A Palaeoenvironmental History of the Paroo and Warrego Regions, Australia: a Multi-Proxy, Multi-Site Approach

A Palaeoenvironmental History of the Paroo and Warrego Regions, Australia: a Multi-proxy, Multi-site Approach

Lucyna M. Gayler BA (Hons)

A thesis submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy (Environmental Science), University of Newcastle.

January 2008

This work contains no material which has been accepted for the award of any other degree or diploma in any university or other tertiary institution and, to the best of my knowledge and belief, contains no material previously published or written by another person, except where due reference has been made in the text. I give consent to this copy of my thesis, when deposited in the University Library, being made available for loan and photocopying subject to the provisions of the Copyright Act 1968.

I hereby certify that the work embodied in this Thesis is the result of original research, the greater part of which was completed subsequent to admission to candidature for the degree.

Acknowledgements

The completion of this research would not have been possible without support from many people and organisations. Thus I would like to extend my sincere thanks to: My family and friends for the unwavering support and patience as well as bravely lending a hand with fieldwork under the blazing desert sun, My husband Maciek for help with the re-development of the pollen database and setting up the digital herbarium, My supervisors Dr Stuart Pearson and A/Prof Tina Offler for loads of encouragement and always ready assistance, Prof Brian Timms for passionate introduction to the Paroo lakes, assistance with trips to the fieldwork sites, and sharing his immense knowledge about the Paroo/Warrego Region, David O’Brian for many fruitful discussions and editorial support, Prof Eric Colhoun for valuable comments on the manustript, Dr Tim Rolph for introduction to the field of mineral magnetic susceptibility and assistance with its measurements as well as the laser particle size analysis, and Prof Greg Skilbeck for assistance with magnetic susceptibility measurements by the Multi-sensor Core Logger, A/Prof Bob Loughran for assistance with 137Caesium analysis and interpretation, Dr Matt Cupper and Dr Jeff Parr for inspiration in the development of new pollen/phytolith processing protocol, Prof Patrick DeDeckker, Dr Adriana Garcia, Dr Doreen Bowdery, Prof Geoff Hope, Dr Mike Macphail, Dr John Magee, Dr Peter Gell, and Joan Powling for their keen interest in this project and help with identifications of different fossils, The Royal Botanic Garden Herbarium, Sydney, and Don McNair for assistance with plant identifications, Ed Rhodes for assistance with OSL dating of the Paroo/Warrego sediments, The technical officers: Chris Dever, Peter Loughran, and Richard Bale for assistance with laboratory work, Olivier Rey-Lescure for tips on image editing, The School of Environmental and Life Sciences and the Department of Geography, University of Newcastle, for financial contribution towards fieldwork and laboratory analysis, Jenny and Mark Handley and Danny MacKellar of Queensland Parks and Wildlife Service, Robin and Rhonda Davis (the managers) and John and Heather Buster (the owners) of Rockwell-Blue Lakes Station, Ray and Wilga Bremner of Muella Station, Garth and Mary Lou Davis of Yandaroo Station, James and Cheryl Hatch of Wombah Station, Bruce Sharp of Comeroo Sation, and Reg and Claudia Collins of Bloodwood Station for their hospitality and access to the lakes.

iii Table of Contents

Acknowledgements iii Table of Contents iv List of Tables xi List of Figures xii Abstract xvii Chapter 1 Introduction 1 1.1 Palaeoenvironmental research in arid and semi-arid inland lakes 1 1.1.1 The beginnings of the arid and semi-arid palaeoenvironmental studies in Australia: large lake research 2 1.1.2 Benefits of records from smaller lakes 2 1.1.3 Benefits of a holistic, multi-proxy, multi-site approach 3 1.2 The Paroo/Warrego Region: an attractive study area 3 1.3 The research objectives 5 1.4 Content overview 5 Chapter 2 The Paroo/Warrego Region: Background information 9 2.1 Introduction 9 2.2 Location of the study area 9 2.3 Climate 11 2.3.1 General outline 11 2.3.2 Historical rainfall patterns 11 2.4 Geology and soils 14 2.5 Geomorphology and hydrology 15 2.5.1 The Paroo and Warrego Rivers 15 2.5.2 Wetlands 17 2.5.3 Erosion 18 2.5.4 The erosion-deposition cycles in a saline lake/playa-lunette system 19 2.6 Vegetation 23 2.6.1 Modern vegetation 23 2.6.1.1 Mulga dominated shrublands and woodlands 23 2.6.1.2 Shrubby and grassy semi-arid woodlands 24 2.6.1.3 Arid riverine chenopod shrubland 24 2.6.1.4 Saline and freshwater wetlands 25 2.6.1.5 Woody shrubs 26 2.6.2 Vegetation history 26 2.7 Fire history 28 2.8 Human impact 29 2.8.1 Aboriginal occupation 29 2.8.2 European settlement 29 Chapter 3 Methods 33 3.1 Introduction 33 3.2 Site selection 33 3.3 Field methods and sample collection 34 3.3.1 Lake and dune cores 34 3.3.2 Surveying 35 3.3.3 Vegetation and modern pollen 35 3.4 Laboratory processing of the samples and cores - summary 36 3.5 Sediment description 37 iv 3.5.1 Colour 37 3.5.2 Stratification and sharpness of boundary 38 3.5.3 Bioturbation 38 3.5.4 Abundance/content 38 3.6 Sediment texture and particle size analysis 38 3.6.1 Sediment texture 38 3.6.2 Particle size analysis 38 3.7 Mineral Magnetic Susceptibility 38 3.7.1 Multi-Sensor Core Logger (MSCL) 39 3.7.2 Bartington MS2B sensor 39 3.7.3 Frequency dependent susceptibility 39 3.8 Organic matter content 40 3.9 Carbonates 40 3.9.1 Hydrochloric acid (HCl) test 40 3.9.2 Loss on ignition 40 3.10 Sediment salinity and pH 40 3.10.1 Electrical conductivity and salinity 40 3.10.2 pH 41 3.11 Gypsum 41 3.11.1 Visual description 41 3.11.2 Drying at 105oC 41 3.11.3 Portable Infrared Mineral Analyser (PIMA) 42 3.12 Mineral composition 42 3.13 Plant identification 42 3.14 Pollen 43 3.14.1 Sampling 43 3.14.2 Pollen extraction from sediment 43 3.14.3 Reference pollen 46 3.14.4 Pollen analysis 46 3.14.5 Pollen diagrams 47 3.15 Biogenic silica: phytoliths, sponge spicules, and diatoms 47 3.15.1 Phytolith analysis 48 3.15.2 Additional phytolith extraction 48 3.15.3 Sponge spicules 48 3.15.4 Diatoms 48 3.16 Other fossils 49 3.16.1 Charophytes, macrophytes, and invertebrates 49 3.16.2 Charcoal 49 3.17 Dating 49 3.17.1 137Caesium 49 3.17.2 Radiocarbon 50 3.17.3 Optically Stimulated Luminescence 50 Chapter 4 Proxies and methods: discussion 51 4.1 Introduction 51 4.2 Sediment colour 51 4.3 Bioturbation 53 4.4 Sediment texture and particle size analysis 53 4.5 Mineral Magnetic Susceptibility 54 4.5.1 Palaeoenvironmental significance 54 4.5.2 Comparability of the results from the Multi-Sensor Core Logger (MSCL) and Bartington MS2B sensor 56 4.5.3 Future directions 57 4.6 Organic matter 57 4.6.1 Selected palaeoenvironmental implications 57 4.6.2 Loss on ignition estimation 58 4.7 Carbonates 59 4.7.1 Palaeoenvironmental significance 59 4.7.2 Comparison of quantification methods 59

v 4.8 Sediment salinity and pH 61 4.8.1 General comments 61 4.8.2 Sediment permeability (relationship with pH and salinity changes) 61 4.8.3 Salinity (electrical conductivity): impacts of gypsum and particle size 62 4.9 Gypsum 62 4.9.1 Habit, location, and orientation within the matrix and their palaeoenvironmental implications 62 4.9.2 Content quantification 64 4.10 Sulphides 65 4.11 Mineral composition 66 4.11.1 Clay minerals 66 4.11.2 Short-Wave Infrared Reflectance (SWIR) 66 4.12 Multi-fossil studies 67 4.12.1 Vegetation proxies: pollen and phytoliths – general comments 67 4.12.2 ‘Unknown type 1’ pollen/spore 70 4.12.3 Pollen: comments on selected aspects of extraction methodology, recovery/preservation trends, and analysis 72 4.12.4 Phytoliths: selected comments on the extraction methodology and analysis 73 4.12.5 Sponge spicules 75 4.12.6 Diatoms 77 4.12.7 Fossil 1 79 4.12.8 Macrophytes (seeds) and charophytes (oospores) 81 4.12.9 Invertebrates: gastropods, ostracods, cladocerans, and chironomids 82 4.12.10 Fossil extraction methods: summary and future progress 84 4.12.10.1 Microfossils including pollen and biogenic silica 84 4.12.10.2 Other micro- and macro-fossils 84 4.13 Charcoal 85 4.14 Continuity of records, accumulation/deflation rates, and dating 86 4.14.1 Accumulation and deflation processes 86 4.14.2 Dating 87 4.14.2.1 137Caesium 87 4.14.2.2 Radiocarbon 88 4.14.2.3 Optically Stimulated Luminescence 88 Chapter 5 Lake Bindegolly 89 5.1 Lake Bindegolly 89 5.1.1 Site and lake overview 89 5.1.2 Results 92 5.1.2.1 Dating: 137Caesium 92 5.1.2.2 Pollen 94 5.1.2.3 Phytoliths 94 5.1.2.4 Other biogenic silica 94 5.1.2.5 Other signs of plant and animal life 96 5.1.2.6 Organic matter and carbonate content 97 5.1.2.7 Magnetic susceptibility 97 5.1.2.8 Gypsum 99 5.1.2.9 Mineral composition: clay content 100 5.1.3 Palaeoenvironmental reconstruction of events 100 Chapter 6 Currawinya Wetlands: Lake Wyara and Lake Numalla 103 6.1 Currawinya Wetlands 103 6.2 Lake Wyara 103 6.2.1 Site and lake overview 103 6.2.2 Results 107 6.2.2.1 Gypsum 108 6.2.2.2 Pollen 109 6.2.2.3 Phytoliths 109 6.2.2.4 Other biogenic silica 109 vi 6.2.2.5 Other signs of plant and animal life 109 6.2.2.6 Magnetic susceptibility 113 6.2.2.7 Mineral composition: clay content 113 6.2.3 Palaeoenvironmental reconstruction of events 113 6.3 Lake Numalla 116 6.3.1 Site and lake overview 116 6.3.2 Results 119 6.3.2.1 Dating: 137Caesium 120 6.3.2.2 Pollen 120 6.3.2.3 Phytoliths 122 6.3.2.4 Other biogenic silica 122 6.3.2.5 Other signs of plant and animal life 122 6.3.2.6 Magnetic susceptibility 123 6.3.2.7 Mineral composition: clay content 123 6.3.3 Palaeoenvironmental reconstruction of events 125 Chapter 7 Rockwell-Wombah system: Mid Blue Lake and Lake Wombah 129 7.1 Mid Blue Lake 129 7.1.1 Site and lake overview 129 7.1.2 Results 135 7.1.2.1 Lake core p1: 137Caesium 137 7.1.2.2 Lake core p1: gypsum 137 7.1.2.3 Lake core p1: pollen 138 7.1.2.4 Lake core p1: phytoliths 138 7.1.2.5 Lake core p1: other biogenic silica 138 7.1.2.6 Lake core p1: other signs of plant and animal life 140 7.1.2.7 Lake core p1: organic matter and carbonate content 140 7.1.2.8 Lake core p1: magnetic susceptibility 142 7.1.2.9 Clay lunette: gypsum 142 7.1.2.10 Clay lunette: pollen 144 7.1.2.11 Clay lunette: phytoliths 144 7.1.2.12 Clay lunette: other biogenic silica 144 7.1.2.13 Clay lunette: magnetic susceptibility 146 7.1.2.14 Swale: gypsum 148 7.1.2.15 Swale: pollen 150 7.1.2.16 Swale: phytoliths 152 7.1.2.17 Swale: other biogenic silica 152 7.1.2.18 Swale: magnetic susceptibility 154 7.1.2.19 Gypsum lunette: gypsum 155 7.1.2.20 Gypsum lunette: pollen 156 7.1.2.21 Gypsum lunette: phytoliths 156 7.1.2.22 Gypsum lunette: other biogenic silica 156 7.1.2.23 Gypsum lunette: magnetic susceptibility 157 7.1.2.24 Gypsum lunette: particle size 159 7.1.3 Palaeoenvironmental reconstruction of events 159 7.2 Lake Wombah 163 7.2.1 Site and lake overview 163 7.2.2 Results 165 7.2.2.1 137Caesium 165 7.2.2.2 Pollen 166 7.2.2.3 Phytoliths and other biogenic silica 168 7.2.2.4 Other signs of plant and animal life 168 7.2.2.5 Magnetic susceptibility 169 7.2.2.6 Gypsum 169 7.2.3 Palaeoenvironmental reconstruction of events 171 Chapter 8 Cuttaburra Channels: Cummeroo Waterhole 175 8.1 Cummeroo Waterhole 175 vii 8.1.1 Site overview 175 8.1.2 Results 175 8.1.2.1 137Caesium 177 8.1.2.2 Pollen 177 8.1.2.3 Phytoliths 180 8.1.2.4 Other biogenic silica 181 8.1.2.5 Other signs of plant and animal life 181 8.1.2.6 Organic matter and carbonate content 181 8.1.2.7 Magnetic susceptibility 181 8.1.2.8 Particle size 181 8.1.2.9 Mineral composition: clay content 183 8.1.3 Palaeoenvironmental reconstruction of events 183 Chapter 9 Bloodwood Station lakes: Lower Bell Lake and Palaeolake 187 9.1 Bloodwood Station lakes 187 9.2 Lower Bell Lake 187 9.2.1 Site and lake overview 187 9.2.2 Results 189 9.2.2.1 Gypsum 190 9.2.2.2 Other crystals: white balls 193 9.2.2.3 Pollen 194 9.2.2.4 Phytoliths 194 9.2.2.5 Other biogenic silica 194 9.2.2.6 Other signs of plant and animal life 194 9.2.2.7 Magnetic susceptibility 196 9.2.3 Palaeoenvironmental reconstruction of events 199 9.3 Palaeolake 203 9.3.1 Site and lake overview 203 9.3.2 Results 203 9.3.2.1 Core B: gypsum 205 9.3.2.2 Core B: pollen 208 9.3.2.3 Core B: phytoliths and other biogenic silica 208 9.3.2.4 Core B: other signs of plant and animal life 208 9.3.2.5 Core B: magnetic susceptibility 208 9.3.2.6 Core K: gypsum 213 9.3.2.7 Core K: pollen 218 9.3.2.8 Core K: phytoliths 218 9.3.2.9 Core K: other biogenic silica 218 9.3.2.10 Core K: other signs of plant and animal life 218 9.3.2.11 Core K: magnetic susceptibility 221 9.3.2.12 Northern red sand dune: gypsum 221 9.3.2.13 Northern red sand dune: pollen 225 9.3.2.14 Northern red sand dune: phytoliths and other biogenic silica 227 9.3.2.15 Northern red sand dune: other signs of plant and animal life 227 9.3.2.16 Northern red sand dune: organic matter and carbonates 228 9.3.2.17 Northern red sand dune: magnetic susceptibility 228 9.3.2.18 Eastern gypsum lunette: gypsum 228 9.3.2.19 Eastern gypsum lunette: pollen 230 9.3.2.20 Eastern gypsum lunette: phytoliths and other biogenic silica 230 9.3.2.21 Eastern gypsum lunette: other signs of plant and animal life 232 9.3.2.22 Eastern gypsum lunette: magnetic susceptibility 234 9.3.3 Palaeoenvironmental reconstruction of events 236 Chapter 10 Yandaroo-Delta Station lakes: Lake Willeroo and Lake Yandaroo 241 10.1 Yandaroo-Delta Station Lakes: Lake Willeroo and Lake Yandaroo 241 10.2 Lake Willeroo 243 10.2.1 Lake and site description 243 10.2.2 Results 244 viii 10.2.2.1 Pollen 244 10.2.2.2 Phytoliths 247 10.2.2.3 Other biogenic silica 248 10.2.2.4 Other signs of plant and animal life 248 10.2.2.5 Carbonates 248 10.2.2.6 Magnetic susceptibility 248 10.2.2.7 Mineral composition: clay content 248 10.2.3 Palaeoenvironmental reconstruction of events 251 10.3 Lake Yandaroo 252 10.3.1 Lake and site description 252 10.3.2 Results 253 10.3.2.1 Augered core: 137Caesium 253 10.3.2.2 Augered core: pollen 256 10.3.2.3 Augered core: phytoliths 256 10.3.2.4 Augered core: other biogenic silica 256 10.3.2.5 Augered core: other signs of plant and animal life 256 10.3.2.6 Augered core: magnetic susceptibility 258 10.3.2.7 PVC core: 137Caesium 260 10.3.2.8 PVC core: pollen 261 10.3.2.9 PVC core: phytoliths 263 10.3.2.10 PVC core: other biogenic silica 264 10.3.2.11 PVC core: other signs of plant and animal life 264 10.3.2.12 PVC core: magnetic susceptibility 264 10.3.3 Palaeoenvironmental reconstruction of events 264 Chapter 11 Palaeoenvironmental history of the Paroo and Warrego Region: regional trends 269 11.1 Introduction 269 11.2 The oldest event: lacustrine phase 269 11.3 Onset of aridity: red sand dune mobilisation (leading into the Last Glacial Maximum?) 272 11.4 Enhanced seasonality/periodicity: lunette formation (LGM to late Pleistocene?) 274 11.4.1 Gypsum-rich ephemeral lake and gypsum dune formation 274 11.4.2 Gypsum deficient ephemeral lake and clay lunette formation 275 11.4.3 Periods of depressed water budgets: soil formation 276 11.4.4 The general weather patterns in the Paroo/Warrego Region from the last glacial phase to the early Holocene 277 11.4.4.1 Wind direction and intensity 277 11.4.4.2 Seasonality/periodicity enhancement and rainfall 278 11.5 Late Pleistocene to Holocene amelioration and late Holocene changes 279 11.6 The late 1890s-1940s drought 281 11.7 General amelioration of conditions in the period of 1950s to present and increase in woody shrubs and trees 282 11.8 Lake Willeroo and Lake Yandaroo: the Warrego River floodplain 283 11.9 Trends in vegetation successions and a lesson in resiliency: a synopsis 284 Chapter 12 Conclusions and future directions 287 12.1 History and dynamics of the Paroo/Warrego landscapes 287 12.2 Multi-proxy, multi-site approach 288 12.3 Future directions 288 References: 291 Appendix 1 Indicative native plant species of four vegetation groups dominant in central part of the Paroo/Warrego Region, adapted from Keith (2004) and Kingsford and Porter (1999) 309 Appendix 2 Core and auger hole locations 319 Appendix 3 ix Pollen/spore/carbonized particle/phytolith extraction method using gravity settling and heavy liquid separation 321 Appendix 4 Modern reference pollen extraction 323 Appendix 5 Pollen types 325 Appendix 6 Phytolith types 329 Appendix 7 Charophytes: identified taxa and their habitat requirements 331 Appendix 8 Invertebrates: taxa list and their habitat requirements 333 Appendix 9 Dating results: 137Caesium, Radiocarbon, and Optically Stimulated Luminescence 335

CD content: The Newcastle Pollen Collection The Paroo/Warrego Herbarium Collection Excel spreadsheets with raw data: • pollen (pollen_counts_cd_attachment.xls) • biosilica (biosilica_2sg_cd_attachment.xls) • other sedimentary characteristics (sed_data_cd_attachment.xls)

x List of Tables

Table 3.1 Different specific gravities (SG) of heavy liquid used by different researchers in pollen and phytolith extraction. 46 Table 4.1 Summary of main factors and processes affecting magnetic susceptibility (MS) and frequency dependent susceptibility (FDS) of sediments and soils. 55 Table 4.2 The clay types recorded in the sediments from the Paroo/Warrego Region and temperatures at which they achieve peak weight losses. 58 Table 4.3 Comparison between carbonate content estimates (%) by LOI method and HCl test. The LOI values were rounded to match the categories used in HCl test. The analysis includes all sample sets presented in this thesis. 60 Table 4.4 Comparison between two methods for visual estimation of gypsum content: from fresh core (A) and from sediments dried at 105oC (B). The analysis excluded sample sets for which no gypsum was recorded and Palaeolake core B for which the gypsum estimate from fresh core was based on 0-2 scale. The analysed sets included: L. Bindegolly core 4; L. Wyara core 2; Mid Blue L.: core p1, clay dune auger hole, swale auger hole, and gypsum lunette auger hole; L. Wombah core 3; Lower Bell L. core G; Palaeolake: core K, northern red sand dune auger hole, and eastern gypsum lunette auger hole. 64 Table 4.5 The main characteristics of pollen and phytoliths relevant to palaeoenvironmental studies. 68 Table 4.6 Summary of pollen and phytolith counts for lake cores and lake and dune augered sediment. 71 Table 4.7 Summary of freshwater sponge spicule occurrence in lake and dune sediments. 76 Table 4.8 Summary of diatom occurrence in lake and dune sediments. 78 Table 4.9 Summary of Fossil 1 occurrence in lake and dune sediments. 80 Table 4.10 Summary of charophyte oospore occurrence in lake and dune sediments. 83 Table 4.11 The summary of modern and fossil invertebrate occurrence within lake sediments. (*core collected from Palaeolake in 1998 by S. Pearson, two chironomid species: Eukiefferiella sp. at 122-130cm and Cladotanytarsus sp. at 172-180cm identified by S. Dimitriadis) 84

xi List of Figures

Figure 1.1 Location of the Paroo/Warrego Region with respect to the main climatic systems. (Source: Bowler et al., 2001; Hesse et al., 2004; McBride and Nicholls, 1983) 4 Figure 1.2 The outline of the thesis content. 6 Figure 2.1 The catchment boundaries of the Paroo and Warrego rivers and the location of the studied lakes. (Image source: Natmap Raster 2003) 10 Figure 2.2 Climate data for Thargomindah. Rainfall is based on observations recorded between 1879 and 2005, evaporation and temperature: 1879-2004, and wind rose and wind speed: 1957-1999. (Source: Bureau of Meteorology, 2006) 12 Figure 2.3 Climate data for Bourke. Rainfall, evaporation, and temperature are based on observations recorded between 1871 and 1996, wind rose: 1881-1996, and wind speed: 1908-1994. (Source: Bureau of Meteorology, 2006) 13 Figure 2.4 Short term rainfall variability for Thargomindah and Bourke based on annual precipitation. (Source: Bureau of Meteorology, 2006) 14 Figure 2.5 Longer term rainfall patterns for Thargomindah and Bourke based on 5 year moving average of annual precipitation. (Source: Bureau of Meteorology, 2006) 14 Figure 2.6 Dust storm approaching Lake Bindegolly. 20 Figure 3.1 The sediment extraction by coring (1 & 2) and augering (3). 34 Figure 3.2 The guidelines for processing sediment cores. 36 Figure 3.3 The guidelines for processing augered material. 37 Figure 3.4 Sediment samples after 80 hours of drying at 105oC. The white particles within the red (top) and grey (bottom) sediment are desiccated gypsum crystals. 42 Figure 3.5 Sieving of the sediment through a 250 µm mesh followed by still settling. 45 Figure 4.1 Comparison between carbonate content (%) estimated by LOI and HCl test for all samples presented in this thesis. 60 Figure 4.2 Unknown type 1 pollen/spore. 72 Figure 4.3 Selected examples of variability in the freshwater sponge spicule forms. 77 Figure 4.4 Fossil 1: a – surface view; b – cross-section view. 79 Figure 4.5 Charophyte oogonia. 82 Figure 5.1 Map of Lake Bindegolly. (Image source: Google Earth 2006; Geological Information Source: Eulo 1:250 000 Geological Series Map Sheet SH 55-1, 1971) 89 Figure 5.2 The cliff on the western margin of dry Lake Bindegolly. The pink hue of the image is caused by approaching dust storm. 90 Figure 5.3 The close-up of the Lake Bindegolly’s western cliff looking down toward the lake. 91 Figure 5.4 Lake Bindegolly: the gibber plain extending westwards from the western cliff edge. 91 Figure 5.5 The eastern shore of Lake Bindegolly during a full stage. 92 Figure 5.6 Sediment description for Lake Bindegolly cores 4a and b. 93 Figure 5.7 Pollen counts for Lake Bindegolly cores 4a & b. 95 Figure 5.8 Biogenic silica counts for Lake Bindegolly cores 4a & b. 96 Figure 5.9 Lake Bindegolly cores 4a & b: sediment diagram. 98 Figure 5.10 Infrared spectrometry data (SWIR) for Lake Bindegolly cores 4a & b. 99 Figure 6.1 Map of Lake Wyara and Lake Numalla. (Main image source: Google Earth 2006; Inset image source: Natmap Raster 2003; Source of geomorphological information: Timms, 1997b, 1998c, 2006) 104

xii Figure 6.2 A line of dead trees along one of the beach ridges on the southern margin of Lake Wyara (located lower and closer to the lake centre than the present line of live trees). Halosarcia spp. in the foreground. 106 Figure 6.3 Sediment description for Lake Wyara core 2. 107 Figure 6.4 Lake Wyara core 2 gypsum: 1a & b – partly dissolved lenticular and lenticular twin gypsarenite at 75cm; 2 – lenticular and lenticular twin gypsarenite with no signs of dissolution at 82cm; 3 – gypsite band within grey clay at 126cm. 108 Figure 6.5 Pollen counts for Lake Wyara core 2. 110 Figure 6.6 Biogenic silica counts for Lake Wyara core 2. 111 Figure 6.7 Lake Wyara core 2: sediment diagram. 112 Figure 6.8 Infrared spectrometry data (SWIR) for Lake Wyara core 2. 114 Figure 6.9 Lake Numalla: A line of dead trees on the lower beach and a line of live trees on the top of the beach ridge in the background. The green ground cover along the beach consists mainly of spiny sedge. 116 Figure 6.10 Lake Numalla: The tree line at the top of the southwestern beach with black box in the foreground of the image. Woody shrubs visible in the background. 118 Figure 6.11 Lake Numalla: Woody shrubs (mainly turpentine and hopbush) on eroding red sands behind the beach. 118 Figure 6.12 Sediment description for Lake Numalla core 1. 119 Figure 6.13 Pollen counts for Lake Numalla core 1. 121 Figure 6.14 Biogenic silica counts for Lake Numalla core 1. 122 Figure 6.15 Lake Numalla core 1: sediment diagram. 124 Figure 6.16 Infrared spectrometry data (SWIR) for Lake Numalla core 1. 125 Figure 7.1 Map of Mid Blue Lake. (Main image source: Eulo SG55-1 airphoto 63, 1995; Inset image source: Google Earth 2006) 130 Figure 7.2 The cliff on the western margin of the dry Mid Blue Lake featured in the transect in Figure 7.4. 131 Figure 7.3 Western cliff of the Mid Blue Lake’s basin. 131 Figure 7.4 Transect across Mid Blue Lake with relative positions of the cores analysed in this study. See map in Figure 7.1 for their exact location. 132 Figure 7.5 Dead shrub in the middle of Mid Blue Lake surrounded by an area of patchy, thin salt crust. 133 Figure 7.6 Mid Blue Lake: a view west from the top of the inner clay lunette. Large samphires in the foreground and smaller ones on the beach (mid-field). 133 Figure 7.7 Mid Blue Lake: a view west from the top of the inner clay lunette showing the low shrub cover of samphires and the soft (fluffy and powdery) clays in the foreground. 134 Figure 7.8 Mid Blue Lake: a view from clay lunette east toward the outer gypsum lunette in the far background. Large samphires in the foreground and low younger samphires visible as a brown stripe between the dunes (in the swale). 134 Figure 7.9 Augering of the outer gypsum lunette looking east with woody shrubs in the far background. 135 Figure 7.10 Sediment description for Mid Blue Lake core p1. 136 Figure 7.11 Mid Blue core p1 gypsum: 1 – gypsite band at 7cm depth; 2 – rosette composed of lenticular gypsarenite crystals at 12cm. 137 Figure 7.12 Pollen counts for Mid Blue core p1. 139 Figure 7.13 Biogenic silica counts for Mid Blue core p1. 140 Figure 7.14 Mid Blue Lake core p1: sediment diagram. 141 Figure 7.15 Infrared spectrometry data (SWIR) for Mid Blue Lake core p1. 142 Figure 7.16 Sediment description for Mid Blue clay lunette auger hole. 143

xiii Figure 7.17 Pollen counts for Mid Blue Lake inner clay lunette auger hole. 145 Figure 7.18 Biogenic silica counts for Mid Blue Lake inner clay lunette auger hole. 146 Figure 7.19 Mid Blue Lake inner clay lunette auger hole: sediment diagram. 147 Figure 7.20 Infrared spectrometry data (SWIR) for Mid Blue Lake inner clay lunette auger hole. 148 Figure 7.21 Sediment description for Mid Blue swale between the lunettes (auger hole). Note: the photo’s scale is severely distorted by perspective. 149 Figure 7.22 Gypsum crystals from Mid Blue Lake swale between the clay and the gypsum lunettes: 1a & b – complex pyramidal to lenticular and simple lenticular selenite at 125cm depth; 2a & b – lenticular twin selenite crystals at 175cm; 3a & b - amorphous selenite at 250cm. 150 Figure 7.23 Pollen counts for Mid Blue Lake swale between the clay and gypsum lunettes (auger hole). 151 Figure 7.24 Biogenic silica counts for Mid Blue Lake swale between the clay and gypsum lunettes (auger hole). 152 Figure 7.25 Mid Blue Lake swale between the clay and gypsum lunettes (auger hole): sediment diagram. 153 Figure 7.26 Infrared spectrometry data (SWIR) for Mid Blue Lake swale between the clay and gypsum lunettes (auger hole). 154 Figure 7.27 Sediment description for Mid Blue Lake gypsum lunette auger hole. Note: the photo’s scale is severely distorted by perspective. 155 Figure 7.28 Pollen counts for Mid Blue Lake outer gypsum lunette auger hole. 156 Figure 7.29 Biogenic silica counts for Mid Blue Lake outer gypsum lunette auger hole. 157 Figure 7.30 Mid Blue Lakes outer gypsum lunette auger hole: sediment diagram. 158 Figure 7.31 Infrared spectrometry data (SWIR) for Mid Blue Lake outer gypsum lunette auger hole. 159 Figure 7.32 Map of Lake Wombah. (Main image source: Brindingabba airphoto 54, 2002; Inset image source: Natmap Raster 2003; Source of geomorphological information: B. Timms, pers. comm. 2007) 164 Figure 7.33 The top of the rocky cliff with a thin layer of red sandy loam (foreground) on the northern margin of Lake Wombah with the beach and drying lake in the background. 164 Figure 7.34 Sediment description for Lake Wombah core 3. 166 Figure 7.35 Pollen counts for Lake Wombah core 3. 167 Figure 7.36 Biogenic silica counts for Lake Wombah core 3. 168 Figure 7.37 Lake Wombah core 3: sediment diagram. 170 Figure 7.38 Infrared spectrometry data (SWIR) for Lake Wombah core 3. 171 Figure 8.1 The location of the Cummeroo Waterhole and cores 3 and 4. (Main image source: Yantabulla airphoto 61, 2003; Inset image source: Google Earth, 2006) 176 Figure 8.2 Cummeroo Waterhole lined with black box and bimble box. 176 Figure 8.3 Dry channel southeast of the Cummeroo Waterhole. 177 Figure 8.4 Sediment description for Cummeroo Waterhole core 4. 178 Figure 8.5 Pollen counts for Cummeroo Waterhole core 4. 179 Figure 8.6 Biogenic silica counts Cummeroo Waterhole core 4. 180 Figure 8.7 Cummeroo Waterhole core 4: sediment diagram. 182 Figure 8.8 Infrared spectrometry data (SWIR) for Cummeroo Waterhole core 4. 183 Figure 9.1 Map of Lower Bell Lake and Palaeolake. (Main image source: Tinchelooka airphoto 339, 1997; Inset image source: Google Earth, 2006; Lower Bell Lake geomorphological information after Timms, 2006) 188 Figure 9.2 Red sand dune northwest of Lower Bell Lake with Callitris glaucophylla in the foreground (right) and mulga in the swale behind the dune. 189

xiv Figure 9.3 Sediment description of Lower Bell Lake core G. 190 Figure 9.4 Lower Bell Lake gypsum: 1 – gypsite band at 6.5cm depth; 2 – gypsite band at 122.3cm; 3a & b – pyramidal to lenticular selenite and gypsarenite with minor signs of dissolution at 83cm; 4a & b – mostly pyramidal to lenticular selenite and gypsarenite at 109cm. 191 Figure 9.5 Lower Bell Lake gypsum: 1a & b – pyramidal to lenticular (single and twin) selenite at 128-131cm depth; 2a & b – pyramidal to semi-prismatic selenite at 132-136cm; 3a & b – lenticular (single and twin) selenite at 154-160cm. 192 Figure 9.6 Selenite in 129-135cm depth section of the Lower Bell Lake core G. 193 Figure 9.7 Pollen counts for Lower Bell Lake core G. 195 Figure 9.8 Biogenic silica counts for Lower Bell Lake core G. 196 Figure 9.9 Lower Bell Lake core G: sediment diagram. 197 Figure 9.10 Infrared spectrometry data (SWIR) for Lower Bell core G. 199 Figure 9.11 Transect across Palaeolake including relative positions of the cores analysed in this study. 204 Figure 9.12 Palaeolake: eastern gypsum lunette with a stand of Casuarina cristata. The inflow channel from Bloodwood Freshwater Lake is located to the left of the picture. Halosarcia spp. in the foreground. 205 Figure 9.13 Red sand dune on the northern shore of the Palaeolake (background). Halosarcia spp. on the beach and partly submerged (foreground). 205 Figure 9.14 Sediment description for Palaeolake core B. Some discoloration of the sediment was caused by light distortion during photographing. 206 Figure 9.15 Palaeolake core B gypsum crystals at 170cm depth: a, b & c – lenticular to pyramidal selenite (single and twin). 207 Figure 9.16 Pollen counts for Palaeolake core B. 209 Figure 9.17 Biogenic silica counts for Palaeolake core B. 210 Figure 9.18 Palaeolake core B: sediment diagram. 211 Figure 9.19 Infrared spectrometry data (SWIR) for Palaeolake core B. 212 Figure 9.20 Sediment description of Palaeolake core K. 214 Figure 9.21 Palaeolake core K gypsum crystals: 1 – rosette of gypsarenite and small selenite at 62cm depth; 2 – irregular (semi-lenticular) gypsum at 81cm depth; 3a & b – mostly lenticular to pyramidal selenite (single to complex) at 95cm; 4a & b – mostly single lenticular to pyramidal selenite and gypsarenite (rarely forming aggregates) at 108cm with minor signs of dissolution along the margins. 215 Figure 9.22 Palaeolake core K gypsum crystals: 1 - selenite at 111cm depth: a – prismatic/pyramidal (twin); b – undefined habit; 2a & b – pyramidal to lenticular selenite and gypsarenite at 122cm; 3a & b – pyramidal to lenticular twin selenite at 127cm. 216 Figure 9.23 Palaeolake core K gypsum crystals: 1a & b – prismatic/pyramidal (twinned) selenite at 134cm depth; 2a & b - mostly lenticular selenite and gypsarenite at 201cm with signs of dissolution (rugged margins) and overgrowth; 3a & b – mostly lenticular selenite at 205cm with signs of secondary precipitation (overgrowth) and dissolution, particularly around the margins. 217 Figure 9.24 Pollen counts for Palaeolake core K. 219 Figure 9.25 Biogenic silica counts for Palaeolake core K. 220 Figure 9.26 Palaeolake core K: sediment diagram. 222 Figure 9.27 Infrared spectrometry data (SWIR) for Palaeolake core K. 223 Figure 9.28 Sediment description for Palaeolake north red sand dune auger hole. 224 Figure 9.29 Palaeolake northern red sand dune gypsum: 1 – gypsite nodule at 320cm; 2a & b – gypsarenite/selenite conglomerate at 340cm; 2c – close up of 2b image. 225 Figure 9.30 Pollen counts for Palaeolake northern red sand dune auger hole. 226 xv Figure 9.31 Biogenic silica counts for Palaeolake northern red sand dune auger hole. 227 Figure 9.32 Palaeolake northern red sand dune auger hole: sediment diagram. 229 Figure 9.33 Infrared spectrometry data (SWIR) for Palaeolake northern red sand dune auger hole. 230 Figure 9.34 Sediment description for Palaeolake eastern gypsum lunette auger hole. 231 Figure 9.35 Palaeolake eastern gypsum dune: 1 – gypsite nodule with sand inclusions within 85- 125cm depth section; 1b – close up of 1a image; 1c – mixture of lenticular gypsarenite and sand at 85-125cm depth; 2 – unidentified crystal balls at 380cm. 232 Figure 9.36 Pollen counts for Palaeolake eastern gypsum lunette auger hole. 233 Figure 9.37 Biogenic silica counts for Palaeolake eastern gypsum lunette auger hole. 234 Figure 9.38 Palaeolake eastern gypsum lunette dune auger hole: sediment diagram. 235 Figure 9.39 Infrared spectrometry data (SWIR) for Palaeolake eastern gypsum lunette dune auger hole. 236 Figure 10.1 Map for Lakes Willeroo and Yandaroo. (Main image source: Toorale SH 550902 airphoto 205, 1997; Inset image source: Google Earth, 2006; geomorphological information after Timms (pers. comm.. 2007) 242 Figure 10.2 Lake Willeroo: scattered rocks on the southern shore. Dead tree trunks scattered over the lake floor in the background. 243 Figure 10.3 Lake Willeroo: dead tree trunk (foreground) and multiple tree stumps (background). 244 Figure 10.4 Sediment description for Lake Willeroo augered sediments. 245 Figure 10.5 Pollen counts for Lake Willeroo auger hole. 246 Figure 10.6 Biogenic silica counts for Lake Willeroo auger hole. 247 Figure 10.7 Lake Willeroo auger hole: sediment diagram. 249 Figure 10.8 Infrared spectrometry data (SWIR) for Lake Willeroo auger hole. 250 Figure 10.9 Dense sedges on the southwestern margin of Lake Yandaroo with the tree and shrub band on the right side of the photo. 253 Figure 10.10 View of Lake Yandaroo from its northeastern shore. Sparse woody shrubs scattered in the foreground. A band of trees and shrubs surrounding the lake is in the background. 254 Figure 10.11 Sediment description for Lake Yandaroo auger hole. The 137Caesium is in mBq/g units as the calculation of mBq/cm2 was impossible due to lack of adequate information on total sample size. 255 Figure 10.12 Pollen counts for Lake Yandaroo auger hole. 257 Figure 10.13 Biogenic silica counts for Lake Yandaroo auger hole. 258 Figure 10.14 Lake Yandaroo auger hole: sediment diagram. 259 Figure 10.15 Infrared spectrometry data (SWIR) for Lake Yandaroo auger hole. 260 Figure 10.16 Sediment description for Lake Yandaroo cores a, b, & d. 261 Figure 10.17 Pollen counts for Lake Yandaroo cores a, b, & d. 262 Figure 10.18 Biogenic silica counts for Lake Yandaroo cores a, b, & d. 263 Figure 10.19 Lake Yandaroo cores a, b, & d: sediment diagram. 265 Figure 10.20 Infrared spectrometry data (SWIR) for Lake Yandaroo cores a, b, & d. 266 Figure 11.1 Sedimentary summary of the lake cores from the Paroo/Warrego Region. The rainfall values are 5 year running means for an average of Thargomindah and Bourke annual precipitation. 271 Figure 11.2 Linear dunes northwest of Lower Bell Lake. (Image source: Brindingabba airphoto 372, 1997) 274

xvi Abstract

The records of environmental change in Australia’s arid zone can be greatly enriched by employing a multi-proxy approach and landscape-scale analysis. This research uses these tools to construct a palaeoenvironmental history of the Paroo/Warrego Region. While the Region’s flow regimes and water balances are characterised by medium-term (decadal) variability (Young, 1999), its hydrological records are inadequately brief. Subsequently, land and water management decisions are based on short term data, risking irreversible damage, desertification or loss of diversity. A better understanding of this highly dynamic landscape can thus improve the land and resource management outcomes.

While dating was constrained by a lack of funds, the Paroo/Warrego history reconstructed from fluvial and aeolian deposits correlated well with events recorded from other inland regions of the Australian continent. In summary, this new research provided evidence of high lake water levels prior to the Last Glacial. The extreme aridity at the onset of Last Glacial caused long term drying of the lakes and mobilisation of the red sand dunes. In latter stages of the glacial phase the aridity gave way to periodic fluctuations between flood and drought events that probably lasted until 16 000 - 14 000 BP. The new climatic regime resulted in formation of gypsum lunettes and later, following reduction in gypsum supply, clay lunettes. The orientation of red sand dunes and lunettes indicates a more northerly extent of the westerlies than in modern times.

Around the late Pleistocene-early Holocene boundary the climate became more stable and wetter, but still somewhat drier than during the pre-Last Glacial lacustrine phase. As a result, the region’s lakes reverted to a permanent and semi-permanent status. A strong aridity signal, comparable to the semi-regular droughts of the Last Glacial, was recorded in the Paroo/Warrego lakes during the late 1890s-1940s period of below average rainfall. It was followed by 50 years of wetter conditions with two extremely wet phases in the 1950s and the 1970s. Finally, the most recent records suggest a new drying trend.

The semi-arid vegetation appears to have adapted to climate variability, with herbs and grasses expanding with the onset of wet conditions before being replaced by Chenopodiaceae as the landscape started to dry. The fresher lake basins and water courses were likely to provide refuge during prolonged arid phases and dispersal foci during intervening wetter periods, thus enabling greater flexibility in response to changes and enhancing resilience. The European land use interfered with the natural cycles and balances, leading to decrease in ground cover, suppression of fire, increase in runoff and catchment erosion, acceleration of sediment

xvii accumulation rates in wetlands, resulting in decline of their water holding capacity, and expansion of woody vegetation.

The research improved the processing protocols, reference databases, and transfer of methods to enable greater sample processing efficiency and improve results. The use of multiple proxies (including biotic and abiotic components) and sites, as well as different depositional features, provided access to a broader picture of environmental change than was previously possible. It also facilitated multi-scale resolution, allowing discrimination between localised responses of individual lakes and regional trends. The full value of this research will come from informing natural resource managers, whose actions will shape the future landscapes of the Paroo and Warrego Region.

xviii Chapter 1

Introduction

“When the lakes of arid regions become extinct, either by reason of evaporation or sedimentation, evidence of their former existence remains inscribed on the inner slopes of their basins or concealed in the strata deposited over their bottoms. These records as a rule are much more lasting than those left by lakes in humid lands…”

Israel C. Russell, 1895 (cited in Currey, 1994)

1.1 Palaeoenvironmental research in arid and semi-arid inland lakes

The palaeoenvironmental history of landscapes and ecosystems is an essential tool in land management for conservation and sustainable development. It can provide information about past changes and cyclicity of climatic and weather patterns that is particularly important in countries like Australia, which are characterised by strong interannual climate variation (Dodson, 1998; Kukla and Gavin, 2004; McKeon et al., 1998). Palaeoenvironmental research can also help to determine the resilience of ecosystems to environmental stresses as well as identify coping-strategies and ‘no return’ points that result in irreversible changes.

In Australia, the palaeoenvironmental research is concentrated in the southeastern corner of the continent, the humid margins of the eastern coast, and Tasmania (e.g. Donders et al., 2007) as those regions offer often plentiful and continuous pollen records, abundance of datable organic material, and locations close to the research centres. In contrast, the vast areas of the continental interior remain largely unstudied, in spite of their great importance as a driving force of the Australian weather and climate patterns. They influence, for example, the strength and northern extent of the westerlies (Bowler, 1978) and regulate the frequencies and loads of dust storms (e.g. Hesse and McTainsh, 2003). In addition, the unique and variable ecosystems of the Australian arid and semi-arid zone, and particularly its wetlands, are recognised as assets of national and international significance, but little is known about their responses to climatic fluctuations.

The importance of the inland drylands creates a great need for better understanding of their history. Their study is, however, often perceived by researchers as risky and with limited

1 promise of satisfactory outcomes due to the variable rates of deposition, unknown deflation history driven by sporadic wind erosion, and scarcity of information about fossil preservation under the dry conditions (e.g. Torgersen et al., 1986).

1.1.1 The beginnings of the arid and semi-arid palaeoenvironmental studies in Australia: large lake research

In spite of the challenges, interest in the environmental information hidden in the arid and semi- arid lakes and dune systems in Australia has a relatively long history. It was to a large degree initiated by a study in the Willandra Lakes in the 1970s (Bowler, 1973, 1980, 1983, 1990). The need for palaeoenvironmental records from inland regions of the continent was further recognised and promoted by setting of a Salt Lakes, Evaporites and Aeolian DepositS (SLEADS) program (Chivas and Bowler, 1986). Within SLEADS’ framework, large multi- disciplinary teams sampled and analysed sediments from a range of large ‘depositional systems’ across Australia; including Lake Buchanan (Chivas et al., 1986), Lake Woods (Hutton et al., 1984), Lake Lewis (Napperby) (Arakel, 1986; English, 2001), Lake Amadeus (Chen et al., 1993; Chen et al., 1991), Lake Eyre (Gillespie et al., 1991; Magee et al., 1995; Magee and Miller, 1998; Magee et al., 2004), Lake Callabonna (Nanson et al., 1998), Lake Frome (Draper and Jensen, 1976; Nanson et al., 1998; Singh and Luly, 1991; Ullman and McLeod, 1986), Lake Tyrrell (Bowler and Teller, 1986; Luly, 1993; Luly et al., 1986; Stone, 2006), and Lake George (De Deckker, 1982b; Singh and Geissler, 1985). The SLEADS project and other studies in drylands were mostly limited to large and mostly permanently waterlogged lake basins that were assumed to represent the most stable sites with consistent deposition history (i.e. to be the least affected by deflation) and to present the best potential for fossil preservation.

1.1.2 Benefits of records from smaller lakes

There are, however, several advantages in studying the undervalued smaller dryland lakes with the most important being their greater sensitivity to changes. While large lakes (several square kilometres in extent) are valuable recorders of major long-term events (Williams et al., 1993), their size buffers them from minor, brief and/or subtle, changes that might pass unrecorded. Furthermore, the larger and deeper the lake, the greater is the lag between the onset of the climatic/environmental change and its impact on the lake’s hydrology, geomorphology, and sediment characteristics.

The smaller lakes, on the other hand, are much more likely to respond to even minor adjustments in the water balance driven by changes in the climatic conditions and catchment characteristics, thus facilitating finer resolution of their palaeoenvironmental records (Kotwicki and Allan, 1998; Williams et al., 1993). In addition, the much more rapid readjustment of the

2 lake basin hydrology and sediment content to the new conditions allows for better dating of the events.

1.1.3 Benefits of a holistic, multi-proxy, multi-site approach

A further refinement of the palaeoenvironmental record from a arid/semi-arid site and expanding it to a landscape-scale can be achieved by employing a novel more holistic, multi- proxy, multi-site approach. The ongoing modification of dryland depositional landscapes, including lake systems, is driven by oscillations between aeolian and fluvial processes (Bullard and McTainsh, 2003). One prominent example of this is the formation of the lake basin-source bordering lunette systems (described by Bowler, 1973), where the lake basin provides a primary storage facility for fluvial (and less frequently aeolian) sediments and the lunette stores the sediments deflated from the lake during a dry phase. As a corollary, to achieve the most complete resolution of the record, sediments from both lake and lunette systems have to be sampled, analysed, and collated.

Furthermore, the alternating cycles of fluvial and aeolian processes, often accompanied by oscillations between freshwater and saline conditions, can result in formation, deposition, and preservation of different proxies. A single-proxy record can be fragmented as the proxy’s occurrence might be limited to a narrow range of conditions. The use of multiple proxies can fill in the gaps present in a single-proxy data set and enrich the record. The complementary use of multiple proxies can be also useful in narrowing down choices in alternative interpretations that are often associated with single-proxy records (Digerfeldt, 1986).

Finally, depending on their size, depth, water chemistry (including salinity levels), the size of the catchment and location within a larger drainage basin, the individual lakes within a single region can react differently to a single environmental event (Macumber, 1991). In addition, the size of the lake can affect the records’ composition, for example, by trapping different proportions of local and extra-local pollen (Holmes, 1994). Thus, expanding a study to include a variety of lake types can enrich the regional scale record as well as provide ways to differentiate between changes occurring within the study area (registered by the lakes with mainly local catchments) and those driven by processes operating several hundreds of kilometres away, for example, in the headwater of the region’s main rivers (affecting lakes with some degree of connectivity with the river system).

1.2 The Paroo/Warrego Region: an attractive study area

The Paroo/Warrego Region is an attractive site for palaeoenvironmental studies due to its unmodified landscape and hydrology, unique location in relation to the major climatic zones and systems, the high diversity of its extensive wetland systems, and the wealth of recent biological and hydrological information. A Paroo/Warrego record also helps to fill a major spatial void in

3 the existing distribution of Quaternary study sites occurring between the coastal areas and the Lake Eyre Basin, and the tropical North and the southeast Australian regions.

Located on the dry northwestern margin of the Murray-Darling Basin, the Paroo and Warrego River catchments supported mainly grazing since the European settlement in 1840s. Its remoteness and harsh climate, including high variability in river flows, protected it from irrigated agriculture and the associated damming and massive water extraction that drastically modified and degraded the landscapes of the remainder of the Basin. Thus the region was able to remain relatively underdeveloped (Kingsford, 1999; Pearson et al., 2003). The low relative relief of the landscape with the low hydraulic gradient results in a sensitive hydrological balance that can quickly react to any changes in water budgets (Macumber, 1991).

Climatically, the region is located within the semi-arid zone on the eastern fringe of the arid interior (Figure 1.1). This makes it a gauge of arid zone expansion and contraction, as sequences of arid/aeolian and fluvio-lacustrine features become preserved within landscape (Bowler, 1990; Williams et al., 1991). The region’s position between the westerly-driven southern westerlies with winter rainfall regime and northern monsoon with summer rainfall (Figure 1.1) provides valuable information about shifts in those two climatic systems. Finally, the proximity to areas with the highest sensitivity of the annual rainfall to the Southern Oscillation Index (SOI) (Figure 1.1) (McBride and Nicholls, 1983; Timms, 2006) can be useful in reconstructions of the past El Niño Southern Oscillation (ENSO) behaviour.

Figure 1.1 Location of the Paroo/Warrego Region with respect to the main climatic systems. (Source: Bowler et al., 2001; Hesse et al., 2004; McBride and Nicholls, 1983) 4 The Paroo/Warrego Region is also unique because of the high concentration and diversity of its wetlands that include saline and freshwater lakes, clay pans, and riverine waterholes (Timms, 1997b). This mosaic of highly productive wetlands supports large populations of aquatic plants, invertebrates, and birds (Kingsford and Porter, 1999; Pearson et al., 2003; Timms, 1993, 1997a, 1997b, 1998a, 2006, 2007, in press) that are likely to leave traces of their existence within the sediments. Many of the lakes have also long-term modern limnological records and geomorphological descriptions stemming from Timms’s nearly 20 year long research program in the region.

1.3 The research objectives

This study was designed to explore and exploit the high resolution records within small lakes in the Paroo/Warrego Region with the aid of a holistic, multi-proxy, multi-site approach. The findings extend and expand the Region’s palaeoenvironmental history beyond the memories and limited documented records of land managers. They contribute to gaining a better understanding of the deeper changes shaping the Australian semi-arid landscapes.

More specifically, the research aimed to address the following objectives: i) to review the performance of some of the standard methodologies and common proxies employed in palaeoenvironmental studies in addressing the challenges posed by the unconventional depositional systems within small lacustrine basins in arid and semi- arid regions; ii) to provide insight into the differences in occurrence and preservation of different proxies within a range of sedimentary environments, including freshwater versus saline and (semi-)permanent versus frequently dry lakes, as well as lacustrine versus lunette deposits; iii) to reconstruct the palaeoenvironmental histories of ten distinctively different lakes within the Paroo/Warrego Region; iv) to collate the individual stories into a single, landscape/regional-scale record of environmental change.

1.4 Content overview

An outline of the thesis content is presented in Figure 1.2. Following this general introduction, Chapter 2 provides more detailed information about the Paroo/Warrego Region including its physical and living landscape as well as some of the complex drivers shaping it in the past and in the present. Chapter 3 contains the descriptions of the methods that were used in this research project.

Figure 1.2 The outline of the thesis content.

6 Chapter 4 contains discussion of proxies and methodologies, providing an interpretative background to the palaeoenvironmental reconstructions for individual lake systems presented in Chapters 5-10. It is intended to reduce the need for repetition of issues shared by the subsequent chapters and provide a landscape-scale perspective on the distribution of selected proxies. The chapter introduces the proxies employed by this project by describing their interpretative value as proven by other researchers. It provides brief assessment of the proxies’ performance within the scope of this study and suggests areas where improvements are still needed to optimise their use and contribution in the future. The chapter also discusses the advantages and shortcomings of selected methods in extraction of the proxy information and interpretational adequacy of their accuracy.

Chapters 5-10 present self-contained descriptions of sites, results, and palaeoenvironmental interpretations for individual lakes that are grouped into wetland systems. Chapter 11 collates the separate stories into a single landscape/regional-scale history and relates it to other records of environmental changes, with particular attention to those originating from inland eastern Australia. Finally, Chapter 12 provides a summary of the main findings, their significance to future arid and semi-arid studies of the highly dynamic environments, and implications to future management of those fragile ecosystems.

Chapter 2

The Paroo/Warrego Region: Background information

2.1 Introduction

This chapter introduces the Paroo/Warrego Region, setting the scene for the later interpretation of the proxy data. The initial section describes the location of the study sites within the catchment area. The following sections provide modern and historical insights into the major factors, including climate, geology and soils, hydrology, vegetation, fire, and land management practices, that play a key role in shaping the region’s landforms by controlling sediment stability and movement.

2.2 Location of the study area

The combined catchment of the Paroo and Warrego Rivers constitutes the most north-westerly part of the Murray-Darling Basin (Figure 2.1) and supports a diversity of wetland systems. In this research, ten lakes from six different wetland groups (e.g. Timms, 1993, 1997a, 1997b, 2006, in press) were selected for detailed sediment analysis (Figure 2.1): i) Lake Bindegolly, ii) Currawinya Lakes: Lake Numalla and Lake Wyara, iii) Blue-Wombah Lakes: Mid Blue Lake and Lake Wombah, iv) Cummeroo Waterhole, v) Bloodwood Station Lakes: Lower Bell Lake and ‘Palaeolake’ (informal name), vi) Delta-Yandaroo Station Lakes: Lake Yandaroo and Lake Willeroo.

The rationale for selecting these particular lakes is provided in Chapter 3 section 3.2, while detailed descriptions of the individual lakes (sites) are provided in relevant sections of Chapters 5-10.

Figure 2.1 The catchment boundaries of the Paroo and Warrego rivers and the location of the studied lakes. (Image source: Natmap Raster 2003)

2.3 Climate

2.3.1 General outline

The studied wetlands lie within the subtropical high pressure belt in the hot grassland (persistently dry) climate zone, close to the eastern boundary of the hot desert (persistently dry) climate zone (Stern et al., 2000). The region is surrounded by vast dry continental plains, preventing frequent penetration of the rain-bearing oceanic air masses as well as any rainfall caused by orographic lifting (Gentilli, 1972). This results in relatively low average annual precipitation with high variability of monthly rainfall, particularly during the summer months, driven by the occurrence of large, sporadic storms (Figures 2.2 and 2.3). Thus, the rainfall events that produce significant runoff are irregular and their effectiveness is often severely limited by the high potential evaporation rates that exceed the average rainfall throughout the whole year (Figures 2.2 and 2.3).

The low rainfall seasonality arises from the location of the study sites within the transition zone between the monsoon driven summer rainfall system of northern Australia and the mid-latitude westerly winter rainfall regimes of southern Australia. A large part of the rainfall’s variability can be attributed to its high correlation to the ENSO (Donders et al., 2007; McBride and Nicholls, 1983), with a rainfall-SOI correlation of r = 0.31 (p<0.001, n = 125) for Thargomindah and r = 0.30 (p = 0.001, n = 117) for Bourke. Temperatures are more predictable, with hot summers and mild to cool winters (Figures 2.2 and 2.3).

The winds in the northwesterly end of the study area (at Thargomindah) are dominated throughout the year by southeasterlies with an increased southwesterly component in spring (Figure 2.2). The wind directions over the southeasterly end of the study area (at Bourke) are more variable with dominant northeasterly to southeasterly winds in summer switching to southeasterlies in autumn (Figure 2.3). A southwesterly component prevails during winter and spring months. The wind speeds stay mostly below 15km/hr throughout the year and display little variability between years (Figures 2.2 and 2.3). While not large, there is a slight seasonal variation in the average wind strength with the highest values in spring and lowest in winter.

2.3.2 Historical rainfall patterns

The rainfall record for the Paroo/Warrego Region extends over nearly 130 years. The annual precipitation can vary substantially from year to year (Figure 2.4). Longer term trends become clear on a plot of moving averages showing wetter and drier periods lasting from few years to a few decades (Figure 2.5). For example, the years between 1896-1947 have been generally drier than the preceding and following decade. Another dry period throughout the 1960s was followed by nearly three decades of medium to higher rainfalls until a sharp drop in 2002. There is a reasonable correlation (r = 0.68, p<0.001) between the annual rainfall values of

Bourke and Thargomindah, with even better agreement shown by the five year moving average (r = 0.76, p<0.001).

Figure 2.2 Climate data for Thargomindah. Rainfall is based on observations recorded between 1879 and 2005, evaporation and temperature: 1879-2004, and wind rose and wind speed: 1957-1999. (Source: Bureau of Meteorology, 2006)

Figure 2.3 Climate data for Bourke. Rainfall, evaporation, and temperature are based on observations recorded between 1871 and 1996, wind rose: 1881-1996, and wind speed: 1908-1994. (Source: Bureau of Meteorology, 2006)

1000 Thargomindah Bourke 900 average annual rainfall for Thargomindah average annual rainfall for Bourke

800

700

600

500

400

300 annual rainfall (mm) rainfall annual

200

10 0

0 1875 1885 1895 1905 1915 1925 1935 1945 1955 1965 1975 1985 1995 2005

Figure 2.4 Short term rainfall variability for Thargomindah and Bourke based on annual precipitation. (Source: Bureau of Meteorology, 2006)

600 Thargomindah Bourke 550 average annual rainfall for Thargomindah average annual rainfall for Bourke

500

450

400

350

300

250

200

15 0

annual rainfall (mm):moving 5 year average 10 0 1875 1885 1895 1905 1915 1925 1935 1945 1955 1965 1975 1985 1995 2005

Figure 2.5 Longer term rainfall patterns for Thargomindah and Bourke based on 5 year moving average of annual precipitation. (Source: Bureau of Meteorology, 2006)

2.4 Geology and soils

The study area (i.e. the central part of the Paroo and Warrego Rivers catchment) is underlain by horizontal Cretaceous sandstones, siltstones, and claystones that were deposited within an inland sea about 100 Ma years ago (Sahukar et al., 2003; Eulo, Yantabulla, and Louth 1:250 000 Geological Series Map Sheets). These sediments are the main water-storing strata of the Great Australian Basin and the source of artesian mound springs, which are common, but mostly inactive, in this region (Sahukar et al., 2003). The exposure of the gently folded Cretaceous sediments has resulted in formation of low hills and plateaus, e.g. southeast of Bloodwood lakes, southwest of Lake Wombah, and west of Lake Wyara (Morgan and Terrey, 1992; Sahukar et al., 2003). Some of the surface sandstones and sands were silicified during the Tertiary to form silcrete (a siliceous duricrust) that forms cliffs and extensive cappings on the plateaus and dissected tablelands, or occurs as desert pavement (gibber) (Goodrick, 1984; 14

Sahukar et al., 2003). The soils that developed over those landscapes are dominated by shallow stony red-brown loams and lithosols, while sandy red earths often form on bottom slopes (Morgan and Terrey, 1992; Sahukar et al., 2003).

The lower areas of the landscape are frequently covered by unconsolidated Quaternary sediments of three major types: black soil and claypan type, soft red type, and hard red type (Goodrick, 1984; James, 1960a). The black soil and claypan type can be found mainly in areas that are subject to relatively frequent flooding and can retain water for extended periods of time, such as floodplains, overflow channels and swamps, playas, and claypans (Goodrick, 1984; Holdaway et al., 2000; James, 1959). The sediment consists usually of grey and brown deep cracking heavy clays or sandy red earths (Goodrick, 1984; James, 1960a; Sahukar et al., 2003).

The alluvial deposits are often bordered by the aeolian sediments of the soft red type that form gently undulating dunefields and sandplains. These deposits are composed mostly of infertile red sands and silts derived from deep weathering of the Cretaceous sandstones and reworking by wind (Goodrick, 1984; James, 1959; Morgan and Terrey, 1992; Sahukar et al., 2003). The dunes are longitudinal and roughly aligned with the southwesterly winds (Goodrick, 1984). In the past some of the dunes have occluded distributary channels (e.g. Lower Bell Lake), but are mostly stabilised at present (Goodrick, 1984; James, 1959, 1960a).

The hard red sediment type describes widespread pavements on talus slopes and gibber plains consisting of colluvial and residual deposits derived mainly from erosion of the silcrete (Goodrick, 1984; James, 1960a). It mostly supports compact, stony to gravely, structureless (but generally deep) red sandy loams and clay loams (James, 1960a).

Lime occurs frequently in most of the soils in the region and is either well mixed with the fine earth or concentrated into hard nodules (James, 1959). Except for the shallow brown clays, all the red and brown clay soils of the sandplains, and about half of the clays on alluvial plains, contain gypsum; its content increasing with depth (Rayment et al., 1983). Large gypsum crystals are common at depth within saline lake sediments (Goodrick, 1984).

2.5 Geomorphology and hydrology

The combined catchments of the Paroo and Warrego Rivers are dominated by low lying (mostly at 75-200m above sea level) flat to slightly undulating sandplains intersected by a network of channels and floodplains as well as low stony ranges.

2.5.1 The Paroo and Warrego Rivers

The Warrego River has its headwaters in Queensland within the Great Dividing Range and the Paroo River starts slightly further south, at the foot of the Range (Figure 2.1). The Paroo River is about 600km long draining a catchment of 73600km2 and the Warrego River flows along a

900km route with a catchment of 69100km2 (Kingsford and Porter, 1999). These large catchments encompass smaller ones, which rely almost solely on the local rainfall (e.g. the Bloodwood Lakes) (Kingsford and Porter, 1999).

In the Paroo River the flows and floods are still unregulated as there are no dams and only very limited water extraction occurs along its course (Alliance Resource Economics, 2002; Timms, 1997b, 1998b). The Warrego River is more affected by water diversion with licensed pumps, surface water storages (including two dams), and irrigation licences (Bureau of Rural Sciences, 2006; Kingsford, 2000). These two rivers are the least regulated within the large Murray- Darling Basin.

The flows of the Paroo and Warrego Rivers are highly variable and unpredictable in terms of timing, duration, and extent, with the Paroo classified as the sixth most variable river in Africa and Australia (Kingsford, 1999; Kingsford et al., 1998; Timms, 1997b; Young, 1999). The high variability of the flow is closely related to the irregularity of rainfall in the headwater area, driven by summer cyclonic depressions (Morgan and Terrey, 1992). In the lower reaches of the Paroo and Warrego rivers, those flows are often called ‘dry floods’ as they move through areas that have not experienced any rainfall (Timms, 2006). Although smaller, the contribution of local rain events in the lower, semi-arid part of the combined catchments is important to the river flows and can even cause flooding (Kingsford, 1999; Young et al., 2006). The Paroo and Warrego rivers generally experience a few minor floods each year and a major one about every five years (Kingsford and Porter, 1999). The floods are, however, neither annually nor even seasonally predictable (Young, 1999).

During low water levels the rivers are isolated but major floods can activate channels that join the Warrego to the Paroo River, mostly via Cuttaburra Creek, and to the Culgoa River (Morgan and Terrey, 1992). During extreme flood events, the Paroo and Warrego can jointly inundate over 800 000 ha (Kingsford et al., 1998). Aside from the flood magnitude, the spatial extent of flooding can be affected by the wetness of the system (particularly the water levels in the associated lakes) and the condition of the floodplain vegetation determining its resistance to flow (Young, 1999). During no flow periods, the rivers exist as a series of highly turbid waterholes (Bunn and Davies, 1999).

In general terms, both Paroo and Warrego Rivers drain into the Darling River, however, in reality their contribution to the Darling’s flow is limited. Their flows tend to dissipate on the southward journey through anastomosing main channels and an extensive network of distributary channels that frequently terminate in swamps and playa systems, such as the Currawinya Wetlands, Lake Wombah, or Lake Peery (Figure 2.1) (Goodrick, 1984; Kingsford and Porter, 1999; Pickard, 1997). Consequently, the Paroo River, for example, flows into the Darling River only in exceptionally large floods (Goodrick, 1984) that occur only a few times 16

per century (Pickard, 1997). More frequently, the Darling River floods north, into the Paroo River (Pickard, 1997; Young, 1999). It was also observed that wind seiches can reverse the river’s flow, pushing the riverine water many kilometres upstream (Pickard, 1997).

2.5.2 Wetlands

The Paroo/Warrego wetlands are highly diverse in terms of vegetation, geomorphology, and hydrology (Kingsford and Porter, 1999). In very broad terms, they can be classified into the following types: saline lakes, small off-river freshwater lakes, large terminal lakes fed by rivers, riverine channels and waterholes, lignum swamps, clay pans, and others (Timms, 1998b). Some of the wetlands may be a combination of two or more of these types.

The high variability of the wetlands manifests itself also in the duration of inundation, ranging from almost permanent to rare and short lived (Timms, 1993). Generally, the length of inundation period is determined by the net inflow (i.e. the total inflow from local rainfall and runoff, and ‘dry’ floods minus the outflow), season (due to variable evaporation rates; Figures 2.2 and 2.3), and lake basin geomorphology (e.g. depth) (Goodrick, 1984; Timms, 1993, 2006). So far little is known about the relationships between wetland filling and flood magnitude, e.g. how big a flood is needed to fill a particular wetland (Timms, 1997b).

While the lake filling regimes are fairly irregular, Timms (2006) reported a strong correlation between the lake filling and El Niño Southern Oscillation (r = 0.62 p>0.001 for Lake Wyara). The larger stream-fed lakes (e.g. Lakes Wyara and Numalla) fill during El Niño years and slowly evaporate in subsequent drier years. Smaller lakes, without significant surface inflows, can fill a few times during wet periods, but dry soon after the rainfall and most remain dry during the La Nina period.

Most of the Paroo/Warrego lakes are of deflationary origins and are usually very shallow and of irregular shape (Timms, 1992). In some cases, however, the lake formation can be contributed to tectonic movements (e.g. Lakes Bindegolly, Wyara, and Wombah) and damming of palaeodrainage channels and valleys (e.g. Lakes Numalla and Lower Bell) with only subsequent modification by deflation (Timms, 1992, 1993). The riverine waterholes, on the other hand, are formed primarily by evorsion (Timms, 1997b).

The Paroo/Warrego wetlands also vary greatly with respect to their salinity (ranging from fresh to crystallising brines) and pH (Timms, 1993). Furthermore, most of the wetlands, including the freshwater lakes, can experience large changes in their salinity as they dry out: from fresh or nearly fresh at full lake stage to hypo-/meso-/hypersaline as the water levels fall, the links to river and other lakes are cut off, and the salts start to concentrate (Timms, 2006). One of the factors controlling the lake salinity is flushing frequency by streams or floods with terminal or

less frequently flushed lakes tending to accumulate large salt concentrations (Timms, 1998b). The riverine waterholes are generally the freshest and most acidic (Timms, 1997b).

Like the majority of the Australian saline lakes/playas, the Paroo/Warrego saline wetlands, with only rare exceptions, are dominated by the sodium (Na) and chlorine (Cl) ions (De Deckker, 1988b; Timms, 1998a) and their sediments are often composed of gypsiferous clays (Timms, 1993). The lakes’ high salinity often provides protection for the sediments from bioturbation by limiting the number of aquatic fauna and especially of burrowing worms (De Deckker, 1988b). However, the dry lake floor can be affected by phytoturbation following colonisation by salt tolerant plants such as samphires (Halosarcia spp.) or by deep cracking of the self-mulching clays.

Saline lakes are also generally more productive than freshwater lakes (Timms, in press) providing greater abundance of potential fossil material. Their biological diversity, however, is usually lower (De Deckker, 1988b). Furthermore, small lakes generally support more species and are more heterogeneous than large lakes, while lakes of the same size are richer in species if they fill episodically rather than seasonally (Timms, in press).

2.5.3 Erosion

It was observed decades ago that soil, climate, and vegetation in the arid and semi-arid zones interact within bounds of a fragile equilibrium (James, 1960c). The introduction of European land use practices into those landscapes has resulted in a large increase in wind and water erosion (Fanning, 1999). Intensive grazing, particularly during drought times, ringbarking and removal of trees for fence posts and steam engine fuel, and damage from introduced pests such as rabbits have all contributed to significant thinning, or even removal of vegetation, and/or its replacement with less desirable species such as woody shrubs (James, 1960c). The decline in vegetation cover, in turn, led to an increased soil instability and its susceptibility to wind and water erosion through sheetwash, rilling, gullying, and wind drift (Pickard, 1994).

The type and amount of erosion are highly dependent on the soil type, with the grey heavy textured soils of the floodplains the least susceptible (James, 1960c). The lighter textured soils, such as the red brown earths and sandy loams, are much more prone to aeolian erosion and can suffer from scalding (i.e. removal of the top soil down to the impermeable horizon), wind sheeting, and sand drift sometimes leading to dune activation (James, 1960c). Water erosion is most common in the hilly and gently undulating areas, although, the slope is often inadequate and the soils too shallow and too easily permeable for extensive gullying (James, 1960c).

A very important type of erosion, although occurring at a much smaller scale, is the undercutting of cliffs on the western and northwestern shores of many Paroo/Warrego lakes (Timms, 2006). The process is effected by waves induced by southeasterly to easterly winds

when water levels are high (Timms, 2006). The erosion of the western cliff faces, combined with progressive sediment accumulation along the eastern shoreline in the form of source bordering dunes (lunettes), led to westward migration of many lakes in the region (Timms, 1992, 2006).

While the extent of erosion can vary spatially, Pickard (1994) suggested for the Western Division (NW to CW NSW) erosion rates 50-90 times greater than the ‘natural’ pre-European rates, while Wasson et al. (1996) went even further, suggesting up to 145-fold increase. A ten year erosion study by Fanning (1994) at Fowlers Gap Station, about 110km north from Broken Hill, yielded an average loss of 3.5mm/year for flat surfaces and 1.8mm/year for hummocky surfaces. It is generally believed that the bulk of the erosion occurred before the 1950s (Pickard, 1994). While increased rainfall and improved land management are cited as the most likely causes for this decline, the exhaustion of erodible material supply is also possible (Pickard, 1994).

The erosion can also be ‘gauged’ by burial of fences. For example, fences constructed before 1905 on the Faulkanhagan and Wannara Creeks floodplains are covered up to 1m (Pickard, 1994). Erosion bared also white cypress (Callitris glaucophylla) stumps on a red sand dune northwest of Lower Bell Lake, exposed and raised Aboriginal heaths on scalded surfaces northwest of Lake Wombah, and led to accelerated sediment influx into Lake Wyara permanently linking the Lake’s islands to the mainland (Timms, 1997b, 1998c). The rapid sedimentation in lake and clay pan basins resulted in substantial reduction of full lake water depths, thus shortening the period during which a lake holds water following a filling event (Timms, 1997b, 1998b, 2006).

Increased surface sediment mobility is also reflected in the frequency and magnitude of dust storms (e.g. Figure 2.6). They can affect the composition of the palaeoenvironmental records as they relocate over short and long distances large amounts of sediment together with modern and fossil pollen, phytoliths, sponge spicules, diatoms, and charcoal (Clarke, 2003; Hesse and McTainsh, 2003). In some cases, the re-deposited material can contribute to preservation of records by providing a protective cover over the primary sediments, for instance, through encroachment of dune sand onto the exposed lake floor.

2.5.4 The erosion-deposition cycles in a saline lake/playa-lunette system

A special semi-arid feature that has formed in response to cycles of erosion and deposition is a saline lake/playa-lunette system (Bowler, 1973; Timms, 1992). In the Paroo/Warrego Region the light grey to white colour of the clay/gypsum lunettes stands out in high contrast to the surrounding red sandy dunes. Commonly, the lunette system consists of a high outer gypsum dune and a lower inner clay-rich dune extending along the eastern shore of the lake (Timms,

1993 and personal observation). While the older gypsum dunes are relicts of past processes, it is possible that at least some of the younger clay lunettes are still active (e.g. at Mid Blue Lake).

Figure 2.6 Dust storm approaching Lake Bindegolly.

The lunette formation depends on a specific set of weather conditions and physical/chemical characteristics of the lake basin. These requirements are particularly well described for lunette systems in southeastern Australia. Initially, in the pre-lunette phase, the humid climate supports a positive water balance: the lake tends to be full, fresh, often deep, and is recharging the groundwater (Bowler and Wasson, 1984). Sandy quartz beaches, and even quartz dunes, are likely to form on the lake’s margin (Bowler, 1973, 1983).

A change to drier/higher evaporation climatic conditions leads to transition from a permanent freshwater to an ephemeral saline lake (Bowler, 1986). As the water levels drop and the solute concentrations increase, the least soluble calcium carbonate starts to precipitate and deposit on the lake floor, followed by gypsum, and finally the most soluble halite (Bowler and Wasson, 1984; Warren, 1982; Williams et al., 1993). Due to the progressive shrinking of the water body, the highest accumulation of surface salt crystals occurs close to the evaporitic centre of the lake (Warren, 1986).

A subsequent influx of freshwater into the lake results in dissolution of the entire surface halite and some of the gypsum deposits (Teller et al., 1982; Warren, 1982). Longer-term preservation

of the precipitated crystals within the lacustrine sediments, and particularly gypsum, depends to a large extent on the characteristics of the subsurface water, particularly its saturation with respect to the particular mineral (Bowler, 1986). The high solubility of halite that results in its complete dissolution during each filling event, greatly reduces its chances of long-term preservation (Warren, 1982).

The drop in the lake water levels also leads to changes in the subsurface waters, as the lake brine, blown towards the dry edges of the lake, percolates down toward the relatively fresh groundwater table increasing its salinity (Bowler, 1986; Bowler and Wasson, 1984). An ongoing repetition of the drying-filling cycle results in rapid transfer of salts from surface to subsurface waters. On reaching supersaturation with respect to sulphates, subsurface gypsum begins to precipitate as small interstitial/displacive crystals as well as coatings on larger sand- sized gypsum grains (Bowler, 1983, 1986; Warren, 1982). Over time, the lake/playa can develop an underlying brine pool, that is more saline than the surrounding and underlying regional groundwater (Bowler, 1986; Macumber, 1991; Teller et al., 1982). The reaching of equilibrium between the brine pool and the lacustrine deposits aids the preservation of crystals within it (Bowler, 1986).

In cases where the lake bed elevation allows evaporation below the water table of the surrounding landscape, the discharge of saline groundwater into the lake basin will enrich its salt content (Bowler and Wasson, 1984). With the onset of a new humid phase, the lake waters might again start to recharge the groundwater, possibly leading to reduction in the lake’s salt content (Bowler and Wasson, 1984; Macumber, 1991).

During the dry lake stage (accompanied by groundwater tables at least 1m below the lake floor) evaporation through the capillary fringe leads to efflorescent growth of fine, closely-spaced crystals of the more soluble salts (e.g. halite) that break the lake surface sediments into small aggregates (Bowler, 1980, 1983, 1986; Bowler and Wasson, 1984). These clay and/or clay-sand aggregates, often referred to as pellets, can also be produced by the mechanical breaking of mud curls or disintegration of a microrelief pattern of domal blisters (Bowler, 1973, 1983). Once the wind speeds become strong enough to overcome the hydroscopic attraction between the pellets (i.e. exceed about 10m/sec), they can be deflated from the lake basin together with the gypsum crystals (Bowler and Wasson, 1984; Timms, 1992; Wasson, 1983). The potential stabilisation of the lake floor by plants is usually prevented by the high salinity of the lake muds and the seasonal/periodic flooding (Bowler, 1973, 1980, 1983).

In Lake Amadeus, however, Chen et al. (1991) attributed the surface gypsum deposits to the discharge of sulphate rich groundwater around the margins of the lake during periods of high water tables. The resultant groundwater pools persist long enough to produce sand size gypsum crystals, which are later deflated and deposited onto the lunette. 21

With persistence of strong and generally unidirectional winds during the dry/low water table season, the layers of the periodically deflated materials accumulate into a crescent-shaped dune with a windward slope usually steeper than the lee slope (Bowler, 1973, 1980, 1983). The presence of marginal vegetation can play an important role in the initial stages of lunette formation by trapping some of the deflated material and, in some cases, the lunette can cap the quartz sand beach or dune (Bowler, 1973, 1983).

Sometimes, the individual layers become preserved as laminae (1cm to <1mm thick) of well sorted quartz and clay pellets, particularly if the deposition rates exceed any soil formation processes (Bowler, 1973, 1983). While the proportions of sand and clay in the lunette reflect the amount of sand in the sediment source, a typical clay lunette contains at least 20% clay (Bowler, 1973). The proportions of clay pellets to quartz grains can, however, vary considerably between dunes and within a single dune (Bowler, 1983). The size of the lunettes is closely related to the source lake basin and in Australia it usually ranges between 5-15m in height rarely exceeding 20m (Bowler, 1973).

After each deflation event, when the conditions become temporarily moister or cooler (thus closer to dew point), the new clay pellet/gypsum layer becomes stabilised as the clays and salts hygroscopically absorb moisture and turn into a sticky coalescing mass (Bowler, 1973, 1980, 1983; Wasson, 1983). As a result, the lunette, unlike sand dunes, becomes resistant to erosion and does not migrate (Bowler, 1973, 1983).

Lunette sediments can contain small and variable amounts of carbonate, sulphate, and chloride, whose distribution is closely related to their solubility (Bowler, 1973, 1983). Chlorides are often removed from the lunette very soon after deposition, with some of them leached back into the lake basin (thus perpetuating high chloride concentrations within the lake) and the groundwater, and some accumulating in the soils downwind of the lunette (Bowler, 1973, 1983). On the other hand, the less soluble carbonates and sulphates, occurring as calcite, dolomite, or gypsum, are permanently removed from the lake basin and retained within the lunette (Bowler, 1973, 1983).

A continued increase of salt concentrations in the subsurface water leads eventually to narrowing of the difference between it and the surface brine (that is generally still more saline) resulting in suppression of vertical salt transfers (Bowler, 1986). The preserved surface salts become more effective in formation of a salt crust, which protects the sediments from further deflation and the lunette formation stops (Bowler, 1983, 1986). The gypsum crystals may, however, continue to form in the subsurface zone (Bowler, 1986).

A prolonged humid phase promotes colonisation of the lunette by plants and initiation of soil formation (Bowler and Wasson, 1984). Intensification of the pedogenic processes leads to leaching of the near surface salts, including carbonates and gypsum (Bowler, 1973, 1983; 22

Bowler and Wasson, 1984). As the leachate-enriched water is returned to the atmosphere, for instance through evapotranspiration, secondary mineral deposition and gypsum precipitation takes place at lower depths of the soil profile (Aref, 2003; Warren, 1982). The generally rapid progress of the re-crystallisation results in formation of many small crystals, i.e. mostly gypsarenite (Warren, 1982).

2.6 Vegetation

2.6.1 Modern vegetation

The study area lies in Mulga Land Bioregion (Sahukar et al., 2003; Wilson, 1999) and, in spite of its arid/semi-arid label, it is characterised by an immense diversity of flora. Consequently, several different vegetation descriptions and classification systems are available for this region (e.g. Beadle, 1948; Casanova, 1999; Cunningham et al., 1992; James, 1960b; Keith, 2004; Kingsford and Porter, 1999). For the purpose of this study, the native vegetation was classified into four major groups: mulga dominated shrublands and woodlands (dominant group), shrubby and grassy semi-arid woodlands (major group), arid riverine chenopod shrubland, and saline and freshwater wetlands (both are minor groups in terms of spatial coverage, but not necessarily in importance) (in sensu Keith, 2004 and Kingsford and Porter, 1999). A separate section was devoted to woody shrubs.

2.6.1.1 Mulga dominated shrublands and woodlands

The most widespread plant formation is the shrublands and woodlands dominated by mulga (Acacia aneura) (James, 1960b; Sahukar et al., 2003; Wilson, 1999). Keith (2004) recognises four main classes within this formation in the Paroo/Warrego area: north-west plain shrublands (and woodlands), gibber transition shrublands and woodlands, stony desert mulga shrublands, and sand plain mulga shrublands. The tall (around 5-7m) north-west plain shrublands occur on the flat to undulating sandplains as different combinations of a wide range of shrub and tree taxa as well as single-species stands (Appendix 1 Group 1) (James, 1960b; Keith, 2004). The shrubland also supports a well developed ground cover dominated by perennial tussock grasses, a 1-3m tall understorey, and sparse emergent trees (up to 10m) represented mostly by ironwood (Acacia excelsa), white cypress pine (Callitris glaucophylla) and poplar box (Eucalyptus populnea).

The gibber transition shrublands and woodlands occur mostly in clayey depressions and drainage lines within the rocky areas that occur in the transition zone between the clay plains in the east and the gibber landscapes in the west (Keith, 2004). The trees within this class can reach heights of up to 10m and the canopy can be dense (particularly within gidgee (Acacia cambagei) stands) to open (usually with multi-species composition) (Appendix 1 Group 1). Brigalow (Acacia harpophylla) becomes more common in the east (Keith, 2004; Wilson, 1999). 23

The understorey consists mainly of scattered chenopod shrubs or perennial tussock grasses. The native ground cover within this class is under increasing pressure from invasion of the introduced buffel grass (Cenchrus ciliaris).

The much lower (generally <4m) stony desert open mulga shrublands, with an understorey of sparse perennial plants and variable densities of ephemeral herbs (Appendix 1 Group 1), occupy stony ranges, downs, and gibber plains (Keith, 2004). The sand plain mulga shrublands stretch over the dunes of the sandplains and are composed mainly of sparse tall shrubs with an open cover of smaller shrubs and ground cover of perennial tussock grasses and ephemeral herbs (Appendix 1 Group 1) (Keith, 2004).

2.6.1.2 Shrubby and grassy semi-arid woodlands

Another important vegetation formation is the shrubby and grassy semi-arid woodlands, of which three classes have been described by Keith (2004) in the central part of the Paroo/Warrego catchment: grassy north-west floodplain woodlands, western peneplain woodlands, and shrubby semi-arid sand plain woodlands. The grassy north-west floodplain woodlands occur mostly on heavy clay soils of the riverine floodplains. They are characterised by an open canopy of 10-20m tall trees, dominated in higher areas by coolibah (Eucalyptus coolabah) with co-dominant poplar box (E. populnea) or grey box (E. macrocarpa) and on the Paroo River floodplains by yapunyah (E. ochrophloia) with a sparse shrub understorey and continuous ground cover with abundance of summer (C4) grasses (Appendix 1 Group 2) (Casanova, 1999; James, 1960a; Keith, 2004).

The western peneplain woodlands within the study area are at the western limits of their distribution and their occurrence is largely confined to drainage lines (Keith, 2004). They are mostly represented by open 10-15m tall Eucalyptus woodlands with open shrub understorey or continuous grassy ground cover (Appendix 1 Group 2). The understorey varies greatly, depending on the soil moisture and texture as well as disturbance history. In general, the well drained sandy soils tend to favour denser and more diverse shrub communities, while the heavier floodplain soils are more likely to support sparse shrub and dense grass cover.

The much drier undulating red-brown sand plains often support open woodlands dominated by belah (Casuarina pauper) and rosewood (Alectryon oleifolius) with an understorey of low chenopod shrubs and ephemeral herbs (Appendix 1 Group 2) (Keith, 2004). The formation is most common in the southern part of the study area.

2.6.1.3 Arid riverine chenopod shrubland

The treeless riverine chenopod shrublands can be found on the deep, often saline, grey-brown clays of the flat alluvial plains and dry lake beds associated with the palaeodrainage systems (Keith, 2004). The shrubland is characterised by open chenopod cover (0.5-2m tall), dominated

by Atriplex spp. and Maireana spp., with scattered ephemeral and perennial herbs and grasses (Appendix 1 Group 3).

2.6.1.4 Saline and freshwater wetlands

The vegetation diversity of the Paroo/Warrego wetlands mirrors their geomorphological variability and can be broadly classified into seven types: salt lakes, freshwater lakes, river channels and waterholes, lignum swamps and overflow plains, Eleocharis swamps, blackbox swamps, and claypans and canegrass swamps (Kingsford and Porter, 1999). Many wetlands might, however, support a combination of these types.

When full, the saline lakes provide attractive habitat for macrophytes and charophytes (Casanova, 1999; Kingsford and Porter, 1999). Many of the aquatic plants can survive and produce seed at higher salinities but freshening of the water might be needed for germination to occur (Keith, 2004). The lake margins are commonly fringed by samphires (Halosarcia spp.) and less often by spiny sedge (Cyperus gymnocaulos), which are frequently reinforced by other sedges (Cyperus spp.), mulka (Eragrostis dielsii), and monkey flower (Mimulus repens) following a flood (Appendix 1 Group 4). Shrubs are often found further away from the water edge and a variety of terrestrial herbs occupy higher ground and adjacent dunes.

The margins of freshwater lakes are often fringed with dense sedges, grasses and herbs such as the red water-milfoil (Myriophyllum verrucosum) (Appendix 1 Group 4) (Kingsford and Porter, 1999). Trees, mainly Eucalyptus spp. and river cooba (Acacia stenophylla), often surround the lakes at slightly higher elevation. The dry lake bed might be invaded by carpet weed (Glinus lotoides), native liquorice (Glycyrrhiza acanthocarpa), and introduced Heliotropium species. Eucalyptus spp. and river cooba, with an understorey of sparse to moderately dense shrubs and ground cover of annual and perennial grasses, annual herbs, and sedges are also common along the margins of river channels and waterholes, while water primrose (Ludwidgia peploides) and duckweed (Lemna spp.) can often grow in water within them (Appendix 1 Group 4) (Casanova, 1999; Kingsford and Porter, 1999).

The blackbox swamps can occur on edges of riverine floodplains or on sandplains (Kingsford and Porter, 1999). Typically, the swamp’s margins are lined with black box (Eucalyptus largiflorens) and poplar box (E. populnea) with ground cover of grasses, sedges, and herbs (Appendix 1 Group 4). Clear water might support aquatic herbs such as red water-milfoil and water nymph (Najas tenuifolia) as well as charophytes.

Lignum (Muehlenbeckia florulenta) swamps with a dense ground cover of grasses and sedges (Appendix 1 Group 4) develop in areas that are flooded every two to eight years (Casanova, 1999; Kingsford and Porter, 1999). The swamp’s margin can be sparsely populated by black box, poplar box, and river cooba, while aquatic herbs might appear after flooding. More

frequently flooded, and mostly more saline, shallow claypans and canegrass swamps support usually sparse to dense cover dominated by canegrass (Eragrostis australasica) (Casanova, 1999; Kingsford and Porter, 1999) (Appendix 1 Group 4). Flooding, however, might encourage growth of some aquatic herbs, charophytes, and mud colonisers.

The Eleocharis swamps are shallow to moderate depth basins within the sandplains, which support extensive cover of sedges (incl. Eleocharis spp.), canegrasses, grasses, and herbs sometimes fringed by occasional lignums (Muehlenbeckia spp.) or nitre goosefoot (Chenopodium nitrariaceum) (Appendix 1 Group 4) (Kingsford and Porter, 1999).

2.6.1.5 Woody shrubs

Many areas, and particularly the sandplains that were affected by overgrazing and soil erosion following European settlement, have experienced loss of perennial grasses accompanied by ‘infestation’ by native woody shrubs, including Dodonaea spp., Eremophila spp. and to a lesser degree Senna spp. (Keith, 2004; Page et al., 2000; Page, 1995; Queensland Parks and Wildlife Service, 1999; Sahukar et al., 2003). Another contributing factor to the increase in woody shrubs may be the decrease in fire frequencies (the main low shrubs’ control agent) with the arrival of European settlers (Drysdale, 1995; Hodgkinson and Harrington, 1985; Page, 1995).

On the other hand, Witt (1993 cited in Queensland Parks and Wildlife Service, 1999) suggested that the woody shrub expansion might be just a part of natural balance related to a particular sequence of weather, and thus it will be replaced in the future by different species in response to subsequent climatic changes. He argued that prolonged drought, such as the one in the first half of the 20th century, could have caused an extensive tree and shrub dieback, resulting in a more open/grassy landscape, while the exceptional rainfall of the 1950s and the 1970s promoted the tree and shrub regeneration (Witt et al., 2006).

The main issue with the expansion of woody shrubs is that they can often reach high densities and substantially reduce the diversity of the original plant community (Morgan and Terrey, 1992; Sahukar et al., 2003). Since the woody shrub species are unpalatable to stock and their proliferation reduces pasture growth (Hodgkinson and Harrington, 1985), they are commonly referred to as ‘woody weeds’. The woody shrubs are common and widespread throughout the study area.

2.6.2 Vegetation history

The earliest record of the Paroo/Warrego Region’s vegetation, that covered the period from the Eocene to the late Pliocene-early Pleistocene boundary, came from palynological reconstruction undertaken by Martin (1997) from bores along the Darling River, near Bourke. One of the more significant findings of this study was transition in the early Pleistocene from a mixture of rainforest taxa and Casuarinaceae forests to open vegetation composed of sparse Casuarinaceae

and Eucalyptus spp. with an understorey rich in Asteraceae, Poaceae, Cyperaceae, and Chenopodiaceae, thus similar to the composition of the area’s modern vegetation.

A more recent ‘fragment’ of the vegetation history was reconstructed by Field et al. (2002) from sediments at Cuddie Springs, about 160km east of Bourke. The undated oldest part of the record supported the presence of a Casuarinaceae forest, which later, at an also undated time, declined in favour of Chenopodiaceae, probably indicating transition to more semi-arid conditions. The continued presence of aquatic taxa, including Cyperaceae, Typha and Myriophyllum, suggested, however, that long term flooding of the lake was common. At around 35 000 BP, the conditions become briefly wetter with the lake supporting Azolla, but the catchment continued to support chenopod shrublands and grasslands with only sparse Casuarinaceae. As a result of the continued amelioration, Casuarinaceae and aquatic taxa levels increased again around 28 000 BP with Poaceae, Asteraceae, and other herbs also present in the record, while Chenopodiaceae declined. The highly disturbed Last Glacial Maximum to modern record is, in its earlier part (~19 000 - 6 000 BP), dominated by an unprecedentedly high proportions of Chenopodiaceae and is barren of pollen in its latter part, except for the surface sample. This surface sample shows the highest Eucalyptus presence recorded in this study and Casuarina levels similar to those observed about 30 000 years ago.

On a more recent and shorter timescale, James (1960c), following extensive surveys in the western NSW area in the 1950s, reported a spread of inedible shrubs, mainly desert (silver) cassia (Senna artemisioides), budda (Eremophila mitchellii), and hopbush (Dodonaea spp.). This trend was confirmed and quantified by analysis of aerial photographs of Currawinya National Park for the period between 1952-1991 by Witt and Beeton (1995). The study has shown that the Park’s woody cover, including Acacia aneura, Eucalyptus spp., other Acacia spp., Eremophila spp., Dodonaea spp., Senna spp., and Hakea spp., has increased from 5-10% to 20% with the major expansion occurring in the 1969-1981 period within the hard mulga, dissected residual, and sand plain mulga landsystems. On the dunefields, trees and shrubs have expanded from 10% of the cover in 1952 to, in some places, 20-50% by 1991, with Dodonaea viscosa and Eremophila sturtii responsible for most of the increase. No changes were, however, detected in the floodplain woodlands and scattered grasslands, nor in the open woodlands and shrublands in the lake areas.

Most of the vegetation changes recorded by Witt and Beeton (1995) were attributed to livestock grazing along with fire suppression (mostly due to decrease in fuel loads) and establishment of extended periods of high summer and winter rainfalls in the years 1973-1977. Soil disturbance from stock, cultivation, road construction, and other earthworks, was also proposed as a contributing factor since the woody shrub species tend to recover faster than grasses on eroded surfaces.

The stable carbon isotope ratio (13C/12C) of sheep faeces under shearing sheds on neighbouring Currawinya and Talyealye properties was used by Witt et al. (1996) to further investigate the vegetation changes in the Paroo Region. The study showed that there was no significant change in the vegetation during the period 1956-1995 on the Talyealye property, which remained over this time mostly open to lightly wooded. This was possibly due to limits on sheep numbers imposed by the lack of permanent water bodies and extensive fires in the 1950s and the 1970s that killed the woody shrubs. The Currawinya property, however, experienced a dramatic shift from perennial grasses (C4) to woody shrubs (C3) around the mid-1970s. The change most likely was driven by lack of fires and overgrazing, since the well watered property was able to support higher sheep numbers for longer periods.

Another study by Witt et al. (2006) of sheep manure deposits at Ambathala Station (just north of the northern end of the Paroo River catchment) provided a slightly longer glimpse into the vegetation changes by covering the period 1930-1995. The analysis of the carbon isotope ratio, leaf cuticle, pollen, and aerial photos showed only very slight changes in proportions of the different plant groups and taxa over time, thus providing no evidence for the widespread expansion of woody taxa postulated by many land managers.

2.7 Fire history

Little is known about the fire history of the area. Hodgkinson and Harrington (1985) suggested that lightning or anthropogenic fires were common in semi-arid areas, particularly in response to fuel load increases following major wet periods. A significant change to the fire regime occurred in the 1880s and the 1890s with the discovery of the Artesian Basin, which resulted in a proliferation of artificial permanent watering places and dramatic increase in densities and spatial distribution of domestic stock as well as native and feral animals (Drysdale, 1995). The subsequent decrease in fuel suppressed fire occurrence to periods often in excess of 50 years (Hodgkinson and Harrington, 1985).

James (1960c) reported sporadic occurrence of fires in the region with a major outbreak, sweeping over more than a million acres, in 1957. He suggested that fires followed by a reasonable summer rainfall assisted germination of mulga seeds. Regular burning was also recommended at that time as a control tool for unpalatable buck spinifex (Triodia mitchellii).

Two extensive fires were reported to have occurred on the Talyealye property, just south of Currawinya, in the 1950s and the 1970s (Witt et al., 1996). Research in the Currawinya National Park (Page, 1995; Witt et al., 1996), however, failed to discover any documented information on its fire history or physical evidence of fires within the park boundaries since the 1950s. An assessment of fuel loads indicated that the lack of grass and fine fuel precluded the

use of fire as a viable tool for reducing woody shrub populations (Page, 1995). The initial assessment was confirmed by failed fire trials (Dollery, 2001).

2.8 Human impact

2.8.1 Aboriginal occupation

The main Aboriginal groups associated with the middle of the Paroo/Warrego Region are Budjiti, Mardigan, Baakandji/Paakantyi, Mulangarpu, and Wilyakali (Alliance Resource Economics, 2002). Although there is little information on the duration of Aboriginal occupation in the area, the archaeological studies confirm their presence at least since the late Holocene (last 2500 years) (Holdaway et al., 2002; Robins, 1999) and limited evidence extends their occupation in the area to the end of the last glaciation (about 14 000 years ago) (Robins and Connolly, 2001). The non-intensive and relatively short periods of occupation of various sites are probably closely related to bursts in food abundance (especially plants, fish, and birds) following flooding (Holdaway et al., 2002; Robins, 1999). Aside from fishing, hunting, and gathering, the Paroo/Warrego indigenous people were also harvesting seeds and using regular burning to promote grass dominance for kangaroos (Goodall, 1999).

After the European settlers took over the waterholes and hunting grounds, many of the Aboriginal inhabitants were forced to settle on stations (Alliance Resource Economics, 2002). Although only a few remain living in the vicinity of the Paroo and Warrego Rivers, many still visit the area to take part in ceremonies and traditional activities, and are involved in management of the national parks (e.g. Bindegolly and Currawinya NPs) (Queensland Parks and Wildlife Service, 1999).

2.8.2 European settlement

The positive reports by first European explorers of the ‘apparently limitless grasslands’ and ‘open woodlands’ in the inland regions encouraged the colonial and later state governments to promote land settlement and pastoralism (Heathcote, 1994). The first settlers reached the Paroo/Warrego Region in the 1840s, but grazing was limited to areas along the rivers and around the permanent or semi-permanent open water bodies (Drysdale, 1995; Goodall, 1999). A trend towards smaller and often unsustainable properties was initiated in 1860 under the ‘Unoccupied Crown Land Occupation Act’, which allowed leases of 64km2 and later 256km2 (Alliance Resource Economics, 2002; Drysdale, 1995). The above average rainfalls and high wool prices at that time promised profitable ventures (Griffiths, 2001). However, drought, falling cattle prices, and rabbit plagues in the following 20 years prompted amalgamations of small leases into large pastoral holdings (Drysdale, 1995).

The discovery of the Artesian Basin in the 1880s and the rapid spread of bores increased water availability, allowing intensification of stocking rates, extension of pastoralism into areas without natural surface water, and an increase in density of settlement (Alliance Resource Economics, 2002; Drysdale, 1995; Goodall, 1999). It also favoured proliferation of native (kangaroos) and feral (rabbits, goats, and pigs) animals that together with the increased stock numbers (reaching over 15 million in the Western Division of NSW in the 1887-1897period) caused possibly irreversible damage to much of the grassland area, particularly through selective grazing of more palatable perennial grasses (Condon and Stannard, 1956; Drysdale, 1995; Pearson et al., 2003).

The use of palatable mulga to feed starving stock during the 1889-1901 drought as well as felling of other trees and large shrubs for fencing timber and for fuel for steam engines that were running bore pumps, further contributed to the land degradation and increase in the soil erosion (Drysdale, 1995; Fanning, 1999; Pickard, 1994). The drought conditions did not, however, prevent a new wave of closer settlement and property division after World War I (Alliance Resource Economics, 2002; Drysdale, 1995). Falling stock prices in 1921, the depression in the 1930s, another drought, and introduction of (unrealistic) minimum stocking rates by the ‘Land Act of 1927’, forced many graziers to overstock their already stressed properties (Drysdale, 1995). The carrying capacity of the Currawinya properties, for example, was assessed in 1929 as 28 000-39 000 head, yet recorded sheep numbers were over 48 000 in 1929, reaching 65 000 in 1944 (Page, 1995).

The above average rainfalls in the 1950s resulted in significant increases in plant biomass, and hence an increase in fire frequencies and suppression of mulga and woody shrubs regeneration in favour of grasslands (Drysdale, 1995). This grassland renewal, together with an increase in wool prices, inspired another governmental ‘Closer Settlement Policy’ in an attempt to improve the efficiency of pasture utilisation (Drysdale, 1995). Once again this led to overstocking as wool prices fell unexpectedly and production costs increased (Drysdale, 1995).

The period of 1960-1990 saw further increases in mulga east of the Warrego River and in woody shrubs west of the river, with a subsequent decrease in pasture areas (Drysdale, 1995). It resulted in a new (and still ongoing) wave of property amalgamation and a decrease in human population (Drysdale, 1995; Pearson et al., 2003). In the Paroo Shire, for instance, the number of properties decreased from 224 in 1960 to 140 in 1993 (Drysdale, 1995).

At present, the main land use in the middle Paroo/Warrego Region is still grazing, beekeeping, and localised mining (e.g. opals around White Cliffs and Yowah) (Alliance Resource Economics, 2002; Pearson et al., 2003). Conservation and tourism land uses are growing strongly with the recognition of ecological values. As a result, several nature reserves and national parks were established in the area, including the Nocoleche Nature Reserve (1979), the 30

Lake Bindegolly National Park (NP) (1991), the Currawinya NP (1991), and more recently in 2002 the Paroo-Darling NP, Paroo-Darling State Conservation Area, and Ledknapper Reserve (2002). Furthermore, the value of Currawinya and Paroo River wetlands was internationally recognised by adding them to the Ramsar list. The most recent efforts for better protection and management of the region resulted in an Intergovernment Agreement for the Paroo River between NSW and QLD (2003), assessment of water availability in the Paroo carried out within the framework of The CSIRO Murray-Darling Basin Sustainable Yields Project (2007), and nomination of the Paroo River Catchment for listing as ‘National Heritage’ under Australian national environment laws (2007).

Chapter 3

Methods

3.1 Introduction

This chapter contains descriptions of methods and procedures utilised in obtaining data presented in Chapters 4-10. The choice of methods used in this research was guided mainly by their usefulness in extraction of the target palaeoenvironmental proxies. However, their simplicity/labour intensity, expense, and access to analytical instruments were also of considerable importance. The evaluation of the used methods in light of their performance in this study and implications for future arid and semi-arid palaeoenvironmental research is provided in Chapter 4.

3.2 Site selection

The selection of the cored lakes was based on the following criteria: i) sedimentary information provided by preliminary coring of several lakes in the Paroo Region in 1998 including Lake Bindegolly, Lake Numalla, Lower Bell Lake, and Palaeolake; ii) advice from B. Timms (pers. comm. 2000), based on his 25 years of limnology research in the Paroo/Warrego Region; iii) representation of a range of drying regimes (with lake area and depth as proxies for the drying frequencies), origins (e.g. deflation playa, waterhole along an ephemeral braided stream channel, tectonic, blocked valley) and chemistry (ranging from fresh to hypersaline); iv) presence of clay and gypsum lunettes; v) information from topographical and geological maps, aerial photos, and satellite images, including geomorphological and geological characteristics of the lake and catchment area, connectivity between lakes, and vegetation densities and type; and vi) accessibility, including entry permits to properties, the ease of vehicle access to the lake, and the ability to venture into the lake with heavy equipment (in some cases, the extreme softness of the sediment prohibited coring at an adequate distance from the shore).

3.3 Field methods and sample collection

3.3.1 Lake and dune cores

Where the lake floor was submerged or moist (all lakes with exception of Lake Willeroo and Lake Yandaroo), a PVC pipe (80 mm diameter) was slide-hammered into the lake floor (Faegri and Iversen, 1989) (Figure 3.1 images 1 & 2) until the pipe would not go any further, or the difference between the distance to the top of the sediment outside and inside of the pipe started to increase (Gayler, 2000). The latter case was attributed to the pipe spearing into the underlying sediment after the pipe was plugged with clay-rich mud. While the inside/outside difference is usually attributed to compression of sediment, particularly in peaty environments, in this study the compression was assumed to be negligible due to the comparatively high density of the pre-cored sediment type, which is composed mostly of tightly packed clays and silts. Where the sediment was dry (e.g. Lake Willaroo basin and lunettes), the sediment was augered by hand bucket auger (Figure 3.1 image 3) and representative samples were collected into plastic bags at measured depths.

Figure 3.1 The sediment extraction by coring (1 & 2) and augering (3).

The coring sites were selected according to the following criteria (in some cases this was achieved by extraction of multiple cores at different positions, in other cases by balancing the various criteria): i) sampling close to the lake’s centre targeted the deepest and usually most continuous fine- textured sediment (Faegri and Iversen, 1989), which, due to the longest periods of inundation, was likely to be least affected by deflation during dry episodes and reworking by waves during filled episodes; ii) sampling close to the eastern margin of the lake basin targeted the longest record in those lakes that are progressively advancing upwind (i.e. in the southwesterly direction) due to undercutting of the upwind shore and deposition of material on the downwind shore (Bowler, 1973, 1983; B. Timms, pers. comm. 1999); iii) sampling far from inflow channels to minimise their interference (Faegri and Iversen, 1989); and iv) accessibility to target (pre-selected) sites, e.g. the surface in Lower Bell and Wyara lakes was too soft to allow walking the long distance required to reach the lake’s centre, while Lake Numalla was too deep to core at the centre of the lake with the available equipment.

The augering sites on the dunes were located close to their highest points. The augering depth was limited by the maximum manageable auger handle length (~5m) and the nature of the sediment, e.g. encountering groundwater table, unconsolidated/extremely sandy sediment, or thick gypcrete crust. After each extraction of the auger bucket, the sediment was carefully laid out on a plastic sheet and depth measurement was taken in the auger hole using a metal measuring tape (Figure 3.1 image 3). Sediment was then described with respect to colour and general stratigraphy. Grab samples were collected at regular intervals and/or middle of distinct sediment units and brought back to the laboratory for further processing.

3.3.2 Surveying

An automated total station was used to survey transects to determine relative relief for the lakes and the dunes, and to provide a datum for interrelation of sediment units. The position of the coring/augering sites was recorded with a GPS unit and/or drafted on sketch maps (Appendix 2).

3.3.3 Vegetation and modern pollen

Due to major time constraints, extensiveness of the study area and the lack of substantial changes in taxa, the vegetation surveys were limited and relatively simple. Vegetation transects were carried out only around the Palaeolake (Gayler, 2000). For other lakes, the vegetation was described by simplified field sketches. The information about the regional vegetation

distribution and composition was derived from published literature (e.g. Chapter 2 section 2.6.1).

Reference pollen was collected opportunistically whenever flowers were encountered around the lakes and on the journey between them. Plant samples were also collected to aid the identification of the source plants.

3.4 Laboratory processing of the samples and cores - summary

The unopened lake sediment cores and bulk samples of augered sediment were transported back to the laboratory, where they were subsampled and subjected to further processing. The processing steps are summarised in Figures 3.2 and 3.3, and the details are provided in the following sections.

Figure 3.2 The guidelines for processing sediment cores.

Figure 3.3 The guidelines for processing augered material.

3.5 Sediment description

3.5.1 Colour

The Revised Standard Soil Color Charts (based on Munsell colour system) (Oyama and Takehara, 1993) were used to describe the colour of moist sediment. The wide range of colour chromas and hues within the lake core, and frequently very gradual transitions between them, as well as the time delay between sampling and description, necessitated some simplifications and approximations of the data.

3.5.2 Stratification and sharpness of boundary

Sediment stratification and the sharpness of boundary between sediment layers was classified according to the revised Troels-Smith system (Kershaw, 1997). The augered sediment allowed only raw estimation of stratification and was too disturbed for the sharpness of boundary determination.

3.5.3 Bioturbation

The presence and nature of bioturbation (mostly roots, root tunnels, mottling, and disruption of laminations) was assessed visually during the description of the lake and dune sediments. The root and root tunnel density was quantified using the 0-4 abundance/content scale (section 3.5.4 below) and was presented as part of the general sediment description figures for the individual cores (Chapters 5-10).

3.5.4 Abundance/content

Abundance/content (e.g. of gypsum, organic material, etc.) was described using a scale from 0 (absent) to 0.5 (rare), 1 (occasional), 2 (considerable/medium presence), 3 (frequent), and 4 (abundant or sole component) (adapted from Kershaw, 1997 and Troels-Smith, 1955).

3.6 Sediment texture and particle size analysis

3.6.1 Sediment texture

The basic texture of the sediment was described using a simple field texture assessment flowchart developed by Nortcliff and Landon (in Rowell, 1994, p.10).

3.6.2 Particle size analysis

In preparation for the analysis, 2ml sediment samples were initially treated with 10% hydrochloric acid (HCl) to remove carbonates, followed by two washes with distilled water and, finally, treatment with 3% hydrogen peroxide (H2O2) to remove organic matter and disperse the sediment (J.-B. Stuut, pers. comm. 2005; T. Rolph, pers. comm. 2005; Murray, 2002). Just before the measurement, the sediment was sieved through 500µm steel mesh.

The particle size distribution was measured using laser particle analyser, Malvern Mastersizer Microplus, with 0.05 - 556µm size range. The resultant 61 size categories were clustered into 10 more general categories (Gale and Hoare, 1991, pp.58-9): medium sand, fine sand, very fine sand, coarse silt, medium silt, fine silt, very fine silt, coarse clay, medium clay, and fine clay.

3.7 Mineral Magnetic Susceptibility

The mineral magnetic susceptibility of samples was measured using two different methods (instruments): Multi-Sensor Core Logger and Bartington MS2B sensor, depending on the nature

of the samples. In both cases, the final values were expressed as mass specific magnetic susceptibility in SI units (m3kg-1), which is the favoured unit for environmental materials (Maher et al., 1999; Smith, 1999).

3.7.1 Multi-Sensor Core Logger (MSCL)

Where sediment core was extracted using a PVC pipe the magnetic susceptibility was measured by a Multi-Sensor Core Logger (Bartington loop sensor MS2C) prior to splitting the core. The magnetic susceptibility was measured in an alternating magnetic field of 0.565kHz at 1cm intervals.

3.7.2 Bartington MS2B sensor

The Bartington dual-frequency sensor MS2B was used to measure the magnetic susceptibility of augered sediment as well as selected samples extracted from the PVC cores, for which the determination of frequency dependent susceptibility was desired.

As part of the preparation for the dual frequency measurements, the samples were dried at 40oC (Maher, 1986, 1998; Walden, 1999). They were gently disaggregated with ceramic mortar and pestle and sieved through a 2mm sieve (Gale and Hoare, 1991; Maher, 1986, 1998), before being transferred into cylindrical plastic pots. The measurements were taken at low frequencies of 0.465kHz and then at high frequencies of 4.65kHz. The mass specific magnetic susceptibility of the samples was calculated using a formula provided by T. Rolph (pers. comm. 2000):

χ 3 /kg)(m × gvaluencalibratio )( /kg ) (m3 /kg) = m χs (g)sampledryofmass (1) where χs is the sample’s mass specific magnetic susceptibility and χm is the sample’s measured magnetic susceptibility value.

3.7.3 Frequency dependent susceptibility

The frequency dependent susceptibility (%), i.e. the difference between the low frequency and high frequency mineral magnetic susceptibility values, allowed detection of ultrafine (<0.03µm) superparamagnetic ferrimagnetic minerals. This was calculated according to a formula (Dearing, 1999a):

3 3 χlf /kg)(m − χhf /kg)(m χ = 3 ×100 (2) fd% χlf /kg)(m

where χfd% is the percentage frequency dependent susceptibility, χlf is the low frequency magnetic susceptibility value, and χhf is the high frequency magnetic susceptibility value.

3.8 Organic matter content

The organic matter content in the sediment was estimated using a loss on ignition (LOI) method modified from Ball (1964) and Rowell (1994). The sediment samples were dried at 40oC, disintegrated with a ceramic mortar and pestle and sieved through a 2mm sieve. After transfer into ceramic crucibles, the samples were dried in an oven for 80 hours at 105oC (Gayler and Pearson, 2003), weighed, and then burned in a furnace at 400oC for 16 hours. The organic matter content was calculated using the following formula:

)( − ignitedofmassgsampledryofmass gsample )( Organic matter(%) = gsampledryofmass )( ×100 (3)

3.9 Carbonates

Hydrochloric acid (HCl) test and loss on ignition were used to determine the carbonate content in the sediment. The two tests were supposed to complement each other and provide more reliable results.

3.9.1 Hydrochloric acid (HCl) test

The HCl calcium carbonate test was carried out on a small (1-1.5cm3) sediment sample following procedures outlined by (Rowell, 1994). Prior to the test, the sediment was dried in an oven at 40oC and crushed with a mortar and pestle to <2mm.

3.9.2 Loss on ignition

The calcium carbonate content was obtained by returning the samples used for organic matter estimations (section 3.8 above) into the furnace and heating them to 550oC for 2hr. The samples were weighted after the ashing and returned to the oven to be heated to 1000oC for 1 hour (Dean, 1974). The calcium carbonate content was then calculated using the following formula:

550° )( − afterMgignitionCafterM 1000° gignitionC )( Calcium carbonate (%) = 550° gignitionCafterM )( ×100 (4) where M is the mass of the sample.

3.10 Sediment salinity and pH

In preparation for the analysis, 10g of sediment (dried at 40oC and then ground to <2mm) was mixed with 50ml of deionised water at a 1:5 ratio and shaken mechanically for 1 hour to dissolve salts (Rayment and Higginson, 1992; Taylor, 1991). The electrical conductivity (EC) was measured before the pH to avoid contamination from the calomel reference electrode.

3.10.1 Electrical conductivity and salinity

Electrical conductivity (EC), used to estimate the concentration of soluble salts in the sediment, was measured with the CON510 bench conductivity meter with built in temperature sensor. 40

The instrument was calibrated to 1412µS buffer solution/standard. The EC values were used as a proxy for salinity with no further corrections.

3.10.2 pH

The pH was measured using TPS pH Cube instrument with pH sensor and automatic temperature compensation. The instrument was calibrated to buffer solutions of pH 4 and 7, and for most samples also pH 10.

3.11 Gypsum

Due to lack of a single simple, quick, and accurate method for evaluating gypsum content and crystal properties, a combination of methods (outlined below) was used to obtain the best estimate of gypsum content and to assess the comparative reliability and accuracy of each of those methods (discussed in Chapter 4 section 4.9).

3.11.1 Visual description

During the initial sediment description, the visible gypsum was classified into three main categories based on the crystal size (Warren, 1982): i) gypsite: more than 50% of the crystals are silt-sized, ii) gypsarenite: at least 50% of the crystals are sand-sized, iii) selenite: more than 50% of crystals are coarser than 2mm.

The abundance of gypsum was described using a 0-4 scale (section 3.5.4 above). The terms gypcrete and gypsum crust were used interchangeably in this study to describe gypsum-rich units (surface or subterranean) of considerable hardness (i.e. hardly or un-breakable by auger, rock-like).

Where present in considerable quantities, selenite, and on occasions gypsarenite, crystals were extracted from the sediment and analysed under the dissecting microscope to assess selected crystal characteristics such as habit, possible erosion of edges and surfaces, and presence of impurities.

3.11.2 Drying at 105oC

When dried for a prolonged period at high temperature, gypsum crystals lose water from their lattice, which is accompanied by a colour change from clear to milky white (Gayler and Pearson, 2003). The dried crystals then become easy to spot and estimate, especially within the often red/orange or grey sediment of the semi-arid region (Figure 3.4). Thus, relative abundance of gypsum on a 0-4 scale (section 3.5.4 above) was determined for each sample, following drying of the sediment at 105oC in preparation for LOI (section 3.8 above). Accurate determination of the crystal size range was not, however, possible as the samples were affected by grinding. 41

Figure 3.4 Sediment samples after 80 hours of drying at 105oC. The white particles within the red (top) and grey (bottom) sediment are desiccated gypsum crystals.

3.11.3 Portable Infrared Mineral Analyser (PIMA)

The gypsum analysis using PIMA was part of a broader mineral analysis and is described in section 3.12 below.

3.12 Mineral composition

Portable Infrared Mineral Analyser (PIMA) was used to obtain short-wave infrared reflectance (SWIR) from flat surfaces of sediment samples that were dried at 40oC and then ground to <2mm. The technique allowed detection of minerals that contain carbonate and hydroxyl radicals (e.g. clays, amphiboles, some sulphates). The mineral composition was defined by the Spectral Geologist V.4 software.

3.13 Plant identification

The majority of plants were identified by the author using Plants of Western New South Wales (Cunningham et al., 1992) and Flora of New South Wales (Harden, 1990, 1991, 1992, 1993), with the latter reference used as guidance for nomenclature. Selected specimens were identified with the assistance of botanist Don McNair (pers. comm. 2003) and the Botanical Information Section at the Sydney Royal Botanic Gardens (pers. comm. 2000, 2003, and 2004). The identified plant specimens were archived in The Don McNair Herbarium, University of Newcastle, and a digital database of the specimens is available on the attached CD.

3.14 Pollen

3.14.1 Sampling

The open sediment cores were sampled for pollen at 10 cm intervals. However, the intervals were reduced to 5cm or additional opportunistic samples were extracted where the laminations were less than 10cm thick. In augered sediments, the pollen sampling targeted the largest unbroken lumps, to minimise the possibility of contamination. The sample size was 2ml, with exception of three samples from the Lower Bell core G (two 5ml samples from depths of 107 cm and 117 cm, and one 10ml sample from a depth of 144 cm). The increase in sample size was necessitated by the high selenite content, which reduced significantly the volume of non- crystal content within the standard (2ml) sample size.

3.14.2 Pollen extraction from sediment

The arid and semi-arid sediments often contain only small concentrations of pollen, and even those are usually in poor state of preservation (e.g. Gayler, 2000; Horowitz, 1992). Furthermore, the sediment is often highly variable and a combination of clayey, silty, and sandy units within one core is often encountered (as demonstrated by Gayler, 2000). Some researchers proposed to use different methods to extract pollen from different sediment types (e.g. Lentfer and Boyd, 2000). Others (e.g. Smith, 1998), however, argued that variation in the extraction procedures, even as small as the change in the order of the processing steps, can introduce differences in the total concentration of recovered pollen and the variability of pollen types. This, in turn, can affect comparability between samples or sites. Thus, it is important that the techniques used in pollen extraction are the least physically and chemically destructive to minimise further pollen damage and can cope with large sample volumes of often highly variable sediment types to eliminate the need for use of different protocols.

In the initial stages of this research, pollen was extracted using a standard hydrofluoric acid (HF)-based method that was employed at that time at the University of Newcastle palynological laboratory (for details see Gayler, 2000). Unsatisfactory results, i.e. low pollen counts combined with large amounts of non-pollen residue such as gypsum and clays, as well as problems related to processing of larger samples, particularly safety concerns related to handling of large amounts of HF, resulted in a search for a better method. An extensive literature review (incl. Bowdery, 1998; Faegri and Iversen, 1989; Horowitz, 1992; Hunt, 1985; Lentfer and Boyd, 1998, 2000; Parr, 2002), contacts and visits to other palynological laboratories, including the University of Melbourne (M. Cupper, pers. comm. 2000) and the Southern Cross University (J. Parr, pers. comm. 2001-3), and laboratory trials, led to compilation of a protocol (Appendix 3) that allowed better pollen recovery with a reasonably

minimised physical/mechanical (abrasion) and chemical (corrosion, extensive expansion) impact on the individual grains. It also improved laboratory safety through elimination of HF.

In line with the new protocol, the samples were first treated with HCl to remove carbonates and dissolve reference Lycopodium spore tablets, followed by 5% tetra-sodium phyrophosphate (Bates et al., 1978; Batten, 1999; M. Cupper, pers. comm. 2000) to disperse clays before sieving through 250µm mesh (Figure 3.5). The clay separation was done by still settling (based on guidelines by Lentfer et al. (2003) and laboratory notes from Laboratory of Paleoecology, Northern Arizona University, unpubl.) (Figure 3.5) for the following reasons: i) The still settling procedure is generally slower (4.5-8 hours settling cycles repeated for a few days) than sieving through a fine (5-8 µm) mesh (the University of Melbourne protocol). However, trying to wash a large clay-rich sample through a fine mesh can be difficult (even when using dispersant), as the mesh constantly clogs (Lentfer and Boyd, 1999) and the sediment can easily spill and became contaminated by splashes. In addition, the already fragile microfossils may be further fragmented and abraded by forcing them through the mesh and by the intensive and/or prolonged friction between the pollen and the mixing agent, such as glass rod, or sand grains and gypsum crystals present within the sample. Finally, stirring or using a vacuum pump may lead to distortion and increase in mesh aperture size (Lentfer and Boyd, 1999) allowing larger than desired particles, including small pollen, to pass through the mesh. ii) While Lentfer and Boyd (1999) reported satisfactory results from a centrifuge settling method, the still settling method, in spite of being more time consuming, seems to be more appropriate to use with the usually very fragile semi-arid fossils. During centrifuging, the clay-rich sediment usually settles so tightly at the bottom of the tube that intensive vortexing and often also agitation with a glass rod is required to re-suspend it (personal experience). Presence of larger hard and angular particles, such as gypsum crystals, combined with intensive vortexing can cause abrasion of pollen grains (Gayler, 2000). The sediment settled on the bottom of a beaker during still settling remains, on the other hand, relatively loose and is easy to resuspend during preparation for the next settling cycle.

The still settling time was calculated using formulas and information in Lentfer et al. (2003) and a considerable margin was added to allow for the actual slower movement of the grains due to non-spherical shape, protrusions on exine, and pore characteristics (Brush and Brush, 1994). Checks of the water discarded from the settling were carried out to confirm that pollen settled to the bottom of the beaker.

Figure 3.5 Sieving of the sediment through a 250 µm mesh followed by still settling.

The pollen was separated from the silt and fine sand fraction by heavy liquid gravity separation using sodium polytungstate (Hart, 1988b; Munsterman and Kerstholt, 1996). During this procedure, the sample was diluted in a liquid of such density (i.e. specific gravity), that after centrifuging the lighter pollen and phytoliths floated on top of the liquid while the heavier mineral particles settled at the bottom of the sample container.

The specific gravities (SG) of heavy liquid used by different researchers for the separation of pollen fall within a range of SG 1.8-2.1 (Table 3.1) with SG 2.0 (used by this author) the most common. Thus, the SG 2.0 provides a good margin for maximum pollen recovery with minimum interference from the mineral fraction, as the density of pollen and organic constituents is generally below 1.7 and that of mineral fragments (including gypsum) above 2.0. While using higher SG to extract both the pollen and phytolith fraction (Table 3.1; also sections 3.15.1 and 3.15.2) was considered, in the end, a two step separation (i.e. two different SGs) was used to maximise the pollen concentrations by reducing the amount of other material recovered with pollen fraction.

Acetolysis was used not only to remove the cellulose (not a very important issue in semi-arid and arid sediments), but more importantly to improve the appearance, and thus identifiability, of the pollen grains as the procedure increases the size of the grains up to 50% (Brown, 1960), darkens their colour, and may also remove some of the exine, making the morphological features more pronounced and easier to distinguish (Brown, 1960; Reitsma, 1969). The acetolysis also removes the cytoplasmic material from within modern pollen. Since all the University of Newcastle reference pollen is subjected to acetolysis, use of this step in processing of fossil material improves the comparability and identifiability. 45

Table 3.1 Different specific gravities (SG) of heavy liquid used by different researchers in pollen and phytolith extraction.

Source Pollen Phytoliths

Cupper M. (pers. comm., 2000) SG 2.0 Faegri and Iversen (1989) SG 1.96-2.1 Horowitz (1992) SG 2.0 Lab. of Paleoecology, Northern Arizona University (unpubl.) SG 1.9 Macphail (2000) SG 2.0 Thun C. (pers. comm. 2000) SG 1.8-2.0

Bowdery (1998) SG 2.3 Lentfer and Boyd (1999; 2000) SG 2.35

Except for the burning caused by prolonged exposure, the contribution of acetolysis to pollen destruction during processing is not clearly defined. While some researchers (Lentfer and Boyd, 2000) believe it may be significant, others (Faegri and Iversen, 1989; Horowitz, 1992) consider its impact on the exine morphology negligible. The lack of solid evidence for the pollen destruction and the multiple advantages in using the acetolysis (outlined above) decided its incorporation into the pollen processing protocol.

The pollen was mounted in glycerol to ensure good preservation and allow movement of grains on the slide after mounting, thus assisting with identification. After transfer onto the glass slide, the pollen was stained with Saffranin to aid differentiation between pollen and non-pollen fossil and highlight the exine details, especially of the damaged grains (Brown, 1960).

3.14.3 Reference pollen

The modern reference pollen samples were processed using a standard method employed at the palynology lab at the University of Newcastle (Appendix 4). The grains were then photographed and edited images were added to the ‘Arid Zone’ section of The Newcastle Pollen Collection digital database (on the attached CD) (Shimeld et al., 2000).

3.14.4 Pollen analysis

The pollen was counted using a Leitz Diaplan microscope at 500x magnification along transects spaced at 2mm intervals to avoid overlap and ensure better coverage of the slide. The unknown, uncertain, and particularly interesting microfossils were photographed with a digital camera mounted on top of the microscope and later consulted with other palynologists. The images also provided a tool for controlling consistency of identifications.

Due to the high variability of pollen densities between samples, the counts were limited to 250 grains or 500 Lycopodium spores, whichever target was reached first. While using a statistical

analysis to determine the appropriate count number (Shimeld, pers. comm. 2006) was considered, it was decided impractical due to often high variability of pollen composition between different lakes, and between lake and dunes combined with usually substantial reduction in pollen richness and concentrations with depth.

The 250 count included both aquatic and terrestrial pollen as well as unidentifiable (e.g. due to damage) pollen grains (Appendix 5). While most pollen studies focus on terrestrial pollen (e.g. Hopf, 1997; Macphail, 2000), this research considered aquatic pollen to be equally important due to the ephemeral nature of the studied lakes.

3.14.5 Pollen diagrams

The pollen diagrams were compiled using the Psimpoll software and edited with CorelDRAW 9. The diagrams, except for the summary, contain raw pollen counts due to the high variability of total pollen counts between the samples and the presence of many samples with very low counts (e.g. Figure 5.7). Constructing a percentage diagram from such data would have presented distorted picture of the pollen distribution within the core.

The summary diagram was based on six groups-categories: Trees and shrubs, Low shrubs, Herbs and grasses, Aquatics, Unknown type 1, and Other. While the Unknown type 1 group consists of just one pollen type (which, so far, has not been identified), its importance for interpretation of the pollen data has justified its inclusion as a separate category in the summary diagram. The Other category contains pollen types that did not fit any of the other categories (e.g. Loranthaceae) or were not identified (incl. the damaged pollen grains). No summary diagrams was constructed for samples with less than 25 pollen grains, as such low counts were considered to introduce too large bias. For isolated samples with ≥25 grains, the peaks were marked with dots connected to the blocks of the summary diagram with broken lines. Dots were also used on the raw count diagrams to highlight the presence of counts equal or less than two grains that would not be noticeable otherwise.

3.15 Biogenic silica: phytoliths, sponge spicules, and diatoms

The change in pollen extraction procedures from dissolution (HF) to gravity separation (heavy liquid) of mineral silica resulted in preservation of biogenic silica such as phytoliths, sponge spicules, and diatoms. To count them, the previously prepared pollen slides (section 3.14.2) were scanned whole at 2mm intervals (i.e. along 8-11 transects) using a Zeiss Microscope at 400x magnification. A lesser number of transects were analysed on slides with a very high density of phytoliths and/or sponge spicules. The scanning was done in a phase contrast mode to highlight the often ‘worn off/wispy’ biogenic silica on a crowded background. Polarization was used to differentiate between biogenic and mineral silica (Bowdery, 1998).

To provide an indication of the density of the material on the slide, the Lycopodium density was roughly estimated on a 1-5 scale: 1 (rare), 2 (infrequent), 3 (frequent), 4 (common), and 5 (abundant).

3.15.1 Phytolith analysis

The total number of phytoliths counted was highly variable. While the counting of a set number of phytoliths, e.g. 100 by Grave and Kealhofer (1999) or 300 by Thorn (2004), was considered, it was impractical due to the absence, or very limited presence, of phytoliths in many of the samples. Furthermore, only a limited selection of phytolith shapes (listed in Appendix 6) was counted due to limited initial expertise in the field combined with time restrictions. The chosen shapes were the most common in sediment and, at the same time, easy to distinguish from non- phytolith biosilica, some mineral silica, and starches.

As with pollen, only raw counts were graphed due to high variability of total phytolith numbers (e.g. Figure 5.8). The summary diagram was constructed only for samples with a total phytolith count of 25 or more. For isolated samples of ≥25 phytoliths peaks were marked with a dot and connected to blocks of summary diagram with broken line. On raw count diagrams, dots were used to mark the presence of two or less phytoliths within a category.

3.15.2 Additional phytolith extraction

It was recognised that since the SG of phytoliths varies between 1.5-2.3 (Lentfer et al., 2003; Wilding and Drees, 1971), the phytoliths with SG >2.0 are missing from the pollen slides. In an attempt to recover them, a heavy liquid (sodium polytungstate) separation was repeated on the sediment left from pollen processing with sodium polytungstate concentrated to 2.3 SG Table 3.1) (Bowdery, 1998; Thorn, 2004), as described in Appendix 3. However, the large amount of sediment retained at this step (≥0.5ml for majority of the samples) and very low phytolith concentrations made the samples unsuitable for phytolith counts. Time restrictions have prevented pursuit of this issue at present time.

3.15.3 Sponge spicules

Sponge spicules were counted at the same time as the phytoliths (section 3.15.1 above). Their abundance was described by 0-4 scale (section 3.5.4 above). Since all encountered spicules fell within the size range of 100-300µm in length and 3-10µm width, they were considered to originate from freshwater sponges, in contrast to the much larger marine sponge spicules (1- 5mm in length and 100µm across) (Clarke, 2003).

3.15.4 Diatoms

The presence or absence of diatoms was recorded while scanning for phytoliths and sponge spicules. No identification were attempted for reasons summarised in Chapter 4 section 4.12.6.

3.16 Other fossils

3.16.1 Charophytes, macrophytes, and invertebrates

Due to time constrains, a separate sampling and extraction of other micro/macrofossils was not undertaken (except for Palaeolake core D). Instead, the residue recovered from the 250µm sieve during pollen processing (section 3.14.2 above and Appendix 3) was scanned under the dissecting microscope. In addition, where fossils were noticed within core, sediment containing mostly gastropods and dense charophyte layers with preserved calcified oogonia was extracted, washed on a 250µm sieve with distilled water, and dried in a 40oC oven.

The Palaeolake core D was analysed for ostracods by P. De Deckker at the Australian National University in the year 2000. As part of the process, 16 samples were extracted at irregular intervals (at 1, 34, 43, 68, 75, 86.5, 94, 101.5, 111, 128, 133.5, 147, 162, 170, 181 and 190cm), treated with 3% solution of hydrogen peroxide (H2O2) to disperse the sample, sieved trough 150µm mesh, and then dried at 40oC.

The recovered charophyte oospores, macrophytes and some invertebrates were identified by Dr Adriana Garcia (pers. comm. 2006). The ostracods and gastropods were identified by B. Timms and P. DeDeckker (pers. comm. 2006).

3.16.2 Charcoal

The presence of large charcoal particles was noted and recorded during scanning of the >250µm ‘refuse’ from pollen processing (section 3.14.2 above and Appendix 3). While microcharcoal was not counted, its presence was considered during pollen counting by noting slides with marked abundance of charcoal-like particles. The charcoal’s identification was based on comparison with charcoal collected from recently burned sites and descriptions provided by S. Mooney (pers. comm. 2005), Mooney et al. (2001), Enache and Cumming (2006) Tolonen (1986).

Nine samples containing high densities of black particles (5 from dune sediments and 4 from Yandaroo and Willaroo lakes) were analysed for macrocharcoal by S. Mooney (pers. comm. 2006).

3.17 Dating

3.17.1 137Caesium

The 137Caesium (137Cs) measurements were carried out on top sediments of six lakes: Lake Bindegolly, Lake Numalla, Mid Blue Lake, Lake Wombah, Cummeroo Waterhole, and Lake Yandaroo. The sediment of the remaining four lakes: Lake Wyara, Lower Bell Lake, Palaeolake, and Lake Willaroo was assessed to be unsuitable for the analysis due to its high

salinity (Gayler, 2000; Longmore et al., 1986; McHenry and Ritchie, 1977). Depending on the sediment characteristics of the individual cores, including total length of the recovered core, suspected sedimentation rates, and sediment texture and density, the individual samples contained 5, 7.5, or 10 cm depth-sections of sediment. They were analysed by the University of Newcastle High Purity Germanium detector (calibrated with Soil 6 – International Atomic Energy Agency). The 137Cs values and associated errors were calculated according to unpublished guidelines by R. Loughran (pers. comm. 2007).

Two 137Cs reference values were calculated for the two extreme ends of the study area (i.e. north-west at Thargomindah and south east at Bourke) using a regression equation for NSW sites by Elliott et al. (1997):

2 cmmBqY += rX += )89.0(072.03.20)/( (5) where Y is the 137Cs reference value and X is the mean annual precipitation (mm).

Thus, a reference value of 29mBq/cm2 was applied to the lakes located north of the QLD/NSW border and 31mBq/cm2 to lakes south of the border. The calculated reference values are consistent with measured reference values of 23.8 mBq/cm2 for Beaconsfield, north of Paroo/Warrego Region in central Queensland, (Elliott et al., 1996) and 28 and 22 mBq/cm2 for Cobar, south of the study sites in central New South Wales (Elliott et al., 1997). All reference values were decay-corrected to year 2006.

3.17.2 Radiocarbon

Two untreated sediment samples from the Palaeolake core B were submitted for the Accelerator Mass Spectrometer (AMS) radiocarbon (14C) analysis at Australian Institute of Nuclear Science and Engineering (AINSE). The number of samples was determined by the available funding.

3.17.3 Optically Stimulated Luminescence

Twelve samples from nine of the lake cores (Lake Bindegolly, Lake Numalla, Lake Wyara, Mid Blue Lake, Lake Wombah, Cummeroo Waterhole, Lower Bell Lake, Palaeolake, and Yandaroo Lake) were selected for the Optically Stimulated Luminescence dating, undertaken by E. Rhodes at the Australian National University. The sample processing and analysis involves a new, experimental technique and is still ongoing.

Chapter 4

Proxies and methods: discussion

4.1 Introduction

The use of multiple proxies and alternative methods for their extraction and analysis is an important contribution of this research to palaeoenvironmental reconstructions from wetland systems in the arid and semi-arid regions. The research provided useful insights into the relevance and performance of selected proxies and methods. This chapter presents those insights and develops an interpretative framework for the palaeoenvironmental reconstructions in Chapters 5-10.

The chapter is framed around a selection of individual proxies. It presents the proxies’ interpretative value as described by other researchers and a general assessment of their contribution to this study. Since the value of recovered proxy data is often determined by the interaction of effectiveness, efficiency, and limitations related to the methods used in its acquisition (through abundance, quality, and accuracy), their evaluation is also included in this discussion. Where relevant, the chapter highlights areas that would benefit from additional experimental work and improvement of reference collections. Finally, it provides brief descriptions of other methods and aspects of the discussed proxies that have potential to contribute valuable information to this type of project, but were not utilised in this study due to resource limitations.

4.2 Sediment colour

Sediment colour can contribute significant information about the sediment’s mineral content as well as processes that affected it, including their intensity and duration, and the diversity of the depositional environment (e.g. Gale and Hoare, 1991; Myrow, 1990; Schwertmann, 1993; Taylor, 1982). Colour is used widely in interpretation of sediments from both aquatic (Lyle, 1983) and dry environments (Callen and Farrand, 1987; Walden et al., 2000; Wasson, 1983).

The sediment colour is often a result of complex interactions between different processes and agents that are still not fully understood. The primary controllers of colour are the ferrous (Fe3+) to ferric (Fe2+) iron ratio responsible for the green/grey to red spectrum and organic carbon content influencing the grey to black spectrum (Lyle, 1983; Myrow, 1990; Schulze et al., 1993). 51

The Fe3+ to Fe2+ ratio can, in turn, be indirectly and directly affected by a variety of physical and biological factors such as grain size (through its relation to specific source minerals and control of permeability), oxidation potential, sedimentation rate (although its effect is still under discussion), groundwater characteristic and migration patterns, sulphate availability, bacterial activity, plant root activity, pyrite content, and the thickness and iron content of grain coatings (Blodgett et al., 1993; Chivas et al., 1986; Gale and Hoare, 1991; Myrow, 1990; Richardson and Daniels, 1993; Taylor, 1982; Wasson, 1983).

As a summary relevant to this study, the high Fe3+/Fe2+ ratio results in red colours and low Fe3+/Fe2+ ratio in greys, with a range of colours between them (Myrow, 1990). The enhancement in Fe3+can result from oxidation of Fe2+, for example, during pedogenic processes, exposure of aquatic sediments to air upon lake drying, recharge into the lake of oxygen rich subsurface waters, or contact with oxidised groundwater (Chivas et al., 1986; Macumber, 1991; Richardson and Daniels, 1993; Schwertmann, 1993). Under anaerobic conditions Fe3+ can be easily reduced to Fe2+ by anaerobic bacteria (Blodgett et al., 1993; Richardson and Daniels, 1993). The bacteria, however, requires carbon (i.e. organic matter) as an energy source for iron reduction (Richardson and Daniels, 1993), thus increasing the chance of preservation of red sediments (at the time of deposition) in generally organic-poor arid and semi-arid environments. The Fe3+ reduction is also slowed down in fine pores of the sediment, thus preserving red mottling that is a remnant of former oxidising environments (Richardson and Daniels, 1993).

Furthermore, Twidale and Wopfner (1990) observed that the intensity of the sand grain colour depends on the time it remains in an aeolian environment, i.e. generally the longer a grain remains within a terrestrial dune system, the redder it becomes as it builds up its crust of iron oxide. However, the colour of a red dune sand encroaching on a channel can lighten as it mixes with the well-washed sand of the channel.

The redness intensity has proven a useful tool in this study as one of the primary indicators of phases of aeolian (red) deposition within fluvial (grey) sediments, of soil formation within an exposed lake basin (red), and, finally, of sediment sourcing areas (e.g. by signalling significant runoff erosion of the surrounding red sandplain or red sand dune mobility). The ‘original’ oxidised red colour of deposited sediments appears to preserve well, even after return of reducing conditions associated with full lake levels, making it an useful proxy.

The colour analysis in this study was limited to broad colour descriptions based on a combination of Munsell chart categories and colour photographs. The important contribution of sediment colour to this research justifies future pursuit of more advanced quantitative methods of colour analysis, such as redness ratings (Harvey et al., 2003) or more complex stereoscopic triaxial scatterplots (Wells, 2002; Wells et al., 2002). This would facilitate the use of colour in statistical analysis and zonations.

4.3 Bioturbation

The bioturbation features, such as roots and faunal burrows, can contribute valuable palaeoenvironmental information about pedogenic and aquatic processes (Behl and Kennett, 1996; O'Geen and Busacca, 2001), and are of particular interest in lakes oscillating between wet and dry conditions. The fossil roots themselves can be preserved in the sediments or their past presence may be indicated by rhizoconcretions (i.e. mineral cementations around the root), replacement cement structures (e.g. of calcite, silica, or other), or downward and laterally branching mottling (Blodgett et al., 1993; Pye, 1983).

The mottling can be either reddish within grey sediments or greyish within red sediments. The red mottles are formed by iron precipitation along root tunnels as oxygen is delivered to the anoxic/submerged sediments by the roots of wetland plants (Richardson and Daniels, 1993). In other cases, the plant roots can translocate the iron from the root’s vicinity and carry it upward in the plant, creating iron depletion zones around the roots characterised by lower redness/increased greyness (Richardson and Daniels, 1993). Plant roots can also oxygenate near surface soil, contributing to its red hue (Richardson and Daniels, 1993).

In addition, the presence of roots within lake sediments can indicate either freshwater (sediment) conditions or the fall in the saline groundwater table below the root penetration zone within a saline lake basin (Bowler and Teller, 1986).

As with the colour, better quantification of the bioturbation in future studies, e.g. by using a bioturbation index developed by Behl and Kennett (1996), could allow inclusion of this proxy in multivariate statistical analysis.

4.4 Sediment texture and particle size analysis

Particle size analysis can be important in determination of change in environmental conditions as it may give clues about changes in the sediment’s source types and the transport agents and/or energies (Digerfeldt, 1986; Teller et al., 1982). Aeolian type sediments, for example, tend to be coarser (fine sand and silt) than the lacustrine fine silts and clays (Gale and Hoare, 1991). In addition, since within a lake basin the coarsest sediments generally settle close to the lake’s margin and the finer particles close to the lake’s centre, a change in particle size might indicate movement of the shoreline in adjustment to shrinkage or enlargement of the lake (Digerfeldt, 1986; Draper and Jensen, 1976; Teller et al., 1982).

In this study, the field method of sediment texture assessment supplied initial information about the textural changes along the cores and was used in selection of subsamples. The laser diffraction particle size analysis, in turn, supplied a tool to quantify the variation in texture. The procedure holds several advantages over other quantitative methods for particle size determination, such as wet or dry sieving or hydrometer, as it allows much smaller sample sizes 53

(particularly important in cores where the sediment amount is limited), is tidier and quicker, and provides much higher data resolution (Gayler, 2000; Murray, 2002).

However, caution had to be applied in data interpretation in relation to coarsely textured samples and those possibly containing gypsum and clay pellets. In the first instance, there is a tendency of the instrument to overestimate the proportion of coarse non-spherical particles (Campbell, 2003). In the latter case, while the total removal of gypsum prior to the particle size analysis (e.g. Chen et al., 1991) was considered, it was not done due to a combination of large number of samples, unfamiliarity with gypsum removal procedures, and time pressures. Consequently, the >500µm crystals were removed during pre-analysis sieving (Chapter 3 section 3.6.2), but the smaller crystals were retained and contribute to the results. While clay pellets were very likely to occur in the type of dune and lake sediments that were analysed (e.g. Bowler, 1973, 1983; Bowler and Teller, 1986; Magee, 1991; Wasson, 1983), their presence and content was not determined within the scope of this study. In spite of these shortcomings, the method fulfilled the requirements of this study.

4.5 Mineral Magnetic Susceptibility

4.5.1 Palaeoenvironmental significance

The magnetic minerals, including iron oxides, iron sulphides, and manganese oxides usually occur at low concentrations, but are of great significance to palaeoenvironmental studies due to their widespread occurrence and high sensitivity to environmental changes (Dearing, 1999b; Gale and Hoare, 1991; Maher, 1986; Thompson and Oldfield, 1986). The most common, simplest, and cheapest technique for studying magnetic properties is analysis of magnetic susceptibility and frequency dependent susceptibility. These magnetic properties are frequently used to correlate multiple cores from a single lake basin (based on common peaks or patterns) and to identify events such as soil formation, soil erosion in the lake catchment, occurrence of fires, and tephra deposition (Table 4.1) (Dearing, 1999b; Gale and Hoare, 1991; Harvey et al., 2003; Maher, 1998; Maher et al., 2002; Thompson and Morton, 1979; Thompson and Oldfield, 1986; Thouveny et al., 1994).

Magnetic susceptibility is mostly controlled by the mineralogy (usually the ferrimagnetic minerals), although, the size and shape of magnetic grains, their spontaneous magnetisation, and other factors may be also significant (Dearing, 1999a; Gale and Hoare, 1991). Since the formation, modification, and translocation of magnetic components in sediments can be effected and affected by a wide range of agents and conditions (physical, chemical, and biological) (e.g. Table 4.1), the aid of auxiliary non-magnetic characteristics of the sediment (e.g. presence of soil structure, pollen record, loss on ignition analyses) can be also important in the final

interpretation of the data (Dearing, 1999b; Gale and Hoare, 1991; Langereis and Dekkers, 1999; Singer et al., 1996; Thompson and Oldfield, 1986).

Table 4.1 Summary of main factors and processes affecting magnetic susceptibility (MS) and frequency dependent susceptibility (FDS) of sediments and soils. Agent/process Effect on magnetic properties Particle size Some studies suggest direct relationship between the particle size and MS (due to diluting effect) while others show inverse or no relationship (Dearing, 1999b; Gale and Hoare, 1991; Thompson and Morton, 1979) ‘Non-magnetic’ content Dilution of magnetic particles by addition/in situ formation/presence of diamagnetic and weakly magnetic materials such as water, organic matter, carbonates, organic silica, and some clays and minerals decreases MS, but does not affect FDS (Dearing, 1999b; Gale and Hoare, 1991; Maher, 1986; Thompson and Oldfield, 1986) Bacterial and algal activity High bacterial populations in some reducing conditions may produce organic magnetite, significantly contributing to the MS (Gale and Hoare, 1991) Reducing conditions Iron oxides can be subject to dissolution by reduction and potential removal/leaching lowering the MS; occurs mainly in waterlogged sediments and high rainfall (>2000mm p.a.) areas (Dearing, 1999b; Gale and Hoare, 1991; Maher, 1986, 1998; Thompson and Oldfield, 1986); reducing conditions also favour formation of iron sulphides (Thompson and Oldfield, 1986) Oxidising Oxidising conditions and neutral to slightly alkaline pH in lakes and soils conditions/acidity facilitate precipitation of iron from solution (e.g. groundwater) (Gale and Hoare, 1991; Thompson and Oldfield, 1986); Low pH values are unfavourable to magnetite precipitation and inhibit bacterial activity (Maher and Thompson, 1999) Weathering/mineral Incorporation of strongly or weakly magnetic minerals from parent content substrates and/or other sources may result in higher or lower MS values; this may be useful in sediment source determinations (Dearing, 1999a; Dearing, 1999b; Maher, 1986; Maher and Thompson, 1999) Pedogenesis Magnetic enhancement (mainly in secondary ferrimagnetic minerals) of soil surface horizons increases both MS and FDS (~6-15%); the process can be suppressed by waterlogging, leaching, and extreme aridity (Dearing, 1999b; Dearing et al., 1996; Gale and Hoare, 1991; Liu et al., 1999; Maher, 1998) Burning (fires) Burning of soils aids conversion of non-ferrimagnetic oxides to much ‘stronger’ magnetite (later oxidised to maghaemite) thus resulting in enhanced MS and FDS (<12%) values; the vegetation destruction by fire can also contribute to increased topsoil erosion (Dearing et al., 1996; Gale and Hoare, 1991; Maher, 1986, 1998; Maher et al., 2002; Maher and Thompson, 1999; Rummery et al., 1979; Thompson and Oldfield, 1986). Climatic/hydrologic factors Regulate weathering, sediment transport and deposition (aeolian/fluvial) within catchment incl. processes, energies, and frequencies/magnitudes of the transport events, as well as in situ production of non-magnetic minerals (Dearing, 1999b)

The mineral susceptibility of sediment analysed in this study fluctuated to a large degree within a value range of 5-15x10-8m3/kg, with peaks rarely reaching 30x10-8m3/kg or higher (except for Palaeolake and Lake Yandaroo). The ‘background’ values probably reflected the low magnetic susceptibility of the dominantly sedimentary bedrock in the region (Dearing, 1999a; Lu Shenggao, 2000).

The enhanced peaks were often accompanied by moderate frequency dependent susceptibility values. Dearing (1999a) considered frequency dependent susceptibility values as low if they fell below 2%, with the highest values around 10-14%. Values above 14% are very rare and are generally considered to be a result of error or contamination. In this study, the high values were sometimes associated with clear signs of soil development, such as root tunnels. Where no signs of in situ pedogenesis was found, it is likely the deposits contained topsoil eroded from the catchment (Gale and Hoare, 1991).

A well developed, older soil might produce a horizon specific vertical profile of magnetic susceptibility and frequency dependent susceptibility that is shaped by leaching of magnetic particles from the upper A horizon and their deposition at lower depths within the B horizon (Maher, 1986). The process was clearly visible in the red sand dune sediments on the margins of the Palaeolake (Chapter 9 section 9.3.2.5).

Magnetic susceptibility might also be affected by the haematite, occurring as part of grain’s red coating, and, possibly also, goethite responsible for yellowish brown to red coatings (Blodgett et al., 1993; Thompson and Oldfield, 1986; Walden et al., 2000). Both minerals are of moderate magnetic susceptibility, which can become important in environments with limited content of the stronger ferrimagnetic minerals such as magnetite and maghaemite (Dearing, 1999a). The redness of the Paroo/Warrego sandplains is evidence of their widespread presence. The importance of haematite and goethite, however, can not be quantified without additional analysis of the magnetic properties of the regional sediments.

4.5.2 Comparability of the results from the Multi-Sensor Core Logger (MSCL) and Bartington MS2B sensor

The magnetic susceptibility in this study was measured by a Multi-Sensor Core Logger (MSCL) and a Bartington MS2B sensor. The values from the two instruments were not directly comparable due to differences in moisture content at the time of measurement (moist and dry respectively), the difference in the frequencies at which the measurements were taken (Chapter 3 section 3.7), and the depth-thickness of the sediment portion that has contributed to the measured values (e.g. Dearing, 1999a). The general trends shown by the two methods were, however, compatible, although the MSCL data might be somewhat underestimated due to the higher water content (Table 4.1).

Based on the experiences gained through this study, the two methods are best considered as complementary rather than as alternatives, since, aside from the general magnetic susceptibility signal, each of them contributes important data not supplied by the other: i) the MSCL allows quick and high resolution (e.g. 1cm interval) magnetic susceptibility measurements while the sediment was enclosed in the coring pipe, allowing identification of zones of particular interest, including those worthy of dual frequency analysis by the MS2B sensor (point iv below), before opening of the cores; ii) the analysis by MSCL does not cause sediment destruction, leaving pristine material for other analysis; iii) the analysis with MS2B sensor is more time consuming as it requires extraction and pre- treatment of samples before measurements, but it can also manage loose (e.g. augered) materials; and iv) the MS2B sensor, unlike the MSCL, allows measurements at two different frequencies (low and high) and thus determination of the frequency dependent susceptibility values.

4.5.3 Future directions

In summary, the mineral susceptibility analyses used in this project provided important palaeoenvironmental data (Chapters 5-10), highlighting the significance of its inclusion in arid and semi-arid zone research. Its usefulness also justifies additional future effort into its study. The study of magnetic properties of the modern arid and semi-arid soils could establish a reference for the strength of the signal recorded in a lake and contribute to better interpretation of the pedogenic signals within the lake sediments, including the issue of in situ versus allogenic (i.e. soil eroded in the catchment) origins. Furthermore, complementary techniques, such as anhysteric remanence (ARM), isothermal remanence (IRM), and analysis of separated magnetic grains by microscopy and X-ray diffraction, could supply additional information about the magnetic properties and identity of magnetic minerals, providing additional insight into the origins of the sediment’s magnetism (e.g. Han and Jiang, 1999; Hounslow and Maher, 1999; Lees, 1999; Maher, 1998; Maher et al., 2002; Maher and Thompson, 1999; Oldfield, 1999; Phartiyal et al., 2003; Thompson and Oldfield, 1986).

4.6 Organic matter

4.6.1 Selected palaeoenvironmental implications

Most of the organic matter in saline lakes is produced by phytoplanktonic, phytobenthonic, and bacterial organisms at brackish salinities or during freshening phases, often characterised by high alkalinity (Warren, 1986). As the salinity increases, many of the organisms die and sink to the bottom of the lake (Warren, 1986). Within oxygen-rich sediments, the deposited organic matter can be greatly reduced by grazing metazoa (Warren, 1986). The preservation of organic

matter is promoted, on the other hand, under anoxic (e.g. high water level) or hypersaline conditions, which significantly suppress the activity of aerobic bacteria (Warren, 1986). While organic matter content can be reduced by an increase in sedimentation rates, the faster burial rate can also contribute to its better preservation as its residence time within the well-oxidised surface zone decreases (Myrow, 1990).

The organic matter content of the Paroo/Warrego sediments is generally low and rarely exceeds 5%. It is so far uncertain to what extent the low values are caused by the relative organic poverty of the semi-arid depositional environments (compared to temperate humid wetlands) and to what extent the responsibility rests with exposure to oxygen during low water level and dry stages.

4.6.2 Loss on ignition estimation

The loss on ignition (LOI) is one of the most widely used and easiest methods for estimating organic matter content (Heiri et al., 2001). Interpretation of the results, however, needs to consider the mineral content of the samples, since loss of adsorbed (surface and interlayer) water and constitutional OH ions can inflate the measured values (Deer et al., 1992; Heiri et al., 2001). The potential bias is thus likely to be highest in sediments poor in organic matter and rich in minerals such as clays and gypsum.

The dehydration of clays is, generally, the most common non-organic and non-carbonate contributor to the LOI weight loss value (Dean, 1974) and list of clays relevant to this project (i.e. detected by PIMA) as well as the temperature ranges of their weight loss are shown in Table 4.2. While their effect on the results is hard to quantify, it is encouraging that the weight loss peaks of the most common of the Paroo/Warrego Region clays, i.e. halloysite and kaolinite, occur outside the temperature range of the organic matter results. The less common montmorillonite has the highest potential to influence the results, however, its exact contribution is hard to quantify.

Table 4.2 The clay types recorded in the sediments from the Paroo/Warrego Region and temperatures at which they achieve peak weight losses. Weight loss temperature Clay type Huang and Jian (1998) Deer et al. (1992) McMeekin (1985)

Halloysite 110oC and 560oC 110oC with gradual loss 200-700oC until 400oC Kaolinite 560oC 400-525oC 450-550oC Montmorillonite 130-200oC 100-250oC and 500-700oC 200-700oC

Illite gradual loss up to ~1000oC with 110oC (with minor loss up 200-700oC slight peaks at ~100oC and 530oC to 350oC) and 350-600oC

It was also noticed early in this research that gypsum can substantially increase the measured weight loss values if no modifications are applied to the standard methodology, such as, for example, great extension of the drying time at 105oC (Gayler and Pearson, 2003 and Chapter 3 section 3.8). However, since the reported potential temperature range of gypsum water loss spans from 100 up to 150oC (Neumann, 1977) or 170oC (Huang and Jian, 1998), the full evaluation of the effectiveness of the described modification (i.e. determination if additional weight losses occur with rise in temperature up to 150oC) requires further research. An increase in the pre-LOI drying temperature might also decrease the contribution to the target weight loss from water released by some clay types.

Finally, there is lack of consistency between sources about the temperatures at which most of the organic matter related weight loss occurs. Heiri et al. (2001), for example, associates it with reactions occurring at 500-550oC, while Ye (1998) believes that most of it occurs between 300- 400oC. While further experimental work might be required to resolve the issue, this study employs the 400oC temperature limit (Chapter 3 section 3.8) to minimise the weight loss contribution from the mostly clay-rich sediments.

4.7 Carbonates

4.7.1 Palaeoenvironmental significance

Increased carbonate content can occur within lake sediments in association with gypsum-clay couplets or buried soil horizons (Bowler and Teller, 1986). In ephemeral lakes, the carbonate precipitates on top of the flood-delivered clays, but before gypsum crystal formation and deposition (Chapter 2 section 2.5.4) (Bowler and Teller, 1986; Warren, 1982). The lack of carbonate layers might indicate suppression of biological carbonate production by persistence of brine salinities that exceed the biota tolerance levels at all times, i.e. even during the freshening phase (Magee, 1991). Their absence can also stem from calcium deficits resulting from its preferential lockup within gypsum compounds (Teller et al., 1982).

The Paroo/Warrego lake sediments have preserved no distinct carbonate layers. The carbonate content was relatively low, remaining mostly below 5%. Higher values were generally limited to surface sediments, buried palaeosols (e.g. Mid Blue Lake, Lake Wombah), and old lacustrine sediments of the large freshwater Lake Numalla.

4.7.2 Comparison of quantification methods

The comparison of results from two standard methods for carbonate content quantification, i.e. loss on ignition (LOI) and HCl test (Chapter 3 sections 3.9.1 and 3.9.2), has shown large discrepancies with an overall correlation value of r = 0.165 (p = 0.01) (Figure 4.1 and Table 4.3). For individual sets of samples (i.e. cores) the correlations (r) ranged from -0.76 to 0.96.

Carbonates 12

6 LOI (%) LOI 4

0 0 2 4 6 8 10 12 HCl test (%)

Figure 4.1 Comparison between carbonate content (%) estimated by LOI and HCl test for all samples presented in this thesis.

Table 4.3 Comparison between carbonate content estimates (%) by LOI method and HCl test. The LOI values were rounded to match the categories used in HCl test. The analysis includes all sample sets presented in this thesis.

CARBONATES HCl test: B (%) 0.1 0.3 0.5 0.75 1 1.5 2 3.5 5 7.5 10 0.1 2 1 0.3 0.5 3 2 1 0.75 511 1 9516 2 21 2 1.5 1831033 3 7453 2 32 2 18 3 2 6 3326 3.5 2316 1 152 Loss on ignition: A 5 6311 1 7.5 2 10 1 A=B: 16 (6.6%) A>B: 177 (72.8%) A

The reasons behind such large variability between sets are unclear. However, the presence of halite and clays, particularly kaolinite, halloysite, and illite, is likely to be responsible for some of the differences, i.e. those related to the overestimation by LOI (Table 4.3), as they contribute to the sample’s weight loss in the temperature range between 550-1000oC (Table 4.2) (Dean, 1974; Huang and Jian, 1998). While the impact of kaolinite and halloysite can be excluded by pre-treatment of the samples through heating to about 600oC prior to the carbonate analysis, the contributions of illite and halite are much harder to separate, as part of their weight loss temperature range overlaps that of carbonates (Huang and Jian, 1998). 60

The HCl test is less precise (Chapter 3 sections 3.9.1) and can be somewhat problematic, particularly in assessment of the visual and audible effects in the lower range of the scale (i.e. below 2%). The most puzzling samples are, however, those that react violently with HCl, but produce little weight loss during ignition. The resolution of this issue, requires further research and possibly the use of other techniques such as X-Ray Diffraction analysis.

4.8 Sediment salinity and pH

4.8.1 General comments

The pH of depositional environments and sediments is important due to its role in fossil preservation and diagenetic processes. An increase in pH values can also facilitate the deposition of metal ions, including iron (Macumber, 1991), possibly affecting the magnetic properties of the sediment.

Generally, lower pH values occur within near surface sediments (i.e. in the oxidation zone) and increase with depth (Macumber, 1991). One of the main factors controlling the pH values is organic matter as decomposition of calcium-rich plants results in higher pH (~7), while that of calcium-poor organisms leads to a substantial increase in acidity (Gale and Hoare, 1991).

In saline environments, addition of fresh water can facilitate ion exchange of H+ in the solution with Na+ from the surface of the sediment matrix, leading to a relative increase in OH- concentrations, and thus increased pH (Blume et al., 2002). The fresh water influx can further exaggerate the pH by dissolution of carbonates (Blume et al., 2002).

In Australian saline lakes and playas the brine pH is often close to neutral (Herczeg and Lyons, 1991) or acid (Long et al., 1992). In the Paroo/Warrego region the lake brine tends to be, however, more alkaline, generally ranging between 8 and 10 (Chapters 5-10). This might indicate a difference in the source of water responsible for lake filling. Herczeg and Lyons (1991), for example, obtained higher pH values from concentrated riverine water (‘rock weathering’ source) than the concentrated rainwater (atmospheric source), which is the dominant source of water in neutral to acid lakes.

Furthermore, Long et al. (1992) suggested that high concentrations of HCO3 ions result in high alkalinity. In the Paroo/Warrego case, however, there was no noticeable difference in alkalinity

between lakes dominated by Na and HCO3 ions (i.e. Lake Willaroo and Lake Yandaroo) and those dominated by Na+ and Cl- ions (others).

4.8.2 Sediment permeability (relationship with pH and salinity changes)

The potential for movement of water and solutes can be significant for preservation of sediment’s depositional characteristics and fossils, particularly in the semi-arid environments, which can be affected by substantial vertical and horizontal groundwater movement

accompanied by changing chemistry. The groundwater movement is to a large extent regulated by sediment permeability, which in turn can be controlled by salinity. High salt concentration results in strong attraction between the sediment matrix and the fine particles attached to its surface (Blume et al., 2002). Lowering of the salinity (e.g. following influx of fresh water) weakens the bonds and allows downward movement of fine particles from coarse to fine textured sediment units, leading to the clogging of pore spaces (Blume et al., 2002). The permeability of the coarsely textured zones, on the other hand, increases. The mobility of the fine particles and the consequent changes in sediment permeability can be further enhanced by an increase in pH (Blume et al., 2002).

4.8.3 Salinity (electrical conductivity): impacts of gypsum and particle size

Gypsum presence can affect the accuracy of the soluble salt estimations (i.e. electrical conductivity measurements) because, if present in large quantities, it does not fully dissolve in the standard 1:5 suspension (Rayment and Higginson, 1992). In addition, as the salts attach more readily to clay particles than sand (Taylor, 1991), the particle size of the sediment might be accountable for some (so far not quantified) of the within-core salinity variations.

4.9 Gypsum

4.9.1 Habit, location, and orientation within the matrix and their palaeoenvironmental implications

Gypsum is a very common mineral in arid and semi-arid lakes, occurring within lake as well as lunette sediments. Since precipitation, shaping, positioning, and modification/preservation of the crystals within the sediments is closely linked to environmental conditions and processes, they can provide valuable information about past changes (Aref, 2003; Bowler and Teller, 1986; Buck and Van Hoesen, 2002; Chen et al., 1991; Macumber, 1992; Magee, 1991; Teller et al., 1982; Warren, 1982).

It is generally accepted that gypsum precipitating from pure supersaturated aqueous solution assumes mostly acicular (needle shaped) prismatic to equant habit (Edinger, 1973; Magee, 1991; Teller et al., 1982). Thus, this form is often linked with formation at the brine surface or at the sediment-water interface once the lake water reaches gypsum supersaturation state (Magee, 1991).

The presence of different ions (e.g. Na+) and/or organic matter favours growth of pyramidal crystals as it depresses the development of the c-axis (Aref, 2003; Bowler and Teller, 1986; Buck and Van Hoesen, 2002; Cody, 1979; Magee, 1991). Since the lake muds are believed to contain a higher concentration of those ions than the lake waters, the occurrence of pyramidal forms is often attributed to subsurface precipitation from groundwater, which was concentrated

to the saturation levels through evaporation at the capillary fringe (Bowler and Teller, 1986; Magee, 1991; Teller et al., 1982). As their location within the sediment is more closely related to the depth at which the groundwater reached saturation levels with respect to gypsum, than with the lake water levels, their interpretation can be difficult.

Macumber (1991) and Cody (1979) link gypsum crystal habit development at the time of depositions also with pH, as low pH tends to produce prismatic forms and high pH results in lenticular shapes. Yet another explanation of the formation of lenticular habits was supplied by Warren (1982), who suggested that the influx of fresh water into a lake basin can result in partial re-dissolution of the prismatic crystals and formation of lenticular forms.

An experiment by Cody and Cody (1988) related the formation of single lenticular crystals to high temperatures and low organic acid concentrations (but still higher organic levels than those associated with prismatic forms). The crystal complexity, including the degree of twinning, was shown to increase with the rise in organic acid content. In addition, the higher temperatures were more likely to produce short cubic prisms while elongated crystals formed at lower temperatures. Furthermore, the presence of halite was found to favour formation of larger single crystals rather than multiple small ones (Cody, 1979; Cody and Cody, 1988).

Very high concentrations of organic compounds can lead to development of gypsum rosettes (Magee, 1991). Although the exact mechanisms of their formation are unknown, they are believed to precipitate from the meteoric water (i.e. groundwater of atmospheric origin) linking their presence to pedogenic processes (Magee, 1991). Furthermore, higher temperatures tend to produce larger and better developed rosettes (Magee, 1991).

One of the most important features of the gypsum record is the occurrence of laminated prismatic gypsum-clay couplets, which indicate strong fluctuation in climatic conditions on seasonal or periodic scale (Bowler and Teller, 1986; Magee, 1991; Warren, 1982). The rapidity of the environmental changes can be deduced from the crystal size and presence of inclusions, with larger, purer crystals representing more constant and slower growth conditions, that is slower changes in salinity (Aref, 2003; Edinger, 1973; Magee, 1991; Warren, 1982). Crystal orientation, sorting, and roundness are, in turn, indicative of in situ formation or transportation/modification (Bowler and Teller, 1986; Magee, 1991; Warren, 1982).

Although the interpretation of gypsum features can be complex and, in some aspects, ambiguous (such as the formation of lenticular form mentioned above), gypsum’s wide-spread occurrence and high interpretative potential makes it worth pursuing. Its value is well illustrated by results in this study (Lake Wyara, Mid Blue Lake, Lower Bell Lake, and Palaeolake in Chapters 6, 7, and 9 respectively), since in spite of the preliminary nature of the analysis, the gypsum derived information contributed significantly to the reconstruction of past changes in environmental conditions. 63

In hindsight, while some of the gypsum characteristics mentioned above might be described by visual assessment of the core and of the extracted (and cleaned) crystals, preparation and analysis of thin sections might be beneficial in future studies as it provides representative sediment samples that can be easily examined under the microscope and are readily accessible for study over longer periods (in contrast to the whole cores) (Bowler and Teller, 1986; Chen et al., 1991; Magee, 1991).

4.9.2 Content quantification

The gypsum content variation within a sediment core can also contribute to the palaeoenvironmental interpretation and, in most cases, the accuracy achieved by visual quantification seems to be adequate. The pursuit of more accurate (but also more elaborate) 2- methods, such as the chemical analysis of SO4 content by Chen et al. (1991) and Magee (1991), might thus be of little additional contributory value in proxy studies.

This study has used two methods to visually quantify gypsum content: one was based on assessment of freshly opened core and the other on samples dried at 105oC prior to loss on ignition analysis (Chapter 3 sections 3.11.1 and 3.11.2). The overall correlation between the methods was strong (r = 0.851, p<0.001) and for most sample sets the differences were small (Table 4.4), although, the estimation from the 105oC drying tended to be, in many cases, slightly higher, probably due to improved gypsum visibility.

Table 4.4 Comparison between two methods for visual estimation of gypsum content: from fresh core (A) and from sediments dried at 105oC (B). The analysis excluded sample sets for which no gypsum was recorded and Palaeolake core B for which the gypsum estimate from fresh core was based on 0-2 scale. The analysed sets included: L. Bindegolly core 4; L. Wyara core 2; Mid Blue L.: core p1, clay dune auger hole, swale auger hole, and gypsum lunette auger hole; L. Wombah core 3; Lower Bell L. core G; Palaeolake: core K, northern red sand dune auger hole, and eastern gypsum lunette auger hole.

o GYPSUM after drying at 105 C (0-4): B ESTIMATE 0 0.5 1 1.5 2 2.5 3 3.5 4 0 75 7651 0.5 3 5 11 1 541 11 1.5 111 2 135 11 2.5 1 11 3 125 4 fresh core (0-4): A 3.5 241 4 2 A=B: 95 (61.7%) A>B: 27 (17.5%) A

In spite of these generally satisfactory results, neither of the methods was free from minor problems. Within the ‘fresh’ sediments, large quartz grains within the clay matrix can be easily confused with gypsum, leading to overestimations. Conversely, while the gypsum stands out clearly in the dried sediments, its white colour may be hard to distinguish from carbonates. Also, grinding and sieving of the sediment prior to the LOI analysis was likely to affect both gypsum size and content within those samples through pulverisation of some crystals and exclusion of those larger than 2mm in diameter. Thus it seems beneficial for both methods to be cross-checked either by each other or by other methods, such as Short-Wave Infrared Reflectance (SWIR) analysis and/or HCl test for carbonates. The SWIR data can be particularly useful, as in addition to confirming gypsum presence, it supplies limited information about its (relative) content based on the strength of the signature.

4.10 Sulphides

Sulphides, such as pyrite (FeS2) and monosulphides (FeS), tend to accumulate on sites with 2- increased concentrations of sulphate (SO4 ), presence of labile carbon, and anoxic conditions that facilitate sulphate reduction by bacteria (Lamontagne et al., 2004; Teller et al., 1982). Those conditions are found within near-surface sediments of many inland saline lakes and they occur at least periodically (Lamontagne et al., 2004).

The presence of sulphides has two main potential impacts on proxies and methods employed by this study: the gypsum content and characteristics, and the SWIR data. Since gypsum dissolution is often the main source of sulphates (Machel, 2001), its formation and preservation within the sulphide zone can be severely limited or even completely halted (Magee, 1991; Teller et al., 1982). Under such conditions, lake surface gypsum deposits were preserved only when the rate of crystal production exceeded the destruction by sulphate reducing bacteria and/or if the bacterial activity was suppressed, for instance, by oxidising conditions (Magee, 1991; Teller et al., 1982). While the bacteria cannot metabolise (grow) during oxygenated low water and dry lake stages, they can successfully survive until the next wet episode (Teller et al., 1982). The highest bacteria concentrations occur usually within the top 20cm and their activity zone moves up as the sediment accumulates and food becomes exhausted at lower depths (Teller et al., 1982).

Of the lakes cored in this study, four (Lakes Bindegolly, Wyara, Wombah, and Palaeolake) supported clear, black to dark grey, sulphide units (Figures 5.6, 6.3, 7.34, and 9.20), however, the deposits were relatively shallow (<20cm). Sulphide traces were also observed in the top 3cm in Lower Bell Lake, but were to a large extent affected by oxidation caused by the increasing exposure of the sediments to air as the lake was drying (Figure 9.3). The presence of sulphide units in these lakes might provide at least a partial explanation for the absence of

prismatic gypsum forms that were encountered in other Australian inland lakes (e.g. Magee, 1991; Teller et al., 1982).

The presence of sulphides can also lower a sample’s reflectance and weaken the spectral absorption features of minerals within that sample during infra-red mineral analysis (Pontual et al., 1997). The effect was clearly visible in this study as the units dominated by sulphides generally produced flatter reflectance spectra than others, including surface sediments of fresher lakes (e.g. Figures 5.10, 6.8, 6.16, 7.38, 8.8, 9.27, 10.8, and 10.15). However, to fully evaluate the impact of sulphides on the infrared spectrometry data, their preservation/destruction during sample drying prior to the analysis needs to be further evaluated.

4.11 Mineral composition

4.11.1 Clay minerals

The presence of certain clay types within sediments, while often resulting from complex sourcing/weathering, transporting and diagenetic processes (Deer et al., 1992; Velde, 1985, 1992), can provide some general clues about environmental conditions at the time of their formation and deposition. The smectites (incl. montmorillonite), for example, often dominate in semi-arid (contrasted seasons and low rainfall) and dry tropical (high temperatures, low seasonal variability, and low rainfall) environments (Velde, 1992). The smectites can be decomposed by strong leaching into kaolinite or halloysite (a hydrated form of kaolinite) (McMeekin, 1985).

Detrital illite and kaolinite dominate around margins of ephemeral shallow lakes while palygorskite are found further from the shore (Velde, 1992). Sepiolite, in turn, tends to dominate the sediments at the lake’s centre (Velde, 1992). Kaolinites are also the dominant clay in most sandstones (Velde, 1985). The kaolinite and illite are, for example, the dominant clay minerals in Lake Tyrrell (Bowler and Teller, 1986).

The lacustrine sediments in the Paroo/Warrego Region are dominated by kaolinite and halloysite (relevant figures and sections in Chapters 5-10), possibly reflecting their location on top of Cretaceous sandstones, siltstones, and claystones (Chapter 2 section 2.4). The less common, but often encountered, montmorillonite is, in turn, consistent with the prevalent semi- arid conditions.

4.11.2 Short-Wave Infrared Reflectance (SWIR)

The spectral reflectance analysis is probably the most widely used technique in determining mineral content and can be also useful in detecting variations in sediment characteristics within a single core and between multiple cores. In the majority of cases, the spectral analysis is carried out with the aid of X-Ray Diffraction (XRD) and/or Scanning Electron Microscopy

(SEM). However, both procedures can involve labour-intensive sample preparation and high costs of analysis. The Short-Wave Infrared Reflectance (SWIR), obtained from the Portable Infrared Mineral Analyser (PIMA), provides an attractive alternative as it allows rapid analysis of sediment. With no or little sample pre-treatment it can be used in the laboratory as well as in the field and the costs are significantly lower.

In spite of some limitations with respect to the type of minerals that can be detected (Chapter 3 section 3.12) (Pontual et al., 1997), the SWIR based mineral content results obtained from PIMA were shown by Velasco et al. (2005) to be comparable to those from XRD analysis. In this study, the technique provided valuable information about gypsum presence and content (as demonstrated in the discussions of gypsum results in Chapters 6, 7, and 9) as well as the clay types. The carbonate presence was not detected, probably due to its relatively small content combined with the tendency for clay to dominate the spectrum in clay-carbonate mixtures (Pontual et al., 1997). The technique also aided differentiation and comparison/correlations between sediment zones (e.g. in Mid Blue Lake).

4.12 Multi-fossil studies

The use of multiple fossil types in palaeoenvironmental reconstruction holds several advantages, particularly in the semi-arid and arid environments. Firstly, they can provide complementary information about vegetation changes (pollen and phytoliths) and rainfall/evaporation regime changes (invertebrates, gastropods, ostracods, aquatic plants, diatoms, and sponges) (Battarbee, 2000; De Deckker, 1982a). The latter group is particularly significant in ephemeral lakes that can respond to changes in precipitation-evaporation balances through transition from fresh to hypersaline stages, resulting in a succession of fresh to salt- tolerant organisms. Secondly, as shown in the latter part of this section, different lakes and dunes can preserve different types of fossils (e.g. Mid Blue Lake and Palaeolake). Hence, the use of multiple lakes and multiple fossils results in a more complete picture of the local and regional changes.

4.12.1 Vegetation proxies: pollen and phytoliths – general comments

The pollen and phytoliths are the two most common types of microfossils used for vegetation reconstructions. They can be used as alternatives and, until recently, pollen was favoured mostly by Quaternary scientists, while archaeologists preferred to work with phytoliths. However, since the fossils display differences in several characteristics from source to preservation and the plant groups/taxa that they represent (Table 4.5), their complementary use might be of great benefit. This was demonstrated in the reconstruction of grassland development in South Africa (Scott, 2002). The attractiveness of simultaneous analysis of both

proxies could be further increased by the development of a protocol that allows extraction of both fossils at the same time, thus significantly reducing the laboratory workload.

Table 4.5 The main characteristics of pollen and phytoliths relevant to palaeoenvironmental studies. Pollen Phytoliths Source: Adult plants: flowers; Juvenile and adult plants: they form within and between cells in inflorescence, leaves, stalks, roots, rhizomes (Bowdery, 1998); Variability of Generally very low; High as single plant can produce many fossils within a different forms depending on the shape of single plant group the cell or the space between cells in (multiplicity): which it was formed; Indicators of: i) Family to genus groups within the i) Sub-families to genera within Dicotyledonous and Monocotyledonous plants as their Monocotyledonous (non- phytolith morphology is most studied Poaceae) plants; and best known, and their phytoliths ii) Poaceae, usually as one group; more abundant and robust compared to Dicotyledonous plants (Hart and Humphreys, 1997); ii) Some of the shapes are specific to C4 or C3 grasses, although, the C4 plants are generally more prolific phytolith producers (Bowdery, 1998; Mulholland, 1989; Thorn, 2004); iii) Dicotyledonous and Monocotyledonous (non-Poaceae) plants often share phytolith forms and thus are hard to classify into separate groups (Bowdery, 1998; Mulholland, 1989; Thorn, 2004); Primary dispersal i) Wide and long-distance spread of Mostly in situ deposition following their and deposition: pollen from wind pollinated release upon decomposition of dead plant plants, particularly trees (Faegri or parts of the plant (e.g. litter) (Bowdery, and Iversen, 1989; Holmes, 1998); 1994); ii) Short-distance (local) dispersal of pollen from animal pollinated plants and from ground cover plants (deposited within 25m from non-grass plants and 800m from grasses) (Faegri and Iversen, 1989; Holmes, 1994); Secondary Pollen transport by wind, water (incl. Transport of phytoliths or plant fragments distribution: wave action and surface water), by wind (identified in dust samples), insects, animals, etc. (Faegri and water, animals, insects, etc, however, Iversen, 1989; Holmes, 1994); with much lower atmospheric input than pollen (Bowdery, 1998); Preservation: i) It is generally believed that high i) Generally stable in both aerobic and pH (e.g. >6) facilitates pollen anaerobic soils (i.e. preserve well in degradation or complete oxidising environments); destruction (Dimbleby, 1957; ii) Solubility is independent from pH 68

Havinga, 1971); however, Bryant below 9, but increases rapidly above et al. (1994) successfully that value (however, recovered from extracted pollen from soils with soils with pH range 3.5-9.8); pH up to 8.9 and found no iii) Larger surface area, high Si/Al ratios, correlation between pH and and higher occluded water content pollen concentration or grain contribute to higher rate of dissolution degradation in arid regions of (as phytoliths dehydrate with age their southwest America and ; susceptibility to dissolution ii) Tschudy (1969 cited in Bryant et decreases); al., 1994) suggested that iv) Presence of organic material may anaerobic conditions (incl. reduce solubility; enhanced CO2 and hydrogen sulphide levels) rather than pH v) Possibly affected by activities of soil promoted preservation; organisms; iii) In arid environments the (Bartoli and Wilding, 1980; Bowdery, frequency of hydration- 1998; Hart and Humphreys, 1997; dehydration cycles through Mulholland, 1989) expansion (on wetting)- contraction (on drying) and large pressures on organic walls during water transformation from liquid to gas might play an important role in pollen preservation (Bryant et al., 1994) Other factors i) Production levels, e.g. high for i) Established plants can continue wind and low for animal phytolith production even during affecting presence pollinated plants (Faegri and extended periods of drought, masking and concentrations: Iversen, 1989); some arid events; ii) Environmental and habitat ii) Decrease in vegetation densities; stresses reducing flowering; iii) Decrease in water budget might lead iii) Dilution by enhanced to decreased availability of monosilic sedimentation; acid, that is necessary to phytolith formation; iv) Dilution by enhanced sedimentation; v) Change in transport processes, e.g. wind speeds, dust storm frequencies, enhanced ‘mobilisation’ of soil following drought or fire event, decrease in runoff amount and/or energy; (Bowdery, 1998)

The distribution of both, pollen and phytoliths, is influenced by secondary relocation by wind, water, insects and animals (Bowdery, 1998; Faegri and Iversen, 1989). However, because primary deposition of phytoliths occurs mostly in situ, they represent more strongly the local vegetation, while pollen tends to contain a more complex mixture of local, regional, and sometimes extra-regional taxa due to the high variability of its dispersal strategies (Table 4.5). In the Paroo/Warrego Region the complexity of pollen origins (and to lesser extent phytoliths) is further exacerbated by the flatness and general openness of the landscape. Such landscape enables longer distance dispersal than would occur, for example, in a well-forested catchment

(Jacobsen and Bradshaw, 1981), as it generally allows the development of higher wind velocities and contains a lesser number of natural traps. Furthermore, the surface water flow directions can also vary (e.g. Chapter 2 section 2.5.1 and site description of Currawinya and Wombah Lakes in Chapters 6 and 7), making the determinations of pollen sources even more complex.

In oxidising environments, the phytoliths generally tend to preserve better than pollen (Table 4.5) due to their resilient silica composition (Bowdery, 1998; Hart and Humphreys, 1997). This was partly proven by this study, as phytoliths were more frequently encountered than pollen in many of the sediment samples (particularly in lower parts of the sediment cores) in Mid Blue Lake lunettes as well as lake cores from Lower Bell Lake, Palaeolake (coreK), and Lake Willeroo (Table 4.6 and relevant pollen and phytolith graphs in Chapters 7, 9, and 10). However, in some cores (i.e. in Lake Bindegolly, Cummeroo Waterhole, Palaeolake coreB), the phytolith concentrations were comparable to pollen, or even lower as, for example, in the lower half of Lake Wyara core, Lake Yandaroo (augered sediments), and Palaeolake gypsum dune.

The pollen and phytoliths are also characterised by different levels of taxa recognition within different plant groups (Table 4.5). While phytoliths can be used to distinguish between various groups within the Poaceae family including C3 and C4 grasses (Mulholland, 1989; Thorn, 2004), they are much less effective in taxa differentiation within the Monocotyledons (non- Poaceae) and Dicotyledons ‘cluster’. This is where pollen holds a strong advantage as discrete pollen types can be often attributed to specific families and even genera.

4.12.2 ‘Unknown type 1’ pollen/spore

A special case of possibly plant related, and immensely valuable, proxy in this study is ‘Unknown type 1’ pollen/spore (Figure 4.2). Its parent plant seems to favour saline lakes in advanced drying or dry floor lake stages as the grain’s presence is often closely bound to lake sediment units containing gypsum laminae and to lunette deposits (e.g. Mid Blue Lake in Chapter 7 and Palaeolake in Chapter 9). In spite of considerable efforts, the identity of the plant remains a mystery; although, it is possible that it is a spore produced by a moss or liverwort (M. Macphail, pers. comm. 2007) or Callitris sp. (E. Colhoun, pers. comm. 2007).

Table 4.6 Summary of pollen and phytolith counts for lake cores and lake and dune augered sediment. Augered Cored lake sediment lake Augered dune sediment sediment Depth (cm)* Depth (cm)** L. Bindegolly (core 4) (core Bindegolly L. 2) (core Wyara L. L. Numalla (core 1) L. (core Blue p1) Midd 3) (core Wombah L. 4) (core Waterhole Cummeroo Lower L. (core Bell G) Palaeolake (core B) Palaeolake (core K) (core) Yandaroo L. L. Yandaroo L. Willeroo Mid Blue L. (clay dune) L. (swallow) Blue Mid Mid Blue L. (gypsum dune) Palaeolake (red sand dune) Palaeolake (gypsum dune) 0 + + + + + + + + + + + + + + + + + + + + 0 + + + + + + + + + + + + + + 5 + + + + + + + + + + 10 + + + + + + + + + + + + + + + + + + 25 + + + + + + 15 + + + + + + + + 20 + + + + + + + + + + + + - + + + + + + + 50 + + + + + + - - + + 25 + + + + + + + + 30 + + + + + + + + + + + + - - + + + + 75 + + + + + + + + 35 + + + + 40 + + + + - - + + + + + + + + + + - + + + 10 0 + + + + + + + + + + 45 + - + - 50 - - + - - - + + + + - - - + + + 12 5 + + + + 55 + - 60 + - + - + + + + + + + + + + + + 15 0 + + + + + + + + + + 65 - + + + + + + + 70 + + + - + - + + + + - + + + 17 5 + + - + + + + - 75 + + + - + + 80 - - + - - - + - + + + + - + 200 + + + + + + 85 90 + - + - + + - + + + 225 - + + + 95 + + 10 0 + - - - + + + + + + + + 250 - + + + + + + + 10 5 - - - - 110 + - + + + + 275 + + + - 115 + - 12 0 + - + - + + 300 - + + + + + 12 5 + - - + 13 0 + - + + 325 - + 13 5 - - - + 14 0 + - - + 350 + + + + 14 5 - - + + 15 0 - + + + 375 - + + + + + 15 5 + - + - 16 0 - + - - - - 400 - + 16 5 - - + - 17 0 + - - - 425 + + 17 5 18 0 + - - - 450 + + + + + - 18 5 - - 19 0 - - 475 19 5 200 + - - - 500 + - 205 - - 210 + - 525 + + + 250-25 pollen grains counted + 24-1 pollen grains counted - 0 pollen grains counted + 250-25 phytoliths counted + 24-1 phytoliths counted - 0 phytoliths counted * Depth of samples +/- 2cm (to allow inclusion of samples at depths within 2cm of the values specified in the table) ** Depth of samples +/- 10cm (to allow inclusion of samples at depths within 10cm of the values specified in the table)

Figure 4.2 Unknown type 1 pollen/spore.

4.12.3 Pollen: comments on selected aspects of extraction methodology, recovery/preservation trends, and analysis

The new gravity/heavy liquid pollen separation protocol formulated in this study (Chapter 3 section 3.14.2 and Appendix 3) achieved better results from the semi-arid sediments than the standard (HF acid) extraction method used in earlier research from the Paroo Region by Gayler (2000). The new method was able to comfortably handle large sediment samples (particularly important for sediment with low fossil concentrations) and to minimise damage to pollen grains. It also reduced the use and risk of exposure to toxic chemicals.

The method’s effectiveness in pollen extraction was assessed by comparing twin sample sets from Palaeolake processed by the standard HF method in the year 2000 (Gayler, 2000) and the gravity/heavy liquid separation method in 2005/2006. Subsequently it was noticed, that while the earlier method resulted in some damage to Lycopodium spores in gypsum-rich samples (Gayler, 2000), no significant deterioration of the spores was observed in the equivalent samples processed according to the latter protocol that significantly reduced the need of intensive

vortexing (Chapter 3 section 3.14.2). The pollen concentrations and taxa richness were similar for both methods.

In more general terms, the highest pollen concentrations were mostly confined to the near- surface (~30cm) sediments and decreased sharply below that depth (Table 4.6). Exceptions were samples from Lake Wyara and Lake Yandaroo, where reasonable pollen concentrations continued to depths of 90 and 175 cm respectively. The richness of pollen types decreased in all sediment cores with Chenopodiaceae, Asteraceae, Poaceae, and, to a slightly lesser extent, Myrtaceae pollen the most persistent and resistant to damage. These grains were also the most easily identified, even when corroded, due to their highly distinctive morphology. On the other hand, grains such as Liliaceae, Cyperaceae, and Callitris were likely to be underestimated, as, even with only a slight amounts of damage, they tended to be harder to differentiate from the non-pollen components.

4.12.4 Phytoliths: selected comments on the extraction methodology and analysis

The use of the gravity/heavy liquid pollen separation method favoured recovery of large quantities of phytoliths from many of the samples (Table 4.6). The phytolith recovery is even more encouraging when it is considered that the counts were likely to underestimate the true concentrations in the sediment.

The phytolith underestimations resulted from limitations of the extraction technique as well as highly selective counts (Chapter 3 section 3.15.1). Because pollen was the target, the extraction method allowed recovery of the lighter (1.5-2.0 SG) fraction of the phytoliths, but not the heavier fraction (2.0-2.3 SG) (Chapter 3 section 3.15.1-2). Furthermore, the analysis of the recovered phytoliths was limited to a selection of the most frequent and most readily recognisable shapes (Appendix 6). Spheres, for example, were omitted from the count as they can be easily confused with starches or mineral quartz, especially under the light microscope. While the use of the polarized light to differentiate between the bio- and mineral silica was considered (Bowdery, 1998), it was impractical in dealing with large phytolith and sample quantities.

The lack of an adequate reference key (especially for Australian plants) (Bowdery, 1998), time limitations preventing from development of phytolith reference collection for the Paroo/Warrego plants, and the need for development of expertise in the phytolith field contributed to further limitation of the phytolith counts. As a result, the tally included only shapes that very closely matched the published descriptions of the target shape-categories. When any doubt existed, items were omitted from the count. The bilobates were least affected

by this selection bias as their shape is the most distinctive, while shapes like polylobate, rondel, and trichome/hair were found to be more problematic.

The promising phytolith results from this study highlight their potential as a palaeoenvironmental proxy in arid environments and should encourage future efforts in refining the methodologies and procedures driving their recovery and analysis. In the development of the protocol, for example, further experimental work is required to investigate what went wrong during the attempted recovery of the heavier phytoliths (Chapter 3 section 3.15.2), including potential instability of the Tasmanian sodium polytungstate (leading to formation of residue), or presence of gypsum (2.3 SG) or other mineral with specific gravity value between 2.0 and 2.3.

The phytolith recognition, classification, and identification, thus interpretation, could be greatly improved by development of a comprehensive, single source reference database, particularly one targeting Australian and multi-plant (not just grass) taxa with strong support of light microscope images that are easily comparable to those viewed on the slides (e.g. similar to the reference pollen management system in The Newcastle Pollen Database). While 93% of Australian plants are endemic, the relevant published (thus accessible) phytolith references are limited mostly to Bowdery’s (1998) collection of selected taxa from central Australia (mostly drawings and SEM photos), Hart’s (1988a; 1992) selection of SEM images from Pilliga (NSW) and Oxford Falls (near Sydney), and Wallis’ (2003) collection of selected taxa from north-west Australia.

One of the significant findings of those studies was the existence of differences in the phytolith characteristics of grass taxa between Australia and North America (Hart and Humphreys, 1997). In addition, while Piperno (2006) concluded from the American research that Chenopodiaceae did not produce phytoliths, Bowdery (1998) recovered phytoliths belonging to Salsola kali in that family. Furthermore, Wallis (2003) has discovered that the same taxa from a range of families growing in different regions (e.g. Kimberly Region and Central Australia) might produce different phytolith shapes and quantities, with arid zone plants tending to be better producers of phytoliths than their tropical and northern semi-arid counterparts. This highlights a risk of relying on non-Australian reference materials in interpretation of Australian samples and supports the need for better understanding of the Australian phytolith production and links between plant taxa and specific phytolith shapes through more extensive research in this field.

Finally, as the phytolith size can be affected by the availability of monosilic acid, and thus water (Bowdery, 1998), the changes in phytolith size can indicate changes in the water budget. Hence, size classification might be worth undertaking in future phytolith studies.

4.12.5 Sponge spicules

Sponge spicules are needle-like siliceous bodies that act as a skeleton for sponge tissues. Since sponges’ presence in the marine and terrestrial ecosystems is usually quite limited, their presence in sediments and soils is generally minor (Clarke, 2003). However, under some conditions (that are, so far, not well understood) their concentrations can become very high (Clarke, 2003). If identified, they can provide valuable information about water properties at the time of their growth, such as pH and salinity (Gammon et al., 2000; Harrison, 1988; Pennak, 1978).

The presence of sponge spicules in sediment is often reported, especially in studies focusing on phytoliths, but is often underutilised as detailed analyses are rarely undertaken. Most frequently the consideration of their presence is limited to ‘absent/present’ to provide evidence of waterlogged conditions or presence of a semi-permanent surface water (Brewer, 1955; Clarke, 2003), changes between lake and mire/bog conditions (Muller et al., 2006), and differentiation between well and poorly drained soils (Schwandes and Collins, 1994). The high and depth- continuous sponge presence in sediments is generally associated with in situ formation and deposition, while very low numbers are more likely to be deposited from dust (Clarke, 2003).

The sponge spicule (especially freshwater taxa) classification and interpretation was for a long time avoided by palaeoecologists due to scarcity of information about the ecology of modern taxa and a long standing confusion about their taxonomy (Harrison, 1988). While the number of publications on individual modern species from North America as well as Europe, Jordan, and Central America has risen considerably in the last two decades, their popularity with palaeoenvironmental scientists has not increased. In Australia, the only recent taxonomy-based interpretation of fossil spicules, seems to be that by Gammon et al. (2000) and this publication is limited to marine species. An annotated bibliography of Australian fossil sponges was compiled by Pickett (1983), however, since most of the publications are single-taxa works, its detailed review and compilation of comprehensive identification key was beyond the time limits of this study.

In the Paroo/Warrego Region, the freshwater sponge spicules were found in many of the pollen slides (Table 4.7 and relevant biogenic silica diagrams in Chapters 5-10) and in all samples from Lake Willeroo they had by far dominated the content. Similar to phytoliths, their full extraction is uncertain as no information was found about their specific gravity, with the only exception of a report by Schwandes and Collins (1994) about sponge spicule extraction with a heavy liquid concentrated to specific gravity of 2.3. In spite of the uncertain recovery rates, this study has shown that sponge spicules can be preserved in arid and semi-arid lake and lunette/dune sediments and show diversity of form (Figure 4.3). A categorisation with respect to shape and surface features was attempted during the counting, however, arbitrary 75

categorisation combined with the lack of comprehensive reference database, made its use in this research untenable.

Table 4.7 Summary of freshwater sponge spicule occurrence in lake and dune sediments. Augered Cored lake sediment lake Augered dune sediment sediment Depth (cm)* Depth (cm)** L. Bindegolly (core 4) 2) (core Wyara L. L. Numalla (core 1) p1) (core L. Blue Midd 3) (core Wombah L. Cummeroo Waterhole (core4) G) (core L. Bell Lower Palaeolake (core B) Palaeolake (core K) (core) Yandaroo L. L. Yandaroo L. Willeroo dune) (clay L. Mid Blue (swallow) L. Mid Blue dune) (gypsum L. Mid Blue Palaeolake sand (red dune) dune) (gypsum Palaeolake 0 - •• • - • • • • • 0 • + + • - • • 5 • + • • • 10 • • • - • + • • + 25 • + • 15 • ••• 20 ••••• + ••• + 50 • + + • - 25 •• • 30 • + • • • + • • + 75 • + + • 35 + • 40 • + - • • • ••- + 100 • + - + - 45 • - 50 --- • • - • + 125 • + 55 - 60 -- • • • • • + 150 • + • • • 65 - • • + 70 --- • • • + 175 • + • - 75 ••+ 80 --- - -• • 200 + • + 85 90 -- • • • 225 + • 95 - 100 -- ••• • 250 + - • - 105 - • 110 - • • 275 + • 115 - 120 - • • 300 + • + 125 - • 130 - • 325 + 135 - + 140 - • 350 + • 145 - • 150 + • 375 + • - 155 -- 160 • -- 400 + 165 -- 170 -- 425 + 175 180 -- 450 --- 185 - 190 - 475 195 200 -- 500 - 205 - 210 - 525 - + abundant + frequent + considerable • occasional • rare - absent

* Depth of samples +/- 2cm (to allow inclusion of samples at depths within 2cm of the values specified in the table) ** Depth of samples +/- 10cm (to allow inclusion of samples at depths within 10cm of the values specified in the table) 76

Figure 4.3 Selected examples of variability in the freshwater sponge spicule forms.

4.12.6 Diatoms

Diatoms can provide valuable information about the changes in lake water chemistry such as salinity, pH, and nutrients (Clarke, 2003; Gasse, 2002; Gasse et al., 1997; Gell, 1997; Gell et al., 2005; Kashima, 2003; Reid et al., 1995). In this study the diatoms analyses was limited to noting their presence on pollen slides (Chapter 3 section 3.15.4). With the exception of the Palaeolake core K, diatoms have been found only in the relatively modern lake sediments (down to ~ 30cm depth) (Table 4.8 and relevant biogenic silica graphs in Chapter 4) and in only three dune sediment samples (two in the Mid Blue Lake gypsum lunette and one in the Palaeolake red sand dune) contained diatoms. While some diatoms still carried enough surface ornamentation to allow identification, many were worn to only an outline visible in phase contrast mode; generally, their preservation condition deteriorated rapidly with depth.

While diatom recovery in this study was unsatisfactory, the reasons for it are unclear. Further experimental work is needed to determine if their absence is a result of chemical dissolution within the sediments, as reported in extremely alkaline or saline waters, particularly if they were under-saturated with respect to silica (Gasse et al., 1997; Reid et al., 1995; Ryves et al., 2001), or due to the pollen extraction protocol. Hence, diatom recovery using a simpler/less destructive method, both in terms of the procedures as well as the number and type of chemicals

used, needs to be tried (e.g. Gell et al., 1999) before full evaluation of the diatom presence in the arid and semi-arid lacustrine sediments can be made.

Table 4.8 Summary of diatom occurrence in lake and dune sediments. Augered Cored lake sediment lake Augered dune sediment sediment Depth (cm)* Depth (cm)** Depth L. Bindegolly (core 4) (core Bindegolly L. 2) (core Wyara L. L. Numalla 1) (core p1) (core L. Blue Midd 3) (core Wombah L. Cummeroo Waterhole (core 4) G) (core L. Bell Lower Palaeolake B) (core Palaeolake K) (core (core) Yandaroo L. Yandaroo L. L. Willeroo (clay Mid dune) L. Blue Mid (swallow) L. Blue (gypsumMid L. dune) Blue Palaeolake sand dune) (red Palaeolake (gypsum dune) 0 ++ - ++++ - ++ 0 ++ ---+ - 5 - +++ - 10 - + --++ ---25 - + - 15 ---- 20 ----+ -----50 -- -- - 25 -- - 30 - + ---+ ---75 --- - 35 -- 40 ------100 -- -+ - 45 -- 50 ------125 -- 55 - 60 ------150 ----- 65 ---- 70 ------175 -- -- 75 --- 80 ------+ 200 --+ 85 90 -- - --225 -- 95 - 100 ------250 - --- 105 -- 110 ---275 -- 115 - 120 --+ 300 -- - 125 -- 130 --325 - 135 -- 140 -- 350 -- 145 -- 150 --375 -- - 155 -- 160 --- 400 - 165 -- 170 -- 425 - 175 180 -- 450 --- 185 - 190 - 475 195 200 -- 500 - 205 - 210 - 525 - + present - absent * Depth of samples +/- 2cm (to allow inclusion of samples at depths within 2cm of the values specified in the table) ** Depth of samples +/- 10cm (to allow inclusion of samples at depths within 10cm of the values specified in the table)

4.12.7 Fossil 1

Fossil 1 (Figure 4.4) is most likely the remains of a living organism, as its reflective characteristics in phase contrast are comparable to those of biogenic silica such as phytoliths and sponges. The fossil was found in some of the pollen samples from Lake Bindegolly, Lake Wyara, Lake Wombah, and Mid Blue Lake clay and gypsum dunes (Table 4.9 and relevant biogenic silica graphs in Chapters 5, 6, and 7). Its pattern of recorded occurrence is of great interest as it provides a potential link between relatively modern sediments (i.e. environmental conditions) in the already mentioned lakes and lunette formation period (probably post-LGM to Holocene). Its identity, however, remains a mystery, in spite of multiple attempts at identification by the community of palynologists and phytolith analysts.

Figure 4.4 Fossil 1: a – surface view; b – cross-section view.

The only identity suggested, so far, is one offered by P. Gell (pers. comm. 2007), which points to Amphora dubia diatom. While the structure of Fossil 1 is generally too coarse for a diatom, its general shape is not unlike Amphora, a genus typical of lowland, brackish waters. The lack of typical diatom surface ornamentation could be explained by partial dissolution, while the spikes might have been formed by secondary deposition of silica around the diatom valve.

Furthermore, a large, heavily silicified Amphora dubia was reported from the gypsum-rich parts of north-western Victoria (P. Gell, pers. comm. 2007 and Gell, 1997). Its robustness may have facilitated its preferential preservation. The occurrence of A. dubia in gypsum-rich habitats and its tolerance of high salinities shows some similarity to the sediments from which Fossil 1 was recovered, i.e. Lakes Bindegolly, Wyara, and Wombah. A modern diatom study from waterholes along Cooper Creek and Warrego River channels (McGregor et al., 2006) has identified seven species of Amphora, but no A. dubia in particular and no Amphora species have been mentioned in a study of algae/diatoms from the Neales and Diamantina Rivers and Cooper Creek in the Lake Eyre Basin (Costelloe et al., 2005).

Table 4.9 Summary of Fossil 1 occurrence in lake and dune sediments. Augered Cored lake sediment lake Augered dune sediment sediment Depth (cm)* Depth (cm)** Depth L. Bindegolly (core 4) (core Bindegolly L. 2) (core Wyara L. 1) (core Numalla L. p1) (core Midd L. Blue 3) (core Wombah L. Cummeroo Waterhole (core 4) G) (core L. Bell Lower Palaeolake B) (core Palaeolake K) (core (core) Yandaroo L. Yandaroo L. L. Willeroo (clay Mid dune) L. Blue Mid (swallow) L. Blue (gypsumMid L. dune) Blue sandPalaeolake dune) (red Palaeolake (gypsum dune) 0 ------0 --+ ---- 5 ----- 10 - + ------25 --+ 15 ---- 20 + ---+ -----50 -- -- - 25 + -- 30 ++ ------75 --+ - 35 -- 40 - + ------100 -- -+ - 45 -- 50 ------125 -- 55 - 60 ------150 ----- 65 ---- 70 ------175 -- -- 75 --- 80 ------200 --+ 85 90 -- - --225 -- 95 - 100 ------250 - --- 105 -- 110 ---275 -- 115 - 120 ---300 - ++ 125 -- 130 --325 - 135 -- 140 -- 350 -- 145 -- 150 --375 -- - 155 -- 160 --- 400 - 165 -- 170 -- 425 + 175 180 -- 450 --- 185 - 190 - 475 195 200 -- 500 - 205 - 210 - 525 - + present - absent * Depth of samples +/- 2cm (to allow inclusion of samples at depths within 2cm of the values specified in the table) ** Depth of samples +/- 10cm (to allow inclusion of samples at depths within 10cm of the values specified in the table)

Aside from its unusual for a diatom robustness and lack of ornamentation, there are also other issues that need to be resolved before confirmation of the Fossil 1 identity as Amphora. For

example, why only this particular diatom shape and size (approx. 65-100 µm) would be affected by the secondary deposition of silica, while other diatoms, sponge spicules, and phytoliths in those samples remained unaffected.

4.12.8 Macrophytes (seeds) and charophytes (oospores)

Both macrophyte and charophyte taxa can supply information about changes in lake or river water characteristics such as salinity, pH, depth, type of sediment, and wave action or flow regimes (Garcia, 1994, 1999a; Garcia and Chivas, 2004). The macrophytes in this study were represented mostly by seeds. With an exception of a single seed in Palaeolake red sand dune at 360cm depth (probably a contaminant), their occurrence was limited to modern (upper 30cm) sediments (relevant core description figures in Chapters 5-10). Thus they were unlikely to play a major role in palaeoenvironmental reconstruction and, while their presence was recorded, they were not subjected to detailed taxonomic identification.

The charophytes, a group of multicellular green Algae (also known as stoneworts), are often encountered in Australian freshwater as well as saline to hyperhaline (especially ephemeral) lakes, creeks, and rivers (Burne et al., 1980; Garcia, 1994, 1999a; Garcia and Casanova, 2003). They can be easily identified in sediments by their characteristic ovoid female reproductive organ (oogonia) (Figure 4.5), that can be preserved as oospore (the zygote and its resistant walls) or, in calcium-rich environments and within a selected group of taxa, as gyrogonite (a calcified content of sterile spiral cells surrounding the zygote). Their presence in ephemeral lakes is of particular interest to palaeoecologists as temporary wetlands can support higher diversity of charophytes than permanent sites (Casanova et al., 2003). Furthermore, living charophytes can tolerate salinities up to 70g/L with shallow water conditions and in inland lakes around northern Spencer Gulf oogonia were found within gypsum and halite deposits (Burne et al., 1980; Garcia et al., 2002). Their germination requires, however, lower salinities (<53g/L), thus periodic supply of rainwater or relatively fresh groundwater (Burne et al., 1980). In favourable conditions, their regeneration can be quite rapid (under two weeks), even after a few years of desiccation (Burne et al., 1980).

While in this study the gyrogonites were mostly dissolved by hydrochloric acid during the initial stages of pollen processing, a reasonable quantity of oospores was recovered and identified from all of the lake sediment cores, but not dune sediments (Table 4.10 and relevant sediment description figures in Chapters 5-10). A list of the identified charophyte taxa and their habitat requirements are provided in Appendix 7. The recovery has confirmed their possible use in any further arid-zone palaeoenvironmental studies and provided an incentive for their inclusion in development of a better multi-micro/macro-fossil extraction method (section 4.12.10.2 below).

Figure 4.5 Charophyte oogonia.

4.12.9 Invertebrates: gastropods, ostracods, cladocerans, and chironomids

Fossil invertebrates, including gastropods, ostracods, cladocerans, and chironomids (e.g. Battarbee, 2000; De Deckker, 1982a, 1982b, 1988b; De Deckker et al., 1982; Hoffmann, 1988; Larocque, 2001) are frequently used in palaeoenvironmental reconstruction research as they can contribute a wide range of palaeoecological information, such as changes in water chemistry and depth, determination of sediment accumulation rates and compaction ratio, and climate seasonality, including seasonal timing of rainfall and lake filling events (De Deckker, 1988a, 1988b).

The quantity of fossil invertebrates recovered from the Paroo/Warrego lake sediment cores was, however, very limited (Table 4.11 and relevant core description figures in Chapters 5-10). Their scarcity can be due to dissolution of calcium carbonate fossils by hydrochloric acid during the initial stages of pollen processing, damage or exclusion of various fossils in other stages of the pollen extraction, and, finally, poor preservation within the sediments. The poor recovery of ostracods from Palaeolake core D (Chapter 3 section 3.16.1), in which specimens were recovered only from the top of the sediment core, suggests poor preservation, although further work is needed in the region to evaluate the preservation rates in different types of wetlands. The differences in preservation of other fossils in different lakes (e.g. pollen and phytoliths in Table 4.6), the identifiable chironomids of considerable age (probably >15 years old) recovered from Palaeolake core 1, and generally satisfactory ostracod recovery from other Australian

ephemeral lakes, strongly support the inclusion of invertebrates as potential proxies in future studies to assess better their full potential.

Table 4.10 Summary of charophyte oospore occurrence in lake and dune sediments. Augered Cored lake sediment lake Augered dune sediment sediment Depth (cm)* Depth (cm)** Depth L. Bindegolly (core 4) (core Bindegolly L. 2) (core Wyara L. L. Numalla 1) (core p1) (core L. Blue Midd 3) (core Wombah L. Cummeroo Waterhole (core 4) G) (core L. Bell Lower Palaeolake B) (core Palaeolake K) (core (core) Yandaroo L. Yandaroo L. L. Willeroo (clay Mid dune) L. Blue Mid (swallow) L. Blue (gypsumMid L. dune) Blue Palaeolake sand dune) (red Palaeolake (gypsum dune) 0 - +++--+ +++0 + • ----- 5 + • - ++ 10 • + ••--• + - 25 • -- 15 • + • - 20 • + - + - • ---• 50 -- -- - 25 •• •• 30 ••- • --- -• 75 - • -- 35 -- 40 • ------• 100 • • -- - 45 -- 50 ------125 -- 55 - 60 ------150 ----- 65 ---- 70 ------175 -- -- 75 --- 80 ------200 --- 85 90 -- - --225 -- 95 - 100 ------250 - --- 105 -- 110 ---275 -- 115 - 120 ---300 -- - 125 -- 130 --325 - 135 -- 140 -- 350 -- 145 -- 150 --375 -- - 155 -- 160 --- 400 - 165 -- 170 -- 425 - 175 180 - • 450 --- 185 - 190 - 475 195 200 -- 500 - 205 - 210 - 525 - + ≥25 oospores + 24-5 oospores • 4-1 oospores - absent * Depth of samples +/- 2cm (to allow inclusion of samples at depths within 2cm of the values specified in the table) ** Depth of samples +/- 10cm (to allow inclusion of samples at depths within 10cm of the values specified in the table)

Table 4.11 The summary of modern and fossil invertebrate occurrence within lake sediments. (*core collected from Palaeolake in 1998 by S. Pearson, two chironomid species: Eukiefferiella sp. at 122-130cm and Cladotanytarsus sp. at 172-180cm identified by S. Dimitriadis)

Depth (cm) of occurrence Lake cores Gastropods Ostracods Cladocerans Chironomids

L. Bindegolly (core 4) 10-15, 26-28 10 L. Wyara (core 2) 30 30 surface L. Numalla (core 1) surface, 5 Mid Blue L. (core p1) 17-18 17-18 surface L. Wombah (core 3) 5 30 Palaeolake (core 1)* 122-130, 172-180 Palaeolake (core B) surface Palaeolake (core D) 0.5-1.5 Palaeolake (core K) surface L. Yandaroo (core) surface -29 L. Yandaroo (augered surface sediment)

4.12.10 Fossil extraction methods: summary and future progress

One of the biggest time costs in the study of any fossil proxies is their extraction. Very often the standard processing methods target a single fossil, rather than allowing a recovery of multiple proxies with the same method. Thus one of the major issues that any multi-fossil research faces is development of a multi-fossil extraction protocol that can reduce the time spent on the sample processing, conserve more sediment for other analysis (particularly valuable for small diameter sediment cores), and decrease the volumes and costs of chemicals consumed during the extraction.

4.12.10.1 Microfossils including pollen and biogenic silica

While some attempts at simultaneous extraction of pollen and phytoliths were already attempted (Frey, 1955; Lentfer and Boyd, 2000 (for tropical areas), and this study), there is still a great need for additional experimental work. Research is needed to improve the recovery rates of diatoms and sponge spicules, including determination of their specific gravity ranges (to adjust the heavy liquid density) and assessment of the impact of the physical and chemical treatments on their survival. For example, there is an opportunity to resolve the uncertainty about the role of different preservatives on diatom dissolution (Gell et al., 1999).

4.12.10.2 Other micro- and macro-fossils

In this study, the analysis of other micro- and macro-fossils was limited to the material recovered from the 250µm sieve during pollen processing (normally considered as waste and

discarded) and a small number of grab sediment samples from depths at which shells were spotted during core description (Chapter 3 section 3.16.1). While this method provided insight into the type and concentrations of micro/macro-fossils without the need for separate processing (thus proving valuable when short of time), it did not support full micro/macrofossils recovery. Some seeds, shells, and charophyte oospores were damaged (fragmented), lost during sieving (e.g. evidenced by charophyte oospores on pollen slides), or were dissolved during the initial stage of sample treatment (e.g. dissolution of shells and gyrogonites by hydrochloric acid). In future research, their better recovery calls for use of gentler (already existing) extraction methods, and ideally a formulation of a single set of procedures that would allow simultaneous extraction of the different fossils.

4.13 Charcoal

Charcoal is widely used to reconstruct records of fire events (e.g. Cupper, 2005; Kershaw et al., 2007; Mooney et al., 2001; Mooney et al., 2007). Its analysis and interpretation is, however, a subject of an ongoing debate, particularly with respect of the micro- versus macrocharcoal and its interpretation as a proxy of extra-regional or local fire events (Asselin and Payette, 2005; Mooney et al., 2001). In spite of contrary views, an increasing number of researchers are turning from micro- to macrocharcoal analysis for reconstructions of local fire events (Black et al., 2006; Enache and Cumming, 2006; Mooney and Black, 2003; Mooney et al., 2007; Power et al., 2006).

In this research, a few sediment units contained considerable amounts of black particles, however, a preliminary macrocharcoal analysis of nine representative samples by S. Mooney (pers. comm. 2006) failed to confirm charcoal presence in most of them. The only exception was for a sample from 425cm depth in Mid Blue clay lunette, and even in that sample the charcoal was less than 1% of total sample content. Similarly, black (charcoal) particles were present only in low background amounts on the pollen slides and none were discovered in the >250µm fraction retained during pollen sieving.

The results seem to indicate that either charcoal does not preserve well within the dryland sediments, and/or the fires do not produce large amounts of charcoal, and/or the charcoal is lost rather than preserved in sediment deposits. The last two explanations are the most likely in the environmental context of the Paroo/Warrego Region. The scarcity of vegetation during dry periods provides very limited amounts of fire fuel and the low densities prevent fires from spreading (Chapter 2 section 2.7). The highest quantities of fuel are produced following a major wet season or flood, with large fires most likely to occur with the onset of a new dry phase. The charcoal is then left on exposed dry surfaces until the next wet phase, often for a considerable length of time (few months to years). Thus, it is vulnerable to removal by wind and long distance, extra-regional transport. Field experiments are needed to test this hypothesis. 85

4.14 Continuity of records, accumulation/deflation rates, and dating

4.14.1 Accumulation and deflation processes

The arid and semi-arid landscapes are shaped by the complex interaction between accumulation and deflation events in both terrestrial and lacustrine environments (Bullard and McTainsh, 2003). The fluvial sediments are delivered by the highly irregular and variable in magnitude runoff and floods. During subsequent dry lake events, basin deflation removes some of the lacustrine muds often without leaving much evidence in the lake sediments (Knight et al., 1995; Timms, 1992, 2006; Twidale and Wopfner, 1990), resulting in gaps in the record that are hard to locate and may be missed altogether. Knight et al. (1995) estimated that more sediment is removed from the Australian continent by atmospheric transport than by river transport; i.e. the dust load carried by wind is much larger than the sediment load carried by rivers.

While some deflated material can travel over long distances (Chen et al., 2002; Hesse and McTainsh, 2003; Knight et al., 1995), some can be found in or around the lake basin, for example forming a source bordering dune. An analysis of such deposits can contribute to filling the gaps in lacustrine sequences. In some cases, older sediment layers deflated from the active modern lake basin might be preserved in marginal areas under protective cover of bordering dunes that capped these sediments before their deflation could occur. Older lacustrine sediments were, for example, preserved under the red sand dune on the northern margin of the Palaeolake (Chapter 9 section 9.3.3) and under other aeolian deposits in Lake Tyrrell (Macumber, 1991). In Lake Eyre, on the other hand, earlier lacustrine sediments persisted as marginal cliffs around the deflation-incised modern playa basin (Magee et al., 1995; Magee and Miller, 1998).

The complex interaction of the magnitude, the frequency, and the duration of accumulation and erosion/deflation events is likely to be manifested in varying sediment accumulation and preservation/retention rates. This means that constant sedimentation rates cannot be assumed and instead a good dating control is critical for establishment of high resolution chronologies. Furthermore, the correlation of undated sediment units can be difficult even within a single lake basin, as, for example, in Lake Eyre (Magee et al., 1995), due to variation in sediment supply and settling across a lake basin. Relying on sediment characteristics to correlate cores from different lakes is even more problematic, as each lake is characterised by different filling and drying regimes as well as sediment sources. While the frequency and length of the wet and dry periods are important in regulating the duration and magnitude of the sediment accumulation and loss, the extent of deflation is also controlled by other factors such as sediment texture, salt content, and the depth of underlying watertable during the dry stages, that also can vary between the individual lakes. Therefore, more intensive dating control will be crucial in future

palaeoenvironmental studies in arid and semi-arid regions to identify the potential gaps and link the reconstructed events to a common timescale.

Within this study, the dating was constrained by budget (section 4.14.2 below). The correlations between sediment sequences from a single lake system as well as those between different lakes were based mostly on obvious common features and trends (Chapters 5-11). The interpretations were made tentatively in expectation that more dates will became available in the future. Thus, the histories of the individual lakes (Chapters 5-10) were presented as sequences of separate events within a history of unknown continuity. For the sake of clarity and brevity, consideration of any gaps was omitted from the reconstructions presented in following chapters, although their potential presence should not be ignored.

4.14.2 Dating

To fix the reconstructed events in time, this study has employed three dating techniques: 137Caesium, Radiocarbon (AMS), and Optically Stimulated Luminescence (Chapter 3 section 3.17).

4.14.2.1 137Caesium

The radioactive isotope 137Caesium (137Cs) was released into the atmosphere in the mid 1950s to the 1970s during thermonuclear weapons testing (Longmore et al., 1986; McHenry and Ritchie, 1977; Ritchie and McHenry, 1990). On contact with the soil surface or water body, the isotope becomes rapidly and firmly adsorbed to finer particles, thus sands tend to be 137Cs poor (Doyle, 1998; Longmore et al., 1986; Ritchie and McHenry, 1990). The comparison of values from an undisturbed reference site and study sites can provide information about relative sediment loss or accumulation rates and the first appearance of 137Cs in the sediments can be used to mark the 1960s (Doyle, 1998; Elliott et al., 1997; Jones et al., 2000; Longmore et al., 1986; Loughran et al., 2002; Ritchie and McHenry, 1990).

Aside from the general links to annual precipitation (through removal from the atmosphere), the 137Cs deposition patterns can be also affected by factors such as storm tracking, density of tall ground cover (of limited relevance in the study area), and timing of rainfall events (Elliott et al., 1997). Furthermore, the use of 137Cs has to be cautious in saline environments (Doyle, 1998; Gayler, 2000; Longmore et al., 1986). The presaturation of the adsorption sites in the clay matrix by sodium (Na+) ions reduces 137Cs adsorption, increasing its potential for advective and diffusive movement (Doyle, 1998; Longmore et al., 1986). The vertical migration of 137Cs within the sediments can be then further enhanced by downward movement of lake brine or fluctuations of the groundwater table (Longmore et al., 1986). Generally, the higher the salinity the higher the amount of the unattached 137Cs and the lower the reliability of the results (Doyle, 1998; Longmore et al., 1986).

For these reasons, in this study only the fresh to moderately saline lakes were analysed with respect to 137Cs (see Chapters 5-10 and Appendix 9 for details). The 137Cs results for those lakes were highly satisfactory, particularly because in many cores the base of the isotope’s profile was overlying units with signs of increased aridity related to the late 19th and early 20th century drought (Figure 2.5), confirming its reliability as a marker of the 1960s.

Due to low sediment volumes available for sampling from the cores and large minimum sample size required for 137Cs analysis, the resolution of 137Cs results is relatively low, varying between 5 and 10cm. So, while throughout the discussion the depth of the 137Cs profile is generally attributed to the maximum depth of the lowest sample with detected 137Cs content (for clarity reasons), it should be remembered that the actual bottom of the 137Cs profile can occur at any depth within that sample. In the future, it might be useful to extract additional short cores adjacent to the main sediment core, specifically for the 137Cs analysis.

4.14.2.2 Radiocarbon

Two uncallibrated radiocarbon (14C) dates have been obtained for lacustrine sediments, with both samples extracted from the Palaeolake core B (Chapter 9 and Appendix 9). These successful results not only contributed to timing of the reconstructed events, but also demonstrated that carbon content of the semi-arid lakes can be adequate for radiocarbon dating. Employment of alternative techniques might be, however, desirable to scrutinise the results, as recommended by Cupper (2006) based on comparative study of radiocarbon and OSL dates from playas in the Murray Basin, southeast Australia.

4.14.2.3 Optically Stimulated Luminescence

Twelve samples for Optically Stimulated Luminescence (OSL) dates (Appendix 9) were extracted and pre-processed by me in the Australian National University OSL laboratory as part of commercial arrangement. However, due in part to staff changes, the laboratory failed thus far to complete the sample processing and analysis.

Chapter 5

Lake Bindegolly

5.1 Lake Bindegolly

5.1.1 Site and lake overview

Lake Bindegolly lies in a downfaulted area to the northeast of a Tertiary Bindegolly fault and is the most southern lake of a system composed of freshwater Lake Hutchinson and saline Lakes Toomaroo and Bindegolly (Figure 5.1) (Timms, 1997b). The lakes are supplied with water exclusively by local streams, mainly Bundilla Creek. All of the lakes are ephemeral and dry out about once in 10 years (Queensland Parks and Wildlife Service, 2007).

Figure 5.1 Map of Lake Bindegolly. (Image source: Google Earth 2006; Geological Information Source: Eulo 1:250 000 Geological Series Map Sheet SH 55-1, 1971) 89

Lake Bindegolly is a relatively deep lake (about 4-5m) and most of the sedimentation seems to be balanced by sediment deflation during dry periods (B. Timms, pers. comm. 2007). In spite of its status of terminal lake, Bindegolly is not very saline, generally ranging from subsaline to mesosaline. The lake possesses a very well developed lunette dune with possible existence of other lunettes (less well developed and to a large degree submerged at higher lake levels) (Figure 5.1) (B. Timms, pers. comm. 2007).

To the west, north, and east Lake Bindegolly is surrounded by red sand plains and to the south by large outcrops of Tertiary silcrete (Eulo 1:250 000 Geological Series Map Sheet SH 55-1, 1971). Much smaller and isolated outcrops of Tertiary silcrete form cliffs on the western margin of the lake raising up to about 25m above the lake floor (Figures 5.2 and 5.3). The cliff tops are covered by desert pavement (gibber) (Figure 5.4).

Figure 5.2 The cliff on the western margin of dry Lake Bindegolly. The pink hue of the image is caused by approaching dust storm.

The lake margins are populated mainly by samphires (Halosarcia spp.) and sedges (Cyperus spp.) (Figure 5.5). The surrounding area to the north, east, and south of Lake Bindegolly is covered by mulga (Acacia aneura) communities, gidgee (A. cambagei) woodlands, Eremophila spp. dominated shrublands and stands of A. ammophila (Queensland Parks and Wildlife Service, 2007). In the west the lake boundary consists of a high rocky cliff extending into sparsely vegetated gibber plains.

Figure 5.3 The close-up of the Lake Bindegolly’s western cliff looking down toward the lake.

Figure 5.4 Lake Bindegolly: the gibber plain extending westwards from the western cliff edge.

Figure 5.5 The eastern shore of Lake Bindegolly during a full stage.

5.1.2 Results

The lake core 4 (in two parts a and b) was selected for the analysis as it contained the longest sediment record. The general description of the sediment from core 4 is provided by Figure 5.6. Pollen and biogenic silica results are presented in Figures 5.7 and 5.8 respectively, while Figures 5.9 and 5.10 show other sedimentary data. Additional descriptions are provided below for selected main proxies.

5.1.2.1 Dating: 137Caesium

All of the 137Cs occurs within the top 20cm of the sediment (Figure 5.6). While some diffusion within the sediments is possible, the evidence of other proxies (e.g. pollen and phytoliths) is generally supportive of 1950s date for the bottom of the 137Cs profile. The total value of 159.72mBq/cm2 ±24.47 when compared to the reference value of 29mBq/cm2 (for Thargomindah) suggests substantial net sediment accumulation (but does not precludes some losses during dry lake phases).

Figure 5.6 Sediment description for Lake Bindegolly cores 4a and b.

5.1.2.2 Pollen

The short pollen record for Lake Bindegolly (Figure 5.7) shows a clear increase (towards modern surface sediments) in tree, tall shrub, and aquatic counts (the latest first appearing at 15cm). Herbs and grasses are consistently recorded, although low shrubs (represented in this core solely by Chenopodiaceae) increase initially to a peak at 30cm, then decrease sharply at around 20cm to recover at 15cm and continue to increase towards the surface. The depression in the Chenopodiaceae count around 20cm is accompanied by decrease in total pollen concentrations (evidenced by a peak in marker spore Lycopodium counts). While decreases in occurrence with depth in some of the groups (e.g. Cyperaceae) might be a result of pollen deterioration due to the harshness of the depositional environment, the relatively strong presence of Chenopodiaceae and Asteraceae, Brassicaceae, and Haloragaceae at 30cm provides support for another explanation: a period of increased aridity (and thus lower pollen production).

An interesting coincidence is the peak at the 20cm depth in Unknown type 1 counts, which, although at this time remains unidentified, tends to appear in association with periods of increased aridity (e.g. its occurrence in Palaeolake described in Chapter 9 section 9.3).

The sample from 35cm depth has been excluded from the interpretation since it was uncertain if the ratio of Chenopodiaceae and Asteraceae pollen to Unknown type 1 is the result of a true relationship or preservation issues. The latter explanation is supported by a dramatic decrease in total pollen counts. The overrepresentation of the Unknown type 1 in this sample is particularly likely since the grain seems to preserve much better than others (e.g. Chapter 9 section 9.3).

5.1.2.3 Phytoliths

Like pollen the phytoliths practically disappear from the sediment below 30cm depth (Figure 5.8). Small phytolith numbers are also recorded for surface counts. The grasses (Poaceae) dominate the record from surface to 25cm depth and then decrease in favour of Dicotyledon/Monocotyledon (non-Poaceae) plants, which peak at 30cm. A depression of all phytolith types and overall counts at around 20cm mirrors the pollen counts suggesting a decrease in vegetation cover.

5.1.2.4 Other biogenic silica

The freshwater sponge spicules have been recorded at depths of 5-40cm, with lower concentrations in more modern sediments (5-10cm) and a slightly higher presence in 15-40cm peaking at 30cm (Figure 5.8). Rare individuals of unidentified Fossil 1 occur at depth of 20- 30cm (except for 22cm). 94

Figure 5.7 Pollen counts for Lake Bindegolly cores 4a & b. 95

Figure 5.8 Biogenic silica counts for Lake Bindegolly cores 4a & b.

5.1.2.5 Other signs of plant and animal life

All the other signs of plant life in the sediment core, including unidentified seeds and plant fragments, and charophyte oospores are also limited to the top 30cm with very limited presence below 10cm (Figure 5.6). Three species of charophytes have been identified: Lamprothamnium heraldii, Nitella cf. ignescens, and Nitella cf. verticillata. As Lamprothamnium sp. can tolerate large changes in salinities (fresh to hypersaline) and both of the Nitella species live in subsaline to hyposaline habitats, their presence is indicative of the present hydrological characteristics of the lake (section 5.1 above).

Occasional Coxiella gilesi shells were found at depths of 12.5, 16, and 27-29.5cm, possibly marking a temporary location of the shoreline (Appendix 8) as well as providing support for the

moderate salinity levels at least during the period represented by the 10-30cm section of the sediment.

A zone of frequent root tunnels occurs at a depth of 39-46cm.

5.1.2.6 Organic matter and carbonate content

Except for the upper 10cm, the organic matter and carbonate content in Lake Bindegolly’s core 4 are relatively low (<5%) and uniform (Figure 5.6). A very slight peak in both materials occurs at 50cm, just below the oxidised (red) zone crisscrossed by a relatively dense network of fine root tunnels.

5.1.2.7 Magnetic susceptibility

The Lake Bindegolly core is characterised by three main peaks/enhancements in magnetic susceptibility: at about 22, 45-60cm (possibly also lower but the signal is distorted below that depth by a void), and 114-126cm (Figure 5.9). All of the peaks are characterised by medium values of the frequency dependency coefficient. The 22cm peak is associated with about 2cm thick band of reddish sediments with moderate upward and downward diffusion. The lack of clear stratification and intensive mottling of the sediments adjacent to this unit might suggest soil development, however, the lack of defined root tunnels and the peak in clay content at this depth is puzzling. An alternative explanation might be the influx of catchment derived fine sediments from topsoil (evidenced by the peak (5.2%) in frequency dependent susceptibility) eroded during a major flood event following reduction in vegetation cover.

The magnetic susceptibility enhancement at about 45-60cm is associated with a slightly lower frequency dependency value (4.4%), a very slight peak in organic matter and carbonates, pronounced increase in pH and sand-sized particles, and evidence of root tunnel just above the unit, suggesting links with soil development. Some uncertainty about this explanation is, however, introduced by the dark grey staining of the unit (Figure 5.6). While the dark grey colour of that stain is similar to that of the sulphide muds at the top of the core, the strong SWIR absorption for that depth does not support sulphide presence (which acts as reflectance suppressant) (Figure 5.10) (Pontual et al., 1997).

The lowest peak in magnetic susceptibility at about 114-126cm occurs at the bottom of a larger (99-131cm) reddish unit overlying grey sediments. The frequency dependent susceptibility of 7-7.8% suggests pedogenic origins of the material, but no clear soil structures are present. The redness of the sediment suggests oxidising, thus drier, conditions, further supporting the soil connection (either through in situ development or influx of topsoil eroded from the catchment/surrounding area).

Figure 5.9 Lake Bindegolly cores 4a & b: sediment diagram.

Figure 5.10 Infrared spectrometry data (SWIR) for Lake Bindegolly cores 4a & b.

5.1.2.8 Gypsum

While small amounts of gypsarenite dispersed though the lake muds were noted in the freshly opened core at depths of 100-131.5cm (Figures 5.6 and 5.9), it was not recorded in the LOI

samples (Figure 5.9), SWIR analysis (Figure 5.10), nor the >250µm fraction from pollen samples (Chapter 3 section 3.14.2). It is possible that the initial observation was confused with large quartz grains that were observed in the pollen samples within this section.

Furthermore, there was no gypsum visible at the 4cm depth, in spite of the suggestion by the Spectral Geologist software (Figure 5.10). However, since the weak absorption at this depth is probably caused by high presence of sulphides (Figure 5.6) that can significantly lower the reflectance and weaken the spectral absorption features of other minerals within the sample (Pontual et al., 1997), the results for this sample are likely to be biased.

5.1.2.9 Mineral composition: clay content

The kaolinite and its hydrated form, halloysite, dominate the sediments down the core (Figure 5.10). The differences in spectra between individual samples are mostly limited to the strength of adsorption (likely to be related to changes in clay content).

5.1.3 Palaeoenvironmental reconstruction of events

Event A: Permanent lake phase (>132cm)

The grey sediments below 132cm are likely to have been deposited under permanent (likely also deep) lake conditions that persisted over a long period (evidenced by low magnetic susceptibility and the thickness of the unit). Such conditions would have been conducive to at least moderate density, permanent vegetation cover on the surrounding landscape.

Event B: Increased aridity, shallow/dry lake phase (~94-132cm)

The increased redness of this section suggests increased input of oxidised material into the lake basin. The decrease in vegetation cover in the surrounding catchment/area (most probably due to intensified aridity) has increased the susceptibility of the soil to erosion. Some of the top soil material has been deposited and preserved in the lake basin (under shallow, ephemeral and/or dry lake conditions). While the evidence of root presence in the form of clearly defined root tunnels is scarce, the intense mottling of this reddish unit does not preclude some colonisation by plants, which could have then acted as a trap for the airborne sediments from the catchment (supported by increased sand-sized particle content). The prolonged aridity has resulted in stripping of the topsoil in the catchment below replacement rates as evidenced by the decreasing content of the magnetic particles (including superparamagnetic particles) towards the top of the unit.

Event C: Amelioration of conditions (~46-94cm)?

The greyness of the sediments in this unit suggests a temporary increase in lake depth and permanency, however, this is the only evidence available to support this interpretation. An 100

increase in sand-sized sediments and magnetic susceptibility at about 65cm might indicate increased input of aeolian sediments, while the water level is still relatively high, thus suggesting a gradual deterioration of the conditions towards aridity.

Event D: Dry lake phase (~40-46cm)

The red (oxidised) sediments, frequent root tunnels, increased organic matter and carbonate content, peak in mineral magnetic susceptibility, and moderate frequency dependency value at this unit suggest development of soil on the lake floor, that in turn suggests prolonged dry conditions. The increase in sand-sized particles suggests entrapment and accumulation of airborne materials, thus likely decreased vegetation cover and increased surface erosion in the surrounding area/catchment.

Event E: Wet phase: the 1880s to the early 1890s (~30-40cm)

The unit is characteristic of lacustrine deposits due to its grey colour and low magnetic susceptibility. The increased proportion of herbaceous pollen (particularly Asteraceae) additionally supports the milder conditions.

Based on the timing of the drought described below and the rainfall data (Figure 2.5), the sediments were likely to accumulate during the period of above average rainfall in the 1880s and the early 1890s.

Event F: Drought: the late 1890s to the 1940s (~20-25cm)

The depression in pollen and phytolith counts is likely to represent a decrease in vegetation cover and pollen production in the surrounding area in response to water shortages combined with increased grazing pressures (described also in Chapter 2 section 2.8.2). The lake at that time was drying (i.e. characterised by low water levels and increased salinity) or dry as indicated by the dominance of Unknown type 1 pollen.

A band of strongly oxidised (reddish) sediments of increased magnetic susceptibility and frequency dependent susceptibility values, as well as lack of clear evidence of root presence/activity, provides further evidence of the increased aridity and likely erosion of the topsoil in the surrounding area.

The timing of the event is based on the location of the unit just below the bottom of the 137Cs profile (thus ‘shortly’ before the 1950s) and the regions rainfall record (Figure 2.5), which indicates prolonged dry conditions in the period between the late 1890s to the 1940s.

101

Event G: Amelioration of conditions in the mid 1950s and 1975 and final deterioration in most recent times (~<20cm)

The general recovery of vegetation following the drought is evidenced by an increase in both pollen and phytolith counts and was likely to be favoured by the medium to above average rainfalls dominant in the second half of the 20th century (Figure 2.5). However, the decrease in phytoliths and Asteraceae/Poaceae pollen, increase in salt tolerant Chenopodiaceae in surface and near surface sediments, disappearance of Coxiella gilesi above 10cm and marked increase in sediment salinity at the surface level seem to imply increasing lake salinity (site dryness?) in the last few years.

102

Chapter 6

Currawinya Wetlands: Lake Wyara and Lake Numalla

6.1 Currawinya Wetlands

Lake Wyara and Lake Numalla are the largest lakes in the Currawinya wetland system. Most of the wetlands were a part of two grazing (sheep and cattle) properties until 1991 when it was declared a national park. The landscape is dominated by sandplains covered mainly with mulga (Acacia aneura) and woody shrubs (mainly turpentine Eremophila sturtii and hopbush Dodonaea viscosa) with frequent specimens of poplar box (Eucalyptus populnea), beefwood (Grevillea striata), emu apple (Owenia acidulate), leopardwood (Flindersia maculosa) and whitewood (Atalaya hemiglauca), and more rare belah (Casuarina cristata), Melaleuca densispicata, and black bluebush (Maireana pyramidata) (Queensland Parks and Wildlife Service, 1999). In the northeast, the Hoods Range supports populations of bastard mulga (Acacia stowardii) and lancewood (Acacia petrea), while the channels and floodplains of the Paroo River and other creeks are lined with yapunyah (Eucalyptus ochrophloia), coolibah (E. coolabah), river red gum (E. camaldulensis), and gidgee (Acacia cambagei) (Queensland Parks and Wildlife Service, 1999).

6.2 Lake Wyara

6.2.1 Site and lake overview

Lake Wyara covers a large area of 3400 ha and is relatively shallow (Timms, 1999, 2006). It has formed in a depression associated with tectonic movements along a Tertiary fault line, just east of the lake, and was further modified by deflation, sedimentation, and wind-induced currents and wave action as evidenced by its curved smooth shoreline and the presence of multiple semi-parallel beaches and spits (Figure 6.1) (Timms, 1997b, 1998c, 1999, 2006). The lake floor is relatively flat and slopes slightly southwards (Timms, 1997b, 1998c).

103

Figure 6.1 Map of Lake Wyara and Lake Numalla. (Main image source: Google Earth 2006; Inset image source: Natmap Raster 2003; Source of geomorphological information: Timms, 1997b, 1998c, 2006)

104

Beaches fringe Lake Wyara along most of the shoreline but are best developed at its northern end. Most of the major beaches survive drowning without much damage and new minor beaches can form in response to changes in the lake water levels (Timms, 1998c). While the height of the beaches increases landwards, they are generally of little value in determination of past lake levels due to the importance of longshore drift in their formation (Timms, 1997b, 1998c). The only reliable witnesses to former water levels are sand and pebble beaches and stranded shingle spits located at a height of 6.7 to 6.95m above the lake floor on the northwestern shore of the lake (Timms, 1997b, 1998c) (Figure 6.1). Spits obscure inlets of two main inflowing creeks: Werewilka in the north and Kaponye in the south, forming estuaries (Timms, 2006). An inconspicuous red clay (non-gypsum) lunette borders the lake in the east (Timms, 2006; B. Timms, pers. comm. 2007).

The Lake Wyara’s depth can reach over 6m on the rare occasion when it overflows, but it usually fluctuates between 4m (considered full) and dry (Timms, 1997b, 1998c, 1999). The lake’s catchment (249 000ha) is practically closed as the lake is filled mainly by local creeks, including a large Werewilka Creek in the north (draining 74% of the catchment) and six minor creeks originating in ranges to the south and west of the lake (Kingsford and Porter, 1999; Timms, 1997b, 1999). The groundwater inflows and outflows are believed to be negligible (Timms, 1998c). Since Wyara’s catchment is rather small, the lake can be filled only by a large rainfall event (<240mm) continuing over few days; such rainfall occurred only 16 times in the last 108 years (Timms, 1999). The period between fills can vary from 2 to 20 years (Timms, 1997b). The full stage generally does not persist for longer than one year, but the dry stage can last up to three years (Timms, 1997b, 1998c). The lake has dried about 19 times in 108 years (Timms, 1997b).

More moderate rainfalls occurring every 0.6-1.4 years will result in only partial filling of the lake (Timms, 1999). An overflow stage, according to the available evidence, can be reached only in the rare events of the Paroo River floodwaters reaching the lake during major flood events (about once in 25 years) (Kingsford and Porter, 1999; Timms, 1997b, 1998c, 1999). The floodwaters enter the lake via Kaponye Creek and leave the same way when the flood level starts to drop. Timms (2006) has shown a strong correlation (r=0.62) between the lake’s full and dry periods and the rainfall-drought periods explained by the Southern Oscillation Index.

Timms (1997b; 1998c) has identified three major strandlines along Lake Wyara’s shores. The first one, at 2.6m above the lake floor, is well marked on the eastern shore by a clay bank lying on the boundary between bare clay and sparse samphires (Halosarcia spp.). The second strandline (representing a ‘lake full’ level), at about 4m above the lake floor, can be distinguished by a clearly visible change in density of the samphire bushes enriched by pigface

105

(Sarcozona praecox), red trefoil (Lotus cruentus), monkey flower (Mimulus repens) and mulka (Eragrostis dielsii) and supported in some places by western boobialla (Myoporum montanum) (Figure 6.2) (Queensland Parks and Wildlife Service, 1999). The third strandline (assumed to represent overflow stage) at about 6.2-6.9m above the lake floor is marked by high beaches (mentioned above) and in several places by a change from low to high shrubs.

Figure 6.2 A line of dead trees along one of the beach ridges on the southern margin of Lake Wyara (located lower and closer to the lake centre than the present line of live trees). Halosarcia spp. in the foreground.

The fluctuating water levels have a major impact on the salinity which ranges from subsaline (2.8g/L) at full lake level to crystalline brine (350g/L) as the lake dries with sodium and chloride the dominant ions (Timms, 1997b, 1998c, 1999, 2006). The water is usually clear and alkaline with a mean pH of 8.6 (ranging between 8.1-10.3) (Timms, 1997b, 1999, 2006).

Wyara is often very productive, and supports a unique biota of limited diversity (Timms, 2006). The clear and nutrient-rich water in Lake Wyara provides (especially at low to moderate salinities <50g/L) attractive habitat for aquatic plants including many charophytes such as Chara, Lamprothamnium, Nitella spp. and Ruppia sp. as well as zooplankton (Timms, 1998c, 1999). The lake’s fauna is dominated by small crustaceans (<2mm), particularly ostracods and cladocerans, that can ‘bloom’ at low to intermediate salinities, but disappear at higher salinities (>60g/L) (Timms, 1999). The presence of particular species is related closely to the water salinity and is independent of seasons (Timms, 1997b). 106

6.2.2 Results

Of two cores recovered from Lake Wyara, core 2 was selected for detailed analysis as it was characterised by seemingly clearer stratification, although both cores were very similar in appearance. Figure 6.3 provides a general description of the sediment and Figure 6.4 shows morphological details of selected gypsum crystals. Pollen and biogenic silica results are shown respectively in Figures 6.5and 6.6, while other sedimentary properties are presented in Figures 6.7 and 6.8. Additional descriptions are provided for the more useful proxies.

Figure 6.3 Sediment description for Lake Wyara core 2.

107

6.2.2.1 Gypsum

There are three noteworthy bands of gypsum in the Lake Wyara core: two gypsarenite bands at 75 and 82cm and a gypsite lamina at 126cm (Figures 6.3 and 6.4). The gypsite lamina (Figure 6.4 image 3) is about 1mm thick and in places broken by void structures (caused by either clay cracking or root penetration) that were filled by overlying, more reddish, sediment. The gypsarenite bands are about 1cm thick and both are composed of lenticular to subhedral pyramidal crystals with slight twinning (Figure 6.4 images 1 & 2), thus suggesting precipitation from sulphate-saturated groundwater some time after the deposition of the sediment (Chapter 4 section 4.9.1) (Aref, 2003; Bowler and Teller, 1986; Buck and Van Hoesen, 2002; Magee, 1991).

Figure 6.4 Lake Wyara core 2 gypsum: 1a & b – partly dissolved lenticular and lenticular twin gypsarenite at 75cm; 2 – lenticular and lenticular twin gypsarenite with no signs of dissolution at 82cm; 3 – gypsite band within grey clay at 126cm.

108

The lack of gypsum lunette adjacent to this lake, together with comparatively low gypsum content of the sediments (Figures 6.7 and 6.8), might indicate a shortage of either one or both of its ionic components (sulphates and calcium) within the lake system. Chemical analysis of the lake water and groundwater would be needed to confirm this theory.

6.2.2.2 Pollen

While the total pollen counts from Lake Wyara stayed above the value of 25 (considered adequate for preparing a summary diagram) from surface to the 90cm depth, the section below 23cm was dominated by a single type: the Unknown type 1 (Figure 6.5). Within the top 23cm there was only a slight increase towards the surface of tree and tall shrub pollen. More notable was the increase in low shrubs (dominated by Chenopodiaceae), accompanied by a slump in herbaceous, and in particular Asteraceae, pollen. Considering the preferential preservation of both types (especially Chenopodiaceae) (Chapter 4 section 4.12.3), this trend is likely to indicate a change in species balance (from less to more salt tolerant).

6.2.2.3 Phytoliths

The phytolith record for Lake Wyara is much shorter than the pollen record and covers only the top 40cm (Figure 6.6). While the proportions of Poaceae to Dicotyledons/Monocotyledons (non-Poaceae) does not vary very much, there is a significant decrease in total phytolith count at a depth of 10cm. However, unlike in Lake Bindegolly, this drop is not replicated in pollen counts.

6.2.2.4 Other biogenic silica

The presence of freshwater spicules within Lake Wyara’s sediments overlaps that of phytoliths and is confined to the upper 40cm of the core (Figure 6.6). The spicule numbers are relatively low with only a very slight increase at 30-40cm depth. Fossil 1 shows up again in the sediments in ‘large’ concentrations (the highest numbers recorded in this study) at 10cm and again, but in much smaller numbers, at 40cm. The first occurrence coincides with significant depression in total phytolith densities and the second with the lowest depth at which phytoliths were recorded. Its unknown identity of Fossil 1, however, limits current conclusions.

6.2.2.5 Other signs of plant and animal life

Most other plant evidence, including seeds and plant fragments, are confined to the top 25cm of sediment (Figure 6.3), although, charophytes persist to a depth of 30cm with a dense band at 27- 28cm. The identified oospores belong to the Lamprothamnium genera that are tolerant of high variations in water salinity and levels (Appendix 7), thus well represent the present conditions at Lake Wyara.

109

Figure 6.5 Pollen counts for Lake Wyara core 2.

110

Figure 6.6 Biogenic silica counts for Lake Wyara core 2.

Multiple sections of root tunnels found occasionally throughout the core may indicate frequent prolonged drying of this part of the lake basin and associated ‘invasion’ by plants, most likely dominated by Halosarcia spp. considering the salinity of the sediment (Figures 6.3 and 6.7).

The dead trees on the lower shoreline around Lake Wyara (Figure 6.2) suggest a long period of persistently low water levels (probably from the late 1890s to the 1940s) followed by a very full stage in around 1950 and overflow conditions in the mid 1970s (Timms, 1998c), where either of the two periods could have resulted in drowning of trees.

111

Figure 6.7 Lake Wyara core 2: sediment diagram.

112

Coxiella gilesi, Diacypris spinosa, and Reticypris (2 different but unidentified species) were recovered from 27-28cm depth. All taxa are indicative of saline environments: the gastropods of hyposaline to mesosaline and the ostracods of hyposaline up to highly hypersaline (hence present conditions) (Appendix 8).

6.2.2.6 Magnetic susceptibility

The Lake Wyara’s core 2 is characterised by one distinctive peak in magnetic susceptibility at about 20cm (Figure 6.7). The peak is associated with a red sediment band, which bears many similarities to the ~22cm band in Lake Bindegolly. While much thinner (about 3mm), this band is also characterised by a higher content of fine particle (particularly clay) sizes and an increased frequency dependent susceptibility value (5.6%). Similarity in the general characteristics and the occurrence depth of those two red bands from Lake Wyra and Lake Bindegolly can possibly indicate their deposition in response to the environmental conditions existing in the region during the same period of time.

The slightly higher magnetic susceptibility below 100cm depth combined with frequency dependent susceptibility of 2.9%, presence of root tunnels, and slight redness of the sediments suggest a prolonged period of dry to shallow (oxidising) lake conditions, allowing some plant establishment (aquatic and/or terrestrial), but little soil development both in situ and source sediment area.

6.2.2.7 Mineral composition: clay content

Like Lake Bindegolly, Wyara’s sediment is dominated by kaolinite and halloysite (Figure 6.8). The differences in spectra absorption between the samples are largely limited to its strength and are therefore most likely are related to the changes in clay content.

6.2.3 Palaeoenvironmental reconstruction of events

Event A: Transition from lacustrine to drier conditions (>125cm)

The grey finely textured (silty clay loam) unit below about 125cm is likely to represent a permanent deep water stage in the lake’s history. The gypsite lamina located close to the top of the unit might be considered a first sign of increasing dryness. As the lake started to dry, the evaporating water reached supersaturation with respect to gypsum resulting in nucleation of gypsite-size crystals and their subsequent deposition on the lake floor; although, analysis of the crystal habit needs to be carried out to confirm its ‘from brine’ origins. The fine crystal size suggests rapid drying (Warren, 1982).

113

Figure 6.8 Infrared spectrometry data (SWIR) for Lake Wyara core 2.

Event B: Dry phase (~125-90cm)

The continued drying of the lake floor resulted in cracking of the clays and breaking up of the gypsite lamina. Meanwhile in the catchment, the progressively drier conditions led to decrease in vegetation cover and increased mobilisation of the surface sediments (characterised, however, by a very weak topsoil signature, as evidenced by the low frequency dependent susceptibility). Some of the airborne particles, with an increased proportion of fine sand, settled on the lake 114

floor filling the voids. The dryness of the lake floor also allowed colonisation by plants, further facilitating entrapment of the aeolian sediment and its incorporation into the underlying lacustrine muds.

Event C: Fluctuating lake level (?) phase (~90-30cm)

While the unit lacks definitive proof of fluctuating lake levels during this phase, such as, for example, gypsum-clay couplets, there are some subtle hints of it in the proxies. The foremost is the dominance throughout the unit of the Unknown type 1 grain, which appears to have strong links with fluctuating lake levels and gypsum dune building conditions (e.g. shown clearly in the analysis of Palaeolake in Chapter 9 section 9.3.2). Furthermore, the unit is composed of mottled greyish and reddish sediments suggesting that the sediment deposited during wetter (high water levels) and drier conditions (influx of oxidised materials) was mixed by plants encroaching onto the lake floor during drier periods (supported by signs of root presence in different densities through most of the unit).

Event D: Wet-dry cycles: late 19th to mid-20th century (?) (18-24cm)

The five couplets of fine laminae of light grey clays and red sediments within this unit suggest influxes of, probably flood derived, clays interspersed with short but intensive periods of shallow/dry lake conditions. The wet period seems to have favoured vegetation growth as evidenced by peaks in both total phytolith and pollen concentrations as well as Asteraceae pollen and grass phytoliths. Periods of high lake productivity are evidenced by concentrations in other fossils such as charophytes, gastropods, and ostracods.

The enhanced magnetic susceptibility accompanied by high frequency dependent susceptibility peaking in tune with the occurrence of the uppermost and thickest of the red bands (at 20cm depth) and the absence of root traces suggest influx of topsoil material from the surrounding area/catchment, which is likely to have suffered at that time from decline in vegetation cover. In view of strong representation of the late 1890s to 1940s drought in other lakes in this study (e.g. Lakes Bindegolly and Numalla), it is possible that this unit, or at least the uppermost part of the red bands, is related to this period. Dating is, however, essential to confirm this interpretation.

Event E: Amelioration and final deterioration of conditions: second half of the 20th century (?) (0-18cm)

The dominantly grey to black (sulphidic) sediments in the upper 18cm suggest a period of wetter conditions preventing the lake from complete drying (the wetter second half of the 20th century?). The temporarily drier conditions are, however, likely to have occurred at about 10cm as implied by increase in redness and dips in pollen and phytolith concentrations. An overall 115

increase in the site’s salinity (e.g. due to increased dryness and/or decreased flood flushing frequencies) is suggested by a steady increase in Chenopodiaceae to Asteraceae ratio and sediment salinity, thus showing some similarity to the most recent lake phase described above for Lake Bindegolly.

6.3 Lake Numalla

6.3.1 Site and lake overview

Lake Numalla is a freshwater, almost permanent lake extending over 2900ha (Timms, 1999). It is also the deepest lake in the region: 6.5m at full stage (Timms, 2006). The lake has formed in a valley of Boorara Creek that became blocked by Paroo River sediments (Timms, 1997b, 1999). Later it was significantly enlarged and deepened by deflation (Timms, 1997b, 1999). The sediment removed by wind from the exposed lake floor was partially deposited at the downwind margins of the lake to form massive sandy beaches, which at present surround the lake along most of its shoreline (Figure 6.9) (Timms, 1997b, 1999). While the western shore supports a single high beach (3-4m above lake floor), two parallel beaches (82 and 44cm lower than the western one) can be found along the eastern and southern shores (Timms, 1997b, 2006). The inner eastern beach becomes submerged at high water levels (Timms, 2006). The well developed spits at the western and northeastern shores (Figure 6.1) indicate the prevalence and strength of the southwesterly (and also hint at the importance of the southeasterly) winds in the region, which are responsible for induced northerly currents (Timms, 1997b, 2006).

Figure 6.9 Lake Numalla: A line of dead trees on the lower beach and a line of live trees on the top of the beach ridge in the background. The green ground cover along the beach consists mainly of spiny sedge.

116

Lake Numalla is fed by both local runoff and the Paroo River floods (Kingsford and Porter, 1999; Timms, 1997b, 1999). The local runoff reaches the lake mainly though three streams flowing from the north and entering the lake through extensive inlets (Figure 6.1). The most important is the Boorara Creek, which has the largest catchment and hence delivers the largest flows (Timms, 1997b). In spite of the important role of the local runoff in sustaining the Lake Numalla’s permanency (with minor water contributions every 0.6 years and major contributions every 1.7 years), only on very rare occasions is it able to effect a complete filling of the lake (Timms, 1999).

In contrast to Lake Wyara, Numalla can be filled by moderate Paroo River floods with minor flood inflows every 0.8 years and major inflows every 1.7 years (Timms, 1999). The floodwater reaches the lake through the southern estuary extending into the Coomburra Waterhole, which is linked to the Paroo River through the Carwarra Creek (Figure 6.1) (Timms, 1997b). When water levels in Lake Numalla become higher than in the Paroo River (either because of flood subsidence or extensive lake filling by Boorara Creek) the flow direction along the Carwarra Creek reverses (Timms, 1997b).

As a result of the frequent filling events, Lake Numalla is almost permanent and it had dried only four times in the last 100 years, with the last drying episode in 2005 (Timms, 1999, 2006). Its permanency was not conducive to lunette development. Instead, a series of elongated and slightly curved claypans east of the present lake shore hint at a possible location of much older shorelines (Figure 6.1) (B. Timms, pers. comm. 2007).

Due to Lake Numalla’s rather complex structure (particularly the presence of bays) the water mixing is difficult, thus salinity, turbidity, and pH can vary spatially and are usually much lower close to inflow areas (Timms, 1999). Generally, however, depending on the lake levels, the water is fresh (0.07mg/L) to subsaline (3.9g/L) (increasing to hypersaline only prior to drying), opaque/turbid, and alkaline (mean pH 8.9 ranging from 6.3-10.2) (Timms, 1997b, 1999).

In spite of the low total invertebrate biomass, compared to other wetlands in the area, the near permanency of the lake and its connections with the Paroo River attract widespread crustacean species and a few riverine species and shrimp (Timms, 1999). The lake’s margins are sparsely to densely inhabited by spiny sedge (Cyperus gymnocaulos) and less frequent saltbushes (e.g. pop bush Atriplex holocarpa) while sheltered bays provide home for red water milfoil (Myriophyllum verrucosum) (personal observation; (Timms, 1997b, 1999). The top of the beach is lined with black box (Eucalyptus largiflorens), western boobialla (Myoporum montanum), belalie (Acacia stenophylla) and needlewood (Hakea leucoptera) (Figure 6.10) (personal observation; Queensland Parks and Wildlife Service, 1999). Remains of flood debris were visible on the lower branches of the trees during a visit in 2003. A line of dead trees half 117

way up the beach is also very prominent on the southern shore (Figure 6.9). Woody shrubs with sparse ground cover of mainly love grass (Eragrostis spp.), button grass (Dactyloctenium radulans), Frankenia gracilis, creeping monkey flower (Mimulus repens), and sesbania pea (Sesbania cannabina) dominate beyond the line of live trees (Figure 6.11).

Figure 6.10 Lake Numalla: The tree line at the top of the southwestern beach with black box in the foreground of the image. Woody shrubs visible in the background.

Figure 6.11 Lake Numalla: Woody shrubs (mainly turpentine and hopbush) on eroding red sands behind the beach.

118

6.3.2 Results

Two lake cores were recovered from Lake Numalla. The water depth limits of the coring equipment were reached close to the steep beach profile and both cores were extracted in the proximity of the shore (Figure 6.1). The longer one (core 1) was used for detailed analysis with general sediment description contained in Figure 6.12, pollen and biogenic silica presented in Figures 6.13 and 6.14, and other sediment data shown in Figures 6.15 and 6.16. The shorter core 2 was used solely for 137Caesium determinations.

Figure 6.12 Sediment description for Lake Numalla core 1.

119

6.3.2.1 Dating: 137Caesium

The coincidence of the bottom of the 137Cs profile with the bottom of the sandy sediment unit (Figures 6.12 and 6.15) suggests their accumulation in the first half of the last century. The lower beach that possibly developed during the dry period of the late 1890s to the 1940s was covered by finer sediments (marked with 137Cs) after a return of wetter conditions in the 1950s. The presence of 137Cs in the sand might be related to reworking of the beach sediments that is particularly likely to occur in a large lake close to the shoreline and/or downward diffusion of the overlying clays into coarsely textured underlying sandy sediment facilitated by the downward movement of the water table during periodic drying episodes.

Furthermore, Lake Numalla is the only lake in this study with total 137Cs (27.14mBq/cm2 ±8.42) lower than the reference value (29mBq/cm2). Given the lake’s permanency, the reworking and redistribution of sediment due to near-shore wave action is a more likely explanation for this phenomenon than sediment loss due to deflation.

6.3.2.2 Pollen

The pollen record for Lake Numalla is also relatively short as it is limited to the top 30cm (Figure 6.13). The oldest sample in the record (at 30cm) is dominated by Chenopodiaceae with small amounts of Asteraceae and very low counts of tree and taller shrub pollen. At 20 to 10cm depth the total pollen counts drop substantially, to recover closer to surface. The drop in concentrations coincides with a sand band representing a former beach (Figure 6.15), thus pollen loss due to sediment reworking or ‘leaching’ can not be ignored as a possible explanation. Noteworthy is also the slight peak in Unknown type 1 associated with the initial decline in pollen concentrations.

Another interesting feature of the pollen data is the change in proportions between Chenopodiaceae, Asteraceae, and trees and shrub taxa (particularly Myrtaceae and Dodonaea) within the top 5cm. As the Chenopodiaceae numbers increase, the other two groups decline.

The Cyperaceae pollen (a sole representative of aquatic taxa) is present throughout the pollen section, but it appears in greatest numbers in the top 5cm. This distribution is possibly due to the difficulty with positive identification of the Cyperaceae pollen even with low damage and/or greater susceptibility to damage. Poaceae pollen is also largely restricted to the top 5cm of the core.

120

Figure 6.13 Pollen counts for Lake Numalla core 1.

121

6.3.2.3 Phytoliths

Like pollen, the phytolith record is limited to the top 30cm (Figure 6.14). However, the total count exceeds 25 only in two samples, with larger peak at 20cm and lower at 5cm. The ratio of Poaceae to Dicotyledons/Monocotyledons (non-Poaceae) is similar in both samples.

Figure 6.14 Biogenic silica counts for Lake Numalla core 1.

6.3.2.4 Other biogenic silica

The presence of freshwater sponges is limited to the same depth section as that of pollen and phytoliths (i.e. top 30cm). It reaches its minimum at 10-13cm and peaks soon after at 5cm (Figure 6.14). The peak is accompanied by the presence of diatoms.

6.3.2.5 Other signs of plant and animal life

The evidence of plant life in the form of plant fragments, seeds, and charophyte oospores are limited to the top 13cm. The oospores belong to two charophyte taxa: Nitella cf. verticillata and Lamprothamnium cf. succinctum. The occurrence of Nitella is largely limited to low

122

salinities (subsaline to hyposaline), thus it is consistent with the modern, generally freshwater, lake system. Lamprothamnium was found only in the 13cm sample, i.e. within the ‘beach’ sediments and its presence may be related to possibly higher salinities (particularly on drying).

Mostly fine root tunnels and occasional rootlets occur within a depth of 20-56cm, decreasing in density with depth (Figure 6.12). This unit is also characterised by increased organic matter content (LOI in Figure 6.15). In addition, occasional plant material is present at a depth of ~20- 30cm. Considering that the unit is directly overlaid by a sand band (a lower beach), it is likely that the roots belonged to sedges (incl. Cyperus gymnocaulos), which occur, in sparse to high densities, at higher levels along modern beach (Figure 6.9). The lack of root traces within the sand section itself may be due to the instability of the sand sediments, which could have been easily reworked by wave action following the rise in the lake levels.

6.3.2.6 Magnetic susceptibility

The magnetic susceptibility of the Lake Numalla core 1 fluctuates in a 4-12x10-8m3/kg range. The lowest values occur in the top 25cm of the core and are associated with high sand content (Figure 6.15). They are probably the result of dilution and/or exclusion by sorting (since the unit represents beach sands).

In spite of the minor variations, there is no clear indication of periods of extended lake floor exposure or unusually high sedimentation of catchment derived oxidised material; i.e. there are no particularly high peaks in magnetic susceptibility or frequency dependent susceptibility values, with the latter ranging from 2-3%. This conclusion is supported by the consistently olive to light grey colour of the sediments with very limited signs of mottling. It is, however, possible that if any such records existed, they were destroyed under persistently reducing conditions.

6.3.2.7 Mineral composition: clay content

The halloysite, kaolinite, and montmorillonite are the dominant clay types in Lake Numalla’s sediment. While not recorded in either Lake Bindegolly or Wyara the presence of montmorillonite is not peculiar as its occurrence is often characteristic of low rainfall (incl. semi-arid and dry tropical) environments (Velde, 1992). Strong leaching can effect its decomposition into kaolinite or halloysite (McMeekin, 1985).

123

Figure 6.15 Lake Numalla core 1: sediment diagram.

124

Figure 6.16 Infrared spectrometry data (SWIR) for Lake Numalla core 1.

6.3.3 Palaeoenvironmental reconstruction of events

Event A: Permanent deep lacustrine phase (?) (~30-109cm)

The grey to olive grey sediments of this unit suggest permanent (deep) water conditions prevalent during its deposition. While the relatively high fine sand content (generally between 40 and 50%) might suggest significant input of aeolian sediments, the magnetic susceptibility data does not show any links with pedogenically modified sediments likely to be sourced from mobilised topsoil. Thus other sand sources, such as bank erosion or beach reworking, are possible. If the latter theory is true, then the increase in the coarser fraction (e.g. at 80cm) might indicate brief phases of persistently lower water levels and beach margin descending closer to the coring site.

125

The lack of any signs of buried beaches (e.g. as the one described below), together with the presence of the shoreline-shaped claypans east of the lake, might suggest a past lake not only deeper, but also extending much further beyond its present boundaries.

Event B: Deterioration of conditions (~20-30cm)

The proxy record from this unit suggests increased water deficit. The fine sand content increased as the water levels dropped and the lake contracted. The Chenopodiaceae dominate the record at 30cm, perhaps in response to the increased aridity and/or salinity related to, for example, less frequent flashing by the Paroo River. The lack of pollen from lower down the cores, however, limits the confidence of this interpretation. The increasing dryness of the site and the lake is further supported by the drop in total pollen concentrations toward the top of this unit (thus decrease in vegetation cover and/or decreased pollen production in times of water stress) as well as the appearance of Unknown type 1, a herald of drying/dry and possibly also saline lake conditions.

Event C: Extended period of low water levels (~7-20cm): 1890s-1940s drought

The sediment of this unit is composed mainly of sand, suggesting lowering of the beach. This in turn would required a persistence of consistently low water level, without, however, complete drying of the lake. Furthermore, this low water level phase was long enough not only to allow formation of the lower beach, but also its colonisation by sedges and trees, whose roots penetrated the underlying sediments. The presence of Lamprothamnium within the beach sands suggests increased salinities in the contacted lake.

Two short-lasting higher water level events (perhaps floods from the Paroo River) left their mark within those beach sands in a form of thin grey clayey bands. The influx of water during these events probably helped to maintain the low water levels within the lake basin and prevented a complete long-term drying.

It is important to note that no other beaches or bands of oxidised sediments have been preserved in the Lake Numalla’s record, possibly suggesting that at no other time in its history, did the lake levels persist for so long at such a low level. Because of the complexity of Lake Numalla filling agents/mechanisms, it is possible that its drying was not only linked to increased regional aridity, but might be related to drier conditions within the catchment headwater area (perhaps due to weakening of the monsoon) resulting in less frequent flooding/flushing by the Paroo River.

The 137Caesium evidence, together with the rainfall data (Figure 2.5), seems to place this event within the period of mostly below average rainfall extending from the late 1890s to the 1940s.

126

Event D: Amelioration and final deterioration of conditions: the 1950s-present (0-7cm)

The amelioration of conditions resulted in increase in lake levels leading to the establishment of a higher beach and burial of the former beach by a new influx of fine grey lacustrine sediments. The general increase in water levels has also contributed to the demise of the former beach vegetation, including trees (Figure 6.9).

The moister conditions are additionally supported by a peak (at 5cm depth) in freshwater sponges as well as tree, shrub and herbs (particularly Asteraceae) pollen and total phytolith numbers. It is likely that this wet phase is closely linked with two periods of exceptionally high rainfall in the 1950s and the 1970s.

The most recent part of the record (represented by the surface sample) saw decline in tree, shrub, herb, and grass pollen in favour of Chenopodiaceae accompanied by a decrease in phytolith concentrations. This suggests increasing aridity and salinity of the site accompanied by a possible reduction in plant cover.

127

Chapter 7

Rockwell-Wombah system: Mid Blue Lake and Lake Wombah

7.1 Mid Blue Lake

7.1.1 Site and lake overview

Mid Blue Lake is the second lake along the drainage line of Number 10 Creek in the Rockwell- Wombah system, which consists of Blue Lakes (including North Blue Lake, Mid Blue Lake, and Lake Bulla) and Lake Wombah (Figures 7.1 and 7.32). The lakes have probably formed as a result of blockages along the course of the creek and were further shaped by deflation.

All of the water responsible for filling Blue Lakes is supplied by local catchments with the majority of it delivered by the Number 10 Creek, which south of each lake is partially blocked by dunes (Timms, 2006). Additional water is supplied to the Mid Blue Lake by a minor creek to the east of the lake. On the western side, the lake’s edge is defined by a cliff of soft sandstones cemented by carbonates (Figures 7.2 and 7.3) that also underlie the lacustrine sediments within the lake’s basin (Timms, 2006).

Mid Blue Lake is bounded on the eastern side by a set of two lunettes: an inner clay lunette and outer gypsum lunette (Figures 7.1 and 7.4). Since deflation of the lake floor was observed in present times (B. Timms, pers. comm. and personal observation) and the lunette surface was covered by a layer of fluffy, loose clay at the time of the fieldwork (Figure 7.7), it is likely that the clay dune is still accumulating (even if at a much slower rate than in the past). The swale between the dunes becomes submerged at high water levels. At the landscape scale, the lake lies within extensive red sand plains (Eulo 1:250 000 Geological Series Map Sheet SH 55-1, 1971).

129

Figure 7.1 Map of Mid Blue Lake. (Main image source: Eulo SG55-1 airphoto 63, 1995; Inset image source: Google Earth 2006)

When full, Mid Blue Lake covers an area of about 210 ha and is 3.4m deep (Timms, 2006). The lake’s salinity fluctuates between 0.7-103g/L depending on the water levels, with a mean pH 8.9 (Timms, in press). The water provides habitat for aquatic macrophytes such as Myriophyllum verrucosum, Lepilaena bilocularis, and Chara spp., as well as abundance of invertebrates including Coxiella gilesi (Timms, in press), which was also encountered in the lake’s sediment core.

When visited during a dry stage, the lake floor was devoid of vegetation, except for samphires (Halosarcia spp.) along the lake margins. Remains of a woody shrub within the basin suggested a longer drying period in the past (Figure 7.5). The swale between the lunettes is probably flooded much less frequently than the lake, as it is densely covered by samphires (Halosarcia spp.) and slightly sparser Frankenia gracilis. The clay lunette is relatively densely covered with matured/older samphires and supports sparse young Eucalyptus trees (Figures 7.6,

130

7.7, and 7.8). The gypsum lunette, on the other hand, is nearly bare (Figure 7.9). The cliff side of the lake supports tree tobacco (Nicotiana glauca) with Casuarina sp. further away. Woody shrubs seem to be the dominant vegetation surrounding the lake on all sides.

Figure 7.2 The cliff on the western margin of the dry Mid Blue Lake featured in the transect in Figure 7.4.

Figure 7.3 Western cliff of the Mid Blue Lake’s basin. 131

Figure 7.4 Transect across Mid Blue Lake with relative positions of the cores analysed in this study. See map in Figure 7.1 for their exact location.

132

Figure 7.5 Dead shrub in the middle of Mid Blue Lake surrounded by an area of patchy, thin salt crust.

Figure 7.6 Mid Blue Lake: a view west from the top of the inner clay lunette. Large samphires in the foreground and smaller ones on the beach (mid-field).

133

Figure 7.7 Mid Blue Lake: a view west from the top of the inner clay lunette showing the low shrub cover of samphires and the soft (fluffy and powdery) clays in the foreground.

Figure 7.8 Mid Blue Lake: a view from clay lunette east toward the outer gypsum lunette in the far background. Large samphires in the foreground and low younger samphires visible as a brown stripe between the dunes (in the swale).

134

Figure 7.9 Augering of the outer gypsum lunette looking east with woody shrubs in the far background.

7.1.2 Results

The Mid Blue Lake sediments are relatively shallow: about 40cm, decreasing slightly in depth towards the western shore (Figures 7.4 and 7.10). The short record results from frequent/intense deflation events that striped the lake sediments to bedrock and deposited at least some of the material on the downwind shore to build a lunette. At least two major lunette building episodes (though each may consist of a few stages) have taken place as evidenced by large clay (inner - younger) and gypsum (outer - older) lunettes (Figures 7.1 and 7.4). The size of the lunettes is indicative of the volumes of sediment removed progressively from the lake in the past.

Four sediment cores including one lake core (core p1) and three augered sediment cores from the clay lunette, the gypsum lunette, and the swale between the dunes were subjected to detailed analysis. Another lake core (core p2) was subsampled for 137Caesium and OSL analysis. Figures 7.10-7.15 and sections 7.1.2.1- 8 present results for the lake core p1, while Figures 7.16-7.20 and sections 7.1.2.9-13 show results for the clay lunette auger hole, Figures 7.21-7.26 and sections 7.1.2.14-18 for the swale between the lunettes, and finally Figures 7.27-7.31 and sections 7.1.2.19-24 for the gypsum lunette.

135

Figure 7.10 Sediment description for Mid Blue Lake core p1.

136

7.1.2.1 Lake core p1: 137Caesium

The 137Cs terminates just above the red oxidised layer at about 15cm depth (Figure 7.10), thus linking the bottom of its profile with a transition from dry to wetter conditions. In spite of the net 137Cs value being nearly four times that of the reference (115.29mBq/cm2 ±11.93 compared to 29mBq/cm2 reference for Thargomindah), suggesting net sediment accumulation. Some loss of sediment from the lake can not be, however, ruled out, particularly that deflation of the lake basin was observed during the field work.

7.1.2.2 Lake core p1: gypsum

The lake sediment core from Mid Blue Lake shows two interesting gypsum features. The first one is a fragment of gypsite lamina bordered by grey clays within a nodule at 7cm (Figure 7.10 and Figure 7.11 image 1). At the time of the coring, the lake, and particularly the top 15cm layer, was very dry. The hammering of the PVC pipe into the lake floor destroyed the original structure of these dry sediments, breaking them into nodules of various sizes. Thus, it is likely that this gypsite lamina was deposited in at least a semi-continuous layer on top of the sediments over at least the deeper part of the lake. However, detailed crystal analysis and more extensive coring are required to confirm this conclusion.

Figure 7.11 Mid Blue core p1 gypsum: 1 – gypsite band at 7cm depth; 2 – rosette composed of lenticular gypsarenite crystals at 12cm.

The gypsum rosettes at a depth of 12cm (Figure 7.10 and Figure 7.11 image 2) suggests soil formation processes affected this part of the lake sediments (Magee, 1991) (also Chapter 4 section 4.9.1). The soil development might have also contributed to the destruction of the original sedimentary structure and the fragmentation of the gypsite lamina. 137

Aside from the features described above, the gypsum presence is suggested by SWIR spectra though most of the core (Figure 7.15) with the highest presence in 25 and 30cm samples. The lack of gypsum in the lowest 5cm is likely to result from dominance of weathering bedrock material in these samples (Figure 7.10). The overall gypsum content in the lake basin is lower than it was in the past as evidenced by the gypsum content of the lake’s source bordering dunes, i.e. the older lunette is richer in gypsum than the younger clay one.

7.1.2.3 Lake core p1: pollen

A pollen record is available for most of the Mid Blue core p1 (Figure 7.12). The lowest samples, 25-35cm, are dominated by Unknown type 1. Its presence within the sediments in the clay and gypsum lunettes supports its proposed link to extreme drying or dry lake floor stages resulting in dune formation (Figures 7.17 and 7.28).

The 10-20cm section of core p1 is characterised by the highest pollen concentrations with relatively stable proportions of low shrubs (dominated by Chenopodiaceae) and herbs and grasses. The surface sample, in turn, records not only a decrease in total pollen concentration but also a marked increase in the proportion of trees and shrubs accompanied by a decrease in mainly Chenopodiaceae and Asteraceae pollen. The latest trend can be due to one of, or a combination of, two factors: preferential pollen preservation (suggested by decreases in counts within this pollen category as well as in total pollen richness with depth) and/or increase in woody shrub presence (suggested by dominance of the Dodonaea pollen) contributing to reduction in ground cover plants with the exception of Poaceae, which tend to increase slightly.

7.1.2.4 Lake core p1: phytoliths

The phytolith record for Mid Blue Lake roughly overlaps in extent that of the pollen (Figure 7.13). There are three main peaks in total phytoliths: one at the surface, and the others at 15- 25cm and 35cm depths, suggesting times of increased plant cover. Both Poaceae and Dicotyledons/Monocotyledons (non-Poaceae) follow this trend with little changes in their proportions.

The most interesting feature of this record is the coincidence of drops in the total phytoliths with the peaks in Unknown type 1 pollen, with the both proxies indicating increased dryness of the lake and surrounding area.

7.1.2.5 Lake core p1: other biogenic silica

Freshwater sponge spicules are the only non-phytolith biogenic silica that occur in significant numbers in the lake sediments. The changes in their concentration are similar to these of phytoliths, further supporting the evidence of three wet phases in the lake core record.

138

Figure 7.12 Pollen counts for Mid Blue core p1.

139

Figure 7.13 Biogenic silica counts for Mid Blue core p1.

7.1.2.6 Lake core p1: other signs of plant and animal life

Plant fragments, seeds, and charophyte oospores occurred within the top 25cm of the lake core (Figure 7.10), i.e. most of the sediment overlying the weathered bedrock. The only identified charophyte species was Lamprothamnium heraldii, an indicator of variable depth and salinity conditions characteristic of the present regimes. The gastropode (Coxiella gilesi) and the ostracodes (Diacypris sp. and Trigonocypris globulosa) found at 17cm, all indicate at least hyposaline conditions, which are in tune with present salinity ranges recorded in the lake (Chapter 7 section 7.1.1).

A root-like mottling occurs at about 16-25cm depth with overlying red sediments intruding into underlying grey sediments (Figure 7.10), suggesting plant colonisation of the lake floor. However, since no roots or fine root patterns were observed, it is possible that these intrusions were caused by clay cracking.

7.1.2.7 Lake core p1: organic matter and carbonate content

The increased carbonate content is closely associated with peaks in organic matter as well as the presence of plant fragments and seeds (Figures 7.10 and 7.14), suggesting links with soil development.

140

Figure 7.14 Mid Blue Lake core p1: sediment diagram.

141

7.1.2.8 Lake core p1: magnetic susceptibility

Only one, but very distinctive, peak in magnetic susceptibility was recorded in the lake core p1 at 20cm depth (Figure 7.14). The peak reached a maximum value of 29.1x10-8m3/kg (nearly three times that of the ‘background’) and was accompanied by a high value of frequency dependent susceptibility of 10.6% (Figure 7.14). While lacking clearly defined roots or root tunnels, the unit is strongly affected by mottling including penetration of overlying red sediment into underlying grey sediment in a manner that can be associated with large root penetration. Slight peaks in organic matter and carbonate content at that depth further support a soil formation event.

Figure 7.15 Infrared spectrometry data (SWIR) for Mid Blue Lake core p1.

7.1.2.9 Clay lunette: gypsum

Gypsum was observed throughout the augered clay lunette sediments, except for the lowest reddish section (below about 412cm), which is similar to the lake sediments and below ~440cm is associated with weathering bedrock materials (Figures 7.16, 7.19, and 7.20). The gypsum

142

occurs in small to medium quantities mostly as non-laminated small gypsite conglomerates and individual gypsarenite crystals.

Figure 7.16 Sediment description for Mid Blue clay lunette auger hole.

143

7.1.2.10 Clay lunette: pollen

The dominance of Chenopodiaceae pollen in surface sediments of the clay lunette reflects the modern plant cover of mostly Halosarcia spp. (Figures 7.17 and 7.7).

The most important feature of these pollen data is, however, the nearly continuous presence of the Unknown type 1, which strengthens its position as indicator of drying/dry and increasingly saline lake conditions that favoured the lunette formation.

7.1.2.11 Clay lunette: phytoliths

The data show initial dominance of Poaceae phytoliths (at 425cm), but their numbers soon decrease in favour of the Dicotyledons/Monocotyledons (non-Poaceae) (Figure 7.18). The latter group dominates the record until a new peak at a depth of 75cm and another one at the surface. This suggests that the vegetation composition through most of the lunette’s formations was different from the modern one, with grasses playing a less significant role.

7.1.2.12 Clay lunette: other biogenic silica

An interesting feature of the clay lunette biogenic silica graphs is the coincidence of two peaks in the total phytolith counts (at 75 and 425cm) with those in Fossil 1 and freshwater sponge spicules (Figure 7.18). The peaks suggest a multi-stage formation of the dunes, rather than a single event. Each peak seems to represent reactivation of dune formation after a fluvial event, with an initial stripping of lake sediments rich in aquatic taxa followed by more mineral sediments sourced from drier environments. There seems to be also some association between the aquatic phase and the dominance of Poaceae phytoliths, and the dominance of Dicotyledons/Monocotyledons (non-Poaceae) and the drier interval.

144

Figure 7.17 Pollen counts for Mid Blue Lake inner clay lunette auger hole.

145

Figure 7.18 Biogenic silica counts for Mid Blue Lake inner clay lunette auger hole.

7.1.2.13 Clay lunette: magnetic susceptibility

Two peaks in magnetic susceptibility were recorded within the sediments of the clay lunette. The slight enhancement in the magnetic susceptibility and frequency dependent susceptibility close to the surface of the clay lunette (Figure 7.19) is most probably related to a limited soil development, considering the apparent presence of permanent plant cover (Figures 7.6, 7.7, and 7.8) together with peaks in carbonate and organic matter content (Figure 7.19). While the frequency dependent susceptibility is rather low (~4.2%) for a topsoil, it is possible that the pedogenic magnetic minerals are progressively diluted by the still active dune building processes.

The most interesting feature of the magnetic susceptibility for the clay lunette is, however, a small peak (17x10-8m3/kg at low frequency) loosely associated with buried reddish sediments (at about 425cm). It is additionally characterised by a slight increase in frequency dependent susceptibility (4.4%) and is supported by an increase in organic matter content (Figure 7.19). The enhanced magnetic susceptibility provides an additional link between this and similar sedimentary units in the swale and gypsum lunette (Figure 7.4), which are likely to represent the oldest sediments within the lake basin. 146

Figure 7.19 Mid Blue Lake inner clay lunette auger hole: sediment diagram.

147

Figure 7.20 Infrared spectrometry data (SWIR) for Mid Blue Lake inner clay lunette auger hole.

7.1.2.14 Swale: gypsum

The uppermost sediment in the swale between the clay and gypsum lunettes does not seem to contain gypsum (Figures 7.21, 7.25, and 7.26), which might be due to a low gypsum content of the deposited sediments (e.g. flood residue) and/or leaching of the gypsum by freshwater perched in the swale following a major lake filling event (Figure 7.4).

The swale contains the only selenite in the Mid Blue Lake basin. The simple lenticular and complex lenticular/pyramidal selenite at depths around 125cm increases to frequent well developed lenticular twins (mostly) at 175cm (Figures 7.21 and 7.22). All of these forms are closely related to precipitation from groundwater (Bowler and Teller, 1986; Magee, 1991; Teller et al., 1982). The absence of a gypsum signature in the SWIR data at 150cm (Figure 7.26) is likely to be due to the removal of selenite before the grinding of these samples.

148

Figure 7.21 Sediment description for Mid Blue swale between the lunettes (auger hole). Note: the photo’s scale is severely distorted by perspective.

At around 200cm the gypsum content in the swale sediments drops and no gypsum was visible at 225cm (Figures 7.21 and 7.22, and sediment description notes). The presence of weathering bedrock fragments highlights links with a similar situations encountered at the bottom of the clay lunette and lake core. An unusual feature is the presence of gypsum ‘agglomerates’ at a 250cm depth (Figures 7.21 and 7.22). J. Magee (pers. comm., 2007) noticed similar gypsum crystal forms (but much larger) in ancient lake terraces near Lake Eyre and interprets them as a product of multiple partial solution and overgrowth events, likely to result from combination of leaching and fluctuations in groundwater table.

149

Figure 7.22 Gypsum crystals from Mid Blue Lake swale between the clay and the gypsum lunettes: 1a & b – complex pyramidal to lenticular and simple lenticular selenite at 125cm depth; 2a & b – lenticular twin selenite crystals at 175cm; 3a & b - amorphous selenite at 250cm.

7.1.2.15 Swale: pollen

The pollen data are limited largely to the top of the sediment unit and reflects the modern vegetation dominated by Chenopodiaceae (i.e. mostly Halosarcia spp.) (Figure 7.23). The limited occurrence of Unknown type 1 in swale sediments (only the top 50cm) may be a result of secondary deposition of lunette material due to water (runoff) erosion or relatively recent colonisation of the swale by the pollen/spore’s parent plant.

150

Figure 7.23 Pollen counts for Mid Blue Lake swale between the clay and gypsum lunettes (auger hole).

151

7.1.2.16 Swale: phytoliths

Like the clay lunette, the swale shows initial dominance of Poaceae, which then decreases in proportion in favour of the Dicotyledon/Monocotyledon (non-Poaceae) phytoliths (Figure 7.24). A short-term grass recovery accompanied by a peak in total phytoliths occurs at a depth of 50cm, possibly indicating a temporary improvement of conditions (wetter phase) and associated boom in plant densities, including grasses.

Figure 7.24 Biogenic silica counts for Mid Blue Lake swale between the clay and gypsum lunettes (auger hole).

7.1.2.17 Swale: other biogenic silica

Similar to clay lunette sediments, the peaks in freshwater sponge spicule concentrations are accompanied by peaks in total phytoliths as well as an increased proportion of grasses (Figure 7.24), indicating a sequence of wet and dry episodes throughout the swale’s existence. Dating (radiocarbon/OSL) is, however, essential to establish any links between the sedimentary histories of the lunette and the swale.

152

Figure 7.25 Mid Blue Lake swale between the clay and gypsum lunettes (auger hole): sediment diagram.

153

7.1.2.18 Swale: magnetic susceptibility

The swale sediments are characterised by a single peak in the magnetic susceptibility at 150- 200cm depth (Figure 7.25). The peak is only slightly stronger than the one in the clay lunette (magnetic susceptibility: 18.8-16.2x10-8m3/kg at low frequency and frequency dependent susceptibility: 5.9-4.7%) and is located at the bottom of the reddish band (about 90-165cm), extending into the underlying greyer and finer sediments. It is possible the fine magnetic particles were leached to the bottom of the coarser layer and redeposited around the transition from the coarser to the finer sediments. The leaching would be facilitated by the perched position of the secondary basin (i.e. the swale) in relation to the floor of the main lake. Such position is likely to promote seepage of water retained within the higher basin into the lower lying lake.

Figure 7.26 Infrared spectrometry data (SWIR) for Mid Blue Lake swale between the clay and gypsum lunettes (auger hole).

154

7.1.2.19 Gypsum lunette: gypsum

Small gypsite conglomerates and gypsarenite crystals are present throughout the outer gypsum lunette (Figures 7.27, 7.30, and 7.31). They dominate the top ~125cm, but below this depth their content decreases markedly. Since the level of the change from white gypsum-rich to reddish gypsum-poorer sediments corresponds roughly to the top of the cliff height, it is possible that the sediments are roughly level with the surrounding landscape (further surveying would be necessary for confirmation). While an allowance has to be made for height differences related to a beach/quartz dune development, it is possible that the reddish sediment unit represents the most easterly shore in the history of the Mid Blue Lake.

Figure 7.27 Sediment description for Mid Blue Lake gypsum lunette auger hole. Note: the photo’s scale is severely distorted by perspective.

155

The lunette itself is also a sign of a much greater past prevalence of gypsum in the lake basin. The decrease in gypsum content of the lake sediments is evidenced by much lower gypsum content in the younger clay lunette.

7.1.2.20 Gypsum lunette: pollen

Very little pollen is present in the gypsum dune sediments (Figure 7.28). The most noteworthy is the presence of the Unknown type 1, which strengthens its links to drying/dry and saline lake floor conditions, required to formation of a source bordering dune (i.e. lunette).

Figure 7.28 Pollen counts for Mid Blue Lake outer gypsum lunette auger hole.

7.1.2.21 Gypsum lunette: phytoliths

There are three peaks in total phytolith counts at 100, 200, and 300cm depths (Figure 7.29). It is noteworthy that the increased phytolith concentrations are accompanied by peaks in the proportions of Poaceae. A decrease in total counts (at 150cm), on the other hand, is characterised by dominance of Dicotyledons/Monocotyledons (non-Poaceae). It implies at least two stages in the lunette’s formation with at least one dune stabilisation phase between them.

7.1.2.22 Gypsum lunette: other biogenic silica

The peaks in total phytoliths are closely associated with peaks in freshwater sponge spicules and Fossil 1. In addition, traces of diatoms were found in the two samples containing the highest phytolith concentrations (at 100 and 200cm). This evidence further strengthens the argument of multi-stage formation of the gypsum lunette suggested in the previous section. 156

Figure 7.29 Biogenic silica counts for Mid Blue Lake outer gypsum lunette auger hole.

7.1.2.23 Gypsum lunette: magnetic susceptibility

The magnetic susceptibility data for the gypsum lunette highlight the boundary between the whitish gypsum rich sediments above 125cm and the underlying red sediments (Figure 7.30). The latter unit is characterised not only by an enhanced magnetic susceptibility within this dune but also by the highest values of magnetic susceptibility (19.9-24x10-8m3/kg at low frequency) and frequency dependent susceptibility (6.4-10.2%) within the whole of the Mid Blue Lake lunette complex. This magnetic signature suggest links with soil formation, although it lacks support from other proxies such as rise in organic matter content and presence of pedogenic structures (although, the latter might have been destroyed by augering). Thus, the magnetic susceptibility can indicate either in situ soil development and/or accumulation of topsoil eroded from the surrounding area that formed a dune, which later provided core for the gypsum lunette.

While the surface gypsum lunette sample has a moderately high frequency dependent susceptibility of 7.6%, it is likely the high value is a result of bias from the low susceptibility values (4.5x10-8m3/kg for low frequency and 4.3x10-8m3/kg for high frequency measurements) rather than enhancement related to soil formation.

157

Figure 7.30 Mid Blue Lakes outer gypsum lunette auger hole: sediment diagram.

158

7.1.2.24 Gypsum lunette: particle size

The particle size data clearly support the boundary between two major sediment units at about 125cm (Figure 7.30). It is, however, possible that the high gypsum content is affecting the results, obscuring the differences between the aeolian and lacustrine units.

Figure 7.31 Infrared spectrometry data (SWIR) for Mid Blue Lake outer gypsum lunette auger hole.

7.1.3 Palaeoenvironmental reconstruction of events

Event A: The oldest (pre-gypsum lunette formation) lake stage

The oldest unconsolidated sediments within the lake basin seem to be the reddish sediments found at the bottom of the clay and gypsum lunettes (Figure 7.4), whose general shape conforms in cross section to a possible shoreline outline. Dating of these sediments is essential to confirm their relationship. If the theory is true, these sediments are likely to have formed the earlier eastern shore of the lake. It is also possible that the unit represents a stabilised and vegetated red sand dune that had accumulated in a past period of aridity. The biogenic silica evidence from that unit points to relatively wet conditions. The lake held water and the surrounding

159

landscape was well vegetated. The strong presence of Poaceae is likely to indicate the dominance of grasslands.

Event B: Increased aridity: gypsum lunette formation

The subsequent arid period resulted in periodic drying of the lake floor (evidenced e.g. by Unknown type 1), exposing it to deflation. The evaporites that were accumulating on the lake’s surface were rich in gypsum. Upon deflation, some of the material was deposited on the downwind eastern shore to form a lunette. The lunette’s formation occurred in at least two stages separated by a wetter phase (evidenced by freshwater sponge spicules and diatom traces), which saw a more permanent lake and increased vegetation cover of the surrounding landscape, including significant grass component.

Event C: Continued/new phase of aridity: clay lunette formation

A continued or new phase of aridity resulted in formation of a new lunette. The lake of this phase was probably reduced in extent compared to the past area, suggesting ever increasing aridity. It is, however, also possible that the lake’s size remained similar but the whole basin moved westwards following the erosion of the western cliff.

Another interesting change in the lake’s characteristics of this period is the apparent decline in the gypsum content of the lacustrine brine and/or the lake floor surface sediments. Since the lunette building processes themselves (i.e. clay efflorescence resulting in production of loose pellets) were still active, the reasons behind the gypsum deficiency must lie either in exhaustion within local sources (groundwater, bedrock), and/or hydrological changes (e.g. change in the chemistry of the water flowing into the lake or increased flushing frequencies), and/or greater influx of clays and silts into the lake diluting the existing gypsum.

Similar to the gypsum lunette, the clay lunette formation occurred in stages. The wet interludes were recorded in the dune sediments by peaks in total phytoliths, grasses, freshwater sponge spicules, and Fossil 1. The rich in aquatic taxa lacustrine sediments were blown out from the lake following the exposure of the lake floor during the more arid periods and trapped by the lunette. The dominance of Poaceae in these sediments suggests flourishing grasslands in the surrounding landscapes. The dry phases, on the other hand, were characterised not only by a general decrease in plant cover (reduced phytolith counts) but also increased importance of Dicotyledon/Monocotyledon (non-Poaceae) plants. The ephemeral state of the lake is also indicated by the presence of Unknown type 1.

It is likely that the clay lunette formation continues in modern times as evidenced by the personal observation of clay deflation from the lake floor during the dry stage, the unconsolidated state of the clay lunette surface at that time to a depth of at least 5cm, and 160

finally, the seemingly young age of the lake sediments. The salts necessary for the clay efflorescence are certainly present in the system, forming patchy, thin salt crusts on the dry lake surface (Figure 7.5), thus inadequate to act as protection, but enough to effect efflorescence. The gypsum presence is also confirmed by the infrared spectrometry data (Figure 7.15). A detailed 137Caesium study of the lake and the clay lunette sediments would be useful in confirming this theory.

Event D: Extreme aridity phase (core p1): a part of event C

A period of extreme aridity combined with high wind velocities resulted in complete stripping of sediment from the lake bed (i.e. exposure of the bedrock). Some of the deflated sediments are likely to be found on the downwind margin of the lake within the clay lunette.

Event E: Amelioration of conditions: new deposition phase (core p1 30-37cm)

An amelioration of conditions resulted in renewed sediment deposition within the lake basin. Strongly mottled grey and red sediments, dominated by silts and clays, suggest a mixture of lacustrine and aeolian sedimentation with a high possibility of exposure of the lake floor and colonisation by plants either periodically or in a single event toward the end of this phase. A decrease in phytoliths and freshwater sponge spicules, together with an increase in Unknown type 1 pollen, imply initially wetter conditions that later become dryer and more saline.

Event F: Lacustrine/dry saline? lake phase (core p1 20-30cm)

The sediment unit representing this phase provides contradictory evidence. The grey colour of the dominant sediments suggests a wet/aquatic depositional environment, thus permanent lake. However, the peak in Unknown type 1 pollen (related to lake drying/dry and possibly saline conditions), together with a dips in freshwater sponge spicules and phytolith concentrations (that suggest a decrease in wetness and vegetation cover), point toward a more arid, ephemeral lake and catchment conditions.

It is important to note here that the inconsistency of the results may have been caused by pre- sampling contamination that resulted from incursion of red sediments from the overlying red units into the sampled unit, for example, by filling cracks that formed during drying of lake muds. However, the lack of replication of the dry condition indicators/signals in the overlying red unit does not seem to support that explanation.

Event G: Drought 1890s-1940s? (core p1 16-20cm)

This sediment section provides also a curious mixture of evidence. The red colour of the unit implies oxidising shallow water to dry lake conditions. The coarser texture, peak in carbonate and organic matter content, and magnetic susceptibility and frequency dependent susceptibility

161

strongly hint at links with pedogenesis, possibly resulting from a mixture of in situ soil development and entrapment of topsoil eroded from devegetated surrounding areas.

A peak in total phytoliths as well as pollen concentrations suggests, on the other hand, increase in vegetation cover (including a large proportion of grasses) within the lake and/or catchment. A peak in Asteraceae counts as well as freshwater sponges points to generally wetter conditions and a waterlogged lake basin. However, the positioning of the unit just below the bottom of the 137Caesium profile attributes the unit formation to the first half of the 20th century, which was characterised by below average rainfall (Figure 2.5).

The most likely reason for these contradictions relate to the low sampling resolution across multiple fine couplets of lacustrine and dry sequences that occur within this unit. In such case, the unit would represent a phase of generally dry conditions interspersed with occasional flood events, resulting in brief booms in herbaceous vegetation.

Event H: Amelioration phase in the 1950s and the 1970s (core p1 ~12-16cm)

The renewed deposition of the grey, finely textured lacustrine muds accompanied by high pollen and phytolith concentrations, particularly at the beginning of the unit, suggest a return to wetter conditions with longer-lasting deeper lake levels and recovery in plant cover. Since the initiation of this unit coincides with the bottom of the 137Cs profile, its accumulation seems to have occurred in the above average rainfall period of the 1950s and the 1970s (Figure 2.5).

Event I: Gradual transition to increasingly dry/ephemeral conditions in post-1970s –modern period (core p1 0-~12cm)

The increased wetness did not last long and the upper part of the unit started to record signs of increased dryness (at about 10cm depth). Disappearance of freshwater sponge spicules, together with an increase in sediment redness and coarse particle content, suggest an increased exposure and dryness of the lake floor, thus a reduction in water depths and filling frequencies as well as an increased contribution of airborne particles to the lake’s floor sediments. The phytolith numbers decrease rapidly, suggesting a decrease in vegetation cover in the surrounding area and thus confirming the increased aridity. The pollen concentrations, on the other hand, are still high with Chenopodiaceae and Asteraceae dominating the record, although, the reappearance of Unknown type 1 hints strongly at increased dryness and salinity. The gypsite lamina at about 7cm further supports the increased dryness as it indicates a recent relatively rapid lake drying event (Warren, 1982).

Furthermore, the gypsum rosettes at a depth of 12cm, together with increased carbonate and organic matter content within the top sediments and the presence of a dead medium-sized shrub in the middle of the lake basin, suggest plant colonisation of the lake floor and in situ soil 162

development. This, in turn, suggests a prolonged dry period in the very recent past. Unfortunately, the plant roots from that event seem to be also responsible for the destruction of the original sedimentary structures in the upper section of the sediment, including the fragmentation of the gypsite lamina, thus affecting the resolution and sequencing of the record.

Aside from increased dryness, the most recent times witnessed also a major change in vegetation composition with increased dominance in trees and shrubs, and particularly the woody weed species dominated by Dodonaea with a minor contribution from Myoporaceae. The ground cover becomes dominated by grasses, which show an increase in both the pollen and phytolith records.

The timing of the gradual drying and increased ephemerality is likely to have occurred in the period of post-1970s, when the rainfall values fluctuated around average values (Figure 2.5).

7.2 Lake Wombah

7.2.1 Site and lake overview

Lake Wombah is the largest lake in the Rockwell-Wombah system, which when full extends over an area of 740ha and reaches depths of up to 2.3m (Timms, 2006). While there is no clear indication on the geological maps, there is a good possibility of tectonic origins of this lake. The presence of a fault line is implied by the straight line and high relief of the Tertiary silcrete cliff (up to 7.5m above the lake level) on the western and northern side the lake (Figure 7.32), combined with a distinct change in geology (B. Timms, pers. comm. 2007; Eulo 1:250 000 Geological Series Map Sheet SH 55-1, 1971)). The considerable amount of talus at the bottom of the slope (Figure 7.33) hints at the importance of southeasterly winds on the lake’s geomorphology during high water level periods.

A set of degraded inner and outer lunettes are present on the eastern shore (Figure 7.32) (Timms, 2006). The outer lunette is gypsum-rich, while the inner lunette, similar to Mid Blue Lake, is mostly composed of clay (B. Timms, pers. comm. 2007). Unlike the Mid Blue Lake clay lunette, however, it also contains a high content of sand due to its modern function of a beach shoreline (B. Timms, pers. comm. 2007). On a larger scale, Lake Wombah lies within an extensive red sand plain bordered south of the lake by Tertiary silcrete outcrops, which are surrounded by Cretaceous sediments overlaid by Quaternary talus (mostly silcrete gravel) (Eulo 1:250 000 Geological Series Map Sheet SH 55-1, 1971 and Yantabulla 1:250 000 Geological Series Map Sheet SH 55-5, ca. 1963).

163

Figure 7.32 Map of Lake Wombah. (Main image source: Brindingabba airphoto 54, 2002; Inset image source: Natmap Raster 2003; Source of geomorphological information: B. Timms, pers. comm. 2007)

Figure 7.33 The top of the rocky cliff with a thin layer of red sandy loam (foreground) on the northern margin of Lake Wombah with the beach and drying lake in the background. 164

In contrast to the Blue Lakes, the Number 10 Creek is insignificant in filling Lake Wombah as it becomes obliterated by dunes southwest of Lake Bulla and reappears near Wombah only as a small stream (Timms, in press). The main filling agent for Lake Wombah is the Paroo River floodwater, which enters and later leaves the lake by the same route, very rarely reaching as far as Lake Bulla (Timms, 2006, in press). Because of the different water sources, the timing of the filling and drying episodes for the Blue Lakes and Lake Wombah can be different as the first responds solely to the local rainfall and the other relies mostly on rain in the upper part of the Paroo River catchment (Timms, 2006). The lake’s salinity fluctuates from subsaline (1.2g/L) to mesosaline (30g/L) with a mean pH value of 9 (Timms, in press).

The filled lake provides habitat for aquatic macrophytes such as Myriophyllum verrucosum, Lepilaena bilocularis and Chara spp. (Timms, in press). The lake’s margins are inhabited mainly by samphires (Halosarcia spp.) and the surrounding landscape is dominated by woody shrubs and mulga (Acacia aneura).

7.2.2 Results

Due to limited access, only sediments on the northwestern margin of the lake were cored in this study. Core 3 was chosen for detailed analysis, including 137Caesium, as it contained the clearest stratigraphy. The sample for OSL dating was extracted from core 2, which was nearly identical to core 3. The general description of core 3 stratigraphy is presented in Figure 7.34, the pollen and biogenic silica data in Figures 7.35 and 7.36 respectively and other sedimentary data in Figures 7.37 and 7.38. Additional descriptions are provided for the selected main proxies.

7.2.2.1 137Caesium

Termination of the 137Cs profile in Lake Wombah seems to occur near the base of a mottled grey and red sediment at 21cm and not far above an orange sediment band at about 24-26cm. The total 137Cs value of 193.00mBq/cm2 ±15.81 is the highest of all the lakes analysed for 137Cs in this study (nearly seven times that of the reference value of 29mBq/cm2). It is possible that the high 137Cs content is related to preferential/selective delivery by the floodwaters of finer (clay) particles, which are the main ‘carriers’ of 137Cs. The coarser particles are more likely to settle out of the floodwater before reaching the lake, particularly since there is no clearly defined channel connecting Lake Wombah with the Paroo River.

165

Figure 7.34 Sediment description for Lake Wombah core 3.

7.2.2.2 Pollen

The pollen record of Lake Wombah is richest in the top 40cm (Figure 7.35). The lower depths are dominated by the Unknown type 1, which is also the only type that persists in interpretable numbers below the 40cm depth. At 30cm depth the Unknown type 1 is replaced by a mixture of Chenopodiaceae, herbs and grasses, and trees and shrubs pollen. At 25cm the herbs and grasses (mainly Tubuliflorae and Liguliflorae Asteraceae types, Brassicaceae, and Poaceae) reach their peak numbers and decrease gradually towards the surface. A reverse trend occurs in Chenopodiaceae and, to a much smaller degree, in the trees and shrubs pollen, whose counts gradually increase towards the surface.

166

Figure 7.35 Pollen counts for Lake Wombah core 3.

167

7.2.2.3 Phytoliths and other biogenic silica

The interpretable phytolith record is limited to the top 30cm, with an exception of an additional sample at 70cm (Figure 7.36). An outstanding feature of the record are two peaks in total phytolith numbers at 70 and 20cm, which are coincident with increases in concentrations of freshwater sponge spicules (from rare to occasional) and at 20cm also with a trace presence of Fossil 1. It is also noteworthy that while for most of the record the Poaceae to Dicotyledons/Monocotyledons (non-Poaceae) ratio is relatively stable, there is a sharp decline in Poaceae at 5cm depth followed by a slight recovery within the surface sediments.

Figure 7.36 Biogenic silica counts for Lake Wombah core 3.

7.2.2.4 Other signs of plant and animal life

Plant fragments and seed were found in most of the samples within the top 30cm of the core (Figure 7.34). This distribution mirrors that of the pollen and phytoliths. Charophyte oospores,

168

on the other hand, were recovered from a single sample at a depth of 25cm. All three oospores belonged to the Lamprothamnium heraldii species, which indicates variable depth and salinity conditions existing at the site in modern times.

Fine plant roots were present in the core within the 12-33cm section, probably indicating prolonged exposure of the lake floor at this site allowing plant invasion (most likely dominated by Halosarcia spp., which are the main species inhabiting the present shore). In addition, a root-like mottling was observed at a depth of 54-72cm.

7.2.2.5 Magnetic susceptibility

The Lake Wombah core 3 is characterised by two distinct peaks in magnetic susceptibility: at about 16-25cm and 55-73cm (Figure 7.37). The upper peak is slightly weaker than the lower one and is accompanied by peaks in clay content and occasional to considerable presence of fine roots (Figure 7.34). In contrast, the lower peak is associated with a secondary peak in sand- sized sediments. Both peaks show higher frequency dependent susceptibility (4.8% for the upper and 6.3% for the lower peak) and are related to reddish units divided by an olive coloured unit. The lower red unit is criss-crossed by a network of fine cracks filled with the overlying olive sediment and relatively densely speckled with unidentified black particles. There are also larger downward oriented inclusions of the olive sediment into the reddish sediments, possibly caused by bioturbation. The overall evidence suggests for both units in situ soil development accompanied by dry lake floor conditions, although, the lower unit is likely to contain also topsoil blown into the lake from the surrounding area.

7.2.2.6 Gypsum

In spite of the evidence of the past gypsum importance in the form of the gypsum lunettes (Figure 7.32), its presence in the cored sediments is limited to observation of occasional gypsarenite crystals at a depth of 7-12cm (Figures 7.34 and 7.37). No gypsum signature showed in the infrared spectrum (Figure 7.38), although its absence in the upper sediment might be due to weakening of the absorption features by sulphates (Pontual et al., 1997), rather than its absence.

169

Figure 7.37 Lake Wombah core 3: sediment diagram.

170

Figure 7.38 Infrared spectrometry data (SWIR) for Lake Wombah core 3.

7.2.3 Palaeoenvironmental reconstruction of events

Event A: Gypsum and clay lunette formation

The lack of grey-red couplet structure that is diagnostic of extreme seasonality necessary for lunette formation (see Palaeolake data in Chapter 9 section 9.3 for more details about this relationship), together with a gypsum deficit in the lake sediments, suggest that the lunettes have formed in earlier times than those represented by the lake core sediment. Considering the location of the cores (Figure 7.32), it is also possible that the coring site at the time of the lunette formation was still northwest of the cliff face (i.e. the cliff has not eroded that far). Thus, the presence of the gypsum lunettes extends the lake’s history beyond the record of the cored sediments.

171

Furthermore, it is probable that the material for the clay lunette came from the sediment unit between about 33-65cm as the presence of Unknown type 1 suggests ephemeral conditions, possibly conducive to the formation of a source bordering dune.

Event B: Mild dry phase (~58-73cm)

The unit represents soil (supported by magnetic susceptibility and frequency dependent susceptibility) that has developed either on talus or aeolian deposits of sand and silt or a mixture of the two. The plant cover of the site (evidenced by roots and increased total phytoliths) contains a high proportion of grasses. The increased concentrations of freshwater sponge spicules might have resulted from seasonal/periodic waterlogging of the site or the spicules have been blown from a pond in a deeper part of the lake’s basin.

Event C: Wetter phase (~33.5-58cm)

The grey colour of the sediments and low magnetic susceptibility are likely to indicate permanent lacustrine conditions during this phase. The high sand content is probably also related to the higher water levels that would be responsible for the reworking and redistribution of the talus material as well as further erosion of the cliff.

The presence of Unknown type 1 throughout that unit seems to contradict that theory. It is possible, however, that the small numbers in the lower part of the unit reflect alternative sources (e.g. extra-basin or shoreline). What is clear is the rise in Unknown type 1 concentrations towards the top of the unit that indicates increasing dryness and heralds a transition towards ephemeral conditions.

Event D: Ephemeral lake phase including the wetter periods of the 1950s and the 1970s (~6- 33.5cm)

The unit is composed of highly mottled grey and red sediments suggesting fluctuations between lacustrine and dry oxidising conditions. The lack of clear stratification might be due to high levels of bioturbation from, for example, bottom feeding carp Cyprinus carpio and birds (particularly waders) (Timms, in press) combined with the relatively shallow depths within the lake.

In spite of the disturbance, some general trends stand out. The first one is the presence of two wetter phases within this period (at ~10 and 20-25cm depth), which supported enhanced plant growth (evidenced by peaks in both pollen and phytolith concentrations) and increased importance of herbs and grasses. The evidence of the earlier wet phase is additionally supported by an increase in freshwater sponge spicule concentration and the occurrence of Fossil 1. Based

172

on the 137Cs and the rainfall data (Figure 2.5), the wet phases are likely to be linked with two main phases of above average rainfall in the 1950s and the 1970s.

The drier conditions prior to the wet phases, as well as during the interval between them, are indicated by an increase in the proportion of Chenopodiaceae as well as a slight decrease in both pollen and phytolith concentrations. An enhanced magnetic susceptibility and frequency dependent susceptibility further support the increased aridity and/or ephemerality of the lake, hinting at a soil signal. The presence of roots is likely to indicate the invasion of the dry lake’s floor by plants and in situ soil development. However, some of the magnetic enhancement could have resulted also from the influx of topsoil that became exposed during the dry phases and was eroded at that time by wind or later, at the onset of a subsequent wet period, by runoff. Both mechanisms point to increased aridity of the catchment.

Event E: Increased aridity and increase in woody shrub cover in post-1970s to modern (~0- 6cm)

Decrease in the total pollen and phytolith concentrations, together with increases in Chenopodiaceae proportions and sediment salinity, indicate increasingly arid conditions towards the top of the lake’s sedimentary history. Because of the close links between the Paroo River flooding and the lake filling events, the changes imply increased dryness not only on a local scale but also hold implications for the Paroo catchment (particularly in the headwater area), suggesting a larger scale aridity.

The last few decades seem also to reflect a major change in vegetation composition. The phytolith evidence points to a major shift from grass to Dicotyledon/Monocotyledon (non- Poaceae) plants, while the pollen shows increases not only in Chenopodiaceae, but also Dodonaea, supporting claims about the recent expansion of woody shrub communities in the region.

173

174

Chapter 8

Cuttaburra Channels: Cummeroo Waterhole

8.1 Cummeroo Waterhole

8.1.1 Site overview

The Cummeroo Waterhole is a deep permanent freshwater waterhole located along the Cuttaburra Creek (Figure 8.1) and it was most likely formed by vertical erosion (evorsion) of the channel floor during floods (Timms, 1997b). The Cuttaburra Creek is a complex system of anabranching streams and channels, waterholes and swamps linking together the Warrego and Paroo Rivers. It is supplied with water by the Warrego River and usually most of the flow is absorbed by the Yantabulla Swamp (just downstream from the Cummeroo Waterhole). During large floods, however, the water, after filling the swamp, can continue down the Cuttaburra channels to join the Paroo River near Wanaaring (Kingsford and Porter, 1999). The Cuttaburra channels are dominated by sandy red earths and grey clays (Morgan and Terrey, 1992).

The banks of the channels around the Cummeroo Waterhole are lined with river cooba (Acacia stenophylla), black box (Eucalyptus largiflorens), bimble box (E. populnea) and lignum (Muehlenbeckia florulenta) (Figures 8.2 and 8.3) (Kingsford and Porter, 1999 and personal observation).

8.1.2 Results

Two cores from the middle of the Cummeroo Waterhole were analysed in this study. Detailed analysis was carried out on core 4 and its general description is shown in Figure 8.4. The pollen and biogenic silica results are presented in Figures 8.5 and 8.6 respectively, while other sedimentary data are illustrated in Figures 8.7 and 8.8. Core 3 was subsampled for 137Caesium and OSL dating.

175

Figure 8.1 The location of the Cummeroo Waterhole and cores 3 and 4. (Main image source: Yantabulla airphoto 61, 2003; Inset image source: Google Earth, 2006)

Figure 8.2 Cummeroo Waterhole lined with black box and bimble box.

176

Figure 8.3 Dry channel southeast of the Cummeroo Waterhole.

A most notable feature of the recovered sediments, which shows clearly in both cores, is a division into two very different sections by a gravel band at a depth of 32-37cm (Figure 8.4). As illustrated below, the sections vary in most aspects: from their fossil content to physical properties.

8.1.2.1 137Caesium

No 137Cs was detected in the sediments of Cummeroo Waterhole (Figure 8.4). This may be due to a combination of substantial stripping of the channel bed during a high flood event in times following the main 137Cs deposition period and post flood deposition of sediments with no or highly diluted 137Cs content. The Yantabulla swamp downstream is a likely 137Cs depository in this system, but needs confirmation by future 137Cs study. Considering that the peak fallout in Australia occurred in 1960s (Longmore et al., 1986), the super-floods that occurred after that date, i.e. in the 1974 and 1976, (Timms, 1997b, Bureau of Meteorology, 2005) are the most likely to be responsible for erosion of the channel bed.

8.1.2.2 Pollen

The pollen record in the top 30cm of the sediment core (Figure 8.5) is dominated by trees and shrubs, and particularly Eucalyptus taxa, with a large component of herbs and grasses pollen, and Chenopodiaceae present in much smaller numbers. While uncommon, the taxa balance is not surprising, since the Cummeroo Waterhole is the freshest of the wetlands included in this study, and the Cuttaburra Channels are densely populated with Eucalyptus spp. (Figure 8.2).

177

Figure 8.4 Sediment description for Cummeroo Waterhole core 4.

178

Figure 8.5 Pollen counts for Cummeroo Waterhole core 4.

179

The pollen diagram indicates three main trends of vegetation changes. First, there is a gradual increase in Eucalyptus concentrations, possibly indicating a gradual recovery of the trees following flood damage. Poaceae, Muehlenbeckia, and Cyperaceae counts are at their peak at 30cm, but considerable counts of other herbs (including Asteraceae) also occur at this depth and might indicate initial colonisation after the flood. Asteraceae reach their peak concentrations at 20cm and from then on there is a gradual decline in all groups, except for trees and shrubs. It is possible that this decline is caused by gradually increasing shading of the ground by the recovering Eucalyptus canopy, grazing, and/or increasing aridity.

8.1.2.3 Phytoliths

The phytolith record for Cummeroo Waterhole is limited to three samples from a 10-30cm depth section (Figure 8.6). The total phytolith numbers increase gradually towards the top of the profile, i.e. 10cm depth, to decrease again at the surface. This trend is followed by both Poaceae and Dicotyledon/Monocotyledon (non-Poaceae) counts. The highest phytolith concentrations overlap those of pollen, suggesting higher biomass during this period.

Figure 8.6 Biogenic silica counts Cummeroo Waterhole core 4.

180

8.1.2.4 Other biogenic silica

While the freshwater sponge spicules are present all through the core, they reach peak numbers (abundant to frequent) at a depth of 10-30cm (Figure 8.6), thus coinciding with the peak phytolith and pollen concentrations.

8.1.2.5 Other signs of plant and animal life

Like phytoliths, plant fragments were recovered only from a depth of 10-30cm (Figure 8.4). Only one sample (at 20cm depth) produced charophyte oospores, which belonged to two taxa: Lamprothamnium sp. and Nitella cf. verticillata, both indicative of brackish conditions. The low number of the recovered charophytes, makes them, however, unreliable indicators of changes.

While the pollen, phytoliths, and plant fragments are all limited to the upper half of the core, the sediment in the lower half is characterised by the presence of occasional fine roots and root tunnels as well as solidified, iron and carbonate coated rootlet tunnels and/or fauna burrows (Figure 8.4). None of those are present in the upper half of the core.

8.1.2.6 Organic matter and carbonate content

The organic matter further highlights the difference between the two sections of the core divided by the gravel band with the sediment above it markedly richer in the organics than the layer below it (Figure 8.7). Also the carbonate content is slightly higher in the upper section than the lower one.

8.1.2.7 Magnetic susceptibility

The magnetic susceptibility in the Cummeroo Waterhole core also reflects the difference between the two main sediment sections with the lower unit characterised by stronger magnetic susceptibility values than the upper one (Figure 8.7). In contrast, the frequency dependent susceptibility values for both units are comparable with 2.9% for the upper and 3.4 and 2.9% values for the lower unit. The higher magnetic susceptibility of the lower unit, combined with the relatively low frequency dependent susceptibility, suggests that the magnetic enhancement is related to the red mottling present below 40cm and indicates only a very weak or absence of pedogenesis.

8.1.2.8 Particle size

The upper part of the core (above the gravel band) is dominated by clays and silts, while the lower part is much more coarsely textured (Figure 8.7). This suggests different sediment sources with the upper one more likely to represent mostly aquatic (from water deposition) and the lower one possibly containing a large proportion of aeolian sediments.

181

Figure 8.7 Cummeroo Waterhole core 4: sediment diagram.

182

8.1.2.9 Mineral composition: clay content

It is interesting to note that the infrared spectrometry is the only proxy not showing any differences between the upper and the lower section of the core, suggesting that the general sediment source area as well as clay diagenetic processes were similar throughout the waterhole’s history (Figure 8.8).

Figure 8.8 Infrared spectrometry data (SWIR) for Cummeroo Waterhole core 4.

8.1.3 Palaeoenvironmental reconstruction of events

Event A: Prolonged dry conditions in the late 1890s to the 1940s? (~37-68cm)

The sediments in the lower half of the core preserved a record of drier conditions, which possibly extended over a longer time period (as implied by the thickness of the sediment unit). The grey colour of the unit suggests waterlogging, while the red mottling implies frequent periodic drying or influx of oxidised materials. The presence of roots and possible fauna burrows within this section, as well as enhanced magnetic susceptibility, supports the occurrence of dry episodes, which allowed colonisation by plants, or at least prevalence of shallow water conditions. On the other hand, the low frequency dependent susceptibility suggests only weak soil development.

183

The over 60% sand content supported by red mottling within this unit suggests an increased influx of aeolian materials that would have been mobilised in the surrounding sandplains under conditions of increased aridity, and possibly also windiness, as well as decreased plant cover.

The timing of the event is uncertain. However, considering the rainfall data (Figure 2.5) and the timing of the flood event directly overlying the unit (described in more detail in the next section), it is likely the event took place during the below average rainfall period in the late 1890s to the 1940s.

Event B: Large flood event: the 1970s (~32-37cm)

This event is suggested by the gravel band, which is also a boundary between two very different sediment units. The gravel is likely to indicate a major erosional (evorsional) event effected by a large flood. The floodwaters could have stripped the channel bed of sediments and brought in gravel, which would have settled in the deeper, thus slower/lower energy parts of the channel, i.e. waterholes.

The most likely to effect these changes is the most recent large flood, which has occurred in the mid 1970s. While the flood of the 1950s was considered, the lack of 137Cs in the waterhole makes it unlikely. Considering that the peak fallout in Australia occurred in the 1960s (Longmore et al., 1986), any sediments that accumulated during that period would be expected to contain some 137Cs. Its absence suggests a removal of the sediment by a later flood, i.e. in the 1970s.

Event C: Wet phase: the 1970s to the early 1990s? (~10-37cm)

The wet conditions of the 1970s produced initially a boom in grasses and aquatic plants (mainly Cyperaceae and lignum Muehlenbeckia), which later decreased in significance in favour of Asteraceae and to a lesser extent Chenopodiaceae. The slow but steady recovery of the Eucalyptus from flood damage is suggested by its gradually increasing pollen contribution. The increased biomass production at that time is reflected by an increase in pollen and phytolith concentrations and an increased recovery of plant fragments. The low magnetic susceptibility and frequency dependent susceptibility values also suggest low contributions of topsoil sediments, thus a more stable catchment, probably due to increased vegetation cover.

Within the waterhole itself, the grey muds suggest low energy aquatic depositional environments. The increased wetness of the system is also evidenced by a peak in populations of freshwater sponges.

184

Event D: Increased aridity: the 1990s to present? (0- ~10cm)

The selection of the 1990s as the transition period between Events C and D was based on a shift towards a lower rainfall regime (Figure 2.5). In the sediment, there are several clues of an increased aridity in the last few years. The Asteraceae and Poaceae gradually decline and the Cyperaceae completely disappears from the record. The decrease in both pollen and phytolith concentrations suggests a decline in plant ground cover. The pollen record becomes more and more dominated by Eucalyptus, which, with its extensive root system, is likely to be the last plant taxon to be affected by the increasing dryness. Furthermore, the increased sediment salinity and decreased freshwater sponge populations are likely to indicate a gradual drying of the waterhole.

185

186

Chapter 9

Bloodwood Station lakes: Lower Bell Lake and Palaeolake

9.1 Bloodwood Station lakes

The Bloodwood Station hosts a complex system of ephemeral wetlands (Figure 9.1) embedded within plains and low undulating hills composed of aeolian red sand and silt (Yantabulla 1:250 000 Geological Series Map Sheet SH 55-5, ca. 1963). The wetlands range in size (from <1 to 820ha in extent and 20-220cm in depth) and salinity (from fresh to hypersaline) (Timms, 1997a). The system derives water exclusively from local rainfall and runoff.

9.2 Lower Bell Lake

9.2.1 Site and lake overview

The Lower Bell Lake formed as a consequence of blockage of Bells Creek by a large red sand dune, which arrived from the northwest (Figures 9.1 and 9.2) (Timms, 2006). While the lake is relatively large (about 185ha when full), it is very shallow (about 30cm) and can remain dry for periods ranging from many months to years (Timms, 2006). The lake is filled mainly by Bells Creek, but some water might flow also from Horseshoe Lake, which is filled by Barton Creek (Figure 9.1) (Kingsford and Porter, 1999).

The eastern margin of Lower Bell Lake supports a system of minor beach and dune ridges and is bordered by an 8m high gypsum lunette (Figure 9.1) (Timms, 2006). The bar across the inlet of Bells Creek and the gypsum mounds through the middle of the lake might be an indication of a development of new, lower shorelines associated with shrinking of the lake (B. Timms, pers. comm. 2007). Gypsum-rich sediments extend all the way to the western edge of the Palaeolake and are covered by 10-60cm of aeolian sand and silt.

The salinity in the Lower Bell can vary from hyposaline to hypersaline depending on the amount of water present (6.7 to 122.9g/L recorded by (Timms, 1998a). The turbidity is very low and mean pH is 8.9 (Timms, 1993).

187

Figure 9.1 Map of Lower Bell Lake and Palaeolake. (Main image source: Tinchelooka airphoto 339, 1997; Inset image source: Google Earth, 2006; Lower Bell Lake geomorphological information after Timms, 2006)

188

Figure 9.2 Red sand dune northwest of Lower Bell Lake with Callitris glaucophylla in the foreground (right) and mulga in the swale behind the dune.

When full, the Lower Bell Lake may support submerged plants, possibly Ruppia megacarpa (Kingsford et al., 1994). The lake’s margins are dominated by samphires (Halosarcia spp.) with occasional saltbushes (i.e. other Chenopodiaceae) (Kingsford et al., 1994). The gypsum lunette supports belah (Casuarina cristata) and occasional Eucalyptus species. Woody shrubs, dominated by turpentine bush (Eremophila sturtii), hop bush (Dodonaea viscosa), and rice flower (Pimelea microcephala), are present in lower numbers on the gypsum-rich sediments between Lower Bell and Palaeolake. The crests of the red sand dunes, northwest of the lake, support stands of Callitris glaucophylla with mulga (Acacia aneura) in the swales (Figure 9.2). A mixture of mulga and woody shrub communities are common on the red sandplains surrounding the lake.

9.2.2 Results

Due to a combination of technical and access problems only one core (G) was recovered from Lower Bell Lake. The general description of the core is presented in Figure 9.3 and a selection of gypsum crystals is shown by Figures 9.4 and 9.5. Pollen and biogenic silica data are illustrated by Figures 9.7 and 9.8 respectively and other sedimentary data are presented in Figures 9.9 and 9.10. Supportive descriptions are provided for a selection of the main proxies.

189

Figure 9.3 Sediment description of Lower Bell Lake core G.

9.2.2.1 Gypsum

The Lower Bell core has two general sections of gypsum occurrence: 0-30cm and 80-165cm (Figures 9.3, 9.9, and 9.10). Most of the selenite and gypsarenite occur as pyramidal to 190

lenticular crystals (Figures 9.4 and 9.5) set within the matrix of the lake sediments (e.g. Figure 9.6), implying precipitation from groundwater (Bowler and Teller, 1986; Magee, 1991; Teller et al., 1982).

Figure 9.4 Lower Bell Lake gypsum: 1 – gypsite band at 6.5cm depth; 2 – gypsite band at 122.3cm; 3a & b – pyramidal to lenticular selenite and gypsarenite with minor signs of dissolution at 83cm; 4a & b – mostly pyramidal to lenticular selenite and gypsarenite at 109cm.

191

Figure 9.5 Lower Bell Lake gypsum: 1a & b – pyramidal to lenticular (single and twin) selenite at 128-131cm depth; 2a & b – pyramidal to semi-prismatic selenite at 132- 136cm; 3a & b – lenticular (single and twin) selenite at 154-160cm.

In most cases, the higher gypsum concentrations are not associated with specific bands/laminae of the sediment (e.g. grey clay or red silty clay), but tend to transgress the boundaries (e.g. Figure 9.6). The general increase in crystal size and density of the gypsum towards the bottom of the core suggests slower and longer crystal formation. However, since it is currently

192

impossible to determine the sequence of groundwater gypsum precipitation or to time the events, the information is of little value in construction of palaeoenvironmental interpretation.

Figure 9.6 Selenite in 129-135cm depth section of the Lower Bell Lake core G.

There are two very distinct gypsite laminae at 6.5cm and 122.3cm (Figure 9.3). In addition, possible traces of very fragmented gypsite laminations occur within the zone of fine grey-red couplets (~78-126cm). Both of the main gypsite laminae lie towards the top of major grey units overlaid by oxidated red sediments, thus suggesting their deposition during a transition from lacustrine to drier lake phases.

The upper gypsite lamina (6.5cm) is underlain by grey sediments heavily mottled with gypsite. The gypsite content decreases with depth. The high gypsum content within the grey sediments is supported by a very strong gypsum signature in the infrared spectrum in this section (Figure 9.10). The reasons for the mottling are, however, unclear.

9.2.2.2 Other crystals: white balls

At depths of 10 and 30cm, the core contained unidentified white crystalline balls that were recovered from pollen samples (Figure 9.3). Since they were also found in the gypsum lunette sediments at Palaeolake, they could be providing another tool for detecting the type of lacustrine sediments that contribute to gypsum lunette formation.

193

9.2.2.3 Pollen

The pollen record in the Lower Bell Lake core is limited to the top 10cm and is dominated by the Chenopodiaceae (Figure 9.7). This is not surprising considering that large areas of the Lower Bell’s basin are well covered by Halosarcia species. The presence of the Unknown type 1 in the record further underlines its links with drying lake conditions and possibly also enhanced lake salinity.

9.2.2.4 Phytoliths

The phytolith record consists of two sections (at depths of 0-5cm and 137-150cm) and an isolated sample at 110cm depth (Figure 9.8). Aside from an increase in Dicotyledon/Monocotyledon (non-Poaceae) phytoliths in comparison to the Poaceae at the 110cm depth, the ratio between the two groups remains similar in the top and bottom sections of the core.

9.2.2.5 Other biogenic silica

Except for two samples (at 50 and 80cm depth) from which they are absent, the freshwater sponge spicules are present throughout the whole core as rare or occasional components (Figure 9.8). Their peak concentrations occur in the 137-150cm section, coinciding with one of the peaks in total phytolith count as well as traces of diatoms at 150cm.

9.2.2.6 Other signs of plant and animal life

Small amounts of plant fragments were found only in the upper two samples, at depths of 0 and 5cm, while seeds were limited to the surface sediments (Figure 9.3). Charophyte oospores were recovered between 0 and 10cm. They belonged to three taxa: Chara fibrosa, Lamprothamnium heraldii, and Nitella cf. verticillata, all of which indicate saline conditions ranging from subsaline (particularly Chara) to hypersaline (Lamprothamnium). These salinities fit within the salinity range recorded in the Lower Bell Lake in modern times (section 9.2.1 above).

Sparse rootlets and root tunnels were found in two sections: 0-36cm and 55-92cm with a section of frequent fine roots between them (i.e. within 36-55cm) (Figure 9.3).

194

Figure 9.7 Pollen counts for Lower Bell Lake core G.

195

Figure 9.8 Biogenic silica counts for Lower Bell Lake core G.

9.2.2.7 Magnetic susceptibility

Lower Bell Lake core’s G magnetic susceptibility data showed five main peaks (Figure 9.9). The peaks seem to be related mainly to a combination of sediment redness and gypsum content, and less consistently to the peaks in the sand content. The most distinctive peaks (at ~5cm, 75cm, and 143cm) appear to correspond to the three thickest and reddest sediment bands. The other sections of the magnetic susceptibility enhancement also tend to occur in core sections with increased redness (mostly in sections dominated by the red-grey couplets) (Figure 9.3).

196

Figure 9.9 Lower Bell Lake core G: sediment diagram. 197

Low magnetic susceptibility, on the other hand, tends to occur in areas with enhanced gypsum content (Figure 9.9). The gypsum impact can be twofold. Firstly (and most significantly), it can act as a diluting agent, decreasing the concentrations of magnetic materials. Secondly, it can suppress the magnetic signature due to its slightly negative magnetic susceptibility of -0.5 to -2.0x10-8m3/kg (Blum, 1997); although the effect on the values will be very small. It is thus possible that the apparent relationship is caused by other independent sedimentary factors or processes.

All of the peaks have relatively high frequency dependent susceptibility ranging from 8 to 10% (Figure 9.9). Since most of the peaks (i.e. at about 5cm, 75cm, and 107cm) are located in sections characterised by clear stratification with little visible disturbance, the sediment was probably already enriched in the fine (superparamagnetic) magnetic particles at the time of deposition in the lake basin. This suggests that the pedogenesis or burning, that led to the formation of the superparamagnetic particles, occurred away from the lake within its catchment (Dearing, 1999a). Since no charcoal was recovered from the sediments, the deposition of eroded topsoil is the most likely explanation.

More likely candidates for in situ soil formation are the magnetic susceptibility peaks at about 60 and 145cm. The upper one is associated with finely mottled grey and reddish sediments accompanied by occasional fine root tunnels. This suggests mixing of lacustrine and oxidised terrestrial sediments under conditions probably drier than those represented by the grey overlying sediments.

The lower of the two peaks (at 145cm) is associated with a red unit overlying brownish grey sediments of most likely aquatic origins. The penetration of the red sediments into the grey unit (supported by its patterns) suggest extensive bioturbation by plant roots, thus a major phase of vegetation occupation on the lake floor and resultant pedogenesis.

Finally, there is, surprisingly, a negative correlation between the magnetic susceptibility and the most frequent root presence in the sediment at the depth of 36-55cm (Figures 9.3 and 9.9). The relatively fine texture of the sediment does not support leaching of the magnetic particles as a likely reason. The dilution by gypsum or colonisation by aquatic plants such as sedges might be the cause of the inconsistency, but further analyses (e.g. frequency dependent susceptibility) are required to explain it.

198

Figure 9.10 Infrared spectrometry data (SWIR) for Lower Bell core G.

9.2.3 Palaeoenvironmental reconstruction of events

Event A: Creek channel, floodplain, or delta? (~150-164cm)

The grey colour of the sediment unit, low gypsum content, and high sand content point respectively to an aquatic/waterlogged and relatively freshwater depositional environment 199

characterised by relatively high energies. This in turn suggests two main interpretational possibilities. The first is that the site was either part of the Bells Creek’s channel or its floodplain, possibly before the damming of its channel by the red sand dune and formation of the lake. The second interpretation suggests that the sediments accumulated in the creek’s delta, where, upon a change from a higher to a lower energy environment, the coarser particles would settle out of suspension. In view of the interpretation presented in the next section, as well as the events registered in the Palaeolake (section 9.3 below), the first interpretation seems more likely to be correct. A better dating of the dune and lake sediments is necessary to resolve this issue.

Event B: Arid phase?: red sand dune mobilisation? (~131.5-150cm)

The high oxidation of this unit hints at extreme aridity. The abundance of phytoliths suggests increased vegetation cover, while the presence of freshwater sponge spicules and diatom traces point to wetter conditions. Finally, the magnetic susceptibility (incl. frequency dependent susceptibility) implies either prolonged in situ soil formation and/or influx of topsoil from the surrounding area. Together they present a rather curious mixture of proxy evidence.

One possible scenario that can explain this evidence is the blocking of the Bell Creek’s channel by a red sand dune mobilised during a time of extreme aridity. While most of the mobilised red sandy sediments would have accumulated in the dune, some would have ended as a fine cover over the rest of the landscape, including the newly formed lake basin. The blockage of the creek’s course by the dune would have resulted in increased retention of water in the new lake following the rare (but probably not entirely absent) rainfall events. The increased wetness of the area would then have favoured plant colonisation and occasional short-term waterlogged conditions would have supported temporary freshwater sponge populations and even some diatoms.

The presence of plants in these wetter areas would be conducive to initiation of pedogenic processes. The plants were also likely to trap some of the aeolian materials, including topsoil that formed in the surrounding area during the preceding wetter phase and was eroded during this phase of increased aridity when the vegetation cover was reduced. The creek’s activity during rare rain events could have also contributed to the redistribution of the red sediments, including the relocation of topsoil materials from its catchment.

Event C: Short wet phase and transition toward increased ephemerality/seasonality (~121- 131.5cm)

A short lacustrine phase represented in this unit deposited a grey sediment unit. The increased water influx into the lake combined with its terminal status possibly resulted in a gradual build-

200

up of gypsum within the lake’s waters. The increasingly arid conditions toward the top of the unit resulted in a concentration of the lake brine and precipitation of gypsite bands. The fluctuations between the drier and wetter conditions led to deposition of at least three sets of grey clay – gypsite couplets. The small crystal size suggests that drying was relatively rapid. The absence of phytoturbation suggests that the lake dry periods were relatively brief.

Event D: Increasing lake dryness: ephemeral lake phase and formation of gypsum lunette (~78-121cm)

The increased aridity demonstrates itself within this unit by the appearance of oxidised red sediment layers in addition to the continued presence of the gypsum, including gypsite, and grey clay layers. The presence of the oxidized layers suggests extended periods of complete lake drying and possible influx of airborne red sediments eroded from the catchment/surrounding area. The grey layers, on the other hand, were probably deposited as a result of extreme rainfall events and/or extensive floods. The presence of gypsite bands, combined with the ephemeral conditions, suggest this unit as a source of material for the formation of the gypsum lunette.

A slightly thicker grey band at about 112cm suggests a short phase of wetter conditions, which would have effected recovery of some of the vegetation in the catchment. This would have rejuvenated the catchment’s soils enriching them in the superparamagnetic grains (responsible for enhanced magnetic susceptibility and frequency dependent susceptibility), phytoliths and even organic matter. However, the next dry episode was likely to result in a new reduction in the plant cover and thus mobilisation of the topsoils, portion of which was deposited and preserved in the lake sediments at a depth of about 110cm. The in situ soil development does not seem to be likely due to scarcity of bioturbation evidence.

Event E: Continuation of ephemeral lake conditions with gypsum deficit (~68-78cm)

While the grey-red couplets in this unit suggest a continuation of the periodic/seasonal lake filling and drying regime, the gypsum disappears from the record, indicating some changes in hydrology and/or gypsum sourcing in the area. The changes seem to be gradual as the decline in frequency and thickness of the gypsite layers is already noticeable towards the top of the previous unit (i.e. Event D above).

Event F: Swampy lake phase with continued gypsum deficit (~33-68cm)

The continued lack of gypsum and signs of extensive bioturbation (evidenced by intense mottling of grey and red sediments) suggest siltation of the lake (i.e. decrease in water depths) and development of a relatively fresher swamp.

201

It is possible that slight amelioration of conditions increased permanency of the lake, resulting in increased deposition of grey sediments. The vegetation recovery was, however, slow, possibly indicating an increase in rainfall but retention of temperatures unfavourable to plant growth. A considerable proportion of the red topsoils from the catchment/surrounding area was subjected to aeolian and/or runoff erosion and subsequent deposition in the vegetated swamp, where it was mixed with the grey sediments by plant roots and animals. As the catchment stabilised, its sediment contributions declined as suggested by a decrease in the red mottling as well as magnetic susceptibility.

Event G: Wet lake phase and return to gypsum-rich conditions (~6-33cm)

The conditions during deposition of this grey sediment unit were likely to be wetter, however, limited signs of oxidation in the form of sparse fine reddish particles are also visible. The processes/conditions responsible for the gypsum influx into the lake also become reactivated, resulting in the highest gypsite content recorded within the core (evidenced by the strongest gypsum signature in the infrared data). The end of this phase is marked by a thick band of gypsite, indicating the final drying of the lake in transition into the next dry phase. Based on the region’s rainfall data (Figure 2.5), this was likely to occur at the end of a wet phase, which concluded in the second half of the 1890s.

Event H: Drought: the 1890s to the 1940s (~6cm)

Considering the evidence of the 1890s to the 1940s drought preserved in the other lakes of the Paroo/Warrego Region, it is likely that the highly oxidised materials within the 3-6cm depth were deposited at that time. The influx of the catchment’s topsoil is supported by the enhanced magnetic susceptibility and frequency dependent susceptibility.

Event I: Amelioration of conditions: ephemeral lake phase post 1940s (~0-5cm)

In the last few decades the lake seems to have fluctuated between wet and dry states, resulting in the deposition of alternative grey and red layers. The modern pollen record in the lake is strongly dominated by the Chenopodiaceae, reflecting their importance within the lake’s basin. The low disturbance of the sediment layer suggest, however, that the colonisation of the coring site by plants was very limited in recent times(no plants were observed at that location during coring). Thus, it appears that the dry lake episodes were too brief or the lake floor too saline to allow plant growth.

202

9.3 Palaeolake

9.3.1 Site and lake overview

The Palaeolake is a small lake (<50ha) that fills only rarely (B. Timms, pers.comm. 2007). The lake’s catchment extends over about 63km2, however, most of the runoff is trapped in Freshwater Bloodwood Lake (Figure 9.1) and continues on to Palaeolake only on those infrequent occasions when Freshwater Lake overflows. To the north and east the Palaeolake is bordered by a large gypsum lunette rising up to 8m above the lake floor (Figure 9.11). To the northeast the lunette is bisected by a channel linking Palaeolake with Freshwater Bloodwood Lake (Figure 9.1).

The salinity in Palaeolake fluctuates within the hyposaline range and B. Timms (pers. comm.) recorded values between 8.6 and 21.8 g/L. When dry, the lake floor is covered by samphires (Halosarcia spp.) and glasswort (Pachycornia spp.) with a minor component of saltbushes (Chenopodiaceae), grasses (Poaceae), forbs and herbs (including Frankenia gracilis, Brachyscome ciliaris, Ixolaena leptosis, Chamaesyce drummondii, and Pimelea trichostachya) (Figures 9.12 and 9.13).

The top of the gypsum lunette is very sparsely vegetated (Figure 9.12), but supports stands of belah (Casuarina cristata) and occasional tree tobacco (Nicotiana glauca) and prickly wattle (Acacia victoriae). The area around the lake is dominated by woody shrubs, mainly turpentine bush (Eremophila sturtii), hop bush (Dodonaea viscosa), and rice flower (Pimelea microcephala) with occasional specimens of western rosewood (Alectryon oleifolius), bimble box (Eucalyptus populnea), whitewood (Atalaya hemiglauca), leopardwood (Flindersia maculate), and wilga (Geijera parviflora). The sand dune crests (even of very low relief) are often favoured by belah populations. A more extensive list of Palaeolake plant species (especially ground cover) is provided by Gayler (2000).

9.3.2 Results

Three lake sediment cores (B, D, and K) and two augered sediment cores (from the northern red sand dune and the eastern gypsum lunette) were used in this study. Detailed analyses were carried out on cores B and K and on the augered sediment, while core D was used for ostracod analysis. The data for core B are presented in Figures 9.14 to 9.19 with additional descriptions in sections 9.3.2.1-5, for core K in Figures 9.20 to 9.27 and sections 9.3.2.6-11, for the northern red sand dune in Figures 9.28 to 9.33 and sections 9.3.2.12-16, and finally for the gypsum lunette in Figures 9.34 to 9.39 and sections 9.3.2.18-21.

203

Figure 9.11 Transect across Palaeolake including relative positions of the cores analysed in this study.

204

Figure 9.12 Palaeolake: eastern gypsum lunette with a stand of Casuarina cristata. The inflow channel from Bloodwood Freshwater Lake is located to the left of the picture. Halosarcia spp. in the foreground.

Figure 9.13 Red sand dune on the northern shore of the Palaeolake (background). Halosarcia spp. on the beach and partly submerged (foreground).

9.3.2.1 Core B: gypsum

Similar to Lower Bell Lake, Palaeolake sediments are characterised by a high gypsum content. In core B most of the gypsite/fine gypsarenite occurs within the top 60cm, mostly as mottling, i.e. as irregular nodules or as patches, which gradually diffuse into the surrounding sediment (Figures 9.14 and 9.19). Except for the reddish uppermost 3cm, the sediments within this depth are grey to reddish grey. While bioturbation is a likely explanation for the mottling (considering the plant presence on the lake floor), the lack of roots or clear root patterns casts some doubt on that explanation.

205

Figure 9.14 Sediment description for Palaeolake core B. Some discoloration of the sediment was caused by light distortion during photographing.

206

The 60 to about 80cm section consisting of grey muds is relatively poor in gypsum (Figures 9.14, 9.18, and 9.19). The 80 to 170cm section (except for 108-119cm depth) is characterised by very clear laminations and bands of grey and reddish sediments. While the fine grey laminae (generally <5mm thick) are composed of clays with no visible gypsum, the thicker and coarser grey bands often contain various densities of scattered gypsarenite crystals. The reddish bands also contain gypsum similar to that of the grey bands with the highest crystal densities occurring in the most oxidised (i.e. reddest) units. It is uncertain if they were deposited from brine or groundwater, or if they represent a mixture of sources.

The final gypsum feature worthy of notice in core B is about 1.5cm thick band of very dense selenite at a depth of 170cm (Figure 9.14). The band is composed of pyramidal to lenticular simple and twin crystals (suggesting their precipitation from groundwater) embedded in oxidised clayey sediments (Figures 9.15 and 9.18 – the selenite was extracted from this sample prior to the particle size analysis). While the band provides a clearly defined boundary between the heavy lake muds (with very low gypsum content if any) at the bottom of the core and the laminated/banded section above, the exact processes/reasons behind its formation and location within the core are still to be determined.

Figure 9.15 Palaeolake core B gypsum crystals at 170cm depth: a, b & c – lenticular to pyramidal selenite (single and twin).

207

9.3.2.2 Core B: pollen

The surface pollen is dominated by Chenopodiaceae, reflecting the modern cover of the lake floor by mainly Halosarcia species (Figure 9.16). The most interesting feature of the pollen data is, however, the presence of the Unknown type 1, which is the only grain that seems to have survived in deeper sediments. The grain’s distribution is closely tied to the section of the core that is composed of the red/gypsum-grey couplets (i.e. within 100-180cm depth with a peak at 140cm), pointing to the association of its source plant with ephemeral (drying/dry saline) lake conditions.

9.3.2.3 Core B: phytoliths and other biogenic silica

The phytolith and freshwater sponge spicule records for core B are very limited and contribute little to the reconstruction of the palaeoenvironmental story (Figure 9.17).

9.3.2.4 Core B: other signs of plant and animal life

Sparse amounts of plant fragments were found throughout core B, mostly in an association with the grey clay/gypsum-rich red sediment couplets (Figure 9.14). This suggests their influx into the lake with the clay bearing floodwaters.

The charophyte and ostracod record is limited in core B to the surface sample (Figure 9.14). The identified charophyte taxa include Lamprothamnium heraldii and Nitella cf. verticillata (subsaline to hypersaline habitats) and the ostracod taxa include Cyprinotus sp. (fresh to hyposaline), Heterocypris sp. (hyposaline), Reticypris sp. and Trigonocypris globulosa (both hyposaline to hypersaline) (Appendix 7 and Appendix 8). All of these species occur within the salinity ranges experienced by Palaeolake in modern times.

9.3.2.5 Core B: magnetic susceptibility

Core B has three large and one small sections of magnetic susceptibility enhancement at 0, 56.5- 70, 120 and 170-174cm (Figure 9.18). All of the peaks are accompanied by increases in frequency dependent susceptibility ranging from 6.5% for the minor peak at 120cm to 8% for 56.5cm, 9.2% for 170cm and 9.9% for surface sediments. All of the peaks are associated with reddish to red sediments, although not all of the red bands are associated with enhanced magnetic susceptibility (through some might have been missed due to the low sampling resolution). In addition, the three upper peaks are confined to sediment units, which show plant presence in the form of fine roots and root tunnels. Thus, all of the evidence points to in situ magnetic enhancement through pedogenesis, although, some contribution from deposition of topsoil eroded in the catchment (transported in dust plume and/or runoff) is also possible.

208

Figure 9.16 Pollen counts for Palaeolake core B. 209

Figure 9.17 Biogenic silica counts for Palaeolake core B.

The largest peak at 170-174cm is the only one that does not contain clear signs of root presence. While there are no clear clues as to the nature of the downward diffusion features of the overlaying red sediments into the underlying muds, it is likely that they have developed as a combination of clay cracking and limited plant invasion (a dense cover would have resulted in much clearer root penetration patterns). Considering the high magnetic susceptibility and frequency dependent susceptibility of this red unit in the lake core, it is reasonable to conclude that most of the superparamagnetic particles (i.e. the soil signature) were formed in the surrounding area rather than in situ.

210

Figure 9.18 Palaeolake core B: sediment diagram. 211

Figure 9.19 Infrared spectrometry data (SWIR) for Palaeolake core B.

212

9.3.2.6 Core K: gypsum

Core K has three main zones of gypsum occurrence: 5-20cm, 60-135cm, and 187-207cm (Figures 9.20, 9.26, and 9.27). The upper zone, similar to core B, supports gypsite. Unlike core B, but similar to Lower Bell Lake core, the gypsite in core K forms a distinctive band (~1cm thick) at about 9cm depth, which then progressively diffuses downwards (accompanied by gradual decrease in abundance) within the grey sediments until it disappears at about 20cm.

The 60-135cm zone is characterised by laminations and bands (1mm to a few centimetres thick) of red, reddish, and grey sediments, with the redder sediments generally of coarser texture than the grey ones. While generally showing a similarity to core B, many of the laminations in core K are much finer and occur in greater numbers. Since core K is located much closer to the channel inlet (i.e. within the delta area) than core B, this may be related to higher sensitivity of this part of the lake basin to individual water influx and drying events. Each creek flow event would dump here the coarser fraction of its load before carrying the finer materials further into the lake. As the lake levels drop, the higher relief of the delta, compared to the rest of the lake basin, would see it dry more frequently.

The gypsum crystal habits of this section are generally larger than those occurring in core B (Figures 9.21, 9.22, and 9.23). A rosette form (Figure 9.21 image 1) was recovered from a depth of 62cm suggesting pedogenic activity. The irregularly shaped gypsum blocks (through showing some slight tendency toward pyramidal form) with obvious inclusions (recovered from the depth of 81cm) (Figure 9.21 image 2) might be also associated with pedogenic processes. The shapeless form of the crystals can result from repetitive partial dilution and renewed growth cycles. This interpretation is supported by considerable presence of fine roots within this section.

The complex structures (nearly rosette-like) of the mostly pyramidal to lenticular gypsum from depths of 95 and 108cm (also showing considerable presence of inclusions) accompanied by root tunnels suggest groundwater origins of the crystals, which were later modified by pedogenic processes (but to a lesser extent than those at the 81cm depth). It is noteworthy that the morphological complexity of the form and the extent of the crystal’s margin dissolution tend to decrease down the core (Figure 9.21 images 2, 3, & 4 and Figure 9.22 images 2 &3). This suggests a decrease in groundwater fluctuations with depth.

213

Figure 9.20 Sediment description of Palaeolake core K.

214

Figure 9.21 Palaeolake core K gypsum crystals: 1 – rosette of gypsarenite and small selenite at 62cm depth; 2 – irregular (semi-lenticular) gypsum at 81cm depth; 3a & b – mostly lenticular to pyramidal selenite (single to complex) at 95cm; 4a & b – mostly single lenticular to pyramidal selenite and gypsarenite (rarely forming aggregates) at 108cm with minor signs of dissolution along the margins.

A peculiar habit is found among selenite crystals recovered from 111 and 134cm depths, that hints at a combination of prismatic and pyramidal development with possible formation of twinning surfaces (Figure 9.22 image 1 and Figure 9.23 image 1). The form suggests formation

215

under conditions (e.g. characterised by a specific surface or subsurface water chemistry) and/or by processes (e.g. related to a drastic change in the depositional environment during crystal formation) that were not recorded in any other lake within this study.

Figure 9.22 Palaeolake core K gypsum crystals: 1 - selenite at 111cm depth: a – prismatic/pyramidal (twin); b – undefined habit; 2a & b – pyramidal to lenticular selenite and gypsarenite at 122cm; 3a & b – pyramidal to lenticular twin selenite at 127cm.

Finally, the gypsum zone at 193-207cm contains about 2cm thick bands of moderately dense lenticular shaped selenite with dense surface deposition (scales-like) of lenticular gypsarenite and loose lenticular gypsarenite (Figure 9.23 image 3). While the lenticular form points to

216

groundwater origins, the size range probably indicates separate precipitation events for the selenite (with more time to grow) and the gypsarenite (a more rapid event and shorter event). The latter event might have been also responsible for scale-like depositions on the larger selenite crystals. These gypsum bands are set within grey sediments bordered to the top and bottom by red silty clays with no signs of gypsum presence (Figures 9.20, 9.26, and 9.27).

Figure 9.23 Palaeolake core K gypsum crystals: 1a & b – prismatic/pyramidal (twinned) selenite at 134cm depth; 2a & b - mostly lenticular selenite and gypsarenite at 201cm with signs of dissolution (rugged margins) and overgrowth; 3a & b – mostly lenticular selenite at 205cm with signs of secondary precipitation (overgrowth) and dissolution, particularly around the margins.

217

9.3.2.7 Core K: pollen

In spite of being rather short, the core K pollen data show an interesting trend, which was also noticed in other lakes of the area, namely a recent decrease in herbs and grasses pollen in favour of Chenopodiaceae (Figure 9.24). A slight recovery of the herbs and grasses occurred in the surface sediments.

The Unknown type 1 was nearly absent from the record, possibly indicating its preference for the main part of the lake bed as habitat rather than the stream’s delta, which is more strongly affected by processes such as sediment sorting, transport, rapid deposition, and reworking.

9.3.2.8 Core K: phytoliths

Of all the Palaeolake cores, core K provided the richest phytolith record (Figure 9.25). The record is characterised by fluctuations in the proportions of the Poaceae and Dicotyledon/Monocotyledon (non-Poaceae) types with Poaceae types dominant at 120, 100, and 80cm, nearly equal proportions of both types at 110, 90, and 5cm, and Dicotyledon/Monocotyledon (non-Poaceae) types by far dominating the 50cm depth. Since, with the exception of the 5cm sample, the highest phytolith counts are coincident with a section of black (organic), grey, and red sediment layers, it is likely the record illustrates cycles of vegetation successions under highly seasonally/periodically dry-wet conditions.

9.3.2.9 Core K: other biogenic silica

The higher freshwater sponge spicule densities are generally associated with depths that have also higher phytolith counts and, at 80 and 120cm, also with traces of diatoms (Figure 9.25). It is thus likely that the sections represent wetter conditions promoting plant growth (hence producing more phytoliths) and more frequent and/or persistent water in the lake supporting sponge populations. Since the spicules are rare in core B samples (Figure 9.17), it is possible that the signal recorded in core K is more closely related to the Freshwater Lake and the spicules were delivered into the lake by the ‘overflow’ stream.

9.3.2.10 Core K: other signs of plant and animal life

Similar to core B, the rare plant fragments found in core K tend to be associated with the unit characterised by the presence of gypsum, dark (organic), grey, and red bands (Figure 9.20), indicating phases of increasing biomass production interlaced with periods of increased aridity.

218

Figure 9.24 Pollen counts for Palaeolake core K. 219

Figure 9.25 Biogenic silica counts for Palaeolake core K.

Except for a single oospore at 180cm, charophyte oospores were found only in relatively modern sediments, i.e. the top 10cm (Figure 9.20). The identified taxa include Lamprothamnium heraldii, Nitella cf. verticillata, and Chara species. The first two indicate variable lake water levels and salinity conditions (subsaline to hypersaline) consistent with those occurring within the lake at the present time. The Chara sp., which favours generally less saline waters than the range recorded for Palaeolake, might have been flushed into the Palaeolake from the fresher Freshwater Lake (a possibility supported by its presence close to the inlet channel). 220

9.3.2.11 Core K: magnetic susceptibility

Two major sections of magnetic susceptibility enhancement occur in core K: at 5cm and within 86-145cm depth (Figure 9.26). Both of these sections are characterised by high frequency dependent susceptibility values of about 11%. The upper magnetic susceptibility peak corresponds to red sediments with sparse signs of root activity (Figure 9.20) suggesting that a large proportion of the enhancement was produced by pedogenic processes in the catchment (eroded and deposited in the lake) with a smaller contribution from in situ soil formation.

The lower section is consistent with the occurrence of red-grey fine to coarse stratification (layering). The two magnetic susceptibility peaks at 100 and 130cm are consistent with the middle of massive red units (with barely visible stratification), while the drop at 120cm is aligned with a slightly thicker grey band. High frequency dependent susceptibility together with the moderate density of root tunnels, as well as a slight increase in organic matter and carbonate content, suggest that the magnetic enhancement resulted from a combination of in situ pedogenic processes and influx of top soil from the catchment.

All of the smaller peaks in magnetic susceptibility (at 58, 77, 156, 186, and 214cm) are also associated with units of distinctly redder sediment. In addition, the 58 and 156cm peaks also show substantial values of frequency dependent susceptibility of 7.3 and 7.5% respectively, suggesting strong links with pedogenic processes (in situ and/or in the catchment).

Of particular interest is also a distinct dip in magnetic susceptibility at 69-77cm associated with a section of grey (lacustrine) sediments underlain and overlaid by red/grey sediment couplets.

9.3.2.12 Northern red sand dune: gypsum

The red sand dune at the northern shore of the Palaeolake contained gypsum throughout most of its augered depth, but in relatively low quantities (Figures 9.28, 9.32, and 9.33). Most of the gypsum within the dune occurs as aggregates of gypsite, gypsarenite, and small selenite crystals with varying amounts of inclusions (mostly sand) (Figure 9.29). Two sections: 20-90cm and below 430cm appear to be gypsum poor (Figures 9.28 and 9.32). The upper one might have been stripped of gypsum through leaching. The lower one is associated with a greyish lake muds covered by red aeolian sediments. Similar muds occur at the bottom of core B, from which gypsum was also absent.

There are also two sections of higher (i.e. considerable) gypsum presence: at 120-230cm and around 360cm depth (Figures 9.28 and 9.32). The higher concentrations, particularly in the upper sections, may be due to reprecipitation of the gypsum leached from the upper horizon. This interpretation is supported by the irregular shape of the gypsum aggregates and the presence of inclusions.

221

Figure 9.26 Palaeolake core K: sediment diagram.

222

Figure 9.27 Infrared spectrometry data (SWIR) for Palaeolake core K. 223

Figure 9.28 Sediment description for Palaeolake north red sand dune auger hole. 224

Figure 9.29 Palaeolake northern red sand dune gypsum: 1 – gypsite nodule at 320cm; 2a & b – gypsarenite/selenite conglomerate at 340cm; 2c – close up of 2b image.

9.3.2.13 Northern red sand dune: pollen

The surface sample from the red sand dune shows a slightly different pollen composition compared to all the other Palaeolake cores. It contains nearly equal proportions of Chenopodiaceae and herbs and grasses with a larger tree and shrub component (Figure 9.30). The difference in the composition seems to be directly controlled by the core’s location and thus the difference in the surrounding vegetation, specifically the presence of woody shrubs with occasional Eucalyptus as well as ground cover supporting higher densities of herbaceous plants and grasses and lower densities of Chenopodiaceae.

A second peak in pollen counts occurs at 360cm, just above the lake mud sediments. While the Asteraceae, Poaceae, and Chenopodiaceae counts are similar to the surface sample, the trees and shrubs are present in much smaller numbers.

225

Figure 9.30 Pollen counts for Palaeolake northern red sand dune auger hole.

226

9.3.2.14 Northern red sand dune: phytoliths and other biogenic silica

Like the pollen, the phytoliths in the red sand dune occur in two peaks: at 0 and 360cm (Figure 9.31). Both peaks seem to show similar proportions between the Poaceae and Dicotyledon/Monocotyledon (non-Poaceae) phytoliths, suggesting similar vegetation composition.

Figure 9.31 Biogenic silica counts for Palaeolake northern red sand dune auger hole.

The presence of higher freshwater sponge spicule densities at 0 and 360cm (Figure 9.31), thus coincident with peaks in pollen and phytolith counts, not only indicates a more frequently filled or more permanent lake, but also supports the use of these peaks as indicators of wetter conditions more conducive to plant growth.

9.3.2.15 Northern red sand dune: other signs of plant and animal life

Plant fragments within the red sand dune were found in small amounts in surface and 360cm samples (Figure 9.28). Their presence further supports the interpretation of pollen, phytolith, and sponge spicule records, that suggested milder, wetter conditions.

227

9.3.2.16 Northern red sand dune: organic matter and carbonates

The organic matter and carbonate contents seem to reflect the division of the recovered sediments into aeolian materials (strongly affected by the pedogenesis) and lacustrine muds, with the latter characterised by slightly smaller amounts of organics and a carbonate deficit (Figure 9.32).

9.3.2.17 Northern red sand dune: magnetic susceptibility

The magnetic susceptibility properties of the sediments in the northern red sand dune additionally highlight the distinction between the overlying red sands and the lacustrine sediments at the bottom of the profile (Figure 9.32). The red sands are characterised by relatively high magnetic susceptibility values (ranging 26.1-50.2x10-8m3/kg at low frequency) accompanied by high frequency dependent susceptibility (9.1-12.1%), which are consistent with dominance of pedogenic processes. The secondary peak at about 180-260cm is more likely to be related to deposition of leached materials within a B horizon than to a buried palaeosol as it coincides with the zone of gypsum accumulation and increased clay content.

In contrast, the lake muds below 380cm are characterised by a decrease in magnetic susceptibility (16.5 to 12x10-8m3/kg at low frequency) and frequency dependent susceptibility (5-3.5%). Both values decrease down the profile with the decrease in content of the (overlying) red sandy sediment.

9.3.2.18 Eastern gypsum lunette: gypsum

The Palaeolake lunette contains over 520cm of gypsum-rich sediments (Figures 9.34, 9.38, and 9.39) with the gypsite and gypsarenite often forming irregular nodules with varying amounts of sand inclusions (Figure 9.35 images 1a & b). The majority of the gypsarenite occurs both as aggregates and as loose lenticular crystals (Figure 9.35 image 1c), suggesting their subsurface formation in lake sediments (e.g. Warren, 1982).

An interesting feature of the lunette’s sediments is the reappearance of the unidentified crystalline balls that were also encountered in the Lower Bell Lake sediments (Figures 9.3, 9.34, and 9.35 image 2). They suggest that the type of sediment encountered in Lower Bell core G was likely to supply materials for the gypsum lunette formation.

In spite of gypsum dominance throughout the lunette deposits, there are visible variations in the red sand content of the sediments (e.g. the occurrence of two distinctive redder/darker bands at 125-200cm and 220-240cm) (Figure 9.34). The redder units a likely to indicate periods of increased instability/mobility of the surrounding red sandplains.

228

Figure 9.32 Palaeolake northern red sand dune auger hole: sediment diagram.

229

Figure 9.33 Infrared spectrometry data (SWIR) for Palaeolake northern red sand dune auger hole.

9.3.2.19 Eastern gypsum lunette: pollen

The most interesting feature of the gypsum lunette pollen data is the dominance, except for the 0-50cm samples, of Unknown type 1 grain (Figure 9.36). This clearly points to the couplet- dominated section of the lake sediments as a source of materials for the lunette formation.

9.3.2.20 Eastern gypsum lunette: phytoliths and other biogenic silica

The gypsum lunette sediments contained very low phytolith concentrations (Figure 9.37), suggesting low plant densities during its formation and/or high sediment accumulation rates.

Only two samples, at 0 and 280cm, recorded rare sponge spicules (Figure 9.37). Their presence at the 280cm depth coincides with a peak in the Unknown type 1 pollen. Considering the general pattern of the grain’s occurrence, this sediment was likely to be removed in the initial stages of the lake floor deflation following a wetter interlude during the ephemeral lake phase.

230

Figure 9.34 Sediment description for Palaeolake eastern gypsum lunette auger hole. 231

Figure 9.35 Palaeolake eastern gypsum dune: 1 – gypsite nodule with sand inclusions within 85-125cm depth section; 1b – close up of 1a image; 1c – mixture of lenticular gypsarenite and sand at 85-125cm depth; 2 – unidentified crystal balls at 380cm.

9.3.2.21 Eastern gypsum lunette: other signs of plant and animal life

Calcified burrows and/or rootlet tunnels were found in very low numbers in the gypsum lunette at depths of 85-125cm and below about 440cm (Figure 9.34). Furthermore, plant fragments were recovered from a sample at a 475cm depth, which, together with increased organic matter and carbonate content (Figure 9.38), suggest a phase of increased plant activity and pedogenic processes, thus a temporary stabilisation of the lunette.

232

Figure 9.36 Pollen counts for Palaeolake eastern gypsum lunette auger hole.

233

Figure 9.37 Biogenic silica counts for Palaeolake eastern gypsum lunette auger hole.

9.3.2.22 Eastern gypsum lunette: magnetic susceptibility

The eastern gypsum lunette is characterised by relatively low magnetic susceptibility (generally <15x10-8m3/kg at low frequency) but moderate to high frequency dependent susceptibility that ranges from 5.4 to 12.1% (Figure 9.38). The upper-most peak in both magnetic susceptibility and frequency dependent susceptibility occurs at the surface and is likely to be related to modern pedogenic processes, particularly because the auger hole was located close to a stand of mature Casuarina trees.

A magnetic susceptibility peak was measured at 180cm (20.1x10-8m3/kg at low frequency) and was accompanied by an increase in frequency dependent susceptibility exceeding 9%. This section of the core (about 100-280cm) contains two bands of redder sediments and fine calcified burrows or root tunnels (Figure 9.34). Such evidence suggests that plants were present on the dune and supports in situ soil development as well as trapping of aeolian red materials that possibly contained topsoil eroded by wind from other parts of the catchment.

234

Figure 9.38 Palaeolake eastern gypsum lunette dune auger hole: sediment diagram.

235

Finally, the bottom section of the augered sediments (>460cm), while showing only slight overall enhancement in magnetic susceptibility (possible effect of dilution in light of the other evidence), is characterised by increased frequency dependent susceptibility (7.3-8.6%) (Figure 9.38). It coincides with an increase in organic matter and carbonate content as well as the presence of fine calcified burrows or root tunnels, but no increase in redness (Figure 9.34), suggesting the presence of plants and an activation of pedogenic processes.

Figure 9.39 Infrared spectrometry data (SWIR) for Palaeolake eastern gypsum lunette dune auger hole.

9.3.3 Palaeoenvironmental reconstruction of events

The correlations between the cores, described below, were based on the similarity of sediment units and trends shown by the different proxies. An independent OSL and radiocarbon dating is necessary to confirm the validity of these interpreted correlations and provide a better timeframe.

236

Event A: Large, deep lake phase (core B >170cm, northern red sand dune >380cm)

The grey, finely-textured muds, characterised by low magnetic susceptibility and frequency dependent susceptibility, suggest permanent lacustrine conditions during their deposition. The lake could have been also deeper than at present, protecting the sediment from oxidation. The gypsum deficit demonstrated by both visual assessment and infrared spectrometry implies a relatively freshwater environment. Furthermore, lacustrine muds, similar to these in the lake core, were found under the northern red sand dune. They provided an evidence of a more extensive lake basin at that stage of the lake history. Considering the upper surface of the mud under the dune is located about 2.5m higher than that in the lake core (Figure 9.11), it seems reasonable to assume that at least some of the difference resulted from the deflation of these lacustrine sediments from the lake basin upon their exposure during subsequent dry episodes. The surrounding well-watered landscape was likely to support dense vegetation cover as well as pedogenic processes. This soil was later eroded during subsequent arid phase.

Event B: Increased aridity: lake drying phase, mobilisation of red sand dunes (core B ~170- 154cm, northern red sand dune ~360-380cm)

Increased aridity resulted in drying of the lake, which is evidenced by the change from grey to red oxidised sediments as well as the first major appearance of gypsum initiated with a dense selenite layer. The latter suggests a decrease in water levels, which led to increased salt concentrations and a change from freshwater to saline conditions. Fine grey bands within the red sediment suggest a seasonal/periodic flooding that probably protected the lake bed from colonisation by plants.

The aridity was likely also to lead to a decline in vegetation cover and mobilised the surrounding red sandplains and dunes, stripping them of topsoil. Some of the mobilised sediments accumulated in a dune formed over the exposed lacustrine muds on the northern margin of the lake, while a finer layer of these red sediments settled over the rest the lake basin.

Event C: Brief amelioration phase: temporarily permanent lake (core B 143-154cm, core K >145cm (?), northern red sand dune ~360cm)

The grey sediments of this unit in core B suggest a temporary return of wetter conditions and a reestablishment of a permanent lake. It is possible that this unit is equivalent to at least one of the grey units below 145cm in core K, due to its location immediately below the grey-red couplets section. The increased plant and freshwater sponge presence in the red sand dune sediments at about 360cm depth provide further evidence for the wetter phase, suggesting a brief period of stability and development of plant cover. Both pollen and phytolith data suggest plant composition similar to the modern one with similar proportions of Chenopodiaceae,

237

Asteraceae, and Poaceae. An exception is the Dodonaea pollen, which is present in distinctly higher numbers in surface sediments.

Event D: Enhanced seasonality/periodicity: ephemeral lake phase, gypsum lunette formation, and continued accumulation of the red sand dune (core B 87-143cm, core K 75-145cm, gypsum lunette, northern red sand dune >360cm)

The gypsum-rich grey-red couplets, occurring in lake cores B and K, suggest fluctuations between dry and filled/flooded lake conditions. The differences in the thickness of the sedimentary layers and in the oxidation intensity (i.e. degree of redness) of the red bands indicate some variability in the length and intensity of the subsequent dry and wet stages, hinting at more irregular, rather than seasonal, events.

It appears also likely that the creek’s delta (core K) was more frequently exposed and less saline than the lake’s main basin (core B), due to its elevated position. Thus, it favoured more frequent plant colonisation as shown by the greater presence of root (than in core B), increased phytolith counts, enhanced magnetic susceptibility and frequency dependent susceptibility, and modification of gypsum crystals through cycles of dissolution and reprecipitation. The high fluctuations in dominance of grasses versus non-grasses suggest frequent changes in plant composition (perhaps as a part of successional cycle) driven by the fluctuating climatic conditions.

A period of prolonged lake dryness (represented by 118-126cm depth in core B and possibly 77- 96cm in core K) occurred around 14 440 years ago, resulting in plant colonisation of the lake floor and development of a weak soil, evidenced by the magnetic susceptibility signature and presence of root-like mottling.

The drying of the lake and associated increase in brine concentration led to precipitation of gypsum. As the lake floor became exposed to deflation, it provided material for the formation of the gypsum lunette on the eastern margin of the lake. The links between the grey-red couplets and the lunette are supported by the exclusive presence within those units of the Unknown type 1 grain.

The fluctuating wet and dry periods became preserved also as distinct whiter and redder bands within the lunette sediments. The whiter units were deposited in the initial stages of the dry lake periods and contain more gypsum. The proportion of the red particles increased as the sandplain became more mobile and the frequency of the dust storms intensified during more advanced stages of the dry episodes. The presence of two units with calcified burrows/rootlet tunnels, on the other hand, suggests two wetter episodes in the lunette’s history.

238

The red bands within the lake sediments not only suggest a lake floor exposed to oxidation but, most likely, also enhanced aeolian erosion within the surrounding area, contributing sediments to the lake basin as well as to the northern red sand dune. Unlike the gypsum lunette, the red sand dune seems to have preserved little of its original sedimentary history. Instead, it appears to be greatly modified by the pedogenic processes with the majority of proxies (including gypsum content and habit, carbonates, magnetic susceptibility, and texture/bulk density) supporting the presence of 0, A (eluviation), and B (illuviation) horizons, suggesting the dune has been stable and vegetated for a long time.

Event E: Amelioration of conditions (core B ~63-87cm, core K ~56-75cm)

The grey unit, characterised by a lower gypsum content and a lower magnetic susceptibility, suggests sedimentation under wetter conditions, possibly indicating a new fresher water phase. The improved conditions, however, did not last long. The gradual increase in reddish particles within the sediment (mostly in the form of highly diffused bands in core B and better defined layers in core K) suggests a slow return to more arid conditions.

Event F: Gradual increase in aridity (core B ~36-63cm, core K ~38-56cm)

The concentration of the red particles became even greater within this unit, suggesting increased erosion of the surrounding sandplains and/or increased incidence of dust storms. The presence of the grey clay component, together with the absence of clear red bands, indicates frequent influx of ‘aquatic’ clays into the lake suggesting that it continued to hold water on at least a semi-permanent basis. The reducing conditions were, however, too weak to reduce all (if any) of the iron oxides deposited in the lake, possibly indicating a low organic matter content and thus a low biomass production in the catchment. Those conditions were present in the lake around 8 000 years ago.

In both cores the unit terminates with a section of rootlets and root tunnel structures accompanied by enhanced magnetic susceptibility and frequency dependent susceptibility, indicating a prolonged dryness of the lake floor, which allowed long-term colonisation by plants and soil development. This evidence of in situ pedogenesis is further supported by the presence of gypsum rosettes within this unit in core K.

Event G: Amelioration of conditions (?) (core B ~0-36cm, core K ~15-38cm )

The unit is composed mainly of grey sediments mottled heavily with fine gypsum. The lack of red mottling suggests relative stability of the surrounding landscape during its deposition. Because of a low filling frequency in modern times and a limited sediment source (i.e. the small catchment), the most recent sedimentation rates are likely to be very low.

239

Event H: Deterioration of conditions (post-1970’s?) (core K 0-15cm )

Unlike core B, core K appears to contain a record of the most recent increase in the aridity in the region. The first indication of drying, following a wetter period of grey sediment deposition, is the precipitation of a gypsite band in response to evaporative concentration of the lake brine. This is followed by a layer of red sediments. Considering the core’s location in the creek’s delta, the sediments were most likely eroded and transported from the catchment by surface runoff. The mobilisation of such large sediment volumes probably required a considerable reduction of vegetation cover in the catchment. This was likely to be caused by a combination of increased aridity and grazing pressures followed by an intense rain event.

The change from wetter to drier and back to wetter events is also reflected in the vegetation record. The Asteraceae and Poaceae proportions decrease and the Chenopodiaceae increase as the conditions deteriorate. The newest wetter phase results in slight recovery of the herbs and grasses, while Chenopodiaceae proportions decrease.

240

Chapter 10

Yandaroo-Delta Station lakes: Lake Willeroo and Lake Yandaroo

10.1 Yandaroo-Delta Station Lakes: Lake Willeroo and Lake Yandaroo

The Willaroo and Yandaroo Lakes lie on the Warrego River floodplain and have formed in response to natural damming of an old river channel (Figure 10.1) (Timms, 1993). Both lakes are filled irregularly, mainly from local runoff (rarely to full capacity), and tend not to hold water for long periods (B. Timms, pers. comm. 2007). Neither of the lakes has overflowed in living history (Timms, 1993).

Another shared characteristic of the lakes is the dominance of sodium (Na) and bicarbonate

(HCO3) ions (Timms, 1993). This sets them apart from the majority of arid and semi-arid lakes that are dominated by sodium and chloride ions. As this ionic characteristic often results from evaporative concentration of riverine waters (Long et al., 1992), this suggests closer links with the river than in any other lake system included in this research.

The area surrounding the two lakes is dominated by a combination of woody shrubs, including hop bush (Dodonaea viscosa), turpentine bush (Eremophila sturtii), and rice flower (Pimelea microcephala) and mulga (Acacia aneura) communities with stands of belah (Casuarina cristata), gidgee (Acacia cambagei), and occasional supple jacks (Ventilago viminalis). The properties are used for sheep and cattle grazing.

241

Figure 10.1 Map for Lakes Willeroo and Yandaroo. (Main image source: Toorale SH 550902 airphoto 205, 1997; Inset image source: Google Earth, 2006; geomorphological information after Timms (pers. comm.. 2007)

242

10.2 Lake Willeroo

10.2.1 Lake and site description

When full, Lake Willeroo extends over 113ha and is generally less than 2.5m deep (Timms, 1993). The lake is alkaline, with mean pH of 9.1, and relatively fresh with recorded salinity values ranging between 0.09-0.52g/L (Timms, 1998a). It lies within the Warrego’s floodplain deposits of black and red loam-sand and clays tightly surrounded by residual and colluvial deposits of angular, poorly sorted sand and gravel (Figure 10.2) that were derived from the parent rock (Louth 1:250 000 Geological Series Map Sheet SH 55-9, 1965).

Figure 10.2 Lake Willeroo: scattered rocks on the southern shore. Dead tree trunks scattered over the lake floor in the background.

While Goodrick (1984) believed the lake could be characterised by long term inundation, long dry periods are evidenced by the remains of dead trees scattered over the lake’s floor (Figure 10.3) (Timms, 1993). A relatively dense belt of live trees (Figure 10.1), mainly black box (Eucalyptus largiflorens) and river cooba (Acacia stenophylla), and shrubs, including western boobialla (Myoporum montanum), rice flower (Pimelea microcephala), and turpentine bush (Eremophila sturtii), extends along the lake’s margins, while rushes (Juncus spp.) fringe the lake’s edge (Goodrick, 1984). When dry, the lake floor is covered with Darling pea (Swainsona greyana) (Goodrick, 1984) and patches of introduced Mexican poppy (Argemone ochroleuca).

243

Figure 10.3 Lake Willeroo: dead tree trunk (foreground) and multiple tree stumps (background).

10.2.2 Results

Due to the extreme dryness of the lake at the time of the fieldwork, only one sediment core was extracted (by augering) from the Lake Willeroo basin. The general description of the core is presented in Figure 10.4, pollen and biogenic silica results in Figures 10.5 and 10.6 respectively, and other sedimentary data in Figures 10.7 and 10.8. Additional descriptions are provided for the main proxy data.

10.2.2.1 Pollen

Pollen recovery was limited mainly to the top 25cm of the core (Figure 10.5). Contrary to the recent trend of Chenopodiaceae replacing herbs and grasses that was recorded in other lakes within this study (see previous chapters), Lake Willeroo registered Chenopodiaceae domination in the past (at 25cm), which became, to a large extent, replaced by herbs and grasses (mainly Asteraceae) with a small contribution of Myrtaceae in most recent times (surface sample). The comparison with the other lakes needs to be, however, cautious due to a coarser sampling resolution within this core combined with an unknown sedimentation rate. A small Asteraceae peak occurs at 100cm depth.

244

Figure 10.4 Sediment description for Lake Willeroo augered sediments. 245

Figure 10.5 Pollen counts for Lake Willeroo auger hole.

246

10.2.2.2 Phytoliths

While recorded at generally low numbers (rarely exceeding 20 per sample), the phytoliths were persistently present throughout the 410cm of the core with major peaks at 0, 150, and 275cm (Figure 10.6). The most interesting feature of this record is the clear dominance (except for the surface sample) of the Dicotyledon/Monocotyledon (non-Poaceae) type. The Poaceae type, in turn, seems to rise to unprecedented importance in modern times.

Figure 10.6 Biogenic silica counts for Lake Willeroo auger hole.

247

10.2.2.3 Other biogenic silica

A feature that sets Lake Willeroo apart from other lakes within the scope of this study is the consistent, extremely high concentrations of freshwater sponge spicules that were not encountered in any other lake (not even in Cummeroo Waterhole, which also recorded high numbers of spicules, but far lower than for Lake Willeroo) (Figure 10.6). There are no clear indications why this particular lake should be such an attractive habitat for sponges.

10.2.2.4 Other signs of plant and animal life

The plant fragments, seeds, and charophyte oospores were relatively rare within the sediment and were not encountered below a 2m depth (Figure 10.4). Only one oospore (from surface sediments) was identified and attributed to Chara cf. globularis. The fresh to subsaline water preferences of this charophyte (Appendix 8) fit well within the modern range of lake salinities.

10.2.2.5 Carbonates

There are three peaks in carbonate content at 100, 175, and 350cm (Figure 10.7). All three peaks are coincident with slight increases in the frequency dependent susceptibility, possibly linked to a weak pedogenesis.

10.2.2.6 Magnetic susceptibility

The magnetic susceptibility and frequency dependent susceptibility values for Lake Willeroo sediments are generally low, ranging 6.2-13.1x10-8m3/kg (at low frequency) for the former and 1.8-4.0% for the latter (Figure 10.7). There are, however, some exceptions. The first one is a peak in magnetic susceptibility at 300cm, reaching a maximum value for the core of 16.5x10- 8m3/kg. While the lack of a corresponding increase in frequency dependent susceptibility excludes pedogenic or fire related enhancement, a coincident peak in sand sized particles might indicate increased influx of oxidised (iron rich) sands of terrestrial or riverine origins.

The maximum values of frequency dependent susceptibility (4.6-5%) occur at a depth of 185- 200cm and two very minor peaks occur at 100cm (3.7%) and 350cm (3.9%). While they are not reflected by overall magnetic susceptibility enhancement trends, they concur with peaks in carbonate content. Since both proxies are often associated with soil development, a very weak soil signature is possible.

10.2.2.7 Mineral composition: clay content

The kaolinite and montmorillonite dominate the sediments throughout the core. This, together with high similarity in shape and strength of the spectra between individual samples, suggests mineral homogeneity of the sediments, thus little, if any, change in sediment sources (Figure 10.8).

248

Figure 10.7 Lake Willeroo auger hole: sediment diagram. 249

Figure 10.8 Infrared spectrometry data (SWIR) for Lake Willeroo auger hole.

250

10.2.3 Palaeoenvironmental reconstruction of events

General comments

The lack of substantial variability in any of the proxies suggests a relatively stable depositional environment throughout the cored history of the lake. The high sponge spicule concentrations hint at persistence of waterlogged conditions, thus failing to record any phases of major water deficient.

Event A, C, and D: Soil-increased plant cover phases: floodplain swamp? (325-350cm, 150- 175cm, 75-100cm)

Each of the three sediment sections (325-350cm, 150-175cm, 75-100cm) is represented by a pair of samples. The older samples within the pairs are characterised by a weak frequency dependent susceptibility enhancement and an increased carbonate content (i.e. a weak soil signature). In addition, the samples in the two upper units (at 175cm and 100cm) are roughly associated with plant fragments. The younger samples within these pairs, in turn, record an increase in phytolith concentrations. The best interpretation of these pairs is an existence of well vegetated floodplain swamp with the bottom section of each of these three sediment unit representing the root zone with weak soil development. The upper sections contain remains of the above ground non-Poaceae vegetation. As the plants were covered by mud during a subsequent flood event, their decomposition resulted in formation of a phytolith-rich sediment units.

The, phytolith-poorer but sponge spicule-rich, grey sediments that accumulated in periods between those floodplain swamp indicate deeper aquatic environments such as a lake or a waterhole.

Event B: Increased aridity: mobilisation of sand dunes? (~300cm)

This mottled light grey and yellow-orange unit with a peak of sand and magnetic susceptibility might indicate a contribution from re-mobilised red sand dunes that were stabilised by plants during the preceding wetter phase. As the red-coloured sediment from the surrounding sand dunes and plains became trapped within the system of riverine channels, it was likely to be reworked by the river currents and/or bleached in a diagenetic environment, particularly under waterlogged conditions (Blodgett et al., 1993). This, in turn, would have led to partial removal of the iron coating and change of colour to a yellow-orange. Reworking of the red sediments (possibly containing catchment’s topsoils) could have also removed any remnant superparamagnetic particles causing similar magnetic susceptibility values at both high and low frequencies, but larger carbonate nodules were likely to be retained.

251

Event E: Most recent times (0-25/100cm)

The upper section of the sediments shows interesting changes in vegetation composition. The pollen record suggests Asteraceae dominance in the past, which was temporarily taken over by Chenopodiaceae. However, most recently the Chenopodiaceae proportions declined again in favour of herbs and grasses (including Asteraceae). There was also some increase in the Myrtaceae. Furthermore, the phytolith record shows an unprecedented increase in Poaceae.

While there are no absolute dates to support this interpretation, it is possible that the first Asteraceae peak occurred during above average rainfall pre-1895 (Figure 2.5). The Chenopodiaceae took over when the conditions deteriorated in the late 1890s to the 1940s. The prolonged drought might have also contributed to establishment of trees in the lake basin. The wetter conditions of the 1950s and the 1970s were likely to be responsible for drowning these trees as well as a new ‘boom’ in Asteraceae within the lake’s catchment. The lower, but still substantial, rainfall regime that persisted to present day was likely to secure a periodic filling that was adequate to prevent a new wave of lake floor colonisation by trees.

The increase in the Poaceae pollen and phytoliths might be, in turn, related not only to the persistence of generally wetter conditions but also pasture promoting activities, such as grazing, clearance of shrubs, and introduction of exotic pasture grasses. A compilation of land management history of the property might be useful in testing of this interpretation.

10.3 Lake Yandaroo

10.3.1 Lake and site description

Lake Yandaroo, when full, covers around 56ha and is relatively shallow with depths below 1m (Timms, 1993). The lake is an alkaline (mean pH of 9.4) fresh to hyposaline lake with recorded salinity ranging between 0.11-4.31g/L (Timms, 1998a). Like Lake Willeroo, it lies within a narrow band of the Warrego River’s floodplain deposits, but unlike Lake Willeroo, it is surrounded by sand plains consisting of red sand and silt often containing secondary amorphous limestone and gypsum (Louth 1:250 000 Geological Series Map Sheet SH 55-9, 1965).

Similar to Lake Willeroo, Lake Yandaroo is surrounded by a relatively dense band of trees and shrubs (Figures 10.1, 10.9, and 10.10), including black box (Eucalyptus largiflorens), coolibah (E. coolabah), western boobialla (Myoporum montanum), rice flower (Pimelea microcephala), and prickly wattle (Acacia victoriae). The lake’s margins are surrounded by rushes, including spiny sedge (Cyperus gymnocaulos), and herbs, including different Malvaceae, Poaceae, and Asteraceae species as well as New Zealand spinach (Tetragonia tetragonoides) (Figure 10.9). On drying the lake floor becomes densely covered by hairy carpet weed (Glinus lotoides).

252

Figure 10.9 Dense sedges on the southwestern margin of Lake Yandaroo with the tree and shrub band on the right side of the photo.

10.3.2 Results

Two sediment cores were collected from Lake Yandaroo. An augered core was extracted from the northern part of the lake in 2001 and a PVC pipe core in three parts (a, b, and d) was extracted from the southern part of the lake in 2004 (Figure 10.1). Both cores were analysed in detail. The results for the augered core are presented in Figures 10.11-10.15 and sections 10.3.2.1-6 and for the PVC core in Figures 10.16-10.20 and sections 10.3.2.7-12.

10.3.2.1 Augered core: 137Caesium

The 137Cs in Lake Yandaroo augered core does not occur below 20cm and appears to be associated with a red sediment unit (Figure 10.11), thus suggesting its deposition post-1950s. The lack of information about the 137Cs mBq/cm2 concentrations for this core, due to a sampling error, prevents comparison with the reference value.

253

Figure 10.10 View of Lake Yandaroo from its northeastern shore. Sparse woody shrubs scattered in the foreground. A band of trees and shrubs surrounding the lake is in the background.

254

Figure 10.11 Sediment description for Lake Yandaroo auger hole. The 137Caesium is in mBq/g units as the calculation of mBq/cm2 was impossible due to lack of adequate information on total sample size.

255

10.3.2.2 Augered core: pollen

Of all the lakes analysed in this study, Lake Yandaroo produced by far the richest pollen record. The pollen in the augered sediment is dominated by Chenopodiaceae and Asteraceae, with all of the other types decreasing sharply in numbers with depth, possibly due to preferential preservation (Figure 10.12). The lowest depth is dominated by Asteraceae pollen, which decreases in proportion toward the depth of 150cm, only to regain its dominance to an even greater degree at 125cm. After this, Chenopodiaceae recover again to a peak concentration at the 75cm depth. They are, however, soon replaced by Asteraceae, which remain the dominant taxa until the present time.

An interesting feature of the pollen data below 100cm is the coincidence of the lower pollen concentrations with Asteraceae dominance and an increase in total concentrations accompanied by an increase in Chenopodiaceae proportions. This possibly suggests higher sediment deposition rates during the wetter (Asteraceae dominated) periods with lower rates during the drier (Chenopodiaceae recovery) periods.

10.3.2.3 Augered core: phytoliths

The phytolith data for Lake Yandaroo cores, in contrast to Lake Willeroo, show a distinct dominance of the Poaceae types (Figure 10.13). Two messages are delivered by the phytolith proxy from this core. The first one is a slight decrease in total counts within the most recent sediments (0-50cm section), suggesting a slight decline in vegetation cover. The second is very slight increases in total counts at 125 and 175cm depths. While these peaks on their own provide only a weak evidence of increased vegetation densities, they lend support to the trends in pollen (above) and sponge spicule (below) records.

10.3.2.4 Augered core: other biogenic silica

The presence of freshwater sponge spicules within the augered sediments is much more limited than in the neighbouring Lake Willaroo (Figure 10.13). Their concentrations fluctuate within the core with most peaks (except for 75cm) coincident with peaks in herbs and grasses as well as peaks in total phytolith counts.

10.3.2.5 Augered core: other signs of plant and animal life

Except for a single charophyte oospore at 100cm, the recovery of charophytes, plant fragments and seeds, and gastropod shells is limited to the upper 25cm of the core (Figure 10.11). The only identified charophyte taxa, Chara cf. globularis, as well as the gastropod taxa, including Isidorella sp., Glyptophysa sp., and Physella sp., indicate a fresh to hyposaline water habitat consistent with the water properties recorded in the lake in modern times.

256

Figure 10.12 Pollen counts for Lake Yandaroo auger hole. 257

Figure 10.13 Biogenic silica counts for Lake Yandaroo auger hole.

10.3.2.6 Augered core: magnetic susceptibility

The red sediment in the upper 20cm of the core has the highest magnetic susceptibility values recorded in any other lake or dune within this study (i.e. 112.8 x10-8m3/kg at low frequency) (Figure 10.14). This magnetic susceptibility is accompanied by high frequency dependent susceptibility of 12.2-12.7%. The underlying greyish sediments are characterised by much lower magnetic susceptibility values of 15.7-17.5x10-8m3/kg and frequency dependent susceptibility ranging 2.2-3.9%. This probably represents a change from fluvial sedimentation within a stable landscape to a massive soil influx from an eroding catchment.

Finally, there is a slight reddening of sediment extracted from the auger hole below about 160cm. This is accompanied by a drop in magnetic susceptibility to 9.1-10.2x10-8m3/kg, but an increase in frequency dependent susceptibility (4.4-6.8%), suggesting some links with pedogenesis or fires.

258

Figure 10.14 Lake Yandaroo auger hole: sediment diagram.

259

Figure 10.15 Infrared spectrometry data (SWIR) for Lake Yandaroo auger hole.

10.3.2.7 PVC core: 137Caesium

The 137Cs in Lake Yandaroo PVC core disappears at a lower depth than in the augered core, i.e. below 30cm depth (Figure 10.16). In contrast to the augered sediments, the 137Cs is present in only half of the red sediment profile. It is likely that the combination of the closer location to the inflow channel (with potentially higher sedimentation rates), the creek’s hydrological characteristics, and higher erosion rates in its catchment are responsible for the differences.

The main conclusion from the 137Cs data is that there was a substantial sediment accumulation in relatively recent times, possibly the last 100 years, based on the depth of the 137Cs profile as well as the total 137Cs content of 138.60mBq/cm2 ±15.08, which is about four and half times that of the calculated reference value for the region, i.e. 31mBq/cm2.

260

Figure 10.16 Sediment description for Lake Yandaroo cores a, b, & d.

10.3.2.8 PVC core: pollen

Like in the augered core, this pollen record is dominated by the Asteraceae pollen with Chenopodiaceae being a second important taxa (Figure 10.17). It is noteworthy that any increases in Asteraceae are accompanied by decreases in Chenopodiaceae and vice versa, representing fluctuations between wetter and drier conditions. A peak in Cyperaceae, herbs, grasses, trees and shrubs and a substantial dip in Chenopodiaceae at 60-70cm depth are a proxy for a particularly wet period in the history of the lake.

261

Figure 10.17 Pollen counts for Lake Yandaroo cores a, b, & d.

262

10.3.2.9 PVC core: phytoliths

The PVC core’s phytolith record can be divided into three main sections: 0-50, 60-70, and 75- 100cm (Figure 10.18). Between 0 and 60cm the proportions between the Poaceae and Dicotyledons/Monocotyledons (non-Poaceae) tend to be stable with grass proportions exceeding those of the non-grass types. A distinct peak in Poaceae at 60-70cm, similar to the one in herbs and grasses in the pollen data, suggests a short phase of vegetation vigour.

Figure 10.18 Biogenic silica counts for Lake Yandaroo cores a, b, & d.

Finally, the section of 75-100cm is characterised by a substantial decrease in total counts accompanied by a switch in dominance from grasses to the Dicotyledon/Monocotyledon (non- Poaceae) types, suggesting a marked change in vegetation densities and composition.

263

10.3.2.10 PVC core: other biogenic silica

The freshwater sponge spicules are present within the core in lower densities in the 80-100cm section and at 0cm depth, and in higher densities within the 10-75cm section, further highlighting the change from drier to wetter conditions that took place at about 75cm (Figure 10.18).

10.3.2.11 PVC core: other signs of plant and animal life

The charophyte oospores, plant seeds and fragments and broken gastropod shells are limited to the upper 40cm of the core, and roughly overlap with the red sediment unit. Similarly roots and roots channels are restricted to this unit, particularly to the 17-52cm depth, suggesting temporary plant invasion of the lake floor during a longer dry period.

10.3.2.12 PVC core: magnetic susceptibility

The magnetic susceptibility values associated with the red sediment in the PVC core are even higher than in the augered core reaching a maximum of 228.3x10-8m3/kg (Figure 10.19). The frequency dependent susceptibility is also high, ranging between 11.5-12%. In spite of some signs of root presence, particularly at 16.5-20cm depth, the dominant clear stratification of the sediment in the form of multiple, thin to thick, layers and marked particle size differences between them suggest that an erosion of the catchment’s topsoil is the major source of the magnetic enhancement rather than in situ pedogenesis.

As in the augered core, the greyish olive sediments below the red unit are characterised by markedly lower magnetic susceptibility of 16.2-29.5x10-8m3/kg as well as frequency dependent susceptibility of 5.9%.

10.3.3 Palaeoenvironmental reconstruction of events

Event A: Pre-lake floodplain (augered core >160cm)

The sediments of this section contain a mixture of grey fluvial clays and reddish sands with a slightly enhanced content of superparamagnetic particles characteristic of soil formation. This indicates a floodplain subjected to deposition of riverine sediments as well as influxes of coarser sands from nearby red sandplains. The area was relatively wet and supported herbs (including Asteraceae), grasses, and possibly also sparse trees.

264

Figure 10.19 Lake Yandaroo cores a, b, & d: sediment diagram.

265

Figure 10.20 Infrared spectrometry data (SWIR) for Lake Yandaroo cores a, b, & d.

Event B: Lake formation (augered core ~150cm)

The grey sediments of this unit mark the start of deposition of mostly finely textured, weakly magnetic aquatic muds that continued until quite recent times (i.e. 20cm depth). At the same time, the area experienced a period of increased aridity as evidenced by a substantial increase in the Chenopodiaceae pollen. It is possible that the increased aridity led to a decrease in vegetation cover, that resulted in an increased instability of the land surface. The increased sediment mobility, and possibly greater sediment loads, contributed to the formation of a dam, which was responsible for the formation of the lake.

Event C: Amelioration phase (augered core ~125cm)

An increase in Asteraceae proportions and freshwater sponge densities within this sediment unit suggests a temporary return of wetter conditions.

266

Event D: Drier, but not arid, phase (augered core ~25-125cm, PVC core >70cm)

The lake continued to accumulate fine grey sediments, while the pollen shows a persistence of approximately equal proportions of Chenopodiaceae and Asteraceae matched by low total phytolith counts dominated by non-grass forms. This might indicate regional aridity (with increased Chenopodiaceae pollen and scarce phytoliths) with locally moist conditions (that continued to support Asteraceae). Since Lake Yandaroo lies close to main channels of the Warrego River, it was more likely to retain moisture for longer periods of time in the midst of a drying landscape than more remote lakes.

Event E: Short wet phase: pre-1895 (?) (augered core ~25cm, PVC core 60-70cm)

The short wet phase is characterised by a renewed vigour in vegetation growth, shown by both pollen and phytolith data. Peaks were recorded in herbs (particularly Asteraceae), grasses, and even some tree and shrub taxa, while Chenopodiaceae proportions slumped. There is also an increase in freshwater sponge populations. Based on the location of this sediment unit with respect to the bottom the 137Cs profile in both cores and the trends in the region’s rainfall (Figure 2.5), it is likely that the unit was deposited during a period of above average rainfall prior to the 1895.

Event F: Relatively moist, high runoff erosion phase: 20th century (augered core 0-25cm, PVC core 0-60cm)

The drought of the late 1890s to the 1940s, combined with heavy overgrazing, stripped the surrounding landscape of its protective vegetation cover. The heavy rainfalls of the 1950s and 1970s (Figure 2.5) resulted in substantial soil erosion in the catchment (evidenced by enhanced magnetic susceptibility data, clear stratification, and limited root presence) and its deposition in the lakes. Considering that the bottom of the 137Cs profile in the PVC core occurs about half way through the red sediments, it is likely that the bulk of the lower portion was eroded and deposited in the 1950s, predating the detectable 137Cs deposition. The 137Cs-rich sediment, in turn, would have been eroded in the 1970s to modern times.

While the pollen and phytolith data suggest high proportions of herbs (mostly Asteraceae) and grasses in the record, it is uncertain if all the fossils were ‘freshly’ incorporated into the sediments at the time of deposition or some of them were recycled from the catchment’s soils, i.e. were produced during the previous wet phase in 1890s and eroded during the subsequent arid phase. The high potential for in situ grain deterioration makes it difficult to differentiate between the authigenic and allogenic pollen based on degree of damage to the grain. Both cores also show an increase in the trees and shrubs, including Eucalyptus, Casuarinaceae, and Sterculiaceae, in most recent times.

267

268

Chapter 11

Palaeoenvironmental history of the Paroo and Warrego Region: regional trends

11.1 Introduction

The lakes of the Paroo/Warrego Region have demonstrated their capacity for preserving high resolution records of environmental changes. Fluctuations in extreme conditions between wet/flood and drought episodes resulted in deposition of distinct, usually fine, sediment layers in the form of grey-red couplets and more extensive, thicker, bands during periods of long-term stability. Auxiliary proxies such as pollen, biogenic silica, and other plant and animal fossils that were preserved within those layers provided additional information on what happened in the lakes and their catchments, in many cases further improving the resolution of the record. This was demonstrated in Chapters 5-10 where sequences of events have been described for each of the lakes.

Following the compilation of the palaeoenvironmental histories of the individual lakes, it became clear that some events have been registered by more than one lake, suggesting responses to larger-scale regional phenomena. This chapter provides a summary of those events and the environmental and climatic changes behind them. It draws links between the records in the lakes of the region and compares them to corresponding changes reconstructed in other parts of the country, and particularly in the inland central and eastern regions. The precise correlations are, however, partly impeded by the limited absolute dating framework available for this study.

11.2 The oldest event: lacustrine phase

The direct evidence for an early lacustrine phase that supported large permanent (possibly also deep) lakes has been, so far, recovered solely from Palaeolake (Chapter 9 section 9.3). Although some of the lacustrine sediments in the lower parts of the Lake Bindegolly, Lake Wyara, and Lake Numalla might have also been deposited during that phase (Figure 11.1),

269

dating is crucial to confirm those links. Less direct evidence of the past, larger extent of the lake basins are the shoreline-shaped claypans east of Lake Numalla as well as high beach ridges on the northwestern margin and an ancient lunette some 400-700m east from the present shoreline in the neighbouring Lake Wyara (Timms, 2006). However, those features are so far un-linked to the cored sediments. Furthermore, it is also possible that the past extensive lakes discovered by Dury (1973) within the Bulloo-Bancannia system, west of the Paroo/Warrego Region, were full/active at that time.

The timing of the lacustrine phase can be attempted by comparison with other studies. West of the Paroo/Warrego Region, in Lake Eyre the last high lake water level event was estimated by Magee et al. (Magee and Miller, 1998; Magee et al., 2004) to occur at 70-60 000 years BP. However, wetter than present conditions, but drier than in the 70-60 000 BP, persisted between 50-30 000 BP (or 55 000 - 40 000 BP as suggested by Nanson et al. (1998)). While enhanced monsoon driven precipitation was likely to be the main agent of maintaining the high water levels in Lake Eyre (Johnson et al., 1999; Magee and Miller, 1998; Magee et al., 2004), Magee and Miller do not preclude the possibility of some contribution from the enhanced winter westerly circulation penetrating north more frequently than in modern times. Both systems would have been as likely to affect the Paroo/Warrego lakes with the main river catchments located north in the monsoon zone and the local catchment receptive to any potential rainfall carried by the westerlies.

In neighbouring Lake Frome, high water levels were reported to occur before 18 000 BP (Bowler et al., 1986). Further south, a period of pre-lunette building high water tables and full fresh lakes was reported to occur in the Willandra Lakes, including Lake Mungo, between about 50 000 and 36 000 BP (Bowler, 1983, 1990, 1998; Bowler et al., 1986) and in Lake Tyrrell before 30 000 BP (Bowler and Teller, 1986). Similarly, Page et al. (1996) have identified a period of increased discharges in the Murrumbidgee River palaeochannels between 55 000 to 35 000 BP (Kerarbury and Gum Creek phases).

On the other hand, lacustrine clays underlying clay-gypsum couplets in the Warrananga and Pine Camp lakes in the Darling Anabranch Region, southwest NSW, dated by Cupper (2005), returned ages of between 9 000 and 63 500 BP and 16 800 and 28 100 BP respectively.

270

Figure 11.1 Sedimentary summary of the lake cores from the Paroo/Warrego Region. The rainfall values are 5 year running means for an average of Thargomindah and Bourke annual precipitation. 271

11.3 Onset of aridity: red sand dune mobilisation (leading into the Last Glacial Maximum?)

Following the phase of high water levels was a period of extreme aridity. It has resulted in mobilisation of red sand dunes, one of which has blocked the course of the Bells Creek, thus effecting the formation of the Lower Bell Lake (Chapter 9 section 9.2). Another one has capped the lacustrine muds on the northern margin of the Palaeolake (Chapter 9 section 9.3). In addition, a relatively thick layer of oxidised sediments was deposited in the newly formed basins of Lower Bell Lake and on top of the (probably partially deflated) lacustrine muds of the Palaeolake (Figure 11.1). It is likely that phases of renewed temporary mobilisation of the red sand dunes (or at least their surface sediments) have continued during the subsequent increased seasonality/periodicity phase, as evidenced by the red sand contributions to the gypsum lunette’s sediments. Similarly, the sandy yellow orange unit preserved in the Lake Willeroo (Figure 11.1) was probably derived from the mobile red sand dunes (Chapter 10 section 10.2). Furthermore, it is also possible that the deeper lakes of the region, such as Lake Bindegolly or Lake Numalla, did not dry completely during this phase and the increased sand mobility was preserved within grey lacustrine clays as a red sandy mottling. Dating of these units would provide a better insight into origins of those sediment units.

While a lot was published on the sand dunes from other parts of the country, there is a distinct lack of information on the linear dunes east of the Cooper Creek catchment, i.e. in the Paroo/Warrego Region, and they are generally missing from any dune occurrence maps (e.g. Hesse and McTainsh, 2003; Wasson et al., 1988; Williams, 1994a; Williams et al., 1993). Therefore, the timing of their mobility can again be speculated only from correlation to extraregional sites.

In Lake Eyre the onset of extreme aridity occurred at about 50 000 years BP and resulted in deflation of large amounts of the formerly deposited lacustrine muds leading to the excavation of the modern basin (Magee et al., 2004). Similarly in the north, the Lynch’s Crater record shows change from open water lake and rainforest taxa towards swampy conditions and sclerophyll taxa around 46 000 BP (Turney et al., 2006). Lake Frome, however, was reported to dry much later, at about 18 000 BP (Bowler et al., 1986).

The activation of the linear dunes in the Lake Eyre Basin occurred somewhat later, at about 40 000 BP, and lasted until the Holocene (Hesse et al., 2004). In the Simpson Desert, Hollands et al. (2006) dated a phase of red sand dune formation between about 35 000 and 25 000 BP, while in its western part Nanson et al. (1995) identified two more recent periods of dune activity at 17

272

000 - 9 000 BP and 5 000 - 0 BP. In the Belarabon-Nulchara Lake region, west of Broken Hill, Williams et al. (1991) suggested deposition of aeolian quartz sands around 45 000 BP.

In southeastern Australia reactivation of the linear dunes resulted, similar to the Paroo/Warrego Region, in their transgression onto the exposed lake floors as well as contribution of silt sized material to lake and shoreline sediments (Bowler, 1978). Bowler (1983; 1998) and Bowler et al. (1986) suggested the phase of the extreme aridity and dune activity was closely related to the Last Glacial Maximum, thus occurring prior to about 18 000 and ceasing at about 15 000 BP. Similarly, Lake Tyrrell sediments recorded a major phase of aeolian activity between about 20 000 and 15 000 BP (Macumber, 1991) and the Warrananga and Pine Camp Lakes showed gradual reduction in vegetation cover leading to the LGM (Cupper, 2005). Hesse et al. (2005), however, suggested the onset of wind erosion in the southern part of the continent as early as about 59 000 - 73 000 years ago.

In the earlier years the increased aridity and dune activation was often attributed to a set of climatic conditions resulting from a combination of decreased humidity, increased summer exposure to radiant energy due to reduced cloud cover, decreased winter temperatures, and increased wind speeds (Bowler, 1978; Bowler and Wasson, 1984). Furthermore, Bowler and Wasson (1984) calculated that to mobilise most of the Australian sand dunes and exceed the erosional resistance of the exposed lacustrine sediments an increase in wind strength by about 20-30% would have been necessary. In support, Ash and Wasson (1983) demonstrated that even in modern times the vegetation cover over most of the Australian desert dunefields is insufficient to prevent major sand movement, pointing to inadequate wind velocity as the main limiting factor of dune formation.

The enhanced wind speed was, however, questioned by Hesse and McTainsh (1999), who, based on the variations in particle size of the dust recovered from the Tasman Sea, concluded that there was no apparent change in wind strength during the LGM to Holocene and instead ascribed a larger role in the dune mobilisation to the significant reduction in vegetation cover (Hesse et al., 2005). The stabilising role of the vegetation was observed much earlier by Bowler and Wasson (1984), who found the mainly low-mobility rounded dunes mostly in wetter parts of the dunefields, while the drier regions were more likely to contain highly mobile high-angle dunes.

Finally, the predominant east/southeast-west/northwest orientation of the linear sand dunes in the region (e.g. Figure 11.2) suggests the westerlies are the main agent in their formation. This issue is discussed in more detail in section 11.4.4.1 below.

273

Figure 11.2 Linear dunes northwest of Lower Bell Lake. (Image source: Brindingabba airphoto 372, 1997)

11.4 Enhanced seasonality/periodicity: lunette formation (LGM to late Pleistocene?)

11.4.1 Gypsum-rich ephemeral lake and gypsum dune formation

The onset of extreme aridity at the approach of the LGM led to falls in groundwater tables, which resulted in conversion of many freshwater lakes into gypsum-rich groundwater discharge basins, setting the scene for the next stage of their palaeoenvironmental history: the formation of gypsum lunettes (Bowler, 1983, 1998). The new phase was characterised by extreme climatic fluctuations, either seasonal, or more likely, irregular interannual in frequency (Chapter 2 section 2.5.4).

In the Paroo/Warrego Region, these new conditions led to deposition of alternate fine layers of grey clays during floods and coarser layers of oxidised red sediments during the dry and windy intervals (Figure 11.1). Initially many of the lakes, including Mid Blue, Wombah, Lower Bell,

274

Palaeolake, and possibly also Lake Bindegolly, were well supplied with gypsum, which upon the lake’s drying became subject to deflation from the lake floor and redeposition on the downwind margin as a lunette (Chapter 2 section 2.5.4).

The lunette structures further support the periodicity of the changes with several sections of sediment accumulated during the drier phases interlaced with units showing signs of temporary dune stabilisation by vegetation, probably during extended wetter phases. The prolonged drier intervals were, in turn, likely to favour plant invasion onto the lake floor and soil development (e.g. Palaeolake at about 14 440 BP) as well as increased incorporation of mobilised red sands into the lunette’s materials. The radiocarbon dating for Palaeolake core B suggests that the phase of enhanced fluctuations in climatic conditions commenced well before 14 440 BP (thus likely at the onset of the glacial phase) and ceased well after that date, but before 7 950 BP (Chapter 9 section 9.3).

The enhanced periodicity and gypsum lunette formation was timed for inland lakes by Warren (1982) to occur at 16 000 to about 14 000 years BP. Chen et al. (1993; 1991), however, obtained a much earlier date of about 45 000 - 60 000 BP for the formation of a gypsum lunette on the margins of Lake Amadeus in central Australia. In the closer located Belarabon-Nulchara Lake region, Williams et al. (1991) dated gypsum lunette with a resulting age range of 18 000 - 16 000 BP. The Lake Frome, in contrast, appears to have remained dry until about 14 000 BP (Bowler et al., 1986).

The Palaeolake’s record seems to correlate well with the initial estimation of an ephemeral phase of floods and droughts in Lake Eyre at 25 000 - 10 000 BP (Magee et al., 1995), but somewhat less with the later amended period of 35 000 -14 000 BP (Magee et al., 2004). While related to gypsum-rich lacustrine deposits, these Lake Eyre records are, however, not linked with any particular gypsum lunette formation event.

Further south, in southeast Australia, Bowler (1973; 1983) has suggested that lunette development took place from around 36 000 BP to 17 000 BP. At Lake Mungo this took place during two main episodic dry-wet lake stages (about 36 000-34 000 and 25 000 - 17/16 000 BP) separated by a brief wet phase (Bowler, 1990; Bowler et al., 1986). A closer ‘fit’ to the Palaeolake estimates is the Lake Victoria record, where the gypsum-rich sediments were dated to have accumulated within the lunette at a much later time, i.e. between about 17-10 000 years BP (Chen, 1995).

11.4.2 Gypsum deficient ephemeral lake and clay lunette formation

While retaining its enhanced seasonality/periodicity character with cycles of full and dry lake stages and continuing to support the clay efflorescence and lake floor deflation processes

275

(Chapter 2 section 2.5.4), the region’s climatic conditions became increasingly arid. The ‘active’ lake basins shrank, resulting in the formation of inner lunettes (e.g. Mid Blue Lake and Lake Wombah in Chapter 7). This new stage of lunette construction seems to be accompanied by a substantial decrease in gypsum availability as these lunettes contain much lesser amounts of the mineral (e.g. Mid Blue Lake). This observation is supported by the gypsum-deficiency observed in the upper part of the grey-red couplet section in Lower Bell Lake core, compared to its lower part (Chapter 9 section 9.2). It is unclear, however, why no distinct clay lunette has developed or was preserved in the Lower Bell’s basin.

One explanation for the gypsum deficiency is the attribution of gypsum supply to groundwater. The prolonged period of increased aridity was likely to result ultimately in a fall of the groundwater table beyond some critical depth that prevented further (seasonal or periodic) recharge into the lake, thus cutting out the gypsum supply.

11.4.3 Periods of depressed water budgets: soil formation

A very interesting period of soil development, that appears not to be limited to just this single lake basin, was dated within the Palaeolake sediments at around 14 440 BP (Figure 11.1 and Chapter 9 section 9.3). This palaeosol appears to have equivalents in other records from two nearby sites, both located just west of Broken Hill. The first one was described by Wasson (1979) within the sediments of Mundi Mundi alluvial fans and its dating turned up ages between 16 570 and 13 010 BP. The second one was described by Williams et al. (1991) in the Belarabon-Nulchara Lake region from a variety of sediments including alluvial fans, dunes, lake margin clays, and palaeochannel alluvium. The dating of the palaeosol resulted in ages between 15 500 and 13 500 BP.

To allow formation of such a soil unit, the conditions would have to remain relatively dry for a considerable period of time, but not too dry to prevent plant growth. Further, sediment accumulation rates would have to decrease significantly to allow progression of the pedogenic processes (Bowler, 1998). Bowler and Teller (1986) suggested that in a saline lake basin, the plant invasion could only take place if the watertable fell to such a depth that it allowed downward leaching of the salts from the near surface sediments. It is unclear, however, to what extent salinity was an issue within the Palaeolake at that time, as there is lack of information on the past salinities. The modern salinities, at least in the Paroo/Warrego Region, do not seem to pose much barrier to plant invasions, particularly by the salt tolerant species such as Halosarcia or Frankenia species.

It is also curious that the palaeosol event roughly coincides with the onset of the driest phase in the Lynch’s Crater in the north, which occurred at about 15 000 BP (Kershaw, 1995), and in the

276

south with the return of high water levels to Lake Frome at about 14 000 BP (Bowler et al., 1986). This suggests a major adjustment and/or shift in the weather patterns at that time with likely intensification of the westerlies and depression of the monsoon.

11.4.4 The general weather patterns in the Paroo/Warrego Region from the last glacial phase to the early Holocene

11.4.4.1 Wind direction and intensity

For a long time it was argued that one of the main factors in lunette formation in southeastern Australia during the last glacial period was the increase in strength, and possibly also frequency, of the winds with a strong westerly component (Bowler, 1973, 1978; Bowler et al., 1986; Bowler and Wasson, 1984; Magee, 1991). Bowler and Wasson (1984) suggested an increase in wind strength up to about 20-30%. The increase in wind speed was attributed by Bowler (1973) to the equatorward compression of the Southern Ocean westerly stream.

The ‘increased wind speed theory’, however, was found to be in disagreement with the conclusions of later dust studies from the Tasman Sea by Hesse and McTainsh (1999; 2003) and Hesse et al. (2004), who concluded that there was no apparent change in wind strength during the Last Glacial to the Holocene. Dodson’s wind reconstruction using the Milankovitch cycle (Dodson, 1998) also suggested weaker westerlies around 18 000 – 11 000 BP (at the time of lunette formation) with subsequent enhancement later, at the Pleistocene-Holocene boundary at about 11 000 – 9 500 BP.

The increased erosion and transport of dust during the last glacial period was instead attributed by Hesse et al. (2005; 2004) to the reduction in plant densities. The idea was, however, not totally new. Much earlier, Bowler (1978) suggested that the westerlies, which originate within the continental interior, even in modern times are capable of advective transport of high amounts of sensible heat, particularly in summer. In the past, the lower sea levels and the resultant increase in the continent’s size were likely to promote further warming of the wind, contributing to the increase in evaporation rates (Bowler, 1978; Bowler and Wasson, 1984). Such hot dry winds would have suppressed plant growth, particularly during summer months (Bowler, 1978), and further stresses would have been placed by cold and frosty winters (Bowler, 1978; Bowler and Wasson, 1984), with temperatures up to 9oC lower than at present, as well as decreased amounts of atmospheric carbon dioxide (Hesse et al., 2004).

In terms of wind patterns, it is generally accepted, and supported by different lines of evidence, that during the last glacial arid phase the westerlies were reaching further north by about: 10o of latitude as suggested by Knight et al. (1995); 5o by Sprigg (1982); and 1-1.5o by Hollands et al. (2006) and Nanson et al. (1995). 277

The interpretation of wind-related features in the Paroo/Warrego Region is in broad agreement with many of the above theories. The location of both the older gypsum and younger clay lunettes on the northeastern to eastern margins of the lake basins and the prevalence of east/southeast-west/northwest orientation of the red sand dunes indicate a legacy of the westerly to southwesterly winds during the drier periods. Considering the modern westerly component is of minor significance and decreases even further northwards in favour of the mainly easterly component (Figures 2.2 and 2.3), the past westerlies had to have extended further north (at least as far as south Queensland) to effect the dune formation. On the other hand, the modern significance of the small westerly component can not be easily disregarded in the face of the evidence of possibly continued, even if at a much lower rate, growth of the clay lunette at the Mid Blue Lake (Chapter 7 section 7.1).

11.4.4.2 Seasonality/periodicity enhancement and rainfall

One of the common theories is that the gypsum lunette formation requires strongly enhanced seasonality with wet winters and hot dry and windy summers (Bowler, 1973; Bowler and Wasson, 1984; Chen et al., 1991; Magee, 1991). The winter rainfall was more likely to be effective in recharging the groundwater and the hot summers were necessary to produce adequate evaporation to allow gypsum precipitation and some deflation of the lake floor (Bowler, 1976, 1978; Chen et al., 1991).

The potential sources of water during that time in eastern Australia were several. In the tropical north, Nanson et al. (1998) proposed enhanced flows/runoff during the LGM due to enhanced monsoon with the monsoonal floods occurring as isolated/occasional but more extreme than modern events. In addition, Bowler and Wasson (1984) suggested that the suppression of plant growth (described in section 11.4.4.1 above) would have resulted in increased runoff, thus increasing the rainfall’s competency in filling of lakes. With the Paroo and Warrego rivers headwaters in the monsoonal summer rainfall zone and the winter rainfall bringing westerlies overhead the study area, it is unclear which of the systems and factors (or all) were responsible for the extensive floods contributing water and fluvial sediments to the Paroo/Warrego lakes.

Although stronger seasonality with an enhanced summer-winter cycle was implicated in the formation of the gypsum lunettes in southeastern Australia (Bowler, 1973, 1983), the frequency of the changes does not seem to fit well the Paroo/Warrego Region. Even assuming the winter rainfall was responsible for filling the lakes, the prevalence of the westerlies during the winter to spring (Figures 2.2 and 2.3), with the highest intensity in winter, lowers the likelihood of effective deflation of the lake basin during summer, when it becomes dry. While the adequacy of weaker summer westerlies or a shift in the timing of prevalent westerlies to match their

278

modern February to April occurrence in southern Australia (Bowler, 1973, 1983) can not be disregarded, their impact would be trivial if the summer monsoonal storms were contributing to the Paroo/Warrego water balances.

Further, even considering loss of some sedimentary couplets due to deflation, their numbers in the Paroo/Warrego lakes are still rather low for annual accumulation over at least two thousand years of the proposed lunette formation. Finally, under the modern climatic conditions, it took nearly 50 years of suppressed rainfall to produce within the lake sediments an oxidised layer roughly similar to the ones present within the couplets section, while shorter term droughts (below 5 years) left no such evidence. Thus, even considering a persistently lower than modern vegetation cover and hotter, drier summers, the extent of the oxidation within the couplets appears too strong to result from a single summer season. In conclusion, while the individual lines of the argument are weak, together they support the periodic (a couple of years to decades cycles), rather than seasonal, fluctuations between the wet and dry phases in the Paroo/Warrego Region.

11.5 Late Pleistocene to Holocene amelioration and late Holocene changes

In most lakes the sediments below the oxidised deposits of the early 20th century drought (discussed in next section), and in the case of the Lower Bell Lake and Palaeolake above the grey-red couplet sections, appear to have been deposited in predominantly aquatic environments with a few signs of periods of increased dryness (Figure 11.1). The lack of information on variability of sediment deposition rates within individual lakes and between the different lakes makes it difficult to produce a high resolution history of changes for that unit as well as correlating the records.

In general terms, the increased wetness seems to correspond to the return of lacustrine conditions, with minor breaks, to the Lake Eyre basin between 10 000 and 3-4 000 years BP (Magee et al., 1995). While highly speculative, it is possible that one of the minor drier intervals during the Lake Eyre wet phase was responsible for prolonged dryness of the shallower and more sensitive Palaeolake and formation of a palaeosol (featured in core B) ‘just’ prior to 7 950 BP (Figure 11.1 and Chapter 9 section 9.3).

In Lake Frome this new wet phase lasted slightly longer, from about 14 000 to 5 000 BP (Bowler et al., 1986). Singh and Luly (1991) attributed this increased water availability to an enhancement of summer monsoon. Amelioration of conditions between 13 000 to 6 000 years BP was also reported from alluvial fans west of Broken Hill by Wasson (1979), while Williams

279

et al. (1991) dated high standlines at the nearby Lake Boola Boolka, located at the distal end of the Talyawalka Anabranch of the Darling River, to 7 500 - 7 100 BP.

To the north, in the Atherton Tablelands, a marked increase in water levels was reported for the early Holocene until about 4 000 BP, with maximum levels at about 8 000 BP (Kershaw, 1995). Very similar results were obtained from another north Australian site, the Groote Eylandt (Shulmeister in Donders et al., 2007). Similarly in the south, an increased moisture phase between about 9 000 and 4 000 BP was reconstructed by Cupper (2005) in the Warrananga and Pine Camp Lakes in the Darling Anabranch Region, while Lake Tyrrell remained permanent between 6 600 and 2 200 BP (Luly, 1993).

Much less apparent are the links of the Paroo/Warrego records with a phase of increased aridity within the last 5 000 years that was widely reported from other sites, e.g. since about ~ 4 000 – 3 000 BP from Lake Eyre (Magee et al., 1995; Magee et al., 2004), after 5 000 BP from Lake Frome (Bowler et al., 1986), and since about 4 500 BP from lakes in the Darling Anabranch Region (Cupper, 2005; Cupper et al., 2000). However, it might be possible that the Paroo/Warrego records from that time period are more closely related to the more variable results, for example, obtained from the Belarabon area, western NSW, which suggested two phases of wetter conditions: 6 500 - 4500 BP and 640 BP – present, with a drier interval at around 2 500 - 640 BP (Wasson, 1976), from Lake Callabonna and Lake Blanche, north of Lake Frome, which showed more positive than modern water balances during the late Holocene (<4-2 000 BP) (Nanson et al., 1998), or from Lake Tyrrell, which remained permanent in the period between 6 600 and 2 200 BP, but was dry between 2 200 and 800 BP (Luly, 1993).

Considering the Paroo/Warrego Region’s latitudinal location, making correlations to other regions without adequate dating becomes even more problematic in view of an argument supplied by Williams (1994b), who suggested that the responses of northern Australia to the postglacial warming appears to be different to that of the southern part of the continent and are driven by two distinctive climatic systems: the tropical summer monsoon and winter westerlies.

Finally, there is a general agreement as to the timing of the southwards shift in the westerlies, which were reported to have occurred in the northwestern Simpson Desert during the last 5 000 years (Hollands et al., 2006; Nanson et al., 1995). Assuming some lag due to the more southerly position of the Paroo/Warrego Region, it is still reasonable to interpret that the change to the southerly-dominated modern wind regimes, as evidenced by the wind roses in Figures 2.2 and 2.3 as well as the spit formation patterns in Lake Numalla (Figure 6.1), occurred relatively recently.

280

11.6 The late 1890s-1940s drought

The below average rainfall conditions persisting in the region from the late 1890s to the late 1940s seem to have left a distinct record within the lake sediments as well as the surrounding landscape. In lake sediments this is represented by an intensely oxidised red horizon often associated with signs of plant invasion. The unit is frequently sandwiched between layers of aquatic grey sediments deposited during the wet periods of the pre-1890s and the post 1940s (Figure 11.1).

A combination of two processes appears to be responsible for the deposition of the red unit, with both of them seemingly rooted in the substantial reduction in the catchment’s vegetation cover resulting from prolonged drought combined with effects of serious overstocking and overgrazing (Chapter 2 section 2.8.2). During the dry phase, the unprotected surface sediment from the surrounding sandplain became mobilised by winds, sometimes taking the form of dust storms, and later trapped by the exposed cracked and/or vegetated lake floors, resulting in deposition of coarsely (fine sand to silt) textured unit. The other mechanism responsible for the deposition of the red sediments was the runoff erosion of the catchment’s surface sediments in the initial stages of the subsequent extremely wet phase. The latter process resulted in deposition of a finely textured clay unit.

The recent, post-European settlement, extensive erosion in catchments and high deposition rates of the red sediments in wetland and creek systems of the arid and semi-arid regions were also noted by other researchers. For example, Timms (1997b; 2006) observed it in many lakes of the Paroo Region, Fanning (1999) on a floodplain at Fowlers Gap about 200km west of Currawinya NP, and Williams et al (1991) in Belarabon Gully about 600km west of Broken Hill.

The effects of the exacerbated post-European sedimentation on individual wetlands as well as regional ecosystems are largely unknown. However, Timms (2006) suggested that many of the Paroo lakes tend to hold water for a shorter amount of time after each filling event, thus prolonging the intervening dry periods and possibly lowering the resilience of the ecosystem to cope with the naturally occurring droughts. Change in the type of sediment from grey to red clays that occurred in the area within the last 100 years exacerbated the problem further. Since the red clays are much less susceptible to deflation than the friable grey clays, the effectiveness of lake floor wind erosion in slowing down or negating the sedimentation rates was severely reduced (Timms, 2006).

Other relicts of the long drought are the buried low beach at Lake Numalla and associated with it a line of dead trees, which mark the outline of the contacted lake basin active at that time (Chapter 6 section 6.3). The new rise in water levels not only resulted in renewed accumulation

281

of aquatic sediments on top of the lower beach, but also drowned the trees growing at lower elevation than the modern tree line. A similar fate was likely met by the trees on the margin of the nearby Lake Wyara (Chapter 6 section 6.2) as well as those that have invaded the Lake Willeroo’s basin (Chapter10 section 10.2), and possibly also some shrubs in Mid Blue Lake (Chapter 7 section 7.1).

It is interesting to note that the late 1890s-1940s drought, in spite of multiple records of drier phases, does not appear to have a similar equivalent in the Paroo/Warrego region’s Holocene past and the intensity of the sediment oxidation is matched only by the deposits from the period of extreme climatic fluctuations during the glacial period. The link is supported by Fossil 1, which although generally associated with lacustrine conditions, was recorded mainly from sediments of Lakes Bindegolly, Wyara, and Wombah that accumulated during the period leading into the drought as well as from the clay and gypsum lunettes at Mid Blue Lake.

Curiously, the drought remains generally unreported from other lakes. Although, it might have contributed to a significant fall in water levels in three maar lakes (Lakes Bullenmerri, Gnotuk, and Keilambete) in southeastern Australia, which was reported to occur during the last 100 years (De Deckker, 1982a).

11.7 General amelioration of conditions in the period of 1950s to present and increase in woody shrubs and trees

The early 20th century drought was succeeded by a wet phase characterised by above average rainfall around 1950 and in the mid 1970s (Figure 2.4), which resulted in rising lake levels that drowned low lying trees and shrubs (e.g. in Lake Willaroo and the Currawinya lakes) and reinstated a higher beach in Lake Numalla. It also renewed sedimentation of the grey fluvial clays in many of the lakes including Lake Bindegolly, Lake Wyara, Lake Numalla, and Mid Blue Lake (Figure 11.1).

In spite of the general amelioration, a change from Asteraceae/Poaceae to Chenopodiaceae dominated vegetation within the last few years, seen for example in Lake Bindegolly, Lake Wyara, Lake Numalla, Lake Wombah, and Palaeolake sediments, and to lesser degree in Cummeroo Waterhole and Lake Yandaroo, as well as the reappearance of the Unknown type 1 in the Mid Blue Lake, suggest a trend toward increased dryness of the region.

Furthermore, the pollen data in this study supported the observed increase in woody shrub populations (Chapter 2 section 2.6.1) and the general transition from grasslands towards woodland/shrubland communities during the last century as records from all of the lakes showed a general recent increase in woody taxa dominated by Eucalyptus spp., Casuarinaceae, and Rutaceae. The increase in woody shrubs, represented mainly by Dodonaea spp. and 282

Myoporaceae, was also recorded in the majority of the lakes, including Lake Bindegolly, Lake Wyara, Lake Numalla, Mid Blue Lake, Lake Wombah, Lower Bell Lake, Palaeolake (core K), and Lake Yandaroo, implying a widespread landscape change.

These conclusions need to be considered with caution due to the high possibility of a bias resulting from differential pollen preservation and the likely disappearance of the tree and shrub pollen down the core that often coincides with the general decline in pollen condition and concentrations. On the other hand, the increase in woody vegetation is supported by a pollen record from Cuddie Springs, west of Bourke, as the Eucalyptus component of the vegetation in the modern (surface) sediments at that site was higher than at any other time in the past and Casuarina levels were similar to those observed about 30 000 years ago (Field et al., 2002).

It is possible that the most recent increase in trees and shrubs is related to the unprecedented grazing pressures, particularly in the post-flood periods. The perennial plants in arid and semi- arid regions need large, therefore infrequent, floods for germination as they can only establish on waterlogged soils that allow development of taproots to access the groundwater (Westbrooke and Florentine, 2005; Westbrooke et al., 2005). The preferential grazing on grasses and herbs disadvantages them in the competition with the less palatable trees and shrubs during the initial stages of plant growth, optimising the establishment potential of the latter (Westbrooke and Florentine, 2005; Witt et al., 2006).

Another factor supporting the expansion of the trees and shrubs might be the post-European settlement reduction in fire occurrence, particularly in the post-flood period characterised by high biomass (thus adequate fuel) and juvenile growth forms of the woody taxa (thus more vulnerable to fire damage). In the later years, shading from the established trees and shrubs might add to the grazing pressures, further suppressing grass densities. The next wet phase is thus likely to meet a changed composition of the seed bank in favour of the tree and shrub component, perpetuating the change from grassland to woodland/shrubland communities.

11.8 Lake Willeroo and Lake Yandaroo: the Warrego River floodplain

The records from Lake Willeroo and Lake Yandaroo (Chapter 10) present a somewhat different story to the other lakes of the Paroo/Warrego Region, probably due to their location within the channel/floodplain part of the Warrego River. Except for the uppermost post-European red sediment layer, Lake Yandaroo appears to have retained permanent water throughout its existence, while the more peripheral Lake Willeroo appears to have been more exposed to swamp-flood fluctuations.

The origins of the lakes are not totally clear. Timms (1993) suggested that they have formed due to natural damming of an old river channel. The nearby Cooper Creek, that also has its

283

headwaters in the monsoonal north, was reported to change from mixed-/sand-load to mud- dominated system between 200 000 to 50 000 years BP (Nanson et al., 1986; Rust and Nanson, 1986) with muds filling the channels around 80 000 - 70 000 BP and again around 40 000 BP, probably due to reduction in flow volumes and energies (Maroulis et al., 2007). It is likely that the same climatic drivers and hydrological responses affected the Warrego River, silting up its channels and leading to formation of the Yandaroo and Willaroo Lakes.

11.9 Trends in vegetation successions and a lesson in resiliency: a synopsis

The pollen and phytolith records from the Paroo/Warrego Region showed an interesting succession of cycles that are closely linked to the wet/flood-drought fluctuations dominating the region’s climate since at least the beginning of the last glacial period. Thus several of the pollen records (from Lake Bindegolly, Lake Wyara, Lake Numalla, Lake Wombah, Cummeroo Waterhole, Lower Bell Lake, Palaeolake (core K) and Lake Yandaroo) showed initial booms in Asteraceae and Poaceae populations triggered by the onset of wet phases, such as these in the 1950s and the 1970s. As the conditions became drier (e.g. towards the year 2000) and the herbs and grasses decreased in significance, the Chenopodiaceae became the dominant taxa. It is possible that the trend was exaggerated by selective grazing.

The occurrence of this succession is supported by records from sheep manure accumulations under shearing sheds on Ambathala Station recovered by Witt et al. (2006). In their study, the breaking of the drought in the 1950s was manifested by a peak in Asteraceae, followed closely by a peak in Poaceae, while Chenopodiaceae proportions remained low through this period. The 1960s, in turn, witnessed a substantial increase in the Chenopodiaceae with a clear decline in the other two taxa.

Another common succession cycle, observed in many of the phytolith records, are the oscillations in the ratio between the Poaceae and Dicotyledon/Monocotyledon (non-Poaceae) types. The dominance of the Poaceae phytolith forms seems to be closely related to periods of lacustrine phases of the lakes, while the non-Poaceae forms increase in proportions during drier intervals. A future improvement in the phytolith identification has the potential to allow better differentiation between the tree, shrub, and other non-Poaceae types, further enriching the record. In summary, the Asteraceae/Poaceae to Chenopodiaceae as well as the Poaceae/non- Poaceae ratios can provide a useful tool in reconstructions of wet-dry cycles, at least in the Mulga region.

Finally, the early 20th century drought provided a valuable insight into the vegetation survival tactics during prolonged dry periods. In Lake Numalla, the plants, including trees, followed the

284

margins of the shrinking lake basin, returning to the higher grounds with the onset of the new wet phase. In Lake Willeroo, the dry lake basin provided a refugia for the stressed vegetation and during the drought supported a moderately dense woodland, which returned to the lake’s shores after the lake started to fill again.

In spite of their seemingly immense importance, the role of the semi-arid wetlands in the survival of plants during long-term droughts remains largely unexplored. This study suggested that by a combination of the low relief of their floors, therefore a decreased distance to the water table, and probably even more significantly their role as runoff collectors during periods of reduced rainfall, the lake basins can act as refugia for plants and possibly also animal species. With the onset of the wetter periods they became the sources for recolonisation. Accordingly, the exacerbated filling of those wetlands with sediment can substantially reduce the resilience of the semi-arid landscapes to cope with climatic fluctuations, leading to a permanent decrease in vegetation cover, higher erosion rates, and even irreversible desertification.

285

286

Chapter 12

Conclusions and future directions

12.1 History and dynamics of the Paroo/Warrego landscapes

Research into the history of the Paroo/Warrego landscape provided insights into its resilience and responses to past fluctuations in water budget and changing weather patterns. While dating of the detailed reconstruction was constrained by the lack of funding, it was still possible to construct a regional history that correlated well with events recorded from northeastern, central, and southeastern inland regions of the Australian continent.

In summary, this research provides evidence of high lake water levels in the Paroo/Warrego Region prior to 30 000 BP. During the Last Glacial, the region experienced extreme aridity leading not only to a long term drying of the lakes, but also mobilisation of the red sand dunes. Sometime during the glacial phase, the aridity gave way to periodic fluctuations between extreme flood and drought events that probably lasted until 16 000 - 14 000 BP. In its initial stages, the new climatic regime resulted in formation of gypsum lunettes and later, following reduction in gypsum supply, clay lunettes. The orientation of both, red sand dunes and lunettes, points to the significance at that time of westerly to southwesterly winds, and thus a more northerly extent of the westerlies than in modern times.

Around the boundary between late Pleistocene and early Holocene the region once again returned to more stable and wetter climatic conditions (but somewhat drier than during the pre- LGM lacustrine phase) supporting permanent to semi-permanent lakes. A new strong aridity signal, comparable to the semi-regular Pleistocene droughts, was recorded in the Paroo/Warrego lakes during the late 1890s-1940s period of below average rainfall. It was followed by half a century of ameliorated conditions with two extremely wet phases in the 1950s and the 1970s.

Vegetation adapted to the highly variable conditions with herbs and grasses flourishing in response to the onset of a wet phase and being replaced by Chenopodiaceae as the landscape started to dry. The fresher lake basins and water courses were likely to provide refuge during prolonged arid phases and dispersal foci during intervening wetter periods enabling greater flexibility and enhanced resistance. 287

The natural cycles were impacted markedly by the introduction of European land use, which included increased grazing pressures during the drier periods, selective and intense grazing during plant regeneration at the onset of wetter intervals, and changes to the fire regimes. This appears to have led to the recent (and seemingly unprecedented) increase in the proportion of woody vegetation. The overall reduction in vegetation cover led also to accelerated catchment erosion and siltation of the various wetlands, decreasing their water storage capacity and subsequently increasing the severity of drought events.

12.2 Multi-proxy, multi-site approach

The reconstruction produced by this research was only possible with the use of the multi-proxy, multi-site approach. The mixed strategy provided a larger quantity of data and broadened the picture of ecosystem responses to environmental changes through integration of a wide range of living and non-living indicators. The complementary study of lake basin together with adjacent lunette and sand dune sediments helped to ‘patch up’ deflation gaps in lacustrine records and enriched the story by providing links between the fluvial and aeolian depositional events. Finally, the use of multiple lakes helped to discriminate between local responses of individual lakes and regional trends expanding the scale dimension of this study.

12.3 Future directions

The research identified also opportunities to improve the protocols of multi-proxy analysis as well as the need for better assessment of the relevance and suitability of standard methodologies in processing of semi-arid sediments. It highlighted issues such as a shortage of multi-fossil processing protocols, scarcity of comprehensive and relevant reference databases for identification purposes, and limited ability of the standard methodologies developed for lacustrine sediments in humid regions to cope with materials accumulating in arid and semi-arid lake systems. While some of these issues were addressed in this research, this foundational work is far from complete. The promising results of this research and the urgent need for palaeoenvironmental information in the face of the modern climate change challenge justify further development of the proxy data methodologies.

Furthermore, the discontinuous and variable sedimentation rates emphasized the need for intensive and multiple dating. A strategic dating of some of the key events, if funds could be found, would tighten the temporal framework for individual lake histories and enable correlation between different sites allowing better resolution of the regional-scale record of environmental changes.

Finally, the research showed that palaeoenvironmental research in desert ecosystems is not just limited to gaining a better understanding of their functioning but also has important implications 288

for a better commercial and protective management of those regions. Since the hydrological data sets for many semi-arid rivers, such as Paroo and Warrego, are rarely sufficiently long to show the medium-term (decadal) variability in flow regimes and water balances (Young, 1999), land and water management decisions are frequently based on short-term data (e.g. the above average rainfalls in the second half of the 20th century). There is also a limited knowledge of the principles governing sustainable functioning of the natural ecosystems. Thus, to avoid loss of diversity, irreversible damage, or desertification, a better understanding of the long-term trends (centennial), the landscape-scale interactions between its living and non-living components, and the mechanisms behind the resilience of the semi-arid ecosystems is urgently needed.

289

290

References:

Alliance Resource Economics, 2002. Social and Economic Analysis of the Paroo River Catchment, Alliance Resource Economics, South Brisbane. Arakel, A.V., 1986. Evolution of calcrete in palaeodrainages of the Lake Napperby area, central Australia, Palaeogeography, Palaeoclimatology, Palaeoecology 54:283-303. Aref, M.A.M., 2003. Classification and depositional environments of Quaternary pedogenic gypsum crusts (gypcrete) from east of the Fayum Depression, Egypt, Sedimentary Geology 155:87-108. Ash, J.E. and Wasson, R.J., 1983. Vegetation and sand mobility in the Australian desert dunefield, Zeitschrift fur Geomorphologie 45:7-25. Asselin, H. and Payette, S., 2005. Detecting local-scale fire episodes on pollen slides, Review of Palaeobotany and Palynology 137:31-40. Ball, D.F., 1964. Loss-on-ignition as an estimate of organic matter and organic carbon in non- calcareous soils, Journal of Soil Science 16:84-92. Bartoli, F. and Wilding, L.P., 1980. Dissolution of biogenic opal as a function of its physical and chemical properties, Soil Science Society of America Journal 44:873-8. Bates, C.D., Coxon, P. and Gibbard, P.L., 1978. A new method for the preparation of clay-rich sediment samples for palynological investigation, New Phytologist 81:459-63. Battarbee, R.W., 2000. Palaeolimnological approaches to climate change, with special regard to the biological record, Quaternary Science Reviews 19:107-24. Batten, D.J., 1999. Small palynomorphs. In: Jones, T.P. and Rowe, N.P. (eds), Fossil Plants and Spores: Modern Techniques, Geological Society, London, pp. 15-9. Beadle, N.C.W., 1948. The Vegetation and Pastures of Western New South Wales, Thomas Henry Tennant, Government Printer, Sydney. Behl, R.J. and Kennett, J.P., 1996. Brief interstadial events in the Santa Barbara basin, NE Pacific, during the past 60 kyr, Nature 376:243-6. Black, M.P., Mooney, S.D. and Martin, H.A., 2006. A >43,000-year vegetation and fire history from Lake Baraba, New South Wales, Australia, Quaternary Science Reviews 25:3003- 16. Blodgett, R.H., Crabaugh, J.P. and McBride, E.F., 1993. The color of red beds - a geological perspective. In: Bigham, J.M. and Ciolkosz, E.J. (eds), Soil Color, Soil Science Society of America, Madison, pp. 127-59. Blum, P., 1997. Physical Properties Handbook: a guide to shipboard measurements of physical properties of deep-sea cores, Ocean Drilling Program Technical Note 26, http://www- odp.tamu.edu/publications/tnotes/tn26/INDEX.HTM. Blume, T., Weisbrod, N. and Selker, J.S., 2002. Permeability changes in layered sediments: impact of particle release, Ground Water 40(5):466-74. Bowdery, D., 1998. Phytolith Analysis Applied to Pleistocene-Holocene Archaeological Sites in the Australian Arid Zone, John and Erica Hedges, Oxford. Bowler, J.M., 1973. Clay dunes: their occurrence, formation and environmental significance, Earth-Science Reviews 9:315-38. Bowler, J.M., 1976. Recent developments in reconstructing late Quaternary environments in Australia. In: Kirk, R.L. and Thorne, A.G. (eds), The Origin of the Australians, Australian Institute of Aboriginal Studies, Canberra, pp. 55-77. 291

Bowler, J.M., 1978. Glacial age aeolian events at high and low latitudes: a Southern Hemisphere perspective. In: van Zinderen Bakker, E.M. (ed.), Antarctic Glacial History and World Palaeoenvironments, Balkema, Rotterdam, pp. 149-72. Bowler, J.M., 1980. Quaternary chronology and palaeohydrology in the evolution of Mallee landscapes. In: Storrier, R.R. and Stannard, M.E. (eds), Aeolian Landscapes in the Semi- arid Zone of South Eastern Australia: Conference Proceedings, Riverina Society of Soil Science, Wagga Wagga, pp. 17-36. Bowler, J.M., 1983. Lunettes as indices of hydrologic change: a review of Australian evidence, Proceedings of the Royal Society of Victoria 95(3):147-68. Bowler, J.M., 1986. Spatial variability and hydrologic evolution of Australian lake basins: analogue for Pleistocene hydrologic change and evaporite formation, Palaeogeography, Palaeoclimatology, Palaeoecology 54:21-41. Bowler, J.M., 1990. Human occupation and environmental change: the ancient record from the Willandra Lakes. In: Noble, J.C., Joss, P.J. and Jones, G.K. (eds), The Mallee Lands: a Conservation Perspective, 152-61. Bowler, J.M., 1998. Willandra Lakes revisited: environmental framework for human occupation, Archaeology in Oceania 33:120-55. Bowler, J.M., Huang Qi, Chen Kezao, Head, M.J. and Yuan Baoyin, 1986. Radiocarbon dating of playa-lake hydrologic changes: examples from northwestern China and central Australia, Palaeogeography, Palaeoclimatology, Palaeoecology 54:241-60. Bowler, J.M. and Teller, J.T., 1986. Quaternary evaporites and hydrological changes, Lake Tyrrell, north-west Victoria, Australian Journal of Earth Sciences 33:43-63. Bowler, J.M. and Wasson, R.J., 1984. Glacial age environments of inland Australia. In: Vogel, J.C. (ed.), Late Canozoic Palaeoclimate of the Southern Hemishpere, Balkema, Rotterdam, pp. 183-208. Bowler, J.M., Wyrwoll, K.-H. and Lu, Y., 2001. Variations of the northwest Australian summer monsoon over the last 300,000 years: the paleohydrological record of the Gregory (Mulan) Lakes System, Quaternary International 83-85:63-80. Boyd, W.E., 1992. A Pollen Flora of the Native Plants of South Australia and Southern Northern Territory, Australia., Royal Geographical Society of Australasia (S.A. Branch), Brewer, R., 1955. The occurence of diatom skeletons and sponge spicules in the soils of New South Wales, The Australian Journal of Science 17(5):177-9. Brown, C.A., 1960. Palynological techniques, Louisiana State University, Baton Rouge. Brush, G.S. and Brush, L.M., 1994. Chapter 3. Transport and deposition of pollen in an estuary: signature of the landscape. In: Traverse, A. (ed.), Sedimentation of Organic Particles, Cambridge University Press, Cambridge, pp. 33-46. Bryant, V.M.J., Holloway, R.G., Jones, J.G. and Carlson, D.L., 1994. 4. Pollen preservation in alkaline soils of the American Southwest. In: Traverse, A. (ed.), Sedimentation of Organic Particles, Cambridge University Press, Cambridge, pp. 47-58. Buck, B.J. and Van Hoesen, J.G., 2002. Snowball morphology and SEM analysis of pedogenic gypsum, southern New Mexico, U.S.A., Journal of Arid Environments 51:469-87. Bullard, J.E. and McTainsh, G.H., 2003. Aeolian-fluvial interactions in dryland environments: examples, concepts and Australia case study, Progress in Physical Geography 27(4):471- 501. Bunn, S.E. and Davies, P.M., 1999. Aquatic food webs in turbid, arid-zone rivers: preliminary data from Cooper Creek, western Queensland. In: Kingsford, R.T. (ed.), A Free-flowing

292

River: The Ecology of the Paroo River, National Parks and Wildlife Service, Sydney, pp. 67-76. Bureau of Meteorology, 2006. Weather/climate data: Thargomindah and Bourke, Australian Government, http://www.bom.gov.au/climate/. Bureau of Meteorology, 2005. Flood warning system for the Paroo River, Australian Government, http://www.bom.gov.au/hydro/flood/qld/brochures/paroo/paroo.shtml. Bureau of Rural Sciences, 2006. Warrego River: River Basin Summary, Australian Government, http://www.connectedwater.gov.au/pdfs/catchment_423_0_summary.pdf. Burne, R.V., Bauld, J. and De Deckker, P., 1980. Saline lake charophytes and their geological significance, Journal of Sedimentary Petrology 50(1):281-93. Callen, R.A. and Farrand, M.G., 1987. Simpson Desert sand colour, Quaterly Geological Notes 104:2-5. Campbell, J.R., 2003. Limitations in the laser particle sizing of soils. In: Roach, I.C. (ed.), Advances in Regolith: Proceedings of the CRC LEME Regional Regolith Symposia, 2003, CRC LEME, Perth, pp. 38-42. Casanova, M.T., 1999. Plant establishment in Paroo wetlands: The importance of water regime. In: Kingsford, R.T. (ed.), A Free-flowing River: The Ecology of the Paroo River, National Parks and Wildlife Service, Sydney, pp. 138-48. Casanova, M.T., Garcia, A. and Porter, J.L., 2003. Charophyte rediscoveries in Australia: what and why?, Acta Micropalaeontologica Sinica 20(2):129-38. Chalson, J.M. and Martin, H.A., 1995. The pollen morphology of some co-occuring species of the family Myrtaceae from the Sydney Region, Proceedings of the Linnean Society of New South Wales 115:163-91. Chen, X.Y., 1995. Geomorphology, stratigraphy and thermoluminescence dating of the lunette dune at Lake Victoria, western New South Wales, Palaeogeography, Palaeoclimatology, Palaeoecology 113:69-86. Chen, X.Y., Bowler, J.M. and Magee, J.M., 1993. Late Cenozoic stratigraphy and hydrologic history of Lake Amadeus, a central Australian playa, Australian Journal of Earth Sciences 40:1-14. Chen, X.Y., Bowler, J.M. and Magee, J.W., 1991. Aeolian landscapes in central Australia: gypsiferous and quartz dune environmnents from Lake Amadeus, Sedimentology 38:519- 38. Chen, X.Y., Spooner, N.A., Olley, J.M. and Questiaux, D.G., 2002. Addition of aeolian dusts to soils in southeastern Australia: red silty clay trapped in dunes bordering Murrumbidgee River in the Wagga Wagga region, Catena 47:1-27. Chivas, A.R. and Bowler, J.M., 1986. Introduction - The SLEADS project, Palaeogeography, Palaeoclimatology, Palaeoecology 54:3-6. Chivas, A.R., DeDeckker, P., Nind, M., Thiriet, D. and Watson, G., 1986. The Pleistocene palaeoenvironmental record of Lake Buchanan: an atypical Australian playa., Palaeogeography, Palaeoclimatology, Palaeoecology 54:131-52. Clarke, J., 2003. The occurence and significance of biogenic opal in the regolith, Earth-Science Reviews 60:175-94. Cody, R.D., 1979. Lenticular gypsum: occurrences in nature, and experimental determinations of effects of soluble green plant material on its formation, Journal of Sedimentary Petrology 49(3):1015-28. Cody, R.D. and Cody, A.M., 1988. Gypsum nucleation and crystal morphology in analog saline terrestrial environments, Journal of Sedimentary Petrology 58(2):247-55.

293

Condon, R.W. and Stannard, M.E., 1956. Erosion in western New South Wales: I. General description of the Western Division, Journal of Soil Conservation Service of NSW 12:192-202. Costelloe, J.F., Powling, J., Reid, J.R.W., Shiel, R.J. and Hudson, P., 2005. Algal diversity and assemblages in arid zone rivers of the Lake Eyre Basin, Australia, River Research and Applications 21:337-49. Cunningham, G.M., Mulham, W.E., Milthorpe, P.L. and Leigh, J.H., 1992. Plants of Western New South Wales, Inkata Press, Melbourne. Cupper, M.L., 2005. Last glacial to Holocene evolution of semi-arid rangelands in southeastern Australia, The Holocene 15(4):541-53. Cupper, M.L., 2006. Luminescence and radiocarbon chronologies of playa sedimentation in the Murray Basin, southeastern Australia, Quaternary Science Reviews 25:2594-607. Cupper, M.L., Drinnan, A.N. and Thomas, I., 2000. Holocene palaeoenvironments of salt lakes in the Darling Anabranch region, south-western New South Wales, Australia, Journal of Biogeography 27:1079-94. Currey, D.R., 1994. Hemiarid lake basins: geomorphic patterns. In: Abrahams, A.D. and Parsons, A.J. (eds), Geomorphology of Desert Environments, Chapman & Hall, London, pp. De Deckker, P., 1982a. Holocene ostracods, other invertebrates and fish remains from cores of four maar lakes in Southeastern Australia, Proceedings of the Royal Society of Victoria 94(4):183-220. De Deckker, P., 1982b. Late Quaternary ostracods from Lake George, New South Wales, Alcheringa 6:305-18. De Deckker, P., 1988a. An account of the techniques using ostracodes in palaeolimnology in Australia, Palaeogeography, Palaeoclimatology, Palaeoecology 62:463-75. De Deckker, P., 1988b. Biological and sedimentary facies of Australian salt lakes, Palaeogeography, Palaeoclimatology, Palaeoecology 62:237-70. De Deckker, P., Bauld, J. and Burne, R.V., 1982. Pillie Lake, Eyre Peninsula, South Australia: Modern environment and biota, dolomite sedimentation, and Holocene history., Transactions of the Royal Society of South Australia 106(4):169-81. Dean, W.E.J., 1974. Determination of carbonate and organic matter in calcareous sediments and sedimentary rocks by loss on ignition: comparison with other methods, Journal of Sedimentary Petrology 44(1):242-8. Dearing, J., 1999a. Chapter 4: Magnetic susceptibility. In: Walden, J., Oldfield, F. and Smith, J.P. (eds), Environmental Magentism: a practical guide. Technical Guide, No. 6., Quaternary Research Association, London, pp. 35-62. Dearing, J.A., 1999b. Chapter 7: Holocene environmental change from magnetic proxies in lake sediments. In: Maher, B.A. and Thompson, R. (eds), Quaternary Climates, Environments and Magnetism, Cambridge University Press, Cambridge, pp. 231-78. Dearing, J.A., Dann, R.J.L., Hay, K., Lees, J.A., Loveland, P.J., Maher, B.A. and O'Grady, K., 1996. Frequency-dependent susceptibility measurements of environmental materials, Geophysical Journal International 124:228-40. Deer, W.A., Howie, R.A. and Zussman, J., 1992. An Introduction to the Rock-Forming Minerals, Hong Kong, Longman. Digerfeldt, G., 1986. Studies on past lake-level fluctuations. In: Berglund, B.E. (ed.), Handbook of Holocene Palaeoecology and Palaeohydrology, John Wiley & Sons, Chichester, pp. 125-43.

294

Dimbleby, G.W., 1957. Pollen analysis of terrestrial soils, New Phytologist 56:2-28. Dodson, J.R., 1998. Timing and response of vegetation change to Milankovitch forcing in temperate Australia and New Zealand, Global and Planetary Change 18:161-74. Dollery, C., 2001. Fire research on Currawinya National Park. In: Page, M., Evenson, C. and Whittington, A. (eds), Research Needs for Managing a Changed Landscape in the Hungerford/Eulo Region, School of Natural and Rural Systems Management, University of Queensland, Gatton, pp. 35-7. Donders, T.H., Haberle, S.G., Hope, G., Wagner, F. and Visscher, H., 2007. Pollen evidence for the transition of the Eastern Australian climate system from the post-glacial to the present-day ENSO mode, Quaternary Science Reviews 26:1621-37. Doyle, G.B., 1998. An Investigation of Recent Sedimentary Deposits within the Lower Hunter River Estuary, NSW., PhD thesis, Dept. of Geography and Environmental Science, University of Newcastle, Newcastle. Draper, J.J. and Jensen, A.R., 1976. The geochemistry of Lake Frome, a playa lake in South Australia, BMR Journal of Australian Geology and Geophysics 1:83-104. Drysdale, A., 1995. Using the history for the future in the mulgalands of south west Queensland. In: Page, M.J. and Beutel, T.S. (eds), Ecological Research and Management in the Mulgalands - Conference Proceedings, University of Queensland Gatton College, Australia, pp. 51-7. Dury, G.H., 1973. Paleohydrologic implications of some pluvial lakes in northwestern New South Wales, Australia, Geological Society of America Bulletin 84:3663-76. Edinger, S.E., 1973. The growth of gypsum: an investigation of the factors which affect the size and growth rates of the habit faces of gypsum, Journal of Crystal Growth 18:217-24. Elliott, G.L., Loughran, R.J., Packer, I., Maliszewski, L.T., Curtis, S.J., Saynor, M.J., Morris, C.D. and Epis, R.B., 1997. A National Reconnaissance Survey of Soil Erosion of Australia: New South Wales, The University of Newcastle, Newcastle. Elliott, G.L., Loughran, R.J., Shields, P. and Powell, B., 1996. A National Reconnaissance Survey of Soil Erosion Queensland, The University of Newcastle, Newcastle. Enache, M.D. and Cumming, B.F., 2006. Tracking recorded fires using charcoal morphology from the sedimentary sequence of Prosser Lake, British Columbia (Canada), Quaternary Research 65:282-92. English, P.M., 2001. Formation of analcime and moganite at Lake Lewis, central Australia: significance of groundwater evolution in diagenesis, Sedimentary Geology 143:219-44. Faegri, K. and Iversen, J., 1989. Textbook of Pollen Analysis. IV Edition, John Wiley & Sons, Chichester. Fanning, P., 1994. Long-term contemporary erosion rates in an arid rangelands environment in western New South Wales, Australia, Journal of Arid Environments 28:173-87. Fanning, P.C., 1999. Recent landscape history in arid western New South Wales, Australia: a model for regional change, Geomorphology 29:191-209. Field, J.H., Dodson, J.R. and Prosser, I.P., 2002. A Late Pleistocene vegetation history from the Australian semi-arid zone, Quaternary Science Reviews 21:1023-37. Frey, D.G., 1955. A differential flotation technique for recovering microfossils from inorganic sediments, New Phytologist 54(2):257-8. Gale, S.J. and Hoare, P.G., 1991. Quaternary Sediments: Petrographic Methods for the Study of Unlithified Rocks, Belhaven Press, New York.

295

Gammon, P.R., James, N.P., Clarke, J.D.A. and Bone, Y., 2000. Sedimentology and lithostratigraphy of Uper Eocene sponge-rich sediments, southern Western Australia, Australian Journal of Earth Sciences 47:1987-103. Garcia, A., 1994. Charophyta: their use in paleolimnology, Journal of Paleolimnology 10:43-52. Garcia, A., 1998. Nitella ignescens sp. nov. and N. ungula sp. nov. (Charales, Charophyta) from Australia, Phycologia 37:53-9. Garcia, A., 1999a. Charophyte flora of south-eastern South Australia and south-western Victoria, Australia: systematics, distribution and ecology, Australian Journal of Botany 47:47-26. Garcia, A., 1999b. Quaternary charophytes from Salina del Bebedero, Argentina: their relation with extant taxa and palaeolimnological significance, Journal of Paleolimnology 21:307- 23. Garcia, A. and Casanova, M.T., 2003. Lamprothamnium heraldii sp. nov. (Charales, Charophyta) from Australia: the first dioecious representative of the genus, Phycologia 42(6):622-8. Garcia, A. and Chivas, A.R., 2004. Quaternary and extant euryhaline Lamprothamnium Groves (Charales) from Australia: Gyrogonite morphology and paleolimnological significance, Journal of Paleolimnology 31:321-41. Garcia, A., Jones, B.G., Chenhalland, B.E. and Murray-Wallace, C.V., 2002. The charophyte Lamprothamnium succinctum as an environmental indicator: a Holocene example from Tom Thumbs Lagoon, eastern Australia, Alcheringa 26:507-18. Gasse, F., 2002. Diatom-inferred salinity and carbonate oxygen isotopes in Holocene waterbodies of the western Sahara and Sahel (Africa), Quaternary Science Reviews 21:737-67. Gasse, F., Barker, P., Gell, P.A., Fritz, S.C. and Chalie, F., 1997. Diatom-inferred salinity in palaeolakes: an indirect tracer of climate change, Quaternary Science Reviews 16:547-63. Gayler, L. and Pearson, S., 2003. Preparation of gypsum-rich sediments for determining organic matter content using loss on ignition, Quaternary Australasia 21(2):30-3. Gayler, L.M., 2000. Potential of a Small Semi-arid Zone Lake for Quaternary Studies in the Paroo, New South Wales., Honours thesis, School of Geosciences, University of Newcastle, Newcastle. Gell, P., 1997. The development of a diatom database for inferring lake salinity, western Victoria, Australia: towards a quantitative approach for reconstructing past climates, Australian Journal of Botany 45:389-423. Gell, P., Tibby, J., Fluin, J., Leahy, P., Reid, M., Adamson, K., Bulpin, S., MacGregor, A., Wallbrink, P., Hancock, G. and Walsh, B., 2005. Accessing limnological change and variability using fossil diatom assemblages, south-east Australia, River Research and Applications 21:257-69. Gell, P.A., Sonneman, J.A., Reid, M.A., Illman, M.A. and Sincock, A.J., 1999. An Illustrated Key to Common Diatom Genera from Southern Australia, Cooperative Research Centre for Freshwater Ecology, Thurgoona. Gentilli, J., 1972. Australian Climate Patterns, Nelson, Adelaide. Gillespie, R., Magee, J.W., Luly, J.G., Dlugokencky, E., Sparks, R.J. and Wallace, G., 1991. AMS radiocarbon dating in the study of arid environments: examples from Lake Eyre, South Australia, Palaeogeography, Palaeoclimatology, Palaeoecology 84:333-8.

296

Goodall, H., 1999. Contesting changes on the Paroo and its sister rivers. In: Kingsford, R.T. (ed.), A Free-flowing River: The Ecology of the Paroo River, National Parks and Wildlife Service, Sydney, pp. 179-200. Goodrick, G.N., 1984. Wetlands of north-western New South Wales, National Parks and Wildlife Service, Sydney. Grave, P. and Kealhofer, L., 1999. Assessing bioturbation in archaeological sediments using soil morphology and phytolith analysis, Journal of Archaeological Science 26:1239-48. Griffiths, T., 2001. One hundred years of environmental crisis, Rangeland Journal 23(1):5-14. Han, J. and Jiang, W., 1999. Particle size contributions to bulk magnetic susceptibility in Chinese loess and paleosol, Quaternary International 62:103-10. Harden, G.J., 1990. Flora of New South Wales, New South Wales University Press, Kensington. Harden, G.J., 1991. Flora of New South Wales, New South Wales University Press, Kensington. Harden, G.J., 1992. Flora of New South Wales, New South Wales University Press, Kensington. Harden, G.J., 1993. Flora of New South Wales, New South Wales University Press, Kensington. Harrison, F.W., 1988. Methods in Quaternary Ecology #4. Freshwater sponges., Geoscience Canada 15(3):193-8. Hart, D.M., 1988a. The plant opal content in the vegetation and sediment of a swamp at Oxford Falls, New South Wales, Australia, Australian Journal of Botany 36:159-70. Hart, D.M., 1988b. A safe method for the extraction of plant opal from sediments, Search 19(5- 6):293-4. Hart, D.M., 1992. A field appraisal of the role of plant opal in the Australian environment, PhD, School of Earth Sciences, Macquarie University, Sydney. Hart, D.M. and Humphreys, G.S., 1997. Plant opal phytoliths: an Australian perspective, Quaternary Australasia 15(1):17-25. Harvey, A.M., Foster, G., Hannam, J. and Mather, A.E., 2003. The Tabernas alluvial fan and lake system, southeast Spain: applications of mineral magnetic and pedogenic iron oxide analyses towards claryfying the Quaternary sediment sequences, Geomorphology 50:151- 71. Havinga, A.J., 1971. An experimental investigation into the decay of pollen and spores in various soil types. In: Brooks, J., Grant, P.R., Muir, M.D., van Gijzel, P. and Shaw, G. (eds), Sporopollenin, Academic Press, London, pp. Heathcote, R.L., 1994. Australia, Longman Scientific and Technical, Harlow. Heiri, O., Lotter, A.F. and Lemcke, G., 2001. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results, Journal of Paleolimnology 25:101-10. Herczeg, A.L. and Lyons, W.B., 1991. A chemical model for the evolution of Australian sodium chloride lake brines, Palaeogeography, Palaeoclimatology, Palaeoecology 84:43- 53. Hesse, P.P., Luly, J.G. and Magee, J.M., 2005. The beating heart: Environmental history of Australia's deserts. In: Smith, M. and Hesse, P.P. (eds), 23oS: Archaeology and Environmental History of the Southern Desserts, National Museum of Australia Press, Canberra, pp. 56-72. Hesse, P.P., Magee, J.W. and van der Kaars, S., 2004. Late Quaternary climates of the Australian arid zone: a review, Quaternary International 118-119:87-102.

297

Hesse, P.P. and McTainsh, G.H., 1999. Last Glacial Maximum to Early Holocene wind strength in the mid-latitudes of the Southern Hemisphere from aeolian dust in the Tasman Sea, Quaternary Research 52:343-9. Hesse, P.P. and McTainsh, G.H., 2003. Australian dust deposits: modern processes and the Quaternary record, Quaternary Science Reviews 22:2007-35. Hodgkinson, K.C. and Harrington, G.N., 1985. The case for prescribed burning to control shrubs in eastern semi-arid woodlands, Australian Rangeland Journal 7(2):64-74. Hoffmann, W., 1988. The significance of chironomid analysis (Insecta: Diptera) for palaeolimnological research, Palaeogeography, Palaeoclimatology, Palaeoecology 62:501-9. Holdaway, S.J., Fanning, P.C., Jones, M., Shiner, J., Witter, D.C. and Nicholls, G., 2002. Variability in the chronology of late Holocene Aboriginal occupation on the arid margin of Southeastern Australia, Journal of Archaeological Science 29:351-63. Holdaway, S.J., Fanning, P.C. and Witter, D.C., 2000. Prehistoric Aboriginal occupation of the rangelands: interpreting the surface archaeological record of far western New South Wales, Australia, The Rangeland Journal 22(1):44-57. Hollands, C.B., Nanson, G.C., Jones, B.G., Bristow, C.S., Price, D.M. and Pietsch, T.J., 2006. Aeolian-fluvial interaction: evidence for Late Quaternary channel change and wind-rift linnear dune formation in the northwestern Simpson Desert, Australia, Quaternary Science Reviews 25:142-62. Holmes, P.L., 1994. 2. The sorting of spores and pollen by water: experimental and field evidence. In: Traverse, A. (ed.), Sedimentation of Organic Particles, Cambridge University Press, Cambridge, pp. 9-32. Hopf, F.V.L., 1997. A Late Glacial and Holocene Vegetation History of Lake St. Clair, Tasmania., Honours, Dept. of Geography and Environmental Science, University of Newcastle, Newcastle. Horowitz, A., 1992. Palynology of Arid Lands, Elsevier, Amsterdam. Hotchkiss, A.T. and Imahori, K., 1988. Additional observations on Nitella verticillata (Characeae) from new locality in New South Wales, Proceedings of the Linnean Society of New South Wales 110(2):187-91. Hounslow, M.W. and Maher, B.A., 1999. Chapter 9: Laboratory procedures for quantitative extraction and analysis of magnetic minerals from sediments. In: Walden, J., Oldfield, F. and Smith, J.P. (eds), Environmental Magnetism: a practical guide. Technical Guide, No. 6, Quaternary Research Association, London, pp. 139-84. Huang, K. and Jian, Y., 1998. Thermal Analysis Curves of Minerals. In: Hatakeyama, T. and Liu, Z. (eds), Handbook of Thermal Analysis, John Wiley and Sons, Chichester, pp. 247- 336. Hunt, C.O., 1985. Recent advances in pollen extraction techniques: a brief review. In: Fieller, N.R.J., Gilbertson, D.D. and Ralph, N.G.A. (eds), Palaeobiological Investigations: Research Design, Methods and Data Analysis, B.A.R., Oxford, pp. 181-7. Hutton, J.T., Prescott, J.R. and Twidale, C.R., 1984. Thermo-luminescence dating of coastal dune sand related to a higher stand of the Lake Woods, Northern Territory, Australian Journal of Soil Research 22(1):15-21. Jacobsen, G.L., Jr. and Bradshaw, R.H.W., 1981. The selection of sites for paleovegetational studies, Quaternary Research 16:80-96. James, J.W., 1959. Erosion Survey of the Paroo-Upper Darling Region, The Journal of Soil Conservation Service of NSW 15(4):315-25.

298

James, J.W., 1960a. Erosion survey of the Paroo-Upper Darling Region. Part II: Land form and Soils, The Journal of Soil Conservation Service of NSW 16(1):41-53. James, J.W., 1960b. Erosion survey of the Paroo-Upper Darling Region. Part III. Vegetation and Pastures, The Journal of Soil Conservation Service of NSW 16(3):185-206. James, J.W., 1960c. Erosion survey of the Paroo-Upper Darling Region. Part IV: Erosion, The Journal of Soil Conservation Service of NSW 16(4):253-76. Johnson, B.J., Miller, G.H., Fogel, M.L., Magee, J.W., Gagan, M.K. and Chivas, A.R., 1999. 65,000 years of vegetation change in Central Australia and the Australian summer monsoon, Science 284:1150-2. Jones, P.A., Loughran, R.J. and Elliott, G.L., 2000. Sedimentation in a semi-arid zone reservoir in Australia determined by 137Cs, Acta Geologica Hispanica 35(3-4):329-38. Kashima, K., 2003. The quantitative reconstruction of salinity changes using diatom assemblages in inland saline lakes in the central part of Turkey during the Late Quaternary, Quaternary International 105:13-9. Keith, D.A., 2004. Ocean Shores to Desert Dunes: The Native Vegetation of New South Wales and the ACT., Department of Environment and Conservation, Hurstville. Kershaw, A.P., 1981. Quaternary vegetation and environments. In: Keast, A. (ed.), Ecological Biogeography of Australia, Dr. W. Junk Publishers, Hague, pp. 82-101. Kershaw, A.P., 1995. Environmental change in Greater Australia, Antiquity 69:656-75. Kershaw, A.P., 1997. A modification of the Troels-Smith system of sediment description and portrayal, Quaternary Australasia 15(2):63-8. Kershaw, A.P., Brentherton, S.C. and van der Kaars, S., 2007. A complete pollen record of the last 230 ka from Lynch's Crater, north-eastern Australia, Palaeogeography, Palaeoclimatology, Palaeoecology 251:23-45. Kingsford, R.T., 1999. What future for our free-flowing rivers and their wetlands - the Paroo River?: An introduction. In: Kingsford, R.T. (ed.), A Free-flowing River: The Ecology of the Paroo River, National Parks and Wildlife Service, Sydney, pp. 6-12. Kingsford, R.T., 2000. Protecting rivers in arid regions or pumping them dry?, Hydrobiologia 427:1-11. Kingsford, R.T., Bedward, M. and Porter, J.L., 1994. Waterbirds and Wetlands in Northwestern New South Wales, National Parks and Wildlife Service, Hurstville. Kingsford, R.T., Boulton, A.J. and Puckridge, J.T., 1998. Challenges in managing dryland rivers crossing political boundaries: lessons from Cooper Creek and the Paroo River, central Australia, Aquatic Conservation: Marine and Freshwater Ecosystems 8:361-78. Kingsford, R.T. and Porter, J., 1999. Wetlands and waterbirds of the Paroo and Warrego Rivers. In: Kingsford, R.T. (ed.), A Free-flowing River: The Ecology of the Paroo River, National Parks and Wildlife Service, Sydney, pp. 25-50. Knight, A.W., McTainsh, G.H. and Simpson, R.W., 1995. Sediment loads in an Australian dust storm: implications for present and past dust processes, Catena 24:195-213. Kotwicki, V. and Allan, R., 1998. La Niña de Australia - contemporary and palaeo-hydrology of Lake Eyre, Palaeogeography, Palaeoclimatology, Palaeoecology 144:265-80. Kukla, G. and Gavin, J., 2004. Milankovitch climate reinforcements, Global and Planetary Change 40:27-48. Lamontagne, S., Hicks, W.S., Fitzpatrick, R.W. and Rogers, S., 2004. Survey and Description of Sulphidic Materials in Wetlands of the Lower River Murray Floodplains: Implications for floodplain salinity management, CRC LEME, CSIRO Exploration and Mining, Bentley.

299

Langereis, C.G. and Dekkers, M.J., 1999. Chapter 10: Magnetic cyclostratigraphy: high- resolution dating in and beyond Quaternary and analysis of periodic changes in diagenesis and sedimentary magnetism. In: Maher, B.A. and Thompson, R. (eds), Quaternary Climates, Environments and Magnetism, Cambridge University Press, Cambridge, pp. 352-82. Larocque, I., 2001. How many chironomid head capsules are enough? A statistical approach to determine sample size for palaeoclimatic reconstructions, Palaeogeography, Palaeoclimatology, Palaeoecology 172:133-42. Lees, J., 1999. Chapter 8: Evaluating magnetic parameters for use in source identification, classification and modelling of natural and environmental materials. In: Walden, J., Oldfield, F. and Smith, J.P. (eds), Environmental Magnetism: a practical guide. Technical Guide, No. 6, Quaternary Research Association, London, pp. 113-38. Lentfer, C.J. and Boyd, W.E., 1998. A comparison of three methods for the extraction of phytoliths from sediments, Journal of Archaeological Science 25:1159-83. Lentfer, C.J. and Boyd, W.E., 1999. An assessment of techniques for the deflocculation and removal of clays from sediments used in phytolith analysis, Journal of Archaeological Science 26:31-44. Lentfer, C.J. and Boyd, W.E., 2000. Simultaneous extraction of phytoliths, pollen and spores from sediments, Journal of Archaeological Science 27:363-72. Lentfer, C.J., Cotter, M.M. and Boyd, W.E., 2003. Particle settling times for gravity sedimentation and centrifugation: a practical guide for palynologists, Journal of Archaeological Science 30:149-68. Liu, X.M., Hesse, P., Rolph, T. and Begét, J.E., 1999. Properties of magnetic mineralogy of Alaskan loess: evidence for pedogenesis, Quaternary International 62:93-102. Long, D.T., Fegan, N.E., Lyons, W.B., Hines, M.E., Macumber, P.G. and Giblin, A.M., 1992. Geochemistry of acid brines: Lake Tyrrell, Victoria, Australia, Chemical Geology 96:33- 52. Longmore, M.E., Torgersen, T., O'Leary, B.M. and Luly, J.G., 1986. Cesium-137 redistribution in the sediments of the playa, Lake Tyrrell, northwestern Victoria. Part 1: Stratigraphy and cesium-137 mobility in the upper sediments, Palaeogeography, Palaeoclimatology, Palaeoecology 54:181-95. Loughran, R.J., Pennock, D.J. and Walling, D.E., 2002. Spatial distribution of caesium-137. In: Zapata, F. (ed.), Handbook for the assessmetn of soil erosion and sedimentation using environmental radionuclides, Kluwer Academic Publications, The Netherlands, pp. 97- 110. Lu Shenggao, 2000. Lithological factors affecting magnetic susceptibility of tropical soils, Zhenjiang Province, China, Catena 40:359-73. Luly, J.G., 1993. Holocene palaeoenvironments near Lake Tyrrell, semi-arid northwestern Victoria, Australia, Journal of Biogeography 20:587-98. Luly, J.G., Bowler, J.M. and Head, M.J., 1986. A radiocarbon chronology from the playa Lake Tyrrell, Northwestern Victoria, Australia, Palaeogeography, Palaeoclimatology, Palaeoecology 54:171-80. Lyle, N., 1983. The brown-green colour transition in marine sediments: A marker of the Fe(III) - Fe(II) redox boundary, Limnology and Oceanography 28(5):1026-33. Machel, H.G., 2001. Bacterial and thermochemical sulfate reduction in diagenetic settings - old and new insights, Sedimentary Geology 140:143-75.

300

Macphail, M., 2000. Palynological processing of Holocene-modern estuarine sediments, pp. 1- 5. Macumber, P.G., 1991. Interaction Between Groundwater and Surface Systems in Northern Victoria, Department of Conservation and Environment, Melbourne. Macumber, P.G., 1992. Hydrological processes in the Tyrrell Basin, southeastern Australia, Chemical Geology 96:1-18. Magee, J.W., 1991. Late Quaternary lacustrine, groundwater, aeolian and pedogenic gypsum in the Prungle Lakes, southeastern Australia, Palaeogeography, Palaeoclimatology, Palaeoecology 84:3-42. Magee, J.W., Bowler, J.M., Miller, G.H. and Williams, D.L.G., 1995. Stratigraphy, sedimentology, chronology and palaeohydrology of Quaternary lacustrine deposits at Madigan Gulf, Lake Eyre, South Australia, Palaeogeography, Palaeoclimatology, Palaeoecology 113:3-42. Magee, J.W. and Miller, G.H., 1998. Lake Eyre palaeohydrology from 60 ka to the present: beach ridges and glacial maximum aridity, Palaeogeography, Palaeoclimatology, Palaeoecology 144:307-29. Magee, J.W., Miller, G.H., Spooner, N.A. and Questiaux, D.G., 2004. Continuous 150 k.y. monsoon record from Lake Eyre, Australia: Insolation-forcing implications and unexpected Holocene failure, Geology 32(10):885-8. Maher, B.A., 1986. Characterisation of soils by mineral magnetic measurements, Physics of the Earth and Planetary Interiors 42:76-92. Maher, B.A., 1998. Magnetic properties of modern soils and Quaternary loessic paleosols: paleoclimatic implications, Palaeogeography, Palaeoclimatology, Palaeoecology 137:25- 54. Maher, B.A., Alekseev, A. and Alekseeva, T., 2002. Variation of soil magnetism across the Russian steppe: its significance for use of soil magnetism as palaeorainfall proxy, Quaternary Science Reviews 21:1571-6. Maher, B.A. and Thompson, R., 1999. Chapter 3: Palaeomonsoons I: the magnetic record of palaeoclimate in the terrestrial loess and palaeosol sequences. In: Maher, B.A. and Thompson, R. (eds), Quaternary Climates, Environments and Magnetism, Cambridge University Press, Cambridge, pp. 81-125. Maher, B.A., Thompson, R. and Hounslow, M.W., 1999. Chapter 1: Introduction. In: Maher, B.A. and Thompson, R. (eds), Quaternary Climates, Environments and Magnetism, Cambridge University Press, Cambridge, pp. 1-48. Maroulis, J.C., Nanson, G.C., Price, D.M. and Pietsch, T., 2007. Aeolian-fluvial interaction and climate change: source-bordering dune development over the past ~100ka on Cooper Creek, central Australia, Quaternary Science Reviews 26:386-404. Martin, H.A., 1997. The stratigraphic palynology of bores along the Darling River, downstream from Bourke, New South Wales, Proceedings of the Linnean Society of New South Wales 118:51-67. McBride, J.L. and Nicholls, N., 1983. Seasonal relationships between Australian rainfall and the Southern Oscillation, Monthly Weather Review 111:1998-2004. McGregor, G.B., Marshall, J.C. and Thoms, M.C., 2006. Spacial and temporal variation in algal-assemblage structure in isolated dryland river waterholes, Cooper Creek and Warrego River, Australia, Marine and Freshwater Research 57:453-66. McHenry, R.J. and Ritchie, J.C., 1977. Physical and chemical parameters affecting transport of 137Cs in arid watersheds, Water Resources Research 13(6):923-7.

301

McKeon, G.M., Hall, W.B., Crimp, S.J., Howden, S.M., Stone, R.C. and Jones, D.A., 1998. Climate change in Queensland's grazing lands. I. Approaches and climatic trends, Rangeland Journal 20(2):151-76. McMeekin, I., 1985. Notes for Potters in Australia: Raw Materials and Clay Bodies, New South Wales University Press, Kensington. Mooney, S.D. and Black, M., 2003. A simple and fast method for calculating the area of macroscopic charcoal isolated from sediments, Quaternary Australasia 21(1):18-21. Mooney, S.D., Radford, K.L. and Hancock, G., 2001. Clues to the 'burning question': Pre- European fire in the Sydney coastal region from sedimentary charcoal and palynology, Ecological Management and Restoration 2(3):203-12. Mooney, S.D., Webb, M. and Attenbrow, V., 2007. A comparison of charcoal and archaeological information to address the influences on Holocene fire activity in the Sydney Basin, Australian Geographer 38(2):177-94. Morgan, G. and Terrey, J., 1992. Nature Conservation in Western New South Wales, National Parks Association of NSW, Sydney. Mulholland, S.C., 1989. Phytolith shape frequencies in North Dakota grasses: a comparison to general patterns, Journal of Archaeological Science 16:489-511. Muller, J., Wüst, R.A.J., Weiss, D. and Hu, Y., 2006. Geochemical and stratigraphic evidence of environmental change at Lynch's Crater, Queensland, Australia, Global and Planetary Change 53:269-77. Munsterman, D. and Kerstholt, S., 1996. Sodium polytungstate, a new non-toxic alternative to bromoform in heavy liquid separation, Review of Palaeobotany and Palynology 91:417- 22. Murray, M.R., 2002. Is laser particle size determination possible for carbonate-rich lake sediments?, Journal of Paleolimnology 27:173-83. Myrow, P.M., 1990. A new graph for understanding colors of mudrocks and shales, Journal of Geological Education 38:16-20. Nanson, G.C., Callen, R.A. and Price, D.M., 1998. Hydroclimatic interpretation of Quaternary shorelines on South Australian playas, Palaeogeography, Palaeoclimatology, Palaeoecology 144:281-305. Nanson, G.C., Chen, X.Y. and Price, D.M., 1995. Aeolian and fluvial evidence of changing climate and wind patterns during the past 100 ka in the western Simpson Desert, Australia., Palaeogeography, Palaeoclimatology, Palaeoecology 113:87-102. Nanson, G.C., Rust, B.R. and Taylor, G., 1986. Coexistent mud braids and anastomosing channels in an arid-zone river: Cooper Creek, central Australia, Geology 14:175-8. Neumann, B.S., 1977. Chapter 12. Thermal techniques. In: Zussman, J. (ed.), Physical Methods in Determinative Mineralogy, Academic Press, London, pp. 605-62. O'Geen, A.T. and Busacca, A.J., 2001. Faunal burrows as indicators of paleo-vegetation in eastern Washington, USA, Palaeogeography, Palaeoclimatology, Palaeoecology 169:23- 37. Oldfield, F., 1999. Chapter 7: The rock magnetic identification of magnetic mineral and magnetic grain size assemblages. In: Walden, J., Oldfield, F. and Smith, J.P. (eds), Environmental Magnetism: a practical guide. Technical Guide, No. 6, Quaternary Research Association, London, pp. 98-112. Oyama, M. and Takehara, H., 1993. Revised Standard Soil Color Charts,

302

Page, K., Nanson, G. and Price, D., 1996. Chronology of Murrumbidgee River palaeochannels on the Riverine Plain, southeastern Australia, Journal of Quaternary Science 11(4):311- 26. Page, M., Beeton, R.J.S. and Mott, J.J., 2000. Grass response to shrub removal in two semi-arid vegetation communities, Rangeland Journal 22(2):220-34. Page, M.J., 1995. Where does Currawinya National Park fit in the conceptual models of change? In: Page, M.J. and Beutel, T.S. (eds), Ecological Research and Management in the Mulgalands - Conference Proceedings, University of Queensland Gatton College, Australia, pp. 37-42. Parr, J.F., 2002. A comparison of heavy liquid flotation and microwave digestion techniques for the extraction of fossil phytoliths from sediments, Review of Palaeobotany and Palynology. Pearson, S., Gayler, L., Hartig, K. and Timms, B., 2003. Ecosystem health in the Paroo: an arid frontier? In: Albrecht, G. (ed.), Proceedings of the Airs Waters Places Transdisciplinary Conference on Ecosystem Health in Australia, University of Newcastle, Newcastle, pp. 252-64. Pennak, R.W., 1978. Fresh-Water Invertebrates of the United States, John Wiley & Sons, New York. Phartiyal, B., Appel, E., Blaha, U., Hoffmann, V. and Kotlia, B.S., 2003. Palaeoclimatic significance of magnetic properties from Late Quaternary lacustrine sediments at Pithoragarh, Kumaun Lesser Himalaya, India, Quaternary International 108:51-62. Pickard, J., 1994. Post-European changes in creeks of semi-arid rangelands, 'Polpah Station', New South Wales. In: Millington, A.C. and Pye, K. (eds), Environmental Changes in Drylands: Biogeographical and Geomorphological Perspectives, John Wiley and Sons, Chichester, pp. 271-83. Pickard, J., 1997. Oral history of the floods on the lower Paroo River. In: Abstracts for The Paroo River Scientific Workshop, Hungerford, 1-2. Pickett, J., 1983. An annotated bibliography and review of Australian fossil sponges. In: Roberts, J. and Jell, P.A. (eds), Memoir 1 of the Association of Australasian Palaeontologists, Association of Australasian Palaeontologists, Sydney, pp. Piperno, D.R., 2006. Phytoliths. A Comprehensive Guide for Archaeologists and Paleoecologists, AltaMira Press, Lanham. Pontual, S., Merry, N. and Gamson, P., 1997. Spectral Interpretation Field Manual, AusSpec International Pty Ltd, Power, M.J., Whitlock, C., Bartlein, P. and Stevens, L.R., 2006. Fire and vegetation history during the last 3800 years in northwestern Montana, Geomorphology 75:420-36. Pye, K., 1983. Early post-depositional modification of aeolian dune sands. In: Brookfield, M.E. and Ahlbrandt, T.S. (eds), Eolian Sediments and Processes, Elsevier, Amsterdam, pp. 197-221. Queensland Parks and Wildlife Service, 1999. Currawinya National Park: Management Plan, Queensland Parks and Wildlife Service, Brisbane. Queensland Parks and Wildlife Service, 2007. Lake Bindegolly National Park, http://www.epa.qld.gov.au/projects/park/index.cgi?noback=1&parkid=101, Rayment, G.E. and Higginson, F.R., 1992. Australian Laboratory Handbook of Soil and Water Chemical Methods, Inkata Press, Melbourne.

303

Rayment, G.E., Walker, B. and Keerati-Kasikorn, P., 1983. Sulfur in the agriculture of Northern Australia. In: Blair, G.J. and Till, A.R. (eds), Sulfur in S.E. Asian and S. Pacific Agriculture, UNE, Indonesia, pp. 228-50. Reid, M.A., Tibby, J.C., Penny, D. and Gell, P.A., 1995. The use of diatoms to assess past and present water quality, Australian Journal of Ecology 20:57-64. Reitsma, T.J., 1969. Size modification of recent pollen grains under different treatments, Review of Palaeobotany and Palynology 9:175-202. Richardson, J.L. and Daniels, R.B., 1993. Stratigraphic and hydraulic influences on soil color development. In: Bigham, J.M. and Ciolkosz, E.J. (eds), Soil Color, Soil Science Society of America, Madison, pp. 109-25. Ritchie, J.C. and McHenry, R.J., 1990. Application of Radioactive Fallout Cesium-137 for Measuring Soil Erosion and Sediment Accumulation Rates and Patterns: A Review, Journal of Environmental Quality 19:215-33. Robins, R., 1999. Clocks for rocks: An archaeological perspective on the Currawinya Lakes. In: Kingsford, R.T. (ed.), A Free-flowing River: The Ecology of the Paroo River, National Parks and Wildlife Service, Sydney, pp. 150-78. Robins, R. and Connolly, M., 2001. Post-European Development and its Implications for Aboriginal Cultural Heritage Management on Currawinya National Park. In: Page, M., Evenson, C. and Whittington, A. (eds), Research Need for Managing a Changed Landscape in the Hungerford/Eulo Region, School of Natural and Rural Systems Management, University of Queensland, Gatton, pp. 15-8. Rowell, D.L., 1994. Soil Science: Methods and Applications, Longman Scientific & Technical, Harlow. Rummery, T.A., Bloemendal, J., Dearing, J. and Oldfield, F., 1979. The persistance of fire- induced magnetic oxides in soils and lake sediments, Annales de Géophysique 35(2):103- 7. Rust, B.R. and Nanson, G.C., 1986. Contemporary and palaeo channel patterns and the late Quaternary stratigraphy of Cooper Creek, Southwest Queensland, Australia, Earth Surface Processes and Landforms 11:581-90. Ryves, D.B., Juggins, S., Fritz, S.C. and Battarbee, R.W., 2001. Experimental diatom dissolution and the quantification of microfossil preservation in sediments, Palaeogeography, Palaeoclimatology, Palaeoecology 172:99-113. Sahukar, R., Gallery, C., Smart, J. and Mitchell, P., 2003. The Bioregions of New South Wales: Their Biodiversity, Conservation and History, National Parks and Wildlife Service (NSW), Sydney. Schulze, D.G., Nagel, J.L., Van Scoyoc, G.E., Henderson, T.L., Baumgardner, M.F. and Stott, D.E., 1993. Significance of organic matter in determining soil colors. In: Bigham, J.M. and Ciolkosz, E.J. (eds), Soil Color, Soil Science Society of America, Madison, pp. 71- 90. Schwandes, L.P. and Collins, M.E., 1994. Distribution and significance of freshwater sponge spicules in selected Florida soils, Transactions of the American Microscopical Society 113(3):242-57. Schwertmann, U., 1993. Relations between iron oxides, soil color, and soil formation. In: Bigham, J.M. and Ciolkosz, E.J. (eds), Soil Color, Soil Science Society of America, Madison, pp. 51-69. Scott, L., 2002. Grassland development under glacial and interglacial conditions in southern Africa: review of pollen, phytolith and isotope evidence, Palaeogeography, Palaeoclimatology, Palaeoecology 177:47-57.

304

Shimeld, P., Hopf, F. and Pearson, S., 2000. Pollen image management: the Newcastle digital collection initiative, Quaternary Australasia 18:13-5. Singer, M.J., Verosub, K.L., Fine, P. and TenPas, J., 1996. A conceptual model for the enhancement of magnetic susceptibility in soils, Quaternary International 34-36:243-8. Singh, G. and Geissler, E.A., 1985. Late Cainozoic history of vegetation, fire, lake levels and climate, at Lake George, New South Wales, Australia, Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 311(1151):379-447. Singh, G. and Luly, J., 1991. Changes in vegetation and seasonal climate since the last full glacial at Lake Frome, South Australia, Palaeogeography, Palaeoclimatology, Palaeoecology 84(75-86). Smith, J., 1999. Chapter 2: An introduction to the magnetic properties of natural materials. In: Walden, J., Oldfield, F. and Smith, J.P. (eds), Environmental Magnetism: a practical guide. Technical Guide, No.6, Quaternary Research Association, London, pp. 5-25. Smith, S.J., 1998. Processing pollen samples from archaeological sites in the southwest United States: an example of differential recovery from two heavy liquid gravity separation procedures. In: Bryant, V.M. and Wrenn, J.H. (eds), New Developments in Palynomorph Sampling, Extraction, and Analysis, American Association of Stratigraphic Palynologists Foundation, 29-34. Sprigg, R.C., 1982. Alternating wind cycles of the Quaternary era and their influences on aeolian sedimentation in and around the dune deserts of south-eastern Australia. In: Wasson, R.J. (ed.), Quaternary dust mantles of China, New Zealand and Australia, Australian National University, Canberra, pp. 211-40. Stern, H., de Hoedt, G. and Ernst, J., 2000. Objective classification of Australian climates, Australian Metorological Magazine 49:87-96. Stone, T., 2006. Last glacial cycle hydrological change at Lake Tyrrell, southeast Australia, Quaternary Research 66:176-81. Taylor, R.M., 1982. Colour in soils and sediments - a review. In: van Olphen, H. and Veniale, F. (eds), International Clay Conference 1981, Elsevier, Amsterdam, pp. 749-61. Taylor, S., 1991. Dryland Salinity - Introductory Extension Notes., Department of Conservation and Land Management (incorporating Soil Conservation Service of NSW), Teller, J.T., Bowler, J.M. and Macumber, P.G., 1982. Modern sedimentation and hydrology in Lake Tyrrell, Victoria, Journal of the Geological Society of Australia 29:159-75. Thompson, R. and Morton, D.J., 1979. Magnetic susceptibility and particle-size distribution in recent sediments of the Loch Lamond drainage basin, Scotland, Journal of Sedimentary Petrology 49(3):801-12. Thompson, R. and Oldfield, F., 1986. Environmental magnetism, Allen & Unwin, London. Thorn, V.C., 2004. Phytolith evidence for C4-dominated grassland since the early Holocene at Long Pocket, northeast Queensland, Australia, Quaternary Research 61:168-80. Thouveny, N., de Beaulieu, J.-L., Bonifay, E., Creer, K.M., Guiot, J., Icole, M., Johnsen, S.J., Jouzel, J., Reille, M., Williams, T. and Williamson, D., 1994. Climate variations in Europe over past 140 kyr deduced from rock magnetism, Nature 371:503-6. Timms, B.V., 1992. Lake Geomorphology, Gleneagles Publishing, Adelaide. Timms, B.V., 1993. Saline lakes of the Paroo, inland New South Wales, Australia, Hydrobiologia 267:269-89. Timms, B.V., 1997a. A comparison between saline and freshwater wetlands on Bloodwood Station, the Paroo, Australia, with special reference to their use by waterbirds, International Journal of Salt Lake Research 5:287-313.

305

Timms, B.V., 1997b. A Study of the Wetlands of Currawinya National Park. A report to the Queensland Department of Environment, Toowoomba, Queensland., University of Newcastle, Newcastle. Timms, B.V., 1998a. Further studies on the saline lakes of the eastern Paroo, inland New South Wales, Australia, Hydrobiologia 381:31-42. Timms, B.V., 1998b. Spatial and temporal variation in inundation of wetlands in the Paroo catchment of the Murray-Darling. In: Williams, W.D. (ed.), Wetlands in a Dry Land: Understanding for Management, Environment Australia, Biodiversity Group, Canberra, pp. 123-9. Timms, B.V., 1998c. A study of Lake Wyara, an episodically filled saline lake in southwest Queensland, Australia, International Journal of Salt Lake Research 7:113-32. Timms, B.V., 1999. Local runoff, Paroo floods and water extraction impacts on the wetlands of Currawinya National Park. In: Kingsford, R.T. (ed.), A Free-flowing River: The Ecology of the Paroo River, National Parks and Wildlife Service, Sydney, pp. 51-66. Timms, B.V., 2006. The geomorphology and hydrology of saline lakes of the middle Paroo, arid-zone Australia, Proceedings of the Linnean Society of New South Wales 127:157-74. Timms, B.V., 2007. The biology of saline lakes of central and eastern inland of Australia: a review with special reference to their biogeographical affinities, Hydrobiologia 576:27- 37. Timms, B.V., in press. The Rockwell-Wombah Lakes, Paroo, Eastern Australia: a ten year window on five naturally salinised lakes, Proceedings of the Linnean Society of New South Wales. Tolonen, K., 1986. Charred particle analysis. In: Berglund, B.E. (ed.), Handbook of Holocene Palaeoecology and Palaeohydrology, John Wiley & Sons, Chichester, pp. 485-97. Torgersen, T., DeDeckker, P., Chivas, A.R. and Bowler, J.M., 1986. Salt lakes: a discussion of processes influencing palaeoenvironmental interpretation and recommendations for future study, Palaeogeography, Palaeoclimatology, Palaeoecology 54:7-19. Troels-Smith, J., 1955. Characterisation of Unconsolidated Sediments. In: Geological Survey of Denmark. IV Series., C.A. Reitzels Forlag Axel Sandal, Kobenhavn, pp. Turney, C.S.M., Kershaw, A.P., James, S., Branch, N., Cowley, J., Fifield, L.K., Jacobsen, G. and Moss, P., 2006. Geochemical changes recorded in Lynch's Crater, Northeastern Australia, over the past 50 ka, Palaeogeography, Palaeoclimatology, Palaeoecology 233:187-203. Twidale, C.R. and Wopfner, H., 1990. Dune fields. In: Tyler, M.J., Twidale, C.R., Davies, M. and Wells, C.B. (eds), Natural History of the North East Deserts, Royal Society of South Australia Inc., Adelaide, pp. 45-60. Ullman, W.J. and McLeod, L.C., 1986. The late-Quaternary salinity record of Lake Frome, South Australia: evidence from Na+ in stratigraphically-preserved gypsum, Palaeogeography, Palaeoclimatology, Palaeoecology 54:153-69. Velasco, F., Alvaro, A., Suarez, S., Herrero, J. and Yusta, I., 2005. Mapping Fe-bearing hydrated sulphate minerals with short wave infrared (SWIR) spectaral analysis at San Miguel mine environment, Iberian Pyrite Belt (SW Spain), Journal of Geochemical Exploration 87:45-72. Velde, B., 1985. Clay Minerals. A Physico-Chemical Explanation of their Occurence, Elsevier, Amsterdam.

306

Velde, B., 1992. Introduction to Clay Minerals: Chemistry, Origins, Uses and Environmental Significance, Chapman and Hall, London. Walden, J., 1999. Chapter 3: Sample collection and preparation. In: Walden, J., Oldfield, F. and Smith, J.P. (eds), Environmental Magnetism: a practical guide. Technical Guide, No. 6., Quaternary Research Association, London, pp. 26-34. Walden, J., White, K.H., Kilcoyne, S.H. and Bentley, P.M., 2000. Namib dune sediments using high field remnance measurements (9T) and Mössbauer analysis, Journal of Quaternary Science 15(2):185-95. Wallis, L., 2003. An overview of leaf phytolith production patterns in selected northwest Australian flora, Review of Palaeobotany and Palynology 125:201-48. Warren, J.K., 1982. The hydrological setting, occurence and significance of gypsum in late Quaternary salt lakes in South Australia, Sedimentology 29:609-37. Warren, J.K., 1986. Shallow-water evaporitic environments and their source rock potential, Journal of Sedimentary Petrology 56(3):442-54. Wasson, R.J., 1976. Holocene aeolian landforms in the Belarabon area, S.W. of Cobar, N.S.W., Journal of Proceedings, Royal Society of New South Wales 109:91-101. Wasson, R.J., 1979. Sedimentation history of the Mundi Mundi alluvial fans, western New South Wales., Sedimentary Geology 22:21-51. Wasson, R.J., 1983. Dune sediment types, sand colour, sediment provenance and hydrology in the Strzelecki-Simpson dunefield, Australia. In: Brookfield, M.E. and Ahlbrandt, T.S. (eds), Eolian Sediments and Processes, Elsevier, Amsterdam, pp. 165-95. Wasson, R.J., Fitchett, K., Mackey, B. and Hyde, R., 1988. Large-scale patterns of dune type, spacing and orientation in the Australian continental dunefield, Australian Geographer 19(1):89-104. Wasson, R.J., Olive, L.J. and Rosewell, C.J., 1996. Rates of erosion and sediment transport in Australia. In: IAHS Publication 236, 139-48. Wells, N.A., 2002. Quantitative evolution of colour measurements: I. Triaxial stereoscopic scatter plots, Sedimentary Geology 151:1-15. Wells, N.A., Konowal, M. and Sundback, S.A., 2002. Quantitative evaluation of colour measurements. II. Analysis of Munsell colour values from the Colton and Green River Formations (Eocene, central Utah), Sedimentary Geology 151:17-44. Westbrooke, M.E. and Florentine, S.K., 2005. Rain-driven episodic flood events: are they a major factor in moulding New South Wales arid land vegetation patterns?, Australian Geographer 36(2):171-81. Westbrooke, M.E., Florentine, S.K. and Milberg, P., 2005. Arid land vegetation dynamics after a rare flooding event: influence of fire and grazing, Journal of Arid Environments 61:249- 60. Wilding, L.P. and Drees, L.R., 1971. Biogenic opal in Ohio soils, Soil Science Society of America Proceedings 35:1004-10. Williams, M.A.J., 1994a. Cenozoic climatic changes in deserts: a synthesis. In: Abrahams, A.D. and Parsons, A.J. (eds), Geomorphology of Desert Environments, Chapman & Hall, London, pp. Williams, M.A.J., 1994b. Some implications of past climatic changes in Australia, Transactions of the Royal Society of South Australia 118(1):17-25. Williams, M.A.J., De Deckker, P., Adamson, D.A. and Talbot, M.R., 1991. Episodic fluviatile, lacustrine and aeolian sedimentation in a late Quaternary desert margin system, central western New South Wales. In: Williams, M.A.J., De Deckker, P. and Kershaw, A.P.

307

(eds), The Cainozoic in Australia: A re-appraisal of the evidence, Geological Society of Australia, Sydney, pp. 258-87. Williams, M.A.J., Dunkerley, D.L., De Deckker, P., Kershaw, A.P. and Stokes, T.J., 1993. Quaternary Environments, Arnold, London. Wilson, B.A., 1999. Mulga Lands. In: Sattler, P. and Williams, R. (eds), The Conservation Status of Queensland's Bioregional Ecosystems, Environmental Protection Agency, Queensland Government, Brisbane, pp. 6/1-35. Witt, G.B. and Beeton, R.J.S., 1995. The regional implications of 'naturalness' in protected area management: a case study from the Queensland Mulgalands. In: Page, M.J. and Beutel, T.S. (eds), Ecological Research and Management in the Mulgalands - Conference Proceedings, University of Queensland Gatton College, Australia, pp. 177-82. Witt, G.B., Luly, J. and Fairfax, R.J., 2006. How the west was once: vegetation change in southwest Queensland from 1930-1995, Journal of Biogeography 33:1585-96. Witt, G.B., Moll, E.J. and Beeton, R.J.S., 1996. Sheep faeces under shearing sheds: a documentary of vegetation change using stable isotope analysis, Rangeland Journal 19(1):109-15. Ye, S., 1998. Thermal Analysis Curves of Plant Materials. In: Hatakeyama, T. and Liu, Z. (eds), Handbook of Thermal Analysis, John Wiley and Sons, Chichester, pp. 239-46. Young, W., Brandis, K. and Kingsford, R., 2006. Modelling monthly streamflows in two Australian dryland rivers: Matching model complexity to spatial scale and data availability, Journal of Hydrology 331:242-56. Young, W.J., 1999. Hydrologic descriptions of semi-arid rivers: An ecological perspective. In: Kingsford, R.T. (ed.), A Free-flowing River: The Ecology of the Paroo River, National Parks and Wildlife Service, Sydney, pp. 77-96.

308

Appendix 1

Indicative native plant species of four vegetation groups dominant in central part of the Paroo/Warrego Region, adapted from Keith (2004) and Kingsford and Porter (1999)

Group 1 Mulga shrublands and woodlands

Class Trees and shrubs Herbs Grasses and grass-like plants North-west Acacia aneura (mulga) Calotis lappulacea Aristida contorta (yellow burr-daisy) (bunched kerosene plain A. excelsa (ironwood) A. homalophylla (yarran) Sclerolaena diacantha grass) shrublands A. victoriae (elegant wattle) (grey copperburr) A. jerichoensis subsp S. uniflora subspinulifera (Jericho Apophyllum anomalum wiregrass) (warrior bush) Dichanthium sericeum Atalaya hemiglauca (Queensland bluegrass) (whitewood) Digitaria brownii Atriplex stipitata (malee (cotton panic grass) saltbush) Eragrostis dielsii Callitris glaucophylla (white (mulka) cypress pine) E. eriopoda Dodonea viscosa (hopbush) (woollybutt) Eremophila mitchellii (budda) E. leptocarpa (drooping E. sturtii (turpentine bush) lovegrass) Eucalyptus populnea (poplar box) Flindersia maculosa (leopardwood) Geijera parviflora (wilga) Grevillea striata (beefwood) Hakea leucoptera (needlewood) H. tephrosperma (hooked needlewood) Myoporum platycarpum (sugarwood) Olearia pimelioides (showy daisy-bush) Pittosporum phylliraeoides (butterbush) Senna artemisioides (silver cassia) Gibber Tall shrubs: Atriplex angulata (fan Chloris truncata

309

transition Acacia aneura (mulga) saltbush) (windmill grass) shrublands and A. cambagei (gidgee) Osteocarpum Dichanthium sericeum A. harpophylla (brigalow) acropterum (water (Queensland bluegrass) woodlands Alectryon oleifolius (western weed) Digitaria spp. rosewood) Sclerolaena lanicuspis Enteropogon acicularis Apophyllum anomalum (woolly copperburr) (curly windmill grass) (warrior bush) S. limbata (pearl Eragrostis parviflora Atalaya hemiglauca copperburr) (weeping lovegrass) (whitewood) E. setifolia (neverfail) Casuarina pauper (belah) Eulalia aurea (slipy Eremophila mitchellii (budda) browntop) Flindersia maculosa Leptochloa digitata (leopardwood) (umbrella canegrass) Grevillea striata (beefwood) Sporobolus caroli (fairy Small understorey shrubs: grass) Atriplex nummularia (old man Paspalidium jubiflorum saltbush) (Warrego grass) Enchylaena tomentosa (ruby saltbush) Maireana pyramidata (black bluebush) Rhagodia spinescens (thorny saltbush) Stony desert Abutilon leucopetalum (lantern Atriplex angulata (fan Aristida contorta bush) saltbush) (bunched kerosene mulga Acacia aneura (mulga) A. lindleyi (baldoo) grass) shrublands A. brachystachya (umbrella Calotis hispidula Astrebla lappacea mulga) (bogan flea) (curly Mitchell grass) A. ramulosa (horse mulga) Osteocarpum Austrostipa nitida A. stovardii (bastard mulga) acropterum (water A. trichophylla A. tetragonophylla (dead weed) A. variabilis finish) Ptilotus spp. Enneapogon avenaceus Apophyllum anomalum Sclerolaena lanicuspis (bottle washers) (warrior bush) (woolly copperburr) Thyridolepis Atalaya hemiglauca S. obliquicuspis mitchelliana (mulga (whitewood) (limestone copperburr) Mitchell grass) Dodonea viscosa (hopbush) Sida corrugata Enchylaena tomentosa (ruby S. phaeotricha (hill saltbush) sida) Eremophila duttonii (harlequin Trichodesma fuchsia-bush) zeylanicum (cattle bush) E. latrobei (crimson turkey- Zygophyllum bush) apiculatum (gallweed) E. sturtii (turpentine bush) Grevillea striata (beefwood) Hibiscus sturtii (hill hibiscus) Maireana pyramidata (black bluebush) M.sedifolia (pearl bluebush) M. tomentosa Nicotiana glauca (tree tobacco)

310

Pittosporum phylliraeoides (butterbush) Prostanthera striatiflora (streaked mint-bush) Rhagodia spinescens (thorny saltbush) Senna artemisioides (silver cassia) Solanum ellipticum (velvet potato bush) S. sturtianum (Thargomindah nightshade) Sand plain Acacia aneura (mulga) Calandrinia baloensis Aristida contorta (broad-leaf parakeelya) (bunched kerosene mulga A. cana (cabbage-tree wattle) A. ligulata (sandhill wattle) C. eremaea (small grass) shrublands A. loderi purslane) A. jerichoensis subsp subspinulifera (Jericho A. victoriae (elegant wattle) Calotis erinacea (tangled burr-daisy) wiregrass) Alectryon oleifolius (western Eragrostis dielsii rosewood) Dissocarpus paradoxus (cannonball burr) (mulka) Atalaya hemiglauca E. eriopoda (whitewood) Myriocephalus stuartii (poached eggs) (woollybutt) Casuarina pauper (belah) Ptilotus polystachyus Eragrostis parviflora Dodonea viscosa (hopbush) (long-tails) (weeping lovegrass) Eremophila longifolia Rhodanthe floribunda Enneapogon avenaceus (berrigan) (common white sunray) (bottle washers) E. sturtii (turpentine bush) Sclerolaena bicornis Eriachne aristidea Flindersia maculosa S. diacantha (threeawn wanderrie (leopardwood) grass) Trachymene glaucifolia Grevillea stenobotrya (sandhill (wild parsnip) Triraphis mollis (purple spider-flower) needlegrass) Hakea leucoptera (needlewood) H. tephrosperma (hooked needlewood) Maireana pyramidata (black bluebush) Rhagodia spinescens (thorny saltbush) Senna artemisioides (silver cassia)

Group 2 Shrubby and grassy semi-arid woodlands

Class Trees and shrubs Herbs Grasses and grass-like plants Grassy north- Trees: Asperula gemella (twin- Astrebla lappacea leaved bedstraw) (curly Mitchell grass) west floodplain Casuarina cristata (belah) C. pauper (black oak) Boerhavia dominii (tarvine) Chloris truncata woodlands Eucalyptus coolabah Calotis hispidula (bogan (windmill grass) Cyperus concinnus 311

(coolibah) flea) (trim sedge) E. largiflorens (black box) Centipeda minima Eleocharis pallens (pale E. camaldulensis (river red (spreeding sneezeweed) spike-rush) gum) Chamaesyce drummondii E. plana (ribbed spike- E. populnea (poplar box) (caustic weed) rush) E. ochrophloia (yapunyah) Einadia nutans (climbing Enteropogon acicularis Shrubs: saltbush) (curly windmill grass) Acacia stenophylla (river Eryngium plantagineum Eragrostis parviflora cooba) (long eryngium) (weeping lovegrass) A. victoriae (elegant wattle) Euchiton sphaericus E. setifolia (neverfail) (Japanese cudweed) Alectryon oleifolius Iseilema (western rosewood) Goodenia fascicularis membranaceum (small (silky goodenia) Flinders grass) Atalaya hemiglauca (whitewood) Haloragis glauca (grey Juncus aridicola raspwort) (tussock rush) Atriplex leptocarpa (slender-fruit saltbush) Ixolaena brevicompta Lachnagrostis filiformis (plains plover-daisy) (blown grass) A. nummularia (old man saltbush) Lepidium hyssopifolium Paspalidium aversum (bent summer grass) Chenopodium nitrariaceum Mimulus gracilis (slender (nitre goosefoot) monkey flower) P. distans Dodonea viscosa (hopbush) Osteocarpum acropterum P. jubiflorum (Warrego (water weed) grass) Enchylaena tomentosa (ruby saltbush) Oxalis chnoodes Sporobolus caroli (fairy grass) Eremophila bignoniiflora Plantago cunninghamii (eurah) (sago weed) Muehlenbeckia florulenta Pseudognaphalium (lignum) luteoalbum (jersey cudweed) Rhagodia spinescens (thorny saltbush) Scleroblitum atriplicinum (purple goosefoot) Sclerolaena birchii (galvanised burr) S. muricata (black roly- poly) Senecio quadridentatus (cotton fireweed) Solanum esuriale (quena) Tribulus terrestris (caltrop) Trigonella suavissima (Cooper’s clover) Wahlenbergia sticta Western Trees: Actinobole uliginosum Aristida jerichoensis (flannel cudweed) subsp subspinulifera peneplain Eucalyptus populnea (poplar box) Alternanthera denticulata (Jericho wiregrass) woodlands E. intertexta (gum (lesser joyweed) Astrebla lappacea coolibah) – on drier/stonier Atriplex spinibractea (curly Mitchell grass) soils Brachyscome lineariloba Austrodanthomia Callitris glaucophylla (hard-headed daisy) caespitosa (ringed (white cypress pine) – on Calandrinia eremaea wallaby grass) light-textured soils (small purslane) A. setacea (small- Shrubs: Calotis cuneifolia (purple flowered wallaby grass) Acacia buxifolia (box- burr-daisy) Austrostipa aristiglumis

312

leaved wattle) C. hispidula (bogan flea) (plains grass) A. calamifolia (wallowa) Centipeda cunninghamii A. scabra (rough A. excelsa (ironwood) (common sneezeweed) speargrass) A. homalophylla (yarran) Cheilanthes sieberi (poison A. variabilis A. rigens (needle wattle) rock fern) Chloris truncata Alectryon oleifolius Chenopodium desertorum (windmill grass) (western rosewood) (desert goosefoot) Dichanthium sericeum Allocasuarina luehmannii Chrysocephalum (Queensland bluegrass) (bulloak) apiculatum (common Elymus scaber everlasting) Atalaya hemiglauca (common wheatgrass) (whitewood) Einadia nutans (climbing Enteropogon acicularis saltbush) Atriplex semibaccata (curly windmill grass) (creeping saltbush) Erodium crinitum (blue Eragrostis lacunaria storkbill) Capparis mitchellii (wild (purple lovegrass) orange) Goodenia cycloptera E. parviflora (weeping (serrated goodenia) Casuarina pauper (belah) lovegrass) G. glauca (pale goodenia) Dodonea viscosa (hopbush) Notodanthonia G. pinnatifida (scrambled semiannularis Eremophila deserti eggs) (Tasmanian wallaby E. longifolia (berrigan) Hyalosperma semisterile grass) E. mitchellii (budda) Isoetopsis graminifolia Sporobolus caroli (fairy Flindersia maculosa (grass cushion) grass) (leopardwood) Minuria leptophylla Thyridolepis Geijera parviflora (wilga) (minnie daisy) mitchelliana (mulga Hakea leucoptera Oxalis chnoodes Mitchell grass) (needlewood) Rumex brownii (swamp Maireana enhylaenoides dock) (wingless fissure-bush) Sclerolaena bicornis M.humillima (goathead burr) M. villosa (silky bluebush) S. uniflora Myoporum platycarpum Sida corrugata (corrugated (sugarwood) sida) Philotheca difformis (small- Solanum esuriale (quena) leved waxflower) Triptilodiscus pygmaeus Senna artemisioides (silver Wahlenbergia gracilis cassia) (Australian bluebell) Shrubby semi- Trees: Cheilanthes sieberi (poison Austrostipa nitida rock fern) arid sand plain Alectryon oleifolius Chloris truncata (western rosewood) Dissocarpus paradoxus (windmill grass) woodlands Casuarina pauper (belah) (cannonball burr) Enneapogon avenaceus Tall shrubs: Omphalolappula concave (bottle washers) Apophyllum anomalum (burr stickseed) Eragrostis dielsii (warrior bush) Pycnosorus pleiocephalus (mulka) Beyeria opaca (smooth (soft billy-buttons) Sporobolus caroli (fairy wallaby bush) Sclerolaena diacantha grass) Eremophila mitchellii (grey copperburr) (budda) S. obliquicuspis (limestone E. sturtii (turpentine bush) copperburr) Exocarpus aphyllus Tetragonia tetragonioides (leafless ballart) Zygophyllum ammophilum Geijera parviflora (wilga) (sand twinleaf)

313

Myoporum platycarpum (sugarwood) Low shrubs: Atriplex stipitata (malee saltbush) Chenopodium curvispicatum Enchylaena tomentosa (ruby saltbush) Maireana pyramidata (black bluebush) M.triptera (three-winged bluebush) Rhagodia spinescens (thorny saltbush)

Group 3 Arid riverine chenopod shrubland

Shrubs Herbs Grasses and grass- like plants Arid Atriplex nummularia Atriplex lindleyi (baldoo) Austrodanthomia (old man saltbush) caespitosa (ringed chenopod A. pseudocampanulata (mealy A vesicaria (bladder saltbush) wallaby grass) shrubland saltbush) Brachyscome lineariloba (hard-headed Chloris truncata Enchylaena daisy) (windmill grass) tomentosa (ruby Calocephlus sonderi (pale beauty Lachnagrostis saltbush) heads) filiformis (blown Malacocera tricornis Daucus glochidiatus (native carrot) grass) Rhagodia spinescens Disphyma crassifolium (round-leaved Sporobolus caroli (thorny saltbush) pigface) (fairy grass) Sclerostegia tenuis Dissocarpus biflorus (twin-horned (slender glasswort) copperburr) Overgrazed/dieback Frankenia connata (clustered sea- areas: heath) Maireana aphylla Ixolaena tomentosa (woolly ixolaena) (cotton bush) Leptorhynchos panaetioidesm (woolly Nitraria billardieri buttons) (nitre bush) Minuria cunninghamii (bush minuria) Sclerolaena muricata Osteocarpum acropterum (water weed) (black roly-poly) Plantago cunninghamii (sago weed) S. tricuspis (giant redburr) Podolepis muelleri (small copper-wire daisy) Rhodanthe corymbiflora (small white sunray) Sclerolaena bicornis (goathead burr) S. brachyptera (short-winged copperburr) S. intricata (tangled poverty bush) S. stelligera (star copperburr) Stda trichopoda (high sida) 314

Vittadinia cuneata (fuzzweed)

Group 4 Wetland systems

Trees and shrubs Herbs Grasses and grass-like plants Salt lakes Shrubs: Charophytes: Cyperus bulbosus (nalgoo) Acacia ligulata (sandhill Chara globularis C. gymnocaulos (spiny wattle) Lamprothamnium sedge) Atriplex spongiosa (pop papulosum C. squarrosus (flat sedge) bush) Nitella lhotzkii Eragrostis dielsii (mulka) Eremophila bignoniiflora Aquatic herbs: (eurah) Lepilaena bilocularis Gunniopsis quadrifida Ruppia spp. (sea tassel) (pig face) Terrestrial herbs: Halosarcia pergranulata (samphire) Epaltes australis (nut head) Mairena pyramidata Frankenia spp. (black bluebush) Lawrencia glomerata Myoporum montanum (golden spike) (boobialla) Lotus cruentus (trefoil) Sclerolaena intricata Mimulus repens (monkey (tangled poverty bush) flower) Senecio gregorii (freshy groundsel) Freshwater Trees: Aquatic herbs: Cyperus gymnocaulos (spiny sedge) lakes Acacia stenophylla (river Myriophyllum verrucosum cooba) (red water-milfoil) Diplachne fusca (beetle Eucalyptus camaldulensis Terrestrial herbs: grass) (river red gum) Glinus lotoides Leptochloa digitata E. largiflorens (black (carpetweed) (umbrella canegrass) box) Glycyrrhiza acanthocarpa Paspalidium jubiflorum E. populnea (poplar box) (native liquorice) (Warrego grass) Heliotropum europaeum Sporobolus mitchelii H. supinum (couch) River Trees: Aquatic herbs: Echinochloa inudata (channel millet) channels and Acacia stenophylla (river Ludwidgia peploides (water cooba) primrose) Eleocharis acuta (common waterholes Eucalyptus camaldulensis Lemna spp. (duckweed) spike-rush) (river red gum) Terrestrial herbs: Eragrostis australasica E. coolabah (coolibah) Carthamus lanatus (saffron (canegrass) E. largiflorens (black thistle) Juncus spp. box) Gnaphalium luteo-album Panicum decompositum E. ochrophloia (Jersey cudweed) (native millet) (yapunyah) Goodenia spp. Paspalidium jubiflorum E. populnea (poplar box) Verbena officinalis (Warrego grass) Shrubs: (common verbena) Sporobolus mitchelii Eremophila divaricata (couch) (spreading emubush)

315

E. polyclada (flowering lignum) Muehlenbeckia florulenta (lignum) Lignum Trees: Aquatic herbs: Cyperus spp. swamps and Acacia stenophylla (river Lemna spp. (duckweed) Echinochloa inudata cooba) Marsilea spp. (nardoo) (channel millet) overflow Eucalyptus largiflorens Myriophyllum verrucosum Eleocharis spp. plains (black box) (red water-milfoil) Juncus spp. E. populnea (poplar box) Leptochloa digitata Shrubs: (umbrella canegrass) Chenopodium Panicum decompositum nitrariaceum (nitre (native millet) goosefoot) Paspalidium jubiflorum Muehlenbeckia florulenta (Warrego grass) (lignum) Stipa spp. (spear grasses) Nitraria billardieri (dillon Sporobolus mitchelii bush) (couch) Eleocharis Shrubs: Aquatic herbs: Cyperus spp. swamps Chenopodium Marsilea angustifolia Diplachne fusca (beetle nitrariaceum (nitre (naroow leaf nardoo) grass) goosefoot) M. drummondii (common Eleocharis acuta (common Muehlenbeckia florulenta nardoo) spike-rush) (lignum) Terrestrial herbs: E. pallens (pale spike-rush) M. horrida (spiny lignum) Alternanthera denticulata Elytrophorus spicatus (joyweed) (spike grass) Goodenia spp. Eragrostis australasica Mentha australis (river (canegrass) mint) Fimbristylis dichotoma Polygonum plebeium Juncus aridicola (tussock (knotweed) rush) Ranunculus spp. Leptochloa digitata (buttercups) (umbrella canegrass) Panicum decompositum (native millet) Paspalidium jubiflorum (Warrego grass) Sporobolus mitchelii (couch) Blackbox Trees: Aquatic herbs: Eleocharis acuta (common spike-rush) swamps Eucalyptus largiflorens Marsilea angustifolia (black box) (naroow leaf nardoo) E. pallens (pale spike-rush) E. populnea (poplar box) M. drummondii (common Eragrostis setifolia nardoo) (neverfail) Myriophyllum verrucosum Juncus aridicola (tussock (red water-milfoil) rush) Najas tenuifolia (water nymph) Terrestrial herbs: Centipeda cunninghamii (sneezeweed)

316

Eryngium plantagineum (eryngo) Glinus lotioides (carpetweed) Heliotropum europaeum H. supinum Mimulus repens (monkey flower) Polygonum plebeium (knotweed) Pratia darlingensis (matted pratia) Ranunculus spp. (buttercups) Stemodia florulenta (blue rod) Charophytes: Chara australis Nitella spp. Claypans Charophytes: Eragrostis australasica (canegrass) and Chara spp. Nitella spp. canegrass Aquatic herbs: swamps Marsilea drummondii (common nardoo) Terrestrial herbs: Bergia trimera (waterfire) Glossostigma diandrum (mudmat) Pratia darlingensis (matted pratia)

317

318

Appendix 2

Core and auger hole locations

Location of the coring sites is based on: Lake and core name GPS coordinates Other sources (WGS 84)

Lake Bindegolly core 4a & b 55 222918E 6892512N

Lake Wyara core 2 55 231338E 6823689N

Lake Numalla core 1 55 236085E 6816470N core 2 notes: ~10m east from core 1

Mid Blue Lake cores p1 & p2 surveyed transect; field notes and sketches auger hole (clay lunette) 55 301237E 6804153N auger hole (between lunettes) 55 301355E 6804160N auger hole (gypsum lunette) 55 301568E 6804171N

Lake Wombah cores 2 & 3 surveyed transect; field notes and sketches

Cummeroo Waterhole cores 3 & 4 field notes and sketches

Lower Bell Lake core G 55 286955E 6734064N

Palaeolake core B surveyed transect; field notes and sketches core K 55 288776E 6734491N auger holes (gypsum lunette and red surveyed transect; field notes and sand dune) sketches

Lake Willeroo auger hole 55 335952E 6670103N

Lake Yandaroo auger hole 55 340234E 6667540N core 55 340517E 6667180N

319

320

Appendix 3

Pollen/spore/carbonized particle/phytolith extraction method using gravity settling and heavy liquid separation

1. Place the sediment samples (1-6 cm3) in 50 ml centrifuge tubes. For larger samples increase the volumes of added chemicals accordingly. 2. Add 2-4 Lycopodium marker-spore tablets (depending on the expected concentration of the target; record the number of tablets used and the batch number). Add 5-10 ml 10% HCl to dissolve carbonates and 10-20 ml distilled water. Mix well. Centrifuge at 3000 rpm for 5 min. Retain supernatant. 3. Fill tubes to 1/2 level with distilled water. Vortex, centrifuge (5 min at 2500 rpm), decant.

4. Add 30-40 ml 5% Na4P2O7.10H2O (tetra-sodium pyrophosphate) [or 4% Na2SiO3 (sodium silicate)] and leave (at least) overnight to disperse clays. A week in dispersant is advisable for larger and clay-rich samples. 5. Sieve through 250 µm mesh into 500-1000 ml beakers (depending on the sample size). 6. Fill the beakers with distilled water, stir and leave to settle (5-6 hrs for 500ml (depending on the initial densities of suspended clay particles); 8hr for 1000ml). Siphon, refill and leave to settle again. Repeat until the water becomes clear at the end of settling (repeat step 4 if large amount of clay is still present – usually necessary for clay-rich samples). Return the remaining sediment into the 50 ml tubes.

7. Add 6 ml (to double the volume of sediment) Na6(H2W12O40).H2O (sodium polytungstate) specific gravity (SG) 2.0 (for pollen) or 2.3 for combined pollen and phytoliths. Mix well. Centrifuge at 1500 rpm for 20 min. Decant sodium polytungstate into additional centrifuge tube. Retain sediment fraction. Note: If pollen concentrations are expected to be low, extract the pollen and light phytoliths fraction first using heavy liquid of 2.0 SG. Reserve the settled sediment fraction. Carry out the acetolysis (steps 11-14) on the extracted pollen. From the reserved sediment extract heavy phytolith fraction using heavy liquid of 2.3 SG (change centrifuge settings to 5 min at 2000 rpm). 8. Top the tubes with decanted sodium polytungstate (and pollen) with distilled water (at least 20-30 ml). Mix well. Centrifuge at 3000 rpm for 5 min. Retain supernatant. (Retain sodium polytungstate for recycling*). 9. To retained sediment fraction from step 7, repeat step 7, decant heavy liquid into tubes containing extracted pollen fraction from step 8. Repeat step 8. Note: Repeat steps 7-8 for third time if sediments are expected to be poor in target material to ensure maximum recovery. 10. Fill tubes to 2/3 level with Glacial Acetic Acid (GAA) to remove water. Vortex, centrifuge, (5 min at 3500 rpm), decant. Repeat step 10.

321

11. Fill tubes with approx. 9mL of Acetic Anhydride and vortex. Add 1 ml sulphuric acid. Place tubes in 90°C water bath for 2 min. 12. Cool in/and centrifuge (5 min at 2500 rpm at 5°C), decant. 13. Fill tubes to 2/3 level with GAA to remove sulphuric acid. Vortex, centrifuge (5 min at 3500 rpm, ambient temp.), decant. 14. Fill tubes to 2/3 level with distilled water. Vortex, centrifuge (5 min at 2500 rpm), decant. When decanting aim to leave 1 ml of water in the bottom of the tube. 15. Mix the sediment with the remaining water from step 15 and transfer the contents to Eppendorf tubes. Add 0.2-1 ml of glycerol to each tube. Do not vortex. Centrifuge for 53 min at 3000 rpm. 16. Using small pipette decant the water from the tube leaving a minute amount of glycerol with the sediment.

* Sodium polytungstate recycling:

17. Filter the sodium polytungstate twice through a 1-2 µm mesh. 18. Concentrate the liquid though evaporation in hot (50oC) water bath.

322

Appendix 4

Modern reference pollen extraction

1. Insert flower heads with pollen into 50ml centrifuge tubes. 2. Fill tubes with 20ml 10% potassium hydroxide (KOH), balance the volumes, vortex. 3. Place open tubes into 90oC water bath for 10 min. 4. Optional sieving: required if flower heads are large: 4.1. Setup clean funnels on clamps with 250ml beakers. 4.2. Place 250µm stainless steel sieves into funnels. 4.3. Rinse the content of the 50ml tubes through the sieves with abundant quantities of distilled water. 4.4. Pour as much of the beakers’ content as possible back into the 50ml tubes, centrifuge, decant. Repeat until beaker is empty. 5. Cool the samples in centrifuge while spinning (setting the temperature at ~ 2oC) for 5 min at 2500 rpm. Alternatively, cool the samples in cold water before centrifuging. Decant the KOH waste into drain with continuous flush. 6. Fill tubes to 2/3 level with distilled water. Vortex, centrifuge (5 min at 2500 rpm, ambient temperature), decant. 7. Fill tubes to 2/3 level with Glacial Acetic Acid (GAA) to remove water. Vortex, centrifuge (5 min at 3500 rpm), decant. Repeat step 7 one to two times depending on the amount of water that is left in the tube after decanting (more needs more repeats). 8. Fill tubes with approx. 9ml of Acetic Anhydride and vortex. Add 1ml of sulphuric acid. Place open tubes in 90°C water bath for 2 min. 9. Cool in/and centrifuge (5 min at 2500 rpm at ~ 2°C), decant. 10. Fill tubes to 2/3 level with GAA to remove sulphuric acid. Vortex, centrifuge (5 min at 3500 rpm at ambient temperature), decant. 11. Fill tubes to 2/3 level with distilled water. Vortex, centrifuge (5 min at 2500 rpm), decant. When decanting aim to leave 1ml of water in the bottom of the tube. 12. Mix the pollen/supernatant with the remaining water from step 11 and transfer the content to Eppendorf tubes. Add 0.2-1ml of glycerol to each tube. Do not vortex. Centrifuge for 53 min at 3000 rpm. 13. Using small pipette decant the water from the tube leaving small amount of glycerol with the pollen.

323

324

Appendix 5

Pollen types

Acacia Aizoaceae Amaranthaceae (Amaranthus type) – the pollen assigned to this category has pores with distinctly larger diameter than those assigned to the Chenopodiaceae type Amaranthaceae (Gomphrena type) Apiaceae (Trachymene type) - the pollen assigned to this category is best characterised by the general description of the Trachymene genus by (Boyd, 1992) and might include pollen from other Apiaceae genera that fit this description Asteraceae (Liguliflorae) Asteraceae (Tubuliflorae) Brassicaceae Caryophyllaceae (Polycarpaea type) - the pollen assigned to this category is best characterised by the general description of the Polycarpaea genus by (Boyd, 1992) and might include pollen from other Caryophyllaceae genera that fit this description Casuarinaceae – while the size classification (Kershaw, 1981) was attempted in this study during counting, the small total number of grains and their widespread and relatively even size distribution made it impracticable to separate them into subcategories Chenopodiaceae Cyperaceae Dodonaea Eucalyptus This category contains all pollen that has been positively identified as belonging to the Eucalyptus genus. None of the counted pollen matched the description of ‘Angophora/bloodwoods’ provided by (Chalson and Martin, 1995). Any further decisions about divisions within this genus were hampered by inadequate reference collection, which contains only 6 species out of over 30 Eucalyptus spp. occurring in the region (incl. 2 collected in the Paroo/Warrego Region, 2 from other arid regions, and 2 from coastal area of the Hunter Valley). Arbitrary divisions based on grain size, wall thickness, and the characteristics of the apertures and the polar island (Boyd, 1992) have also been unsuccessful due to lack of consistent relationships between any combination of those features with some, like grain size and wall thickness, spread evenly through the whole range of recorded values. Euphorbiaceae (Chamaesyce type) – the pollen assigned to this category is best characterised by general description of the Euphorbia genus by (Boyd, 1992) and the Chamaesyce drummondii (Euphorbia drummondii Boiss.) image in The Newcastle Pollen Database Fabaceae (Type A) – this category includes grains with general characteristics of the Fabaceae family displayed mostly by Lotus, Swainsona, and Trigonella genera (Boyd, 1992); The Newcastle Pollen Database) Frankenia Goodeniaceae

325

Gyr ostemonaceae Haloragaceae Lamiaceae – the pollen matched the Teucrium racemosum image from The Newcastle Pollen Database Liliaceae Loranthaceae Muehlenbeckia – this pollen type matches that of Muehlenbeckia florulenta in the Newcastle Pollen Database Myoporaceae Myrtaceae (damaged) – this category contains pollen that displays the general characteristics of the Myrtaceae family but was too damaged or obscured to be subdivided any further Myrtaceae (other) – this category contains pollen of non-Eucalyptus genera of the Myrtaceae family Other – this category contains pollen that was encountered within a particular core low numbers in only one sample and within top 10 cm of the sediment (considered modern); it was used where the number of taxa in the top 10cm was large to improve the clarity of the pollen diagram Papaveraceae – this category contains pollen with characteristics displayed by Argemone ochroleuca in The Newcastle Pollen Database Pimelea – this category includes shrubs and forbs Plantago Poaceae Portulacaceae – the pollen in this category matches the image of Calandrinia pumila in The Newcastle Pollen Database Proteaceae (Grevillea type) – the name ‘Grevillea’ was used since this genus is the most abundant Proteaceae in the region (Cunningham et al., 1992), however, the count might include triporate pollen from other Proteaceae genera Rutaceae (~40 µm) – pollen with general characteristics of the Rutaceae family but with larger grain size than Geijera type. Rutaceae (Geijera type) - the pollen in this category matches the image of Geijera parviflora in The Newcastle Pollen Database Scrophulariaceae (Mimulus type) – the pollen in this category matches the image of Mimulus repens in The Newcastle Pollen Database Scrophulariaceae (Stemodia type) – the pollen in this category matches the image of Stemodia florulenta in The Newcastle Pollen Database Senna Sida/Lawrencia/Boerhavia type - the pollen assigned to this category is best characterised by general description of the Sida (Malvaceae) and Boerhavia (Nyctaginaceae) genera by (Boyd, 1992) and the Lawrencia glomerata and Boerhavia cocinea images in The Newcastle Pollen Database Solanaceae (Nicotiana type) – the pollen in this category displays the genral characteristics of the representants of the Nicotiana genus in The Newcastle Pollen Database; the grain size varies between 30-45 µm Solanaceae (Solanum type) – the pollen in this category displays the genral characteristics of the representants of the Solanum genus in The Newcastle Pollen Database; the grain size varies between 20-25 µm 326

Sterculiaceae (Keraudrenia type) - the pollen in this category matches the image of Keraudrenia integrifolia in The Newcastle Pollen Database better than other genera within this family Tribulus Typha Unidentified (damaged) – this category contains grains that can still be recognised as pollen, but are too damaged (worn out, folded, crumpled, fragmented) to be ever positively identified uk c4_02_04 – unknown pollen grain (photo 1 below) uk c4_32_02 – unknown pollen grain (photo 2 below)

Unknown pollen grains: 1 - uk c4_02_04 and 2 - uk c4_32_02.

Unknown type 1 Wahlenbergia

327

328

Appendix 6

Phytolith types

Arc-triangle - described by Bowdery’s (1998) group E (Poaceae) and group 2 (Dycotyledon and Monocotyledon – non Poaceae); no information was found to allow differentiation between those two groups Bi- and n-lobate: Bilobe – described by Bowdery’s (1998) groups A1-6 (Poaceae) Cross - described by Bowdery’s (1998) group B7 (Poaceae) and Thorn (2004) Polylobate - described by Bowdery’s (1998) groups B9 (Poaceae) and Thorn (2004) Ornamented rectangle (= elongate – crenulated) – the shapes counted for this category are best characterised by the description of Bowdery’s (1998) group F (Poaceae) and group J (Dycotyledon and Monocotyledon – non Poaceae); the initially limited phytolith expertise of the author disallowed differentiation between those groups during counting Rondel - described by Bowdery’s (1998) group C (Poaceae) and Thorn (2004) Saddle - described by Bowdery’s (1998) group C (Poaceae) and Thorn (2004) Trichome/hair - described by Bowdery’s (1998) groups Q (Dycotyledon and Monocotyledon – non Poaceae) and Thorn (2004)

329

330

Appendix 7

Charophytes: identified taxa and their habitat requirements

General comment A modern charophyte study within the Paroo wetlands recorded the presence of Nitella cf. ungula, Chara fibrosa, and Lamprothamnium species in Bloodwood and Blue Lakes (Garcia and Casanova, 2003). Taxa list Chara – Chara (incl. fibrosa) is a common genus in the Paroo lakes and one of the dominant charophyte taxa in many of the subsaline lakes of the region (Timms, 1993) Chara fibrosa – lives at 0.2-0.7m depth in very clear water with salinity 1-2g/L and pH 6-7; widespread in Australia (Garcia, 1999a); Chara (cf.) globularis – is a cosmopolitan species inhabiting a wide range of latitudes, altitudes, and depths (0.2-30m below water surface) and is widely distributed within Australia. It can withstand wide changes in temperature (even freezing). It prefers quiet fresh to hyposaline (0-5g/L) alkaline waters in creeks, lakes, and ponds with muddy, sandy and clayish bottoms, sometimes with calcareous remains (Garcia, 1994, 1999a). Lamprothamnium – is the only charophyte taxa that can tolerate a wide range of salinity as well as water levels and are widespread in ephemeral lakes. It grows in tropical and temperate areas in usually shallow waters (a few cm to 2m), which are a subject to intensive insolation resulting in high water temperatures (often up to 30oC). It prefers clear, calm to wave dominated waters (the latter often caused by wind action on shallow water) with sandy or sandy-muddy bottom. Germination usually occurs after filling of dry or almost dry basin with freshwater supplied by rainfall or floods. The euryhaline characteristic of Lamprothamnium and ability to adapt to changing environments results in great variability in the size as well as degree of calcification of oospores and gyrogonites, making the species separation difficult. Their presence in sediments indicates non-marine environments with salinity varying from fresh to hypersaline (generally between 2-3g/L to 70g/L, but rarely below 2-3g/L) and water bodies retaining water for at least 70 consecutive days. (Garcia, 1999a; Garcia and Chivas, 2004; Garcia et al., 2002) Lamprothamnium heraldii – the species was first described from Lake Gidgee (part of the Bloodwood lake system), Blue Lakes (incl. Mid Blue Lake), and Palaeo Lake (called Palaeolake in this study) (Garcia and Casanova, 2003) Lamprothamnium (cf.) succinctum – prefers alkaline waters with variable salinities (a few to 70g/L) and can tolerate water depths between a few cm to 2m. As a heliophile, favours clear water and sandy or sandy-muddy bottoms. Its occurrence records are limited to coastal lagoons in NSW. (Garcia, 1999b; Garcia and Chivas, 2004; Garcia et al., 2002) Nitella – taxa common in the Paroo lakes, often one of the dominant charophytes in subsaline lakes (Timms, 1993) Nitella (cf.) ignescens – taxa endemic to Australia as it has been found only in NSW and Victoria. Lives in subsaline shallow waters (a few centimetres) in soft sediments but also

331

found among hard rocks) along the shoreline. It is often exposed to desiccation and wave action. (Garcia, 1998, 1999a) Nitella (cf.) verticillata – inhabits brackish (subsaline to hyposaline) waters (Hotchkiss and Imahori, 1988)

332

Appendix 8

Invertebrates: taxa list and their habitat requirements

Gastropoda Coxiella – can withstand periods of desiccation (but not found in lakes that remain dry for a few years, e.g. in Central Australia) and high salinities by sealing its aperture with the operculum. The genus is not found in fresh water habitats. In Mid Blue Lake it was found living at salinities between 16.8-20.8g/L (with maximum of 34g/L). No living specimens have been found at great water depths and its presence indicates depths of less than 25m and most likely <6m. After death the shell fills with decay gas and floats, often towards the shore, where it can accumulate in extensive layers (hence good indication of shoreline). Otherwise, they are very sparsely distributed across dry lake floor. (Chivas et al., 1986; De Deckker, 1982a; De Deckker et al., 1982; Timms, in press) Coxiella gilesi Glyptophysa sp. – freshwater taxa; found by Timms (1993) in Lakes Yandaroo and Willaroo at 0-4.7g/L salinities Isidorella sp. – occurs in freshwater to subsaline environments (<3g/L) (B. Timms, pers. comm. 2006) Physella sp. – introduced taxa occurring in slightly hyposaline environments (3-5g/L) (B. Timms, pers. comm. 2006) Ostracoda Cyprinotus (edwardi?) – recorded in Mid Blue Lake and Lake Wombah at salinities of 3-33g/L (Timms, in press) and in Yandaroo and Lower Bell lakes at salinities 0.17-24.6 (however occurring usually below 10g/L) (Timms, 1993) Diacypris – recorded in Lake Wyara at salinities up to 50g/L (Timms, 1997b, 1998c), in Mid Blue Lake and Lake Wombah below 180g/L (Timms, in press), and in Lower Bell Lake between 5.2-263.8g/L, where it was one of the dominant taxa at higher salinities (Timms, 1993) Diacypris spinosa – occurring at salinity range 5-40g/L (De Deckker, 1988a) Heterocypris – occurring in Paroo at low salinities (3-20g/L) with maximum 41g/L at Mid Blue Lake and Lake Wombah (Timms, 1993, in press); elsewhere reported by De Deckker (1988a) at 5-10g/L salinities Reticypris - recorded in Lower Bell at salinities 5.2-122.9g/L becoming the dominant taxa at higher salinities (Timms, 1993). Also reported from Mid Blue Lake and Lake Wombah (Timms, in press). Elsewhere reported by De Deckker (1988a) at salinities below 140g/L. Trigonocypris globulosa – recorded in Lake Wyara at salinities below 10g/L (Timms, 1997b, 1998c); in Mid Blue Lake and Lake Wombah at 9.1-16.8g/L with max 68g/L (Timms, in press) and in Lower Bell Lake at 2.2-122.9g/L (Timms, 1993) Chironomidae Eukiefferiella sp. Claodotanytarsus sp. 333

334

Appendix 9

Dating results: 137Caesium, Radiocarbon, and Optically Stimulated Luminescence

137Caesium data

Equivalent to 137 137 Cs Cs Total Core: depth (cm) core no.: 2 (mBq/cm2) depth (cm): (mBq/g) (mBq/cm ) L. Bindegolly core 4: 159.723 0-10 3.973 ±1.336 47.009 ±15.809 ±24.473 10-20 6.815 ±1.130 112.715 ±18.682 20-30 Not detected 30-40 Not detected L. Numalla core 2: Core 1: 27.137 0-7.5 0-8.5 1.988 ±0.675 19.972 ±6.783 ±8.423 7.5-15 8.5-17 0.386 ±0.269 7.165 ±4.994 15-22.5 17-23 Not detected 22.5-30 23-35 Not detected Mid Blue L. core p1: 115.286 0-7.5 5.3674 ±1.305 26.942 ±6.552 ±11.928 7.5-15 9.428 ±1.064 88.344 ±9.967 15-22.5 Not detected 22.5-30 Not detected L. Wombah core 3: 193.000 0-10 5.024 ±1.082 49.300 ±10.618 ±15.814 10-15 12.232 ±1.440 78.451 ±9.233 15-20 9.249 ±1.023 65.250 ±7.216 20-25 Not detected 25-30 Not detected Cummeroo Waterhole core 3: Core 4: 0-5 0-5 Not detected 10-15 10-15 Not detected 15-20 15-20 Not detected 20-25 20-25 Not detected 25-30 25-30 Not detected 35-40 35-40 Not detected L. Yandaroo (augered hole): 0-10 13.790 ±0.647 10-20 2.692 ±0.560 20-30 Not detected 30-40 Not detected L. Yandaroo (core): 138.601 0-10 3.513 ±0.638 50.468 ±9.167 ±15.076 10-20 1.928 ±0.531 32.610 ±8.986 20-30 3.282 ±0.467 55.523 ±7.906 30-40 Not detected 40-50 Not detected

335

Radiocarbon (14C AMS) data

Site (core no.) Depth (cm) Conventional 14C age δ(13C) correction Lab code

Palaeolake (B) 54-56 7950 ± 110 -25 OZG886 119-121 14440 ± 150 -25 OZG887

Optically Stimulated Luminescence data

*Equivalent to core Depth Site (core no.) Age estimate Lab code no. (depth from (cm) surface)

Lake Bindegolly (4b) 133-139 K0696

Lake Wyara (2) 92-98 K0701

Lake Numalla (1) 32-39 K0695

Mid Blue Lake (p2) 28.5-32 K0702 p1 (27.5-30.5 cm)

Lake Wombah (2) 36-40 Awaiting K0703 3 (36-40 cm)

Cummeroo Waterhole (3) 44-49 K0699 4 (42-47 cm) results Palaeolake (B) 201-206 K0694 Palaeolake (K) 81-85 K0697 145-150 K0698

Lower Bell Lake (G) 68-73 K0692 153-158 K0693

Lake Yandaroo (cC) 81-86 K0700 cD (81-86 cm) * ‘the equivalent to core’ was specified where the sedimentcore, for which results are presented further in this chapter, was different from the core sampled for OSL; the interpolation was based on combination of several different sediment characteristics, visual as well as quantitative.

336