SEDIMENTOLOGY AND TAPHONOMY OF A TETRAPOD FOSSIL ACCUMULATION IN THE TRIASSIC BURGERSDORP FORMATION OF THE KAROO BASIN Dr Frederik Petrus Wolvaardt ORCID Number: 0000-0001-8117-558X A dissertation submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Master of Science. Cape Town, 2021 ii DECLARATION I declare that this dissertation is my own, unaided work. It is being submitted for the Degree of Master of Science at the University of the Witwatersrand, Johannesburg. It has not been submitted before for any degree or examination at any other University. F. P. Wolvaardt On this 12th day of August 2021 at Cape Town. iii ABSTRACT Field investigations of a rare accumulation of tetrapod fossils in the Triassic Burgersdorp Formation of the Karoo Basin reveal the palaeoenvironmental and behavioural mechanisms that led to its genesis. A total of 140 skulls and skeletons of seven tetrapod taxa and samples of three burrow cast geometries were collected with details of their depositional environment setting. These data are interpreted and integrated to reconstruct the palaeoenvironment and taphonomic pathways leading to accumulation, burial and preservation of this anomalous occurrence. The sedimentological facies analysis reveals a high sinuosity meandering river and floodplain setting. A semi-arid climate, with seasonal rainfall caused periodic rise and fall of the water table and floodplain ponds. Herbivorous tetrapods dominated by procolophonids, with dentitions adapted to browse fibrous reedbeds, congregated around pond margins. The animals scratch-burrowed in the floodplain surface for thermo-regulation and possibly estivation. Combined physical and behavioural factors caused the higher-than-average probability of tetrapod death, burial and preservation at this site. iv In memory of the two people who introduced me to the wonderful world of fossils. My father George Sebastiaan Wolvaardt 1938 to 2012 and world-renowned fossil-finder James William Kitching 1922 to 2003. v ACKNOWLEDGEMENTS During the course of this study, I have become deeply indebted to many individuals. My gratitude is therefore extended to the following persons and Institutes for their hospitality and help in various ways: To my supervisors Prof. Roger Smith and Prof. John Hancox for their many inputs, enthusiasm, support, critical counsel and encouragement during this work. I am particularly grateful to have been part of Prof. Smith’s Karoo field team since 2003 and his continuous motivation for me to become a palaeontologist. During these field trips I learned a great deal about palaeontology. Prof. Bruce Rubidge is thanked for his support of this project. I am indebted to the late Prof. James Kitching for first pointing out the Lemoenfontein locality to me in 1976. To my friend Dr Michael Strong for the many hours of debate about rocks and fossils during field trips to the Karoo. To Mr André van der Walt of the farm Lemoenfontein for permission to work on the farm over many years and his interest in fossils. To numerous colleagues and members of staff at the Evolutionary Studies Institute at University of the Witwatersrand. To the Iziko South African Museum for curating, preparing and providing access to the Lemoenfontein fossil collection. I am grateful for the many hours of laboratory and field work support from Ms Claire Browning and her interest in the project. I would also like to thank Ms Zaituna Skosan, Ms Tiffany-Lea Van Zyl and Mr Sibusiso Mtungata for the excellent preparation of the Lemoenfontein fossils and for making access to the collection possible. To my sister Léta for hosting my field camps on the family farm, Gladde Grond, and my brother Wimpie for always being on the lookout for new fossil localities. To my mother Alta for being the de facto curator of the initial Lemoenfontein fossils until 2003. Last but not least, to my wife Jessie and my children Laura and Luca, for giving up many potential family holidays while I was camping in the Karoo. vi TABLE OF CONTENTS 1. INTRODUCTION 1 1.1 Hypotheses tested 4 1.2 Objectives 4 1.3 Aims 5 1.4 Dissertation overview 6 2. MATERIALS AND METHODS 10 2.1 Field Based Observations 10 2.2 Laboratory and collections-based data capture 11 2.3 Analyses 12 3. SEDIMENTOLOGY AND LITHOSTRATIGRAPHY OF LEMOENFONTEIN 16 3.1 Overview of the lithostratigraphy of the main Karoo Basin 16 3.2 Lemoenfontein fossil site: geological setting 19 3.3 Basis for facies sequence description 23 3.4 Description of the Lemoenfontein facies sequences 29 Cycle 0 29 Cycle 1 30 Cycle 2 32 Cycle 3 32 Cycle 4 33 Cycle 5 34 Cycle 6 35 Cycle 7 35 Cycle 8 36 Bamboesberg Member 36 3.5 Detailed description and interpretation of the bonebed interval of Cycles 0 and 1 39 3.6 Interpretation of depositional environment 63 4. BURROW CAST ICHNOLOGY OF LEMOENFONTEIN 66 4.1 Background information on Triassic burrow casts 66 4.2 Lemoenfontein burrow casts 71 4.3 Reniform burrow casts 73 vii Reniform burrow architecture 73 Surficial morphology of reniform burrow casts 75 Reniform burrow fills 76 4.4 Lemoenfontein sub-circular burrow casts 76 Sub-circular burrow architecture 76 Surficial morphology of sub-circular burrow casts 77 Burrow fill 79 4.5 Qualitative analysis of Lemoenfontein burrow casts 81 4.6 Reniform burrow discussion 85 4.7 Sub-circular burrow discussion 90 4.8 Potential burrow-makers 92 Lemoenfontein reniform burrow trace makers 92 Lemoenfontein sub-circular burrow trace makers 98 Palaeoenvironmental significance of crayfish burrows 102 4.9 Putative Langbergia burrow cast and its biostratigraphic significance 104 5. CREVASSE SPLAY SURFACE ICHNOLOGY OF LEMOENFONTEIN 108 5.1 General description 108 5.2 Current crescents 110 5.3 Scratch-digging traces 115 5.4 Scoyenia ichnofacies 122 5.5 Sequence of deposition and trace events 123 6. VERTEBRATE TAPHONOMY OF LEMOENFONTEIN 125 6.1 Literature review of studies relevant to the taphonomic analysis of the Lemoenfontein bonebed 125 6.2 Lemoenfontein: Spatial and stratigraphic distribution of fossil bones 130 6.3 Lemoenfontein: Taphonomic analysis of fossil bones 132 Taphonomic data recorded 132 Lemoenfontein Taphonomic classes 132 Physical characteristics descriptors 133 Interpretation descriptors 134 Perimineralisation and nodule formation 137 viii 6.4 Taphonomy of tetrapod fossils in Gully 1 139 Teratophon SAM-PK-K10174 (Field number 117) 139 Teratophon SAM-PK-K011599 (Field number L18-2) 140 Teratophon Field number L18-1 147 Teratophon Field number L19-1 147 Teratophon Field number L20-10 148 Procolophonid SAM-PK-K10171 150 6.5 Taphonomy of tetrapod fossils in Bonebed 1: 150 Taphonomic Group 1: Isolated skull with articulated lower jaw 151 Taphonomic Group 2: Articulated skeletons encased in nodules that take the shape of the carcass 152 Taphonomic Group 3: Articulated skeletons curled up in a 3-D spiral, encased in nodules 156 Taphonomic Group 4: Isolated skulls and articulated postcranial elements SAM-PK-10201 (Field numbers 122 A/B, and L19-2 to L19-8) 157 6.6 Taphonomy of tetrapod fossils in Gullies 2 and 3: 160 Trirachodon SAM-PK-K10163 (Field number 98) 160 6.7 Taphonomic insights 162 7. VERTEBRATE PALAEONTOLOGY OF LEMOENFONTEIN 169 7.1 Overview of the Cynognathus AZ biostratigraphy 169 Lower Cynognathus AZ (Langbergia-Garjainia Subzone) (Olenekian) 170 Middle Cynognathus AZ (Trirachodon-Kannemeyeria Subzone) (Early Anisian) 170 Uppermost Cynognathus AZ (Cricodon-Ufudocyclops Subzone) (Late Anisian) 171 7.2 Systematics of the Lemoenfontein fauna 173 7.3 Discussion of the Lemoenfontein fauna 189 8. LEMOENFONTEIN PALAEOENVIRONMENTAL RECONSTRUCTION AND THE ORIGIN OF BONE ACCUMULATION 193 8.1 Overview of bonebed definition, classification, and genesis. 193 8.2 Triassic Karoo Bone Accumulations 198 Literature review of studies relevant to the sedimentologic and taphonomic analyses of the Lemoenfontein bone accumulation. 198 8.3 Lemoenfontein fossil accumulation 203 ix 8.4 Reconstruction of the once living Cynognathus AZ T-K Subzone ecosystem 205 Sedimentological interpretations 205 Crevasse splay sole surface interpretations 205 Burrow cast interpretations 206 Taphonomic interpretation of the bone accumulation 207 Faunal community interpretations 208 Palaeoenvironmental reconstruction 209 8.5 Origin of bone accumulations at Lemoenfontein 213 8.6 Comparison to other bone accumulations in the Triassic strata of the Karoo Basin 216 8.7 Comparison to the Lifua Member of Tanzania and the Santa Maria Formation of Brazil 218 9. CONCLUSIONS 220 REFERENCES 223 APPENDIX A: LITHOSTRATIGRAPHIC LOG 246 A.1 Legend 246 A.2 Lithostratigraphic log 247 APPENDIX B: TETRAPOD FOSSIL DATABASE 252 APPENDIX C: BURROW CAST DATA 263 x LIST OF FIGURES Figure 3.1: Geographic location of the Lemoenfontein bone accumulation site. 21 Figure 3.2: Google Earth Image of the locality and the measured section. The Lemoenfontein hill is in the foreground. This section crosses eight ledge forming sandstones each of which represents the base of a fining- upward third order cycle. The dashed lines show the path of the measured section from the channel of the spruit to the top of the hill and its extension from the second ledge forming sandstone (the stratigraphic equivalent of the top of the sandstone cap on the hill) up to the Bamboesberg Member. The final one being the Bamboesberg Member. The beds containing the bone accumulation are indicated in light yellow. Note that although the green portion of the section is following the downstream direction in the spruit, it is traversing upwards strategraphically due to the dip of the beds towards the southeast. 22 Figure 3.3: Condensed stratigraphic section showing the third order cycles, lithofacies codes, interpreted facies associations, depositional environments, and depositional system. The lithofacies codes, overbank facies associations and channel facies associations used to characterize the beds are provided in Table 3.1, Table 3.2 and Table 3.3, respectively. The legend for the sedimentary structures is provided in Appendix A. 38 Figure 3.4: Stratigraphic section of the beds that host the vertebrate fossil and burrow accumulation (-12 m to 26 m). Expanded sections provide an interpretation of the pedogenic sedimentary structures, the seasonal water table variation and the stratigraphical exten of the two types of burrow casts in the fossil hosting beds. The legend for the sedimentary structures is provided in Appendix A. 42 Figure 3.5: (A) BCk-horizon consisting of massive mudstone. Measuring staff = 1 m. (B) Skolithos bioturbation of surface (multiple light-grey disks, < 0.5 cm diameter, two examples arrowed). (C) Discrete Stage II calcareous nodules. Scale bars = 5 cm 44 Figure 3.6: (A) Strongly developed structural Bg-horizon consisting of prismatic peds and BCss-horizon with slickensides. Note the modern alluvium consisting of A, B and C soil horizons overlying the ancient palaeosols. (B) Calcified rootlets. (C) Mottling. Scale bar in B = 5 cm. Ruler in C = 10 cm x 10 cm. 46 Figure 3.7: Claystone coated slickensides in massively bedded silty mudstone. These pedogenic slickensides are convex-concave slip surfaces that form during expansion/contraction in expansive clay soils such as vertisols. Scale bar = 10 cm. 47 xi Figure 3.8: (A) BCk-, BCt- and calcareous nodular horizon, k. (B) Embedded procolophonid skull in nodular horizon (white arrow = snout, black arrow = back of skull). (C) Reniform burrow penetrating nodular horizon. Ruler in C = 10 cm x 10 cm, scale bar in B = 3cm. 51 Figure 3.9: (A), (B) Sandstone sheet, SS (arrowed), above calcareous horizon, BCk. (C) Dish- shaped sandstone lens (1–2 m diameter). Scale bar in B and C = 20 cm. 52 Figure 3.10: (A) Structural Bg-horizon containing numerous isolated, reworked and articulated procolophonid skulls and postcranial elements (B) and (C). White arrows = snouts of two skulls in dorsal view, red arrow = small skull in lateral view with teeth visible. Scale bars in B and C = 2 cm. 54 Figure 3.11: Crevasse splay sandstone at Lemoenfontein locality. (A) View of crevasse splay sandstone outcrop. (B) Circular current crescent casts on sole surface. (C) Casts of interpreted vertebrate scratch digging activity (arrowed) on sole surface. Ruler in A = 10 cm x 10 cm, Scale bar in B and C = 5 cm. 55 Figure 3.12: Schematic diagram showing the vertical stratigraphic position of the different types of soil horizons and the relative position of pedogenic and diagenetic carbonate formation within these. Also shown is the region where ground water influenced the formation of pedogenic calcrete. The rooted horsetail stands are not to scale and have been enlarged for clarity. (Diagram modified from Figure 4 in Tabor and Myers (2015)). 60 Figure 4.1: Reniform burrow casts. (A) B19-5, gently curving inclined burrow cast showing chevron pattern of ridges (Note the penetration by secondary smaller burrow). (B) B19-2, bilobed cross-section. (C) Sets of narrow oblique ridges on the latero-ventral surface of the B19-5 burrow cast. Two types of these sets are highlighted according to their angle with respect to the long axis of the burrow cast (Low angle: yellow dashed lines and arrows. High angle: red dashed lines). Ruler in A = 10 cm x 10 cm, scale bar in B and C = 2 cm. 74 Figure 4.2: Sub-circular burrow casts: (A) Burrow B19.7-2 with enlargement chamber near the top (total excavated length 80 cm). (B) Cross-section showing a diagenetic colour alteration halo. (C) Knobby and hummocky surficial morphology over printing sub-horizontal scrape marks (selected examples highlighted with yellow dashed lines). (D) Knobby and hummocky surficial burrow morphology from Hasiotis and Mitchell (1993, Fig. 6-E). Scale bar in B and C = 5 cm. 78 xii Figure 4.3: Sub-circular burrow casts. (A) Burrow B19-10 with less prominent surface scrape marks or ridges transverse to the burrow axis (selected examples highlighted with white dashed lines). (B) Surficial burrow morphology from Hasiotis and Mitchell (1993 Fig. 6-B). Scale bar = 2 cm. 80 Figure 4.4: (A) Plot of horizontal to vertical diameter. (B) Comparison of diameter ratios for the two types of burrows. 82 Figure 4.5: (A) Comparing ramp angles for the two types of burrows. (B) Plot of Ramp angle as a function of diameter ratio. (C) Plot of ramp angle versus burrow size. (D) Comparing burrow sizes for the two types of burrows. 83 Figure 4.6: Microgomphodon in the burrow terminal chamber (pictures provided by V. Fernandez, August 2019). (A) Burrow chamber surface showing scratch marks (arrowed). (B) Micro CT scan showing the articulated Microgomphodon skeleton. Scale bar = 10 cm. 94 Figure 4.7: (A) Giant reniform burrow terminal chamber from the study locality. (B) Distinct reniform cross-section (width = 15 cm, mid-level height = 7.5 cm, maximum height 12.5 cm). Scale bar = 5 cm. 96 Figure 4.8: Schematic model of continental freshwater crayfish burrowing near a water source where there is a seasonal variation in the ground water table (based on Abouessa et al., 2015, Fig. 8). (A) Mid-season water level and burrow depth; (B) Dry season water level, older burrows are extended down to the new water level and new burrows are excavated all the way to the lowered water level, note the change in the burrow orientation from sub-vertical to oblique below the mid-level enlargements; (C) Wet-season water level, older burrows are partially filled in with mud or sand and can get penetrated by smaller new burrows. (This is a diagrammatic representation and the burrows and reedbeds are not to scale, the former have been enlarged for clarity). 103 Figure 4.9: Large reniform burrow cast at 1 m of the Lemoenfontein stratigraphic section. (A) location in siltstone cliff face (arrowed). (B) View showing T- junction. Hammer length = 33 cm (arrowed). (C) Left shaft and terminal chamber. (D) Right shaft and terminal chamber. Ruler = 10 cm. 105 Figure 4.10: Comparison of the Lemoenfontein bifurcating burrow cast (A) with the Langbergia burrow casts (B) and palaeoenvironmental reconstruction (C.) B and C are Figures 6 and 13 of Groenewald et al. (2001). Hammer length in A = 33 cm. Scale bar in B = 15 cm. 106 Figure 5.1: (A) View of tabular crevasse splay sandstone that crops out at the Lemoenfontein study site. (B) Skolithos vertical burrows opening onto sharp top surface of crevasse splay sandstone. Examples arrowed. Ruler in (A) = 10 x 10 cm, Scale bar in (B) = 15 cm. 109 xiii Figure 5.2: (A) Cluster of aligned current crescents. Note Cynodontipus traces in the foreground. (B) Desiccation crack dissecting current crescents. Foreground scale bar in A = 15 cm, Scale bar in B = 5 cm. 111 Figure 5.3: (A) Stem casts preserved in 3-D. (B) Horizontal stem compressions on top surface of crevasse splay sandstone (arrowed). (C) Pith casts preserved as the central peg in each crescent. (D) Vertically preserved stem cast showing horizontal nodes and vertical ridges around the stem. Scale bar in A = 5 cm, scale bar in B = 10 cm, Scale bar in C= 4 cm, scale bar in D = 3 cm. 112 Figure 5.4: Typical examples of Horseshoe Vortex simulation and experimental results. Diagrams from (A) Jahangirzadeh et al. (2014), (B) Zhao et al. (2010), (C) Yagci et al. (2017). (D) Modern example of a scour around a sub-round pebble (picture taken after an overbank flood at the Lemoenfontein locality). Ruler in D = 10x10 cm. 113 Figure 5.5: Lemoenfontein scratch-like traces. (A) Slab containing an elongated trace and a single crescentic trace (arrowed in black) and a possible Rhynchosauroides footprint (arrowed in white). (B) Elongated trace showing a bilobed architecture. (C) Crescentic trace showing opposing scrapes. (D) Cluster of three crescentic traces. Scale bars = 5 cm. 117 Figure 5.6: Examples of Cynodontipus terminal burrow traces from Olsen et al. (2020), Figure 23. Scale bar = 1 cm. 118 Figure 5.7: (A) Example of crescentic scrape marks from Lemoenfontein (B) Colour coded traces of scrape marks that occur in sets with four to five subparallel scrapes with the same direction, depth, and length, this being indicative of the number of claws on a tetrapod manus or pedes. Scale bar = 1 cm. 118 Figure 5.8: (A) Trace of the rhynchosaur manus (SAM-PK10180). (B) Superposition of this trace on a Cynodontipus specimen from Lemoenfontein. (C) and (D) Superposition of this trace on a possible Rhynchosauroides print from Lemoenfontein. Scale bars = 1 cm. 121 Figure 5.9: Scoyenia ichnofacies. (A) Narrow softground meniscate feeding trace (arrowed). (B) Wide firmground striated meniscate traces. Scale bars = 1 cm. 122 Figure 5.10: Sequence of events that allowed the Cynodontipus, Scoyenia, and Rhynchosauroides traces to be made on the mud surface as well as the preservation of Sphenophyte stem casts. Drawing by Claire Browning (2021). 123 xiv Figure 6.1: Drone image showing the Lemoenfontein hill in the foreground with the three exposures of the fossiliferous beds, Gully 1, Bonebed 1, and Gullies 2 and 3 in the distance. 131 Figure 6.2: Teratophon (SAM-PK-K10174). (A) In-situ during excavation, showing association with Reniformichnus, and vertical position relative to the calcareous horizon. (B) Specimen post-excavation showing calcium carbonate surface coating. (C) Specimen post-preparation. Scale bar = 5 cm. 140 Figure 6.3: (A) L18-2 lying in the base of erosion gully in the orientation and attitude that it was found, which although apparently ex-situ it is still considered to be in its original stratigraphic position. Note the association with Reniformichnus and its stratigraphic level relative to the calcareous horizon. (B) L18-2 before preparation showing the detached skull block and colour mottling attributed to hydromorphic gleying of the mudrock surrounding the specimen. (C) Specimen after mechanical preparation showing the presence of 2 fully articulated curled-up Teratophon skeletons, one on top of the other. Top arrow indicates position of the second skull. Dashed line traces the inferred still-covered articulated spine, and the bottom arrow indicates the string of articulated caudal vertebra attributed to the second specimen. (D) Lateral view of second Teratophon skull lying dorsal-up beneath the still articulated, but slightly dislodged, thoracic ribs of the top specimen. Note the position of two gastralia ribs ( arrowed) that had also become slightly dislodged from the top specimen. This taphonomic style suggests that the two sub-adult Teratophon had their skin intact when they became arranged into this configuration, and that they were most likely both alive. An underground burrow setting is therefore the most likely scenario. (E) Right lateral view of uppermost specimen. The two skulls have small differences in shape and size that may be attributed to ontogeny or sexual dimorphism. All scale bars = 5cm. 146 Figure 6.4: (A) Teratophon (L18-1), skull embedded in calcareous nodular horizon. (B) L18-1 Specimen afer preparation. (C) L19-1 Teratophon lower jaw embedded in calcareous nodular horizon. (D) L19-1 Teratophon lower jaw after preparation. (E) In-situ Teratophon articulated skeleton (L20- 10) coated in cemented mudstone. Identification based on triangular shape of skull (arrowed). Scale bars in A = 3 cm and B = 4 cm, scale bar in C and D = 2 cm, scale bar in E = 10 cm. 149 Figure 6.5: Examples of Taphonomic Group 1: Isolated skull with articulated lower jaw. (A) and (B) Microgomphodon skull (L20-3), encased in carbonate nodule as found in the field. (B) Trirachodon skull (L19-17) encased in carbonate nodule as found in the field. (D) Very rare Lumkuia skull (L19- 19) encased in carbonate nodule as found in the field. Scale bar in A, B and C = 3 cm, scale bar in D = 2 cm. 152 xv Figure 6.6: Examples of Taphonomic Group 2: Articulated skeletons encased in nodules that take the shape of the carcass. (A) Microgomphodon (SAM- PK-K11601), skeleton encased in carbonate nodule that takes the shape of the carcass. This is the first Microgomphodon specimen with a fully articulated skeleton attached to the skull. (B) Left lateral view of SAM- PK-K11601 skull, showing possible preburial damage (arrowed) and excellent preservation quality of the bone. (C) Microgomphodon oligocynus (SAM-PK-K10160), skull with articulating cervical vertebrae and manus encased in carbonate nodule (articulated manus removed from nodule during preparation). (D) Eohyosaurus wolvaardti, SAM-PK- K10180, skull was encased in carbonate nodule before preparation. Note articulated right manus on temporal region. Scale bar in A = 15 cm, scale bar in B = 3 cm, scale bar in C and D = 2 cm. 155 Figure 6.7: (A) Examples of juvenile Teratophon articulated skeletons tightly curled- up and encased in carbonate nodules. (A) and (B) L18-5 dorsal and lateral views. (C) Teratophon (L19-21), note the position of the manus beneath the skull and 3-dimensional preservation of the articulated ribs in life position. (D) Teratophon (L19-23B), articulated juvenile skeleton. Scale bars = 3 cm. 157 Figure 6.8: (A) Structural Bg-horizon, source of numerous isolated, reworked, and articulated skulls and postcranial elements. (B) Procolophonid skull, SAM-PK-K10208. (C) Trirachodon skull SAM-PK-K10207, note evidence of reworking in that the snout has been weathered off and "re-sealed" after final burial with an iron-rich crust. (D) semi-articulated postcranial elements. Scale bars = 3 cm. 158 Figure 6.9: Trirachodon SAM-PK-K10163 (A) Ventral view of the anterior half of the burrow tunnel-shaped nodule containing skull and skeleton showing the orientation of the skull that is downwards and to the right. Note the presence of the maxillary platform that is characteristic of Trirachodon. (B) Dorsal view of the cylindrical nodule as found in the field (left squamosal arrowed). (C) Dorsal view of skull. Scale bars = 3 cm. 161 Figure 7.1: (A) Right lateral and (B) left lateral views of the rhynchosaur, Eohyosaurus wolvaardti, SAM-PK-K10159, (holotype, from Butler et al., 2015). Right lateral (C), dorsal (D) and left lateral (E) photos of prepared skull of SAM-PK10180, Eohyosaurus wolvaardti. Note articulated right manus lying across the temporal fenestra on the right side. Scale bars = 2 cm. 174 Figure 7.2: (A) Right lateral, (B) dorsal and (C) left lateral photos of prepared skull of SAM-PK-K10762, Trirachodon berryi from the Lemoenfontein locality. Note prominent maxillary platform (arrowed). Scale bar = 2 cm. 176 xvi Figure 7.3: (A) Ventral view of Langbergia modisei, holotype specimen NMQR 3255 (from Abdala et al., 2006). Note the absence of a prominent maxillary platform (arrowed). (B) Ventral view of partially prepared skull of L19- 17, Trirachodon and (C) ventral view of prepared skull of SAM-PK10762, Trirachodon berryi from the Lemoenfontein locality. Note prominent maxillary platforms in B and C (arrowed). (D) Lateral view of L20-12, Cricodon, showing sectorial postcanines (arrowed). Scale bar = 2 cm. 178 Figure 7.4: (A) and (B) Right lateral views of the prepared skull of the holotype of Lumkuia fuzzi, Specimen BP/1/2669 (Picture in A by C. Kammerer (2018), drawing in B from Hopson and Kitching, 2001). (C) Dorsal view showing possible parietal foramen endocast (arrowed), (D) right lateral view and (E) left lateral view of the new prepared skull of Lumkuia, Specimen L19- 19. Scale bar = 2 cm. 179 Figure 7.5: (A) Microgomphodon oligocynus (SAM-PK-K10160), dorsal, left lateral and ventral views of the skull with articulating cervical vertebrae and manus that was encased in carbonate nodule (articulated manus removed from nodule during preparation), Image from (Abdala et al., 2014). (A) Microgomphodon (SAM-PK-K11601), skeleton encased in carbonate nodule that takes the shape of the carcass. This exceptional specimen is the first Microgomphodon specimen with a fully articulated skeleton attached to the skull. (C) Dorsal view of its skull. (D) Right lateral view of its skull. Scale in A bar = 1 cm, Scale bars in B, C and D = 3 cm. 181 Figure 7.6: (A) Teratophon spinigenis (SAM-PK-K010174). (B) Dorsal and right lateral view of Teratophon spinigenus (skull of L18-2). (C) Dorsal and left lateral views of the holotype of Thelerpeton oppressus (BP/1/4538, from Modesto and Damiani (2003)). Scale bar = 2 cm. 183 Figure 7.7: Ontogenetic range of Teratophon skull sizes from Lemoenfontein. (A) Juvenile (SAM-PK-K10190). (B) Sub-adult (SAM-PK-K011599, L18-2). (C) Adult (SAM-PK-K010174). Scale bars = 4 cm. 184 Figure 7.8: Comparison of juvenile Teratophon to Thelerpeton. (A) and (B) Thelerpeton paratype (BP/1/4586), dorsal and right-lateral views, respectively. (C) and (D) Juvenile Teratophon (SAM-PK-K10190), dorsal and right-lateral views, respectively. Scale bars = 4 cm. 185 Figure 7.9 (A) Teratophon specimen 18-1 showing prominent right quadratojugal horn after preparation. (B) Thelephon specimen SAM-PK-K10172. Scale bars = 4 cm. 187 Figure 7.10: (A) Dorsal view, (B) left lateral view, and (C) right lateral views of Thelephon contritus (PK-K010162). Note the expanded marginal teeth in B and C (arrowed). Scale bar = 2 cm. 188 xvii Figure 7.11: Relative frequency of the fossils of different genera collected from the Lemoenfontein locality. 189 Figure 8.1: (A) Spatial abundance of tetrapod fossils in Bonebed 1. The size of each circle correlates to the number of specimens assigned to the same GPS coordinate. (B) Spatial distribution of tetrapod fossils covering Gully 1, Bonebed 1 and Gullies 2 and 3. Each white dot represents an individual specimen (Black rectangle corresponds to the area in (A)). 204 Figure 8.2: Schematic palaeoenvironmental reconstruction of the once-living ecosystem at the base of the Cynognathus AZ T-K Subzone at the Lemoenfontein locality. The landscape was dominated by a floodplain dotted with seeps and standing water bodies that became the preferred habitat of small herbivorous and carnivorous cynodonts, archosaurs, therocephalians and procolophonids. These semi-permanent ponds were surrounded by sphenophyte reed beds that acted as source of food. The stratigraphic sequence has been simplified from that of Lemoenfontein. Active and older buried burrows and bones are indicated. Also, areas of down wasting of the floodplain following the abandonment of previously active channels as well as areas of floodplain deposition due to sediment laden overbank flood events is also presented schematically. 212 xviii LIST OF TABLES Table 2.1: Data collection and analysis process. Columns identify the main elements of the study and their progression from left to right delineates the data collected and methods used to obtain the partial and final interpretations. These steps were applied for each of the main aspects of the study i.e., sedimentology, vertebrate fossils, and trace fossils. The insights from these topics combine for an integrated interpretation. 15 Table 3.1: Lithofacies Classification 23 Table 3.2: Overbank facies associations (adapted from Colombera et al., 2013) 24 Table 3.3: Channel Facies Associations (adapted from Colombera et al., 2013) 25 Table 4.1: Lemoenfontein burrow casts - basic dimensions 75 Table 4.2: Lemoenfontein burrow casts characteristics and comparison to similar burrow casts 87 Table 4.3: Skull width of the possible burrow makers 94 Table 6.1: Taphonomic classes with characteristic descriptors (adapted from Smith, 1989) 136 Table 8.1: Summary of the various biogenic and physical bone accumulation mechanisms. Specific examples and influencing factors of these mechanisms are also provided. Summary based on Rogers et al., 2007. 196 Table 8.2: Summary of the possible post concentration taphonomic pathways and their resultant signatures. Summary based on Rogers et al., 2007. 197 1 1. INTRODUCTION The main Karoo Basin of South Africa is infilled with a near-continuous sequence of sedimentary rocks deposited over a period of almost 120 million years (Catuneanu et al., 2005). The rocks accumulated in a retro-foreland basin (Catuneanu et al., 1998) in southwestern Gondwana from the Late Carboniferous (300 Ma) to the Early Jurassic (183 Ma) (Duncan et al., 1997). Karoo Supergroup strata cover two thirds of the surface area of South Africa, although much of this is sub-outcrop that is covered by Cenozoic deposits. Based on their lithological and sedimentological characteristics, several groups can be defined within the Karoo Supergroup. From oldest to youngest, these are Dwyka, Ecca, Beaufort, Stormberg and Drakensberg groups (Johnson, 1976; SACS, 1980). Deposition of the Karoo succession began during the melt-out stages of the Late Carboniferous Gondwanan glaciation and terminated with the early stages of breakup recorded by the vast basaltic flows that form the bulk of the Drakensberg Group (Duncan et al., 1997). The main Karoo Basin is the largest and deepest (5 km at its maximum in the southern foreland trough) of several connected or partially connected basins on Gondwana (Hancox, 1998; Catuneanu et al., 2005; Scheiber-Enslin et al., 2015). The succession thins significantly towards the north, where it rests on the stable margin of the basin and against the upwarping forebulge (Catuneanu et al., 2005; Rutherford et. al, 2015). The Beaufort Group is divided into a lower Adelaide Subgroup and an upper Tarkastad Subgroup (SACS, 1980; Catuneanu et al., 2005). These dominantly fluvial deposits are made up of fining-upward cycles of sandstones, siltstones, and mudstones (Catuneanu et al., 2005). The Tarkastad Subgroup can be distinguished by its greater sandstone-to-mudstone ratio (Rutherford et al., 2015). The Burgersdorp Formation is the final depositional episode of the Beaufort Group. It is paraconformably overlain by the Molteno Formation, which is the basal unit of the Stormberg Group. The Burgersdorp Formation consists mainly of mudrocks with inter- bedded sandstones in fining-upward cycles of a few metres to tens of metres in thickness, with a maximum thickness of 1000 metres in its southernmost outcrops (Johnson et al., 2 2006). It is considered to be Early to Middle Triassic in age (Hancox, 1998, p. 11). Recent radiometric dates from biostratigraphic correlatives in South America suggest an early Late Triassic age (Marsicano et al., 2016; Ottone et al., 2014), however there is significant controversy regarding this interpretation (Kammerer and De los Angeles Ordoñez, 2021). An extensive study of the Burgersdorp Formation is reported by Hancox (1998) and other geological descriptions are provided by Karpeta and Johnson (1979), Dingle et al. (1983), Johnson (1976, 1984), Hiller and Stavrakis (1984), Johnson and Hiller (1990), Kitching (1995), Hancox (2000), Neveling et al. (2005) and Rutherford et al. (2015). Burgersdorp Formation rocks consist of isolated, lenticular, feldspathic fine to medium- grained channel sandstones, common tabular fine-grained crevasse splay sandstones and argillaceous overbank mudrocks. The sandstones often overlay a basal lag consisting of reworked mudclast conglomerate (Hancox, 1998, p. 55). The common greyish-red (5R 4/2) to dusky-red (5R 3/4) massive overbank fines consist mainly of siltstones and mudstones and may contain sand-filled mudcracks and vertebrate and invertebrate burrows. Playa lake deposits are present in the form of well-laminated reddish mudrocks with pedocrete (cemented soil) horizons (Neveling, 2002; Hancox et al., 2010). It is generally accepted that the Burgersdorp Formation mudrocks and sandstones were deposited by northwesterly flowing meandering rivers, during a warm, arid to semi-arid climatic interval (Hancox, 1998). These high sinuosity rivers transported a mixed sediment load and regularly flooded, resulting in overbank traction and suspended load deposition on the floodplain surface. Some granular sheetwash beds as well as shallow-ephemeral floodplain channel fills and bedload dominated braided river deposits are also present, particularly in the lower part of the Burgersdorp Formation. Lacustrine palaeoenvironments also occur in the lower Burgersdorp Formation in the northern part of the Karoo Basin (Groenewald, 1996; Hancox et al., 2010). The rocks of the Beaufort Group preserve a rich tetrapod fossil record that has allowed for the establishment of seven assemblage zones, with nine subzones (Smith et al., 2020). Due to the abundance of body and trace fossils, their excellent preservation and their near- continuous temporal record, the biostratigraphy of the Beaufort Group has become a global 3 standard for the correlation of Middle Permian to Middle Triassic terrestrial sequences (Lucas, 1998; Hancox, 2000; Neveling et al., 2005). The Triassic strata of the Beaufort Group consists of the Early Triassic Lystrosaurus declivis Assemblage Zone (AZ) and the Early to Middle-Triassic Cynognathus AZ (Rubidge et al., 1995; Botha and Smith, 2020; Hancox et al., 2020). The Cynognathus AZ corresponds lithostratigraphically to the Burgersdorp Formation. Although there are few lithostratigraphic marker horizons in the Burgersdorp Formation, the Cynognathus AZ has been subdivided into three subzones based on differences in stratigraphic (and likely temporal) distribution of fossil tetrapod taxa (Shishkin et al., 1995; Hancox et al., 1995; Hancox, 1998; Hancox et al. 2020). The exceptional fossil record of the Beaufort Group also provides extensive data for the study of terrestrial vertebrate evolution. Studies on the taphonomy of Triassic bonebeds in the main Karoo Basin are few (Smith, 1980; Smith, 1989; Smith and Kitching, 1997). Kitching (1963) first drew attention to the occurrence of “bonebeds” in the Middle Triassic exposures in the Joe Gqabi (Burgersdorp) district, located on the farms Cragievar and Winnaarsbaken. The term “bonebed” has since been more rigidly defined (Rogers et al., 2007). Here, the term is used as a general term to reflect the anomalous accumulation of vertebrate fossils. Kitching also pointed out a locality on the farm, Lemoenfontein, in the Xhariep (Rouxville) district, where he found what he interpreted to be cynodont burrow casts and a few Trirachodon and procolophonid skulls (pers. comm., 1976). Subsequent exploration of this locality, first in my personal capacity and then followed by a formal field trip to the locality in 2003 accompanied by Dr Smith (then curator of Karoo Palaeontology at Iziko SA Museum, Cape Town) resulted in the discovery of a variety of well-preserved fossils of several different taxa and the identification of this locality as a potential bonebed. The Lemoenfontein locality has so far delivered over 140 identifiable skulls of at least six different taxa, many attached to articulated skeletons. The taxa from this bonebed have several unique features in common, such as herbivory, small body size and a high degree of articulation. To date, the collection includes two holotypes and one potential paratype. Along with the body fossils at least three different types of interpreted burrow cast geometries have also been identified at this field site. 4 1.1 Hypotheses tested Based on the information collected during this study five different hypotheses were tested. • The sedimentology and taphonomy of the locality indicate a floodplain lake or pond margin paleoenvironment, colonised in the long term by a diversity of taxa and supporting abundant vegetation in the wet season. • The changing water table (calcareous nodular horizon), sheet wash burrow fill, slickensides, textural B horizons, pedogenesis and mud crack casts point to seasonal variation in rainfall (and possibly to periods of drought). • The bone accumulation mechanism at this locality is a combination of intrinsic biogenic and physical concentration agents. Climatic rather than tectonic factors were the overriding influence on accumulation of sediments and bones at this site. • The sedimentary strata hosting the Lemoenfontein bonebed are at the transition between the Cynognathus AZ Langbergia-Garjainia (L-G) Subzone and the Trirachodon-Kannemeyeria (T-K) Subzone. • Procolophonids dug underground burrows or scratched-out near surface holes in the floodplain soils surrounding ponds and lakes. 1.2 Objectives The main purpose of this study is to identify the palaeoenvironmental and behavioural mechanisms that led to the accumulation of tetrapod fossils at the Lemoenfontein locality. Firstly, field data were gathered by studying the Lemoenfontein sedimentology, the burrow casts, other traces, and the tetrapod fossil faunal diversity and taphonomy. Secondly, these data were integrated and interpreted to reconstruct the palaeoenvironment and taphonomic pathways that explain the anomalous accumulation of tetrapod fossils. 5 As a secondary objective, it is intended that this detailed investigation of the Lemoenfontein concentration of fossil bones will make important contributions to our understanding of the bone accumulation mechanisms, and in doing so, generate information that will help to reconstruct the palaeoenvironments and palaeoecology of Middle Triassic tetrapods in the main Karoo Basin of South Africa more accurately. It has been observed that procolophonid fossils often occur in clusters in Triassic Karoo sedimentary rocks and it has been postulated that they were burrowing animals (Colbert and Kitching, 1975; Sidor et al., 2008; Bordy and Krummeck, 2016; deBraga, 2003). Many of these citations referred to Kitching’s verbal communication as reported by Groenewald (1991) about possible procolophonid fossil remains in a burrow. No formal study to test the validity of these observations has been done to date, and this is one aspect that the present study aims to address. 1.3 Aims To gain an understanding of the bone accumulation mechanisms and generate information to reconstruct the palaeoenvironments, this study adopted several aims. • A systematic search of the locality to collect remaining skulls and skeletons in ex-situ calcareous nodules within the surface “float”, as well as those still embedded in the mudstone or preserved in burrow casts. • Preparation of a representative sample of the fossils from this locality and with the help of experts in the taxonomy of the various groups, achieve a positive identification to species level, if possible. • Search the collections of the Evolutionary Studies Institute, Iziko South African Museum and National Museum to include all previously collected fossils from this locality in the database. • Analyse and describe the biodiversity of the tetrapod assemblage at the bonebed locality. • Identify and describe the different taphonomic styles of fossilization at the locality. 6 • Sample and log the different types of burrow casts from the locality. Use the systematic methods of Miller et al. (2001) to identify and classify the burrow casts and to identify the possible burrow producers. • Measure and sample a detailed sedimentological section from the lowest mudrock exposures below the bonebed locality up to the base of the Bamboesberg Member of the Molteno Formation and plot the occurrence of the fossils and the burrow casts on this log. Locate existing fossils and burrows relative to the B horizon and calcareous nodular horizon that demarcate the fossiliferous beds at this locality and make informed interpretations about the trace makers, their reasons for burrowing and the possibility that burrows are terminated just above the ancient water table. • Based on the insights from sedimentology and taphonomy, propose a mechanism that explains the anomalous accumulation of bones at the locality and derive conclusions about the ancient floodplain processes, depositional environments and climate. 1.4 Dissertation overview An extensive literature survey was performed at the beginning of this study. This survey interrogated relevant studies on the Triassic aged strata in the main Karoo Basin, with particular focus on the Burgersdorp Formation. The results of the literature survey are presented as an introduction to each of the core chapters of this dissertation according to the main topic under investigation in that chapter. The methods and materials used for field, laboratory and collections-based data capture are described in Chapter 2. It also contains a description of the approach used to analyse and interpret this information. The sedimentology and lithostratigraphy of Lemoenfontein are described in Chapter 3. First, the results of the literature survey of these aspects within the main Karoo Basin, with emphasis on the Burgersdorp Formation are provided. This is followed by a description of the 7 geographical and geological setting. The third-order upward-fining facies sequences that extend from the informal Eldorado marker up to the Bamboesberg Member are described, along with a detailed analysis of the beds hosting the bone accumulation in the first cycle. Chapter 3 concludes with an interpretation of the depositional environment. Chapter 4 presents a full analysis of the burrow cast ichnology of the Lemoenfontein site. The chapter starts with the results of the literature survey of Triassic burrow casts. An overall description of the burrow casts is then provided. The results of the detailed analysis of the reniform and sub-circular burrow casts that considered burrow architecture, surficial morphology, burrow fill and characteristic burrow measurements are presented. The chapter concludes with a discussion of the potential burrow-makers and the palaeoenvironmental significance of these burrows. The ichnology of the crevasse splay sole surface is presented in Chapter 5. A general description of the sandstone is provided, followed by a discussion of the flow induced current crescents, casts and imprints of sphenophyte stems, desiccation cracks, various worm, insect, and arthropod traces, as well as tetrapod scratch-digging traces. A discussion of its similarity to the Scoyenia ichnofacies is followed by an interpretation of the palaeoenvironmental significance of these traces. The taphonomy of the tetrapod fossils is addressed in Chapter 6. This chapter starts with the results of the literature survey of the vertebrate taphonomy in the Karoo Basin, followed by an introduction to the spatial and stratigraphic distribution of the fossil bones on Lemoenfontein. Descriptors for the observable taphonomic characteristics and the interpretation of these characteristics are introduced. These descriptors are then used to identify taphonomic groups to which a representative sample of the fossils could be assigned in a systematic way. Based on this analysis, insights are derived about the palaeoenvironment and the animal behaviour. 8 A systematic description of the Lemoenfontein tetrapod fauna is presented in Chapter 7. The chapter starts with an overview of the Cynognathus AZ biostratigraphy. Seven different faunal types from the main families that are present: Rhynchosauria (one genus), Cynodontia (three genera), Therocephalia (one genus), and Procolophonidae (two genera) are identified at genus and species level. The relative frequency of occurrence of the taxa is calculated and presented in graphic form. Key aspects of these taxa, with a focus on dentition, are described and analysed. The insights gained from this assessment are then used to interpret the palaeobehaviour of the fauna as well as to identify the biostratigraphic position of the Lemoenfontein site. In Chapter 8 the insights gained from the investigation of the Lemoenfontein sedimentology, the crevasse splay sole surface, the burrow casts, the taphonomy, and the fauna are summarised and combined to interpret how palaeoenvironmental and behavioural factors contributed to the concentration of tetrapod fossils at the Lemoenfontein locality. The chapter starts with a description of a generic conceptual framework for the genesis and analysis of vertebrate skeletal concentrations, as well as the results of the literature survey of Triassic Karoo Basin bone accumulations. The accumulation of fossil skulls and skeletons at Lemoenfontein is described and each occurrence is plotted on aerial photographs of the site. The palaeoenvironment and once-living ecosystem at the base of the Cynognathus AZ T-K Subzone is then reconstructed by integrating the evidence from the various investigations. A schematic illustration of this palaeoenvironment is provided. This reconstruction is then used to develop the most parsimonious explanation for the origin of the concentration of fossil bones at Lemoenfontein. The chapter concludes by comparing the Lemoenfontein accumulation to other bone accumulations in the Triassic strata of the Karoo Basin, as well as to bone accumulations in the Lifua Member of Tanzania and the Santa Maria Formation of Brazil. A final summary of the key insights and conclusions is provided in Chapter 9. These conclusions are presented in a way that is consistent with the main aims and objectives as set out above in this introduction. 9 Appendix A contains the full stratigraphic log from the informal Eldorado marker up to the base of the Bamboesberg Member of the Molteno Formation. Appendix B contains the database of all the tetrapod fossil specimens that were collected and considered as part of this study. Appendix C contains the basic burrow cast parameters as recorded in the field and in the laboratory. The dissertation is concluded with the full list of references in the Reference List. 10 2. MATERIALS AND METHODS 2.1 Field Based Observations The field work for this study was conducted over two ten-day fieldtrips in April 2019 and September 2020. It was mainly focused on the bone accumulation site on the farm Lemoenfontein (44) in the Xhariep (Rouxville) district on the road between Aliwal North and Bethulie. Additional bonebed localities on the farms Winnaarsbaken and Cragievar in the Joe Gqabi (Burgersdorp) district (Kitching, 1963) were surveyed for comparative purposes. A stratigraphic section was logged through the Burgersdorp Formation strata from the informal Eldorado marker exposures below the locality up to the Bamboesberg Member of the Molteno Formation. The strata at this locality are mostly horizontal and, in some cases, slightly tilted from the horizontal, dipping towards the southeast. A Jacob staff with Abney level, tape measure, a handheld Global Positioning System (GPS) device and compass were used to measure the vertical sedimentological sections and record palaeocurrent directions. Lithological textures and rock colours were noted during logging. Samples of the various silt- and mudstones, crevasse splay sandstones, crevasse channel sandstones, palaeosols and calcareous nodules were collected for study and reference. A systematic search for vertebrate and invertebrate body fossils, coprolites, burrow casts, trackways and other traces was performed during two field trips to the locality. The GPS coordinates of each in-situ fossil were logged. The lithology (mudstone, siltstone, sandstone) and other sedimentary structures of the host matrix were recorded along with its colour using a Munsell Soil Color Chart (1975 edition). Photographs of embedded fossils were taken. Following the taphonomic methods of Smith (1989), the attitude (dorsal up, ventral up, lateral up etc.), degree of articulation, pre-burial weathering or breakage and evidence of perimineralisation were recorded for each fossil. Drone-sourced aerial photographs of the locality were also taken to provide an up-to-date base map for accurate positioning of fossils and sample localities. The methods of Miller et al. (2001) and Bordy and Krummeck (2016) were used to record the following measurable parameters that characterise burrows: diameter, width/height ratio, 11 architecture, branching, penetration depth and angle, surface markings, burrow lining (mud chips), producer in burrow and occurrence density (per square meter of outcrop). Samples of the burrow casts were collected. The study area contains a distal crevasse splay sandstone with a sole surface that has preserved in minute detail the surface features of the underlying mudrock palaeosurface. A 20 m x 2 m strip of the sandstone bed was lifted in slabs to expose the casts of invertebrate and vertebrate activity as well other flow-induced features on the basal surface of this sandstone. Samples were collected, photographed and the presence of various biogenic and current induced markings were recorded. 2.2 Laboratory and collections-based data capture The Karoo fossil databases of the Evolutionary Studies Institute at the University of the Witwatersrand in Johannesburg, the Iziko South African Museum in Cape Town, as well as the National Museum in Bloemfontein were interrogated for additional records of fossils and burrow casts from the Lemoenfontein locality. These specimens were incorporated into the analyses. A selection of the fossils in these collections as well as a representative sample of new specimens recovered during fieldwork for this project were prepared using pneumatic percussion drills up to a degree of completion that allowed positive identification to species level. Preparators were assigned at the Iziko South African Museum and at an external independent preparation laboratory under the supervision of Professor Smith. To reduce the likelihood of fruitless hours of preparation with no positive identification, arrangements were made at the start of this study for several of the specimens to undergo micro-CT scanning and rendering to 3-D models. However, due to the 2020 COVID19 lockdown restrictions in South Africa, access to these facilities was not possible. Therefore, only the mechanically prepared specimens were identified to the taxonomic level of genus and species. High resolution photographs were taken under controlled conditions. 12 All fossils were allocated a taphonomic class based on the scheme devised by Behrensmeyer (1978) and Smith (1989) and as extended in this study. Using the taphonomic classes that were retained for this locality, information was derived about the depositional environment and climate during deposition of the sediments at this locality. A generic conceptual framework for the genesis and analysis of vertebrate skeletal concentrations is provided by Rogers and Kidwell (2007). This framework was utilised to categorise the bonebeds genetically and to derive insights about the biogenic mechanisms (intrinsic versus extrinsic) or the physical agents that led to the bone accumulations. Using the systematic approach for characterizing burrows developed by Miller et al. (2001) and implemented by Bordy and Krummeck (2016), the measured burrow parameters were used to allocate burrows to an ichnotaxon and to identify possible burrow producers. 2.3 Analyses The lithology of the of the rocks that make up the stratigraphic section were analysed by recording the physical characteristics such as colour, texture, grain size and composition of the collected samples. Age data were obtained by relative dating using biostratigraphic correlation of tetrapod taxa. Miall (1985, 1996) defined a systematic nomenclature for lithofacies codes and fluvial architectural elements and Hancox (1998) applied these to the Burgersdorp Formation. These codes were adapted and used to interpret the sedimentological information recorded during the logging of the vertical stratigraphic section. The data generated for the beds containing the tetrapod bone accumulations was correlated with the interpretations derived from the overall lithological and sedimentological analyses. The GPS coordinates of fossils and ichnofossils were plotted on a Google Map image that was overlaid with a high-definition drone-captured aerial image of the area. Clustering and other 13 characteristics were noted. The fossil occurrences were recorded in a database and relative frequencies were then calculated. The taphonomy of the tetrapod bone accumulation was analysed by extending the methods of Behrensmeyer (1978) and Smith (1989), with consideration given to factors such as the taxa present, degree of articulation, pre-burial weathering, breakage, or perimineralisation. Representative taphonomic styles were defined and the specimens were assigned accordingly. Using the outcome of the systematic approach for characterizing burrows (Miller et al., 2001), interpretations were derived about the social behaviour and lifestyle of the possible burrow producers. The insights gained from the lithology, sedimentology, taphonomy, fossil taxa, ichnofossils and burrow analyses were used to reconstruct the palaeoenvironment that existed during the deposition of the sediments as well as to identify the mechanisms or combinations thereof that led to the bone accumulations. The sedimentological and taphonomic data are integrated into a taphonomic pathway description. The analyses were concluded by a discussion of the palaeoclimatic and/or tectonic drivers of environmental conditions that led to the bone accumulation. A reconstruction of the once living Cynognathus AZ ecosystem at the base of the T-K Subzone at this locality was provided, along with an interpretation of the floodplain processes and tetrapod community behaviours. Comparisons of the Lemoenfontein sedimentology, taphonomy and faunal composition were made with similar-aged bone accumulation sites in the Manda Beds of Tanzania (Smith et al., 2017) and the Middle Triassic Santa Maria Formation of Brazil (Bertoni-Machado and Holz, 14 2006). Based on these, the biostratigraphic correlation and robustness of the interface of the L-G and T-K Subzones was confirmed. The process of analysis followed during this study is illustrated in Table 1.1. The columns identify the major steps taken during the study and the progression from left to right identifies the data collected and methods used to obtain the partial and final interpretations. These steps were applied for each of the main aspects of the study i.e., sedimentology, vertebrate fossils, and trace fossils. The insights from these topics combined for an integrated interpretation. 15 Table 2.1: Data collection and analysis process. Columns identify the main elements of the study and their progression from left to right delineates the data collected and methods used to obtain the partial and final interpretations. These steps were applied for each of the main aspects of the study i.e., sedimentology, vertebrate fossils, and trace fossils. The insights from these topics combine for an integrated interpretation. Aspect Field Methods Data Collection Lab Methods/Analyses Interpretative parameters Integrated Interpretation Sedimentology Measure stratigraphic section. Locate fossils on section. Photographs of fossiliferous beds. Lithology. Sedimentary structures. Pedogenesis. Colours. Palaeocurrents. Confirmation of lithology based on samples. Transcription of field records. Assignment of lithofacies codes and fluvial architectural elements. Depositional environment and timing. Derivation of conclusions about the ancient floodplain processes, depositional environment, and climate. Derivations of conclusions about the behaviour of the animals that colonized the site. Proposal of a mechanism that explains the anomalous accumulation of bones. Vertebrate fossils Systematic search and collection. Excavation. Encasement in plaster jackets. Photographs of in situ fossils. Drone based aerial photographs. GPS coordinates. Taphonomic parameters. Host matrix. Preliminary taxonomic identification. Fossil preparation. Taxonomic identification. Taphonomic modes definition. Characterisation of the bone accumulation. Museum data base search. High resolution photographs of specimens. Animal behaviour. Taphonomic pathway. Ichnofossils Systematic search and collection. Excavation. Photographs of in situ burrow casts and traces. Record characteristics (architecture, surficial morphology). Burrow cast data: Ramp angle. Architecture and morphology. Vertical and horizontal diameter. Penetration depth. Palaeosurface physical and biological traces. Measure burrow characteristics. Record surficial morphology and burrow architecture. Systematic classification. Identification of burrow-makers. High resolution photographs of specimens. Animal behaviour. Depositional environment. 16 3. SEDIMENTOLOGY AND LITHOSTRATIGRAPHY OF LEMOENFONTEIN 3.1 Overview of the lithostratigraphy of the main Karoo Basin The main Karoo Basin (MKB) of South Africa represents a near continuous sequence of sediment deposition over a period of almost 120 million years (Catuneanu et al., 2005). Up to 8 km of basin-fill sediments accumulated in a broadly east-west trending foreland basin (de Wit, 1992, Catuneanu, 2005) in southwestern Gondwana from the Late Carboniferous (300 Ma; Visser and Dukas, 1979) to the Middle Jurassic (183 Ma; Duncan et al., 1997). The Karoo sequence of sedimentary rocks has the lithostratigraphic rank of Supergroup and its outcrops (and sub-outcrops) cover two thirds of South Africa. Based on sedimentological characteristics, several groups can be defined within the Karoo Supergroup. Starting with the oldest, these are named the Dwyka, Ecca, Beaufort, Stormberg and Drakensberg groups (SACS, 1980; Catuneanu et al., 2005; Johnson et al., 2006). Karoo sediment accumulation began at the end of the Carboniferous as the super continent emerged from an extended glaciation and continued relatively uninterrupted until the beginning of the breakup of Gondwana, heralded by the vast magma flows that form the Drakensberg Group (Duncan et al., 1997). The Main Karoo Basin is the largest and deepest of several connected or partially connected basins across southern Gondwana (Hancox, 1998). The thickest deposits in the main Karoo Basin occur in the southeast, proximal to the Cape Fold Belt where they reach a maximum of 5 km (Scheiber-Enslin et al., 2015). Most of the formations thin significantly to the distal northeast (Rutherford et al., 2015), however reciprocal flexure of the crust has created some areas of overthickening and removal/non-deposition in the central and northern areas (Bordy et al., 2004; Catuneanu et al., 2005). In the southern Karoo Basin, the Dwyka Group (320-280 Ma) glaciogenic deposits accumulated subaqueously mainly from floating ice and outwash from retreating glaciers. The confined post-glacial meltwaters resulted in Ecca Group marine and deltaic strata that conformably overlie the Dwyka Group. These are followed by the non-marine Beaufort Group and after a five-to-ten-million-year hiatus, the Stormberg Group. 17 The Beaufort Group is divided into a lower Adelaide Subgroup (Middle-Late Permian Koonap- Middleton, and Balfour formations) and upper Tarkastad Subgroup (Early to Middle Triassic Katberg and Burgersdorp formations) (SACS, 1980; Catuneanu et al., 2005). These alluvial successions are made up of stacked fining-upward cycles of various scales and complexity but with similar sandstone, siltstone, and mudstone lithologies. (Catuneanu et al., 2005). The Tarkastad Subgroup overall has a greater sandstone-to-mudstone ratio than the Adelaide Subgroup (Johnson, 1976, page 243). The Stormberg Group overlies the Beaufort Group. It is made up of the Late Triassic Molteno formation, the Triassic-Jurassic Elliot Formation and the Jurassic Clarens Formation. There is a stratigraphic hiatus between the Burgersdorp Formation and the Molteno Formation (Hancox 1998, Catuneanu et al., 2005) as well as at the lower and upper contacts of the Lower Elliot Formation (Bordy et al., 2004). The Molteno Formation consists of the upward-coarsening Bamboesberg and Indwe members (Hancox, 1998) and the informally named Tsqima succession (Bordy et al., 2005). The Beaufort Group fill of the Karoo Basin was controlled by pulsed orogenesis of the Cape Fold Belt (CFB) resulting from subduction of the Palaeo-Pacific plate beneath the Gondwana plate (Hancox, 1998; Catuneanu et al., 1998; Catuneanu et al., 2005). This tectonic activity created the reciprocal flexural profile of the Karoo Basin with a deep retro-foreland basin proximal to the orogenic load and a peripheral bulge distal to the load (Catuneanu et al., 1997; Catuneanu et al., 1998). During a loading event the proximal foreland undergoes subsidence, and the peripheral bulge undergoes uplift. This creates accommodation space in the foreland, which in turn leads to retrogradation of coarse sediments. During the unloading that follows the opposite occurs, i.e., uplift of the proximal foreland and subsistence of the distal foreland, leading to a progradation of coarse sediments. The response of coarse clastic deposition is therefore 180 degrees out of phase with the tectonic pulse events (Hancox 1998). Hancox (1998) was able to correlate the deposition of the arenaceous Katberg and Molteno formations to periods of orogenic unloading that resulted in progradation of coarse clastic wedges into the basin. Similarly, the deposition of the predominantly argillaceous Balfour and 18 Burgersdorp formations was correlated to orogenic loading tectonism leading to retrograde deposition of coarse sediments confined to the southern margin of the basin. Thus, the deposition of Triassic sediments in the Karoo retro-foreland basin was controlled by periods of limited or no tectonic activity during which channel-dominated fluvial systems prograded (basinward), followed by periods of pulsed tectonism in the CFB, creating accommodation space and leading to sourceward retrogradation of the channel-dominated systems. This model for deposition control is out of phase with the previous sedimentary model of Hancox and Rubidge (1998) and the tectonostratigraphic model of Groenewald (1996). The Burgersdorp Formation is the final depositional episode of the Beaufort Group. As described above, it is para- to un-conformably overlain by the Molteno Formation of the Stormberg Group. It is of Early to Middle Triassic age with an overall thickness of up to 1000 metres in its southern parts (Johnson et al., 2006) and consists of mudrocks with inter- bedded sandstones in upward-fining cycles of a few meters to tens of meters thick. An extensive study of the Burgersdorp Formation is reported in Hancox (1998). Other geological descriptions are provided by Karpeta and Johnson (1979), Dingle et al. (1983), Johnson (1976, 1984), Hiller and Stavrakis (1984), Johnson and Hiller (1990), Kitching (1995) and Hancox (2000), Neveling et al. (2005), and Rutherford et al. (2015). Most of the Burgersdorp Formation was deposited by northwesterly flowing meandering rivers during a warm, arid to semi-arid climatic interval (Hancox, 1998). These rivers formed a high sinuosity, mixed-load fluvial system. Suspension-load dominated deposition occurred on the floodplains during and following overbank flood events. Some sheetwash, shallow- ephemeral and bedload-dominated sand sheets deposited by low-sinuosity and possibly braided rivers are also present, particularly in the lower part of the Burgersdorp Formation (Hancox, 1998). This depositional scenario resulted in a succession of isolated, lenticular, feldspathic channel sandstone bodies, common tabular crevasse splay sandstone sheets and thick tabular beds of pedogenically-modified overbank mudrocks. The channel sandstone bodies invariably overlay an eroded surface commonly marked with patches of intraformational basal lag conglomerate consisting of reworked mudclasts and calcareous nodules. The thick beds of massive greyish-red (5R 4/2) to dusky-red (5R 3/4) overbank fines 19 consist mainly of siltstones and mudstones and contain some sand-filled mudcracks, vertebrate and invertebrate burrows and vertebrate body fossils. Playa-lake deposits are present in the form of well-laminated greyish-red (10R 4/2) mudrocks with pedocrete (cemented soil) horizons. Lacustrine laminated mudrocks also occur in the lower Burgersdorp Formation in the northern part of the Karoo Basin (Groenewald, 1996; Hancox et al., 2010). 3.2 Lemoenfontein fossil site: geological setting The farm Lemoenfontein (44) is situated in the Southern Free State, 14 km to the northwest of the town of Aliwal North on the road to Bethulie. It is one of several farms (Gladde Grond, Betjies Kraal, Blom and Mooiplaas) that lie in a wide semi-circular valley that opens towards the southwest with the Orange River forming the southern boundary (Figure 3.1). The northern, eastern, and western flanks of the valley are formed predominantly by mudrock exposures, up to 160 m in height, of the Early (Scythian) to Middle (Anisian) Triassic Burgersdorp Formation. These exposures correspond biostratigraphically to the Trirachodon- Kannemeyeria (T-K) Subzone of the Cynognathus AZ, (Smith et al., 2020). The mudrock exposures are capped at 1479 m in the north and northwest by the prominent Bamboesberg Member sandstone that forms the lower part of the Molteno Formation (Hancox 1998, pp. 158-159). The mid-valley elevation is at 1303 m. The valley is bisected by a large ephemeral erosion “sluit” or “spruit” that runs from north to south towards the Orange River. Water from the catchment area in the north flows to the Orange River via this spruit resulting in numerous shallow gullies eroded into fossil-bearing mudrocks. The bonebed under investigation occurs in dark reddish-brown (5YR 4/3) mudrock exposures around and to the north of an isolated hill on the farm Lemoenfontein. This hill is situated to the west of the erosion spruit and southeast of the main homestead on the farm (Figure 3.1). The Lemoenfontein hill forms the southernmost and lowermost of eight stepped plateaus that extend into the valley from the base of “Vaalkop”, the prominent hill on the highest of these plateaus that is capped by a sandstone that forms the base of the Indwe Member (middle Molteno Formation). Stratigraphically these steps extend from the Bamboesberg Member (basal Molteno Formation) in the north down towards the south on the western 20 side of the spruit (see Figures 3.1 and 3.2) to some 25 m above the informally named Eldorado marker (Neveling, 2004) or “middle marker” (Hancox, 1998). The plateaus, including Lemoenfontein hill are topped by erosion- resistant sandstones with the more easily- eroded mudrocks forming slopes that steepen with increased elevation up to the Bamboesberg Member. The fossil-rich Lemoenfontein hill is topped by a 2.8 m thick multi-storied sandstone that has been erosively separated from the sandstone body that forms the lowest of the stepped plateaus, and it is now linked by a mudrock-dominated saddle exposure. The sandstone capping the lowest plateau is a 1 m thick tabular sandstone and it is likely to have originally been connected with the sandstone capping the Lemoenfontein hill. The basal surface of this sandstone corresponds stratigraphically (and in elevation) with the top surface of the sandstone that caps the hill. The interlinking saddle contains the main fossil-bearing dark reddish-brown (5YR 4/3) mudstone exposures, and there are parallel erosion gullies (depth of approximately 1 m – 1.5 m) on both east and west sides of the saddle. Similar erosion gullies exposing fossil-bearing reddish-brown (5YR 4/3) mudstones occur on the eastern flank of the main body of the ridge. The exposed mudrock slopes and surfaces are densely covered by a float of small to medium sized (3 to 15 cm) smooth-surfaced brown-weathering, ellipsoidal, calcareous nodules, and irregularly shaped sandstone slabs. A sedimentological log of approximately 175 metres was measured through eight 3rd-order fining-upward cycles (Hancox, 1998; Miall, 2014) from the bed of the spruit up to the top of the first prominent sandstone of the Bamboesberg Member (Figure 3.2). Standard field methods, including Jacob’s staff and sight level as well as a Munsell geological rock-colour chart (2009 revision) were used to record a stratigraphic log to an accuracy of 5 cm (see Appendix A for the full log). The sedimentary rocks of this section are typical of the T-K Subzone of the Cynognathus AZ and consist mainly of massive or horizontally stratified mudrocks with inter-bedded sandstones in cycles of a few meters to tens of metres in thickness. The stratigraphic positions of all in-situ vertebrate fossils, burrow casts, fossil- bearing conglomerates and coprolites were mapped onto the log. 21 Figure 3.1: Geographic location of the Lemoenfontein bone accumulation site. 22 6 2 3 4 5 7 8 -12 m to 0 m 0 m to 25 m 25 m to 163 m Section Bamboesberg Member Lemoenfontein hill Eldorado marker Fossiliferous beds 6 1 Indicates base of third order cycle Crevasse splay surface Figure 3.2: Google Earth Image of the locality and the measured section. The Lemoenfontein hill is in the foreground. This section crosses eight ledge forming sandstones each of which represents the base of a fining-upward third order cycle. The dashed lines show the path of the measured section from the channel of the spruit to the top of the hill and its extension from the second ledge forming sandstone (the stratigraphic equivalent of the top of the sandstone cap on the hill) up to the Bamboesberg Member. The final one being the Bamboesberg Member. The beds containing the bone accumulation are indicated in light yellow. Note that although the green portion of the section is following the downstream direction in the spruit, it is traversing upwards strategraphically due to the dip of the beds towards the southeast. 23 3.3 Basis for facies sequence description The lithofacies codes, overbank facies associations and channel facies associations used to characterize the Lemoenfontein beds are provided in Table 3.1, Table 3.2, and Table 3.3, respectively. Table 3.1: Lithofacies Classification Facies Code Lithofacies Sedimentary Structure Description* Sp Sandstone Horizontal, parallel, planar stratification Parallel and non-inclined (upper-phase plane bed high velocity laminar flow). Sm Sandstone Massive Crevasse splay. St Sandstone Trough cross- stratification Sets of arcuate foresets with curved bounding surface resulting from, downstream migration of 3-D bedforms over an uneven channel floor with scour depressions under moderate to high energy turbulent flow. Ss Sandstone Scoured surface Channel bottom, erosional bounding surface, commonly elongate depressions parallel to flow overlain with mud clast gravel lags resulting from high energy turbulent flow in channel thalweg. Sl Sandstone Lenticular Floodplain depressions and distributary channels. Sr Sandstone Ripple cross- stratification Moderate turbulent flow over rough sandy beds form downstream migrating ripples. With abundant sediment supply the migration of one ripple on the stoss side of another results in climbing ripple cross- stratification. Spcb Sandstone Planar cross- stratification Parallel, inclined planar foresets dipping in a single direction resulting from downstream migration of straight- crested bar forms on a flat surface. Sei Sandstone Massive to cross- stratified, fine to medium grained sandstone, containing intra formational clasts. Clasts are predominantly mud- and siltstone, with varying amounts of intra-formational debris including fossil bone, coprolites and fragmented and particulate plant material. This facies is further characterised by the total absence of extra formational clasts. (Hancox, 1998). SSp Siltstone Horizontal, parallel, planar stratification and lamination. Parallel and non-inclined thin planar beds resulting from sedimentation of suspended silt during waning phase of unconfined overbank flows. Mp Mudstone Horizontal, parallel, planar stratification and lamination Parallel and non-inclined laminated thin beds that result from suspension settling of fine-size sediment or precipitation from solution, low energy-slack-water pools, ponds and lakes. Mm Mudstone or Claystone Massive Contain few or no visible internal lamina, overbank or abandoned channel deposits Cc Calcium carbonate - limestone Nodules Pedogenic or early diagenetic in origin * Adapted for the Burgersdorp Formation from Li and Zhang (2017). 24 Table 3.2: Overbank facies associations (adapted from Colombera et al., 2013) Overbank Facies Associations Notation Key Characteristic Corresponding Subenvironment Levee LV Levee elements typically take the form of tapering wedges that thin away from channel-belt margins and are slightly elevated above the rest of the floodplain; their base may be poorly defined, and internal accretion surfaces may offlap and/or downlap, showing dip angles up to 10°, although 2 to 5° are more common, associated with the sloping topography bordering channels; (palaeo) flow direction is usually oriented at high angle with the channel border. They have distinctive internal erosion surfaces. Although levees may develop at smaller scales (for example, crevasse channel levees), LV elements usually represent the sedimentary and geomorphic expression of the most proximal over-bank deposition next to channel-belt margins. Smith (1990) distinguished inner- from outer -bank levees based on their facies sequence and the fact that outer-bank levees have more deeply incised scour surfaces Crevasse splay CS These elements are tongue-shaped sandstone bodies bordering channel- belt margins. These bodies thin away from the channel margins, as they interfinger or grade laterally into other elements, and they tend to have flat, sharp, and slightly erosive bases although tabular stratification is common, internal accretion surfaces usually downlap, dipping at low angle to angle of repose, as they record the progradation of the splay onto the floodplain or into standing bodies of water. Upper surfaces are sharp and scoured by runoff channels in the proximal part and more interdigitating with mudrock distally. These elements represent the sedimentary and geomorphic product of splay progradation and aggradation through the periodic unconfined flow from crevasse channels tapping discharge from the channel-belts during floods. Distal overbank fines/ floodplain lake DF Interbedded thinning upwards fine sandstone-siltstone cycles and horizontally stratified fine-grained sandstones and contains thin siltstone lenses. These elements are interpreted to be distal floodplain deposits, typically around the margins of distal floodplain lakes. Proximal overbank fines PF These elements consist of tabular or prismatic siltstone bodies in which laterally persistent depositional increments tend to be vertically stacked and bounded by planar surfaces, demonstrating an overall aggradational character. Pedogenic alteration is common. Isolated calcareous nodules and nodular horizons are common. The presence of slickensides in the siltstone results in a crumbly weathering pattern. Well-developed B- horizons are present (Smith, 1990). These elements are the sedimentary expression of vertically aggrading floodbasins, in which traction and suspension settling from episodic subaerial unconfined flows with a high sediment load is the dominant process; bedload deposition of mud- aggregates on the floodplain can also produce fine-grained floodplain units (Rust and Nanson, 1989). 25 Table 3.3: Channel Facies Associations (adapted from Colombera et al., 2013) Channel Facies Associations Notation Key Characteristic Corresponding Subenvironment Vertically aggraded channel-fill CHv These elements are characterized by downstream-elongated incisional concave-upward bases, on which depositional increments are overall vertically-stacked, either concentrically or onlapping the channel margin, resulting in the dominantly horizontal orientation of planar, undulating or scour-like internal second and third order bounding surfaces. Although CH elements may be locally composed of small-scale downstream, oblique, lateral or upstream-accretion increments, they significantly lack the laterally persistent inclined accretion foresets that is typical of barform deposits. These elements generally represent the overall aggradational infill of active low sinuosity and anastomosing channels. Lateral accretion barform CHl These elements are characterized by sharp, sub-horizontal to slightly concave-upward, and often erosional bases, on which depositional increments are laterally stacked, with dip direction at high angle with respect to the palaeoflow direction, and dip angle up to 25°, generally showing off- lapped upper terminations and interdigitations with mudrocks of the levee facies. These elements represent the infill of active channel belts by inner bank attached, laterally expanding and downstream migrating bars, most commonly typified by meander point bars. Downstream accretion barform DA These elements are characterized by sub-horizontal to slightly concave- upward and often erosional bases, on which depositional increments are stacked at low angle with respect to palaeoflow, determining a dominance of low-angle downstream-dipping second and third-order bounding surfaces. DA elements may be locally composed of oblique, lateral, or upstream accretion increments, but the overall preponderance of downstream growth is their key character. These elements represent the infill of active channel-belts by downstream-migrating bars typical of braided channel systems. Chute channel fill HO This element type encompasses major scour-hollow fills, characterized by incisional concave-upward scoop-shaped bases, and by infill through accretion on inclined or horizontal surfaces or on a combination of both These elements represent the infill of deeply incised trough shaped scours within channel belts, for example by the migration of mouth- bars into deep confluence scours or by infilling of flood-related scours during waning-flood stage. This is typical of an abandoned channel-fill common in meander cut-offs (ox bow lakes) and chute channels. 26 The description of the sedimentary rocks in the Lemoenfontein succession is based on the hierarchy of sedimentological cycles on the fluvial floodplain as defined by Miall (2014) as a framework. These cycles can briefly be summarised. • First order cycles are formed by allogenic causes and typically represent the entire succession of gradual upward-fining deposition following an initial orogenic up-lift or unloading event. In the context of this study an example of a first order cycle is represented by the the deposition of Triassic sediments in the Karoo retro-foreland basin during periods of limited or no tectonic activity when the facies prograded basinward and resulted in the succession from the base of the Katberg Formation to the base of the Bamboesberg member of the Molteno Formation. • Second order allogenic cycles reflect isostatic adjustments or major climate change events. Isostatic adjustments can occur due to major erosion and deposition events that leads to secondary loading or unloading in different parts of the floodplain. The paraconformable contact between the Burgersdorp and Molteno formations as evidenced on Lemoenfontein is a result of such a second order erosional cycle. • Third order cycles are autogenetic to the fluvial system and refer to rapidly changing events that occur when geomorphic thresholds are exceeded after periods of metastable equilibrium. Such events are represented in the Burgersdorp Formation sedimentary rocks by the arrival, migration or abandonment of fluvial channels or avulsions that lead to rapid truncation of a particular mode of deposition on the floodplain. The third order fining upward cycles at Lemoenfontein will be used as the main framework for this facies description. • Fourth order cycles are caused by periodic erosion or other detailed responses of the fluvial system to the changes brought about by the first to third order cycles. For example, these are represented by the various erosional surfaces in the rocks on Lemoenfontein. • Fifth order cycles relate to seasonal variations such as rainfall or temperature and encompass other significant non-seasonal hydrological events such as major floods that occur significant time periods apart. The consequence of seasonality is present on Lemoenfontein in the form of evidence for seasonal variation of the groundwater table, rapid sediment deposition or long-term floodplain depositional stasis. 27 Interpretation of the palaeoenvironments and sub-environments is made for each cycle through the logged section. Specifically, lithofacies codes (Miall, 1985, Li and Zhang, 2017) are assigned to the third order sedimentary strata in the section. Using this lithofacies classification and the guidance provided in Smith (1990), the depositional facies are assigned to four facies associations (architectural elements as per Miall (1985)). • Channel (CH) – Typical features are sandstone bodies with concave-up erosional bases, comprising variety of stacked bar complexes, single or multistoried accretion units, fingers, or lenses. Vertebrate fossils are rare and typically confined to basal lag conglomerates. • Levee/Channel-Bank (CB) – Typically consists of rapidly alternating fine sandstone and siltstone beds overlying simple or multi-storied channel sandstones. Small scale scour surfaces overlain with lenticular sandstones frequently occur. Thin mudstone veneers cap siltstone and sandstone beds and record features such as rippled surfaces, run-off rills or sand-filled desiccation cracks. Fossils are common but are usually disarticulated single elements such as skulls or lower jaws. • Proximal Floodbasin (PF) – Accumulated from the base of the meanderbelt slope on level areas of the floodplain. Consists of massive siltstones and minor mudstones that are horizontally or climbing ripple cross-stratified. Are frequently interrupted by thick (>1 m) crevasse splay sandstones with sharp non-erosional bases. Mudstone veneers may occur with run-off rills and sand-filled desiccation cracks. Isolated calcareous nodules and nodular horizons are common. The presence of slickensides in the siltstone results in a crumbly weather pattern. Well-developed B-horizons are present. Fossil skulls with and without articulated lower jaws are common and articulated skeletons within burrows can occur. Carbonate precipitation around fossil bones is common. • Distal Floodbasin (DF) – These low accretion rate deposits occur in depressions far away from the main channel. These deposits consist mainly of thinly-bedded mudstones interbedded with thin siltstone and sandstone beds. Desiccation cracks are common, but calcareous nodules are rare. Vertebrate fossils are rare and consist of small, isolated elements. Soil formation is usually inhibited by a water table that is near the surface (Smith, 1990). 28 A similar definition is provided by Zwoliński (1992). Four distinct floodplain zones are introduced. Zone 1 is immediately adjacent to the river channel and shows the most varied lithology and morphology. Zone 2 consists mainly of clays and muds and covers most of the proximal side of the floodplain. Zone 3 corresponds to the distal part and contains very little depositional material. Zone 4 corresponds to higher valley walls or higher floodplain surfaces that only get inundated in the most severe of overbank floods. Also overbank flood events on a lowland meandering river floodplain are categorized into 6 distinct phases as follows: • Phase 1 – Rising of water and bank modification stage; • Phase 2 – Floodplain inundation and initial deposition; • Phase 3 – Flood peaks and widespread transport and deposition; • Phase 4 – Falling of water stages and high intensity deposition; • Phase 5 – Cessation of overbank flow and final deposition; and • Phase 6 – Post-flood transformation of overbank forms and deposits. These definitions are very useful for the interpretation of the sedimentary rocks in the Lemoenfontein stratigraphic section, particularly for the detailed description of Cycles 0 and 1. 29 3.4 Description of the Lemoenfontein facies sequences The detailed description of the sedimentary rocks in the Lemoenfontein succession that follows is based on the framework described above. The interpretation of the Lemoenfontein palaeoenvironments and sub-environments derived from this description is schematically presented in Figure 3.3 for the entire 175 m of the stratigraphic log. Cycle 0 The entire Cycle 0 is exposed in the stream bed and banks of the Lemoenfontein spruit. The base of the incomplete Cycle 0 is covered in river sand and the section extends from the - 12 m to - 5 m in the measured section. The beds that make up this cycle are tilted due to a dolerite dyke intrusion. This dip is 9° with a strike of 115° (southeast). Therefore, to log this part of the stratigraphic section, measurements were taken in the downstream direction of the spruit (green dashed line in Figure 3.2). The lower 1.5 m consists of massively bedded siltstone. The light-grey (5Y 7/1) bottom half of this siltstone consists of 12 to 20 cm thick beds with clay veneered surfaces. In the pale-olive (5Y6/3) top half these surfaces become irregular and hummocky with a pustular matted texture. Continuing up section a 1 m thick light-grey (5Y 7/1) blocky-weathering fine siltstone contains a single reniform burrow cast followed by 50 cm of pale-olive (5Y 6/3) blocky-weathering coarse siltstone that contains one subcircular burrow cast that extends into the fine siltstone below. This is overlain with an upward-fining light-grey (5Y 7/1) siltstone (1.4 m thick) with two horizons of elongated prismatic peds 15 cm apart and an 80 cm thick reddish-brown (5YR 4/3) massive mudstone with rootlets and isolated calcareous nodules that are approximately 4 cm in diameter. Its top surface is bioturbated by numerous vertical Skolithos burrows, 5 to 9 mm in diameter. Directly above this mudstone is a strongly developed 20 cm thick, mottled silty mudstone bed consisting of elongated prismatic peds overlain by 1 m thick massively bedded reddish-brown (5YR 4/3) silty mudstone. The massive mudstone features abundant near vertical striated slip surfaces or slickensides that are randomly orientated. The beds containing the nodules and peds have undergone pedogenesis and are like those in Cycle 1 that host the tetrapod body fossils and burrow casts. An interpretation of these pedogenic features is provided in Section 3.4 below. The evidence of pedogenesis and seasonal water table variation (peds, 30 slickensides, calcareous nodules) leads to the interpretation that these mudrocks were deposited in a proximal overbank floodplain setting (PF facies association). Cycle 1 Cycle 1 extends from the -5 m level up to the base of the second major sandstone ledge at 23.5 m (yellow dashed line in Figure 3.2). The base of Cycle 1 consists of a fine-grained, light- grey (5Y 7/1), multi-storied sandstone that extends from -4.4 m to 0 m. The first storey is 40 cm thick and horizontally (parallel) stratified. This storey overlies a pale-olive (5Y 6/3), 30 cm thick siltstone with a ped horizon at its base. The second 1 m thick storey is trough cross- stratified. The large (8 m wide, 50 cm deep) elongated trough cross-beds that occur lower down in this storey reduce in size near its top (4 m wide, 25 cm deep). Some horizontal plant stem compressions are also preserved on this surface. The third storey is 1.9 m thick. Its first 50 cm contains thin horizontal stratification and laminae. This is followed by thicker stratification and laminae in the next 25 cm. The top surface of this bed is bioturbated by Skolithos. This is followed by another 75 cm and 40 cm thick horizontally and thinly bedded elements topped by relatively large wavelength, low amplitude (“washed-out”) mega ripples. The final storey is 1.23 m thick and is also horizontally stratified. This multi-storied sandstone (-4.4 m to 0 m) is interpreted to be the Eldorado marker of Neveling (2004). The justification of this interpretation and the biostratigraphic significance of this marker is explained Chapters 4 and 7. This sandstone complex is interpreted as the result of high energy in- channel sedimentation in a low- to medium-sinuosity river, and is assigned to the CH facies association. A 60 cm pale-olive (5Y 6/3) siltstone bed showing blocky weathering occurs directly on top of the Eldorado marker. This is followed by a 1.2 m thick light-grey (5Y 7/1) sandy-siltstone. This bed hosts a large bifurcating reniform burrow cast (horizontal diameter = 16 cm and vertical diameter 11 cm). Above this bed is a 50 cm thick pale-olive (5Y 6/3) mudstone bed that contains slickensides. It is initially massively bedded but incipient horizontal stratification occurs towards its top. A 25 cm thick horizontally stratified light-grey (5Y 7/1) fine sandstone follows, which in turn is overlain with a 40 cm thick pale-olive (5Y 6/3) siltstone. Above this 31 siltstone is a 20 cm thick sandstone with an erosional base with runnels. This is followed by two sets of 60 cm thick siltstone/sandstone couples. The rest of the Cycle 1 sedimentary succession is dominated by tabular massive dull reddish- brown (2.5YR 5/4) and greyish-red (5R 4/2) mudstone beds between 20 – 40 cm thick interbedded with a single yellowish-grey (5Y 7/2) 10 cm thick fine-grained sandstone of (interpreted to be a crevasse splay sandstone, CS, at 13 m of Figure 3.3). The mudstone beds display evidence of pedogenesis suggesting seasonal ground water level fluctuations. This is interpreted from the structure of the mature textural B-horizon, two well-developed calcareous nodular horizons, slickensides and cutan formation. Sandstone-filled desiccation cracks penetrating the massive mudstone are present at 9.5 m and 13 m. It is interpreted that the stacked mudstone beds were deposited rapidly in overbank traction and suspended load deposition from sediment-laden overbank floods, but the mature B-horizon and calcareous nodular horizons point to long periods of floodplain stasis with little or no sediment accumulation. The mudstones are very rich in tetrapod fossils. These fossils are predominantly articulated skulls with skeletons and isolated skulls with articulated lower jaws and are frequently coated in thin to thick calcareous skins. The mudstones also host tetrapod and other burrow casts. Based on the massive nature of the mudrocks, these are assigned the lithofacies code Mm. The presence of a crevasse splay sandstone, the evidence of pedogenesis and seasonal water table variation, and the other factors mentioned above, leads to the interpretation that these mudrocks have been deposited in a proximal overbank floodplain setting (PF facies association). Cycle 1 is the host of the bone accumulation that forms the focus of this study and is therefore discussed in detail in Section 3.4 where fourth and fifth order cycles are also considered. These mudstone beds are anomalously thick. To account for this, a possible mud aggregate depositional mechanism is explored further in the detailed description of Cycle 1 in Section 3.4. 32 Cycle 2 Cycle 2 extends from the base of the fine-grained horizontally (parallel) stratified sandstone at 23.5 m up to the base of the sandstone that forms the third significant ledge at 38 m. (Figure 3.2). This yellowish-grey (5Y 7/2) channel sandstone has an erosional base (with runnels) and consists of two storeys. There is a thin clay pebble conglomerate at the contact between the two storeys. The top surfaces show parting lineation. The uppermost storey contains large concave-up erosional channel surfaces which are filled with horizontally- stratified sandstone (lithofacies code Sp, see Figure 3.3). This sandstone complex is interpreted as the result of high energy in-channel sedimentation in a low- to medium- sinuosity river and assigned the CH facies association. A 1 m thick, horizontally-stratified, fine-grained olive-grey (5Y 3/2) tabular sandstone (lithofacies code Sp) follows directly above the stratigraphic level of the channel sandstone. This interpreted crevasse splay sandstone has runnels and current crescents at its generally sharp base and Skolithos on its upper surface. By its nature, this sandstone is proximal overbank and assigned to the PF facies association. The crevasse splay sandstone is followed by a 10 m thick interval of horizontally stratified dusky-red (5R 3/4) to olive-grey (5Y 3/2) mudstone beds (lithofacies code Mp). The lower 2 m hosts isolated oblate and irregular-shaped calcareous nodules. Due to its vertical superposition on a crevasse splay sandstone, as well as the evidence for pedogenesis, this mudstone interval is interpreted to be deposited in a proximal overbank environment and assigned PF facies association. No vertebrate fossils were found in Cycle 2. Cycle 3 Cycle 3 extends from 38 m to about 55 m up to the base of the sandstone assembly that form the fourth prominent ledge in the succession. The base of Cycle 3 consists of a 2.8 m thick, olive-coloured (5Y 5/4), fine-grained sandstone that is multi-storied. The different storeys have different structures and textures; from the bottom up – mudstone flakes, a prominent 33 concave-up erosional channel surface which is filled with planar stratified sandstone, trough cross-stratification, diagenetic concretions, erosional contact surfaces, rib and furrow structures with a final transition from horizontal to massive stratification near its top (assigned to the lithofacies codes St, Sp, Sm and Ss). This sandstone is interpreted to be part of a medium- to high-sinuosity channel facies assemblage (CH). This channel assemblage is followed by a 14 m thick mudrock interval that is dominated by olive-coloured (5Y 5/4) mudstone beds. Interbedded in this mudstone units are two crevasse splay sandstones (one planar- and one massively-bedded) and a single siltstone bed showing blocky weathering (lithofacies codes Sp, Mp, SSp and Sm.) The mudstone is thickly-bedded (10 - 30 cm thick tabular units) and contains features such as slickensides, colour mottling and calcified fissures. This mudstone interval is interpreted to have