE micrograph highlighting textural complexity of canga and the region mapped by NanoSIMS in Fig. 5.8. (B) Microfossils form around weathered grains, with abiotic iron oxide (hematite) precipitates further in filling pore spaces...... 115 Fig. 5.8. BSE micrograph (A) and NanoSIMS elemental micrographs (50 × 50 µm) of the region highlighted by the white square in Fig. 5.7 for Al (B), Ti (C), Cr (D) and Fe (E). At the top of the micrograph, a recombinant grain showing a classic weathering front with titanium and chromium present but depleted in the weathered portion of the grain compared with the less weathered portion. Relatively newly formed iron-rich and microbial cements are depleted in titanium and chromium. The titanium-rich fragment at the bottom of the micrographs shows a complex restructuring during weathering to form (titanium mineral name) surrounded by goethite. Minor amounts of the original mineral are also preserved between the microfossils. Raman colour map (F): blue = goethite, red = hematite, white = anatase, yellow = rutile. . 116 Fig. 5.9. (A) Low magnification BSE micrograph showing aluminium-enrichment within pore spaces. Raman colour map (B): blue = goethite, red = hematite, yellow = gibbsite. Note: gibbsite identification is affected by grain orientation. (C) Some filamentous and rod-shaped microfossils (white arrows) are preserved within the goethite cements...... 118 Fig. 5.10. NanoSIMS elemental micrographs (50 × 50 µm) for the region indicated in Fig. 5.9, demonstrating aluminium (A) enrichment within some pore spaces. Silicon (B) may be slightly enriched within the aluminium-rich pore spaces, whereas titanium (C), chromium (D), iron (E) and thorium (F) are each depleted...... 119 Fig. 6.1. Google Earth image of the Serra Sul region of the Carajás, Pará, Brazil, highlighting locations of the drill core samples in close proximity to the S11D mine and Lakes Amendoim and Violão...... 135 Fig. 6.2. Backscattered electron SEM micrographs highlighting the presence of iron oxide mineralised cell envelopes present throughout canga in the Serra Sul region, Carajás, Pará, Brazil...... 141 Fig. 6.3. (A) Backscattered electron SEM micrograph highlighting encrusted cell envelopes (white arrows) within a recombinant grain composed of detrital fragments and recent iron oxide precipitates infilling pore spaces. (B) Low resolution backscattered electron SEM micrograph of the highly weathered recombinant grain containing microfossils. White arrows highlight the edges of the grain, which is enriched in titanium (see Fig. 6.5A)...... 142 Fig. 6.4. Synchrotron-based X-ray fluorescence microscopy of a canga subsample from drill core 1 from a depth of approximately 4.9 m. Titanium (A) is enriched in rims around highly weathered grains (white arrows) and within detrital grains (asterisk). Vanadium (B) shows similar weathering patterns to titanium. Chromium (C), iron (D) and cobalt (E) are mostly strongly enriched in precipitates that form within the pore spaces in canga. Colour scale bar indicates relative element concentrations from low (black) to high (white)...... 144 Fig. 6.5. Synchrotron-based X-ray fluorescence microscopy of a canga subsample from drill core 2 from a depth

xxii

of approximately 4.2 m. Titanium (A) displays unusual rims around grains. Chromium (C), manganese (D), iron (E) and cobalt (F) are primarily enriched within the detrital fragments in the canga. Raman spectroscopy (F) of a titanium coated grain highlights that goethite (blue) has precipitated around the hematite-rich grain with microcrystalline anatase (green) disseminated throughout the goethite precipitates and in some of the microscale hematite (red) grains. Colour scale bar indicates the relative element concentrations from low (black) to high (white). Raman colour map: blue = goethite, red = hematite, green = anatase...... 145 Fig. 6.6. Backscattered electron SEM micrograph from sample DC1_4.02 m, highlighting the distribution of various bacteriomorphic fossils, including filamentous structures (white arrow) and Sporosarcina-like cells (black arrows) within the goethitic cements. (B) Low resolution backscattered electron SEM micrograph highlighting the subhedral dodecahedral form of a martite mineral embedded within abiotic and microfossil cements within canga. (C) XANES analysis from the area represented by the white dashed rectangle in Fig 6.6 (B) revealed ferrous iron is enriched within the martite (with kenomagnetite) mineral within the altered and relict grains throughout canga compared with the surrounding abiotic and microfossil cements. Colour scale bar indicates the proportion of ferrous iron to total iron from low (black) to high (white)...... 149 Fig. 6.7. Model of the influences on the evolution of canga plateaus in the Serra Sul, Carajás Mineral province in northern Brazil. The weathering and erosion of the surrounding metavolcanic units and mafic dykes and sills enriches the goethitic cements with aluminium, silicon, phosphorus, titanium, vanadium and zirconium. The very high phosphorus concentrations in the goethite cements (averaging 0.53 wt.%) also indicates a continuous phosphorus influx. Aerosols that from Africa that deposit phosphorus in the Amazon Basin (Okin et al., 2004) may also be incorporated in the ferruginous duricrusts. Organic acids exuded by plants roots and microorganisms associated with the canga mobilises titanium, which reprecipitates around nearby grains as microcrystalline anatase. Acid-induced weathering of martite releases ferrous iron, which re-precipitate to form goethite cements (Reaction 6.3*)...... 152 Fig. 7.1. (A) Google Earth image highlighting the canga plateaus of the Serra Sul (S11 regions) de Carajás Mineral Province, Pará, Brazil. Within the canga plateau are Lakes Violão and Amendoim (sample locations indicated for each), with the S11D mine towards the East. (B) Photograph of the edge of Lake Violão highlights the red-stained rocks where streams flow over the canga into the lake...... 163 Fig. 7.2. Photographs of microbial biofilms that grew at the edges of Lake Amendoim (A and B) and Lake Violão (C – F). (A) Microbial ‘globular’ biofilms form along grass roots that grow next to the stream within a relatively stagnant pool. (B) Fresh dark (black) biofilm next to the stream was sampled. Dried, flaky biofilm coated the canga above the stream. (C) An aqueous seep that flowed through the canga (karst-like feature) stained the rock below red. No iron in solution was detected here. (D) Iron-stained microbial biofilms coated the rocks in regions where organic matter accumulates within a stream that flowed in Lake Violão. (E) Foam that accumulated at the edge of Lake Violão. (F) An iron-stained microbial biofilm within a stream...... 165 Fig. 7.3. Photograph of the polished section of the canga that formed at the edge of Lake Violão. The black rectangle highlights the representative region used to calculate cement (~23%) to fragment (~77%) surface area ratio. The low cement to fragment ratio is consistent with Dorr’s (1964) description of a rock called

xxiii

‘canga rica.’ (B) Backscattered electron micrograph of large microbial fossils and small encrusted cell envelopes within the pore spaces (white arrows) that contribute to the formation of cements within the canga rica...... 171 Fig. 7.4. (A) Backscattered scanning electron micrograph revealing limited numbers of sporadically emplaced cell envelope structures within heterogeneous cements. (B) Low magnification backscattered scanning electron micrograph of cements forming next to a large, weathered fragment within the canga. The inner black portion of the black-and-yellow line traces approximately outlines the heterogeneous cements that contain microbial remnants and are considered to have a ‘biological texture’...... 172 Fig. 7.5. Backscattered scanning electron micrograph of a meniscus cement within the canga rica sample. (A) Poorly preserved cocci, baccili and possible remnants of filamentous microfossils contribute to the formation of the ferruginous cement. (B) Low magnification micrograph highlighting the heterogeneous nature of the biologically-textured cements. Nanoindentation tests for the microbial cements, neighbouring goethite cements, hematite-rich fragment (highlighted in Fig. 7.3A) and resin (n – number of

tests, x̅ H – average hardness, 휎H – standard deviation for hardness, x̅ Er – average reduced modulus, 휎H – standard deviation for reduced modulus)...... 173 Fig. 7.6. Phylum level classification of unique 16S rRNA sequences associated with microbial communities and biofilms collected from steams and seeps around Lake Amendoim (AS-GB and AS-BlB), the surface (VP- SB) and bottom (VP-BB) of an oxidised ephemeral pool surrounding Lake Violão, at the edge of Lake Violão (LV-F) and streams and seeps that flowed into Lake Violão (VS-RB, VS-OB and VS-FeS). Phyla below 1.5% relative abundance are grouped into other including, Altiarchaeota, Archaea_unclassified, BRC1, Chlamydiae, Cloacimonetes, Crenarchaeota, Cyanobacteria, Deferribacteres, Deinococcus- Thermus, Dependentiae, Diapherotrites, Elusimicrobia, Epsilonbacteraeota, Euryarchaeota, FBP, FCPU426, Fibrobacteres, Firestonebacteria, Fusobacteria, GAL15, Gemmatimonadetes, Hydrogenedentes, Kiritimatiellaeota, Latescibacteria, Lentisphaerae, Margulisbacteria, Nitrospinae, Nitrospirae, Omnitrophicaeota, Rokubacteria, Spirochaetes, Tenericutes, Thaumarchaeota, WOR-1, WPS- 2, WS1, WS4, Zixibacteria...... 177 Fig. 7.7. Heatmap analysis of 16S rRNA gene sequences showing the most abundant OTUs (OTU clustered at a distance ≤ 0.03) in each of the biofilms that formed in the streams that feed Lake Amedoim and Lake Violão and within Lake Violão. The scale bar represents relative abundance within samples from red (most abundant) to black (least abundant), with white indicating undetected OTUs. The nearest named representative in the public domain and its accession number are provided for each OTU, with blue text used to highlight putative iron-oxidising lineages and red for putative iron-reducing lineages...... 180 Fig. 8.1. Schematic model of experimental setup for the bioreactor treatment. A peristaltic pump set at a pump rate of ~ 21.6 mL/hour for 8 hours/day provided the bioreactor with a renewed nutrient source. With a few exceptions, this pump regime was used for 24 weeks for the 26-week experiment. The fermentation-driven bioreactor produced approximately 80 ppm of ferrous iron, which was allowed to ‘overflow’ onto the artificial slope. A thick biofilm covered the surface of the bioreactor, which was anoxic below the surface. An identical uninoculated control experiment, fed with sterilised anoxic milliQ water was run simultaneously. Solutions initially stayed at the surface before falling to the bottom of the experiment, then gradually rising and dripping out of the experiment at the ‘runoff’ point. At the end of the

xxiv

experiment, the bioreactor and control slopes were subdivided into six regions for sampling: ‘R’ denotes bioreactor treatment with ‘C’ used for control samples. P1 (0 – 8 cm), P2 (8 – 16 cm) and P3 (16 – 25 cm) indicates the distance from the control and bioreactor overflow points, with ‘S’ and ‘D’ denoting surface and depth samples, respectively. Photographs of the experimental setup, highlighting mineralised vs unmineralised surfaces of the bioreactor treatment and the control treatment, respectively...... 195 Fig. 8.2. Low magnification backscattered electron micrographs highlighting meniscus-style cements (thin arrows), partially mineralised biofilms (dashed arrow) and cracked ‘bridging’ structures. High magnification insets in (B) highlight these mineralised biofilms are composed of rod- and cocci-shaped microbial casts...... 201 Fig. 8.3. Backscattered electron micrographs of the surface biofilm that coated bioreactor treatment slopes. High magnification micrograph (A) revealed the high cell density and evidence of binary-fission (arrow) within fossilised biofilms. Low magnification micrograph (B) demonstrates the continuous nature of the mineralised surface biofilm on the bioreactor treatment slope. Nanoindentation tests for the microfossil cements (Regions 1 and 2), neighbouring goethite cements formed prior to experiment (Region 3),

hematite-rich fragment (Region 4) and resin (Region 5). Note: n – number of tests, x̅ H – average hardness

(GPa), 휎H – standard deviation for hardness (GPa), x̅ Er – average reduced modulus (GPa), 휎H – standard deviation for reduced modulus (GPa)...... 203 Fig. 8.4. Backscattered electron micrograph of mineralised biofilms coating grains (thin arrows). Micrograph (B) revealing different stages of mineralisation: partially mineralised biofilms with encrusted cell envelopes (thin arrow) and microscale fragments dispersed throughout the biofilm with high magnifcation inset revealing the preservation of rod-shaped bacteriomorphic casts...... 204 Fig. 8.5. Backscattered electron micrograph of the surface biofilm in the bioreactor treatment slope, approximately 24 cm from the overflow of the bioreactor, revealing large (ca. 5 m is diameter), eukaryotic-type cell structures that had been encrusted in iron oxide minerals. Smaller (ca. 1 m in diameter), cocci-shaped cell envelopes were also immobilised by the ferrous iron-rich solutions (arrows)...... 205 Fig. 8.6. Heatmap analysis of 16S rRNA gene sequences from water samples collected after 13, 17 and 23 weeks of experiment runtime from the bottom and top of the bioreactor and control reactor chambers as well as from the runoff collected. Relative abundances greater than 5% are indicated for the most prevalent OTUs. Simultaneous samples were collected to determine aqueous geochemistry using ICP- OES (Appendix 3; Tables S1 – S9). Iron-reduction in the bioreactor was driven by fermentation (Gagen et al., 2019), with putative iron-reducing microorganisms listed in blue text. The nearest named isolate in the public domain and its accession identification code is provide for each OTU. Aqueous elements that significantly (P < 0.05) correlated (r > 0.6) with OTUs are listed. OTUs were clustered at a distance of ≤ 0.03...... 207 Fig. 8.7. Heatmap analysis of 16S rRNA gene sequences from rock samples collected at the end of the experiment from various distances (P1 = 0 – 8 cm; P2 = 8 – 16 cm; P3 = 16 – 25 cm) away from the overflow of the bioreactor (R) and the control (C) reactor. Microbial lineages of the surface samples (S) were compared with samples collected from depth (D). The slope in the bioreactor treatment were dominated by Gram-negative microbial lineages, OTUs 1, 5 and 6, with relative abundances greater than

xxv

5% provided. The nearest named isolate from public domains and its accession identification code are listed for each OTU. Rock material from the control reactor and bioreactor were also sampled after dehydration. OTUs were clustered at a distance of ≤ 0.03...... 208 Fig. 8.8. Heatmap analysis of internal transcribed spacer 2 (ITS2) fungal gene sequences from rock samples collected at the end of the experiment from various distances (P1 = 0 – 8 cm; P2 = 8 – 16 cm; P3 = 16 – 25 cm) away from the overflow of the bioreactor (R) and the control (C) reactor...... 209 Fig. 9.1. (A) An ephemeral pool, common in depressions in the canga plateaus, with a surface sheen biofilm. (B) Unstained transmission electron micrograph, revealing the heterogeneous iron oxide minerals that nucleate on the cells’ surfaces and extracellular polymeric substances...... 219 Fig. 9.2. A focal series of backscattered electron micrographs of encrusted cell envelope structures that have been mineralised in the goethite-rich veins throughout iron-rich substrates. As iron-rich solutions percolate throughout the profile, microorganism become fossilised, altering the hydrological connectivity...... 221 Fig. 9.3. Photograph of hematite-rich fragments that are cemented on a slope, highlighting the relatively fast formation of some iron-rich cements in canga...... 222 Fig. 9.4. A conceptual model of the first stage of a bioremediation strategy for iron ore mines. Specific proportions of crushed waste material (canga) cover the first two or three benches at the base of the closed mine. Iron-reducing bioreactors are placed above loose material. Solutions from the bioreactor are drip- fed throughout the slope. The partially-degraded carbon source leaving the bioreactor would promote heterotrophic microbial growth at the surface to naturally aggregate grains. The iron-rich solutions (up to 300 ppm) produced in the bioreactor and evenly distributed throughout the artificial slope would mineralise the biofilms, forming ferruginous biocements. Continued iron precipitation after cell fossilisation will continue to strengthen these cements. Once the lower section is stabilised, these processes would be repeated for the next three pit benches. Solutions would be collected at the base of the pit and recycled, creating a closed-loop system...... 225

List of Tables Table 3.1. Bulk chemical (XRF) and mineralogical characteristics and the primary fossilisation texture with the ferruginous duricrust sample and the goethite-cemented sample associated...... 56 Table 6.1. Major, minor and trace element sample concentrations determined using X-ray fluorescence (wt.%)...... 139 Table 6.2. . Correlation matrix for drill core 1 – bulk sample. Positive correlations (r > 0.6) are highlighted in yellow; negative correlations (r < -0.6) are highlighted in blue...... 140 Table 6.3. Correlation matrix for drill core 2 – bulk sample. Positive correlations (r > 0.6) are highlighted in yellow; negative correlations (r < -0.6) are highlighted in blue...... 140 Table 6.4. Perceived enrichment factors in canga from the Serra Sul compared with supergene enrichment factors from iron ores (saprolite) around the world (Carajás and Urucum, South America, Mount Tom Price, Australia and Maremane Dome, South Africa). Atypical enrichment factors within the canga are highlighted in yellow...... 150

xxvi

Table 7.1. . Water chemistry for sampled locations. Cation concentrations determined using ICP-OES. A dash (-) is used to indicate were a measurement was not taken...... 175 Table 8.1. Average water chemistry (ppm) ± standard error over the 26-week experiment. Sample taken from top and bottom of bioreactor showed no statistical difference and were combined into a single data set. See Appendix 3 - Tables S1 – S9 for complete set of data...... 206

List of abbreviations % parts per hundred

°C Celsius

ΔG°f Gibbs free energy of formation under standard conditions

μm micrometre

μXRD micro X-ray diffraction aq aqueous

BCM biologically controlled mineralisation

BIFM biologically influenced mineralisation

BIM biologically induced mineralisation

BSE-SEM backscattered electron scanning electron microscopy dH2O dionized water

DL detection limit

DO dissolved oxygen

EDS energy dispersive x-ray spectroscopy

EPS extracellular polymeric substances

FE-SEM field emission scanning electron microscopy

FIB-SEM focused ion beam scanning electron microscopy g grams

Ga Giga-annum (billion years)

Gt gigatons

xxvii

h hours

ICP-OES inductively coupled plasma-optical emission spectroscopy keV kiloelectronvolt kJ kilojoules km kilometre kV kilovolt

L litre m meter

M moles per litre

Ma milli-annum (million years)

Mm millimetre mM millimoles per litre mol moles msec millisecond nm nanometre

NanoSIMS nanoscale secondary ion mass spectrometry pH potential of hydrogen (-log[H+]) ppb parts per billion ppm parts per million s second

SAED selected area electron diffraction

SE secondary electron

SEM scanning electron microscopy t metric tons

xxviii

TEM transmission electron microscopy

TM trademark vol. volume wt.% weight percent

XFM X-ray fluorescence microscopy

XRD X-ray diffraction

xxix

Chapter 1 Introduction and thesis structure

1.1. Introduction Mountains are literally moved to produce the world’s resources. As society aims to move towards a circular economy, increasing pressures are applied to the sustainability of the resources sector. Toy and Griffith (2001) provide an evaluation on the changing societal values that influence political and commercial reform towards effective reclamation practices for iron ore mines in Brazil. The massive scale of iron ore mining in Brazil presents a logistical challenge for remediation. In Brazil, Vale export more than 300 million tonnes of high-grade iron ore, making them the world’s largest iron ore producers (United States Geological Survey, 2018). Iron ore reserves in Brazil exceed 10s of billions of tonnes, with long-term large-scale mines creating serious environmental challenges.

Perhaps these challenges are best exemplified by the tragic collapse of two upstream iron ore tailings dams in recent years (Hinman, 2019; Segura et al., 2016). These catastrophic events resulted in the deaths of mining employees and civilians as well as some of Brazil’s largest environmental disasters with potential liabilities of up to US$55 billion (Armstrong et al., 2019). Personally, these disasters forced me to re-evaluate the potential implications of developing novel biocements as a part of my thesis research. The initial focus of this work was reasonably descriptive and characterisation-based: could we demonstrate Dorr’s (1964) hypothesis that microbiology plays a role in the formation and evolution of the iron-rich duricrusts (locally known and referred to throughout this thesis as canga) that caps iron ore deposits in Brazil. After successfully determining the presence and distribution of biosignatures with canga, the thesis shifted focus, aiming to use our understanding of the naturally occurring processes that stabilised canga horizons to accelerate the re-cementation of crushed mine waste. These results have far-reaching implications for the remediation of iron ore mines and stabilisation of mine tailings (waste).

With the number tailings dams collapses doubling in the past 20 years (Armstrong et al., 2019), slope stabilisation has become one of the greatest challenges facing the mining industry. To prevent future disasters, collaborative efforts between researchers, mining executives and political advisors will be required to develop new strategies for mine site remediation and long- term sustainability.

Chapter One

1.2. Thesis structure and research objectives This document consists of nine chapters, where Chapter One (Introduction), Chapter Two (Literature Review) and Chapter Nine (Discussions and Conclusions) summarise the primary findings and interpretations of the thesis, providing a holistic overview of natural canga evolution and implications for iron ore mine bioremediation strategies. Chapters Three to Eight are primary, stand-alone research articles, either published, in review or to be submitted to the journals identified on the first page of each chapter.

The primary research chapters each include a relevant introduction and conclusions synopsis; therefore, Chapter One and conclusions within Chapter Nine have been kept brief to allow for the document to be read in its entirety without unnecessary repetition. Chapter Two (Literature Review) offers a broader overview and contextualisation of the thesis and, together with Chapter Nine, provides a working strategy for iron ore mine remediation in Brazil.

The aims of this contribution were to understand the microbiological influences on the redox- based geochemical cycling of iron oxide minerals within canga environments, which is fundamental to the formation of new iron cements (Monteiro et al., 2014). I therefore sought to determine the occurrence and distribution of biosignatures within canga. With the identification of prevalent biosignatures throughout canga, we conducted thorough investigations to:

1. characterise the geochemical environments that contributed to microbial fossilisation

and long-term preservation of organic and inorganic biosignatures in iron-rich

environments (Chapters Three and Four);

2. understand the role microorganisms play in metal mobility, mineral dissolution and re-

precipitation within iron-rich duricrusts (Chapter Five) and apply this to our

understanding of the evolution of ferruginous duricrusts (Chapter Six).

To understand the modern day geochemical cycling of iron, I studied the iron-rich seeps that flow into hydraulically restricted dissolution lake features (Chapter Seven) that form with canga in the Carajás Mineral Province, Pará, Brazil. The field environments offered inspiration for the relatively rapid formation of incredibly durable iron cements and contributed to the design of an experiment that aimed to model these conditions and accelerate slope stabilisation and the reformation of an iron-rich crust (Chapter Eight).

31 Thesis structure and introduction

References

Armstrong M, Petter R, Petter C. Why have so many tailings dams failed in recent years? Resources Policy 2019; 63.

Dorr JVN. Supergene iron ores of Minas Gerais, Brazil. Economic Geology 1964; 59: 1203- 1240.

Hinman P. Brazil: New vale mine disaster is one more corporate failure. Green Left Weekly 2019: 15.

Monteiro HS, Vasconcelos PM, Farley KA, Spier CA, Mello CL. (U–Th)/He geochronology of goethite and the origin and evolution of cangas. Geochimica et Cosmochimica Acta 2014; 131: 267-289.

Segura FR, Nunes EA, Paniz FP, Paulelli ACC, Rodrigues GB, Braga GÚL, et al. Potential risks of the residue from Samarco's mine dam burst (Bento Rodrigues, Brazil). Environmental Pollution 2016; 218: 813-825.

Toy TJ, Griffith JJ. Changing surface-mine reclamation practices in Minas Gerais, Brazil. International Journal of Surface Mining, Reclamation and Environment 2001; 15: 33- 51.

United States Geological Survey. Iron Ore Statistics and Information. 2018. USGS Reston, VA, 2018, pp. 88-89.

Chapter 2 Strategising the bioremediation of iron ore mines

Alan Levetta*, Emma Gagena, Anat Paza, Paulo Vasconcelosa, Gordon Southama

a School of Earth and Environmental Sciences, University of Queensland, Brisbane, Queensland 4072, Australia

To be combined with Discussion Chapter and submitted to Earth Science Reviews

*author to whom correspondence should be directed for all Chapters. [email protected]

Chapter Two

Abstract Iron ore mine remediation presents a unique challenge, distinct from remediation efforts of other lateritic deposits. The primary differences are related to soil development, or lack thereof. Unique rupestrian ecosystems, including a suite of naturally rare and several endemic plant species, are associated with the ironstone (canga) outcrops that commonly cap iron ore mines. To reinstate native ecosystems post-mining, developing a substrate similar to the iron-rich cap rock is required. These iron-rich duricrusts have evolved by the ongoing dissolution and re- precipitation of iron oxide minerals, continuously forming new iron cements. Fluctuating redox conditions between the microscale anaerobic niches and the broadly oxidising environment are critical to the iron cycling that contributes to the evolution and stabilisation of these surface duricrusts. Understanding the mechanisms that contributes to natural iron cement formation allows for the development of novel biotechnologies that aim to accelerate these processes. These biotechnologies would offer an environmentally-friendly, circumneutral (pH 5 – 7) strategy for iron ore mine remediation. Successful remediation programmes would aim to accelerate the formation of iron cements necessary to stabilise crushed materials on hillslopes, restore a functional hydrology and regenerate a substrate similar to an iron-rich duricrust for revegetation using native species. Here, we review recent advances in understanding the biological process that contribute to canga evolution. Discussion surrounding their implications for strategies that aim to remediate iron ore mines are expanded in Chapter Nine.

Bioremediation of iron ore mines

2.1. Introduction Remediating and decommissioning the giant iron ore mines in Brazil remains a significant challenge. Some of the world’s largest iron ore mines are located in the Carajás Mineral Province (Fig. 2.1). Unlike the remediation of other lateritic mines, including bauxite deposits that focus on topsoil return and seeding (Grant and Koch, 2007), iron ore remediation prioritises the regeneration of the iron-rich duricrust that typically cap iron deposits (Dorr, 1964; Levett et al., 2016).

The massive iron ore provinces in Brazil are, in part, enriched in iron by the extensive weathering of laterally continuous Archean banded iron formations (BIFs) (Cloud, 1973). During weathering, the silica-rich bands are leached away, enriching the remaining rock in iron oxide, predominantly hematite or martite after magnetite (Rosière and Rios, 2004). The ultimate product of weathering in these systems is the formation of a relatively hard, porous, iron-rich duricrust (canga; Fig. 2.2A) that prevents the erosion of the underlying friable iron ore within the saprolite (Dorr, 1964). Canga is extremely resistant to physical erosion, with cosmogenic isotope studies demonstrating these duricrusts to be amongst the longest-lived continuously exposed land surfaces on Earth (Monteiro et al., 2018a; Monteiro et al., 2018b; Shuster et al., 2012).

Despite being of iron ore grade (commonly > 55 wt.% Fe), the high phosphorus (0.1 – 0.5 wt.%) and alumina (ca. 3 – 10 wt.%) concentrations of the canga in Brazil make it an uneconomic by-product of iron mining (Dorr, 1964). To access the underlying high-grade iron ore, canga is stripped and stockpiled, destroying the macrobiota that has associated with these outcrops (Toy and Griffith, 2001). Successful remediation of iron ore mines would restore hydraulic function and provide a substrate for revegetation of native ecosystems. This could be achieved by regenerating a substrate similar to the canga outcrops that currently cap highly weathered BIFs in Brazil. Here, we review the biological mechanisms that naturally contribute to canga evolution and propose a model for the environmentally-friendly reclamation of iron ore mines that leverages these natural processes.

35 Chapter Two

Fig. 2.1. Digital elevation model of the Carajás region generated using the Shuttle Radar Topography Mission (SRTM) from the USGS Earth Explorer (https://earthexplorer.usgs.gov/). N1 highlights some canga regions in Serra Norte, S11 highlights canga regions in Serra Sul, Carajás Mineral Province, Pará.

Bioremediation of iron ore mines

2.2. What is canga? Canga is primarily composed of detrital fragments of hematite and BIF, cemented together by authigenic goethite (Figs. 2.2B – 2.2C). Dorr (1964) recognised that a renewable iron source must be supplied to canga to prevent its erosion and proposed that bacteria may play a role in canga evolution. Indeed, many of the iron oxides within canga exhibit recycled textures, highlighting the geochemical cycling of iron that occurs within these horizons (Fig. 2.2D). More recently, Monteiro et al. (2014) presented robust geochronological and geochemical evidence, postulating that the flora and microbiota associated with canga continuously drive iron oxide dissolution. The redox cycling capacity of iron is critical to the longevity of canga, continuously forming new iron cements that combat physical erosion. The iron oxide minerals in canga have undergone several generations of recycling to form these complex cements (Fig. 2.2B – 2.2E) (Sahoo et al., 2017). For example, Monteiro et al. (2014) demonstrated that many of the goethite minerals in canga suitable for dating had a minimum age of ca. 2 Ma, for a period of ca. 50 Ma, indicating these goethite minerals may have been recycled 25 times. Understanding the natural biological processes that contribute to the geochemical cycling of iron within canga will be fundamental to developing strategies for iron mine remediation. Harnessing and promoting these natural processes may be used to accelerate the re-cementation of canga, creating an iron-rich crust akin to canga substrates for revegetation programmes.

2.3. Drivers for the remediation of iron ore mines The scale of iron ore mining operations in Brazil poses a threat to the unique ecosystems that have adapted to survive in association with these ironstone outcrops (Jacobi and Carmo, 2011; Jacobi et al., 2007). Ironstone outcrops represent a relatively harsh environment: little to no soil development limits available nutrients (Messias et al., 2013); rock surfaces can reach 70 ºC and the UV exposure can be extreme. In addition, the rocks are enriched in potentially toxic metals when mobile; for example, aluminium, iron and manganese. Therefore, plants growing on canga are commonly metallotolerant or rarely hyperaccumulator species (Schettini et al., 2018). Canga-associated flora in Brazil are diverse and naturally rare, including many endemic species (Carmo and Jacobi, 2016; Gastauer et al., 2012; Nunes et al., 2015; Viana et al., 2016). In the Carajás Mineral Province, canga-associated vegetation are declining as a result of mining related activities: canga-associated vegetation covered an area of area of 77.2 km2 in 2016 compared with 105.2 km2 in 1973 (Souza-Filho et al., 2019).

The high diversity of canga ecosystems is reflected across all scales. At the quadrant scale, diverse niches creates complex vegetative structures and physiognomies, resulting in high

37 Chapter Two alpha (within-site) diversities (Jacobi et al., 2007). At the profile-scale, edaphic factors are the primary drivers responsible for vegetation zonation in the canga-associated ecosystems in the Carajá Mineral Province (Nunes et al., 2015). Finally, at the state- and continental-scales, canga ecosystems appeared to have evolved in relative isolation as inselbergs, reflected by high beta (ration of regional to local) diversities (Jacobi et al., 2007; Mota et al., 2018). For example, different canga outcrops in the Quadrilátero Ferrífero (QF), separated by only 32 km, share only 27% plant species in common (Jacobi et al., 2007). Similarly, canga outcrops in the Carajás support at least 856 species of seed plants while 946 species have been identified in the QF, with only 96 species (less than 12%) common to both states (Mota et al., 2018). While these complexities are beyond the scope of this review, they all require consideration for mine remediation programmes that aim to reinstate native vegetation.

2.4. Biologically-mediated iron oxide dissolution mechanisms in canga The exact mechanisms that promote iron reduction in canga remain enigmatic but are likely to be accelerated by biological processes controlled by plants, fungi and microorganisms that grow in direct association with the canga (Gagen et al., 2019). Though a number of processes are likely to contribute to iron oxide dissolution, we will only cover the main mechanisms thought to contribute to iron oxide dissolution in canga (Gagen et al., 2019). Despite canga having relatively high phosphorus concentrations (typically 0.1 – 0.5 wt.%) (Dorr, 1964), the goethite-rich nature of the rock means most of the phosphorus is mineral bound, primarily associated with aluminium-substituted goethite (Ruan and Gilkes, 1996; Schulze and Schwertmann, 1984). To release the phosphorus and make it available for assimilation, plants exude a variety of organic acids which promote the reductive dissolution of nearby iron oxide minerals (Reaction 2.1) (Neumann and Römheld, 1999). Microbial biofilms and lichen that cover the surface of most of canga are also likely to exude organic acids, contributing to iron oxide mineral dissolution (Bertrand et al., 1999; Gadd, 2007). Secondly, plants and fungi are also known to acidify their local environment within the rhizosphere, promoting the dissolution of less-well crystalline iron oxides (Reaction 2.2) and the release of mineral-bound phosphorus (Hinsinger, 2001).

2+ − − CH3COOH (푙) + 8FeOOH (Goethite)+ 2H2O (푙) → 8Fe (푎푞) + 2HCO3 (푎푞) + 14OH (푎푞) (2.1)

+ 3+ FeOOH (Goethite) + 3H (푎푞) → Fe (푎푞)+ 2H2O (푙) (2.2)

Bioremediation of iron ore mines

Fig. 2.2. Weathering profile from S11D region. Canga forms a highly resistant cap above the friable iron ore within the saprolite (A). Canga is highly heterogeneous, primarily composed of detrital fragments that have undergone several generations of recycling, cemented together by authigenic iron oxide precipitation (B and C). Many of the iron concretion textures have organic- like textures (D). Dissolution-depressions within of ferruginous duricrusts form ephemeral pools that are often covered in iron sheens indicating the interaction of iron with microbial biofilms (E).

39 Chapter Two

Finally, iron oxide minerals may serve as the ultimate electron acceptors for microbial respiration in anoxic niches throughout canga, potentially driven by fermentation of organic compounds (Reaction 2.3) (Gagen et al., 2019; Parker et al., 2018). Canga environments are likely to have been broadly oxic throughout their formation and evolution (Monteiro, 2017). Therefore, iron reduction is likely to be restricted to specific niches and serve as the rate- limiting step for the formation of iron oxide cements.

2+ − 퐶6퐻12푂6 (푙) + 8FeOOH (Goethite)+ 2H2O (푙) → 2CH3COOH (푙) + 8Fe (푎푞) + 2HCO3 (푎푞) +

− 14OH (푎푞) (2.3)

2.5. Iron oxidation, biomineralization and fossilisation processes in canga Given the instability of ferrous iron within the broadly circumneutral, oxic canga environments, iron oxidation and hydrolysis to form mineral precipitates (Reaction 2.4) is likely to be relatively rapid.

2+ + 4퐹푒 (푎푞) + 푂2 + 6퐻2푂(푙) → 4퐹푒푂푂퐻(퐺표푒푡ℎ𝑖푡푒) + 8퐻 (푎푞) (2.4)

These processes can occur abiotically or be influenced by microorganisms. Neutrophilic iron- oxidising microorganisms, including Gallionella and Sideroxydans lineages, have been identified in iron seeps and streams in weathered banded iron formation environments (Gagen et al., 2018) and canga (Chapter Seven). Mineralised iron stalks and sheath structures are rarely found in the fossil record of canga; therefore, classic, neutrophilic iron-oxidising microorganisms appear to play a limited role in cement formation (Levett et al., 2019). Abiotic iron oxide precipitation in canga is likely to initially coat the existing grains (Fig. 2.3). Several generations of iron oxide mineral cycling (Monteiro et al., 2014) may be required to form sufficient cements (Fig. 2.3)

Throughout this contribution we adopt the terminology used by Dupraz et al. (2009) and others to adequately differentiate between various modes of biomineralisation. These include, biologically controlled mineralisation (BCM; for example, intracellular magnetite mineralisation controlled by magnetotactic bacteria (Kopp and Kirschvink, 2008)), biologically induced mineralisation (BIM; for example, extracellular hydrous ferric oxide mineralisation by neutrophilic iron-oxidising bacteria (Emerson and Ghiorse, 1993) and biologically influenced mineralisation (BIFM; passive mineralisation of extracellular organic polymers driven by electrostatic interaction (Ferris et al., 1988). Here, and throughout this thesis, we focus exclusively on extracellular mineralisation (BIM/BIFM) and for the most part on BIFM. Intracellular mineralisation is only discussed after cell death as part of

Bioremediation of iron ore mines permineralisation. After fossilisation, minerals may continue to precipitate and infill intracellular voids (Levett et al., 2016). The direction of mineral growth appears to be a critical marker when cell envelope structures are degraded: minerals form at the membrane and grow inwards, rather than nucleating at central-source and radiating outwards (Benzerara et al., 2004; Cosmidis et al., 2013).

Fig. 2.3. Model to highlight the main mechanisms hypothesised to play a role in iron reduction and iron oxidation in canga. In bold are the mechanisms that could be used to promote the biogeochemical cycling of iron in circumneutral environments to accelerate the reformation of iron-rich concretions for iron ore remediation. Backscattered electron micrographs highlight iron mineralised biofilms that form meniscus-type concentrations, generated in an experiment in Chapter Eight.

41 Chapter Two

There is also an important distinction between biomineralisation and microbial fossilisation. Biomineralisation represents the active or passive mineralisation of cell envelope and extracellular organic polymers (for example, peptidoglycan of gram-positive bacteria, the outer membrane of gram-negative bacteria, extracellular polymeric substances (EPS) and sheath structures). Biomineralisation represents the first stages of microbial fossilisation; however, cells are still able to maintain metabolic function during biomineralisation (Benzerara et al., 2011; Kappler et al., 2005; Phoenix and Konhauser, 2008). In contrast, microbial fossilisation is the contribution of microbial-influenced structures and minerals to the lithology. Samples prepared as a rock by dehydration in an oven may contain evidence of microbial fossilisation, whereas samples dehydrated using solvent replacement may contain evidence of biomineralisation.

Biosignatures may be directly or indirectly associated with preserved microbial fossils. Microbial fossils (direct biosignatures) can be extraordinarily resilient, surviving within the geological record for billions of years (Schopf, 1993). However, to understand the role microorganisms play within the environments they survive in, it is also important to develop our knowledge of indirect biosignatures. These may include but are not limited to: isotopic, elemental or mineralogical (Banfield et al., 2001) distributions induced or altered by microbial life during microbial-mineral interactions (for example, biocorrosion); influence on grain distributions and stabilisation (Westall and Cavalazzi, 2011) in sedimentary environments (for example, stromatolites) and organic or inorganic by-products of bacterial metabolism.

2.6. Conclusions Remediation of iron ore mines will benefit from the development of novel biotechnologies that accelerate the redox-based cycling of iron oxide minerals. Microbial dissimilatory or fermentation-driven iron reduction offers a method to reduce iron oxide minerals at circumneutral pH and place iron into solution within an optimised bioreactor. Accelerating the growth of microbial biofilms promotes the aggregation of fine grains and provide a framework for passive iron oxide biomineralisation. Fossilising microbial biofilms may reduce the need for multiple generations of iron oxide recycling to form physically stable cements. Scaling these biotechnologies appropriately may provide a novel solution for iron ore remediation programs, which would be transferable to other commodities, greatly advancing long-term sustainability in the mining sector.

Bioremediation of iron ore mines

Acknowledgements We acknowledge support the Vale S.A.-UQ Geomicrobiology initiative and the Australian Research Council Linkage Program (LP140100805) to G. Southam and P. Vasconcelos. The authors acknowledge the facilities and the scientific and technical assistance of the Australian Microscopy and Microanalysis Research Facility at the Centre of Microscopy and Microanalysis, at the University of Queensland. Alan Levett acknowledges the support from the Australian Government Research Training Program.

References Banfield JF, Moreau JW, Chan CS, Welch SA, Little B. Mineralogical biosignatures and the search for life on Mars. Astrobiology 2001; 1: 447-465.

Benzerara K, Menguy N, Guyot F, Skouri F, de Luca G, Barakat M, et al. Biologically controlled precipitation of calcium phosphate by Ramlibacter tataouinensis. Earth and Planetary Science Letters 2004; 228: 439-449.

Benzerara K, Miot J, Morin G, Ona-Nguema G, Skouri-Panet F, Ferard C. Significance, mechanisms and environmental implications of microbial biomineralization. Comptes Rendus Geoscience 2011; 343: 160-167.

Bertrand I, Hinsinger P, Jaillard B, Arvieu J. Dynamics of phosphorus in the rhizosphere of maize and rape grown on synthetic, phosphated calcite and goethite. Plant and Soil 1999; 211: 111-119.

Carmo FFd, Jacobi CM. Diversity and plant trait-soil relationships among rock outcrops in the Brazilian Atlantic rainforest. Plant and Soil 2016; 403: 7-20.

Cloud P. Paleoecological significance of the banded iron-formation. Economic Geology 1973; 68: 1135-1143.

Cosmidis J, Benzerara K, Gheerbrant E, Estève I, Bouya B, Amaghzaz M. Nanometer-scale characterization of exceptionally preserved bacterial fossils in Paleocene phosphorites from Ouled Abdoun (Morocco). Geobiology 2013; 11: 139-153.

Dorr JVN. Supergene iron ores of Minas Gerais, Brazil. Economic Geology 1964; 59: 1203- 1240.

Dupraz C, Reid RP, Braissant O, Decho AW, Norman RS, Visscher PT. Processes of carbonate precipitation in modern microbial mats. Earth-Science Reviews 2009; 96: 141-162.

Emerson D, Ghiorse WC. Ultrastructure and chemical composition of the sheath of Leptothrix discophora SP-6. Journal of Bacteriology 1993; 175: 7808-7818.

Ferris FG, Fyfe WS, Beveridge TJ. Metallic ion binding by Bacillus subtilis: implications for the fossilization of microorganisms. Geology 1988; 16: 149-152.

43 Chapter Two

Gadd GM. Geomycology: biogeochemical transformations of rocks, minerals, metals and radionuclides by fungi, bioweathering and bioremediation. Mycological Research 2007; 111: 3-49.

Gagen EJ, Levett A, Paz A, Gastauer M, Caldeira CF, Valadares RBdS, et al. Biogeochemical processes in canga ecosystems: armoring of iron ore against erosion and importance in iron duricrust restoration in Brazil. Ore Geology Reviews 2019; 107: 573-586.

Gagen EJ, Levett A, Shuster J, Fortin D, Vasconcelos PM, Southam G. Microbial diversity in actively forming iron oxides from weathered banded iron formation systems. Microbes and Environments 2018; 33: 385-393.

Gastauer M, Teixeira Braga Messias MC, Alves Meira Neto JA. Floristic composition, species richness and diversity of campo rupestre vegetation from the Itacolomi State Park, Minas Gerais, Brazil. Environment and Natural Resources Research 2012; 2.

Grant C, Koch J. Decommissioning Western Australia's first bauxite mine: co-evolving vegetation restoration techniques and targets. Ecological Management & Restoration 2007; 8: 92-105.

Hinsinger P. Bioavailability of soil inorganic P in the rhizosphere as affected by root-induced chemical changes: a review. Plant and Soil 2001; 237: 173-195.

Jacobi CM, Carmo FF. Life-forms, pollination and seed dispersal syndromes in plant communities on ironstone outcrops, SE Brazil. Acta Botanica Brasilica 2011; 25: 395- 412.

Jacobi CM, Carmo FF, Vincent RC, Stehmann JR. Plant communities on ironstone outcrops: a diverse and endangered Brazilian ecosystem. Biodiversity and Conservation 2007; 16: 2185-2200.

Kappler A, Schink B, Newman DK. Fe(III) mineral formation and cell encrustation by the nitrate-dependent Fe(II)-oxidizer strain BoFeN1. Geobiology 2005; 3: 235-245.

Kopp RE, Kirschvink JL. The identification and biogeochemical interpretation of fossil magnetotactic bacteria. Earth-Science Reviews 2008; 86: 42-61.

Levett A, Gagen E, Shuster J, Rintoul L, Tobin M, Vongsvivut J, et al. Evidence of biogeochemical processes in iron duricrust formation. Journal of South American Earth Sciences 2016; 71: 131-142.

Levett A, Gagen EJ, Diao H, Guagliardo P, Rintoul L, Paz A, et al. The role of aluminium in the preservation of microbial biosignatures. Geoscience Frontiers 2019; 10: 1125-1138.

Messias M, Leite M, Meira Neto J, Kozovits A, Tavares R. Soil-vegetation relationship in quartzitic and ferruginous Brazilian rocky outcrops. Folia Geobotanica 2013; 48: 509- 521.

Monteiro H, Vasconcelos P, Farley K. A combined (U‐Th)/He and cosmogenic 3He record of landscape armoring by biogeochemical iron cycling. Journal of Geophysical Research: Earth Surface 2018a; 123: 298-323.

Bioremediation of iron ore mines

Monteiro HDS. Paleoenvironmental evolution of continental landscapes through combined high-resolution geochronology and δ18O ion microprobe analysis of goethite. School of Earth Sciences. PhD thesis. The University of Queensland, 2017.

Monteiro HS, Vasconcelos PM, Farley KA, Spier CA, Mello CL. (U–Th)/He geochronology of goethite and the origin and evolution of cangas. Geochimica et Cosmochimica Acta 2014; 131: 267-289.

Monteiro HS, Vasconcelos PMP, Farley KA, Lopes CAM. Age and evolution of diachronous erosion surfaces in the Amazon: combining (U-Th)/He and cosmogenic 3He records. Geochimica et Cosmochimica Acta 2018b; 229: 162-183.

Mota NFdO, Watanabe MTC, Zappi DC, Hiura AL, Pallos J, Viveros RS, et al. Amazon canga: the unique vegetation of Carajás revealed by the list of seed plants. Rodriguésia 2018; 69: 1435-1488.

Neumann G, Römheld V. Root excretion of carboxylic acids and protons in phosphorus- deficient plants. Plant and Soil 1999; 211: 121-130.

Nunes JA, Schaefer CE, Ferreira Júnior WG, Neri AV, Correa GR, Enright NJ. Soil-vegetation relationships on a banded ironstone 'island', Carajás Plateau, Brazilian Eastern Amazonia. Anais da Academia Brasileira de Ciências 2015; 87: 2097-2110.

Parker CW, Auler AS, Barton MD, Sasowsky ID, Senko JM, Barton HA. Fe (III) reducing microorganisms from iron ore caves demonstrate fermentative Fe (III) reduction and promote cave formation. Geomicrobiology Journal 2018; 35: 311-322.

Phoenix V, Konhauser K. Benefits of bacterial biomineralization. Geobiology 2008; 6: 303- 308.

Rosière CA, Rios FJ. The origin of hematite in high-grade iron ores based on infrared microscopy and fluid inclusion studies: the example of the Conceição mine, Quadrilátero Ferrífero, Brazil. Economic Geology 2004; 99: 611-624.

Ruan HD, Gilkes RJ. Kinetics of phosphate sorption and desorption by synthetic aluminous goethite before and after thermal transformation to hematite. Clay Minerals. 31, 1996, pp. 63.

Sahoo PK, Guimarães JTF, Souza-Filho PWM, da Silva MS, Júnior WN, Powell MA, et al. Geochemical characterization of the largest upland lake of the Brazilian Amazonia: impact of provenance and processes. Journal of South American Earth Sciences 2017; 80: 541-558.

Schettini AT, Leite MG, Messias MCT, Gauthier A, Li H, Kozovits AR. Exploring Al, Mn and Fe phytoextraction in 27 ferruginous rocky outcrops plant species. Flora 2018; 238: 175-182.

Schopf JW. Microfossils of the early Archean apex chert: new evidence of the antiquity of life. Science 1993; 260: 640-646.

Schulze DG, Schwertmann U. The influence of aluminium on iron oxides; X, Properties of Al- substituted goethites. Clay Minerals 1984; 19: 521-539.

45 Chapter Two

Shuster DL, Farley KA, Vasconcelos PM, Balco G, Monteiro HS, Waltenberg K, et al. Cosmogenic 3He in hematite and goethite from Brazilian “canga” duricrust demonstrates the extreme stability of these surfaces. Earth and Planetary Science Letters 2012; 329: 41-50.

Souza-Filho PWM, Giannini TC, Jaffé R, Giulietti AM, Santos DC, Nascimento WR, Jr., et al. Mapping and quantification of ferruginous outcrop savannas in the Brazilian Amazon: a challenge for biodiversity conservation. PLOS ONE 2019; 14: e0211095.

Toy TJ, Griffith JJ. Changing surface-mine reclamation practices in Minas Gerais, Brazil. International Journal of Surface Mining, Reclamation and Environment 2001; 15: 33- 51.

Viana PL, Mota NFdO, Gil AdSB, Salino A, Zappi DC, Harley RM, et al. Flora of the cangas of the Serra dos Carajás, Pará, Brazil: history, study area and methodology. Rodriguésia 2016; 67: 1107-1124.

Chapter 3 The role of aluminium in the preservation of microbial biosignatures

Alan Levetta, Emma J. Gagena, Hui Diaob, Paul Guagliardoc, Llew Rintould, Anat Paza, Paulo M. Vasconcelosa, Gordon Southama

a School of Earth and Environmental Sciences, University of Queensland, Brisbane, Queensland 4072, Australia b Centre for Microscopy and Microanalysis, University of Queensland, Brisbane, Queensland 4072, Australia

c Centre for Microscopy, Characterisation and Analysis, University of Western Australia, Perth 6009, Western Australia, Australia

d School of Chemistry, Physics & Mechanical Engineering, Queensland University of Technology, Brisbane, Queensland 4001, Australia

Citation for this publication:

Levett, A., Gagen, E.J., Diao, H., Guagliardo, P., Rintoul, L., Paz, A., Vasconcelos, P.M. and Southam, G. (2019). The role of aluminium in the preservation of microbial biosignatures. Geoscience Frontiers 10, 1125-1138.

Chapter Three

Abstract Demonstrating the biogenicity of presumptive microfossils in the geological record often requires supporting chemical signatures, including isotopic signatures. Understanding the mechanisms that promote the preservation of microbial biosignatures associated with microfossils is fundamental to unravelling the palaeomicrobiological history of the material. Organomineralisation of microorganisms is likely to represent the first stages of microbial fossilisation and has been hypothesised to prevent the autolytic degradation of microbial cell envelope structures. In the present study, two distinct fossilisation textures (permineralised microfossils and iron oxide encrusted cell envelopes) identified throughout iron-rich rock samples were analysed using nanoscale secondary ion mass spectrometry (NanoSIMS). In this system, aluminium is enriched around the permineralised microfossils, while iron is enriched within the intracellularly, within distinct cell envelopes. Remarkably, while cell wall structures are indicated, carbon and nitrogen biosignatures are not preserved with permineralised microfossils. Therefore, the enrichment of aluminium, delineating these microfossils appears to have been critical to their structural preservation in this iron-rich environment. In contrast, NanoSIMS analysis of mineral encrusted cell envelopes reveals that preserved carbon and nitrogen biosignatures are associated with the cell envelope structures of these microfossils. Interestingly, iron is depleted in regions where carbon and nitrogen are preserved. In contrast aluminium appears to be slightly enriched in regions associated with remnant cell envelope structures. The correlation of aluminium with carbon and nitrogen biosignatures suggests the complexation of aluminium with preserved cell envelope structures before or immediately after cell death may have inactivated autolytic activity preventing the rapid breakdown of these organic, macromolecular structures. Combined, these results highlight that aluminium may play an important role in the preservation of microorganisms within the rock record.

The role of aluminium in biosignature preservation

3.1. Introduction Microbial fossils provide insights into the environmental conditions in which they existed, revealing information about how microorganisms interacted with their environments. Microorganisms may be fossilised in diverse environmental conditions and preserved by interactions with various elements and minerals. Silicification (Konhauser et al., 2004), calcification (Riding, 2000) and ferrugination in acidic (Preston et al., 2011) and neutral environments (Salama et al., 2013) may be responsible for the preservation of bacteriomorphic structures. Within iron-rich environments, electrostatic interactions between iron cations and net negative cell envelopes have been proposed to drive the biomineralisation of microorganisms, which has been postulated to represent the first stage of microbial fossilisation (Ferris et al., 1988; Li et al., 2013; Li et al., 2014). The present study aimed to determine the chemical biosignatures associated with fossilised microorganisms in near-neutral iron-rich environments.

Aluminium is the third most abundant element within the Earth’s crust, with only oxygen and silicon more abundant. Despite the plenitude of aluminium, it serves no known biological function and is generally toxic in labile forms to most microorganisms (Exley and Birchall, 1992). Aluminium mobility is primarily controlled by pH, although the solubility of minerals containing aluminium and organic acids also influence the release and subsequent stability of aluminium ions in solution (Bache, 1986). Below pH 5, the prominent ionic aluminium species 3+ 3+ is Al(H2O)6 , typically referred to as Al or free aluminium, which is considered to be the most inimical to biota (Macdonald and Martin, 1988). Insoluble Al(OH)3 precipitates reach a maximum at approximately 6.2, which coincides with the minimum solubility of free aluminium (Martin, 1986). At circumneutral pH, aluminium is relatively immobile and generally considered nontoxic (Bache, 1986). In alkaline solutions (pH ≥ 7.4), Al(OH)3 - precipitates begin to redissolve, forming Al(OH)4 complexes (Macdonald and Martin, 1988).

Aluminium binding to cell envelope structures of Escherichia coli (Guida et al., 1991) has been demonstrated. Gibbsitised fungi that have been interpreted to play a role in active Al precipitation in bauxite pisoliths in Western Australia have also previously been reported (Anand and Verrall, 2011); however, the role of aluminium in the preservation of organic biosignatures has not previously been reported.

Determining the biogenicity of bacteriomorphic structures within the geologic record based can be misleading (Marshall et al., 2011). Chemical or isotopic signatures to support the

49 Chapter Three preservation of microbial fossils are often required to ascertain the biological origins of microfossils (Brasier et al., 2015). Here, two distinct microbial fossilisation textures identified in iron-rich environments using scanning electron microscopy were examined using nanoscale secondary ion mass spectrometry (NanoSIMS), revealing a role for aluminium in the preservation of structural ± organic biosignatures.

3.2. Materials and methods 3.2.1. Site description and sample collection 3.2.1.1. Carajás sample - ferruginous duricrust A ferruginous duricrust hand-sample was collected from eastern aspect of the Carajás mineral province in the State of Pará, Brazil (Vale S.A. N1 site). The hand-sample collected was of a well consolidated ferruginous duricrust that caps highly weathered banded iron formation (BIF). The duricrust sample was collected in proximity to a large lake and was likely to have been previously submerged during the summer months (November – March) when the Carajás mineral province receives the bulk of its annual rainfall (approximately 1800 mm). The ferruginous duricrust sample from the lake-edge was extremely competent, hard and appeared to be less porous than ferruginous duricrust that formed away from the lake-edge. Ferruginous duricrusts in the Carajás typically cap the weathering profile of BIFs and are associated with little to no soil profile. The sample was collected directly from the surface using a rock hammer in the complete absence of any soil. Consistent with the general description of ferruginous duricrusts that cap BIFs by Dorr (1964), the sample contains detrital fragments of high-grade specular hematite cemented by secondary bands of vitreous and dull goethite. Secondary goethite bands that indicate the flow of iron-rich solutions made this sample of interest to investigate the presence of microbial fossils and their role in the evolution of ferruginous duricrusts (Levett et al., 2016).

3.2.1.2. Quadrilátero Ferrífero sample – goethite-cemented vein A hand sample from an exposed goethite-cemented vein was collected from the saprolite (depth of approximately 15 m) of a highly weathered banded iron formation profile from Serra do Gandarela, Quadrilátero Ferrífero in the State of Minas Gerais, Brazil (Fig. 3.1). The Quadrilátero Ferrífero (QF) has a tropical sub-humid climate with an approximate annual rainfall of 1900 mm during the summer months (November – March). Temperatures in the QF are typically between 13 and 30 °C year round with temperatures of up to 70 °C recorded over bare rocks (Jacobi and Carmo, 2011). The geology and geochemistry of the QF have been previously presented (Spier et al., 2003). The goethite-cemented vein was closely associated

The role of aluminium in biosignature preservation with the roots of small shrubs, which had become coated in iron oxide precipitates (Fig. 3.1A). The sample contains well consolidated goethite and hematite with dull secondary goethite precipitates within pore spaces and covering grain surfaces (Fig. 3.1B).

Fig. 3.1. Photograph of goethite cemented vein in the exposed saprolite from Serra do Gandarela, Quadrilátero Ferrífero (QF), State of Minas Gerais, Brazil. The white arrow indicates the region from which the hand sample was collected, approximately 15 m below the surface. (B) Photograph of goethite-cemented hand sample highlighting the presence of iron oxide coated roots.

3.2.2. Sample characterisation Bulk sample pH, elemental composition and mineralogy of samples (see Fig. 3.1) from the Carajás mineral province (permineralised microfossils) and the Quadrilátero Ferrífero (encrusted cell envelopes) were determined to characterise the samples. For all analyses, rock samples were crushed to less than 62 µm using a ring and puck mill. The bulk sample pH was determined by mixing 1g of sample with 5 mL of ultrapure water, agitating for 2 hours and measuring the pH of the solution.

Bulk sample chemical compositions were determined using X-ray fluorescence (XRF) spectroscopy at the Australian Laboratory Services (Analytical Geochemistry). Samples were

51 Chapter Three fused with a lithium tetraborate:lithium metaborate flux (including lithium nitrate as an oxidising agent) with a ratio of 12:22. Fused samples were poured into a platinum mould and analysed using XRF spectroscopy. Loss on ignition was calculated at 1000 °C.

Fourier transform infrared spectroscopy (FTIR) was used to identify the mineralogy of the grains associated with the permineralised microfossils and the encrusted cell envelopes. Briefly, a Nicolet iS50 FTIR spectrometer coupled with a continuum infrared microscope equipped with a Mercury-Cadmium-Telluride detector was operated in attenuated total reflection (micro-ATR) mode using a germanium crystal. Samples were analysed for 64 scans using an effective 40 × 40 µm spot size. Spectra were collected in the mid-infrared spectrum (4000 – 650 cm-1) with a spectral resolution of 4 cm-1.

3.2.3. Electron Microscopy Polished petrographic thin sections were made from rock samples from the Quadrilátero Ferrífero and the Carajás mineral province to characterise the microstructure of each rock sample. The microstructure of both samples was analysed to determine the role of microorganisms in the biogeochemical cycling of iron within highly weathered BIFs (Levett et al., 2016). A BAL-TEC MSC-010 sputter coater was used to coat petrographic thin sections with 10 nm iridium. A JEOL JSM-7100F Field Emission scanning electron microscope (FE- SEM) equipped with an energy dispersive X-ray spectrometer (EDS) was used to examine polished petrographic thin sections. High resolution SEM micrographs were acquired using accelerating voltages of 3 – 15 kV, with low voltages used for discerning surface features in secondary electron mode. Sample surfaces were cleaned using a XEI Scientific Evactron 25 Decontaminator RF Plasma Cleaning System and samples were degassed at 50 °C for a minimum of 12 hours prior to examination.

3.2.4. Nanoscale secondary ion mass spectrometry 3.2.4.1. Sample preparation Permineralised microfossils were polished following the methods described by Levett et al. (2016). Briefly, ferruginous duricrust samples were hand polished using new, fixed SiC adhesives before a submicrometre diamond polish. Samples were adhered to a 1-inch diameter probe mount using Araldite (HY951). Samples were not embedded in an epoxy resin. All samples were coated with 5 – 10 nm gold prior to NanoSIMS analysis. Regions possessing microfossils (Fig. 3.2) were identified using BSE-SEM and targeted at high magnification (Fig. 3.3) for NanoSIMS (Fig. 3.4).

The role of aluminium in biosignature preservation

Encrusted microbial cell envelopes (Fig. 3.5) identified using BSE-SEM were prepared for nanoscale secondary ion mass spectrometry (NanoSIMS) using a FEI SCIOUS Focused Ion Beam – Scanning Electron Microscope (FIB-SEM) DualBeam system with lift-out capabilities. A large lamella (approximately 40 × 60 µm and 5 µm in depth) was extracted from polished petrographic thin sections using an up-scaled transmission electron microscope lamella preparation technique (Heaney et al., 2001). Briefly, large trenches were milled on both sides of the area of interest (Fig. 3.6), followed by a U-cut using a 50 nA gallium probe at an accelerating voltage of 30 kV. The lamella was extracted vertically using the EasyLift Manipulator system, allowing a previously unexposed cross-sectional area to be analysed using NanoSIMS. Sample preparation using the FIB-SEM preserved the structure of the encrusted cell envelopes (Fig. 3.7) and allowed for sample preparation away from resin, reducing potential sample contamination. The lamella was attached to a copper half grid using ion beam induced platinum deposition. The lamella was polished using decreasing beam currents to remove sample surface striations. A final step of low energy FIB polishing was carried out at 3 nA to reduce the ion beam damage on the sample’s surface.

3.2.3.2. NanoSIMS analysis Elemental maps were acquired for encrusted microbial cell envelopes and permineralised microfossils using the CAMECA NanoSIMS 50 (CAMECA, Paris, France) at the University of Western Australia, which allows for the simultaneous collection of 5 ions. In this experiment, to maximise the spatial resolution of the NanoSIMS, elemental maps were acquired using the Cs+ ion beam, which was focused to a diameter of approximately 50 nm with a beam current of 0.3 pA. The primary Cs+ beam was used to sputter secondary ions to investigate biosignatures preserved within an iron-rich matrix. For permineralised microfossils 12 - 28 - 31 - 43 - 72 - C2 , Si , P , AlO , FeO secondary ions were collected. Preliminary maps of permineralised microfossils revealed that CN⁻ cluster ions were not preserved, therefore only - 12 - 12 14 - C2 ions were measured in high-resolution maps. For encrusted cell envelopes, C2 , C N , 31P-, 43AlO-, 72FeO- secondary ions where collected. It should be noted that nitrogen cannot be measured directly using NanoSIMS, therefore the carbon-nitrogen cluster ion was measured to determine the presence of nitrogen biosignatures associated with microfossils. Aluminium- oxide and iron-oxide cluster ions were measured to allow iron and aluminium to be mapped with carbon, nitrogen and phosphorus simultaneously, collecting secondary ions from a single plane. High magnification elemental micrographs were acquired by rastering the beam over a field of view of 8 × 8 µm at a resolution of 256 × 256 pixels with a dwell time of 40 – 50 ms

53 Chapter Three per pixel. Samples were pre-sputtered using the Cs+ ion beam to remove any surface contamination and implant Cs+ into the sample surfaces.

FIJI software was used to produce NanoSIMS chemical micrographs of permineralised microfossils and encrusted cell envelopes. Semi-quantitative NanoSIMS maps are presented as black and white intensity maps, with white areas having a higher relative elemental concentration compared with darker regions (Hoppe et al., 2013). For permineralised microfossils, the colocalisation between phosphorus and aluminium for the permineralised microfossils was determined based on the Pearson correlation coefficient (r) for the entire region analysed using NanoSIMS. For encrusted cell envelopes, cross-sections of individual microfossils (n = 2) were analysed to determine the correlation of preserved carbon and nitrogen with aluminium and iron. To achieve this, cross-sections of individual cells were plotted as pixel distance against ion counts. The Pearson correlation coefficient (r) was calculated to determine the correlation between carbon and nitrogen with iron and aluminium.

3.3. Results Field Emission scanning electron microscopy was used to reveal two distinct microfossil textures preserved within iron-rich environments: permineralised microfossils (Figs. 3.2 – 3.4) and cell envelope structures encrusted in iron oxide minerals (Figs. 3.5 – 3.8). Nanoscale secondary ion spectrometry (NanoSIMS) highlighted that aluminium is enriched around permineralised microfossils, which appears to govern the structural preservation of permineralised microfossils (Fig. 3.4). For encrusted cell envelopes, NanoSIMS reveals that aluminium is enhanced in regions with preserved carbon and nitrogen biosignatures, which are likely to be remnant organic cell envelope structures (Fig. 3.8). The correlation of aluminium with organic carbon and nitrogen is highlighted by intensity cross-sections of individual cells (regions of interest; ROI) where ion counts are plotted against pixel distance (Fig. 3.9). Iron concentrations are depleted in all regions where carbon and nitrogen are preserved (Fig. 3.9). The carbon and nitrogen biosignatures associated with the encrusted cell envelopes are referred to as organic biosignatures.

3.3.1. Permineralised microfossils The lake-edge ferruginous duricrust sample in which the permineralised microfossils are identified has an approximate pH of 6.5. X-ray diffraction data indicate that goethite and hematite are the dominant minerals present in the duricrust sample with minor kaolinite. The duricrust sample is primarily composed of Fe (52.64%), Al2O3 (8.63%) and SiO2 (4.92%), with

The role of aluminium in biosignature preservation

minor TiO2 (0.85%) and P (0.227%). All other cations are below 0.1% (Table 3.1). Microfossil morphologies are typically rod-shaped, ranging from 0.5 – 1 m in diameter and 1.5 – 2 m in length (Fig. 3.3). Evidence of cell-replication (Fig. 3.3; arrows) and colony formation highlight these structures represent fossilised microorganisms. Relatively large microbial clusters (biofilms) are present throughout the ferruginous duricrust sample (Fig. 3.2). Microfossils within the lake-edge sample had putatively grown within the lake environment (pH 5.6 – 5.8).

55 Chapter Three

Table 3.1. Bulk chemical (XRF) and mineralogical characteristics and the primary fossilisation texture with the ferruginous duricrust sample and the goethite-cemented sample associated.

Permineralised microfossils were analysed for the preservation of chemical biosignatures using NanoSIMS in the location highlighted in Fig. 3.3. NanoSIMS analysis highlights that aluminium and phosphorus are enriched around all permineralised microfossils, with phosphorus distribution showing an affinity with aluminium (Fig. 3.4; r = 0.41). Phosphorus

The role of aluminium in biosignature preservation enrichment around permineralised microfossils cannot be considered an organic biosignature; it is simply a consequence of the preferential sorbtion of phosphorus to aluminium-substituted iron oxide minerals (Ruan and Gilkes, 1996; Schulze and Schwertmann, 1984). In contrast, iron is enriched within the intracellular regions of microbial fossils. Carbon and nitrogen biosignatures are also not preserved in association with permineralised microfossils (Fig. 3.4). In the absence of organic biosignatures associated with permineralised microfossils, the enrichment of aluminium around permineralised microfossils is responsible for the structural preservation of the microorganisms. The structural preservation of microfossils indicates that the iron-aluminium oxide minerals had not undergone a significant alteration or dissolution event.

3.3.2. Encrusted cell envelopes The pH of the goethite-cemented vein cross-cutting the saprolite in which the encrusted cell envelopes were identified was approximately 6.3. The primary mineralogy of goethite and hematite is supported by the high iron content (57.9 wt.%). Relatively low bulk-rock Al2O3

(2.54 wt.%) and SiO2 (1.55 wt.%) concentrations are present with minor TiO2 (0.2 wt.%) and P (0.240 wt.%; Table 3.1). Encrusted microbial cell envelopes are preserved throughout the goethite-rich vein that cross-cut the hematite-enriched saprolite in Serra do Gandarela, Quadrilátero Ferrífero (Fig. 3.5). In contrast to the permineralised microfossils, SEM-EDS and FTIR microspectroscopy reveals that mineralised microbial cell envelopes are preserved within the pore spaces between with gibbsite clasts. Secondary electron micrographs reveal that the three-dimensional (3D) structure of rod-shaped microfossils is preserved within pore spaces (Fig. 3.5B). Encrusted cell envelopes are typically rod-shaped and approximately 1 µm in diameter and 1.5 µm in length (Fig. 3.5B). Ferruginised plant roots could not be identified in thin samples using scanning electron microscopy, therefore the distribution of microfossils with respect to plant roots is not apparent.

57 Chapter Three

Fig. 3.2. Backscattered electron scanning electron micrographs highlighting rod-shaped microfossils (A) and large permineralised microbial biofilms (B) within an iron-rich duricrust capping iron ore deposits in the Carajás Mineral Province, State of Pará, Brazil. Permineralised microfossils formed around highly weathered kaolinite-rich clasts.

The role of aluminium in biosignature preservation

Fig. 3.3. High magnification backscattered electron micrograph of rod-shaped permineralised microfossils, with evidence of colony formation and cell replication (arrows). The rectangle highlights the region analysed using NanoSIMS (see Fig. 3.4).

A variety of fossilisation textures were associated with the encrusted cell envelopes (Fig. 3.7). The carbon and carbon-nitrogen NanoSIMS elemental micrographs highlight that cell envelope biosignatures associated with the mineralised cell envelopes are preserved (Fig. 3.8). The intracellular regions of fossilised cell envelopes are infilled with secondary iron oxide minerals to varying degrees, ranging from completely void (Fig. 3.8; white arrow), partially infilled (Fig. 3.8; Region of Interest (ROI) 1) and completely infilled microfossils (Fig. 3.8; ROI 2). The completely infilled microfossil shares a similar chemical signature with the permineralised microfossils: iron-enriched intracellularly and aluminium enriched around the microfossil (Fig. 3.8). The carbon and nitrogen intensities are lower for the completely infilled microfossil compared with the partially infilled and void microfossils (Fig. 3.8 – 3.9).

59 Chapter Three

Fig. 3.4. NanoSIMS micrographs of permineralised microfossils presented as black and white intensity maps, with white areas having a higher relative elemental concentration compared with darker regions. NanoSIMS micrographs reveal that organic biosignatures are not preserved with permineralised microfossils. Aluminium is enriched around the microfossils, while iron is enriched within the intracellular regions of microfossils. Phosphorus colocalises with aluminium (r = 0.41). The composite micrograph highlights aluminium (green) enrichment around microfossils, with iron (blue) is enriched within intracellular regions. All micrographs are 8 × 8 µm.

The role of aluminium in biosignature preservation

Fig. 3.5. Field Emission scanning electron micrographs of mineral encrusted cell envelope structures identified within a goethite cemented vein that cross-cut the saprolite of a weathered banded iron formation in the Serra do Gandarela, Quadrilátero Ferrífero in the State of Minas Gerais, Brazil. (A) Backscattered electron micrograph highlighting the preservation of encrusted cell envelopes within pore spaces. (B) Secondary electron micrograph of encrusted cell envelopes highlighting the three-dimensional preservation of encrusted cell envelopes that had formed in pore spaces.

61 Chapter Three

Cross-sections of individual microfossils reveals that aluminium is enhanced in regions with preserved carbon and nitrogen (organic) biosignatures (Fig. 3.9). A cross-section of a partially infilled microbial fossil (Fig. 3.8; ROI 1) highlights that aluminium is enhanced with the preserved cell envelope structures and correlates well with preserved carbon (r = 0.67) and nitrogen (Fig. 3.9; r = 0.76). In contrast, the relative iron concentration is depleted in locations and poorly correlates with preserved carbon (r = 0.13) and nitrogen (r = 0.24) biosignatures. The relative iron concentration appears to be reduced for all microfossils in regions where carbon and nitrogen biosignatures are preserved (Fig. 3.8). Similarly, a cross-section of the completely infilled cell (ROI 2) reveals that aluminium is slightly enriched in regions where organic biosignatures are preserved (Fig. 3.9). Aluminium positively correlates with the preserved carbon (r = 0.76) and nitrogen (r = 0.88) biosignatures (Fig. 3.9) in the completely infilled microfossil (Fig. 3.8). In comparison, iron negatively correlates with preserved organic biosignatures and aluminium (Fig. 3.9).

3.4. Discussion The microfossils in the present study are all preserved within iron-rich rocks, with a bulk iron concentration of approximately 52.64 wt.% for the ferruginous duricrust (permineralised microfossils) sample and 57.90 wt.% for the goethite-cemented vein (encrusted cell envelopes). Iron biomineralisation has been suggested to represent the first stages of microbial fossilisation and may be essential for the structural integrity of cell components after cell death (Ferris et al., 1988; Li et al., 2013; Li et al., 2014; Miot et al., 2009a; Miot et al., 2009b; Miot et al., 2011; Picard et al., 2015; Schädler et al., 2009). Therefore, iron was hypothesised to drive the fossilisation of the microfossils presented here. The reduced iron concentration and the enrichment of aluminium associated with preserved organic biosignatures (Figs. 3.8 – 3.9) suggests that aluminium complexing with cell envelope structures before or immediately after cell death may have been responsible for the preservation of organic biosignatures associated with encrusted cell envelopes. Aluminium complexation with organic cell envelopes appears to resist autolytic degradation of cell envelope structures following cell death and contributes to the preservation of organic biosignatures associated with encrusted microfossils (Figs. 3.8 – 3.9).

Carbon and nitrogen biosignatures were not preserved with the permineralised microfossils. In the absence of preserved organic biosignatures, aluminium enrichment around permineralised

The role of aluminium in biosignature preservation

Fig. 3.6. Backscattered electron scanning electron micrograph of encrusted cell envelopes that had infilled pore spaces between gibbsite-rich clasts. The rectangle highlights the regions from which a lamella was extracted using a focused ion beam scanning electron microscope for NanoSIMS analysis (see Figs. 3.7. – 3.8.). microfossils was responsible for the structural preservation of the permineralised microfossils within iron-rich duricrusts (Figs. 3.2 – 3.4). The NanoSIMS micrographs presented here indicate that aluminium may play a critical role in the structural and physicochemical preservation of microfossils within iron-rich environments.

Iron-aluminosilicate minerals have been demonstrated to mineralise cell envelope structures of microorganisms (Ferris et al., 1987). To date, research has focused on the role of iron in microbial fossilisation and biosignature preservation (Ferris et al., 1988; Li et al., 2013; Miot et al., 2011; Picard et al., 2015). Iron-organic complexes have been proposed to represent the initial stages of microbial fossilisation in iron-rich environments (Li et al., 2013), resisting the degradation of organic biosignatures (Picard et al., 2015). The data presented here indicate that aluminium ions binding with microbial cell envelope structures may increase the successful fossilisation and the preservation of organic biosignatures within the geologic record.

The iron-rich duricrust sample and the goethite cemented vein both had a circumneutral pH, therefore aluminium should be relatively immobile within these environments (Bache, 1986). Within the duricrust sample, permineralised microfossils (Fig. 3.2) were routinely identified

63 Chapter Three within proximity to highly weathered kaolinite-rich clasts that may have been actively weathered by microorganisms via the exudation of organic acids. The encrusted microbial cell envelopes (Fig. 3.5) were identified in the pore space between gibbsite grains within the goethite-cemented vein. Given the stability of gibbsite in circumneutral environments, the production of organic acids is likely to control the release of aluminium from gibbsite clasts (Bache, 1986). Weathering of gibbsite grains (Fig. 3.1) may have released aluminium ions into solution, contributing to microbial fossilisation and preservation. Aluminium and phosphorus can be present in unweathered BIFs in Brazil (Dorr, 1973), which can be enriched and maintained within the ferruginuous duricrusts and the saprolite of highly weathered BIFS (Dorr, 1964).

Fig. 3.7. High magnification backscattered electron Field Emission micrograph of encrusted cell envelope sample that had been extracted using a focused ion beam scanning electron microscope from the region highlighted in Fig. 3.6. A variety of fossilisation textures were associated with the encrusted cell envelopes. The rectangle represents the region analysed using NanoSIMS (see Fig. 3.8).

The role of aluminium in biosignature preservation

Fig. 3.8. NanoSIMS analysis of mineral encrusted cell envelopes are displayed as black and white intensity maps, with higher concentrations represented by white areas and lower concentrations displayed as darker regions. NanoSIMS micrographs highlight the preservation of carbon and nitrogen biosignatures associated with the microbial cell envelope. Phosphorus was distributed throughout the sample and only enhanced due to edge effects associated with the hollow encrusted cell envelopes. Aluminium and iron are present throughout the sample but aluminium appears to be slightly enriched where preserved organic biosignatures are preserved (see Fig. 3.9). In contrast, iron concentrations are reduced in regions where organic carbon and nitrogen are preserved. The composite micrographs highlights the poor correlation of iron (blue) with preserved organic nitrogen (green). All micrographs are 8 × 8 µm.

65 Chapter Three

3.4.1. Permineralised microfossils In the absence of chemical biosignatures associated with the permineralised microfossils, bacteriomorphic structures were determined to represent remnant microorganisms using the guide by Westall (1999). The biogenicity of permineralised microfossils was determined based on cell size (1 – 2 µm in length), shape (rod-shaped with curved ends), evidence for cell replication (Fig. 3.3) and microbial colony formation (Fig. 3.2). Permineralised microbial biofilms within ferruginous duricrusts have previously been demonstrated to colonise in proximity to and along the grain boundaries of kaolinite-rich clasts (Levett et al., 2016). The preservation of subcellular structures including periplasmic structures may be preserved associated with microfossils despite the pseudomophic mineral replacement of the majority of organic carbon associated with the initial microbial cell (Cosmidis et al., 2013). High resolution micrographs of permineralised microfossils identified in iron-rich duricrusts that highlights the preservation of rod-shaped microfossils with evidence of binary fission, cell envelope bilayers and filamentous bacteriomorphs have previously been published (Levett et al., 2016). NanoSIMS analysis of permineralised microfossils identified within an iron-rich duricrust that capped iron ore deposits demonstrates that aluminium was critical to the structural preservation of microfossils. All permineralised microfossils shared a consistent fossilisation chemical signature: aluminium enriched around the microfossils and iron enriched within the remnant cell. Organic biosignatures are not preserved in association with the permineralised microfossils, possibly replaced by the pseudomorphic precipitation of iron and aluminium oxide minerals that have structurally preserved the microfossils (Cosmidis et al., 2013).

3.4.2. Encrusted cell envelopes The preservation of organic carbon and nitrogen associated with the encrusted cell envelope fossils (Fig. 3.8) agrees with the 3D preservation of rod-shaped microfossils (Fig. 3.5B). Cosmidis et al. (2013) observed similar fossilisation textures within a coprolite, with calcium phosphate minerals forming around microfossils and infilling remnant cells to varying extents. NanoSIMS analysis of encrusted cell envelopes highlights that aluminium is enhanced in association with preserved carbon and nitrogen and may play a direct role in the preservation of organic biosignatures associated with the microbial cell envelopes.

Aluminium has previously been demonstrated to bind to cell envelope structures in E.coli, inhibiting cell growth (Guida et al., 1991). Preserved carbon and nitrogen biosignatures associated with microfossils must have resisted enzymatic and oxidative degradation despite being fossilised within a highly oxidising environment, evidenced from the primary iron oxide

The role of aluminium in biosignature preservation minerals being goethite and hematite. Iron was relatively depleted in locations where carbon and nitrogen biosignatures are preserved (Fig. 3.8); however, iron oxide precipitates have contributed to cell envelope preservation by extending the mineral encrustation around the microfossils. The additional mineralisation of cell envelopes is likely to have reduced the exposure of carbon and nitrogen biosignatures to chemical and enzymatic degradation. These results indicate the resistance of aluminium to changes in oxidation potential may promote the preservation of organic biosignatures (Bache, 1986).

Encrusted cell envelopes with differing degrees of secondary iron oxide infilling and organic biosignature preservation allows for the investigation of microfossil weathering. In the present study, internal mineral precipitates nucleated at the membrane-cytoplasm boundary, forming a continuous layer parallel with cell membrane (Fig. 3.8; ROI1). Secondary mineral growth may then continue from the membrane-cytoplasm boundary and infill the cytoplasm of remnant cells. These observations are consistent with laboratory (Benzerara et al., 2004) and environmental studies (Cosmidis et al., 2013) investigating internal mineral precipitates. Analysis of the partially infilled microbial fossil (Fig 3.8; ROI 1) highlights that aluminium concentrations continue to decrease in intracellular regions where no organic carbon or nitrogen is preserved. In contrast, iron concentrations are slightly increased in intracellular regions (Fig. 3.9). For the completely infilled cell (Fig 3.8: ROI 2), carbon and nitrogen associated with the cell envelope were reduced, indicating that it may represent a more advanced stage of environmental weathering and recrystallisation compared with partially infilled encrusted cell envelopes. The completely infilled cells shared a similar chemical signature with the permineralised microfossils: iron enriched within intracellular regions and aluminium-enriched around the microfossil.

The preservation of organic biosignatures associated with microorganisms are typically investigated by increasing temperature and pressure to simulate diagenesis (Beveridge et al., 1983; Li et al., 2014; Oehler and Schopf, 1971; Picard et al., 2015). The reduced carbon and nitrogen signal in the completely infilled cell (Fig. 3.8; ROI 2) indicates that the continuous precipitation of minerals during weathering, not exposure to increased temperatures and pressures, may be a limiting factor for the preservation of organic biosignatures. Therefore, low temperature and pressure weathering experiments are required to assess the preservation of organic biosignatures associated with microfossils in iron-rich environments.

67 Chapter Three

3.4.3. A model for the role of aluminium in microbial fossilisation In agreement with the fossilisation textures presented here, Londono et al. (2017) demonstrated aluminium colocalised with cell envelope structures (in comparison with intracellular regions), while iron concentrations were enriched within intracellular regions. Aluminium complexation with phosphate groups of plasma membranes has been demonstrated to destabilise the membrane structures and disrupt the membrane permeability (Deleers et al., 1985; Deleers et al., 1986), which may allow iron to permeate into the intracellular regions and restricting intracellular aluminium transportation (Londono et al., 2017).

The intensity cross-section of the completely infilled microbial fossil (Fig. 3.8; ROI 2) displays a similar elemental distribution to the permineralised microfossils (Figs. 3.2 – 3.4): aluminium enriched around the cell envelope and iron enriched within the intracellular regions of the microfossil. Microbial plasma membrane structures that resist the transportation of aluminium ions into intracellular regions (Londono et al., 2017) may have resulted in the nucleation of aluminium-enriched minerals on microbial cell surfaces. Aluminium may then be further enriched by the preferential inclusion of aluminium into these aluminium-enriched minerals. The pseuomorphic precipitation of iron and aluminium minerals from solution is likely to replace the organic biosignatures. The permineralised microfossils may represent a more advanced stage of microfossil weathering compared with the encrusted cell envelopes that showed a similar chemical signature when completely infilled (Fig. 3.8; ROI 2).

The enrichment of aluminium around preserved microfossils and the enrichment of iron within intracellular regions requires explanation. Trivalent iron and aluminium are known to have a strong affinity for anions capable of donating oxygen. Inorganic and organic phosphates therefore present ideal ligands for aluminium and iron complexation. Membrane phosphate end groups have been proposed to have the following preferential affinity for iron and aluminium cations: Fe3+ > Al3+ > Fe2+ (Zatta et al., 2002). Therefore, the poor Fe correlation with preserved organic biosignatures presented here is unexpected and is likely to be explained by aluminium forming an effectively irreversible complex with extracellular organic substances. Aluminium has been demonstrated to bind tenaciously with Mg-dependent enzymes, with exchange rates of approximately 105 slower than Mg. The formation of enzymatically unavailable aluminium- organic complexes may contribute to resisting the degradation of organic biosignatures (Ferris et al., 1988). Additional work is required to demonstrate the preservation of aluminium-organic complexes associated with mineral encrusted microbial cell envelopes.

The role of aluminium in biosignature preservation

Fig. 3.9. Intensity line plots of ion counts plotted against pixel distance cross-cutting microfossils for a partially infilled microfossil (Fig. 3.8; ROI 1) and a completely infilled microfossil (Fig. 3.8; ROI 2). For ROI 1, aluminium is enhanced in association with the organic biosignatures and was slightly enriched in regions where organic carbon and nitrogen are preserved. Iron poorly correlates with organic carbon and nitrogen and is slightly increased in intracellular lumen. Similarly, for the completely infilled cell (Fig. 3.8; ROI 2), aluminium is enhanced in regions with the preserved organic carbon and nitrogen. Note, iron is out of phase with the preserved organic biosignatures. Phosphorus is relatively consistent for both ROI 1 and 2, highlighting these cells were not affected by edge enhancements.

69 Chapter Three

3.4.4. The relative age of microfossils Geochronological data has demonstrated that the ferruginous duricrusts that cap BIFs in Brazil tend to increase in age with depth (Monteiro et al., 2014; Shuster et al., 2012). The top 10 m of the ferruginous duricrust from the Carajás (N4C) mineral province has an approximate average age of 8.25 Ma and samples from the top 2 cm of the profile have an approximate age of 0.9 Ma (Shuster et al., 2012). The ferruginous duricrust sample containing permineralised microfossils was collected directly from the surface (Levett et al., 2016). Therefore, the permineralised microfossils presented here are likely to date to less than 0.9 Ma. Biological mechanisms have been postulated to drive the biogeochemical cycling of iron within the ferruginous duricrusts that cap BIFs (Monteiro et al., 2014). Microfossils in the uppermost crust are therefore likely to have experienced an increased exposure to iron-rich solutions, mineral precipitation and weathering, which appears to have replaced organic biosignatures.

Within a single profile, the saprolite has been demonstrated to have mineralised prior to the overlying ferruginous duricrust (Monteiro et al., 2014). Goethite grains from the saprolite of the Gandarela Syncline have produced dates of approximately 25 to 55 Ma (Monteiro et al., 2014). Therefore, the encrusted cell envelopes identified within the vein that cross-cut the saprolite are likely to have a greater age than the permineralised microfossils identified in the ferruginous duricrust.

3.5. Conclusions High magnifciation NanoSIMS maps of permineralised microfossils and mineral encrusted microbial cell envelopes identified in iron-rich rocks indicates that aluminium may play an important role in the structural and physiochemical preservation of microbial fossils. In the absence of preserved organic biosignatures, aluminium enrichment around permineralised microfossils governs the structural preservation of microorganisms within an iron-rich duricrust that caps iron ore deposits in Brazil. The pseudomorphic precipitation of aluminium and iron oxide minerals associated with permineralised microfossils appears to replace the preservation of organic biosignatures associated with microfossils in iron-rich rocks. For mineralised cell envelope structures, aluminium is enhanced in regions with preserved organic carbon and nitrogen, indicating that aluminium complexation with microbial cell envelopes inhibits the enzymatic and oxidative degradation of organic biosignatures.

The role of aluminium in biosignature preservation

Acknowledgements We acknowledge support from the Vale S.A.-UQ Geomicrobiology initiative and the Australian Research Council Linkage Program (LP140100805) to G. Southam and P. Vasconcelos. The authors acknowledge the facilities and the scientific and technical assistance of the Australian Microscopy and Microanalysis Research Facility at the Centre of Microscopy and Microanalysis, at The University of Queensland. Alan Levett acknowledges that support from the Australian Government Research Training Program.

References Anand R, Verrall M. Biological origin of minerals in pisoliths in the Darling Range of Western Australia. Australian Journal of Earth Sciences 2011; 58: 823-833.

Bache BW. Aluminium mobilization in soils and waters. Journal of the Geological Society 1986; 143: 699.

Beveridge T, Meloche J, Fyfe W, Murray R. Diagenesis of metals chemically complexed to bacteria: laboratory formation of metal phosphates, sulfides, and organic condensates in artificial sediments. Applied and Environmental Microbiology 1983; 45: 1094-1108.

Brasier MD, Antcliffe J, Saunders M, Wacey D. Changing the picture of Earth's earliest fossils (3.5–1.9 Ga) with new approaches and new discoveries. Proceedings of the National Academy of Sciences 2015; 112: 4859-4864.

Deleers M, Servais J-P, Wülfert E. Micromolar concentrations of Al3+ induce phase separation, aggregation and dye release in phosphatidylserine-containing lipid vesicles. Biochimica et Biophysica Acta (BBA) - Biomembranes 1985; 813: 195-200.

Deleers M, Servais J-P, Wülfert E. Neurotoxic cations induce membrane rigidification and membrane fusion at micromolar concentrations. Biochimica et Biophysica Acta (BBA) - Biomembranes 1986; 855: 271-276.

Dorr JVN. Supergene iron ores of Minas Gerais, Brazil. Economic Geology 1964; 59: 1203- 1240.

Dorr JVN. Iron-formation in South America. Economic Geology 1973; 68: 1005-1022.

Exley C, Birchall D. The cellular toxicity of aluminium. Journal of Theoretical Biology 1992; 159: 83-98.

71 Chapter Three

Ferris FG, Fyfe WS, Beveridge TJ. Bacteria as nucleation sites for authigenic minerals in a metal-contaminated lake sediment. Chemical Geology 1987; 63: 225-232.

Ferris FG, Fyfe WS, Beveridge TJ. Metallic ion binding by Bacillus subtilis: implications for the fossilization of microorganisms. Geology 1988; 16: 149-152.

Guida L, Saidi Z, Hughes MN, Poole RK. Aluminium toxicity and binding to Escherichia coli. Archives of Microbiology 1991; 156: 507-512.

Heaney PJ, Vicenzi EP, Giannuzzi LA, Livi KJ. Focused ion beam milling: a method of site- specific sample extraction for microanalysis of Earth and planetary materials. American Mineralogist 2001; 86: 1094-1099.

Hoppe P, Cohen S, Meibom A. NanoSIMS: technical aspects and applications in cosmochemistry and biological geochemistry. Geostandards and Geoanalytical Research 2013; 37: 111-154.

Jacobi CM, Carmo FF. Life-forms, pollination and seed dispersal syndromes in plant communities on ironstone outcrops, SE Brazil. Acta Botanica Brasilica 2011; 25: 395- 412.

Konhauser KO, Jones B, Phoenix VR, Ferris G, Renaut RW. The microbial role in hot spring silicification. AMBIO: A Journal of the Human Environment 2004; 33: 552-558.

Levett A, Gagen E, Shuster J, Rintoul L, Tobin M, Vongsvivut J, et al. Evidence of biogeochemical processes in iron duricrust formation. Journal of South American Earth Sciences 2016; 71: 131-142.

Li J, Benzerara K, Bernard S, Beyssac O. The link between biomineralization and fossilization of bacteria: insights from field and experimental studies. Chemical Geology 2013; 359: 49-69.

Li J, Bernard S, Benzerara K, Beyssac O, Allard T, Cosmidis J, et al. Impact of biomineralization on the preservation of microorganisms during fossilization: an experimental perspective. Earth and Planetary Science Letters 2014; 400: 113-122.

Londono SC, Hartnett HE, Williams LB. Antibacterial activity of aluminum in clay from the Colombian Amazon. Environmental Science & Technology 2017; 51: 2401-2408.

Macdonald TL, Martin BR. Aluminum ion in biological systems. Trends in Biochemical Sciences 1988; 13: 15-19.

Marshall CP, Emry JR, Marshall AO. Haematite pseudomicrofossils present in the 3.5-billion- year-old Apex Chert. Nature Geoscience 2011; 4: 240.

Martin RB. The chemistry of aluminum as related to biology and medicine. Clinical Chemistry 1986; 32: 1797.

Miot J, Benzerara K, Morin G, Kappler A, Bernard S, Obst M, et al. Iron biomineralization by anaerobic neutrophilic iron-oxidizing bacteria. Geochimica Cosmochimica Acta 2009a; 73: 696-711.

The role of aluminium in biosignature preservation

Miot J, Benzerara K, Obst M, Kappler A, Hegler F, Schädler S, et al. Extracellular iron biomineralization by photoautotrophic iron-oxidizing bacteria. Applied and Environmental Microbiology 2009b; 75: 5586-5591.

Miot J, Maclellan K, Benzerara K, Boisset N. Preservation of protein globules and peptidoglycan in the mineralized cell wall of nitrate‐reducing, iron (II)‐oxidizing bacteria: a cryo‐electron microscopy study. Geobiology 2011; 9: 459-470.

Monteiro HS, Vasconcelos PM, Farley KA, Spier CA, Mello CL. (U–Th)/He geochronology of goethite and the origin and evolution of cangas. Geochimica et Cosmochimica Acta 2014; 131: 267-289.

Oehler JH, Schopf JW. Artificial microfossils: experimental studies of permineralization of blue-green algae in silica. Science 1971; 174: 1229-1231.

Picard A, Kappler A, Schmid G, Quaroni L, Obst M. Experimental diagenesis of organo- mineral structures formed by microaerophilic Fe(II)-oxidizing bacteria. Nature Communications 2015; 6.

Preston L, Shuster J, Fernández-Remolar D, Banerjee N, Osinski G, Southam G. The preservation and degradation of filamentous bacteria and biomolecules within iron oxide deposits at Rio Tinto, Spain. Geobiol. 2011; 9: 233-249.

Riding R. Microbial carbonates: the geological record of calcified bacterial–algal mats and biofilms. Sedimentology 2000; 47: 179-214.

Ruan HD, Gilkes RJ. Kinetics of phosphate sorption and desorption by synthetic aluminous goethite before and after thermal transformation to hematite. Clay Minerals. 31, 1996, pp. 63.

Salama W, El Aref M, Gaupp R. Mineral evolution and processes of ferruginous microbialite accretion–an example from the Middle Eocene stromatolitic and ooidal ironstones of the Bahariya Depression, Western Desert, Egypt. Geobiol. 2013; 11: 15-28.

Schädler S, Burkhardt C, Hegler F, Straub K, Miot J, Benzerara K, et al. Formation of cell- iron-mineral aggregates by phototrophic and nitrate-reducing anaerobic Fe (II)- oxidizing bacteria. Geomicrobiology Journal 2009; 26: 93-103.

Schulze DG, Schwertmann U. The influence of aluminium on iron oxides; X, Properties of Al- substituted goethites. Clay Minerals 1984; 19: 521-539.

Spier C, de Oliveira S, Rosière C. Geology and geochemistry of the Águas Claras and Pico Iron Mines, Quadrilátero Ferrífero, Minas Gerais, Brazil. Mineralium Deposita 2003; 38: 751-774.

Westall F. The nature of fossil bacteria: a guide to the search for extraterrestrial life. J. Geophys. Res. Planets 1999; 104: 16437-16451.

73 Chapter Three

Zatta P, Kiss T, Suwalsky M, Berthon G. Aluminium(III) as a promoter of cellular oxidation. Coordination Chemistry Reviews 2002; 228: 271-284.

Chapter 4 Characterisation of iron oxide encrusted microbial fossils

Alan Levetta, Emma J. Gagena, Llew Rintoulb, Paul Guagliardoc, Hui Diaod, Paulo M. Vasconcelosa, Gordon Southama

aSchool of Earth and Environmental Sciences, University of Queensland, Brisbane, Queensland 4072, Australia

bCentral Analytical Research Facility, Institute of Future Environments, Queensland University of Technology, Brisbane, Queensland 4001, Australia

cCentre for Microscopy, Characterisation and Analysis, University of Western Australia, Perth 6009, Western Australia, Australia

dCentre for Microscopy and Microanalysis, University of Queensland, Brisbane 4072, Queensland, Australia

As an aspiring geomicrobiologist, contributing to our understanding of the preservation of biosignatures in the geological record felt like a rite of passage.

A work in progress, this is my contribution so far…

Manuscript in review with Scientific Reports

Chapter Four

Abstract Robust methods for the characterisation of microbial biosignatures in geological matrices is critical for developing mineralogical biosignatures. Studying microbial fossils is fundamental for our understanding of the role microorganisms have played in elemental cycling in modern and ancient environments on Earth and potentially Mars. Here, we aim to understand what promotes the fossilisation of microorganisms after the initial stages of biomineralisation, committing bacteriomorphic structures to the geological record within iron-rich environments. Mineral encrusted cell envelope structures were routinely identified within a goethite-rich vein that cross-cut the saprolite (iron ore) of a weathered banded iron formation (BIF) system in the Quadrilátero Ferrífero, Brazil. The preservation of potential organic and mineralogical biosignatures associated with these fossils was characterised using the following high- resolution analytical techniques: scanning and transmission electron microscopy, focused ion beam scanning electron microscopy, nanoscale secondary ion mass spectrometry, synchrotron- based Fourier transform infrared spectroscopy and Raman spectroscopy. Electron microscopy demonstrated that mineral nucleation associated with a range of cell envelope structures typically followed the extant cell templates. These biologically-influenced iron-rich minerals are microcrystalline with minimal secondary growth. In contrast, intracellular mineralisation formed larger minerals that grew inward from the cell membrane to infill intracellular voids after cell death. A three dimensional reconstruction of encrusted cell envelopes in a fossilised biofilm suggests that microorganisms may be able to replicate, during the initial stages of mineralisation. Carbon and nitrogen signatures are preserved associated with the cell envelope structures; however, there were no conclusive mineralogical biosignatures associated with the mineralised cell envelopes highlighting the classical importance of morphology and elemental biosignatures in determining the biogenicity of bacteriomorphic structures.

Biosignature characterisation

4.1. Introduction In 1996, the geoscience community was challenged with a question: What evidence is required to prove the existence of life (McKay et al., 1996)? Characterisation of the ALH84001 meteorite presented 20 nm diameter tubular nanofossil structures, indirect organic signatures and mineral formation anomalies as evidence for past life. Intense scrutiny of these controversial results has led to questions regarding the robustness of these signatures as evidence of life; however, the work has been instrumental in stimulating scientists to better understand microbial fossilisation and the array of microbial biosignatures that may be preserved (Davila et al., 2008).

The surface of Mars is extremely inhospitable for life as we know it, with records of life on Mars likely to be restricted to mineralogical biosignatures (Banfield et al., 2001). Therefore, in-depth characterisation of microbial fossils here on Earth is required to develop new biosignatures that may be preserved in the geological record, particularly mineralogical biosignatures. Low temperatures and pressures, little-to-no atmospheric protection from ionising radiation and oxidising geological conditions provide little optimism for finding extant life. Future Mars Rover missions are being designed to collect, encapsulate and store samples from below the surface (up to 2 m depth) for later collection and transportation back to Earth. The potential success of such rover-based missions requires identifying key near-surface environments on Earth conducive to microbial fossilisation (McMahon et al., 2018). This article responds to a call for additional work and understanding of microbial fossilisation in pore and fracture filling near-surface environments (McMahon et al., 2018).

Here, we have correlated a range of high-resolution analytical techniques to characterise well- preserved iron oxide encrusted microbial cell envelopes fossilised in vein structures below the surface (~ 15 m depth) to aid in the search for robust microbial biosignature targets. Studying the mineralogy associated with microfossils assists in constraining the environments in which microorganisms existed and their role in altering the biogeochemistry of their local environment. The continued development of nano- and microscale analytical techniques provides scientists with an increasingly large toolkit to more effectively characterise bacteriomorphic structures and determine their biogenicity. This article offers insights into the effectiveness of various analytical methods when assessing biogenicity of bacteriomorphic structures and the development of robust biosignature targets. In addition, this manuscript serves a timely reminder that we, as a scientific community, must maintain high standards for

77 Chapter Four what we accept as microbial fossils as set out by Westall (1999) to avoid ambiguity in the literature.

4.2. Materials and methods Naturally fossilised microorganisms were identified within a goethite-rich vein that cross-cut an iron ore deposit in the Quadrilátero Ferrífero, Minas Gerais, Brazil from a depth of approximately 15 m (Levett et al., 2019a; Monteiro et al., 2018). The goethite-rich vein formed during the weathering of the hosting BIF and the vein minerals, likely to be carbonate or sulphide (Monteiro et al., 2018). (U-Th)/He geochronology of goethite fragments from this sample (G-12-11) yield precipitation ages ranging from 40 to 30 Ma (Monteiro et al., 2018), demonstrating the preservation of these biosignatures over geologic time.

4.2.1. Scanning electron microscopy Polished petrographic thin sections (100 µm thick) were prepared by dehydrating the vein rock fragment at 40 ºC overnight and embedding in EpoxiCure 2 epoxy. Thin sections, coated in 10 nm iridium using a BAL-TEC MSC-010 sputter coater, were examined using a JEOL7100 scanning electron microscope in backscattered electron mode at an accelerating voltage of 15 kV to identify fossilised microorganisms and regions of interest for further analysis.

4.2.2. Focused ion beam scanning electron microscopy 4.2.2.1 Sample preparation For all analyses except Raman spectroscopy, sample preparation of mineralised cell envelopes was performed using a FEI Scious focused ion beam scanning electron microscope (FIB-SEM) DualBeam system with lift-out capabilities to extract lamella. For nanoscale secondary ion mass spectrometry (NanoSIMS) and synchrotron-based Fourier transform infrared spectroscopy (FT-IR), procedures outlined in Levett et al. (2019a) were followed for sample preparation. Briefly, a large lamella, 40 µm × 80 µm and 4 µm thick, was prepared by upscaling the typical TEM lamella preparation technique (Heaney et al., 2001). The heterogeneous nature of the fossilised biofilms and the fact that lamellae were extracted vertically from the thin section meant the preparation of a quality section was somewhat serendipitous. For transmission electron microscopy, a standard-sized lamella (approximately 4 × 4 µm and 0.1 µm thick) was prepared from a region previously used to remove a large lamella, thereby offering a view into the ‘subsurface’ in these thin sections.

Biosignature characterisation

4.2.2.2. Three dimensional (3D) visualisation of fossilised biofilms Microfossils encrusted in iron oxides were imaged using slice-and-view technology on the FIB- SEM for high-resolution 3D imaging. A platinum deposition was used to protect an area of approximately 25 × 25 µm containing abundant fossilised microorganisms. A gallium probe of 30 nA was used to mill trenches surrounding the area of interest to prevent shadowing effects during imaging. Slices with a thickness of 100 nm and approximately 20 µm in depth were milled using a 3 nA gallium ion probe. A series of backscattered scanning electron (BSE) micrographs were accumulated using an accelerating voltage of 2 kV during the automated slice-and-view process. Amira 6.1. (FEI Visualisation Sciences Group) software was used to assemble the electron micrographs to reconstruct the three-dimensional structure of the fossilised microorganisms.

4.2.3. Nanoscale secondary ion mass spectrometry (NanoSIMS) High-resolution elemental maps were acquired using the CAMECA NanoSIMS 50 at the University of Western Australia. Elemental maps were acquired using a Cs+ ion beam, which was focused to a diameter of 50 – 60 nm, using a current of 0.3 pA. The primary Cs+ ion beam 12 - 12 14 - 27 16 - 56 16 - was used to sputter the following secondary ions: C2 , C N , Al O and Fe O . High- resolution elemental micrographs were acquired from a 12 × 12 µm region of interest by rastering the beam with a dwell time of 80 msec per pixel at a resolution of 256 × 256 pixels. To remove any surface contamination and implant Cs+ into the sample’s surface, lamella were pre-sputtered for 10 min using the Cs+ ion beam. Semi-quantitative maps are presented as grayscale intensity maps, with white regions indicating a higher relative abundance (Hoppe et al., 2013). FIJI software using the OpenMIMS plugin was used for data analysis.

4.2.4. Infrared microspectroscopy 4.2.3.1. Fourier transform infrared (FT-IR) microspectroscopy To determine the nature of organic biosignatures preserved associated with the encrusted cell envelope structures, a FIB lamella was analysed using infrared microspectroscopy at the Australian Synchrotron (Clayton, Australia). A Bruker V80v FT-IR spectrometer with a photovoltaic liquid nitrogen cooled mercury-cadmium-telluride (MCT) detector system (Bruker Optik GmbH, Ettlingen, Germany) was operated in transmission mode with a beamsize of 4.17 µm to acquire mid-infrared spectra (3800 – 900 cm-1). Spectra were obtained by accumulating 128 scans with a resolution of 4 cm-1. The sample was moved in a raster fashion with a step size of 2 µm. Data were processed using OPUS 8.0 software (Bruker Optik, GmbH, Ettlingen, Germany). To produce the infrared map, all spectra were integrated between 2910

79 Chapter Four

-1 and 2940 cm to highlight regions enriched with aliphatic hydrocarbon moieties of CH2. The infrared map is displayed as a contoured heat-map, with the largest integrated absorbance areas represented by pink regions.

4.2.3.2. Raman microspectroscopy Raman spectra from a petrographic thin section were acquired using a WITec Alpha 300 series Raman equipped with a 532 nm laser, operated at 0.53 mW. The laser was focussed with a Zeiss Plan-Neofluar 100×/1.30 Oil objective using Zeiss Immersol 518 F immersion oil. The region of interest was mapped by raster motion in X and Y increments of 0.25 m with a dwell time of 12 s. To generate mineral maps, Classical Least Squares (CLS) methods used each spectrum as a linear combination of individual constituent spectra, plus error. Using this method, CLS can decompose a spectrum into a set of constituent scores of individual components to produce semi-quantitative mineral maps. Reference spectra of goethite and lepidocrocite for CLS calculations were extracted from within the data set. To limit the interference by broad fluorescence, the background was removed from all spectra prior to analysis. Data analysis, including CLS calculations, and instrument control were performed using Project Four software and WITec Control Four, respectively.

4.2.5. Transmission electron microscopy Rock samples, prepared for transmission electron microscopy using the focused ion beam scanning electron microscopy, were examined using a FEI Technai F20 field emission scanning transmission electron microscope (FEG-S/TEM) operated in bright field mode at 200 kV.

4.3. Results Scanning electron microscopy revealed the extraordinary preservation of cell capsule structures and potentially extracellular polymeric substances (EPS; Fig. 4.1). A 3D reconstruction of the fossilised biofilm (Fig. 4.2) highlights that as minerals continue to precipitate in association with the microorganisms, preserving the encrusted cell envelope structures. This process occurs in generations; secondary iron oxide minerals infill the extracellular regions between preserved cell envelopes as new microorganisms are fossilised within the pore space. As the microorganisms are continuously fossilised, the pore space of the fracture is infilled (Video 4.1).

Biosignature characterisation

Fig. 4.1. Backscattered electron micrographs highlight the preservation of the cell envelopes structures, possibly including extracellular polymeric substances.

81 Chapter Four

Fig. 4.2. (A) Backscattered electron micrograph of fossilised biofilm from which 3D reconstruction was created (see Video 4.1). (B) 3D orthogonal sections highlighting the preservation of mineralised cell envelopes during cell replication.

Biosignature characterisation

Video 4.1. 3D reconstruction of mineral encrusted cell envelopes that have been preserved and infilled pore spaces within a secondary goethite-rich vein that crosscuts the saprolite of banded iron formations in the Quadrilátero Ferrífero, Minas Gerais, Brazil. Video available at: https://drive.google.com/drive/folders/1Y_nRv6xOtafPC369UtNdVlwExK-XIzk2?usp=sharing

NanoSIMS analysis revealed preserved organic carbon and nitrogen associated with the cell envelope structures of microfossils (Fig. 4.3). Synchrotron-based FT-IR analysis highlighted the enrichment of organic biosignatures associated with regions containing microfossils, but it was not able to resolve individual cell envelope structures (Fig. 4.4). The preserved organic biomarkers appeared to be aliphatic methylene moieties (CH2), highlighted by the bands at 2857 and 2928 cm-1 (Fig. 4.4), corresponding to the antisymmetric and symmetric stretching

83 Chapter Four of saturated hydrocarbons, respectively. Microfossils were preserved within an iron-rich matrix surrounding gibbsite grains, highlighted by bands at 3379, 3448, 3528, 3621 cm-1 (Fig. 4.4).

Fig. 4.3. (A) Backscattered scanning electron micrograph of FIB lamella highlighting the region mapped using NanoSIMS (white square). (B) Composite image highlighting the relative distributions of carbon (red), nitrogen (green) and iron (blue). Carbon (C), nitrogen (D), aluminium oxide (E) and iron oxide (F) NanoSIMS elemental micrographs highlighting the preservation of carbon and nitrogen associated with the cell envelope structures.

Biosignature characterisation

Fig. 4.4. (A) Backscattered scanning electron micrograph of FIB lamella highlighting the region mapped using synchrotron-based Fourier transform infrared spectroscopy (FT-IR). (B) Contoured heatmap generated by integrating all spectra between 2910 and 2940 cm-1, corresponding to the antisymmetric stretch of aliphatic moieties (CH2). Infrared spectrum of a region containing methylene moieties (blue spectrum), associated with the microfossil-rich regions of the lamella. Red spectrum highlights the minerals around microfossils are gibbsite- rich.

85 Chapter Four

Raman spectroscopy revealed both goethite [훼-FeOOH] and minor lepidocrocite [훾-FeOOH] precipitating around cell envelopes, with lepidocrocite minerals forming relatively pure phases in the mineralised regions between microfossils (Fig. 4.5). Transmission electron microscopy indicated that there was no clear mineralogical difference between iron oxide minerals that precipitated in association with the cell envelopes structures and those that precipitated within the matrix (Fig. 4.6). A transmission-based microXRD experiment conducted at Advanced Light Source (Microdiffraction Beamline 12.3.2) also did not reveal a clear mineralogical biosignature associated with the iron oxide encrusted microfossils (data not shown).

Biosignature characterisation

Fig. 4.5. (A) Backscattered scanning electron micrograph highlighting region mapped using Raman spectroscopy, which revealed iron oxide minerals associated with microfossils are minor lepidocrocite (B) and goethite (C). Lepidocrocite is enriched within the extracellular regions between fossilised cell envelopes.

87 Chapter Four

Transmission electron microscopy of microfossils revealed that biologically influenced mineralisation (BIFM) of the cell envelope appears to follow the direction of the cell envelope structure (Figs. 4.6 – 4.7). In contrast, iron oxide minerals that precipitate in the intracellular regions grow towards the centre of the remnant cell, permineralising the fossil (Fig. 4.6). These post-fossilisation mineral precipitates appear to be larger than the iron oxide minerals associated with the cell envelope (Fig. 4.6).

There was no clear mineralogical difference between iron oxides that had precipitated in association with the cell envelopes structures and those that has precipitated within the matrix of the sample (Fig. 4.6). Raman spectroscopy revealed both goethite [훼-FeOOH] and minor lepidocrocite [훾-FeOOH] precipitating around cell envelopes, with lepidocrocite minerals forming relatively pure phases in the mineralised regions between microfossils (Fig. 4.7).

Biosignature characterisation

Fig. 4.6. (A) Bright field transmission electron micrograph of a TEM lamella containing a mineralised cell envelope. Selected area electron diffraction patterns could not conclusively distinguish intracellular minerals (B), biologically-influenced mineralisation associated with the cell envelope structure (C) and minerals within the matrix between fossilised cell envelopes.

89 Chapter Four

4.4. Discussion Here, we present abundant mineralised cell envelope structures fossilised within a goethite- rich vein structure that cross-cuts a weathering BIF (Levett et al., 2019a). These indisputable microbial fossils provide a superb opportunity to understand potential mineralogical biosignatures preserved during microbial fossilisation by authigenic minerals. Rock fissures, fractures and pore spaces present ideal locations for the preservation of microorganisms as elements in solution percolate throughout the weathering profile. The multivalent oxidation states of iron and its low solubility in circumneutral pH environments make iron-rich regions

Biosignature characterisation an attractive target for microbial fossilisation in near-surface environments, both on Earth and potentially Mars.

Cell envelope structures reduce the stability of cations in solution, altering the mineral phases that would otherwise have precipitated abiotically. If these biominerals resist recrystallisation, unique mineral phase distributions may be preserved as mineralogical biosignatures (Banfield et al., 2001; Li et al., 2013). In this study, crystal orientation and crystal size of the microbially- influenced iron oxide minerals appears to be altered by the microbial cell envelopes (Figs. 4.1 and 4.2) but no clear mineralogical biosignatures were preserved (Figs. 4.6 and 4.7). Although characterisation techniques used here are not exhaustive, there was no clear mineralogical difference between iron oxides that precipitated around the cell envelope and neighbouring iron oxide minerals in the matrix (Figs. 4.6 – 4.7), despite the apparent enrichment of aluminium associated with the cell envelope (Fig. 4.4). Therefore, the organic-mineral complexes that resist recrystallisation during the initial stages of mineral formation (Banfield et al., 2000) do not appear to be preserved in million-year-old fossils.

In this study, mineral precipitates associated with cell envelope structures are extremely fine- grained compared with the post-death intracellular mineral precipitates (Fig. 4.6 – 4.7). Cations binding to active sites on the cell envelope (Fein et al., 1997) appears to create multiple mineral nucleation sites, restricting crystal growth. In contrast, post-death mineral precipitates within the intracellular voids have fewer mineral nucleation sites and, therefore, are allowed to grow in a less restricted manner. Consistent with Cosmidis et al. (2013), intracellular mineral precipitates always grow from the cell envelope inward to fill the intracellular void.

The 3D characterisation of iron oxide encrusted microbial fossils, provides an important opportunity to produce orientated reconstructions useful in the search for fossilised bacteria or biofilm (Video 4.1). Though FIB-SEM was used in this study, synchrotron-based nanotomography now offers non-destructive 3D reconstructions with submicron spatial resolution (Müller et al., 2018). These technical developments offer unparalleled opportunities to understand the mechanisms that contribute to microbial fossilisation. The 100 nm resolution of the 3D reconstruction produced for multiple sections of each microbial fossil. The microfossils examined here are typically cocci-shaped and approximate 1 m in diameter. Sarcina-like multicellular packet structures (for example, see Fig. 6.6) are never observed in the fossilised biofilm characterised here; however, rarely, paired cells are preserved that share a cytoplasm in a single section (Fig. 4.2). These textures provide evidence that the fossilised

91 Chapter Four microorganisms presented here may be able to replicate during the initial stages of biomineralisation as has been previously postulated (Benzerara et al., 2011; Phoenix and Konhauser, 2008; Phoenix et al., 2000).

Microorganisms fossilised by authigenic mineral nucleation, rather than the binding of sediments within the biofilm (Newman et al., 2016), provide valuable insights into the environmental conditions in which the living microorganism existed. These biogenic minerals may also provide information on influence of the microorganisms on their surrounding environment. Previous experiments have demonstrated the precipitation of lepidocrocite associated with neutrophilic iron-oxidising microorganisms (Chan et al., 2011) and nitrate- dependent iron-oxidising bacteria (Larese-Casanova et al., 2010). Given the microfossil structures in this study are generally cocci-shaped, they are unlikely to represent sheath structures of classic Leptothrix-type neutrophilic iron-oxidising bacteria (Emerson and Ghiorse, 1992), though filaments do exist (see Video 4.1; 38 – 40 s, top right-hand corner). In addition, the low nitrogen availability in these environments (Messias et al., 2013) suggests that microfossils are also unlikely to represent nitrate-dependant iron oxidisers. The apparent binding of aluminium with cell envelope structures (Levett et al., 2019a), indicates that these microorganisms were likely to have been preserved by the passive nucleation of minerals on the cells’ surfaces (Ferris et al., 1988; Li et al., 2013). The formation of lepidocrocite together with goethite (Fig. 4.5) suggests that the pH was between 5 – 7. Lepidocrocite forms preferentially to goethite under slightly slower oxidation rates of iron (Cornell and Schwertmann, 2003), indicating reduced partial pressures of oxygen in pore spaces below the surface compared with atmospheric conditions.

In this BIF weathering profile, many cells contribute to mineral nucleation by the passive interaction of cations with the net negative cell envelope (Ferris et al., 1988). Even amongst the cells that contribute to biomineralisation (Levett et al., 2019b), few are likely to achieve a state of ‘fossilisation’, whereby they are preserved in the geological record. Extensive mineralisation is required to achieve microbial fossilisation and preservation. Based on earlier findings (Levett et al., 2019a), aluminium binding irreversibly with cell envelope structures appears to play an important role in the preservation of organic biosignatures; however, synchrotron-based FT-IR analysis could not resolve aluminium-organic complexes.

The preservation of microorganisms in the geological record is rare. As such, an abundance of well-preserved microfossils in any environment requires careful consideration. To resist the

Biosignature characterisation breakdown of cellular components, particularly cell envelope structures and potentially EPS, rapid and extensive mineralisation is required. The influence of water in microbial fossilisation within the lithosphere is also likely to be critical. While fine-grained, generally amorphous iron oxide precipitates readily nucleate on cell envelope structures in iron-rich aqueous environments (Ferris et al., 1988), for example, 2-line ferrihydrite (Kennedy et al., 2004); these cells are unlikely to be fossilised within water saturated environments. Therefore, following this initial stage of biomineralisation during a cells exposure to cation-rich solutions, periods of drying appear to be imperative for fossilisation. During dehydration, any remaining ions in solution (in this case, predominately iron and aluminium), would be concentrated, accelerating additional mineral nucleation on the cells’ surfaces. Alternating wet and dry periods may be required to promote additional mineralisation. In this scenario, additional metals in solution would be provided during wet periods, which may allow for the recrystallisation of existing iron oxide minerals and the additional precipitation of new iron oxide minerals (Legrand et al., 2004). During drying periods, newly mineralised microorganisms may be committed to the geological record, contributing to preservation of relatively large microfossil clusters.

The organic compounds associated with the cell envelopes are likely to be preserved by the electrostatic-driven nucleation of aluminium and iron oxide minerals within relatively oxidising environments (Ferris et al., 1988; Levett et al., 2019a; Monteiro et al., 2018). As the microfossils are continuously exposed to aluminium and iron-rich solutions, the mineralised cell envelopes appears to act as a filter; iron is allowed into the cell whereas aluminium is enriched around the cell envelope (Levett et al., 2019a; Londono et al., 2017). The structure of the cell envelop appears to restrict aluminium transfer into the cell (Londono et al., 2017), possibly even after cell death. Aluminium may also continue to be enriched around the cell envelope as it preferentially precipitates with existing aluminium-substituted iron oxide minerals that have previously nucleated on the cells’ surfaces (Cornell and Schwertmann, 2003). In this way, even after all the organic components of the cell envelope are replaced and the cell has been completely permineralised, aluminium enrichment around the cells may help to preserve the bacteriomorphic structure within the geologic record (Levett et al., 2016; Levett et al., 2019a).

Appropriate sample preparation for high-resolution analytical work is critical. Many techniques require a polished surface to spatially resolve distinctions between minerals influenced by microorganisms compared with ‘abiotic’ mineral precipitates. To study microfossils, sample preparation using a FIB-SEM offers a number of benefits including,

93 Chapter Four targeted preparation of localised regions of interest without introducing organic contaminants. As a destructive sample preparation technique, great care and skill is required when preparing microfossil lamella using a FIB-SEM; however, this sample preparation technique is highly versatile (Heaney et al., 2001). Samples can be made thin enough to be analysed using transmission X-ray and infrared sources but also robust enough for NanoSIMS, a destructive secondary ion technique. Therefore, FIB-SEM sample preparation allows for highly targeted, polished sample preparation and for correlation between several different analytical datasets, as demonstrated in this study. Ultrathin samples (~100 nm thick) can also be prepared for high- resolution transmission electron microscopy (HR-TEM) and scanning transmission X-ray microscopy.

The redistribution of elements that contribute to near-surface microbial fossilisation in rock pore spaces and fissures is fundamental to targeting drill regions for the identification of microfossils on samples from Mars. Organic biosignatures and abundant microbial fossils preserved in iron-rich environments highlights the potential to target iron-rich regions on the Martian surface for the search of potential microbial biosignatures. In depth characterisation of indisputable microbial fossils combining a suite of nano- and microscale analytical techniques sets important benchmarks for the identification of biosignatures within the geological record. Additional studies of natural microbial fossils in a variety of environments is required to understand potential biosignatures preserved in different environments and aim to develop new robust biosignatures, for example, mineralogical biosignatures.

4.5. Conclusions Iron oxide microfossils, preserved in near surface environments by the extensive mineralisation of the cell envelope structures, prevents the degradation of the originally soft, organic biosignatures. A remarkable variety of microfossil textures are preserved. Following cell envelope mineralisation, continued exposure to iron-rich solutions results in the permineralisation of microfossils, as iron oxide minerals grow inwards to infill intercellular voids. Microorganisms appear to be able to metabolise and replicate even while partially mineralised. As microorganisms are fossilised, they can infill pore and fissure spaces within near surface rocks. The creation of massive iron oxide plateaus during fossilisation of these bacteria and the development of robust biosignatures in these fossils provides a potentially important geomorphological target for the search for life, or remnants of life (fossils) on Mars.

Biosignature characterisation

Acknowledgements We acknowledge support from the Vale S.A.-UQ Geomicrobiology initiative and the Australian Research Council Linkage Program (LP140100805) to G. Southam and P. Vasconcelos. This research was undertaken on the IRM beamline at the Australian Synchrotron, part of ANSTO. We thank Mark Tobin and Pimm Vongsvivut for assisting with XFM and XANES data collection at the Australian Synchrotron. The authors acknowledge the facilities and the scientific and technical assistance of the Australian Microscopy and Microanalysis Research Facility at the Centre of Microscopy and Microanalysis, at the University of Queensland. We thank Graeme Auchterlonie and Han Gao for assistance operating the transmission electron microscope. We wish to acknowledge the assistance of the staff of Central Analytical Research Facility (CARF) at Queensland University of Technology (QUT) for access to analytical instrumentation, supported by the Faculty of Science and Engineering at QUT. Alan Levett acknowledges the support from the Australian Government Research Training Program.

References Banfield JF, Moreau JW, Chan CS, Welch SA, Little B. Mineralogical biosignatures and the search for life on Mars. Astrobiology 2001; 1: 447-465.

Banfield JF, Welch SA, Zhang H, Ebert TT, Penn RL. Aggregation-based crystal growth and microstructure development in natural iron oxyhydroxide biomineralization products. Science 2000; 289: 751-754.

Chan CS, Fakra SC, Emerson D, Fleming EJ, Edwards KJ. Lithotrophic iron-oxidizing bacteria produce organic stalks to control mineral growth: implications for biosignature formation. International Society for Microbial Ecology Journal 2011; 5: 717-727.

Cornell RM, Schwertmann U. The iron oxides: structure, properties, reactions, occurrences and uses. Weinheim, Germany: John Wiley & Sons, 2003.

Davila AF, Fairén AG, Schulze-Makuch D, McKay CP. The ALH84001 case for life on Mars. In: Seckbach J, Walsh M, editors. From fossils to astrobiology: records of life on Earth and search for extraterrestrial biosignatures. Springer Netherlands, Dordrecht, 2008, pp. 471-489.

95 Chapter Four

Emerson D, Ghiorse WC. Isolation, cultural maintenance, and taxonomy of a sheath-forming strain of Leptothrix discophora and characterization of manganese-oxidizing activity associated with the sheath. Appl. Environ. Microbiol. 1992; 58: 4001-4010.

Fein JB, Daughney CJ, Yee N, Davis TA. A chemical equilibrium model for metal adsorption onto bacterial surfaces. Geochimica et Cosmochimica Acta 1997; 61: 3319-3328.

Ferris FG, Fyfe WS, Beveridge TJ. Metallic ion binding by Bacillus subtilis: implications for the fossilization of microorganisms. Geology 1988; 16: 149-152.

Hoppe P, Cohen S, Meibom A. NanoSIMS: technical aspects and applications in cosmochemistry and biological geochemistry. Geostandards and Geoanalytical Research 2013; 37: 111-154.

Kennedy C, Scott S, Ferris F. Hydrothermal phase stabilization of 2-line ferrihydrite by bacteria. Chemical Geology 2004; 212: 269-277.

Larese-Casanova P, Haderlein SB, Kappler A. Biomineralization of lepidocrocite and goethite by nitrate-reducing Fe(II)-oxidizing bacteria: effect of pH, bicarbonate, phosphate, and humic acids. Geochimica et Cosmochimica Acta 2010; 74: 3721-3734.

Legrand L, Mazerolles L, Chaussé A. The oxidation of carbonate green rust into ferric phases: solid-state reaction or transformation via solution. Geochimica et Cosmochimica Acta 2004; 68: 3497-3507.

Levett A, Gagen E, Shuster J, Rintoul L, Tobin M, Vongsvivut J, et al. Evidence of biogeochemical processes in iron duricrust formation. Journal of South American Earth Sciences 2016; 71: 131-142.

Levett A, Gagen EJ, Diao H, Guagliardo P, Rintoul L, Paz A, et al. The role of aluminium in the preservation of microbial biosignatures. Geoscience Frontiers 2019a; 10: 1125- 1138.

Levett A, Gagen EJ, Southam G. Small but mighty: microorganisms offer inspiration for mine remediation and waste stabilisation. Microbiology Australia 2019b; 40: 190-194.

Li J, Benzerara K, Bernard S, Beyssac O. The link between biomineralization and fossilization of bacteria: insights from field and experimental studies. Chemical Geology 2013; 359: 49-69.

Londono SC, Hartnett HE, Williams LB. Antibacterial activity of aluminum in clay from the Colombian Amazon. Environmental Science & Technology 2017; 51: 2401-2408.

McKay DS, Gibson EK, Thomas-Keprta KL, Vali H, Romanek CS, Clemett SJ, et al. Search for past life on Mars: possible relic biogenic activity in Martian meteorite ALH84001. Science 1996; 273: 924-930.

Biosignature characterisation

McMahon S, Bosak T, Grotzinger J, Milliken R, Summons R, Daye M, et al. A field guide to finding fossils on Mars. Journal of Geophysical Research: Planets 2018; 123: 1012- 1040.

Messias M, Leite M, Meira Neto J, Kozovits A, Tavares R. Soil-vegetation relationship in quartzitic and ferruginous Brazilian rocky outcrops. Folia Geobotanica 2013; 48: 509- 521.

Müller S, Pietsch P, Brandt B-E, Baade P, De Andrade V, De Carlo F, et al. Quantification and modeling of mechanical degradation in lithium-ion batteries based on nanoscale imaging. Nature communications 2018; 9: 2340.

Newman SA, Mariotti G, Pruss S, Bosak T. Insights into cyanobacterial fossilization in Ediacaran siliciclastic environments. Geology 2016; 44: 579-582.

Phoenix V, Konhauser K. Benefits of bacterial biomineralization. Geobiology 2008; 6: 303- 308.

Phoenix VR, Adams DG, Konhauser KO. Cyanobacterial viability during hydrothermal biomineralisation. Chemical Geology 2000; 169: 329-338.

Westall F. The nature of fossil bacteria: a guide to the search for extraterrestrial life. J. Geophys. Res. Planets 1999; 104: 16437-16451.

Chapter 5 Microbial weathering signatures in lateritic ferruginous duricrusts

Alan Levetta, Paulo M. Vasconcelosa, Emma J. Gagena, Llew Rintoulb, Carlos Spiera, Paul Guagliardob, Gordon Southama

a School of Earth and Environmental Sciences, University of Queensland, Brisbane, Queensland 4072, Australia

b School of Chemistry, Physics and Mechanical Engineering, Queensland University of Technology, Brisbane, Queensland 4001, Australia

c Centre for Microscopy, Characterisation and Analysis, University of Western Australia, Perth 6009, Western Australia, Australia

Citation for this publication:

Chapter Five

Abstract Iron-rich duricrusts (canga) that blanket iron ore deposits in tropical regions in Brazil have a complex formation and evolution. The direct role of microorganisms in the biogeochemical cycling of iron and remarkably, aluminium, which is critical for the preservation of these duricrusts, is becoming more apparent. In this study we combine nanoscale secondary ion mass spectrometry (NanoSIMS) with Raman spectroscopy and scanning electron microscopy to distinguish highly weathered and recycled grains, portions of which have been metasomatised by iron in solution, from new mineral precipitates within canga. Microbially-accelerated weathering of grains within canga appears to promote iron, aluminium and titanium oxide mineral dissolution. Anatase forms at the boundaries of rutile grains, highlighting titanium dissolution and re-precipitation at the mineral-scale. Titanium is absent from the authigenic cements in canga. Aqueous ferrous iron, which is sensitive to changes in the oxidation potential of the solution, readily oxidises and re-precipitates in the pore spaces in proximity to preserved microorganisms, leaving aluminium as the most abundant cation in solution. Changes in solution pH and/or organic chemistry causes aluminium to precipitate from solution as gibbsite that infills pore spaces as the final geological texture. Evidence for microbially-promoted weathering includes the fossilisation of microorganisms along grain boundaries, in which aluminium, titanium, chromium and iron are depleted. Immobile elements including titanium, chromium and thorium, associated with rounded fossilised biofilms also indicate that microorganisms have completely weathered grains and become fossilised in place, texturally replacing the grain in the process. Microorganisms also play a role in iron and aluminium precipitation, providing sites for mineral nucleation. These results highlight the microscale changes in oxidation potential, pH and organic chemistry throughout canga that are likely to be influenced by the microbiome and associated flora. Understanding the role of microorganisms in iron biogeochemical cycling will contribute to the re-cementation of iron- rich duricrusts after the completion of iron ore mining, providing a substrate for revegetation using native plant species. This study also sheds light on direct (microfossils) and indirect (weathering) biosignatures in iron-rich geological materials.

99 Geochemical signatures of microbial weathering

5.1. Introduction Canga is a goethite-cemented breccia that forms an iron-rich duricrust that commonly caps iron ore deposits in Brazil. Composed primarily of iron oxide minerals that are relatively resistant to weathering, the formation and evolution of canga is strongly influenced by climatic controls, topography, nature of the source rock as well as the native flora and microbiome (Levett et al., 2016; Monteiro et al., 2018a; Monteiro et al., 2014; Monteiro et al., 2018b; Shuster et al., 2012; Spier et al., 2018).

The redox cycling of iron within canga is pivotal to the extremely low erosion rates of canga, helping to create one of the longest-lived exposed surfaces on Earth (Monteiro et al., 2018b). Iron(III) oxide minerals are locally dissolved placing Fe(II) into solution, which is generally unstable in these oxic and circumneutral environments, oxidising and re-precipitating to form new iron oxide cements that aid the stabilisation of iron-rich material in the canga. In contrast, more mobile elements that do not readily undergo redox cycling (for example, silicon) tend to remain in solution and are lost through weathering. Therefore the redox cycling of iron and the distinct differences between the solubility of Fe(II) and Fe(III) appear to be critical to the long- term preservation of canga.

Geochronological and geochemical evidence indicates that goethite in surface canga has undergone repeated dissolution and re-precipitation, suggesting that plants and the rhizospheric microbiome promote iron oxide dissolution near the surface. For example, the iron oxide minerals in canga have undergone many generations of recycling over the past 50 Ma, contributing to the textural complexity of canga (Monteiro et al., 2014). Greater than expected 3He isotopes have also been identified at depths of up to 10 m throughout the canga profiles in the Carajás, indicating the transport of material across the canga profile over million year timescales, which is attributed to bioturbation from burrowing terminates (Shuster et al., 2012). Acquisition of phosphorus is one reason that plants and the microbiome associated with canga may be significant contributors to the dissolution of goethite. Plants are known to secrete organic acids to release and uptake phosphorus incorporated into iron oxides, thereby promoting mineral dissolution (Neumann and Römheld, 1999). Minor impurities, including Al, Ti, V, Cr, Mn, Co, Ni, Cu, Zn and Pb adsorbed to goethite or incorporated into their structure may also be released to weathering solutions during this process (Monteiro et al., 2018b).

The canga plateaus in the Serra Sul are extremely resistant to erosion, dropping less than 30 m over the last 80 Ma (Monteiro et al., 2018b). The focus of this manuscript is to identify

100

Chapter Five microbial signatures of weathering and precipitation that occur in situ since the deposition and development of the ferruginous duricrust. The microbially-accelerated weathering within the ferruginous duricrusts is essential for their continuous evolution, dissolving typically poorly soluble minerals in circumneutral environments. The low solubility of the elements within these minerals causes them to re-precipitate within the duricrusts, forming new cements that are critical for canga erosion resistance (Monteiro et al., 2014). These processes are essential for the preservation of the massive iron ore deposits in Brazil, and possibly Australia (Gagen et al., 2019).

Direct evidence of microorganisms and plants contributing to mineral weathering and dissolution in canga is limited. Fossilised microorganisms, referred to as microfossils, have previously been identified in surface samples of canga (Levett et al., 2016) and within goethite- cemented veins that crosscut the iron-enriched saprolite (Levett et al., 2018) of highly weathered banded iron formations (BIFs). In each of these cases, the hand samples were specifically selected based on their macroscopic biological and geological textures, including an association with plant roots and goethite bands that suggested the flow of iron in solution (Levett et al., 2016; Levett et al., 2018).

In the present study, a canga drill core from the southern region (Serra Sul) of the Carajás mineral province in Brazil was subsampled and examined to determine the role of microorganisms in the weathering and the precipitation of iron oxide minerals. Distinguishing the geochemical signatures of microbially-accelerated and chemically driven mineral weathering is critical to establish the biological processes that contribute to canga evolution. Understanding these mechanisms will be important as post-mining rehabilitation efforts in iron ore provinces are focused on promoting the re-formation of canga crusts after the completion of iron ore mining (Gagen et al., 2019; Gagen et al., 2018; Levett et al., 2016). The present work is also fundamental for the development of indirect (biogeochemical weathering signatures) and direct (preservation of microfossils) biosignatures.

5.2. Geological and environmental setting The Carajás Mineral Province is located along the eastern margin of the Southern Amazon Craton, Pará, Brazil. The region is of significant geological importance, containing world class iron and manganese deposits, as well as Cu-Au and Au-PGE mineralisation (Grainger et al., 2008). The southern aspect of the Carajás Mineral Province (Serra Sul de Carajás; Fig. 5.1) contains the world’s largest iron ore deposits (S11D), with projected exports of up to 90 million

101 Geochemical signatures of microbial weathering tonnes of high-grade iron ore per year. Alongside these world class iron ore deposits, biological studies have revealed the regions ecological importance (Nunes et al., 2015). The canga outcrops host predominately herbaceous and shrubby plants which are juxtaposed by the surrounding montane forests (Fig. 5.1B). The flora associated with canga outcrops are highly diverse. For example, Mota et al. (2018) reported more than 850 seed-producing plant species associated with canga in Carajás including endemic species, with 545 species present in the Serra Sul region. Canga provides a substrate for this highly specialised ecosystem that has evolved along with the duricrusts to survive in these relatively harsh environments. The Serra Sul de Carajás also contains karstic lakes (Sahoo et al., 2016) and cave formations (Piló et al., 2015), which each contribute to the flora and fauna diversity. To access the high-grade iron ore, the ferruginous duricrust must be removed, destroying the naturally rare biota associated with these formations. Worldwide, the ecosystems associated with these duricrusts are threatened by the scale of iron ore mining (Gibson et al., 2010; Jacobi et al., 2007). Efforts to reinstate these ecosystems after the completion of mining require an understanding of the natural processes that contribute to the formation and ongoing evolution of these ferruginous duricrusts.

A geological map of the Serra Sul de Carajás is presented in Sahoo et al. (2016) and a cross- section of the region is provided in Monteiro et al. (2018b). Briefly, the Archean Xingu Complex forms the basement of the Serra Sul de Carajás and is primarily composed of granitoids and gneisses. Rocks from the Itacaiúnas Supergroup [2.73 – 2.84 Ga] and the Águas Claras Formation [~2.64 – 2.68 Ga] make up the Serra Sul de Carajás. The Itacaiúnas Supergroup is composed of mafic to intermediate low-grade metavolcanic rocks (Grão-Pará Group) with mafic sills (Parauapebas and Cigarra Formations) and banded iron formations (Carajás Formation). Towards the east of the Serra Sul de Carajás, these units are overlain by the Águas Claras Formation, which is composed of fluvial sedimentary rocks (Grainger et al., 2008; Machado et al., 1991; Trendall et al., 1998).

The Carajás region has a tropical climate. The average daily-high temperature typically ranges from 26 – 28 ºC with monsoonal rainfall precipitating in the wet season (November to May) (Sahoo et al., 2016). Typical annual rainfalls of 1800 – 2300 mm have contributed to the development of extensive laterites, fundamental to the iron enrichment that has produced some of the world’s largest iron ore reserves. The ferruginous duricrusts that cap these iron ore deposits have some of the lowest erosion rates of any continuously exposed surface (Monteiro et al., 2018a; Monteiro et al., 2018b; Shuster et al., 2012).

102

Chapter Five

Fig. 5.1. Geographical context of the study site in the Serra Sul de Carajás Mineral Province, Pará, Brazil. (A) Digital elevation model of the Carajás region generated using the Shuttle Radar Topography Mission (SRTM) from the USGS Earth Explorer (https://earthexplorer.usgs.gov/). (B) Google Earth image of the Serra Sul canga plateau and the drill core location. (C) Remaining portion of the drill core sample after a billet (D) was prepared to make thin section samples.

103 Geochemical signatures of microbial weathering

5.3. Methods and materials 5.3.1. Sample collection and bulk characterisation A canga drill core collected by the Vale exploration team in the Carajás Mineral Province, Pará, Brazil was subsampled (Fig. 5.1C and D). The vertical drill core was collected in the Serra Sul in the Carajás Mineral Province (coordinate: 6.40039117 ºS, 50.39797396 ºW). The subsample was collected from a depth of approximately 4 m, was approximately 12 cm in length and well- consolidated. The sample contained detrital hematite fragments, typically less than 2 cm in length, consolidated by complex secondary, iron-rich cements (Fig. 5.1C).

The chemical composition of a representative sample, including cement and fragmented material, was determined at the Australian Laboratory Services (Analytical Geochemistry) using the iron ore XRF fusion method (ME-XRF21n). Briefly, the sample, was pulverised to less than 70 microns using a ring-and-puck mill before being fused with a lithium tetraborate:lithium metaborate (12:22) flux, which included lithium nitrate as an oxidising agent. Following fusion, the samples were analysed in a platinum mould with a subsample taken for measuring the loss on ignition volatiles at 1000 °C.

5.3.2. Microscale sample characterisation To characterise the microstructure of the canga, a polished petrographic thin section was prepared from the drill core sample and examined using a JEOL7100 Field Emission scanning electron microscope (FE-SEM) in backscattered electron mode at an accelerating voltage of 15 kV. Prior to examination, sample surfaces were cleaned using a XEI Scientific Evactron 25 Decontaminator RF Plasma Cleaning System, degassed at 50 °C for approximately 12 hours, and then coated with 10 nm iridium using a Quorum Q150T sputter coater. A photograph of the thin section and regions analysed is provided in Appendix 1.

Regions of interest identified using the SEM were mapped using the CAMECA NanoSIMS 50L to produce high-resolution semi-quantitative elemental maps. A Hyperion (H200) RF plasma O- source was used for all analyses. Regions of interest were mapped by rastering the beam over fields of view of 50 × 50 μm at a resolution of 512 × 512 pixels. Prior to imaging, each area was presputtered with the primary beam to a dose of more than 1 × 1017 ions·cm-2.

Each region was imaged twice: first with a beam of 15 pA and a dwell time of 7 ms per pixel to target the major elements of 27Al and 56Fe. To improve sensitivity, the beam current was then increased to 250 pA to target minor/trace elements, including 28Si, 48Ti, 52Cr, 63Cu, 248ThO and 254UO. For minor and trace elements, eight planes were recorded, each with a dwell time

104

Chapter Five of 6 ms per pixel. Increasing the beam current enhances sensitivity at the cost of spatial resolution; for example, compare the aluminium and chromium NanoSIMS maps. Individual planes were aligned and stacked using the OPEN MIMS plugin in the FIJI software package to produce elemental maps (Schindelin et al., 2012). Individual elemental micrographs are displayed as relative intensities, where white areas have a higher relative elemental concentration compared with darker regions (Hoppe et al., 2013). All elemental maps are available in Appendix 1, including some maps not used in this study.

Four complex geological textures throughout the canga thin section were mapped using nanoscale secondary ion mass spectrometry (NanoSIMS) to distinguish the element distribution of weathered grains and newly formed precipitates that form the cements in canga. Microfossils that had formed around grain boundaries or in close association with geological textures of interest throughout canga were targeted. The geochemical signatures for several examples of weathering and precipitation or cement formation are examined.

5.3.3. Raman spectroscopy The mineralogy of the regions mapped using NanoSIMS was subsequently determined using Raman spectroscopy. Minor translational misalignments between the region mapped using NanoSIMS (solid white square) and Raman spectroscopy (dashed white square) are noted.

Raman spectra were acquired with a WITec Alpha 300 series Raman microscope equipped with a 532 nm laser operating at less than 0.6 mW to avoid thermal transformation of the iron- rich minerals. The laser was focused with a Zeiss 50× objective of 0.7 NA, which formed a confocal sample volume approximating a cylinder of 0.5 m diameter and interacting with the top 2 – 3 m of the sample. The four areas of interest of 50 x 50 m were mapped by raster motion in X and Y increments of 1 m, with a dwell time of 16 s per spectrum. Classical Least Squares (CLS) methods were used to generate mineral maps. By treating each spectrum as a linear combination of individual constituent spectra, plus error, CLS can be used to decompose a spectrum into a set of constituent scores if reference spectra for the various components are known. A constituent score represents a semi-quantitative measure of its relative at that particular point on the sample. Mineral maps were generated by plotting the mineral score as a function of sample position. Reference spectra of hematite, gibbsite, goethite, anatase and rutile for use in the CLS calculations were identified from within the data set by visual inspection and comparison with known standards (Hanesch, 2009; Ocana et al., 1992; Ruan et al., 2001). A background was removed from the spectra prior to CLS analysis to limit the interference by

105 Geochemical signatures of microbial weathering broad fluorescence that occurred at various positions on the sample. Instrument control and data analysis including CLS calculations were performed using WITec Control Four and Project Four software. Representative examples of all minerals identified in each region are provided in Appendix 1.

5.4. Results 5.4.1 Bulk sample characterisation The bulk sample is primarily composed of hematite and goethite, supported by the very high iron content of 62.8 wt.%. Alumina (Al2O3: 3.35 wt.%), silica (SiO2: 0.43 wt.%), phosphorus

(P: 0.194 wt.%) and TiO2 (0.24 wt.%) are the major other constituents, with all other elements below 0.1 wt.%. The bulk Cr2O3 concentration is approximately 0.015 wt.%.

5.4.2. Geochemical signatures of microbial activity in canga The granular material in canga can be enriched in aluminium compared with the surrounding iron-rich cements; however, the structure of these grains is goethitic (Fig. 5.2A; white arrows). Mineralised cell envelopes, preserved around goethite grains were identified throughout the canga thin section (Fig. 5.2A). Microfossils are commonly rod-shaped and 1 – 5 µm in length, preserved by the mineralisation of the cell envelop structures (Fig. 5.2B).

Aluminium was relatively enriched in the less weathered portions of grains (Fig. 5.3B; green lines). Highly immobile elements such as titanium and chromium were the most useful for distinguishing weathered grains from new cements at the microscale. The titanium signatures indicate that the area highlighted by the blue line (Fig. 5.3C) forms part of a highly weathered grain, in which the titanium is preserved. Raman spectroscopy reveals that within the central portion of the grain, highlighted by the green lines, the titanium-rich fragments are predominately anatase (white). In contrast, there is little to no titanium within the iron-rich region highlighted by the red-line (Fig. 5.3C). Chromium is enriched within the central, less weathered portion of the grains compared with the outer rinds that have been extensively weathered (Fig. 5.3D). The Raman mineralogical maps highlight that both the new cements (Fig. 5.3A; red line) and the highly weathered portion of the large grain are goethite (blue) (Fig. 5.3F). Some hematite (red) is also present in the bands ‘above’ the grains with minor detrital anatase (Fig. 5.3F; white). A large, more euhedral rutile (yellow) clast is preserved within the outer portion of the highly weathered grain (Fig. 5.3F).

106

Chapter Five

Fig. 5.2. Low magnification backscattered electron micrograph highlight microfossils forming around grains in canga (A). The white square represents the regions mapped by NanoSIMS (see Fig. 5.3.). (B) High magnification micrograph demonstrating the influence of microorganisms on the precipitation of iron oxide minerals from supersaturated solution throughout canga. Microfossils are commonly rod-shaped and fossilised via microbially-influenced biomineralisation.

107 Geochemical signatures of microbial weathering

Fig. 5.3. Scanning electron micrograph (A) and NanoSIMS elemental micrographs (50 × 50 m) of the region highlighted by the white square in Fig. 5.2 for Al (B), Ti (C), Cr (D) and Fe (E). Aluminium is enriched with in the less weathered portions of the grains, highlighted by the green lines. Titanium and chromium are preserved in the highly weathered portion of the grain, highlighted by the blue line. Newly formed iron-rich cements (red line), including cements formed by microbial fossilisation, and are depleted in titanium and chromium. Raman spectroscopy (F) highlights the phase transformation of the titanium-rich minerals from rutile to anatase. Raman colour map: blue = goethite, red = hematite, white = anatase, yellow = rutile.

108

Chapter Five

The fossilised microbial biofilms throughout canga can be extensive, contributing to the cementation of relatively large areas (Fig. 5.4A). Various microfossil textures are preserved (Fig. 5.4B). Generational growth, evidenced by microfossil lamellae, are highlighted by the solid black arrow (Fig. 5.4B). Rod-shaped bacteriomorphic structures are preserved within the relict matrix material of the canga, with little-to-no pore space surrounding these microfossils (Fig. 5.4B: dashed black line). The white arrows highlight concentric microbial growth and fossilisation around highly weathered material within canga (Fig. 5.4B: white oval). Aluminium, titanium and chromium (Fig. 5.5A – C) were strongly enriched in the less- weathered portion of the recycled grains highlighted by the green line in Fig. 5.5A. In comparison, the authigenic iron precipitates highlighted by the red line in Fig. 5.5A, are depleted in the relatively immobile titanium, chromium and thorium. This contrasts with the preserved microbial biofilms preserved within weathered grains (Fig. 5.5C; white arrows) that have higher relative concentrations of thorium, titanium and chromium. Raman spectroscopy indicates that goethite (blue) is the predominant mineral phase with anatase (white), the only titanium-bearing mineral present in the grain (Fig. 5.5F).

109 Geochemical signatures of microbial weathering

Fig. 5.4. Low resolution BSE micrograph highlighting the continuous and relatively large areas containing fossilised microorganisms. The white square approximately represents the region mapped using NanoSIMS (see Fig. 5.5). (B) High-resolution insert showing the preservation of rod-shaped microfossils that form lamellae (black arrow) around a weathered grain (white oval).

110

Chapter Five

Fig. 5.5. NanoSIMS elemental micrographs (50 × 50 µm) of the region highlight by the white square in Fig. 5.4. Aluminium (A), titanium (B), chromium (C), and thorium (E) are relatively enriched in the grain to the left. In comparison, the iron-enriched region around this grain shows a depletion in the titanium and chromium indicating that these are newly formed iron precipitates are not part of the weathered grain. Titanium, chromium, thorium and uranium signatures are preserved within the rounded regions where microfossils are preserved. These geochemical signatures demonstrate that the microfossils have actively weathered and texturally replaced the grain, with the chemical signatures of the immobile elements partially preserved. Raman colour map (F): blue = goethite, white = anatase.

111 Geochemical signatures of microbial weathering

The preservation of rounded biofilms in the drill core sample is relatively common (Fig. 5.6). For example, two relict grains (Fig. 5.6A) appear to have been completely weathered and texturally replaced with microfossils (Fig. 5.6B). The region highlighted in Fig. 5.6C provides details of the iron cycling that has taken place. On the left-hand side, the spherical grain is surrounded by microfossils, which are preserved by mineral nucleation (aluminium and iron oxides) on the cell envelope surface (Fig. 5.6C; white arrow). In contrast, the microfossils on the right-hand side of the grain have been permineralised by the continued exposure to iron- rich solutions (Fig. 5.6C; black arrow). Microfossils have also been preserved both around and within a weathered grain (Fig. 5.6D). The microfossils form lamellae around the grains (for example, Fig. 5.4B) and the grains have weathered to the point where other microorganisms have been able to enter into the grain, where they have been fossilised in place (Fig. 5.6D). With the continuum of mineral weathering and microbial preservation (Fig. 5.6E), the grain may be completely texturally replaced by the microfossils, demonstrating that the microfossils (the final product) are relatively stable and resistant to weathering.

112

Chapter Five

Fig. 5.6. A series of field emission BSE micrographs demonstrating the preservation of microfossils that are at various stages of weathering and texturally replacing rounded material. The weathered grains all appear to have microfossil lamellae banding, which may be a result of fossilisation at different time periods. (A) An overview of two, almost circular grains (white arrows) that have been completely weathered and replaced by microfossils. (B) High magnifcation BSE micrograph demonstrating the nature of microbial fossilisation. (C) Microfossils are initially preserved around a grain. Two lamellae bands of microfossils can be seen. The outermost band may represent the first microfossils. As the grain is weathered, additional microfossil lamellae may replace the grains texture. The microfossils on the right have been permineralised, whereas the microfossils of the left are hollow mineralised cell envelopes. (D) Abundant rod-shaped microfossils forming around a highly weathered grain. Some

113 Geochemical signatures of microbial weathering microfossils have been preserved within the grain. (E) Microfossils forming around and within (high-resolution inset) a grain.

Within canga, weathered and highly recycled materials are cemented together by generational iron precipitation events (Fig. 5.7A). These weathered grains are often surrounded by layers of ferruginised biofilms (Fig. 5.7B) with further abiotic iron-rich precipitates infilling pore spaces.

In this example, aluminium (Fig. 5.8B), titanium (Fig. 5.8C), chromium (Fig. 5.8D) and iron (Fig. 5.8E) all appear to be slightly depleted in the most weathered outer portion of the grain (Fig. 5.8A; blue line) compared with the less weathered inner portion (Fig. 5.8A; green line). The ferruginised microbial biofilms are goethitic, primarily composed of iron and aluminium oxides and depleted in titanium and chromium (Figs. 5.8B – E). Hematite (red; Fig. 5.8F) precipitates have infilled some of the pore spaces. The large titanium-rich fragment is primary composed of rutile with anatase forming along the edge (Fig. 5.8F). This transformation, during weathering, highlights the dissolution of rutile, which was likely to have originated as ilmenite

[FeTiO3] from the surrounding metavolcanic units, and the re-precipitation of anatase, which precipitates at low temperatures (Dachille et al., 1968). Titanium is therefore ‘mobilised’ but rapidly redistributed into new, sedimentary-based mineral assemblages at a microscale.

114

Chapter Five

Fig. 5.7. (A) Low-magnification BSE micrograph highlighting textural complexity of canga and the region mapped by NanoSIMS in Fig. 5.8. (B) Microfossils form around weathered grains, with abiotic iron oxide (hematite) precipitates further in filling pore spaces.

115 Geochemical signatures of microbial weathering

Fig. 5.8. BSE micrograph (A) and NanoSIMS elemental micrographs (50 × 50 µm) of the region highlighted by the white square in Fig. 5.7 for Al (B), Ti (C), Cr (D) and Fe (E). At the top of the micrograph, a recombinant grain showing a classic weathering front with titanium and chromium present but depleted in the weathered portion of the grain compared with the less weathered portion. Relatively newly formed iron-rich and microbial cements are depleted in titanium and chromium. The titanium-rich fragment at the bottom of the micrographs shows a complex restructuring during weathering to form (titanium mineral name) surrounded by goethite. Minor amounts of the original mineral are also preserved between the microfossils. Raman colour map (F): blue = goethite, red = hematite, white = anatase, yellow = rutile.

116

Chapter Five

Aluminium-enriched precipitates are relatively common in the pore spaces throughout canga. The aluminium-enriched precipitates examined in this study (Figs. 5.9 – 5.10) contain evidence of permineralised microfossils within the nearby goethite cements (Fig. 5.9B; white arrows). Microbial activity does not appear to have been directly associated with the gibbsite mineral precipitation (Fig. 5.9B). Minor silicon-enrichment also occurs within the gibbsite precipitates (Fig. 5.10B), which are depleted in titanium, chromium, iron and thorium (Fig. 5.10C – F). These gibbsite precipitates highlight that aluminium, not iron, was the dominant element in solution in this region before precipitating as the final geological texture to completely infill these pore spaces.

117 Geochemical signatures of microbial weathering

Fig. 5.9. (A) Low magnification BSE micrograph showing aluminium-enrichment within pore spaces. Raman colour map (B): blue = goethite, red = hematite, yellow = gibbsite. Note: gibbsite identification is affected by grain orientation. (C) Some filamentous and rod-shaped microfossils (white arrows) are preserved within the goethite cements.

118

Chapter Five

Fig. 5.10. NanoSIMS elemental micrographs (50 × 50 µm) for the region indicated in Fig. 5.9, demonstrating aluminium (A) enrichment within some pore spaces. Silicon (B) may be slightly enriched within the aluminium-rich pore spaces, whereas titanium (C), chromium (D), iron (E) and thorium (F) are each depleted.

119 Geochemical signatures of microbial weathering

5.5. Discussion NanoSIMS mapping of microfossils within canga revealed a geochemical distinction between highly weathered grains and newly formed cements. Within the ferruginous duricrust, the most obvious geochemical signature of a relict, highly weathered and recycled grain is offered from mapping titanium and chromium, each of which are depleted in the highly weathered portions of the grains. During weathering and recycling of the grains, pore spaces develop and allow for the inclusions of detrital materials, for example, the rutile grain incorporate into the weathered portion of the grain (Fig. 5.3F). These highly weathered portions of grains are then likely to be metasomatised by iron in solution. Therefore, the iron precipitates that embed the rutile grain can be considered authigenic. These processes highlight the continued cycling within the duricrusts environment that can contribute to the formation of a single grain in canga (Monteiro et al., 2014; Sahoo et al., 2017b).

Importantly, the abundance of microfossils preserved throughout the goethite cements in this subsurface canga sample reveal an intimate role of microorganism in the evolution of these duricrusts. Though difficult to quantify, the routine identification of large microbial biofilms throughout canga indicates that substantial proportions of the canga cements can be attributed to microorganisms (for example, Fig. 5.4A).The presence of microfossils identified throughout the sample corroborates geochemical and geochronological evidence that canga evolution is strongly influence by the associated biome. It also provides strong evidence that microbial activity contributes to in situ weathering throughout the canga profile. This highlights that subsurface microorganisms are able to survive sporadically throughout the canga pore spaces, altering the geochemical conditions. The distinct microfossil textures and geochemical signatures in canga highlight that microorganisms appear to play an important role in both mineral weathering and precipitation. These processes appear to occur concurrently, possibly contributing to the weathering of the original grain and its textural replacement.

5.5.1. A model for microfossil textural replacement A revelation from the NanoSIMS elemental mapping was the preservation of immobile elements, including titanium, chromium and thorium associated with spherical-shaped fossilised biofilms (Figs. 5.4 – 5.5). Here we will explore the two main interpretations to explain these results.

In the first model, the titanium, chromium and thorium signatures are from microscopic detrital fragments that have been transported and settled in pore spaces in association with the rounded

120

Chapter Five fossilised biofilms. Such a pore space may have been created by the growth of a plant root, which can grow to depths of several meters in canga (Nunes et al. 2015). The plant root would likely have exuded organic acids to grow into the rock, supporting microbial growth. After the death of the plant, the root may have decayed and been degraded by microorganisms. In this scenario, iron and aluminium in solution flow through the connected pore space created by the root and fossilised the microorganisms. Resistant detrital fragments containing titanium, chromium and thorium are transported as microscopic fragments that settle in the pore space, producing the chemical signatures in Fig. 5.5; white arrows. If we explore this idea further, it would be expected that microscopic fragments of anatase, rutile or other resistant detrital materials, even 1 µm in diameter, would be detected by Raman spectroscopy (beam size ~ 0.5 µm); however, limited anatase was identified. Further, the rounded fossilised biofilms are approximately 30 µm in diameter, which even for plant root hairs, is extremely small. The sample was collected from 4 m below the surface and the extremely slow erosion rates (Monteiro et al., 2018b) indicate that this is unlikely to have drastically changed over the last several million years. Canga-associated plants on the upper plateau of the Serra Sul region of the Carajás are broadly characterised as herbaceous and shrubby, with larger montane forests forming in the lower transition zones, where organic matter accumulates (Nunes et al., 2015). The fragmented nature of canga may allow greater root penetration but this would be restricted to the tallest shrubs and herbaceous plants, producing the largest root systems. Finally, these titanium, chromium and thorium signatures are not detected in non-rounded fossilised biofilms (Figs. 5.3 and 5.8), where microscopic detrital fragments would also be able to settle. Therefore the geochemical signatures of highly immobile elements associated with rounded fossilised biofilms do not appear to be from transported detrital fragments.

The second (and preferred) model involves the growth of microorganisms around a highly recycled grain that has incorporated titanium and chromium, most likely from the weathering of the surrounding mafic rock units (see Section 5.5.4.). In this scenario, microbially-influenced weathering attacks the grains structure. During the microbially-influenced weathering, some of the iron and aluminium that is leached from the grain nucleates on the cell surfaces and fossilises the microorganisms. The most immobile and biologically incompatible elements (for example, titanium, chromium and thorium) are rapidly immobilised and redistributed as microscopic precipitates or mineral inclusions. No microcrystalline anatase minerals were observed as recorded with fossils in metamorphic rocks in New Zealand (Galvez et al., 2012). As weathering continues, the pore spaces in the grain become enlarged and the ‘pioneer’

121 Geochemical signatures of microbial weathering microorganisms move into the grain (Fig. 5.6E). These microorganisms continue to weather the grain, dissolving iron and aluminium, and becoming fossilised by passive mineral nucleation on the cells’ surface (Ferris et al., 1988). As this process is repeated, the grain becomes texturally replaced by microfossils, with titanium, chromium and thorium discarded as ultrafine inclusions. In support of this hypothesis, Männik et al. (2009) demonstrated that microorganisms are capable of living in spaces smaller than their own diameters and can maintain mobility when pore space diameters exceeds their own diameters by only ~30%. Therefore, microorganisms growing around the grain that promote or accelerate its dissolution may be able to enter the weathered material. The second hypothesis is also more consistent with the textures presented in Figs. 5.6D and 5.6E, where varying degrees of the original grains are still preserved.

5.5.2. Microbially-accelerated weathering signatures The fragmented material in canga has been continuously weathered, physically transported and disrupted by the monsoonal rain. Anatase and other highly resistant minerals (for example, zircon) have been incorporated into the recycled grains during this extensive geochemical cycling for periods of 10s of millions of years (Monteiro et al., 2018b). As a consequence of the continued recycling, the original lithology is rarely preserved within these grains.

The biome associated with canga in tropical regions has been proposed to accelerate the weathering of the relatively stable Fe(III) oxide minerals in canga (Levett et al., 2016; Levett et al., 2020; Monteiro et al., 2014; Parker et al., 2013; Paz et al., 2020); however, to date, no direct evidence of microbially-accelerated weathering has been demonstrated and the mechanisms remain unclear. This manuscript provides the strong, though not unequivocal, evidence that microorganisms actively alter the geochemical conditions in canga and consequently contribute to weathering throughout the canga profile. Here, we investigate the preferential textures, element mobility and mineralogical transformations that are likely to have taken place under the influence of microbial weathering, as evidenced by the preservation of large microbial biofilms throughout the canga cements.

High magnifcation NanoSIMS elemental micrographs provide evidence of microbially- accelerated weathering within canga as they likely weather and texturally replace grains. The depletion of aluminium, titanium, chromium and iron from grains surrounded by microfossils, also indicates in situ weathering, accelerated by microorganisms (Fig. 5.8). The titanium-rich fragment in Fig. 5.8 indicates that rutile appears to be weathering and redistributed at the grain

122

Chapter Five boundaries as anatase. The redistribution of titanium from rutile to anatase at the mineral-scale is consistent with previous studies that note the transformation of ilmenite and rutile to anatase in the ferruginous mottled zone and note an absence of rutile from the most intensely weathered ferruginous duricrusts (Du et al., 2012). Titanium mobility and re-distribution is likely to be influenced by organic complexation from the nearby growth of microorganisms (Neaman et al., 2005).

Surfaces waters in the Serra Sul de Carajás typically have a pH between 5 – 6 (Sahoo et al., 2016; Silva et al., 2018). In this circumneutral environment, iron oxide dissolution is restricted to suboxic and anoxic niches created by organic carbon degradation (Levett et al., 2020). During iron reduction, which is stimulated during the wet seasons in the Carajás (Levett et al., 2020), aluminium is also placed into solution (Sahoo et al., 2017a). Iron and aluminium dissolution in these microscale niches throughout canga play an important role in the continued formation of secondary cements and, thus, the stability of canga horizons (Levett et al., 2020; Monteiro et al., 2014).

Fossilised microorganisms at the interface of weathered minerals (Fig. 5.8), the apparent textural replacement of grains by microfossils (Fig. 5.5) and the redistribution of rutile to anatase provide strong evidence that microorganisms actively contribute to weathering throughout canga. These microbially-accelerated weathering signatures are important for the development of indirect biosignatures, where microfossils are not preserved.

Consistent with Monteiro (2017), uranium concentrations were relatively depleted compared with thorium signal in the surface duricrust sample. For all regions associated with microfossils that were mapped here, the uranium signal was predominately noise Appndix ). In contrast, the thorium in the ferruginous duricrust samples appears to be virtually insoluble and maintained in the highly weathered relict grains (Fig. 5.5). For a profile in the Igarapé Bahia gold deposit, uranium in the surface duricrust was 12.2 ppm compared with thorium concentrations of 23.9 ppm (Monteiro, 2017). In contrast, at depths of ~ 80 m, the uranium concentrations were strongly enriched (150 – 300 ppm) compared with thorium (0.01 – 0.03 ppm), indicating the 2+ uranium is oxidised and transported as 푈푂4 (푎푞) in solution and re-precipitates at depth 2+ (Monteiro, 2017). The complexation of 푈푂4 (푎푞) by organic matter is likely to assist in the transportation on uranium to depth throughout these profiles, with inorganic ions (for example, carbonates and phosphate complexes) not readily available (Ganesh et al., 1997).

123 Geochemical signatures of microbial weathering

5.5.3. Mineral nucleation: microbial influences and abiotic precipitates The iron, titanium and aluminium that are placed into solution in canga are not transported far. The titanium present in the less weathered (inner) portion of grains as rutile is dissolved and re-precipitated as anatase along the exposed surfaces of the titanium-rich grain boundaries (Fig. 5.8). Of the iron and aluminium leached from grains and placed into solution, some appears to re-precipitate on nearby cell surfaces, fossilising at least some of the microorganisms responsible for weathering (Fig. 5.6). Of the dominant cations remaining in solution (iron and aluminium), iron is most likely to re-precipitate from solution first because aqueous ferrous iron is sensitive to changes in oxidation potential. Aqueous ferrous iron can be transported and precipitates around grains, now goethite (Fig. 5.3), or can infill pore spaces that microorganisms survive in, now recrystallised to hematite (Fig. 5.8). In comparison, the remaining aluminium in solution becomes relatively enriched and continues to be transported, precipitating as the final geological texture in select pore spaces as gibbsite (Figs. 5.9 – 5.10). The lack of zonation throughout the gibbsite precipitates supports the idea that little-or-no iron is left in solution when gibbsite forms. Minor silicon, which is relatively immune to changes in the oxidation potential compared with iron, is also present in these gibbsite precipitates. Minor trace elements, including uranium, continue to be transported in solution (Monteiro, 2017). Aluminium re-precipitation is most likely to be induced when meteoric waters influenced by microbial activity are mixed with fresh solutions, changing the pH and/or organic chemistry. Alternatively, evaporation during dry periods may promote the supersaturation of aluminium, causing it to re-precipitate. Acidic solutions and organocomplexation of aluminium are likely to be controlled by the native biota.

The biofilms throughout canga influence the precipitation of iron and aluminium minerals by providing a site for mineral nucleation on cell envelope surfaces (Ferris et al., 1988; Li et al., 2013). The mineralised cell envelopes can contribute to the formation of microbial cements within the canga (Levett et al., 2016). The preservation of organic biosignatures may be promoted by aluminium binding essentially irreversibly with the cell envelope, preventing the breakdown of the organic biosignatures (Levett et al., 2018). As microfossils continue to be exposed to iron-rich solutions, they can become permineralised (Fig. 5.8), which typically replaces the preserved organic biosignatures, leaving the microfossils with an aluminium enriched cell envelope-like structure (Levett et al., 2018). The microbial biofilms also provide an organic framework that can aggregate minerals and provides a template for mineral nucleation and cement formation (Levett et al., 2020). In the absence of microbial biofilms,

124

Chapter Five authigenic mineral precipitation would primarily coat mineral grains and may form a less effective cement.

5.5.4. Canga: extremely stable but biogeochemically dynamic Cosmogenic 3He measurements indicate that canga erosion rates in the Serra Sul are approximately 0.08 m Ma-1, making these duricrusts some of the most erosion-resistant landforms on Earth (Monteiro et al., 2018b). At first glance, the extreme stability of these surfaces appears to contradict the biological activity, highlighted in this study. However, in reality, the biological activity that promotes mineral dissolution is vital for the formation of new cements. Without continued mineral precipitation forming cements, the fragmental material would simply erode and be lost through the weathering channels. During the development of these duricrusts, extensive weathering has removed the most soluble elements, enriching only the most immobile elements. Microorganisms that promote in situ (within the duricrust) dissolution of iron and aluminium only influence canga on the scale of microenvironments, as opposed to large-scale climatic changes. Restricting microbial activity that creates reducing environments to ‘hotspots’ throughout canga is critical for the re- precipitation of minerals. Ubiquitous biological activity that promotes reducing conditions throughout canga would create an imbalance between mineral dissolution and precipitation, leading to the loss of iron and aluminium from the duricrusts. Cave formations throughout canga in the Serra Sul provide evidence of weathering and mineral precipitation imbalances. Defects to the structural integrity of canga may generate a collapse, though it is much more likely that the weathering imbalance is from more susceptible minerals, including carbonates and silica than iron oxide dissolution (Monteiro et al., 2018a). This conclusion has important implications for iron ore mine remediation strategies that may aim to harness microbially- accelerated iron oxide dissolution in situ. Over stimulating microbial activity that induce reducing conditions and promote iron oxide dissolution may result in the loss of iron from the system.

Within canga, the conditions that appear to promote mineral dissolution are short-term and readily altered. Changes to the oxidation potential, pH and organic compounds causes ions in solution to re-precipitate, preventing their loss from the duricrust. In addition, these mineral precipitates can cement loose grains that have been eroded and deposited from the surrounding metavolcanic units, thereby incorporating them into the duricrust.

125 Geochemical signatures of microbial weathering

The presence of rutile and chromium within the recycled grains indicates that material from the surrounding metavolcanic units are incorporated into canga, though it should be noted that metals in resistant minerals can also be incorporated into duricrusts from aeolian sources (Brimhall et al., 1988). Many of the grains with titanium and chromium signatures are strongly altered, with no primary mineralogy remaining. Previously reported geochronological evidence indicates that the surrounding metavolcanic units are eroded at higher rates than the canga

(Monteiro et al., 2018a). The extremely low concentrations of titanium ([TiO2]BIF = 0.017 wt.%) and chromium ([Cr2O3]BIF = 0.005 wt.%) in the least weathered version of the protolith

(Carajás Formation) indicate that titanium ([TiO2]canga = 0.24 wt.%) and chromium

([Cr2O3]canga = 0.015 wt.%) in the canga studied here are, in part, derived from external sources (Klein and Ladeira, 2002). This is particularly evidenced by the presence of detrital rutile in the canga samples, which is shown to weather to anatase (Fig. 5.8F) and is notably absent from other intensely weathered duricrusts (Du et al., 2012). Through these processes, minor elements (chromium, titanium and thorium) and detrital mineral inclusions from the metavolcanic units or aeolian sources are likely incorporated into the iron oxides throughout canga.

5.6. Conclusions The data presented here highlight the concurrent weathering and precipitation of iron oxide minerals within canga and the influences of microorganisms on each process. Titanium and chromium provide useful geochemical signatures to differentiate newly formed cements from weathered and/or metasomatised grains. Phase transformation of rutile to anatase provides conclusive evidence of weathering, likely to be influenced by the surrounding microorganisms, and titanium dissolution and re-precipitation at the mineral-scale. Iron dissolution is likely to be the limiting step for cementation of canga and is critical for canga formation, evolution and preservation. Iron dissolution is also likely to be controlled by hotspots of biological activity. Microbial biofilms provide an organic framework for mineral nucleation and cement formation. Abiotic mineral precipitates of goethite, hematite and gibbsite are also critical to the formation of new cements. The re-precipitation of iron and aluminium around grains and within pore spaces highlight the microscale changes in geochemical environments within canga.

Acknowledgements We acknowledge support the Vale S.A.-UQ Geomicrobiology initiative and the Australian Research Council Linkage Program (LP140100805) to G. Southam and P. Vasconcelos. The authors would like to acknowledge the Australian Microscopy & Microanalysis Research Facility, AuScope, the Science and Industry Endowment Fund, and the State Government of

126

Chapter Five

Western Australian for contributing to the Ion Probe Facility at the University of Western Australia. The authors acknowledge the facilities and the scientific and technical assistance of the Australian Microscopy and Microanalysis Research Facility at the Centre of Microscopy and Microanalysis, at the University of Queensland. We wish to acknowledge the assistance of the staff of Central Analytical Research Facility (CARF) at Queensland University of Technology (QUT) for access to analytical instrumentation, supported by the Faculty of Science and Engineering at QUT. Alan Levett acknowledges the support from the Australian

Government Research Training Program.

References Brimhall GH, Lewis CJ, Ague JJ, Dietrich WE, Hampel J, Teague T, et al. Metal enrichment in bauxites by deposition of chemically mature aeolian dust. Nature 1988; 333: 819.

Dachille F, Simons P, Roy R. Pressure-temperature studies of anatase, brookite, rutile and TiO2-II. American Mineralogist: Journal of Earth and Planetary Materials 1968; 53: 1929-1939.

Du X, Rate AW, Gee MM. Redistribution and mobilization of titanium, zirconium and thorium in an intensely weathered lateritic profile in Western Australia. Chemical Geology 2012; 330: 101-115.

Ferris FG, Fyfe WS, Beveridge TJ. Metallic ion binding by Bacillus subtilis: implications for the fossilization of microorganisms. Geology 1988; 16: 149-152.

Galvez ME, Beyssac O, Benzerara K, Menguy N, Bernard S, Cox SC. Micro-and nano-textural evidence of Ti (–Ca–Fe) mobility during fluid–rock interactions in carbonaceous lawsonite-bearing rocks from New Zealand. Contributions to Mineralogy and Petrology 2012; 164: 895-914.

Ganesh R, Robinson KG, Reed GD, Sayler GS. Reduction of hexavalent uranium from organic complexes by sulfate-and iron-reducing bacteria. Appl. Environ. Microbiol. 1997; 63: 4385-4391.

Gibson N, Yates CJ, Dillon R. Plant communities of the ironstone ranges of South Western Australia: hotspots for plant diversity and mineral deposits. Biodiversity and Conservation 2010; 19: 3951-3962.

Grainger CJ, Groves DI, Tallarico FHB, Fletcher IR. Metallogenesis of the Carajás Mineral Province, Southern Amazon Craton, Brazil: varying styles of Archean through

127 Geochemical signatures of microbial weathering

Paleoproterozoic to Neoproterozoic base- and precious-metal mineralisation. Ore Geology Reviews 2008; 33: 451-489.

Hanesch M. Raman spectroscopy of iron oxides and (oxy) hydroxides at low laser power and possible applications in environmental magnetic studies. Geophysical Journal International 2009; 177: 941-948.

Hoppe P, Cohen S, Meibom A. NanoSIMS: technical aspects and applications in cosmochemistry and biological geochemistry. Geostandards and Geoanalytical Research 2013; 37: 111-154.

Jacobi CM, Carmo FF, Vincent RC, Stehmann JR. Plant communities on ironstone outcrops: a diverse and endangered Brazilian ecosystem. Biodiversity and Conservation 2007; 16: 2185-2200.

Klein C, Ladeira EA. Petrography and geochemistry of the least altered banded iron-formation of the Archean Carajás Formation, Northern Brazil. Economic Geology 2002; 97: 643- 651.

Levett A, Gagen E, Shuster J, Rintoul L, Tobin M, Vongsvivut J, et al. Evidence of biogeochemical processes in iron duricrust formation. Journal of South American Earth Sciences 2016; 71: 131-142.

Levett A, Gagen EJ, Diao H, Guagliardo P, Rintoul L, Paz A, et al. The role of aluminium in the preservation of microbial biosignatures. Geoscience Frontiers 2018: 1-12.

Levett A, Gagen EJ, Vasconcelos PM, Zhao Y, Paz A, Southam G. Biogeochemical cycling of iron: implications for biocementation and slope stabilisation. Science of The Total Environment 2020; 707: 136128.

Li J, Benzerara K, Bernard S, Beyssac O. The link between biomineralization and fossilization of bacteria: insights from field and experimental studies. Chemical Geology 2013; 359: 49-69.

Machado N, Lindenmayer Z, Krogh T, Lindenmayer D. U-Pb geochronology of Archean magmatism and basement reactivation in the Carajás area, Amazon shield, Brazil. Precambrian Research 1991; 49: 329-354.

Männik J, Driessen R, Galajda P, Keymer JE, Dekker C. Bacterial growth and motility in submicron constrictions. Proceedings of the National Academy of Sciences USA 2009; 106: 14861-14866.

128

Chapter Five

Monteiro HS, Vasconcelos PM, Farley KA, Spier CA, Mello CL. (U–Th)/He geochronology of goethite and the origin and evolution of cangas. Geochimica et Cosmochimica Acta 2014; 131: 267-289.

Mota NFdO, Watanabe MTC, Zappi DC, Hiura AL, Pallos J, Viveros RS, et al. Amazon canga: the unique vegetation of Carajás revealed by the list of seed plants. Rodriguésia 2018; 69: 1435-1488.

Neaman A, Chorover J, Brantley SL. Implications of the evolution of organic acid moieties for basalt weathering over geological time. American Journal of Science 2005; 305: 147- 185.

Neumann G, Römheld V. Root excretion of carboxylic acids and protons in phosphorus- deficient plants. Plant and Soil 1999; 211: 121-130.

Ocana M, Garcia‐Ramos JV, Serna CJ. Low‐temperature nucleation of rutile observed by Raman spectroscopy during crystallization of TiO2. Journal of the American Ceramic Society 1992; 75: 2010-2012.

Parker CW, Wolf JA, Auler AS, Barton HA, Senko JM. Microbial reducibility of Fe (III) phases associated with the genesis of iron ore caves in the Iron Quadrangle, Minas Gerais, Brazil. Minerals 2013; 3: 395-411.

Paz A, Gagen EJ, Levett A, Zhao Y, Kopittke PM, Southam G. Biogeochemical cycling of iron oxides in the rhizosphere of plants grown on ferruginous duricrust (canga). Science of The Total Environment 2020: 136637.

Piló LB, Auler AS, Martins F. Carajás National Forest: iron ore plateaus and caves in southeastern Amazon. Landscapes and landforms of Brazil. Springer, 2015, pp. 273- 283.

Ruan H, Frost R, Kloprogge J. Comparison of Raman spectra in characterizing gibbsite, bayerite, diaspore and boehmite. Journal of Raman Spectroscopy 2001; 32: 745-750.

Sahoo PK, Felix Guimarães JT, Martins Souza-Filho PW, Sousa da Silva M, Maurity CW, Powell MA, et al. Geochemistry of upland lacustrine sediments from Serra dos Carajás, Southeastern Amazon, Brazil: implications for catchment weathering, provenance, and sedimentary processes. Journal of South American Earth Sciences 2016; 72: 178-190.

Sahoo PK, Guimarães JT, Souza-filho PW, Bozelli RL, de Araujo LR, de Souza Menezes R, et al. Limnological characteristics and planktonic diversity of five tropical upland lakes from Brazilian Amazon. Annales de Limnologie-International Journal of Limnology. 53. EDP Sciences, 2017a, pp. 467-483.

129 Geochemical signatures of microbial weathering

Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, et al. Fiji: an open-source platform for biological-image analysis. Nature Methods 2012; 9: 676.

Silva MSD, Guimarães JTF, Souza Filho PWM, Nascimento Júnior W, Sahoo PK, Costa FRD, et al. Morphology and morphometry of upland lakes over lateritic crust, Serra dos Carajás, southeastern Amazon region. Anais da Academia Brasileira de Ciências 2018; 90: 1309-1325.

Spier CA, Levett A, Rosière CA. Geochemistry of canga (ferricrete) and evolution of the weathering profile developed on itabirite and iron ore in the Quadrilátero Ferrífero, Minas Gerais, Brazil. Mineralium Deposita 2018; 54: 983-1010.

Trendall AF, Basei MAS, de Laeter JR, Nelson DR. SHRIMP zircon U–Pb constraints on the age of the Carajás formation, Grão ParáGroup, Amazon Craton. Journal of South American Earth Sciences 1998; 11: 265-277.

130

Chapter 6 Microbiological and chemical influences on the evolution of iron- rich duricrusts in the Serra Sul de Carajás, Pará, Brazil

Alan Levetta, Paulo M. Vasconcelosa, Michael M. W. Jonesb, Llew Rintoulb, Anat Paza, Emma J. Gagena, Gordon Southama

a School of Earth and Environmental Sciences, University of Queensland, Brisbane, Queensland 4072, Australia

bCentral Analytical Research Facility, Institute of Future Environments, Queensland University of Technology, Brisbane, Queensland 4001, Australia

Manuscript in review with Chemical Geology

Chapter Six

Abstract The development and evolution of iron-rich duricrusts (canga) that cap weathered banded iron formations (BIFs) in the Serra Sul of the Carajás, northern Brazil, has been essential to armour these landscapes, allowing for the enrichment and preservation of vast iron ore deposits. Here, we present a detailed biogeochemical study of canga drill cores, collected for exploration purposes, to understand metal mobility and iron oxide transformations. Microorganisms, which play a significant role in mineral recycling, are commonly fossilised throughout canga and volumetrically account for significant portions of the iron cements. During microbially- accelerated mineral weathering, titanium is placed in solution and reprecipitates around nearby grains as microcrystalline anatase. X-ray absorption near edge structure (XANES) analysis of the canga demonstrate that the grains are enriched in ferrous iron, which is relatively easily released during acid-induced weathering. The release of ferrous iron from altered magnetite during acid extraction experiments, as well as pitting throughout martite grains, supports the nonredox-based transition of magnetite to martite via a ferrous iron-deficient magnetite (kenomagnetite). The ferrous iron released during the weathering of magnetite-rich rocks, providing an iron source for the generation of new cements within canga. Synchrotron-based X-ray fluorescence microscopy revealed titanium-rich grains with a dodecahedral form that are likely to be magnetite grains originating from the surrounding mafic sills and/or the metavolcanic and metasedimentary pile that overlies the Carajás BIF. The enrichment of aluminium, phosphorus, silicon, titanium, vanadium and zirconium in the cements also suggest contributions from sources other than the protolithic BIF. The transportation and incorporation of these materials within the canga cements highlights the continuous formation of new cements during mineral recycling and suggests the Carajás has undergone a degree of relief inversion. The markedly high phosphorus concentrations in the canga cements compared with any of the possible protoliths, either the underlying BIF or the surrounding metavolcanic units, indicates an external, possibly aeolian, phosphorus source.

132

Trace element behaviour in canga horizons

6.1. Introduction The formation and ongoing evolution of ferruginous duricrusts that cap iron ore deposits in Brazil continues to be a research topic of geological (Monteiro et al., 2018a; Monteiro et al., 2014; Monteiro et al., 2018b; Shuster et al., 2012; Spier et al., 2018), ecological (Jacobi and Carmo, 2011; Jacobi et al., 2007; Messias et al., 2013; Viana-Silva and Jacobi, 2012) and microbiological importance (Gagen et al., 2019; Gagen et al., 2018; Levett et al., 2016; Levett et al., 2019a). The indurated iron-rich duricrusts that cap iron ore deposits in Brazil are goethite- cemented breccias, referred to as canga (Dorr, 1964). The formation and iron enrichment of canga is primarily via leaching of silica and other more soluble elements during intensive weathering of the underlying Archean banded iron formations (BIFs) (Trendall et al., 1998). Biogeochemical cycling of iron provides the ultimate control for the ongoing evolution and long-term stability of these ancient landforms (Monteiro et al., 2018b).

Weathering of the Carajás BIF over 10s of millions of years removes silica as H4SiO4 (l), enriching secondary iron oxide minerals (predominantly hematite [Fe2O3]) within the saprolite, which is mined as high-grade iron ore. Partial reductive dissolution of hematite and magnetite in the near-surface environment (upper 20 – 40 m of the weathering profile), and the local reprecipitation of soluble iron as goethite cements, generates the brecciated rock composed of BIF and hematite clasts cemented by authigenic goethite. The canga is porous but strongly cemented, allowing solutions to percolate through and continue to weather the underlying saprolite (Dorr, 1964), promoting propagation of the canga-formation front (Monteiro et al., 2014; Monteiro et al., 2018b). The underlying friable iron ores are protected from erosion by the formation of canga (Dorr, 1964).

Cosmogenic 3He concentrations demonstrate that canga horizons in the Serra Sul de Carajás are some of the most stable landforms on Earth; erosion is combatted by continuously forming ferruginous cements within canga (Monteiro et al., 2018a). The geochemical, redox-based cycling of iron within canga is likely to be driven by biological processes (Monteiro et al., 2014). This is supported by the prevalence of fossilised microorganisms throughout canga (Levett et al., 2016; Levett et al., 2019a), which can constitute a significant proportion of the duricrust cements and contribute to relatively rapid cementation (Levett et al., 2019b; Levett et al., 2020).

In the absence of a soil profile, microorganisms, insects and plants that grow within canga environments must interact, both chemically and physically, directly with the rock to obtain

133 Chapter Six nutrients and occupy structural niches. Canga represents a relatively harsh environment, containing very few available nutrients, with most of the phosphorus (up to 0.6 wt.% in this study) inorganically bound to the plentiful and diverse iron oxide minerals (Dorr, 1964; Jacobi and Carmo, 2011). Plants have been generally shown to exude organic acids under phosphorus deficient conditions (Neumann and Römheld, 1999), promoting the release of mineral-bound phosphorus into solution for assimilation by the plant and contributing to iron dissolution in canga (Paz et al., 2020).

This manuscript aims to examine the influence of microorganisms on the evolution of iron-rich duricrusts; their role in mineral dissolution, precipitation and their impacts on element mobility, including titanium and chromium, throughout canga profiles. The minor and trace element analysis of canga also offers insights into the influx of exogenous materials that play a role in the formation and evolution of these duricrusts.

6.2. Geological setting Sahoo et al. (2016) provides a geological map of the Serra Sul, Carajás Mineral Province in northern Brazil (Fig. 6.1). The Carajás Formation overlies low-grade mafic to intermediate metavolcanic units that comprise the Parauapebas Formation and mafic sills (Cigarra Formation), which together form the Grão Pará Group (Olszewski et al., 1989). An unconformity separates the Grão Pará Group from the underlying Xingu Complex, which is comprised of Mesoarchean granitoids and gneisses (Feio et al., 2013). Overlying the Grão Pará Group are the Igarapé Cigarra and Aguas Claras Formations, composed of mixed volcanics, sandstone and siltstone (Trendall et al., 1998). The southern region (Serra Sul) of the Carajás Mineral Province contains the largest iron ore mine in the world (S11D; established 2017), which is in proximity to ecologically unique Lake Violão and Lake Amendoim (Fig. 6.1). Canga caps the weathered banded iron formations and the surrounding metavolcanic units (Monteiro et al., 2018b); canga on the volcanic units is thought to be a consequence of the local transport of BIF and hematite fragments from the higher elevation BIF to the lower volcanic units and the in situ cementation of the detrital materials by authigenic iron cements (Tolbert et al., 1971).

The Serra Sul region has a tropical climate, with annual daily-high temperatures typically 26 – 28 °C (Sahoo et al., 2016). Protracted (millions of years) monsoonal rainfall (Monteiro, 2017) during the wet season (November – May), has extensively weathered the region including the Carajás Formation (BIF), helping to create some of the world’s largest high-grade iron ore

134

Trace element behaviour in canga horizons deposits, including Serra Norte, Serra Leste and Serra Sul (Grainger et al., 2008; Tolbert et al., 1971).

Fig. 6.1. Google Earth image of the Serra Sul region of the Carajás, Pará, Brazil, highlighting locations of the drill core samples in close proximity to the S11D mine and Lakes Amendoim and Violão.

6.3. Materials and methods 6.3.1. Sample characterisation Well-cemented ‘in situ’ material was collected from across two 12 m sections of drill core, providing a profile for the canga in the Serra Sul (Fig. 6.1). Drill core subsamples were crushed to less than 70 micrometres using a ring and puck mill to characterise bulk chemistry: X-ray fluorescence (XRF) and iron-extraction experiments, which were used to model the Fe(II) and microbial-reducible Fe(III) available in canga.

The chemical composition of bulk samples throughout the profile was quantified using XRF at the Australian Laboratory Services (Analytical Geochemistry). Prior to analysis, samples were

135 Chapter Six fused with a lithium tetraborate:lithium metaborate (12:22) flux, which included lithium nitrate as an oxidising agent. Samples were analysed in a platinum mould following fusion. Loss on ignition (LOI) volatiles were determined at 1000 °C.

The iron-extraction experiments were carried out using protocols modified slightly from Lovley and Phillips (1987). Briefly, to determine the amount of acid-extractable Fe(II) present throughout the canga profile, approximately 0.1 g of crushed canga was reacted with 5 mL of 0.5 N HCl in a glass scintillation vial. After 1 hour, the crushed canga and HCl solution were homogenised by inverting several times and a 0.1 mL aliquot was added to 5 mL of 1 g/L ferrozine in 50 mM HEPES (N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid) buffer. This was repeated three times to determine reproducibility of the extraction. The ferrozine and reactant where mixed thoroughly and filtered through a 0.22 µm pore diameter (Millex polyethersulfone (PES)) filter. The Fe(II) concentration was determined by measuring absorption at 562 nm using a UV-Visible spectrophotometer and quantified using a

FeSO4·7H2O standard curve (r = 0.9999) . The process was repeated using a duplicate sample, except the extractant was 5 mL of 0.25 M hydroxylamine·hydrochloride in 0.25 N HCl. The biologically available (readily reducible) Fe(III) present in canga was determined by subtracting the hydrochloric acid extractable iron from the hydroxylamine·hydrochloric acid extractable iron.

6.3.2. Electron microscopy To characterise the microstructure of all drill core subsamples, polished petrographic thin sections were examined using JEOL7100 Field Emission scanning electron microscope (FE- SEM) equipped with an energy dispersive X-ray spectrometer. Prior to examination, samples were degassed at 50 °C for a minimum of 12 hours and coated with either 10 nm of carbon or iridium using BAL-TEC MSC-010 sputter coaters. An accelerating voltage of 15 kV was used to acquire backscattered electron micrographs.

6.3.3. Synchrotron-based X-ray fluorescence microscopy Quartz and glass petrographic thin sections of drill core subsamples were polished to a thickness of approximately 30 µm and coated with carbon for analysis using synchrotron-based X-ray fluorescence microscopy (SXFM) at the Australian Synchrotron, Clayton, Australia (Paterson et al., 2011). A monochromatic 18.5 keV X-ray beam was focused to approximately 2 µm using a Kirkpatirck-Baez mirror pair, and scanned across regions of interest with a sampling interval of 2 µm in both the horizontal and vertical directions with an effective dwell

136

Trace element behaviour in canga horizons time of 0.33 ms per pixel. Excited fluorescent photons were collected with a 384 element Maia detector (Siddons et al., 2014) and analysed with GeoPIXE (Ryan et al., 2014), producing elemental maps from phosphorus to zirconium inclusive. Elemental maps were quantified using known elemental platinum, iron and manganese foils.

Spatially resolved X-ray absorption near edge spectroscopy (XANES) data was collected using the scanning parameters described above with 148 non-uniformly spaced energies from 6.962 keV to 7.462 keV. Data were first conditioned for noise using the data collected at an incident energy of 7.462 keV (Jones et al., 2017) to eliminate pixels with poor signal, before spectra were normalised and background subtracted. The degree of iron oxidation was determined using the energy where the rising edge is equal to a normalised intensity of 0.8, with a higher energy denoting greater degree of oxidation (Berry et al., 2018; Nedoseykina et al., 2010). To reduce the influence of noise present in the raw per-pixel spectra, the rising edge was fitted with a pair of Pseudo-Voigt functions and the energy where the fit equalled a normalised intensity of 0.8 used to determine the degree of iron oxidation. Spectra with a poor fit (R2 < 0.98) were omitted from further consideration.

6.3.4 Raman Spectroscopy The mineralogy of titanium-rich and titanium-poor grains that were mapped using synchrotron- based XFM was determined using Raman spectroscopy. In addition, the titanium-enriched rim of a grain boundary was mapped using Raman spectroscopy to determine the mineral phases and their distribution. For all analysis, a WITec Alpha 300 Raman microscope equipped with a 532 nm laser was operated at less than 0.6 mW to acquire Raman spectra. The laser, focused with a Zeiss 50× objective of 0.7 NA, formed a confocal sample volume with a cylindrical- shape that interacted with the top 2 – 3 m of the sample and had a diameter of ~ 0.5 m. A dwell time of 64 s was used to acquire single point spectrum for mineral identification. To acquire the Raman spectroscopy map, the titanium-rich grain boundary was mapped by raster motion in increments of 2 m to cover the rectangular (100 x 50 m) region of interest with data accumulated for 16 s per spectrum. The background was removed from all spectra prior to analysis to produce mineralogical maps using Classical Least Squares (CLS) method. For CLS analysis, each spectrum is deconvoluted into a mineral score by using reference spectra to account for the individual constituents of each spectrum, accounting for error. The resulting constituent (mineral) score represents a semi-quantitative measure of its relative concentration at the specific site on the sample. Reference spectra for hematite, goethite and anatase were

137 Chapter Six identified by manual inspection of the data set. WITec Control Four and Project Four software packages were used instrument control and data analysis.

6.4. Results 6.4.1. Bulk sample characterisation The two drill cores from the canga profile (S11D, Pará, Brazil, Vale S.A. site) were primarily composed of goethite and hematite, reflected in the high iron concentrations (56 – 65 wt.%; Table 6.1). A two-tailed t-test assuming unequal variance indicated that phosphorus, titanium, vanadium, zirconium and LOI concentrations were significantly greater in the cements compared with the fragments (P < 0.01; Table 6.1). The average aluminium concentrations for the two drill cores are 3.96 wt.% and 1.73 wt.%, respectively. Aluminium and iron are the dominant cations within canga and are inversely proportional, which is reflected by the negative aluminium-iron correlations for both drill core 1 (r = -0.81; Table 6.2) and drill core 3 (r = -0.84; Table 6.3). The increased relative surface area of aluminium-substituted goethite provides additional binding sites for phosphorus sorption (Schulze and Schwertmann, 1984). Therefore, the positive aluminium-phosphorus correlation values for drill core 3 (r = 0.79; Table 6.3) was expected but it is surprisingly low for drill core 1 (r = 0.33; Table 6.2). The low phosphorus-aluminium correlation and the relatively high aluminium concentration for drill core 1 indicates that the aluminium is not substituted into goethite minerals but instead forms gibbsite (Levett et al., 2019a). Vanadium and titanium dioxide (TiO2) also correlated with aluminium for both drill cores (r ≥ 0.76; Tables 6.2 and 6.3).

Acid-extraction experiments using 0.5 N HCl indicated that approximately 10 – 20 ppm Fe(II) was present in the bulk canga samples (Table 6.1). To determine the component of canga and the minerals that contained the relatively easily acid-extractable Fe(II), the acid extraction assays were conducted using cement-only and fragment-only portions of drill core 3. The Fe(II) concentrations were significantly (P < 0.001) higher in the detrital fragments (approximately 40 ppm) compared with the canga cements (approximately 1 ppm; Table 6.1), highlighting a source of accessible Fe(II) within the specular mineral proportions of canga.

138

Trace element behaviour in canga horizons

Table 6.1. Major, minor and trace element sample concentrations determined using X-ray fluorescence (wt.%).

*Labile ferrous iron concentrations determine using a 0.5 M HCl extraction **Biologically-reducible ferric iron determined using a 0.25 M hydroxylamine·hydrochloride in 0.25 M HCl extraction Significant (p<0.01) difference between fragment and cement concentrations are highlighted in yellow, with higher average concentrations highlighted in orange and lower concentrations highlighted in blue.

139 Chapter Six

Table 6.2. . Correlation matrix for drill core 1 – bulk sample. Positive correlations (r > 0.6) are highlighted in yellow; negative correlations (r < -0.6) are highlighted in blue.

Table 6.3. Correlation matrix for drill core 2 – bulk sample. Positive correlations (r > 0.6) are highlighted in yellow; negative correlations (r < -0.6) are highlighted in blue.

6.4.2. Electron microscopy Microfossils, identified by the preservation of cell envelope structures, were present throughout drill core subsamples to depths of approximately 5 m (Figs. 6.2, 6.3 and 6.6). Microfossils were likely to be preserved below 5 m; however, the high number of samples requiring characterisation at the microscale and the possible degradation of bacteriomorphic structures in these older canga samples (Shuster et al., 2012) precluded the opportunity to determine the biogenicity of bacteriomorphic structures identified in these samples (data not shown). Rod- shaped microfossils preserved by the mineralisation of their cell envelope structures occur

140

Trace element behaviour in canga horizons around highly weathered grains throughout the canga (Fig. 6.2A), with large biofilms commonly preserved (Fig. 6.2B). The encrusted cell envelopes (Fig. 6.3A; white arrows) were preserved within a grain that had undergone generations of recycling, comprised of iron precipitates infilling vein-like structures and incorporating detrital materials (Fig. 6.3B). Microfossils in the recycled grain had a lower cell density (Fig. 6.3A) compared with other biofilms fossilised through canga (for example, Fig. 6.2).

Fig. 6.2. Backscattered electron SEM micrographs highlighting the presence of iron oxide mineralised cell envelopes present throughout canga in the Serra Sul region, Carajás, Pará, Brazil.

141 Chapter Six

Fig. 6.3. (A) Backscattered electron SEM micrograph highlighting encrusted cell envelopes (white arrows) within a recombinant grain composed of detrital fragments and recent iron oxide precipitates infilling pore spaces. (B) Low resolution backscattered electron SEM micrograph of the highly weathered recombinant grain containing microfossils. White arrows highlight the edges of the grain, which is enriched in titanium (see Fig. 6.5A).

142

Trace element behaviour in canga horizons

6.4.3. Chemical and mineralogical mapping of select canga regions Synchrotron-based X-ray fluorescence microscopy (SXFM) highlights the complex textures within canga and reveals surprising elemental behaviour, particularly that of classically immobile elements such as titanium and chromium. For sample DC1_4.90 m, the iron-rich precipitates within the pore spaces of canga (Fig. 6.4D; arrows) are also enriched with elements that are not derived from the protolithic BIF. For example, these massive, primarily goethitic precipitates are enriched with cobalt (Fig. 6.4E; arrows) and a slight enrichment in chromium (Fig. 6.4C; arrows) compared with the canga matrix. In contrast, titanium and minor vanadium rims surround highly weathered clastic material (Figs. 6.4A and 6.4B; arrows). Raman spectroscopy indicated the titanium-rich grain (Fig. 6.4A; asterisk) was hematite. The dodecahedral form of this grain indicates that it is likely to be an altered magnetite grain.

Similarly in subsample DC3_5.20 m, titanium was enriched around euhedral hematite and relict, highly recycled grains (Fig. 6.5A; white arrows). Chromium (Fig. 6.5B), manganese (Fig. 6.5C) and cobalt (Fig. 6.5E) do not display similar patterns to the titanium and are primarily enriched within the grains. Raman spectroscopy revealed the fragments enriched with chromium (marked with an asterisk; Fig. 6.5B) are hematite. The hematite-rich fragments were also enriched in manganese (Fig. 6.5C), and cobalt (Fig. 6.5E). The XFM micrographs revealed local, small-scale mobilisation of titanium preserved within the duricrusts. Titanium is leached from titanium-bearing grains and reprecipitates around nearby grains (Fig. 6.5A; arrows). The optical image in Fig. 6.5F highlights the region mapped using Raman spectroscopy, with the large hematite-rich fragment in the top of the image (Fig. 6.5F). The high relief of this fragment meant it was out of focus with respect to the rest of the region so it not assigned a mineralogical value in the Raman map but spot analysis confirmed it is hematite. The Raman spectroscopy maps reveal that titanium does not substitute into the structure of the goethite (Fig. 6.5F; blue) precipitates but forms microcrystalline anatase (Fig. 6.5F; green) inclusions within the precipitates. Larger anatase and hematite (red) minerals were distributed around the fragment coatings (Fig. 6.5F).

A pitted grain from subsample DC1_4.02 m that has a dodecahedral (subhedral) form, likely to be altered magnetite (martite), was mapped using XANES to determine the distribution of the ferrous iron within the canga. The altered magnetite grain was surrounded by a diverse array of microfossils, including filamentous microfossils (Fig. 6.6A; white arrow) and cross-

143 Chapter Six sections of Sporosarcina-type (Mazanec et al., 1965) cell groupings (Fig. 6.6A; black arrows). The pitted grain and other highly altered grain remnants are embedded within microfossil and abiotic goethitic cements (Fig. 6.6B). The XANES analysis, corroborated the iron extraction experiments, revealing directly that Fe(II) is enriched within the relict grains compared with the surrounding goethitic cements (Fig. 6.6C).

Fig. 6.4. Synchrotron-based X-ray fluorescence microscopy of a canga subsample from drill core 1 from a depth of approximately 4.9 m. Titanium (A) is enriched in rims around highly weathered grains (white arrows) and within detrital grains (asterisk). Vanadium (B) shows similar weathering patterns to titanium. Chromium (C), iron (D) and cobalt (E) are mostly strongly enriched in precipitates that form within the pore spaces in canga. Colour scale bar indicates relative element concentrations from low (black) to high (white).

144

Trace element behaviour in canga horizons

Fig. 6.5. Synchrotron-based X-ray fluorescence microscopy of a canga subsample from drill core 2 from a depth of approximately 4.2 m. Titanium (A) displays unusual rims around grains. Chromium (C), manganese (D), iron (E) and cobalt (F) are primarily enriched within the detrital fragments in the canga. Raman spectroscopy (F) of a titanium coated grain highlights that goethite (blue) has precipitated around the hematite-rich grain with microcrystalline anatase (green) disseminated throughout the goethite precipitates and in some of the microscale hematite (red) grains. Colour scale bar indicates the relative element concentrations from low (black) to high (white). Raman colour map: blue = goethite, red = hematite, green = anatase.

145 Chapter Six

6.5. Discussion The powder XRF and XFM data highlight the high concentrations of major and minor elements in the cements of canga compared with the hematite (martite)-rich detrital fragments (Table 6.1; Figs. 6.4 – 6.5). Where the canga plateaus outcrop in the Serra Sul, the overlying metavolcanic units (Trendall et al., 1998) have eroded and appear to have contributed exogenous material to the ferricretes. Titanium signatures revealed variable sources for iron oxide minerals within the canga. For example, the grain in Fig. 6.4 (asterisk) with a dodecahedral form, typical of magnetite, is relatively rich in titanium. Similarly, the titanium- rich grains in Fig. 6.5 are pitted, showing textures typical of martite minerals (Klein and Ladeira, 2002). In comparison, the euhedral titanium-poor martite grain (Fig. 6.5A; yellow asterisk) is likely to originate from the underlying BIF, with manganese contributions from the manganese-rich dolomites during alteration (Dalstra and Guedes, 2004; Silva et al., 2013). Hydrothermal magnetite minerals associated with BIFs in Western Australia have low aluminium, titanium, vanadium, chromium, manganese and cobalt contents (Nadoll et al., 2014). Knipping et al. (2015) also demonstrate that igneous magnetite has a high titanium and vanadium content compared to hydrothermal magnetite. Therefore, the titanium-rich fragments are likley to be altered magnetite that originated from the surrounding metavolcanic units, rather than hydrothermal magnetite or altered magnetite from the protolithic BIF. As these surrounding metavolcanic units, mafic dykes and fluvial sedimentary sequences (Trendall et al., 1998) erode, the chemically stable minerals have been incorporated within the canga (Fig. 6.7). The incorporation of the surrounding metavolcanic units is corroborated by the enrichment of aluminium, silicon, titanium, vanadium and zirconium within the cements of canga compared with the fragments (Table 6.1). The incorporation of these foreign materials preferentially in the canga cements highlights the relatively high rates of re-cycling within these iron-rich duricrusts (Monteiro et al., 2018b; Sahoo et al., 2017). New cements that are formed during the biogeochemical cycling of iron oxide minerals prevents the erosion of these exogenous materials. The transport, deposition and incorporation of surrounding units into the ferruginous duricrust tentatively suggests that Serra Sul has undergone relief inversion.

6.5.1. Titanium and chromium mobility and redistribution The XFM results presented here (Figs. 6.4 – 6.5), highlight localised mobility of titanium, leached from titanium-rich fragments and re-precipitating to coat the surrounding grains. Titanium mobility is highly likely to be influenced by presence of organic acids as well as the local physiochemical conditions. Neaman et al. (2005) used titanium to normalise element

146

Trace element behaviour in canga horizons mobility in a basalt using a 0.01 M citrate extraction under differing partial pressures of oxygen. Citrate treatments of basalt under oxic and anoxic conditions released 0.13% and 0.2%, respectively, of titanium into solution compared with <0.01% in ligand-free control experiments, regardless of oxygen. Organic acids exuded by plants and the microbiome associated with canga are likely to contribute to the titanium-mobility and redistribution within the duricrusts. The role of microorganisms in titanium mobility is supported by the preservation of microfossils (Fig. 6.3) within a nearby grain, also coated in titanium (Fig. 6.5A). Therefore, the titanium mobility and re-distribution around grains may be considered an indirect signature of microbial activity.

Titanium still appears to behave as an immobile element at the profile-scale, precipitating out of solution around nearby grains rather than being lost through weathering channels. The titanium mobility and redistribution presented here requires consideration when using titanium as an immobile element for mass balance calculations for both closed (Nesbitt, 1979) and open (Brimhall et al., 1988) weathering systems. Titanium precipitated around grain boundaries is likely to be lost during sample collection and processing.

Chromium is slightly enriched in the goethite-rich precipitates within the pore spaces of canga (Fig. 6.4C), highlighting its mobility and re-distribution within circumneutral environments. During Cr(VI) reduction, organics can complex the chromium to maintain mobility; however, these soluble complexes are not formed by mixing Cr(III) species with organics in a neutral buffer (Puzon et al., 2005). Therefore, mobilisation of chromium within canga, a predominately circumneutral environment, indicates solutions percolating throughout canga are highly − oxidising, as chromium is likely to be mobilised by oxidation to HCrO4(aq) species (Beverskog and Puigdomenech, 1997). Uranium is also depleted within goethite minerals in canga, 2+ oxidised and transported as UO2(aq), depleting near surface uranium concentrations (Monteiro et al., 2018b). Though sub- to anoxic niches occur throughout canga and contribute to the biogeochemical cycling or iron, chromium and uranium mobilisation highlights solutions are predominantly highly oxidising.

6.5.2. Chemical stability of iron oxide minerals with canga The XANES analysis combined with the acid extraction experiments provide a valuable model for the weathering patterns of iron oxide minerals within canga, particularly altered magnetite minerals. Ferrous iron, enriched within the fragments of canga compared with the cements (Table 6.1; Fig. 6.6C), is relatively easily released into solution by weak acid concentrations

147 Chapter Six

(0.5 N HCl). Ferrous iron can be incorporated in the structure of hematite (Skomurski et al., 2010); however, the ferrous iron released during the acid extraction is likely to be remnant from the original magnetite minerals, now martite containing domains of kenomagnetite. A relatively stable proportion of ferrous iron that can be released during acid weathering lends further plausibility to the nonredox transformation of magnetite to hematite (Reaction 6.1) postulated by Ohmoto (2003).

+ 2+ 퐹푒3푂4 + 2퐻 → 퐹푒2푂3 + 퐹푒 + 퐻2푂 (6.1)

This reaction indicates that acid weathering releases ferrous iron and enriches ferric iron within the mineral, which re-orders to form martite with kenomagnetite. This reaction, which accounts for a mineral volume loss of 32.2%, is also supported by the pitting throughout the altered magnetite fragments. In comparison, the oxidation of ferrous iron within magnetite (Reaction 6.2) leads to an increase in volume of 1.66% (Mücke and Cabral, 2005).

4퐹푒3푂4 + 푂2 → 6퐹푒2푂3 (6.2)

In oxic environments, the ferrous iron released is unstable and likely to precipitate as goethite (Reaction 6.3), contributing to the formation of new cements within canga. Understanding these weathering patterns may be valuable when selecting a starting material for the remediation of iron ore mines to accelerate the formation of iron-rich cements (Gagen et al., 2019; Gagen et al., 2018; Levett et al., 2016).

2+ + 4퐹푒 + 푂2 + 6퐻2푂 → 4퐹푒푂푂퐻 + 8퐻 (6.3)

Ferrous iron interactions with iron oxide minerals are of broad interest to biogeochemists and more work is required to understand these interactions and electron transfer mechanisms and atom exchange more clearly in natural systems (Frierdich and Catalano, 2012; Latta et al., 2012).

148

Trace element behaviour in canga horizons

Fig. 6.6. Backscattered electron SEM micrograph from sample DC1_4.02 m, highlighting the distribution of various bacteriomorphic fossils, including filamentous structures (white arrow) and Sporosarcina-like cells (black arrows) within the goethitic cements. (B) Low resolution backscattered electron SEM micrograph highlighting the subhedral dodecahedral form of a martite mineral embedded within abiotic and microfossil cements within canga. (C) XANES analysis from the area represented by the white dashed rectangle in Fig 6.6 (B) revealed ferrous iron is enriched within the martite (with kenomagnetite) mineral within the altered and relict grains throughout canga compared with the surrounding abiotic and microfossil cements. Colour scale bar indicates the proportion of ferrous iron to total iron from low (black) to high (white).

149 Chapter Six

6.5.3. Sources of phosphorus within canga For many of the world’s iron ore deposits (excluding goethite-rich ores and those influenced by postgenetic alteration), the average enrichment factors of phosphorus is less than those for iron (ranging from 1.63 to 2.42), except at the Sandur schist belt (Gutzmer et al., 2008). Phosphorus enrichment in supergene-derived iron ores never exceeded that of aluminium (Gutzmer et al., 2008). In contrast, phosphorus concentrations in the ferruginous carapaces that cap iron ore deposits are extremely variable. Here, phosphorus concentrations within the cements (0.53 wt.%) and fragments (0.31 wt.%) are at least an order of magnitude greater than phosphorus concentrations in the least altered Carajás Formation (BIF), which averages less than 0.01 wt.% (Klein and Ladeira, 2002; Sahoo et al., 2017; Sahoo et al., 2015). Therefore, phosphorus enrichment in the canga in the Serra Sul is well beyond the maximum iron enrichment factors during supergene weathering (Table 6.4). Phosphorus enrichment within canga in South America is well documented (Dorr, 1973; Spier et al., 2018). Here, we expand on mechanisms of phosphorus mobility and postulate that aeolian sources are significant contributors to phosphorus enrichment.

Table 6.4. Perceived enrichment factors in canga from the Serra Sul compared with supergene enrichment factors from iron ores (saprolite) around the world (Carajás and Urucum, South America, Mount Tom Price, Australia and Maremane Dome, South Africa). Atypical enrichment factors within the canga are highlighted in yellow.

Mass balance calculations for typical lateritic profiles shows that phosphorus is a mobile element, depleted by acidic complexation during weathering (Hill et al., 2000; Walker and Syers, 1976). Phosphorus behaviour throughout lateritic profiles is highly irregular and rarely

150

Trace element behaviour in canga horizons correlates with iron concentrations. For example, in a well-studied lateritic profile (Tansvaal Supergroup in southern Africa) differences in phosphorus concentrations between the iron-rich and iron-poor mottled zones are negligible (~ 0.04 wt.% phosphorus); however, the phosphorus concentrations are greatly increased (0.21 wt.% phosphorus) in the overlying laterite (Beukes et al., 2002). In the Hamersley region of Western Australia, phosphorus is depleted in iron-rich carapaces compared with the underlying iron ore (Morris, 1994). These authors note that hardcaps blanketing the iron ore in the Hamersley are dehydrated, transforming goethite to hematite and excluding phosphorus during the process. As a consequence, phosphorus is enriched within the most recent goethite precipitates, before being ultimately leached to the underlying units (Dukino et al., 2000; Morris, 1994). Therefore goethite, particularly aluminium-substituted goethite, appears to maintain the high phosphorus concentrations in the ferruginous duricrusts in tropical environments; however, this does not provide a source for the phosphorus.

As discussed, the goethitic cements in canga in the Carajás are strongly influenced by the associated biome; microorganisms and plants associated with the canga contribute to the biogeochemical cycling of iron oxide minerals as a they exude organic acids to release and uptake mineral-bound phosphorus (Monteiro et al., 2018b; Paz et al., 2020). Even if the canga- associated biome has developed extremely effective strategies to uptake the newly available phosphorus, these processes will ultimately contribute to the lability of phosphorus, supported by the relatively rare formation of secondary iron phosphates (Spier et al., 2018). Therefore, the biome in the Carajás contributes to phosphorus depletion in canga, not enrichment. As such, to maintain such high phosphorus concentrations within highly recycled canga cements, the phosphorus source is likely to be renewable. The surrounding metavolcanic rock units that contributed to titanium-enrichment have low phosphorus concentrations (< 0.035 wt.%) (Feio et al., 2013), and, given the topographic inversion, these phosphorus sources are non- renewable.

151 Chapter Six

Fig. 6.7. Model of the influences on the evolution of canga plateaus in the Serra Sul, Carajás Mineral province in northern Brazil. The weathering and erosion of the surrounding metavolcanic units and mafic dykes and sills enriches the goethitic cements with aluminium, silicon, phosphorus, titanium, vanadium and zirconium. The very high phosphorus concentrations in the goethite cements (averaging 0.53 wt.%) also indicates a continuous phosphorus influx. Aerosols that from Africa that deposit phosphorus in the Amazon Basin (Okin et al., 2004) may also be incorporated in the ferruginous duricrusts. Organic acids exuded by plants roots and microorganisms associated with the canga mobilises titanium, which reprecipitates around nearby grains as microcrystalline anatase. Acid-induced weathering of martite releases ferrous iron, which re-precipitate to form goethite cements (Reaction 6.3*).

Atmospheric phosphorus deriving from Africa has been demonstrated to supplement the neighbouring Amazon Basin (Okin et al., 2004). African aerosols produced by biomass burning

152

Trace element behaviour in canga horizons in Africa have been demonstrated to supply up to 50% of the phosphorus deposited annually in the Amazon (Barkley et al., 2019). These phosphorus-rich aerosols are therefore likely to provide the continual phosphorus sources for canga-associated ecosystems in the Carajás (Fig. 6.7). Given the low bioavailability of phosphorus within canga, these atmospheric sources may provide essential phosphorus sources for the productivity of canga ecosystems (Okin et al., 2004). Direct measurements of aeolian deposition within Serra Sul are required to confirm these hypotheses.

6.6. Conclusions The detailed geochemical analysis and electron micrographs presented here sheds light on the complex evolution of canga horizons in the Serra Sul, Carajás in northern Brazil. Titanium mobility, acid-induced weathering experiments and the preservation of abundant microfossils highlights the important role microorganisms play in mineral weathering and formation of new cements. Minor and trace element analysis of the grain and cement portions of the canga reveals the influx of the exogenous materials to the canga profile. The transport, deposition and incorporation of lithic elements, tentatively suggests the Serra Sul plateaus have undergone some relief inversion. Markedly increased phosphorus concentrations in the canga cements also indicate a possible renewable aeolian phosphorus source, in line with Amazonian phosphorus sources being derived from Africa.

Acknowledgements We acknowledge support from the Vale S.A.-UQ Geomicrobiology initiative and the Australian Research Council Linkage Program (LP140100805) to G. Southam and P. Vasconcelos. This research was undertaken on the XFM beamline at the Australian Synchrotron, part of ANSTO. We thank Daryl Howard for assisting with XFM and XANES data collection at the Australian Synchrotron. The authors acknowledge the facilities and the scientific and technical assistance of the Australian Microscopy and Microanalysis Research Facility at the Centre of Microscopy and Microanalysis, at the University of Queensland. We wish to acknowledge the assistance of the staff of Central Analytical Research Facility (CARF) at Queensland University of Technology (QUT) for access to analytical instrumentation, supported by the Faculty of Science and Engineering at QUT. Alan Levett acknowledges the support from the Australian Government Research Training Program.

References Barkley AE, Prospero JM, Mahowald N, Hamilton DS, Popendorf KJ, Oehlert AM, et al. African biomass burning is a substantial source of phosphorus deposition to the

153 Chapter Six

Amazon, Tropical Atlantic Ocean, and Southern Ocean. Proceedings of the National Academy of Sciences USA 2019; 116: 16216-16221.

Berry AJ, Stewart GA, O'Neill HSC, Mallmann G, Mosselmans JFW. A re-assessment of the oxidation state of iron in MORB glasses. Earth and Planetary Science Letters 2018; 483: 114-123.

Beukes NJ, Dorland H, Gutzmer J, Nedachi M, Ohmoto H. Tropical laterites, life on land, and the history of atmospheric oxygen in the Paleoproterozoic. Geology 2002; 30: 491-494.

Beverskog B, Puigdomenech I. Revised pourbaix diagrams for chromium at 25–300 ºC. Corrosion Science 1997; 39: 43-57.

Brimhall GH, Lewis CJ, Ague JJ, Dietrich WE, Hampel J, Teague T, et al. Metal enrichment in bauxites by deposition of chemically mature aeolian dust. Nature 1988; 333: 819.

Dalstra H, Guedes S. Giant hydrothermal hematite deposits with Mg-Fe metasomatism: a comparison of the Carajás, Hamersley, and other iron ores. Economic Geology 2004; 99: 1793-1800.

Dorr JVN. Supergene iron ores of Minas Gerais, Brazil. Economic Geology 1964; 59: 1203- 1240.

Dorr JVN. Iron-formation in South America. Economic Geology 1973; 68: 1005-1022.

Dukino RD, England BM, Kneeshaw M. Phosphorus distribution in BIF-derived iron ores of Hamersley Province, Western Australia. Applied Earth Science 2000; 109: 168-176.

Feio G, Dall’Agnol R, Dantas E, Macambira M, Santos J, Althoff F, et al. Archean granitoid magmatism in the Canaã dos Carajás area: implications for crustal evolution of the Carajás province, Amazonian craton, Brazil. Precambrian Research 2013; 227: 157- 185.

Frierdich AJ, Catalano JG. Controls on Fe (II)-activated trace element release from goethite and hematite. Environmental Science & Technology 2012; 46: 1519-1526.

Grainger CJ, Groves DI, Tallarico FHB, Fletcher IR. Metallogenesis of the Carajás Mineral Province, Southern Amazon Craton, Brazil: varying styles of Archean through Paleoproterozoic to Neoproterozoic base- and precious-metal mineralisation. Ore Geology Reviews 2008; 33: 451-489.

Gutzmer J, Chisonga BC, Beukes NJ, Mukhopadhyay J. The geochemistry of banded iron formation-hosted high-grade hematite-martite iron ores. Reviews in Economic Geology 2008; 15: 157-183.

154

Trace element behaviour in canga horizons

Hill I, Worden R, Meighan I. Geochemical evolution of a palaeolaterite: the Interbasaltic Formation, Northern Ireland. Chemical Geology 2000; 166: 65-84.

Jacobi CM, Carmo FF. Life-forms, pollination and seed dispersal syndromes in plant communities on ironstone outcrops, SE Brazil. Acta Botanica Brasilica 2011; 25: 395- 412.

Jacobi CM, Carmo FF, Vincent RC, Stehmann JR. Plant communities on ironstone outcrops: a diverse and endangered Brazilian ecosystem. Biodiversity and Conservation 2007; 16: 2185-2200.

Jones MWM, Hare DJ, James SA, de Jonge MD, McColl G. Radiation dose limits for bioanalytical X-ray fluorescence microscopy. Analytical Chemistry 2017; 89: 12168- 12175.

Klein C, Ladeira EA. Petrography and geochemistry of the least altered banded iron-formation of the Archean Carajás Formation, Northern Brazil. Economic Geology 2002; 97: 643- 651.

Knipping JL, Bilenker LD, Simon AC, Reich M, Barra F, Deditius AP, et al. Trace elements in magnetite from massive iron oxide-apatite deposits indicate a combined formation by igneous and magmatic-hydrothermal processes. Geochimica et Cosmochimica Acta 2015; 171: 15-38.

Latta DE, Bachman JE, Scherer MM. Fe electron transfer and atom exchange in goethite: influence of Al-substitution and anion sorption. Environmental Science & Technology 2012; 46: 10614-10623.

Levett A, Gagen E, Shuster J, Rintoul L, Tobin M, Vongsvivut J, et al. Evidence of biogeochemical processes in iron duricrust formation. Journal of South American Earth Sciences 2016; 71: 131-142.

Levett A, Gagen EJ, Diao H, Guagliardo P, Rintoul L, Paz A, et al. The role of aluminium in the preservation of microbial biosignatures. Geoscience Frontiers 2019a; 10: 1125- 1138.

Levett A, Gagen EJ, Southam G. Small but mighty: microorganisms offer inspiration for mine remediation and waste stabilisation. Microbiology Australia 2019b; 40: 190-194.

Lovley DR, Phillips EJP. Rapid assay for microbially reducible ferric iron in aquatic sediments. Applied and Environmental Microbiology 1987; 53: 1536-1540.

Mazanec K, Kocur M, Martinec T. Electron microscopy of ultrathin sections of Sporosarcina ureae. Journal of Bacteriology 1965; 90: 808-816.

Messias M, Leite M, Meira Neto J, Kozovits A, Tavares R. Soil-vegetation relationship in quartzitic and ferruginous Brazilian rocky outcrops. Folia Geobotanica 2013; 48: 509- 521.

155 Chapter Six

Monteiro HS, Vasconcelos PM, Farley KA, Spier CA, Mello CL. (U–Th)/He geochronology of goethite and the origin and evolution of cangas. Geochimica et Cosmochimica Acta 2014; 131: 267-289.

Morris R. Detrital iron deposits of the Hamersley Province. AMIRA-CSIRO Iron Ores of the Hamersley Province Project P75G, report 1994; 76.

Mücke A, Cabral AR. Redox and nonredox reactions of magnetite and hematite in rocks. Chemie der Erde-Geochemistry 2005; 65: 271-278.

Nadoll P, Angerer T, Mauk JL, French D, Walshe J. The chemistry of hydrothermal magnetite: a review. Ore Geology Reviews 2014; 61: 1-32.

Neaman A, Chorover J, Brantley SL. Implications of the evolution of organic acid moieties for basalt weathering over geological time. American Journal of Science 2005; 305: 147- 185.

Nedoseykina T, Kim MG, Park S-A, Kim H-S, Kim S-B, Cho J, et al. In situ X-ray absorption spectroscopic study for the electrochemical delithiation of a cathode LiFe0.4Mn0.6PO4 material. Electrochimica Acta 2010; 55: 8876-8882.

Nesbitt HW. Mobility and fractionation of rare earth elements during weathering of a granodiorite. Nature 1979; 279: 206-210.

Neumann G, Römheld V. Root excretion of carboxylic acids and protons in phosphorus- deficient plants. Plant and Soil 1999; 211: 121-130.

Ohmoto H. Nonredox transformations of magnetite-hematite in hydrothermal systems. Economic Geology 2003; 98: 157-161.

Okin GS, Mahowald N, Chadwick OA, Artaxo P. Impact of desert dust on the biogeochemistry of phosphorus in terrestrial ecosystems. Global Biogeochemical Cycles 2004; 18.

Olszewski WJ, Wirth KR, Gibbs AK, Gaudette HE. The age, origin, and tectonics of the Grão Pará Group and associated rocks, Serra dos Carajás, Brazil: Archean continental volcanism and rifting. Precambrian Research 1989; 42: 229-254.

Paterson D, Jonge MDd, Howard DL, Lewis W, McKinlay J, Starritt A, et al. The X‐ray fluorescence microscopy beamline at the Australian Synchrotron. AIP Conference Proceedings 2011; 1365: 219-222.

156

Trace element behaviour in canga horizons

Puzon GJ, Roberts AG, Kramer DM, Xun L. Formation of soluble organo− chromium (III) complexes after chromate reduction in the presence of cellular organics. Environmental Science and Technology 2005; 39: 2811-2817.

Ryan C, Siddons D, Kirkham R, Li Z, De Jonge M, Paterson D, et al. Maia X-ray fluorescence imaging: capturing detail in complex natural samples. Journal of Physics: Conference Series. 499. IOP Publishing, 2014, pp. 012002.

Sahoo PK, Souza-Filho PWM, Guimarães JTF, da Silva MS, Costa FR, Manes C-LdO, et al. Use of multi-proxy approaches to determine the origin and depositional processes in modern lacustrine sediments: Carajás Plateau, Southeastern Amazon, Brazil. Applied Geochemistry 2015; 52: 130-146.

Schulze DG, Schwertmann U. The influence of aluminium on iron oxides; X, Properties of Al- substituted goethites. Clay Minerals 1984; 19: 521-539.

Siddons DP, Kirkham R, Ryan CG, De Geronimo G, Dragone A, Kuczewski AJ, et al. Maia X-ray microprobe detector array system. Journal of Physics: Conference Series 2014; 499: 012001.

Silva RE, Hagemann S, Lobato L, Rosiere C, Banks D, Davidson G, et al. Hydrothermal fluid processes and evolution of the giant serra norte jaspilite-hosted iron ore deposits, Carajás mineral Province, Brazil. Economic Geology and the Bulletin of the Society of Economic Geologists 2013; 108: 739-779.

Skomurski FN, Rosso KM, Krupka KM, McGrail BP. Technetium incorporation into hematite (α-Fe2O3). Environmental Science & Technology 2010; 44: 5855-5861.

157 Chapter Six

Tolbert GE, Tremaine JW, Melcher GC, Gomes CB. The recently discovered Serra dos Carajás iron deposits, Northern Brazil. Economic Geology 1971; 66: 985-994.

Viana-Silva FE, Jacobi CM. Myrmecofauna of ironstone outcrops: composition and diversity. Neotropical Entomology 2012; 41: 263-71.

Walker T, Syers JK. The fate of phosphorus during pedogenesis. Geoderma 1976; 15: 1-19.

158

Chapter 7 Biogeochemical cycling of iron: implications for biocementation and slope stabilisation

Alan Levetta*, Emma J. Gagena, Paulo M. Vasconcelosa, Yitian Zhaob, Anat Paza, Gordon Southama

a School of Earth and Environmental Sciences, University of Queensland, St. Lucia, QLD, Australia b School of Mechanical and Mining Engineering, University of Queensland, St. Lucia, QLD, Australia

Citation for this publication:

Chapter Seven

Abstract Microbial biofilms growing in iron-rich seeps surrounding Lake Violão, Carajás, Brazil serve as a superb natural system to study the role of iron cycling in producing secondary iron cements. These seeps flow across iron duricrusts (referred to as canga in Brazil) into hydraulically restricted lakes in northern Brazil. Canga caps all of the iron ore deposits in Brazil, protecting them from being destroyed by erosion in this active weathering environment. Biofilm samples collected from these seeps demonstrated heightened biogeochemical iron cycling, contributing to the relatively rapid, seasonal formation of iron-rich cements. The seeps support iron-oxidising lineages including Sideroxydans, Gallionella, and an Azoarcus species revealed by 16S rRNA gene sequencing. In contrast, a low relative abundance of putative iron reducers; for example, Geobacter species (<5% of total sequences in any sample), corresponds to carbon limitation in this canga-associated ecosystem. This carbon limitation is likely to restrict anoxic niches to within biofilms. Examination of a canga rock sample collected from the edge of Lake Violão revealed an array of well- to poorly-preserved microbial fossils in secondary iron cements. These heterogeneous cements preserved bacterial cell envelopes and possibly extracellular polymeric substances within the microfossil iron-rich cements (termed biocements). Bacterial iron reduction initiates the sequence, and intuitively is the rate-limiting step in this broadly aerobic environment. The organic framework of the active- and paleo- biofilms appear to provide a scaffold for the formation of some cements within canga and likely expedites cement formation. The accelerated development of these resilient iron-rich biocements in the lake edge environment compared with surroundings duricrust-associated environments may provide insights into new approaches to remediate mined land, aiding to stabilise slopes, reduce erosion, restore functional hydrogeology and provide a substrate akin to natural canga for revegetation using endemic canga plant species, which have adapted to grow on iron-rich substrates.

160

Lake-edge field characterisation

7.1. Introduction Some of the world’s longest-lived landforms resist erosion because they are iron-cemented (Monteiro et al., 2014; Shuster et al., 2005). To strengthen these erosion-resistant landforms, physical weathering and erosion must be combatted by the formation of new cements. These cements form as a result of the geochemical cycling of cations, particularly iron, aluminium and silicon, from the micro- to the macroscale. Geochemical cycling of cations at the macroscale promotes weathering and cementation at distinct landscape positions. For example, during the weathering of banded iron formations (BIFs) in semi-arid environments, silicon may be leached from the BIF and transported to lower horizons where it redeposits, stabilising these lower landforms (Morris, 1983). In contrast, the physiochemical conditions that influence the solubility of iron (both ferrous and ferric) in tropical climates are more complex, leading to short- and long-range iron transport (Yamaguchi et al., 2007). Biological processes that influence redox conditions, including availability of complexing agents and reductants, alter the geochemical cycling of iron (Colombo et al., 2014). For short-range geochemical cycling of iron, these erosion-resistant horizons must contain distinct microenvironments; some that promote mineral dissolution and others conducive to precipitation (Yamaguchi et al., 2007).

Landscapes degraded by anthropogenic activity, including land cleared for road construction, building and mining, may, in principle, be stabilised by iron cements. Understanding these natural systems, where weathering and subsequent re-precipitation of iron minerals contributes to long-term stabilisation of the landscape, provides valuable insights from which to glean strategies for effective surface stabilisation.

Microbially-promoted stabilisation techniques typically target calcium carbonate precipitation for cement formation (DeJong et al., 2010); however, iron biogeochemical cycling may provide a more chemically resistant alternate for landform stabilisation. Iron-reducing microorganisms utilise a variety of mechanisms including outer membrane c-type cytochromes (Lovley et al., 2004; Richardson, 2000), nanowire production (Reguera et al., 2005) and soluble redox-active small molecules (for example, flavins) (Breuer et al., 2015) to shuttle electrons to iron (III) minerals, placing ferrous iron into solution. In circumneutral environments, ferrous iron is relatively soluble compared with ferric iron. In oxic solutions, ferrous iron is rapidly oxidised to ferric iron, which is unstable and precipitates as hydrous ferric oxides (Rentz et al., 2007). Therefore, iron-reducing microorganisms and changes in oxidation potential at a constant circumneutral pH can promote the geochemical cycling of iron between solution and mineral precipitates at the microscale (Roden et al., 2004).

161 Chapter Seven

In the present study, we sought to understand the contributions of natural biogeochemical iron cycling to the cementation of ferruginous (iron cemented) duricrusts in the iron ore ridges at Serra Sul (S11D), Serra dos Carajás, Pará, Brazil. We focused on lake edge environments as biological ‘hotspots’ where iron dissolution and re-precipitation is active in the present day.

7.2. Geological and environmental setting Surface water runoff into local hydrologically restricted dissolution (karstic) depressions within the ferruginous duricrust in Serra Sul contributes to the formation of lake environments such as Lake Violão and Lake Amendoim (Fig. 7.1). In some locations, seeps that flow into the lake support thick biofilms interspersed with fresh iron oxide precipitates (Fig. 7.2). The biofilms and the surrounding rocks suggests that these microbial ‘hotspots’ promote the biogeochemical cycling of iron.

Lake Violão and Lake Amendoim sit at an elevation between approximately 700 - 720 m. Sahoo et al. (2017) recently characterised the geochemistry of the lakes and Silva et al. (2018) described the local geomorphology, which have developed over lateritic ferruginous duricrusts that cap high grade iron ore. Lake Violão ranges from oligotrophic in the dry season (June to October) towards eutrophic in the wet season (November to May) whereas Lake Amendoim is classified as ultra- to slightly-oligotrophic throughout the year (Sahoo et al., 2017). Lake Violão has a surface area of approximately 0.27 km2, 7.77 km circumference and a catchment area of approximately 1.83 km2, whereas Lake Amendoim is slightly smaller; surface area, perimeter and basin catchment area are 0.126 km2, 6.89 km and 1.20 km2, respectively (Silva et al., 2018). The climate in the Serra Sul is tropical, with average daily-high temperatures of ~26 °C year round and an annual rainfall between 1800 – 2300 mm, predominately (approximately 80%) falling in the wet season (Silva et al., 2018). The average water temperatures for the lakes is typically ~ 27 C; pH (6.3 – 7.9), dissolved oxygen (7.1 – 7.9 mg/L), electrical conductivity (6.1 – 6.9 s/cm) and oxidation-reduction potential (259 – 389 mV) do not vary significantly between wet and dry seasons and are not stratified throughout the major lake bodies (Sahoo et al., 2016; Sahoo et al., 2015). The lakes are currently protected by Brazilian Law restricting mining (Silva et al., 2018) but the nature of the ore body and recent developments in iron ore mining technologies that reduce environmental impact have allowed the development of the world’s largest iron ore mine (S11D) in close proximity to the protracted lakes (Fig. 7.1). The ferruginous duricrusts predominately support shrubby and herbaceous rupestrian plant species with less vegetation (Fig. 7.1) compared with the surrounding montane and Capão Island forests (Nunes et al., 2015).

162

Lake-edge field characterisation

Fig. 7.1. (A) Google Earth image highlighting the canga plateaus of the Serra Sul (S11 regions) de Carajás Mineral Province, Pará, Brazil. Within the canga plateau are Lakes Violão and Amendoim (sample locations indicated for each), with the S11D mine towards the East. (B) Photograph of the edge of Lake Violão highlights the red-stained rocks where streams flow over the canga into the lake.

163 Chapter Seven

Each lake has steeply dipping sides, typically greater than 20 and occasionally approaching 50 (Silva et al., 2018). The steeply dipping slopes along the lake edges are primarily composed of detrital fragments of hematite cemented together by thin (mm-scale) authigenic goethite. The low cement:fragment ratios in these rocks are consistent with Dorr’s (1964) description of a rock called ‘canga rica.’ The thin section of the canga sample from the edge of Lake Violão highlights the low cement:fragment ratio (Fig. 7.3A). The cements are heterogeneous and deceptively robust: a well-directed, firm strike with a rock hammer is required to fracture the rock along planes of weakness in the cements. The canga rica preferentially fractures unevenly into sheet-like rock samples, typically less than 5 cm thick. The steeply dipping canga rica that forms at the edge of Lakes Violão and Amendoim is of particular interest to this study as it highlights that ferruginous cements can be formed relatively quickly, stabilising loose fragments on the slope that would otherwise be eroded into the lake. The water chemistry and microbial populations of streams that flow into the lakes (Fig. 7.2) as well as the canga rica substrate that forms the lake edge (Fig. 7.1B) were examined at the microscale to understand the role that microorganisms may play in the formation and evolution of these ferruginous cements.

164

Lake-edge field characterisation

Fig. 7.2. Photographs of microbial biofilms that grew at the edges of Lake Amendoim (A and B) and Lake Violão (C – F). (A) Microbial ‘globular’ biofilms form along grass roots that grow next to the stream within a relatively stagnant pool. (B) Fresh dark (black) biofilm next to the stream was sampled. Dried, flaky biofilm coated the canga above the stream. (C) An aqueous seep that flowed through the canga (karst-like feature) stained the rock below red. No iron in solution was detected here. (D) Iron-stained microbial biofilms coated the rocks in regions where organic matter accumulates within a stream that flowed in Lake Violão. (E) Foam that accumulated at the edge of Lake Violão. (F) An iron-stained microbial biofilm within a stream.

165 Chapter Seven

7.3. Methods and materials 7.3.1. Locations for sample collection Sampling locations for biological ‘hotspots’ of iron cycling, was guided by the presence of aqueous ferrous iron. To indicate where aqueous ferrous iron in solution exceeded 1 mg L-1, approximately 0.1 mL of water was pipetted into 1 mL of 1 g L-1 ferrozine, a modification of the ferrozine assay developed by Stookey (1970). Positive ferrozine assays indicating ferrous iron concentration greater than 1 mg L-1 were recorded for seeps at Lake Amendoim (AS-GB; Fig. 7.2A and AS-BlB; Fig. 7.2B), the surface and bottom of an ephemeral pool (VP-SB and VP-BB) and two of the three seeps sampled at Lake Violão (VS-OB; Fig. 7.2D and VS-FeS; Fig. 7.2F). Two sample locations, including one of the seeps at Lake Violão (VS-RB; Fig. 7.2C) and ‘foam’ that formed by wind blowing across the lake (LV-F; Fig. 7.2E) did not produce a strong ferrozine reaction, indicating ferrous iron in solution was less than 1 mg L-1. Note that these data were only used to target hotspots of iron-cycling and a complete set of samples for water chemistry, microscopy (light and electron microscopy; data not shown) and DNA sequencing were subsequently collected. Biofilm samples were collected from the field into sterile Falcon tubes (15 and 50 mL), with DNA preserved by adding LifeGuardTM to a final concentration of approximately 10% (aq). All samples were kept cool (~ 4 C) in the field (less than eight hours) until they could be frozen. To collected corresponding samples for water chemistry, approximately 10 mL of water was filtered through a 0.22 m syringe filter into sealed serum vials containing nitrogen gas phase (and using a vent needle to avoid pressurisation during filtration) in the field to minimise ferrous iron oxidation and precipitation out of solution.

Where water flowed down the slope away from the orange-coloured biofilms, additional water samples were collected into a nitrogen filled vial to monitor the iron in solution. In addition, where water appeared relatively stagnant (for example, an ephemeral pool within the duricrust) a dissolved oxygen profile was taken using a FireSting GO2 (Optical Oxygen Meter, Pyroscience GmbH, Aachen, Germany) with a robust probe (3 mm diameter). The water pH was measured in the field for locations using MColorpHastTM pH strips 4.0 – 7.0 (Merck KGa, Darmstadt, Germany). A rock sample from the edge of Lake Violão (Fig. 7.1B) was also collected for scanning electron microscopy (Figs. 7.3 – 7.4).

7.3.2. Scanning electron microscopy A representative canga rica rock sample fractured from the lake edge around Lake Violão was dehydrated in a 40 °C oven overnight and embedded in an epoxy resin (EpoxiCure 2) to

166

Lake-edge field characterisation produce a petrographic thin section, which was polished to a thickness of 100 m. A final 0.25 m diamond abrasive was used to achieve a submicron polish for examination using a scanning electron microscope (SEM). A JEOL7100 SEM in backscattered electron mode with an accelerating voltage of 15 kV was used to examine the petrographic thin section. The sample was degassed at 50 °C overnight and coated with 10 nm iridium using a Quorum Q150T sputter coater prior to examination. ImageJ was used to estimate the surface area cement:fragment ratio for a representative portion of the cross-section (Fig. 7.3A; black rectangle).

7.3.3. Nanoindentation Nanoindentation tests were used to characterise the mechanical properties of various cements and a hematite-rich fragment throughout the canga thin section using A TI-900 Hysitron Triboindentor® equipped with a diamond Berkovich tip (radius ca. 100 nm). A load-control function with loading, holding and unloading times of 10 s, 10 s and 15 s, respectively. An indentation load of 5 mN was used to determine the hardness (H) and reduced modulus (Er), which were calculated based on the load-displacement (P-h) curves. Two textures of goethite- rich lamella cements were tested: ‘abiotic’ mineral precipitates (goethite cements) and those that appear to be influenced by microorganisms (microbial cements). For each of these textures, 4 × 4 indentation arrays with 10 m spacing between each indent were used. Dissolution pitting within the hematite fragment required indent locations to be selected individually on polished surfaces, with each indentation location more than 10 m apart. The heterogeneous nature of the sample at a scale of 100 nm occasionally resulted in individual loading curves being unreproducible and these were subsequently removed for statistical analysis. A two-tailed t- test assuming equal variance indicated there was no statistical difference between the hardness (p = 0.37) and the reduced modulus (p = 0.94) for the different arrays of microfossil cements tested; therefore, these data were combined for statistical analysis.

7.3.4. Water chemistry Water samples were kept at room temperature (~ 25 C) and transferred to the laboratory, where they were acidified to a final HNO3 concentration of 7% (aq) using concentrated (70% (aq))

HNO3. Water samples were transferred to Teflon tubes and digested in a MARS Xpress microwave at 160 °C for 10 min and 170 °C for a further 10 min. Digested samples were diluted to a final concentration of 5% (aq) HNO3 and analysed using a Perkin Elmer Optima 7300DV inductively coupled plasma optical emission spectroscopy with an argon plasma gas (15 L min- 1). Samples were analysed for the following soluble elements (detection limit in ppm provided

167 Chapter Seven in parentheses): aluminium (0.0012), barium (0.00004), calcium (0.0005), iron (0.0003), potassium (0.0003), magnesium (0.0001), sodium (0.0002), sulphur (0.0002) and zinc (0.0002).

7.3.5. DNA extraction, sequencing and sequence analysis To extract DNA, samples were thawed and centrifuged at 10,000×g for 10 min, the supernatant was removed and the remaining biomass was collected for DNA extraction following the DNeasy Powersoil Kit (Qiagen, Hilden, Germany).

The V6 – V8 portion of the 16S rRNA gene was amplified from the extracted DNA using the universal primers 926f and 1392r, modified to contain Illumina specific adapter sequence; 926F: 5’- TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGAAACTYAAAKGAATTGRCGG- 3’ and 1392wR: 5’- GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGACGGGCGGTGWGTRC-3’. Libraries were prepared using the workflow outlined by Illumina (#15044223 Rev.B), with the exception that the NEBNext® UltraTM II Q5 Mastermix (New England Biolabs #M0544) was used in standard PCR conditions. Agencourt AMPure XP beads (Beckman Coulter, Brea, CA, USA) were used to purify the resulting PCR amplicons before the purified DNA was indexed using the Illumina Nextera XT 834 sample Index Kit A-D (Illumina FC-131-1002, San Diego, CA, USA) in standard PCR conditions (aforementioned PCR mastermix) to assign unique 8- bp barcodes. Indexed amplicons were pooled in equimolar concentrations and sequenced using a MiSeq Sequencing System (Illumina) with paired end sequencing (V3 300-bp chemistry) in accordance with the manufacture’s protocol at the Australian Centre for Ecogenomics, The University of Queensland.

Sequences were processed using MOTHUR as per the MiSEQ standard operating procedures (https://www.mothur.org/wiki/MiSeq_SOP, accessed on 20th April, 2019) with minor alterations to allow for processing forward reads only. Briefly, forward reads were trimmed on quality using a quality average of 35 across a sliding window size of 50 nt before the primer was removed. Reads were subsequently trimmed to 250 nt and ambiguous bases and homopolymeric structures (8 nt repeats) were removed before alignment using the SILVA SSU Ref NR 99 v132 reference database (Pruesse et al., 2007). Putative chimeras identified using the silva132.gold alignment database as a reference were removed and the remaining sequences were classified using the SILVA SSU Ref NR 99 v132 database. Sequences classified as

168

Lake-edge field characterisation

Eukaryota, Chloroplast, Mitochondria or unknown were also removed. Sequences were clustered into Operational Taxonomic Units (OTUs) at a distance of ≤ 0.03 for further processing. Libraries were subsampled to the smallest number of unique OTUs (8128 OTUs) and sample coverage, species richness (Sobs indicator), Shannon-Weaver diversity index and Shannon evenness index were determined using the calculators in MOTHUR. Singletons (OTUs that only occur once in the entire dataset) were then removed before generating a heatmap to visualise shared OTUs. Nucleotide basic local alignment search tool (BLASTn) (Altschul et al., 1990), using the NCBI and non-redundant nucleotide collection, was used to compare the major OTUs with publicly available sequences. Sequences were initially compared by excluding uncultured and environmental samples. OTUs that demonstrated less than 97% identity with a named species were re-analysed including all publicly available sequences. Sequences have been submitted to the NCBI Sequence Read Archive under BioProject number PRJNA554967.

7.4. Results 7.4.1. Characterisation of canga rica from the edge of Lake Violão The canga that formed at the edge of Lake Violão was very competent (non-friable), with microscale fragments and secondary goethite-rich precipitates cementing large hematite-rich fragments, ranging from less than 1 cm to greater than 5 cm (Fig. 7.3A). The canga was primarily composed of the detrital fragments (surface area of approximately 77%) with the cement material only accounting for approximately 23% of the surface area of the sample (Fig. 7.3A).

Bacteriomorphic structures resembling previously characterised canga-associated microfossils (Levett et al., 2016; Levett et al., 2019) were commonly identified throughout the goethite-rich cements in the canga (Fig. 7.3). Fewer, relatively large microfossils (ranging from 3 – 5 µm) are interpreted as relicts of microbial eukaryotes; for example, algae or fungi (Fig. 7.3). Smaller, approximately 1 µm, microbial fossils are also evident in the pore spaces in proximity to the larger microfossils (Fig. 7.3; white arrows).

Encrusted cell envelopes are preserved throughout the cement portion of the canga rica (Fig. 7.4A; white arrows). These bacteriomorphic cell envelope structures are relatively sparse compared with other biofilms identified in canga (Levett et al., 2016) but are frequently identified and give much of the cement throughout the canga rica sample a ‘biological-texture.’ For example, microbial fossils are frequently identified in the iron cements between the large

169 Chapter Seven detrital fragments (Fig. 7.4B; within the black portion of the black-and-yellow line traces). Lamellae are also common throughout the biologically-textured cements (Fig. 7.4B; white arrows). The fragment has also undergone weathering, evidenced by some of the smaller portions of the fragment maintaining limited connections with original fragment (Fig. 7.4B; dashed black line).

Spherical, rod-shaped and elongated structures are remnants of cocci, bacilli and filamentous structures that are preserved throughout the iron-rich cements within the canga (Fig. 7.5A), which can form large (Fig. 7.5B) and relatively robust structures that cement large fragments (Fig. 7.3). The average hardness of abiotic cements (goethite cements = 7.5 ± 0.6 GPa) was significantly (p < 0.001) harder than cements influenced by microorganisms (microfossil cements = 1.4 ± 0.4 GPa; Fig. 7.5B). As a comparison, the average hardness for the hematite- rich fragment was 13.9 ± 0.8 GPa and the resin was 0.21 ± 0.0017 GPa (Fig. 7.5B).

170

Lake-edge field characterisation

Fig. 7.3. Photograph of the polished section of the canga that formed at the edge of Lake Violão. The black rectangle highlights the representative region used to calculate cement (~23%) to fragment (~77%) surface area ratio. The low cement to fragment ratio is consistent with Dorr’s (1964) description of a rock called ‘canga rica.’ (B) Backscattered electron micrograph of large microbial fossils and small encrusted cell envelopes within the pore spaces (white arrows) that contribute to the formation of cements within the canga rica.

171 Chapter Seven

Fig. 7.4. (A) Backscattered scanning electron micrograph revealing limited numbers of sporadically emplaced cell envelope structures within heterogeneous cements. (B) Low magnification backscattered scanning electron micrograph of cements forming next to a large, weathered fragment within the canga. The inner black portion of the black-and-yellow line traces approximately outlines the heterogeneous cements that contain microbial remnants and are considered to have a ‘biological texture’.

172

Lake-edge field characterisation

Fig. 7.5. Backscattered scanning electron micrograph of a meniscus cement within the canga rica sample. (A) Poorly preserved cocci, baccili and possible remnants of filamentous microfossils contribute to the formation of the ferruginous cement. (B) Low magnification micrograph highlighting the heterogeneous nature of the biologically-textured cements. Nanoindentation tests for the microbial cements, neighbouring goethite cements, hematite-rich fragment (highlighted in Fig. 7.3A) and resin (n – number of tests, x̅ H – average hardness, 휎H – standard deviation for hardness, x̅ Er – average reduced modulus, 휎H – standard deviation for reduced modulus).

173 Chapter Seven

7.4.2. Water chemistry Aqueous iron concentrations, presumably as ferrous iron given the circumneutral pH conditions (pH 5 – 6) and a strong reaction with ferrozine in the field, ranged from 15.8 ppm to below detection limit (Table 7.1). Soluble iron was below detection limit at the perimeter of Lake Violão associated with the foam (Fig. 7.2E). Similarly, the seep (Fig. 7.2C) that flowed through the pore spaces in canga (for example, Fig. 7.3A) had no detectable soluble iron; however, the biofilm and the rock surface at the overflow contained fresh iron oxide precipitates (red), indicating oxidation of aqueous ferrous iron and subsequent precipitation.

The highest soluble iron concentrations were typically recorded in association with well- developed mm-scale microbial biofilms. Soluble iron decreased with increasing distance from the well-developed microbial biofilms (Table 7.1; VS-OB samples and VS-FeS samples). Soluble iron concentrations did not appear to change with water depth in the stream pool or in the oxidised ephemeral pool that was sampled, indicating effective mixing of solutions within this water body (Table 7.1; AS-GB samples and VS-SB – VS-BB). Dissolved oxygen concentrations were relatively high in this small water body, maintaining 20% air saturation even at a depth of approximately 15 cm within an ephemeral pool (Table 7.1; sample 21E). The alkaline-earth (magnesium and calcium) and alkaline (sodium and potassium) metals were enriched within the ephemeral pool compared with all other sample locations.

174

Lake-edge field characterisation

Table 7.1. Water chemistry for sampled locations. Cation concentrations determined using ICP- OES. A dash (-) is used to indicate were a measurement was not taken.

175 Chapter Seven

7.4.3. Microbial community structures Proteobacteria were a dominant phylum in all samples, accounting for 26 – 72% of all sequences per sample (Fig. 7.6). Biofilms supported by the surface streams and seeps that flowed into Lake Violão (VS-RB, VS-OB and VS-FeS) also contained a relatively high proportion of Patescibacteria (11 – 31%) and bacteria unclassifiable below the phylum level (7 – 16%), with sample VS-RB also containing a relatively high proportion of Firmicutes (10%; Fig. 7.6). Aside from Proteobacteria, Planctomycetes (22%), Armatimonadetes (20%) and Actinobacteria (10%) were also dominant in Lake Violão (LV-F; Fig. 7.6). Other major phyla at the surface (VP-SB) and bottom (VP-BB) of the ephemeral pool samples were Actinobacteria (15%) and Planctomycetes (11%; Fig. 7.6). Strong similarities in the microbial community structure at the surface and bottom of the sampled ephemeral pool were consistent with mixing solutions and the absence of broad anoxic conditions (VP-SB and VP-BB; Fig. 7.6). In contrast, the black biofilm associated with the surface stream at Lake Amendoim (AS- BlB) contained abundant Acidobacteria (10%), while Plantomycetes (11%) and Firmicutes (10%) accounted for large proportions of the total sequences in the globular biofilm (AS-GB), which was sampled approximately 10 m downstream from the black biofilm (Fig. 7.6).

176

Lake-edge field characterisation

Fig. 7.6. Phylum level classification of unique 16S rRNA sequences associated with microbial communities and biofilms collected from steams and seeps around Lake Amendoim (AS-GB and AS-BlB), the surface (VP-SB) and bottom (VP-BB) of an oxidised ephemeral pool surrounding Lake Violão, at the edge of Lake Violão (LV-F) and streams and seeps that flowed into Lake Violão (VS-RB, VS-OB and VS-FeS). Phyla below 1.5% relative abundance are grouped into other including, Altiarchaeota, Archaea_unclassified, BRC1, Chlamydiae, Cloacimonetes, Crenarchaeota, Cyanobacteria, Deferribacteres, Deinococcus-Thermus, Dependentiae, Diapherotrites, Elusimicrobia, Epsilonbacteraeota, Euryarchaeota, FBP, FCPU426, Fibrobacteres, Firestonebacteria, Fusobacteria, GAL15, Gemmatimonadetes, Hydrogenedentes, Kiritimatiellaeota, Latescibacteria, Lentisphaerae, Margulisbacteria, Nitrospinae, Nitrospirae, Omnitrophicaeota, Rokubacteria, Spirochaetes, Tenericutes, Thaumarchaeota, WOR-1, WPS-2, WS1, WS4, Zixibacteria.

177 Chapter Seven

Sequencing coverage of the samples ranged from 61 to 98% (Appendix 2). The Shannon- Weaver diversity index ranged from 3.98 for iron foam in Lake Violao to >7 for an iron seep flowing into the same lake (Appendix 2). The Shannon evenness index reflected the same trend, with the least even microbial community the lake foam (0.58) and the most even microbial community the iron seep (0.89; Appendix 2). Of the eukaryotic sequences that were recovered with the primers used in this study (926f and 1392r; see Section 7.3.5.) the greatest proportion of eukaryotes to the total community was in the surface biofilm of the ephemeral pool (VP-SB; 17%). The foam from Lake Violão, the bottom biofilm of the ephemeral pool at Lake Violao and the globular biofilm from an iron seep flowing into Lake Amendoim had between 5 and 9% eukaryotesin the total sequences. All other biofilms associated with iron seeps contained fewer than 5% eukaryotic sequences (Appendix 2).

Microbial lineages affiliated with well-known iron-oxidising microorganisms were present throughout water stream samples from Lake Amendoim (AS-GB and AS-BlB) and Lake Violão (VS-RB, VS-OB and VS-FeS; Fig. 7.7). For example, OTU008 demonstrated 100% identity across the sequenced region of the 16S rRNA gene to a well-characterised neutrophilic iron-oxidising microorganism, Sideroxydans lithotrophicus (Weiss et al., 2007) and was present in all surface stream samples (Fig. 7.7). OTU034, OTU038 and OTU040 present in the surface stream (LS and VS) samples, also appear to be members of a well-described family of iron- and sulphur-oxidisers, the Gallionellaceae (Fig. 7.7). OTU038 was also detected in the ephemeral pool (VP) samples. OTU 010 and OTU 042 demonstrated 98% identity across the sequenced region of the 16S rRNA gene to Burkholderiales bacterium GJ-E10, a recently described chemolithotrophic iron-oxidising microorganism (Fukushima et al., 2015). OTU011, a dominant lineage from a surface stream at Lake Violão (sample VS-FeS; Fig. 7.7), may represent a novel iron-oxidising lineage. There is genomic evidence that the nearest named isolate (95% identity across the sequenced 16S rRNA gene region to Azoarcus sp. CIB CP011072) has iron oxidation capabilities. Specifically, the genome of Azoarcus sp. CIB encodes three decaheme cytochromes that demonstrate very high homology to MtoA c-type cytochromes (e-value = 1e-126 to MtoA Gallionella capsiferriformans ES-2) and each of these genes are followed by genes encoding proteins demonstrating high homology with MtoD (e- value = 2e-34 to Sideroxydans lithotrophicus MtoD). MtoA and MtoD are the key enzymes in electron transport for iron oxidation (Liu et al., 2018a).

Known iron-reducing lineages in canga-associated surface waters were less abundant than putative iron oxidisers. OTUs 025 and 049, present in the surface streams (AS-GB, AS-BlB,

178

Lake-edge field characterisation

VS-RB, VS-OB and VS-FeS) but absent from the ephemeral pool (VP-SB and VP-BB) and the Lake Violão (LV-F) samples, demonstrated strong (>99%) 16S rRNA gene identity to the known iron-reducing lineage Geobacter (Fig. 7.7). Well-described iron-cycling microorganisms were less prevalent in the sample collected from the boundary of Lake Violão (LV-F) and the oxidised ephemeral pool (VP-SB and VP-BB).

179 Chapter Seven

Fig. 7.7. Heatmap analysis of 16S rRNA gene sequences showing the most abundant OTUs (OTU clustered at a distance ≤ 0.03) in each of the biofilms that formed in the streams that feed Lake Amedoim and Lake Violão and within Lake Violão. The scale bar represents relative abundance within samples from red (most abundant) to black (least abundant), with white indicating undetected OTUs. The nearest named representative in the public domain and its accession number are provided for each OTU, with blue text used to highlight putative iron-oxidising lineages and red for putative iron-reducing lineages.

180

Lake-edge field characterisation

7.5. Discussion The iron oxide cements that consolidate the ferruginous duricrust at the edge of Lake Violão appear to form relatively quickly, possibly within decades to millennia rather than on a scale of millions of years as demonstrated for typical canga horizons (Monteiro et al., 2014). The steeply dipping slopes at the lakes’ edges together with monsoonal rainfall during wet seasons would quickly erode loose detrital fragments into the lakes if cements were not formed relatively rapidly. The low cement:fragment ratio of the canga rica (approximately 1 part cement to 3 parts fragment) at the lake edge also supports a relatively rapid formation of cements. In addition, the canga rica that forms at the lakes’ edges preferentially fractures in sheet-like rock samples (less than 5 cm thick), indicating that these surface rocks had been conglomerated more recently and independently from the underlying duricrust.

Microorganisms appear to play an important role in relatively rapid formation of these cements at the edges of both Lakes Amendoim and Violão. An array of well- to poorly- preserved microbial fossils, commonly observed throughout the goethitic cements, highlights the involvement of microorganisms in the cementation of canga at the lake edge (Figs. 7.3 – 7.5). The microbial biofilms that grow throughout the seeps and streams at the lakes’ edges appear to aggregate detrital fragments and provide an organic framework for the nucleation of authigenic hydrous ferric oxide minerals. As the microbes become fossilised, they contribute to the formation of robust cements within the canga. These organic frameworks appear to expedite the formation of cements throughout canga rica; in their absence, the soluble iron in solution would simply precipitate out to coat fragments and form ferruginous pisolith structures instead of robust cements.

The formation of ferruginised microbial cements within the canga is likely to be limited by the reduction of iron oxide minerals (Gagen et al., 2018). Microbial biofilms in the streams and seeps that supplement Lakes Violão and Amendoim coincide with increased soluble iron concentrations. Up to 15 ppm of iron in solution was recorded in association with extensive biofilm, indicating rapid iron cycling. In comparison, the average dissolved iron in the lakes is less than 0.5 ppm (Sahoo et al., 2017). The gentle flow of surface water and the small size of the seeps are also likely to allow for the development of extensive biofilms compared to within the open lake. These extensive biofilms likely create anoxic niches that accommodate and promote microbial-induced iron reduction. The low relative abundance of sequences affiliated with well-known iron-reducing microorganisms in the biofilms appears to be inconsistent with the elevated soluble iron concentrations (Fig. 7.7). However, the stoichiometry of the microbial

181 Chapter Seven iron reduction equation highlights that a single hydrated carbon molecule can account for up to four ferrous iron cations being placed into solution (Reaction 7.1). Therefore, a low concentration of hydrated carbon may support a relatively small but consistent population of iron-reducing microorganisms. Inorganic and organic acids generated by microbial processes may also promote the reductive dissolution of iron (III) oxide minerals (Schwertmann, 1991).

3+ 2+ + 퐶퐻2푂 + 4퐹푒 + 퐻2푂 → 4퐹푒 + 퐶푂2 + 4퐻 (7.1)

Together, the electron microscopy of the rock samples, the water chemistry and microbial community analysis highlight that microorganisms play important roles in the weathering of iron oxide minerals in canga, placing iron into solution. Microbial cell envelopes provide sites for the nucleation of iron oxide minerals to form stable ferruginous microbial cements. Each of these processes takes place within microenvironments within the canga horizon, creating a highly resistant, ‘self-healing’ duricrust (Monteiro et al., 2014).

7.5.1. Ferruginous biocement formation: industrial applications By understanding these natural microbial processes that contribute to the relatively rapid cementation, it is possible to develop new biotechnologies that target the biogeochemical cycling of iron to chemically and physically stabilise degraded landscapes. The redox-based biogeochemical cycling of iron that occurs throughout canga is accelerated within the lake edge environments. Within canga lake edge environments, surface water and carbon-rich nutrients from nearby plants (for example, Fig. 7.2F) support the enhanced growth of diverse microbial biofilms compared with the rest of the canga horizon. These microbial hotspots create an environment that accelerates iron cycling. Respiration of carbon sources from the nearby plants depletes dissolved oxygen and creates anoxic microenvironments, potentially isolated to within the biofilm (Table 7.1), accelerating the rate-limiting dissolution of iron oxide minerals (Gagen et al., 2018). Harnessing these natural process will contribute to the development of new biotechnologies with various applications. For example, these novel biocements may be used to physically and chemically stabilise hillslopes and create mineralised surfaces that restore the natural hydrogeology of landscapes degraded by anthropogenic activities. Given the chemical stability of goethitic cements in canga over millions of years (Monteiro et al., 2014), these biotechnologies also provide alternative solutions for protecting sulphide-based waste piles, reducing the generation of the acid mine drainage (Liu et al., 2018b). Accelerating the formation of long-term hydrogeochemically stable ferruginous biocements may provide effective chemical and physical barriers to protect sulfidic waste piles.

182

Lake-edge field characterisation

7.5.2. Limitations of natural iron reduction Trace element analysis demonstrates that the environments that contributed to canga formation and evolution are overall oxidising (Monteiro, 2017). Relatively high oxygen concentrations in the seeps studied here may indicate that available organic carbon is a limitation for the generation of broadly anoxic condition that are required for microbially-induced iron reduction. Carbon limitations for iron reduction are also supported by increased iron and aluminium in solution during the wet season compared with the dry season (Sahoo et al., 2017). In the natural environment, carbon limitations may be important for the re-oxidation of iron, preventing its loss through weathering channels. Gagen et al. (2018) postulated that carbon limitations and overall oxic conditions with iron-rich duricrust environments are critical for the net generation of iron oxide minerals. Therefore, broadly oxic conditions and carbon-limitations appear to be essential for the short-range biogeochemical cycling of iron that occurs in canga horizons and contributes to their extraordinary resistance to physical erosion (Monteiro et al., 2018). For industrial applications, these limitations may be overcome by isolating microbial iron reduction to optimised bioreactors before releasing the soluble iron into the broadly oxic environment to create chemically and physically stable surfaces.

7.5.3. Contributors to biocement formation: iron oxidation and microbial fossilisation: A number of putative iron-oxidising microorganisms, including Sideroxydans, Gallionella, and an Azoarcus species were identified in the biofilms. These putative iron-oxidising microbial lineages may be underrepresented in the microbial fossils preserved within the cements of the canga (Figs. 7.3 – 7.4). Chemolithotrophic microorganisms that enzymatically oxidise iron as an energy source have typically developed strategies to prevent becoming encrusted in iron oxide minerals (Chan et al., 2011; Hegler et al., 2010; Miot et al., 2009). Gallionella species commonly produce extracellular stalks as a nucleation site for hydrous ferric oxides (Chan et al., 2011). The poorly crystalline iron (III) oxide precipitates that may be produced by iron- oxidising microorganisms may also be readily reduced by iron-reducing bacteria; for example, Geobacter species within the biofilm (Roden, 2003; Roden, 2012; Roden et al., 2004). Therefore, microbially induced iron oxide minerals are unlikely to contribute to biocement formation in canga unless they are no longer viable or inactive.

Instead, the microbial fossils preserved within the canga are likely to be non-iron-oxidising lineages within the microbial community that have not evolved mechanisms to avoid encrustation (Loiselle et al., 2018). Microbial cell envelopes have a net negative charge, forming electrostatic interactions with iron in solution, which together with the low solubility

183 Chapter Seven of iron in circumneutral environments, provide sites for the nucleation of iron (III) minerals (Ferris et al., 1988; Li et al., 2013). These electrostatic interactions are passive but can metal binding competes with protons in metabolising cells (for example, protons from membrane- induced proton motive force) (Mera et al., 1992). Therefore, dead cells may initially bind metals faster than metabolising cells. Quantitatively, cell envelope structures are likely to represent the most important and active sites for biologically influenced mineralisation (Fein et al., 1997). Aluminium in solution also plays a role in the preservation of these biosignatures (Levett et al., 2019), binding irreversibly with the cell envelope structures while iron is enriched within the intracellular void (for example, Fig. 7.3 inset).

A continuous supply of soluble iron can completely fossilise and permineralise entire biofilms (Levett et al., 2016), producing new iron-rich minerals that can aggregate and cement fragments. In the cements investigated here, the microbial fossils were relatively poorly preserved and the cell population density appears to be low (Figs. 7.3 – 7.5). While direct biosignatures (for example, microfossils presented in Fig. 7.4B) are relatively sparse, the surrounding iron oxide minerals are extremely heterogeneous (Fig. 7.4B), suggesting their precipitation was influenced by the microbial biofilm. Biofilms in the streams are extensive and closely associated with the canga substrates (Figs. 7.1B and 7.2). The EPS matrix, which can account for 90% of the dry mass of a biofilm (Flemming and Wingender, 2010), appears to be completely replaced by the iron oxide minerals with only the resistant cell envelope structures preserved as direct biosignatures within the heterogeneous cements. The non- uniform mineral formation and element distribution, as indicated by the backscattered electron micrographs of many of the authigenic iron oxides (for example, Fig. 7.4), gives much of the cement a biological texture (Fig. 7.4B).

Lamellae in the biologically-textured cements (Fig. 7.4B) highlight that the cementation is generational, incrementally developing through time. These lamellae bands may be indicative of wet and dry seasons, where microbial activity and biofilm formation are less well developed during dry seasons. It is interesting to note that Sahoo et al. (2017) recorded an approximate 10-fold increase in the dissolved iron and aluminium in Lakes Amendoim and Violão during the wet season (November to May) compared with the dry season. These data are consistent with our hypothesis that the microbial processes that promote iron oxide mineral dissolution are accelerated during the wet seasons, when biologically available organic matter (washed from the surrounding vegetation) would be more readily available, aiding microbial respiration and the development of anoxic niches. As iron mineralised biofilms are exposed to dehydrating

184

Lake-edge field characterisation conditions during the dry season, they may come fossilised and contribute to the formation of new iron-rich biocements in these erosion resistant horizons.

Quantifying the portions of cements in the canga rica sample that are biologically influenced during formation is difficult due to the low density of direct biosignatures. If we assume the biologically-textured cements represent the fossilisation of entire biofilms over multiple generations, a significant portion of the cement in the canga rica that forms on the lakes’ edges can be considered to have been influenced, and accelerated, by the growth of microbial biofilms.

7.6. Conclusions Iron-rich cement formation within ferruginous duricrusts that cap iron ore deposits in Brazil appears to be accelerated by ‘hotspots’ of microbiological activity. A relatively low but consistent population of iron-reducing microorganism (Geobacter lineages) that survive within microscale anoxic niches within well-developed biofilms appear to be responsible for driving the reductive dissolution of iron oxide minerals. These iron-reducing microorganisms support the growth of abundant and relatively diverse iron-oxidising populations, which contribute to the precipitation of fresh iron oxide minerals. Cements with canga are formed when ferrous iron in solutions is oxidised and subsequently nucleates on nearby cell envelope structures, fossilising microorganisms and possibly the extracellular polymeric substances associated with the biofilms. The biofilms provide an important framework to guide the formation of iron-rich cements within the canga. Understanding these natural processes may contribute to new technologies enabling enhanced physical and chemical stabilisation of degraded land sites and the generation of protective barriers for sulfidic mine waste, potentially reducing acid mine drainage.

Acknowledgements We acknowledge support from the Vale S.A.-UQ Geomicrobiology initiative and the Australian Research Council Linkage Program (LP140100805) to G. Southam and P. Vasconcelos. The authors acknowledge the facilities and the scientific and technical assistance of the Australian Microscopy and Microanalysis Research Facility at the Centre of Microscopy and Microanalysis, at the University of Queensland. Alan Levett acknowledges the support from the Australian Government Research Training Program.

185 Chapter Seven

References Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. Journal of Molecular Biology 1990; 215: 403-410.

Breuer M, Rosso KM, Blumberger J, Butt JN. Multi-haem cytochromes in Shewanella oneidensis MR-1: structures, functions and opportunities. Journal of The Royal Society Interface 2015; 12: 20141117.

Colombo C, Palumbo G, He J-Z, Pinton R, Cesco S. Review on iron availability in soil: interaction of Fe minerals, plants, and microbes. Journal of Soils and Sediments 2014; 14: 538-548.

DeJong JT, Mortensen BM, Martinez BC, Nelson DC. Bio-mediated soil improvement. Ecological Engineering 2010; 36: 197-210.

Fein JB, Daughney CJ, Yee N, Davis TA. A chemical equilibrium model for metal adsorption onto bacterial surfaces. Geochimica et Cosmochimica Acta 1997; 61: 3319-3328.

Ferris FG, Fyfe WS, Beveridge TJ. Metallic ion binding by Bacillus subtilis: implications for the fossilization of microorganisms. Geology 1988; 16: 149-152.

Flemming H-C, Wingender J. The biofilm matrix. Nature Reviews Microbiology 2010; 8: 623.

Fukushima J, Tojo F, Asano R, Kobayashi Y, Shimura Y, Okano K, et al. Complete genome sequence of the unclassified iron-oxidizing, chemolithoautotrophic Burkholderiales Bacterium GJ-E10, isolated from an acidic river. Genome Announcements 2015; 3: e01455-14.

Hegler F, Schmidt C, Schwarz H, Kappler A. Does a low-pH microenvironment around phototrophic Fe(II)-oxidizing bacteria prevent cell encrustation by Fe(III) minerals? FEMS Microbiology Ecology 2010; 74: 592-600.

Levett A, Gagen E, Shuster J, Rintoul L, Tobin M, Vongsvivut J, et al. Evidence of biogeochemical processes in iron duricrust formation. Journal of South American Earth Sciences 2016; 71: 131-142.

Levett A, Gagen EJ, Diao H, Guagliardo P, Rintoul L, Paz A, et al. The role of aluminium in the preservation of microbial biosignatures. Geoscience Frontiers 2019; 10: 1125-1138.

Li J, Benzerara K, Bernard S, Beyssac O. The link between biomineralization and fossilization of bacteria: insights from field and experimental studies. Chemical Geology 2013; 359: 49-69.

186

Lake-edge field characterisation

Liu X, Shi L, Gu J-D. Microbial electrocatalysis: redox mediators responsible for extracellular electron transfer. Biotechnology Advances 2018a; 36: 1815 - 1827.

Liu Y, Wu S, Nguyen TAH, Southam G, Chan T-S, Lu Y-R, et al. Microstructural characteristics of naturally formed hardpan capping sulfidic copper-lead-zinc tailings. Environmental Pollution 2018b; 242: 1500-1509.

Loiselle L, McCraig M, Dyar M, Léveillé R, Shieh S, Southam G. A spectral comparison of jarosites using techniques relevant to the robotic exploration of biosignatures on Mars. Life 2018; 8: 61.

Lovley DR, Holmes DE, Nevin KP. Dissimilatory Fe(III) and Mn(IV) Reduction. Advances in Microbial Physiology. Volume 49. Academic Press, 2004, pp. 219-286.

Mera MU, Kemper M, Doyle R, Beveridge T. The membrane-induced proton motive force influences the metal binding ability of Bacillus subtilis cell walls. Applied Environmental Microbiology 1992; 58: 3837-3844.

Monteiro HS, Vasconcelos PM, Farley KA, Spier CA, Mello CL. (U–Th)/He geochronology of goethite and the origin and evolution of cangas. Geochimica et Cosmochimica Acta 2014; 131: 267-289.

Morris R. Supergene alteration of banded iron-formation. Developments in Precambrian geology. 6. Elsevier, 1983, pp. 513-534.

Pruesse E, Quast C, Knittel K, Fuchs BM, Ludwig W, Peplies J, et al. SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Research 2007; 35: 7188-7196.

Reguera G, McCarthy KD, Mehta T, Nicoll JS, Tuominen MT, Lovley DR. Extracellular electron transfer via microbial nanowires. Nature 2005; 435: 1098-1101.

Rentz JA, Kraiya C, Luther GW, Emerson D. Control of ferrous iron oxidation within circumneutral microbial iron mats by cellular activity and autocatalysis. Environ. Sci. Technol. 2007; 41: 6084-6089.

187 Chapter Seven

Richardson DJ. Bacterial respiration: a flexible process for a changing environment. Microbiology 2000; 146: 551-571.

Roden EE. Fe(III) oxide reactivity toward biological versus chemical reduction. Environmental Science & Technology 2003; 37: 1319-1324.

Roden EE. Microbial iron-redox cycling in subsurface environments. Biochem Society Transactions 2012; 40: 1249-56.

Roden EE, Sobolev D, Glazer B, Luther GW. Potential for microscale bacterial Fe redox cycling at the aerobic-anaerobic interface. Geomicrobiology Journal 2004; 21: 379- 391.

Schwertmann U. Solubility and dissolution of iron oxides. Plant and Soil 1991; 130: 1-25.

Shuster DL, Vasconcelos PM, Heim JA, Farley KA. Weathering geochronology by (U-Th)/He dating of goethite. Geochimica et Cosmochimica Acta 2005; 69: 659-673.

Stookey LL. Ferrozine---a new spectrophotometric reagent for iron. Analytical Chemistry 1970; 42: 779-781.

Weiss JV, Rentz JA, Plaia T, Neubauer SC, Merrill-Floyd M, Lilburn T, et al. Characterization of neutrophilic Fe(II)-oxidizing bacteria isolated from the rhizosphere of wetland plants and description of Ferritrophicum radicicola gen. nov. sp. nov., and Sideroxydans paludicola sp. nov. Geomicrobiology Journal 2007; 24: 559-570.

Yamaguchi KE, Johnson CM, Beard BL, Beukes NJ, Gutzmer J, Ohmoto H. Isotopic evidence for iron mobilization during Paleoproterozoic lateritization of the Hekpoort paleosol profile from Gaborone, Botswana. Earth and Planetary Science Letters 2007; 256: 577- 587.

188

Chapter 8 Biocement stabilisation of an artificial slope and the reformation of iron-rich crusts Alan Levetta, Emma J. Gagena, Yitian Zhaob, Paulo Vasconcelosa, Gordon Southama

a School of Earth and Environmental Sciences, University of Queensland, Brisbane, Queensland 4072, Australia b School of Mechanical and Mining Engineering, University of Queensland, St. Lucia, QLD, Australia

Manuscript accepted with major revisions in Proceedings of National Academy of Sciences of the United States of America

Chapter Eight

Abstract Novel biotechnologies are required to remediate iron ore mines and address the increasing number of tailings (mine waste) dam collapses worldwide. In this study, we aimed to accelerate iron reduction and oxidation to stabilise an artificial slope. An open-air bioreactor was inoculated with a mixed consortium of microorganisms capable of reducing iron. Fluid from the bioreactor was allowed to overflow onto the artificial slope. Carbon sources from the bioreactor fluid promoted the growth of a surface biofilm within the artificial slope, which naturally aggregated the crushed grains. The biofilms provided an organic framework for the nucleation of iron oxide minerals. Iron-rich biocements stabilised the artificial slope and were significantly more resistant to physical deformation compared with the control experiment. These biotechnologies highlight the potential to develop strategies for mine remediation and waste stabilisation by accelerating the biogeochemical cycling of iron.

190

Slope stabilisation experiment

8.1. Introduction Australia and Brazil are the world’s largest producers of high grade iron ore, together exporting approximately 1 billion tonnes per year to provide the world with building resources (United States Geological Survey, 2018). The massive financial incentives of mining threaten the rare ecosystems that survive on the indurated duricrusts that cap iron ore provinces in both Australia (Gibson et al., 2010) and Brazil (Nunes et al., 2015). Recent work has revealed the native biome associated with these iron-rich duricrusts (canga) plays an important role in the biogeochemical cycling of iron oxide minerals, generating microbial and abiotic cements that contribute to the long-term stability of canga (Levett et al., 2016; Monteiro et al., 2014). In Brazil, this microscale cycling of iron and continuous formation of new cements within these duricrusts make them some of the most erosion-resistant surfaces on Earth (Monteiro et al., 2018). Canga- associated microbiomes in these environments offer an array of biotechnological potential (Caneschi et al., 2018). The present study aimed to develop a biotechnological approach to canga stabilisation for mine remediation. It was, in part, inspired by the microbial communities that contribute to iron cementation in natural duricrusts on the sloped edge of lakes in canga environments (Levett et al., 2020). New biotechnologies that stabilise canga for iron ore remediation may also be transferable to the stabilisation of tailings dams, which are urgently needed given the doubling of dam failures in the last 20 years (Armstrong et al., 2019), among other potential uses.

Mine surface remediation post-mining remains a worldwide issue that prevents the reclamation of thousands of mine sites (Mendez and Maier, 2007). Harnessing the knowledge gained by studying natural duricrust evolution (Gagen et al., 2019a; Levett et al., 2016; Levett et al., 2020; Monteiro et al., 2018; Monteiro et al., 2014; Spier et al., 2018), we aimed to promote biogeochemical cycling of iron in order to accelerate the reformation of an iron-rich crust. The ultimate goal is to regenerate a habitat similar to pre-mining conditions, allowing for revegetation of plant species adapted to growing on iron-rich duricrusts. Reforming an iron- rich crust will also assist with restoring hydraulic function as well as chemical and physical stabilisation of mine waste material. Large, open cast iron ore mine operations will require the exposed pit surfaces to be rejuvenated without being completely infilled, therefore stabilising waste material on artificial slopes is essential for remediation efforts.

Prior to mining, the iron-rich duricrusts are typically stripped and stock-piled to access the high-grade iron ore below. Canga, primarily composed of goethite and hematite, contains a high aluminium and phosphorus content, typically making it a by-product of iron ore mining

191 Chapter Eight and an ideal starting material for remediation efforts. We aimed to stabilise a gentle artificial slope (approximately 10°) of crushed canga using microorganisms to accelerate the redox cycling of iron oxide minerals and the formation of new iron cements.

This experiment physically separated a ‘reactor’ (left), and an ‘artificial slope’, of crushed canga (Fig. 8.1). A peristaltic pump provided solutions to the reactors, which were allowed to ‘overflow’ onto the slope (Fig. 8.1). The solutions that flowed down the slope were collected at the base of the experiment, referred to as the ‘runoff’. The bioreactor treatment was inoculated with a recently described iron-reducing microbial consortium (Gagen et al., 2019b) and provided with a simplified nutrient medium (see Methods 8.2). The ‘reactor’ of an uninoculated control experiment run in parallel, was fed with anoxic, sterilised milliQ water. We present here data showing that the surface of an artificial slope of canga fragments was stabilised using microbial iron reduction and subsequent microbially-influenced iron oxide precipitation. A parallel, uninoculated control system failed to produce iron biocements

8.2. Materials and methods 8.2.1. Experimental Vessels Each experimental vessel was composed of a polyacrylic plastic containing two components: a rectangular ‘reactor’ portion on the left and a trapezoid-shaped ‘artificial slope’ portion on the right-hand side. An ‘overflow’ from the reactor portions provided solutions to the artificial slopes and the water ‘runoff’ from the slope was collected. Each reactor portion contained crushed (< 2 mm fragments) canga fragments, predominately goethite and hematite with an average iron concentration of 56 wt.%. The canga fragments were collected from the Serra Sul canga plateau (GPS coordinates from Google Earth: -6.4004028; -50.3504527777), Carajás Mineral Province, Pará, Brazil. The crushed canga substrate was sieve separated and mixed by mass in ratios consistent with standard cement: 3 parts greater than 2 mm, 3 parts 0.25 – 2 mm and 1 part less than 0.25 mm before being sterilised (autoclave settings: 121 ºC held for 30 min at a minimum of 210 kPa). Approximately 4 kg of crushed canga was added to the artificial slope portion of each experiment. An uninoculated control experiment provided direct comparison with a bioreactor experiment, which was inoculated with a culture enriched in bacteria capable of using iron (III) as a terminal electron acceptor (Gagen et al., 2019b). The iron-reducing inoculant is a fermentation-driven microbial consortium dominated by a Telmatospirillum metagenome assembled genome (Gagen et al., 2019b). A peristaltic pump displacing 0.36 mL min-1 fed both the control and bioreactor. Anoxic, sterile milliQ water was provided to the control while the reactor was supplemented with a simplified Geobacter

192

Slope stabilisation experiment

medium (DSMZ 579: 2.5 g/L NaHCO3; 1.50 g/L NH4Cl; 0.60 g/L NaH2PO4; 0.1 g/L KCl; 3.36 g/L D-Glucose). Dissolved oxygen (FireStingGO2, Pyrosciene) and pH for the control reactor and bioreactor were monitor throughout the experiment. The experiment was run continuously for 24 weeks. Pump times during the initial setup period varied from 12 to 8 hours per day to set a balance in iron-reduction and maintaining the microbial population. Ferrous iron concentrations in the initial set up period were determined using a Ferrozine assay (Lovley and Phillips, 1987). At the end of the two-week testing period, pumps were active for at least 8 hours every day, except for 5 days in week 14 and 10 days throughout weeks 21 – 22, during which the bioreactor was re-inoculated with the original culture and pumps were turned off for 10 days. The experimental vessels were housed within a growth chamber (Conviron Adaptis A1000) maintaining a ‘day-time’ (light intensity: 700 µmol) temperature of 30 ºC and a ‘night- time’ (no light) temperature of 25 ºC, with each time period being 12 hours

8.2.2. Drop-cone test At the completion of the experiment, the slopes resistance to erosion was approximated using a cone drop test, rock samples from the slope portion of the control and reactor experiments were collected to determine the microbial community. For the drop cone test, the vertex of cone was aligned within 0.5 mm of the surface of the slope and allowed to drop for 3 s, after which the cone penetration depth was recorded. An approximate mass of 130 g was added to the bar (80 g) to increase the penetration force. The additional of the weights restricted the maximum fall of the cone to approximately 19 mm. The slope portion of the experiment was divided into 1 cm grids, with a drop cone measurement taken at each point, expect where large clasts affected the penetration. If the cone penetrated more than 10 mm, a minimum 2 cm gap in each direction was left to avoid affected surfaces.

8.2.3. Scanning electron microscopy To examine the microstructure of the cements, consolidated fragments from the slope portion of the reactor experiment were taken from 0 – 8 cm, 8 – 16 cm and 16 – 25 cm along the surface from both the bioreactor and control systems at the cessation of the experiment. Samples were also collected from the bioreactor and control reactor themselves. To preserve the microbial components, consolidated fragments were prepared using a Pelco Biowave microwave. Samples were fixed in approximately 2.5% glutaraldehyde (2 min on - 2 min off - 2 min on, 80 W, under vacuum) and dehydrated using the following acetone series: 20%, 40%, 60%, 80%, 90%, 3 × 100% (each step: 40 s, 250 W, no vacuum). The dehydrated samples were then gradually infiltrated and replaced with an Embed 812 resin using acetone: resin ratios of 3:1,

193 Chapter Eight

1:1, 1:3 and twice with 100% resin (each step: 3 min, 250 W, under vacuum), after which they were cured at 60 ºC for 48 hours.

Polished petrographic thin sections of the resin-embedded samples were examined using a JEOL7100 scanning electron microscope equipped with energy dispersive X-ray spectroscopy (SEM-EDS) in backscattered electron mode at an accelerating voltage of 15 kV. No staining treatments were used on the samples; all cell envelope mineralisation was from natural processes. Prior to examination, samples were degassed at 50 °C for a minimum of 12 hours and coated with 10 nm iridium using a Quorum Q150T sputter coater

8.2.4. Water chemistry To determine iron added to solution, water chemistry was monitored on an approximately weekly basis using inductively coupled plasma optical emission spectroscopy (ICP-OES) from: (i) the base of each reactor, (ii) the surface of each reactor and the (iii) runoff from both the control and reactor experiment. High evaporation rates in the growth chamber may have concentrated ions in the runoff samples, which were collected overnight (10 – 12 hours) in clean beakers. On three occasions, insufficient sample volumes precluded the opportunity to test runoff samples. Filtered water samples were acidified to a concentration of 7% HNO3 and digested in a MARS Xpress microwave (10 min at 160 ºC; 10 min at 170 ºC) in Teflon tubes.

Water samples were then diluted to 5% HNO3 and analysed using a Pelkin Elmer Optima 7300DV ICP-OES with an argon flow rate of 15 L/min. The difference between iron in solution in the reactor and the leachate was assumed to have precipitated in the slope portion of the experiment.

194

Slope stabilisation experiment

Fig. 8.1. Schematic model of experimental setup for the bioreactor treatment. A peristaltic pump set at a pump rate of ~ 21.6 mL/hour for 8 hours/day provided the bioreactor with a renewed nutrient source. With a few exceptions, this pump regime was used for 24 weeks for the 26-week experiment. The fermentation-driven bioreactor produced approximately 80 ppm of ferrous iron, which was allowed to ‘overflow’ onto the artificial slope. A thick biofilm covered the surface of

195 Chapter Eight the bioreactor, which was anoxic below the surface. An identical uninoculated control experiment, fed with sterilised anoxic milliQ water was run simultaneously. Solutions initially stayed at the surface before falling to the bottom of the experiment, then gradually rising and dripping out of the experiment at the ‘runoff’ point. At the end of the experiment, the bioreactor and control slopes were subdivided into six regions for sampling: ‘R’ denotes bioreactor treatment with ‘C’ used for control samples. P1 (0 – 8 cm), P2 (8 – 16 cm) and P3 (16 – 25 cm) indicates the distance from the control and bioreactor overflow points, with ‘S’ and ‘D’ denoting surface and depth samples, respectively. Photographs of the experimental setup, highlighting mineralised vs unmineralised surfaces of the bioreactor treatment and the control treatment, respectively.

8.2.5. DNA extraction, sequencing and sequence analysis To extract DNA for water samples, 2 mL samples from the bottom, top and runoff collection for both the bioactive and uninoculated control experiments was centrifuged at 10,000×g for 10 min and the supernatant was removed to enrich the biomass. Runoff water samples were collected overnight (ranging between 8 – 10 hours). For the rock samples, approximately 2 g of material was collected for DNA extraction. Samples included material from the surface (S) and at depth (D), which were collected from various positions along the slope away from the overflow: 0 – 8 cm (P1), 8 – 16 cm (P2) and 16 – 25 cm (P3) for both the bioreactor (R) and control (C) systems (Fig. 8.1). For all samples, DNA was extracted using the DNeasy Powersoil Kit (Qiagen, Hilden, Germany).

Amplicon sequencing and PCR amplification

The 16S rRNA gene (V6 – V8 region) was amplified from the extracted DNA using the universal primers 926f and 1392r, modified to contain Illumina specific adapter sequence; 926F: 5’- TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGAAACTYAAAKGAATTGRCGG- 3’ and 1392wR: 5’- GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGACGGGCGGTGWGTRC-3’. For fungal communities, the internal transcribed spacer 2 (ITS2) region was targeted using the ITS3 (5’- GCATCGATGAAGAACGCAGC -3’) and ITS4 (5’- TCCTCCGCTTATTRATATGC -3’) primers (White et al., 1990) modified to contain Illumina specific adapter sequence (ITS3: 5’- TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGCATCGATGAAGAACGCAGC - 3’ and ITS4: 5’- GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGTCCTCCGCTTATTRATATGC -

196

Slope stabilisation experiment

3’). To prepare libraries, the workflow outlined by Illumina was followed with the exception that the NEBNext® UltraTM II Q5 Mastermix (New England Biolabs #M0544) was used in standard PCR conditions. The resulting PCR amplicons were purified using Agencourt AMPure XP beads (Beckman Coulter, Brea, CA, USA). Purified DNA was indexed using the Illumina Nextera XT 834 sample Index Kit A-D (Illumina FC-131-1002, San Diego, CA, USA) in standard PCR conditions (aforementioned PCR mastermix) to assign unique 8-bp barcodes. Equimolar concentrations of the indexed amplicons were sequenced using a MiSeq Sequencing System (Illumina) with paired end sequencing (V3 300-bp chemistry) in accordance with the manufacture’s protocol at the Australian Centre for Ecogenomics, The University of Queensland.

Sequence analysis

Water and rock samples were processed using MOTHUR (https://www.mothur.org/wiki/MiSeq_SOP, accessed on 19th February, 2019), which was used in accordance with the MiSEQ operating procedures. Minor alteration were made to process forward reads only and processing fungal DNA with various lengths. Forward reads were trimmed on quality (quality average = 35; sliding window size = 50 nt) before the primer was removed. Reads were subsequently trimmed to 250 nt. Ambiguous bases and homopolymeric structures (8 nt repeats) were removed. For 16S rRNA reads, alignment using the SILVA SSU Ref NR 99 v132 reference database (Pruesse et al., 2007). Potential chimeras were identified by comparing sequences with the silva132.gold reference database and subsequently removed. Remaining sequences were classified using the SILVA SSU Ref NR 99 v132 reference database and sequences classified as Eukaryota, Chloroplast, Mitochondria or unknown were removed. Fungal ITS reads were classified using the UNITE reference database (Kõljalg et al., 2005). Potential chimeras were identified and subsequently removed using a self-referencing method. A distance of ≤ 0.03 was used to cluster sequences into Operational Taxonomic Units (OTUs) and singletons (OTUs with only one sequence) were removed to produce all figures. Libraries were subsampled to the smallest number of OTUs (16S rRNA-water samples = 6747 OTUs; 16S rRNA-rock samples = 15548 OTUs; Fungal (ITS)-rock sample = 15744 OTUs) for diversity analysis and generating heatmaps to visualise shared OTU distributions and measures of Jaccard (Chao et al., 2005) and Yue and Clayton (Yue and Clayton, 2005) dissimilarity. For 16S rRNA-water samples, OTU associations with water chemistry data was correlated using Pearson’s coefficient, with correlations greater than 0.6 or less than −0.6 reported. Values below the detection limit in the water chemistry datasets were assigned concentrations of 50%

197 Chapter Eight of the detection limit to reduce statistical correlation bias (Croghan and Egeghy, 2003). A principal coordinate (PCoA) analysis was also performed for the 16S rRNA-water samples to visualise clustering using the Jaccard distances between the libraries. In addition, correlations (Pearson’s correlation coefficient) with axis of the PCoA plots and the water chemistry datasets were reported. The nucleotide basic local alignment search tool (BLASTn) (Altschul et al., 1990), using the NCBI and non-redundant nucleotide collection, was used to compare the major OTUs with publicly available sequences for all sample types. Sequences were initially compared by excluding modelled, uncultured and environmental samples. All sequences have been submitted to the NCBI Sequence Read Archive; 16S rRNA sequences have been submitted under BioProject number PRJNA599307 and ITS sequences have been submitted BioProject number PRJNA602296.

8.2.6. Nanoindentation To characterise the mechanical properties of the mineralised microbial cements, nanoindentation tests were performed on the thin section from R-P1-S (Fig. 8.1) using a TI- 900 Hysitron Triboindentor® equipped with a diamond Berkovich tip (radius ca. 100 nm). A load-control function (10 s loading; 10 s holding and 15 s unloading) with an indentation load of 5 mN was used to determine the hardness (H) and the reduced modulus (Er), which were computed based on the recorded load-displacement (P-h) curves. For comparison several locations were tested, including, two of the mineralised biofilms (microfossil cements) (Fig. 8.3: Regions 1 and 2), a natural cement that consolidated recycled fragments (Fig. 8.3 Region 3), a hematite-rich clast (Fig. 8.3; Region 4) and the resin (Fig. 8.3; Region 5). With the exception of the natural cement, all other structures were relatively homogenous; therefore, 4 × 4 indentation arrays with 10 m spacing between each indent were performed on the structures. Individual points where selected for the natural cements with at least 10 m spacing between each other. Unreproducible load-displacement curves that occurred due to submicron variations in the textures were removed for statistical analysis. A two-tailed t-test assuming equal variance indicated there was no statistical difference between the hardness (p = 0.25); however, the reduced modulus indicated a significant difference (p < 0.005) for the reduced modulus of the microfossil cements. Significant differences for both the hardness and reduced modulus for all other textures were determined (p < 0.001). Methods used for nanoindentation were identical to those used in (Levett et al., 2020), allowing for a direct comparison between synthetic biocements produced in this experiment and naturally made biocements occurring in the field.

198

Slope stabilisation experiment

8.3. Results 8.3.1. Formation of iron biocements After 26 weeks of wetting-drying cycles, the artificial slope in the bioreactor system was mineralised with a ferruginous surface biofilm (Fig.8.1). The microbial growth in the bioreactor system had aggregated the crushed canga fragments in the slopes and biomineralisation of these biofilms formed cements between the crushed fragments. Cracks on the artificial slope within the bioreactor system highlighted that the biomineralised concretions had consolidated and hardened the material over the 26-week experiment. In comparison, the artificial slope within the uninoculated control system supported fungal growth but the crushed canga fragments showed no visual evidence of aggregation. The drop-cone test confirmed that the bioreactor experimental set up (average penetration = 0.7 mm; standard deviation = 0.75 mm; Appendix 3;Video S8.1 and S8.2) was more resistant to unidirectional physical damage than the control (average penetration = 14.9 mm; standard deviation = 4 mm; Appendix 3; Video S8.3).

After completion of the experiment, the reactor artificial slope was roughly broken up using a hammer and sterile 50 mL tube to collect profile samples. Surface samples within 8 cm from the bioreactor overflow (R-P1-S; Fig. 8.1) were the most-well consolidated. During profile sampling, large portions of the slope in the R-P1-S region remained intact after being dislodged from the experiment, including one large aggregate with an approximate surface area of 10 × 8 cm, consolidated to a depth of approximately 5 cm. In comparison, surface samples at the lower portion of the slope (R-P3-S; Fig. 8.1) displayed comparatively more fungal growth, were less mineralised and consolidation was restricted to approximately the top 2 cm. Over the 16 hour daily drying phases, evaporation dehydrated the entire experiment to approximately 2 cm below the height of the runoff outlet. Little to no consolidation was observed below the dehydration height (R-P1-D, R-P2-D, R-P3-D; Fig. 8.1).

8.3.2. Iron fossilisation of microorganisms Mineralised biofilm aggregates were not evident in any of the material collected from the control reactor or throughout the artificial slope in the control system. In the bioreactor system, a spectrum of partly mineralised to fossilised biofilms formed concretionary coatings between the crushed starting material (Figs. 8.1 – 8.5). Examining the microstructure of consolidated fragments from the slope of the bioreactor treatment revealed extensive ferruginised microbial biofilms in R-P1-S samples (Fig. 8.2), forming meniscus-type biocements between grains (Fig. 8.2A; thin arrows). These ferruginised biofilms had formed around and connected individual crushed fragments, some more than 0.5 mm apart. Large (mm-scale), partially mineralised

199 Chapter Eight biofilms also formed aggregate structures between large grains helping to consolidate this loose material (Fig. 8.2A; dashed arrow). Thick, heavily mineralised biofilms that formed ‘bridging’ structures in Fig 8.2D were commonly observed (Fig. 8.2A; thick arrow). These highly mineralised bridging structures (Fig. 8.2A; thick arrow) had cracked during sample processing, indicating that these microbial concretions form semi-rigid structures. Continued exposure to aqueous iron infilled the pore spaces between the encrusted cell envelopes, leaving bacteriomorphic void spaces and fossilising the relatively large biofilm (Fig. 8.2B). Rod- and cocci-shaped microbial fossils, typically 1 µm in diameter, were commonly identified throughout the fossilised biofilms (Fig. 8.2B). Microbial mineralisation is well-developed in the centre of fossilised biofilms, with less well-developed fossilisation towards the edges of the biofilms (Fig. 8.2B; high-resolution insets). Microfossils were typically bacteriomorphic casts preserved with little or no cell envelope structure observed (Fig. 8.2B; high magnification insets)

200

Slope stabilisation experiment

Fig. 8.2. Low magnification backscattered electron micrographs highlighting meniscus-style cements (thin arrows), partially mineralised biofilms (dashed arrow) and cracked ‘bridging’ structures. High magnification insets in (B) highlight these mineralised biofilms are composed of rod- and cocci-shaped microbial casts.

201 Chapter Eight

Cell structures (voids) are abundant in the fossilised biofilms, which provided active sites for mineral nucleation (Fig. 8.3A). Evidence of binary fission, indicating preservation during growth, occurs throughout the fossilised biofilms (Fig. 8.3A; white arrow). The high cell abundance and relatively small size of the bacteria provides a plethora of bioactive surfaces to facilitate and guide mineral nucleation. Mineralisation of cell envelopes and intercellular regions contributed to a high mineral to void space (casts of the microfossils) ratio and increased the robustness of the biocements. Nanoindentation tests on two regions of well- mineralised biocements indicated the hardness was 1.1 ± 0.3 GPa and 1.3 ± 0.2 GPa and a reduced modulus (stiffness) of 18.2 ± 3.6 GPa and 22.2 ± 2.4 GPa (Fig. 8.3; Regions 1 and 2). In comparison, naturally formed goethitic concretions that hold together the highly recycled material within canga had a harness of 3.1 ± 0.6 GPa and a reduced modulus of 56.1 ± 10.0 GPa (Fig. 8.3; Region 3). The hematite fragment (Fig. 8.3; Region 4) and the resin (Fig. 8.3; Region 5) had a hardness of 10.5 ± 0.7 GPa and 0.17 ± 0.0015 GPa, respectively.

The initial growth of the microbial biofilm is vital, providing an organic scaffold structure that promotes aggregation of fragments, which can then be ferruginised to improve consolidation. The large ferruginised biofilm in Fig. 8.3B is representative of the surface biofilm of the slope within R-P1-S samples. Ferruginised biofilms were most prominent at the surface of the slope, though they existed down to approximately 5 cm within R-P1-S samples. For example, the microbial biofilms approximately 2 cm below the surface of the slope also contributed to the aggregation of fragments (Fig. 8.4A), with varying degrees of mineralisation further consolidating the fragments (Fig. 8.4B). The ferruginised biofilms coated many grains (for example, Fig. 8.4A; thick arrow). The high-resolution inset (Fig. 8.4B; thin arrow) highlights highly mineralised microbial biofilms apparently dominated by rod-shaped microorganisms with only the bacteriomorphic casts preserved. These fossilised biofilms are surrounded by microbial biofilms in which only the reactive cell envelopes have been mineralised (Fig. 8.4B; thick arrow). The biofilms also aggregate microscale fragments and small clastic material is also embedded in some of the biofilms (Fig. 8.4B; thin arrow).

202

Slope stabilisation experiment

Fig. 8.3. Backscattered electron micrographs of the surface biofilm that coated bioreactor treatment slopes. High magnification micrograph (A) revealed the high cell density and evidence of binary-fission (arrow) within fossilised biofilms. Low magnification micrograph (B) demonstrates the continuous nature of the mineralised surface biofilm on the bioreactor treatment slope. Nanoindentation tests for the microfossil cements (Regions 1 and 2), neighbouring goethite cements formed prior to experiment (Region 3), hematite-rich fragment

(Region 4) and resin (Region 5). Note: n – number of tests, x̅ H – average hardness (GPa), 휎H – standard deviation for hardness (GPa), x̅ Er – average reduced modulus (GPa), 휎H – standard deviation for reduced modulus (GPa).

203 Chapter Eight

Fig. 8.4. Backscattered electron micrograph of mineralised biofilms coating grains (thin arrows). Micrograph (B) revealing different stages of mineralisation: partially mineralised biofilms with encrusted cell envelopes (thin arrow) and microscale fragments dispersed throughout the biofilm with high magnifcation inset revealing the preservation of rod-shaped bacteriomorphic casts.

204

Slope stabilisation experiment

The lower portion of the artificial slope (R-P3-S; 16 – 25 cm) was dominated by organic (likely fungal) aggregates with less mineralisation observed (Fig. 8.5). Surface biofilms provided a physically and chemically restrictive barrier above aggregated grains (Fig. 8.5A). Presumably, less soluble iron was available further away from the overflow to fossilise the microorganisms; therefore, only cell envelope structures are mineralised with intercellular regions remaining unmineralised (Fig. 8.5B). Large rod-shaped or filamentous cells, 5 – 10 µm in diameter appear to be fossilised fungal structures. Microscale cocci and/or rods, more consistent with the size of bacteria (ca. 1 µm in diameter), are preserved in close proximity to the fungi (Fig. 8.5B; arrows).

Fig. 8.5. Backscattered electron micrograph of the surface biofilm in the bioreactor treatment slope, approximately 24 cm from the overflow of the bioreactor, revealing large (ca. 5 m is diameter), eukaryotic-type cell structures that had been encrusted in iron oxide minerals. Smaller (ca. 1 m in diameter), cocci-shaped cell envelopes were also immobilised by the ferrous iron-rich solutions (arrows).

205 Chapter Eight

8.3.3. Open-air bioreactor The bioreactor was inoculated with a fermentation-driven microbial consortium capable of reducing crystalline iron oxide minerals (Gagen et al., 2019b). The bioreactor, which overflowed onto the treatment slope, produced an average aqueous iron concentration of 73.0 ± 8.97 ppm of throughout the experiment (Table 8.1), while the runoff (from the slope into a collection beaker) averaged only 3.50 ± 2.48 ppm. The bioreactor was anoxic below the surface biofilm (Fig. 8.1); dissolved oxygen was less than 0.2% air saturation within 1 mm from the surface, allowing ferrous iron to stay in solution. Consequently, there was no difference in the aqueous iron concentration in the top compared with the bottom of the bioreactor. Dissolved oxygen in the control reactor was between 90 – 100% air saturation throughout the experiment. The pH in the bioreactor was 6.1 compared with 5.2 in the control reactor.

Table 8.1. Average water chemistry (ppm) ± standard error over the 26-week experiment. Sample taken from top and bottom of bioreactor showed no statistical difference and were combined into a single data set. See Appendix 3; Tables S1 – S9 for complete set of data.

8.3.4. Microorganisms in the bio- and control reactors The microbial communities within solutions contained within the bioreactor and the control reactor were measured at depth and at the surface of the bioreactor, three times throughout the course of the experiment. There were no striking differences in the major microbial lineages in samples collected from the top compared to the bottom of the reactor, consistent with water chemistry being similar at these depths. For example, Telmatospirillum from the inoculum, was the major lineage, which was consistent with the inoculum (OTU2, average 25 ± 14% of total community sequences), and a Burkholderia was predominant (24 ± 7% of total community sequences, Fig. 8.6). Other major OTUs that developed in the bioreactor included two Morganella species (OTU8 and OTU9, with 16 ± 2 % and 10 ± 2% respectively) and a

206

Slope stabilisation experiment

Pantoea agglomerans (OTU10, 8 ± 2%), a putative iron reducer (Francis et al., 2000). Microbial communities in the runoff collected from the bioreactor slope were dominated by an Achromobacter (OTU1, 25 ± 6 % of sequences), an Enterobacter (OTU5 14 ± 5%) and Telmatospirillum sequences (OTU2, 15 ± 8%; Fig. 8.6).

8.3.5. Microorganisms fossilised to create biocements The Telmatospirillum and Burkholderia sequences were also detected from time to time in the water from the control reactor (17 ± 22% and 16 ± 22% of total sequences respectively; Fig. 8.7), likely as a result of cross-contamination by aerosols, which is unsurprising given the open nature of the experiments incubated in the same growth chamber. A Ralstonia (OTU7) was also predominant in the water from the control (17 ± 14%). The control runoff was dominated by a Crenobacter (OTU 19 ± 16%) and the Ralstonia species (OTU7, 12 ± 1%; Fig. 8.6)

Fig. 8.6. Heatmap analysis of 16S rRNA gene sequences from water samples collected after 13, 17 and 23 weeks of experiment runtime from the bottom and top of the bioreactor and control reactor chambers as well as from the runoff collected. Relative abundances greater than 5% are indicated for the most prevalent OTUs. Simultaneous samples were collected to determine aqueous geochemistry using ICP-OES (Appendix 3; Tables S1 – S9). Iron-reduction in the bioreactor was driven by fermentation (Gagen et al., 2019), with putative iron-reducing microorganisms listed in blue text. The nearest named isolate in the public domain and its accession identification code is provide for each OTU. Aqueous elements that significantly (P < 0.05) correlated (r > 0.6) with OTUs are listed. OTUs were clustered at a distance of ≤ 0.03.

After the experiment had dried, rock samples were collected at depth and at the consolidated surface along the slope, to determine microbial lineages that contributed to consolidation of the

207 Chapter Eight artificial slope in the bioreactor system. At both the surface and at depth, Achromobacter was the dominant lineage detected in the bioreactor slope (40 ± 5% of sequences), while an Enterobacter (OTU5; 17%) and a Stenotrophomonas (OTU6; 9%) were predominant lineages in the consolidated surface samples in the most consolidated material 0 – 8 cm from the overflow (Fig. 8.7). The Telmatospirillum (OTU2) was a predominant lineage at depth within the rock samples (9 ± 4% ) compared with only 2 ± 0.5% relative abundance in the well cemented surface samples (Fig. 8.7). Rock samples collected along the control slope were dominated by Methylobacterium mesophilicum (OTU3, 62 ± 2% at the surface and 29 ± 4% at depth; Fig. 8.7).

Fig. 8.7. Heatmap analysis of 16S rRNA gene sequences from rock samples collected at the end of the experiment from various distances (P1 = 0 – 8 cm; P2 = 8 – 16 cm; P3 = 16 – 25 cm) away from the overflow of the bioreactor (R) and the control (C) reactor. Microbial lineages of the surface samples (S) were compared with samples collected from depth (D). The slope in the bioreactor treatment were dominated by Gram-negative microbial lineages, OTUs 1, 5 and 6, with relative abundances greater than 5% provided. The nearest named isolate from public domains and its accession identification code are listed for each OTU. Rock material from the control reactor and bioreactor were also sampled after dehydration. OTUs were clustered at a distance of ≤ 0.03.

Samples collected from the surface were also sequenced for fungal linages (Fig. 8.8.). Surface samples from the position 1 from the bioreactor system (R-P1-S) was dominated by lineages

208

Slope stabilisation experiment that shared 100% identity across the internal transcribed spacer 2 (ITS2) region with Purpureocillium lavendulum (fungal OTU1; 65%) and Penicillium citrinum (fungal OTU5; 32%; Fig. 8.8). Fungal OTUs 2 (99.7% Scedosporium minutisporum), 3 (100% Gibberella fujikuroi) and 4 (100% Fusarium pseudensiforme) were dominant further away from the overflow (8 – 25 cm from the overflow) in the reactor system (Fig. 8.8). The control system contained almost all the top 10 fungal OTUs distributed at various proportions throughout the slope, with the exception of OTUs 3 (100% Gibberella fujikuroi) and 4 (100% Fusarium pseudensiforme), which were absent from P3 region (16 – 25 cm) from the overflow. Dominance of one fungal OTU over another in the control system was most likely to be dependent on which colonies were sampled as the geochemistry was consistent throughout surface of the control slope.

Fig. 8.8. Heatmap analysis of internal transcribed spacer 2 (ITS2) fungal gene sequences from rock samples collected at the end of the experiment from various distances (P1 = 0 – 8 cm; P2 = 8 – 16 cm; P3 = 16 – 25 cm) away from the overflow of the bioreactor (R) and the control (C) reactor.

8.4. Discussion Accelerating the biogeochemical cycling of iron effectively consolidated crushed fragments of canga in this experiment, reforming an iron-cemented layer akin to a precursor of an iron-rich duricrust (Fig. 8.1). The biofilms in the artificial slope provided an essential organic scaffold for mineralisation. Biofilm growth between grains naturally aggregated crushed material and

209 Chapter Eight provided the framework for the mineral precipitates. These biofilms often formed meniscus- type cements similar to those characterised in natural canga samples (Levett et al., 2020). The formation of an iron-rich crust and stabilisation of mine waste has been identified as a critical limitation to the sustainability of iron ore mining (Levett et al., 2016). The re-cementation of the iron ore overburden (canga) may allow for the re-vegetation of native and naturally rare plant species that existed prior to mining (Nunes et al., 2015). This experiment provides direct evidence that ferruginisation of microorganisms is an effective, circumneutral pH method to accelerate authigenic iron oxide cement, mitigating the need for acid weathering to form crustal cements (Liu et al., 2018). Fossilising surface biofilms on artificial slopes may also help to restore hydraulic function to degraded landscapes such as iron ore mines. The accelerated biogeochemical cycling of iron presented here offers the opportunity to restore substrates similar to natural iron-rich duricrusts associated with mine sites worldwide. Further, these biotechnologies may be used for the chemical and physical stabilisation of mine tailings and the reclamation of degraded land sites, representing a prodigious advancement for sustainability across the mining sector.

In this study, aqueous iron supplied to the slope from the bioreactor averaged 73 ppm (0.073 g L-1). Using an average pump rate of 1.21 L/week for 24 weeks over the 26-week experiment 2+ indicates that only a total of 2.1 g of Fe(aq) was added to the slope. The biologically influenced hydrous ferric oxide minerals in this study were not identifiable using infrared techniques (Raman and micro Fourier transform infrared); however, they are assumed to be poorly crystalline ferrihydrite, with an approximate formula of Fe10O14(OH)2·~H2O (Kennedy et al., 2004; Michel et al., 2010). Assuming the iron precipitated as ferrihydrite, this only adds a total of approximately 3.2 g of hydrous ferric iron oxide to the 1 kg of crushed canga that was cemented within the top 2 – 5 cm of the experiment system. As such, it is estimated that < 1 wt.% of iron in solution would be required to consolidate the surface fragments of the crushed waste material. Considering the extremely small quantities of iron added to the system, the biocements were extraordinarily resistant to uniaxial deformation at both the macroscale (drop cone tests; Appendix 3; Video S1 – 3) and the nanoscale (nanoindentation tests; Fig. 8.3). The microfossils appeared to create a structure with similar properties to a closed-cell metal foam; highly porous with relatively high resistance to plastic deformation (Idris et al., 2009). The average hardness of microfossil cements here (1.2 ± 0.3 GPa) was only slightly less (p < 0.05) than microfossil cements identified in natural canga samples (1.4 ± 0.3 GPa; (Levett et al., 2020)).

210

Slope stabilisation experiment

The synthesis of ‘biocements’ as demonstrated here offers insights into the fossilisation and preservation of potential biosignatures in iron-rich environments. The electrostatic interactions between aqueous iron and the net negative charge of microbial cell envelopes drives passive, microbially-influenced iron oxide precipitation (Ferris et al., 1988; Li et al., 2013). Cell envelope structures in this experiment appear to be poorly preserved compared with microfossils identified within the field, suggesting that iron alone may not be sufficient for the long-term preservation of organic microbial biosignatures (Levett et al., 2019). The Telmatospirillum (OTU2) lineage was enriched at depth compared with the surface samples (Fig. 8.6), suggesting oxic conditions within the dehydration zones (uppermost 2 – 5 cm) at the surface. Oxic conditions at the surface appears to play an important role in the fossilisation of the microorganisms, thereby, contributing to the stabilisation of the crushed canga.

Cell structures play an important role in biomineralization (Beveridge, 1989). The three most abundant OTUs, likely to be responsible for the biocements that formed with the artificial slope in the bioreactor system, were all Gram-negative microbial lineages. Gram-negative and Gram- positive microorganisms each provide active sites for mineral nucleation, although additional binding sites may be associated with the cell envelope structures of Gram-positive microorganisms (Beveridge and Fyfe, 1985). In a long-term fossilisation experiment it was concluded that Gram-positive bacteria were more readily silicified than their Gram-negative counterparts (Orange et al., 2014). Here we report the complete and relativity rapid ferruginisation of Gram-negative microorganisms (Figs. 8.2 – 8.4). The active sites for mineral nucleation on microbial surfaces are also likely to be cation specific. In a study restricted to Gram-negative bacteria, iron complexed with the cell envelope structures of only one of the four related Pseudomonas aeruginosa strains (Langley and Beveridge, 1999). Those authors suggested that both A- and B-band lipopolysaccharides may influence overall cell surface properties, promoting the formation of mineral precipitates. In this experiment, electron microscopy also revealed the encrustation of large, eukaryotic cells in R-P3-S region (Fig. 8.5). Fungi are known to play important roles in metal immobilisation (Gadd, 2007); however, limited studies have investigated the role of fungi in iron biomineralisation (Anand and Verrall, 2011; Oggerin et al., 2016). The encrustation of the fungal mat growing in the R-P3-S region demonstrates that iron-rich solutions can fossilise a variety of cell envelope structures. Therefore, algae and fungi that naturally grow on the iron-rich canga substrates may provide valuable organic scaffolds for the consolidation of waste rock.

211 Chapter Eight

Fungal growth alone on the control slope was not enough to aggregate crushed fragments. Preliminary experiments also indicate that supplementing the crushed canga with aqueous iron solutions (up to 300 ppm) in the absence of a microbial biofilm does not consolidate crushed fragments (data not shown); rather, the iron in solution simply coats grains, enlarging pisoliths. Growth of a microbial scaffold appears to be an essential first step for the formation of authigenic cements between grains. Ferruginisation of the biofilm acts to immobilise the microorganisms, preserving the microbial biofilms and consolidating the crushed mine waste. Together the growth of these microorganisms and their subsequent or concomitant mineralisation creates a chemically stable biocement that promotes physical stabilisation.

The effect of mineralisation on microbial growth remains enigmatic (Benzerara et al., 2011). Cyanobacteria and BoFeN1(nitrate-reducing iron oxidiser) can survive and possibly even benefit from extracellular biomineralization (Kappler et al., 2005; Phoenix and Konhauser, 2008; Phoenix et al., 2000). In this study, the large continuous biofilms and possible evidence of cell replication within mineralised biofilms (Fig. 8.3) indicates that microbes maintain the ability to metabolise and reproduce even when partially encrusted in iron oxides.

Given the critical importance of these microorganisms to cement formation, the growth of a microbial scaffold prior to exposure to iron-rich solutions may be a more effective method for aggregation and material consolidation compared with the concurrent microbial growth and iron exposure that was employed in this experiment. This may be particularly pertinent if the ferruginisation of the microbial biofilms is too restrictive for microbial growth. Alternatively, layering may be important – large and well-developed biofilms may not be ferruginised entirely during exposure to iron, reducing the permanence and strength of the microbial cements after degradation and recycling of the biological scaffold.

The laboratory-scale experiment presented here provides strong evidence that accelerating the biogeochemical cycling of iron can be utilised as a new biotechnology for the stabilisation of fragments. Field trials are required to determine if these biotechnologies can be scaled appropriately. In particular, the delivery of the iron feedstock needs to be controlled. A gradational change in iron mineralisation was observed between the top and the bottom of the artificial slope. To appropriately scale these concepts, the iron-rich solutions would have to be distributed evenly throughout the field site. A gravity-driven drip irrigation system could distribute the ferrous iron solutions from numerous bioreactors throughout the site without introducing oxygen.

212

Slope stabilisation experiment

Optimisation of microbial reduction within the bioreactor is also required to enhance long- term, automated iron oxide dissolution. Complete characterisation of the geophysical properties of the consolidated materials and understanding site-specific hydraulic, geological and environmental conditions will also be essential for field-scale application of this technology. Grain size is an important consideration for microbial cements to be effective. Microbial aggregation is best suited to relatively fine-grained or crushed materials (for example, tailings). Larger (cm- to dm-scale) rock fragments have a reduced surface area to volume ratio, which will reduce the effectiveness of microbial aggregation and microbially- mineralised cements.

8.5. Conclusions Promoting the biogeochemical cycling of iron oxide minerals appears to be an effective method to stabilise mine waste at circumneutral pH. Fermentation-driven iron reduction within an open bioreactor was able to produce an average aqueous iron concentration of 80 ppm. Carbon sources from the bioreactor promoted microbial growth on the article slope, which naturally aggregated grains and provided an organic framework for mineral nucleation. The ferruginisation of microbial biofilms produced a biocement that significantly improved unidirectional physical damage at the macro- (drop-cone test) and nanoscale (nanoindentation). Design improvements are required to appropriately scaleup microbially-accelerated iron cementation for mine waste stabilisation and iron ore mine remediation, but this experiment highlights the future potential to improve the long-term sustainability of the mining sector using these biotechnologies.

Reference Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. Journal of Molecular Biology 1990; 215: 403-410.

Anand R, Verrall M. Biological origin of minerals in pisoliths in the Darling Range of Western Australia. Australian Journal of Earth Sciences 2011; 58: 823-833.

213 Chapter Eight

Armstrong M, Petter R, Petter C. Why have so many tailings dams failed in recent years? Resources Policy 2019; 63.

Beveridge TJ. Role of cellular design in bacterial metal accumulation and mineralization. Annual Reviews of Microbiology 1989; 43: 147-171.

Beveridge TJ, Fyfe WS. Metal fixation by bacterial cell walls. Can. J. Earth Sci. 1985; 22: 1893-1898.

Caneschi WL, Felestrino ÉB, Fonseca NP, Villa MM, Lemes CGdC, Cordeiro IF, et al. Brazilian ironstone plant communities as reservoirs of culturable bacteria with diverse biotechnological potential. Frontiers in Microbiology 2018; 9.

Chao A, Chazdon RL, Colwell RK, Shen TJ. A new statistical approach for assessing similarity of species composition with incidence and abundance data. Ecology Letters 2005; 8: 148-159.

Croghan C, Egeghy PP. Methods of dealing with values below the limit of detection using SAS. Southern SAS User Group 2003; 22: 24.

Ferris FG, Fyfe WS, Beveridge TJ. Metallic ion binding by Bacillus subtilis: implications for the fossilization of microorganisms. Geology 1988; 16: 149-152.

Francis CA, Obraztsova AY, Tebo BM. Dissimilatory metal reduction by the facultative anaerobe Pantoea agglomerans SP1. Appl. Environ. Microbiol. 2000; 66: 543-548.

Gadd GM. Geomycology: biogeochemical transformations of rocks, minerals, metals and radionuclides by fungi, bioweathering and bioremediation. Mycological Research 2007; 111: 3-49.

Gagen EJ, Zaugg J, Tyson GW, Southam G. Goethite reduction by a neutrophilic member of the alphaproteobacterial genus Telmatospirillum. Frontiers in Microbiology 2019b; 10: 2938.

Idris MI, Vodenitcharova T, Hoffman M. Mechanical behaviour and energy absorption of closed-cell aluminium foam panels in uniaxial compression. Materials Science and Engineering: A 2009; 517: 37-45.

Kappler A, Schink B, Newman DK. Fe (III) mineral formation and cell encrustation by the nitrate‐dependent Fe (II)‐oxidizer strain BoFeN1. Geobiology 2005; 3: 235-245.

214

Slope stabilisation experiment

Kennedy C, Scott S, Ferris F. Hydrothermal phase stabilization of 2-line ferrihydrite by bacteria. Chemical Geology 2004; 212: 269-277.

Kõljalg U, Larsson KH, Abarenkov K, Nilsson RH, Alexander IJ, Eberhardt U, et al. UNITE: a database providing web‐based methods for the molecular identification of ectomycorrhizal fungi. New Phytologist 2005; 166: 1063-1068.

Langley S, Beveridge T. Effect of O-side-chain-lipopolysaccharide chemistry on metal binding. Appl. Environ. Microbiol. 1999; 65: 489-498.

Levett A, Gagen E, Shuster J, Rintoul L, Tobin M, Vongsvivut J, et al. Evidence of biogeochemical processes in iron duricrust formation. Journal of South American Earth Sciences 2016; 71: 131-142.

Levett A, Gagen EJ, Diao H, Guagliardo P, Rintoul L, Paz A, et al. The role of aluminium in the preservation of microbial biosignatures. Geoscience Frontiers 2019; 10: 1125-1138.

Li J, Benzerara K, Bernard S, Beyssac O. The link between biomineralization and fossilization of bacteria: insights from field and experimental studies. Chemical Geology 2013; 359: 49-69.

Lovley DR, Phillips EJP. Rapid assay for microbially reducible ferric iron in aquatic sediments. Applied and Environmental Microbiology 1987; 53: 1536-1540.

Mendez MO, Maier RM. Phytostabilization of mine tailings in arid and semiarid environments—an emerging remediation technology. Environmental Health perspectives 2007; 116: 278-283.

Michel FM, Barrón V, Torrent J, Morales MP, Serna CJ, Boily J-F, et al. Ordered ferrimagnetic form of ferrihydrite reveals links among structure, composition, and magnetism. Proceedings of the National Academy of Sciences 2010; 107: 2787-2792.

Monteiro HS, Vasconcelos PM, Farley KA, Spier CA, Mello CL. (U–Th)/He geochronology of goethite and the origin and evolution of cangas. Geochimica et Cosmochimica Acta 2014; 131: 267-289.

215 Chapter Eight

Oggerin M, Tornos F, Rodriguez N, Pascual L, Amils R. Fungal iron biomineralization in Rio Tinto. Minerals 2016; 6: 37.

Orange F, Dupont S, Goff OL, Bienvenu N, Disnar J-R, Westall F, et al. Experimental fossilization of the thermophilic Gram-positive Bacterium Geobacillus SP7A: a long duration preservation study. Geomicrobiology Journal 2014; 31: 578-589.

Phoenix V, Konhauser K. Benefits of bacterial biomineralization. Geobiology 2008; 6: 303- 308.

Phoenix VR, Adams DG, Konhauser KO. Cyanobacterial viability during hydrothermal biomineralisation. Chemical Geology 2000; 169: 329-338.

United States Geological Survey. Iron Ore Statistics and Information. 2018. USGS Reston, VA, 2018, pp. 88-89.

White TJ, Bruns T, Lee S, Taylor J. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. PCR protocols: a guide to methods and applications 1990; 18: 315-322.

Yue JC, Clayton MK. A similarity measure based on species proportions. Communications in Statistics-Theory and Methods 2005; 34: 2123-2131.

216

Chapter 9 Discussion, implications and conclusions

The primary aims of this thesis were to (i) determine the presence and distribution of biosignatures within canga (9.1.), (ii) understand how microbial processes influence canga evolution (9.2.-9.3.) and (iii) develop strategies that could be used to accelerate iron cementation of crushed canga to reform an iron-rich crust (9.4.-9.5.).

9.1. Distributions of biosignatures in canga After close to 1000 hours examining canga thin sections at the microscale using a field emission scanning electron microscope (FE-SEM) during this thesis, I feel confident in concluding that microorganisms play an important role in both mineral dissolution and precipitation within canga. Bacteriomorphic structures were identified in almost every thin section I examined. Many of the microfossils formed abundant biofilms that appeared to contribute to concurrent mineral dissolution and, as these microorganisms were fossilised, the formation of new cements. The heterogeneity of canga is truly remarkable. Transferring samples between different analytical techniques, particularly from electron microscopy to infrared spectroscopy, required removing the conductive coating off samples by polishing with a 0.25 m diamond paste. Polish for slightly too long and I might completely remove the targeted microfossils; keep polishing and an entirely ‘new’ perfectly preserved biofilm may be exposed. All within a ca. 10 – 20 m depth profile.

The thesis work has demonstrated the importance of combining analytical techniques to understand processes of microbial fossilisation and the influences of microorganisms on elemental and mineralogical distributions, which are representative of the evolving duricrust. These data play an important role in the identification and characterisation of direct and indirect biosignatures within the geological record. Remarkably, nanoscale elemental mapping of preserved encrusted cell envelopes revealed a role for aluminium in the long-term (up to 25 Ma) (Monteiro, 2017) preservation of organic and inorganic biosignatures associated with microfossils (Chapter Three). High-resolution mineralogical characterisation could not conclusively discern changes in mineralogy between microbially influenced iron oxide minerals and iron oxide minerals that had precipitated within intercellular voids (Chapter Four).

Chapter Nine

9.2. Canga: a reservoir for iron-associated microbial life At first glance, the longevity of canga appears juxtaposed by the dynamic nature of the canga- associated ecosystems, which promote mineral weathering. In actuality, the biological processes that contribute to the geochemical cycling of iron are essential for the extraordinary erosion resistance of these horizons (Monteiro et al., 2014). The porous nature of canga allows water to pool at various depths throughout the canga profile, which is typically 10 m thick (Dorr, 1964). At the surface, dissolution depressions within the duricrust form ephemeral pools containing evidence of iron-associated microbial life (for example, surface sheens; Fig. 9.1A). Microbial biofilms are frequently encrusted in iron-rich precipitates, immobilised on the cell envelope structures (Fig. 9.1B). Microorganisms surviving in these iron-rich environments have not developed strategies to avoid encrustation, although they must still be able to metabolise and replicate during the initial phases on biomineralisation. Therefore, a delicate balance exists between microbial survival, where they directly and indirectly promote iron reduction, and microbial fossilisation, thus, contribute to the formation on new iron cements in canga (Fig. 9.2).

218

Discussions

Fig. 9.1. (A) An ephemeral pool, common in depressions in the canga plateaus, with a surface sheen biofilm. (B) Unstained transmission electron micrograph, revealing the heterogeneous iron oxide minerals that nucleate on the cells’ surfaces and extracellular polymeric substances.

Below the surface, pore spaces are difficult to study directly but are likely to be critical to the longevity of canga. As iron oxide minerals are dissolved and re-precipitated, these pore spaces constantly evolve, altering the hydrological connectivity and creating diverse niches that support a wide variety of microbial life (Fig. 9.2). The unique microbial communities that survive in these iron-rich environments have substantial biotechnological potential (Caneschi et al., 2018). Given the abundance of iron oxide minerals in canga, it is inevitable that microbial communities capable of using iron as part of energy metabolism would have an ecogenetic advantage. Microbial communities within canga may co-evolve and depend on only a few lineages capable of shuttling electrons to the iron oxide minerals using a variety of mechanisms,

219 Chapter Nine including nanowires (Reguera et al., 2005) and/or via redox-active molecules (for example, flavins) (Breuer et al., 2015).

Paleomicrobial activity throughout the canga profile was studied by collecting drill core samples. Combining nanoscale elemental mapping with Raman spectroscopy-based mineral identification, I concluded that microorganisms were able to completely weather grains. Microbially-mediated mineral weathering is likely to be accelerated by the exudation of organic acids, possibly a response to nutrient deficiencies (potassium and phosphorus) that are mineral bound. During mineral weathering, relatively insoluble metals (for example, aluminium and iron) are placed into solution, which subsequently nucleates on nearby cell envelope surfaces. As these processes are repeated, microfossils texturally replace the grains in canga, leaving only traces of the most insoluble metals (thorium and minor titanium; Chapter Five).

In Chapter Six, elemental mapping of canga at the cm-scale using synchrotron-based X-ray fluorescence microscopy (SXFM), revealed that titanium was not immune to the weathering effect of microorganisms. In regions associated with fossilised microorganisms, titanium precipitates coated nearby grains, likely having been complexed by organic acids exuded by microorganisms and re-precipitated as anatase [TiO2] around nearby grains. Targeted chemical analysis also indicated that a renewable source of phosphorus is supplied to canga in the Carajás Mineral Province, Pará, Brazil, plausibly deriving from African aerosols (Barkley et al., 2019; Okin et al., 2004). X-ray absorption near edge structure (XANES) confirmed that ferrous iron was enriched within the detrital portions of canga compared with the goethitic cements. The ferrous iron from these fragments was relatively easily released from the fragments using a low concentration acid extraction (0.5 N HCl), supporting the non-redox based transformation of magnetite [Fe3O4] to hematite [Fe2O3] via kenomagnetite, a ferrous iron deficient variety of magnetite (Ohmoto, 2003).

220

Discussions

Fig. 9.2. A focal series of backscattered electron micrographs of encrusted cell envelope structures that have been mineralised in the goethite-rich veins throughout iron-rich substrates. As iron- rich solutions percolate throughout the profile, microorganism become fossilised, altering the hydrological connectivity.

9.3. Accelerated canga formation in natural environments The natural environment studied in Chapter Seven highlighted the relatively rapid formation of canga, almost appearing to defy gravity (Fig. 9.3). These natural environments inspired the notion that canga re-cementation can be accelerated and stabilise slopes. In these environments,

221 Chapter Nine extensive microbial biofilms were associated with iron-rich seeps that flowed down relatively steep slopes (up to 50) (Silva et al., 2018) into depression lakes within the duricrusts. Aqueous iron concentrations were greatest in association with the well-developed biofilms, highlighting that microorganisms were responsible for the dissolution of the iron oxide minerals. Examination of the canga samples that formed at the lake-edges revealed large portions of the meniscus-type cements had been influenced by microbial biofilms.

Fig. 9.3. Photograph of hematite-rich fragments that are cemented on a slope, highlighting the relatively fast formation of some iron-rich cements in canga.

9.4. Promoting iron-cycling to accelerate re-cementation Armed with knowledge of relatively rapid iron cement formation on the slopes of the lake- edge, an experiment was designed to model these conditions (Chapter Eight). By promoting iron reduction within an open-air bioreactor and allowing the resulting iron-rich solutions to follow over crushed canga fragments, the resistance of the surface to unidirectional damage was significantly improved. Iron minerals can also nucleate on nearby cell envelopes structures (Ferris et al., 1987; Ferris et al., 1988; Schultze-Lam et al., 1995). This thesis highlights the importance of microbial biofilms as an organic framework to ‘guide’ mineral nucleation and reduce the number of mineral cycling events required to form iron cements (Chapter Eight).

222

Discussions

Therefore, accelerating iron cement formation in canga may be achieved through a two-stage process: microbial growth followed by mineralisation of the biofilm (Fig. 9.2). These processes may occur concurrently or as separate phases. These results highlight that microbial biofilms, which naturally aggregate grains, provide a suitable organic framework to guide mineral nucleation and promote the stabilisation of crushed mine waste.

Together, these results confirm the innate role of microorganisms in the evolution of some of the world’s most physically stable landforms (Monteiro et al., 2018a; Monteiro et al., 2018b; Shuster et al., 2012). They also show that accelerating the biogeochemical cycling of iron oxide minerals may contribute to the development of important biotechnologies for the chemical and physical stabilisation of mine waste, imperative for iron ore remediation strategies in Brazil.

9.5. Harnessing biogeochemical processes for iron ore bioremediation One of the greatest challenges of strategising mine remediation is scaling chemical processes that occur at a nanoscale to geological (km) scales. Microbial biofilms and mats naturally form aggregates, trapping and binding microscale fragments (Burne and Moore, 1987), which would otherwise be readily eroded. These microscale fragments are likely to play an important role in cement formation in canga horizons. Microbial biofilms offer advantages over phytostabilisation strategies (Mendez and Maier, 2007), as microbes are able to survive in low pH and in high metal environments. Bioremediation efforts that aim to use microbial processes take advantage of the widespread distribution of microorganisms, relatively rapid and sustained replication rates and high surface area to volume ratios, which provide reaction sites for biologically-influenced mineralisation (BIFM) of iron oxide minerals. Remarkably, microorganisms are also able to manoeuvre in pore spaces only 30% larger than their own diameter (Männik et al., 2009), allowing them to grow within the microscale pore spaces between crushed fines.

The grain size distribution of the starting material is critical for the successful formation of biocements. In a patent for calcium phosphate biocements, specific ratios of very fine (0.1 – 1 µm), fine (1 – 40 µm) and coarse (40 – 300 µm) materials were required for improved compressive strength (Wenz and Driessens, 2002). The energy requirements and volume of material required for iron ore remediation, precludes the use of these material size distributions for canga re-cementation. Instead, ratios of 3 parts fine-sand (< 0.1 mm), 3 parts course-sand (0.1 – 2 mm) and 1 part gravel (2 – 20 mm) cemented together by chemical iron oxide

223 Chapter Nine precipitates influenced by microbial biofilms are proposed but require strength testing and refining.

Long term, iron ore mining will require the stabilisation of material on relatively steep gradients (up to 45º). Therefore, efforts must be focused on stabilisation of artificial hillslopes. These technologies will be directly transferable to mine waste (and potentially to tailings) stabilisation for other commodities. To account for the monsoonal rainfall and to minimise erosion, the slopes of iron ore pits will have to be stabilised in sections, from the base up. Slopes would have to be sufficiently cemented during the dry (less wet) season, allowing only approximately six months to produce a resilient structure.

Combining the concepts summarized here, a model that would represent the first stage of iron ore mine remediation is provided (Fig. 9.4). Initially, waste material (canga) with a specific grain size distribution would be distributed to cover the bottom two or three pit benches, creating a slope of approximately 30 – 40º. Iron-reduction would be compartmentalised within a bioreactor that was fed with a carbon source and placed above the crushed material. Filtered ferrous iron-rich solutions together with the degraded carbon sources from the bioreactor would be gravity fed to cover the waste material using a pipe network to prevent the abiotic oxidation of the iron in solution during the distribution around the remediation site. The carbon source would promote heterotrophic microbial growth and the formation of a biofilm at the surface. The biofilms would naturally aggregate fine grains and provide an organic framework for mineralisation by the iron-rich solution. Iron biomineralisation will fossilise the microbial biofilms, driving the formation of microbiologically influenced iron cements rather than simply coating grains. Continued iron precipitation after the initial fossilisation of the microbial biofilms would continue to strengthen the existing iron cements. The mineralised biofilms would aggregate and stabilise crushed canga, creating an iron-rich crust that would aid in the restoration of a functional hydrology and provide a substrate for revegetation programmes aiming to restore the native ecosystems. The solutions from the bioreactor would be collected at the base and re-cycled within a closed-loop system. After the stabilisation of the lower bit benches (approximately 12 months), artificial slopes covering the next benches could be established for stabilisation, possibly while maintaining the lower systems.

224

Discussions

Fig. 9.4. A conceptual model of the first stage of a bioremediation strategy for iron ore mines. Specific proportions of crushed waste material (canga) cover the first two or three benches at the base of the closed mine. Iron-reducing bioreactors are placed above loose material. Solutions from the bioreactor are drip-fed throughout the slope. The partially-degraded carbon source leaving the bioreactor would promote heterotrophic microbial growth at the surface to naturally aggregate grains. The iron-rich solutions (up to 300 ppm) produced in the bioreactor and evenly distributed throughout the artificial slope would mineralise the biofilms, forming ferruginous biocements. Continued iron precipitation after cell fossilisation will continue to strengthen these cements. Once the lower section is stabilised, these processes would be repeated for the next three pit benches. Solutions would be collected at the base of the pit and recycled, creating a closed- loop system.

9.6. Final observations Microorganisms have a profound impact on Earth terrestrial surface conditions, controlling the chemistry, mineralogy and aggregate structures throughout soils and the geological record. The distribution of microbial biosignatures and their geochemical and geophysical effects in canga were studied in detail using state-of-the-art analytical instruments. Understanding the natural microbiological processes that contribute to canga evolution was used to accelerate the stabilisation of crushed canga in a laboratory-scale experiment. Accelerating the

225 Chapter Nine biogeochemical cycling of iron may serve as an important strategy for iron ore remediation and potentially the chemical and physical stabilisation of mine tailings associated with other commodities, such as base metals.

References Barkley AE, Prospero JM, Mahowald N, Hamilton DS, Popendorf KJ, Oehlert AM, et al. African biomass burning is a substantial source of phosphorus deposition to the Amazon, Tropical Atlantic Ocean, and Southern Ocean. Proceedings of the National Academy of Sciences USA 2019; 116: 16216-16221.

Breuer M, Rosso KM, Blumberger J, Butt JN. Multi-haem cytochromes in Shewanella oneidensis MR-1: structures, functions and opportunities. Journal of The Royal Society Interface 2015; 12: 20141117.

Burne RV, Moore LS. Microbialites: organosedimentary deposits of benthic microbial communities. Palaios 1987; 2: 241-254.

Dorr JVN. Supergene iron ores of Minas Gerais, Brazil. Economic Geology 1964; 59: 1203- 1240.

Ferris FG, Fyfe WS, Beveridge TJ. Bacteria as nucleation sites for authigenic minerals in a metal-contaminated lake sediment. Chemical Geology 1987; 63: 225-232.

Ferris FG, Fyfe WS, Beveridge TJ. Metallic ion binding by Bacillus subtilis: implications for the fossilization of microorganisms. Geology 1988; 16: 149-152.

Männik J, Driessen R, Galajda P, Keymer JE, Dekker C. Bacterial growth and motility in submicron constrictions. Proceedings of the National Academy of Sciences USA 2009; 106: 14861-14866.

Mendez MO, Maier RM. Phytostabilization of mine tailings in arid and semiarid environments—an emerging remediation technology. Environmental Health perspectives 2007; 116: 278-283.

Monteiro HS, Vasconcelos PM, Farley KA, Spier CA, Mello CL. (U–Th)/He geochronology of goethite and the origin and evolution of cangas. Geochimica et Cosmochimica Acta 2014; 131: 267-289.

226

Discussions

Ohmoto H. Nonredox transformations of magnetite-hematite in hydrothermal systems. Economic Geology 2003; 98: 157-161.

Okin GS, Mahowald N, Chadwick OA, Artaxo P. Impact of desert dust on the biogeochemistry of phosphorus in terrestrial ecosystems. Global Biogeochemical Cycles 2004; 18: 1-9.

Reguera G, McCarthy KD, Mehta T, Nicoll JS, Tuominen MT, Lovley DR. Extracellular electron transfer via microbial nanowires. Nature 2005; 435: 1098-1101.

Schultze-Lam S, Ferris FG, Konhauser KO, Wiese RG. In situ silicification of an Icelandic hot spring microbial mat: implications for microfossil formation. Canadian Journal of Earth Sciences 1995; 32: 2021-2026.

Wenz R, Driessens F. Biocements having improved compressive strength. In: Merck Patent GmbH DD, editor. 6,495,156. Google Patents, U.S. Patent, 2002.

227

Appendix 1 Supplementary information for Chapter Five

Fig. S5.1. (A) Photograph of petrographic thin section of canga from a depth of ~ 4 m from the Serra Sul, Carajás, Pará, Brazil (6.40039117 ºS, 50.39797396 ºW). (B) Backscattered electron micrograph highlighting the regions analysed, each containing abundant numbers of microbial fossils.

Appendix 1

Fig. S5.2. NanoSIMS analysis of a weathered grain shown in Figs. 5.2 – 5.3.

229 Supplementary Information for Chapter Five

Fig. S5.3 NanoSIMS analysis of a texturally replaced grain in canga shown in Figs. 5.4 – 5.5.

230

Appendix 1

Fig. S5.4. NanoSIMS analysis of weathered grains and new cements in canga shown is Figs. 5.7 – 5.8.

231 Supplementary Information for Chapter Five

Fig. S5.5. NanoSIMS analysis of aluminum-enriched regions within canga shown in Figs. 5.9 – 5.10.

232

Appendix 1

233 Supplementary Information for Chapter Five

234

Appendix 1

235 Supplementary Information for Chapter Five

236

Appendix 1

237 Supplementary Information for Chapter Five

238

Appendix 1

239 Supplementary Information for Chapter Five

240

Appendix 2 Supplementary information for Chapter Seven

Note: AS-GB = Lake Amendoim Seep – Globular Biofilm (Fig. 7.2A) AS-BlB = Lake Amendoim Seep – Black Biofilm (Fig. 7.2B) VP-SB = Lake Violão Pool – Surface Biofilm (Not shown) VP-BB = Lake Violão Pool – Bottom Biofilm (Not shown) VS-RB = Lake Violão Seep – Red Biofilm (Fig. 7.2C) VS-OB = Lake Violão Seep – Organic Biofilm (Fig. 7.2D) LV-F = Lake Violão – Foam (Fig. 7.2E) VS-FeS = Lake Violão Seep – Iron seep (Fig. 7.2C) *Shannon: Shannon diversity index for an OTU definition (https://www.mothur.org/wiki/Shannon) **Sobs: the observed richness (https://www.mothur.org/wiki/Sobs) ***Shannon Evenness: Shannon index-based measure of evenness

Appendix 3 Supplementary information for Chapter Eight

Video S8.1. Video of slope of bioreactor at position 3 (16 – 25 cm from the overflow) taken after drop cone test to show mineralised biofilm at surface. Note, at end brittle deformation of the mineralised biofilm. Video available at: https://drive.google.com/drive/folders/1Y_nRv6xOtafPC369UtNdVlwExK-XIzk2?usp=sharing

Please note: Best to use Google Chrome. PowerPoint slide will need to be downloaded.

Appendix 3

Video S8.2. Video of bioreactor treatment slopes at position 3 (16 – 25 cm from the overflow). To demonstrate resilience of surface, the cone was dropped from around 5 mm. Note, this measurement was not included in the data set. Video available at: https://drive.google.com/drive/folders/1Y_nRv6xOtafPC369UtNdVlwExK-XIzk2?usp=sharing

Please note: Best to use Google Chrome. PowerPoint slide will need to be downloaded.

243 Supplementary Information for Chapter Eight

Video S8.3. Video of control treatment slopes at position 3 (16 – 25 cm from the overflow). Video available at: https://drive.google.com/drive/folders/1Y_nRv6xOtafPC369UtNdVlwExK- XIzk2?usp=sharing

Please note: Best to use Google Chrome. PowerPoint slide will need to be downloaded.

244

Appendix 3

Table S8.1. Water chemistry collected from slope stabilisation experiment (Chapter Eight) at week 15. All concentrations in ppm.

Sample ID Al Ca Cr Cu Fe K Mg Mn Na Ni P S R-Bottom 0.00 0.06 0.00 0.00 74.09 54.23 0.00 0.55 814.43 0.02 61.94 0.46 R-Top 0.00 0.03 0.00 0.00 71.60 53.62 0.00 0.54 804.69 0.00 61.85 0.46 R-Overflow 0.00 1.25 0.00 0.00 1.25 129.76 0.50 10.70 1860.01 0.00 0.00 91.14 C-Bottom 0.00 0.00 0.00 0.07 0.01 0.62 0.00 0.01 8.94 0.05 0.12 0.77 C-Top 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.02 0.00 0.00 0.72 C-Overflow 0.00 0.49 0.00 0.00 0.01 0.68 0.23 1.05 0.17 0.00 0.00 0.59

Table S8.2. Water chemistry collected from slope stabilisation experiment (Chapter Eight) at week 16. All concentrations in ppm.

Sample ID Al Ca Cr Cu Fe K Mg Mn Na Ni P S R-Bottom 0.02 0.22 0.00 0.00 109.45 62.87 0.00 0.58 902.51 0.00 51.75 0.47 R-Top 0.01 0.25 0.00 0.00 109.12 62.48 0.00 0.57 900.32 0.01 51.19 0.43 R-Overflow 0.01 1.33 0.00 0.00 17.64 138.72 0.44 14.25 1995.75 0.00 0.00 38.71 C-Bottom 0.00 0.10 0.00 0.00 0.02 0.18 0.00 0.00 0.57 0.00 0.00 1.30 C-Top 0.01 0.06 0.00 0.00 0.02 0.05 0.00 0.00 0.13 0.00 0.00 1.22 C-Overflow 0.00 0.89 0.00 0.00 0.02 1.32 0.35 1.35 0.59 0.00 0.00 1.29

Table S8.3. Water chemistry collected from slope stabilisation experiment (Chapter Eight) at week 17. Dash (-) represents samples not collected. All concentrations in ppm.

Sample ID Al Ca Cr Cu Fe K Mg Mn Na Ni P S R-Bottom 0.00 0.31 0.00 0.02 83.49 61.64 0.00 0.41 909.62 0.02 56.82 0.41 R-Top 0.00 0.25 0.00 0.00 84.88 61.64 0.00 0.42 912.46 0.00 57.93 0.40 R-Overflow ------C-Bottom 0.00 0.09 0.00 0.00 0.00 0.03 0.00 0.00 0.40 0.00 0.00 1.19 C-Top 0.00 0.12 0.00 0.00 0.00 0.00 0.00 0.00 0.19 0.00 0.00 1.23 C-Overflow ------

245 Supplementary Information for Chapter Eight

Table S8.4. Water chemistry collected from slope stabilisation experiment (Chapter Eight) at week 19. All concentrations in ppm.

Sample ID Al Ca Cr Cu Fe K Mg Mn Na Ni P S R-Bottom 0.00 0.20 0.00 0.01 21.28 53.96 0.00 0.07 818.70 0.00 138.93 0.23 R-Top 0.00 0.43 0.00 0.00 33.77 52.78 0.00 0.13 800.70 0.00 109.76 0.32 R-Overflow 0.00 0.31 0.00 0.00 2.63 91.64 0.08 1.40 1196.20 0.00 0.03 14.78 C-Bottom 0.00 0.55 0.00 0.01 0.05 13.39 0.00 0.00 170.78 0.00 14.98 18.26 C-Top 0.00 0.14 0.00 0.00 0.00 10.98 0.00 0.00 139.55 0.00 12.45 15.13 C-Overflow 0.00 0.27 0.00 0.00 0.00 10.99 0.10 0.06 141.55 0.00 0.00 60.22

Table S8.5. Water chemistry collected from slope stabilisation experiment (Chapter Eight) at week 23. All concentrations in ppm.

Sample ID Al Ca Cr Cu Fe K Mg Mn Na Ni P S R-Bottom 0.02 0.37 0.01 0.18 56.51 61.25 0.00 0.21 879.67 0.02 82.72 0.26 R-Top 0.02 0.31 0.00 0.00 57.33 60.60 0.00 0.21 882.38 0.00 76.61 0.23 R-Overflow 0.01 0.26 0.00 0.01 0.32 221.85 0.05 0.22 3387.10 0.00 0.05 44.05 C-Bottom 0.01 0.22 0.00 0.00 0.00 1.02 0.00 0.00 9.78 0.00 0.29 1.48 C-Top 0.01 0.20 0.00 0.00 0.00 0.82 0.00 0.00 8.94 0.00 0.27 1.22 C-Overflow 0.03 0.48 0.00 0.00 0.06 37.60 0.21 0.11 438.68 0.00 0.00 141.30

Table S8.6. Water chemistry collected from slope stabilisation experiment (Chapter Eight) at week 25. Dash (-) represents samples not collected. All concentrations in ppm.

Sample ID Al Ca Cr Cu Fe K Mg Mn Na Ni P S R-Bottom 0.02 0.30 0.00 0.03 140.15 68.84 0.34 0.60 897.91 0.01 17.11 0.86 R-Top 0.05 0.29 0.00 0.00 162.15 72.54 0.43 0.73 948.62 0.00 11.53 0.99 R-Overflow ------C-Bottom 0.01 0.16 0.00 0.01 0.03 0.69 0.00 0.00 8.39 0.00 0.37 1.84 C-Top 0.01 0.18 0.00 0.00 0.00 0.58 0.00 0.00 7.92 0.00 0.39 1.90 C-Overflow ------

246

Appendix 3

Table S8.7. Water chemistry collected from slope stabilisation experiment (Chapter Eight) at week 26. All concentrations in ppm.

Sample ID Al Ca Cr Cu Fe K Mg Mn Na Ni P S R-Bottom 0.03 0.27 0.00 0.00 47.83 57.55 0.00 0.16 834.60 0.00 68.55 0.43 R-Top 0.03 0.21 0.00 0.00 28.23 58.47 0.00 0.12 828.87 0.00 66.15 0.39 R-Overflow 0.04 0.21 0.00 0.01 0.17 169.24 0.00 0.12 2456.37 0.00 0.14 69.03 C-Bottom 0.02 0.17 0.00 0.00 0.00 0.45 0.00 0.01 3.64 0.00 0.00 0.75 C-Top 0.02 0.18 0.00 0.00 0.00 0.33 0.00 0.01 2.56 0.00 0.00 0.69 C-Overflow 0.02 0.69 0.00 0.00 0.00 22.76 0.49 0.41 269.88 0.00 0.00 55.18

Table S8.8. Water chemistry collected from slope stabilisation experiment (Chapter Eight) at week 27. Dash (-) represents samples not collected. All concentrations in ppm.

Sample ID Al Ca Cr Cu Fe K Mg Mn Na Ni P S R-Bottom 0.02 0.23 0.00 0.00 44.86 57.77 0.00 0.13 795.75 0.00 55.63 0.21 R-Top 0.01 0.26 0.00 0.00 40.02 58.29 0.00 0.12 798.18 0.00 54.76 0.30 R-Overflow 0.01 0.33 0.00 0.01 1.78 217.70 0.09 1.57 3196.13 0.00 0.21 5.52 C-Bottom 0.01 0.13 0.00 0.00 0.00 0.32 0.00 0.01 1.48 0.00 0.08 0.41 C-Top 0.01 0.16 0.00 0.00 0.00 0.11 0.00 0.01 1.07 0.00 0.07 0.35 C-Overflow ------

Table S8.9. Water chemistry collected from slope stabilisation experiment (Chapter Eight) at week 27. All concentrations in ppm.

Sample ID Al Ca Cr Cu Fe K Mg Mn Na Ni P S R-Bottom 0.02 0.40 0.00 0.00 74.52 58.09 0.01 0.21 833.94 0.00 48.05 0.37 R-Top 0.03 0.41 0.00 0.00 74.95 58.75 0.01 0.21 836.51 0.00 48.26 0.31 R-Overflow 0.01 0.53 0.00 0.00 0.69 176.98 0.26 1.34 2708.83 0.00 0.08 28.64 C-Bottom 0.00 0.17 0.00 0.00 0.00 0.25 0.00 0.00 1.94 0.00 0.09 0.45 C-Top 0.00 0.22 0.00 0.00 0.00 0.01 0.00 0.00 0.75 0.00 0.09 0.45 C-Overflow 0.03 1.03 0.00 0.00 0.00 6.58 0.59 0.92 73.80 0.00 0.00 3.06

247