1 Early humans recipe for processing foods at Blombos Cave, 2 South Africa (85-82 ka) 3 4 Supplementary Information 5

6 Cynthia Larbey1†, Karen L. van Niekerk2, Christopher S. Henshilwood2,3, Martin, K.

7 Jones1 8

9 1 McDonald Institute for Archaeological Research, University of Cambridge, U.K.

10 2 SFF Centre for Early Sapiens Behaviour (SapienCE), University of Bergen, Norway

11 3 Evolutionary Studies Institute, University of the Witwatersrand, Johannesburg, South Africa 12 13 † Corresponding author [email protected] https://orcid.org/0000-0002-9632-2746 14 15 Table of Contents 16 17 S.1 The Blombos site – Introduction 2 18 S.2 Combustion events at Blombos Cave 3 19 S.3 Sampling 4 20 S.4 The flotation experiments 6 21 S.6 Modern reference collection 11 22 S.7 Summary of results 15 23 References for Supplementary Information 16 24 25 Table of Figures 26 27 Figure S.1: Blombos Cave site plan ...... 2 28 Figure S.2: The south section before excavation 2013 ...... 3 29 Figure S.3: The south section of Blombos Cave, with phasing and OSL ...... 3 30 Figure S.4: Two combustion events in section at Blombos ...... 4 31 Figure S.5: Sample locations, south section, Blombos Cave ...... 6 32 Figure S.6: Flotation Experiment 1 ...... 7 33 Figure S.7: Flotation Experiment 2 ...... 8 34 Figure S.8: Plants around Blombos Cave ...... 13 35 36 Table of Tables 37 38 Table S.1: Samples, contexts, dates and processing ...... 5 39 Table S.2: Modern root and tuber samples collected from around Blombos Cave ...... 14 40 Table S.3: Botanical Results from Blombos Cave ...... 15 41 42 43

1 44 45 S.1 The Blombos site – Introduction 46 47 The fieldwork at the MSA site of Blombos Cave was conducted in 48 November/December 2013 (Spring in South Africa). During that season, excavation 49 focussed on the south section (Figure S.1, marked by red box) and in photo at Figure 50 S.2. The following describes the fieldwork collection and laboratory analysis 51 processes. 52 53

54 55 Figure S.1: Blombos Cave site plan showing excavated areas and south section excavated in 2013 – outlined with 56 red box (Image adapted from original by Magnus Haaland 1)

2 57 58 Figure S.2: The south section before excavation 2013 (Image: Magnus Haaland) 59 60 Figure S.3 shows the phasing and dating of the sequence of the south section at 61 Blombos Cave. 62

63 64 Figure S.3: The south section of Blombos Cave, with phasing and OSL (optically stimulated luminescence) (Image: 65 Magnus Haaland 1) 66 67 S.2 Combustion Events at Blombos Cave 68

3 69 To identify parenchyma samples as food, it was the aim to take botanical samples, 70 where possible, through hearths. No intact hearths were identified at Blombos but 71 remnant combustion features were either identified during excavation (see Table S.1) 72 or in section (Figure S.4). 73

74 75 Figure S.4: Two combustion events in section at Blombos (both at the base of CGAC 76 level – 85-90 kya), Blombos Cave, South Africa (Images: C. Larbey) 77 78 The combustion event in the left image of Figure S.4 exhibits lenticular shaping; this

79 shaping may be associated with hearth cleaning 2,3. Given the location of these 80 combustions events, cleaning activity or trampling would seem likely in hearths close 81 to the mouth of the cave. 82 83 S.3 Sampling 84 85 The unique nature of the Blombos site and its specific preservation conditions were 86 established in an experiment with flotation of archaeological charcoal versus modern 87 charcoal. This experiment is detailed below. Flotation is the common method for 88 collecting charred botanical samples, such as wood charcoal and parenchyma. The 89 flotation process involves depositing sediment from archaeological contexts into water

90 and collecting charred fragments as they float to the top 4. Dry sieving can also be 91 used and is successful in recovering desiccated samples, but charred archaeological 92 botanical samples can be fragile and damaged/destroyed when subject to mechanical

93 stress 4. As with many plant fragments, parenchyma is often preserved because it is 94 burned and its cell structure survives as carbon. However, starchy tubers with 95 parenchyma tissue preserved have also been found in waterlogged deposits, where the 96 oxygen is exluded (and thereby all the microbes that would normally breakdown

4 97 starch) such as at the Bronze Age size of Must Farm in the UK 5. Plant remains 98 including parenchyma with the cell structure in tact have been found when desiccated

99 slowly such as at Mons Claudianus 6. 100 101 The combustion events identified during excavation at contexts CCC F7b (74.6±3.9) 102 and CFD G7a 76.7±4.8 (see Table S.1 and Figure S.3) were sampled for flotation. 103 Small block botanical samples were floated and despite visible charred remains when 104 dry, no charred fragments floated to the surface. This is because of the level of 105 phosphatisation of charred fragments from the cave that were also identified in the 106 micromorphology micrographs. The remaining combustion events were identified in 107 section and block samples were taken through the section. These samples were dried 108 and micro-excavated in the laboratory. 109 110 Table S.1: Samples, contexts, dates and processing * = those samples taken from what appeared to be a combustion 111 event in the section but yielded little/no evidence Context OSL Date Sampling Context Processing Method Type (kya) Method Off-hearth CFA F7b 70±4 Micro-excavated sample Cut from Combustion Flotation CCC F7b 74.6±3.9 quadrants event experiment as putative 68.8±4.6 - CFB-CFC * Micro-excavated hearths 75.5±5 identified Combustion Flotation CFD G7a 76.7±4.8 during event experiment excavation Combustion 76.7±4.8 - CF h Micro-excavated event 78.8±5.6 Combustion CGAB h1 81±4 Micro-excavated event CGAC Combustion 85±6 Micro-excavated Cut from G7b event section CGAC * 85±6 Micro-excavated H7a CGAC I6c * 85±6 Micro-excavated

5 112

0 0.5m

113 114 Figure S.5: Sample locations, south section, Blombos Cave 115 116 S.4 The flotation experiments 117 118 Trowelled sediment at Blombos is dry sieved in a 4mm mesh and very little charcoal is 119 recovered in this process. During the excavation at Blombos, the charcoal in the 120 sediment was moist, giving it a buttery texture that smeared when the trowel scrapped 121 across it. The charcoal fragments in this moist condition were also extremely fragile to 122 handle. Two experiments were run to test whether flotation would be a viable method 123 for the collection of botanical remains. Samples were collected from a hearth that 124 became apparent during excavation and these were used for the experiment. 125 126 Flotation experiment 1 127 128 The Objective: to float sediment known to contain charcoal fragments and to observe 129 the outcome. 130

6 131 The Method: A 250 ml block sample of archaeological hearth sediment in which 132 charcoal fragments were visible (Figure S.6 - left) was floated in a small bucket with 133 fresh water. 134 135 The Outcome: These samples were placed in water in a small bucket flotation. No 136 charred plant remains floated to the surface and the experiment produced just muddy 137 water (Figure S.6 - right). 138

139 140 Figure S.6: Flotation Experiment 1: (Left) a 250 ml block of sediment from CFD G7a (76.7±4.8 kya), Blombos 141 Cave; (right) result of flotation (Images: C. Larbey) 142 Flotation experiment 2 143 144 The Objective: a) To observe clearly the reaction of archaeological charcoal in both 145 fresh water and seawater, without surrounding sediment, which muddies the water. b) 146 To observe any difference between fresh and seawater. 147 148 The Method: Two fragments of dried archaeological charcoal from Blombos Cave and 149 two fragments of modern charcoal of equal size were floated. One fragment each of 150 archaeological and modern charcoal were floated in a jug of fresh water and one 151 fragment of each in a jug of filtered seawater. 152 153 The Result: The modern charcoal floated in both fresh and seawater. Both fragments of 154 archaeological charcoal ‘exploded’ on entry into the fresh and sea water and sank to 155 the bottom (Figure S.7). 156

7 157 158 Figure S.7: Flotation Experiment 2: Floating modern charcoal fragment and archaeological fragment ‘exploded’ and 159 sunk to the bottom of fresh water (Image: C. Larbey) 160 161 Flotation was not possible and block sediment samples would have to be taken back to 162 the UK for micro-excavation. 163 164 S.5 Laboratory process 165 166 The laboratory process is summarised in Table S.2. 167 168

8 169 Table S.2: Laboratory Process 170 171 Samples weighed and volumes taken on 172 Easyweigh GT4800 digital scales 173 ß 174 Air-Dried Samples 175 ß 176 Micro-excavated under a stereo-microscope 177 (Nikon SMZ800 x10, x20, x40) 178 Featherweight spring forceps used to sort 179 sample and select artefacts and plant fragments 180 ß 181 Fragments were mounted onto Agar Scientific 25mm diameter, 8mm pin 182 SEM stubs, covered with 25 mm carbon adhesive discs, with fragment numbers 183 marked in white ink 184 ß 185 Each stub was then analysed in a Hitachi TM 3000 environmental 186 scanning electron microscope. X30 and x40 magnifications were used 187 for whole fragment images and x300-x500 used for parenchyma and 188 morphological feature images. 189 ß 190 Micrograph images taken and initial analysis recorded 191 ß 192 Charred fragments remained stored on SEM stubs and curated in labelled 193 Agar Scientific SEM stub boxes (14 pin type) 194 195 Weights and volumes 196 197 The volumes were taken in a volume-marked glass beaker. 198 199 200 201

9 202 Air-drying 203 204 The block sediment samples were air-dried; if placed in the Leec electric drier, the 205 blocks of sediment become hard and difficult to separate. The air-drying of the block 206 samples made the charred fragments capable of being handled, although they were still 207 very fragile. In an excavation context, charcoal from Blombos Cave had the 208 consistency of butter, but if removed with surrounding sediment and left to dry out in a 209 box, the charred fragments became viable, that is; capable of being handled in the 210 normal way. This sampling technique was introduced to the sampling regimes at both 211 sites, with a significant increase in charcoal collection as a result. 212 213 Micro-excavating and sorting 214 215 Visible artefacts such as vertebrate remains, mollusc shells and lithics were removed 216 via an initial screening process. Larger samples were sub-sampled to fifty percent by 217 volume using a riffle box. During the sorting process, vertebrate remains (burned and 218 unburned and macro and micro-mammal), lithics and mollusc shells were recovered. 219 and these were recorded according to standard excavation procedures. These artefacts 220 were counted, bagged and labelled (but not washed). All subsequent screening was 221 performed under the stereo-microscope to identify putative charred plant remains. At 222 the end of screening each sample, the putative plant fragments were mounted onto 223 SEM stubs, labelled and placed into SEM storage boxes. Wood charcoal fragments 224 were included in the assemblage for SEM analysis and proved a good comparison to 225 distinguish parenchyma and root secondary xylem from wood. 226 227 SEM analysis 228 229 The SEM used was an Hitachi TM 3000 desktop environmental scanning electron 230 microscope, where no gold sputter coating was necessary so that samples are available 231 for future analysis. Analyses were made on full vacuum, in ‘Analyses’ mode. A whole 232 fragment micrograph was taken for each fragment and then subsequent images were 233 taken as appropriate, and analysed further. 234 235

10 236 Parenchyma analysis 237 238 Initial parenchyma identification was made at the time of scanning and characteristics 239 were recorded directly onto an excel spreadsheet. This process also meant that 240 fragments analysed early in the research were re-visited after the benefit of greater 241 knowledge and experience of the assemblage. 242 243 The aim of this process was to identify the presence/absence of cooked starchy plant 244 tissue from hearths. Identification to species, or family was ideal but as it is 245 difficult to identify species from fragments of parenchyma, the minimum requirement 246 was, if possible, to identify charred parenchyma with its associated characteristics and 247 other botanical remains such as seeds. 248 In addition to the modern reference collection created in this research and described 249 below, much of the identification of parenchyma and general morphological features 250 of starchy plant structures has come from the reference books created from the

251 research by Jon Hather 7-10. Other publications from Lucy Kubiak Martens, Sarah

252 Mason and Victor Paz have been particularly useful 11-18. 253 254 Curation 255 256 The archaeological fragments are curated and stored on the SEM stubs in labelled stub 257 boxes to minimize the amount of handling. Stub boxes are stored in the George Pitt 258 Rivers Laboratory. 259 260 The import of samples and laboratory processes complied with the DEFRA Directive 261 2008/61/EC and were imported into the UK under the appropriate licence for each 262 year. The samples were also exported from South Africa under the appropriate licences 263 from National Heritage Resources Act 25 1999. 264 265 S.6 Modern reference collection 266 267 The analysis of parenchyma and features of starchy plant tissue is limited by a lack of 268 regional reference materials. Whilst text books are invaluable, there are invariably 269 characteristics that might help in identifications and understanding regional variations.

11 270 There are no such publications from South Africa as no such research studies have 271 been conducted. It has been invaluable to be able to compare archaeological material 272 with the reference collection. The collection created a better understanding of the 273 biomes around the cave, and the variety of biomes meant that year-round plant food 274 sources would have been available through varying climate conditions. In the 275 laboratory, there was a tuber-bearing species Aspargus sprengeri that did not survive 276 the drying process because the tubers were almost entirely comprised of water, and 277 other species, such as Watsonia sp. corms that required longer burning because the 278 corms have evolved to be resistant to wild fires. The comparison of SEM micrographs 279 has supported the identification of features and in the case of two fragments to family. 280 The following describes the method used to collect and create the modern parenchyma 281 reference collection. 282 283 Across the Cape Floristic Region (CFR) the availability of geophyte species is: (a) 284 highest from winter to early summer (July-December); (b) lowest from mid-summer- 285 early autumn (January-April); (c) variable in the autumn (May-June); and (d) the

286 February- March summer period, is considered the ‘carbohydrate-crunch’ period 19. De 287 Vynck et al.’s research aimed to establish the foraging potential of carbohydrate 288 resources in the CFR. This study established that hunter-gatherers would be able to 289 forage sufficient carbohydrates to feed a family, year round across all vegetation types. 290 291 Around Blombos Cave is a natural ‘rock garden’ of succulent and geophyte species. 292 The surrounding landscape is a mosaic of primarily Strandveld, interspersed with 293 Renosterveld, coastal littoral vegetation, and Albany thicket, the latter characterised by 294 milkwood, (Sideroxylon inerme). Proteas can be found in remnant Fynbos about five 295 kilometres inland. Patches of rushes (Typha sp.) near the beach indicated the presence 296 of fresh water seep (Figure S.8). 297

12 298

299 300 Figure S.8: Plants around Blombos Cave. Top left: succulent with taproot; Top right: 301 reeds (Typha sp.) indicating fresh water seep; Bottom left: milkwood thicket 302 (Sideroxylon inerme) and; Bottom right: wild pincushion proteas (Leucospermum 303 cuneiforme) in inland remnant fynbos, close to Blombos Cave (Images: C. Larbey) 304 305 The samples collected can be found at Table S.3 306 307

13 308 Table S.2: Modern root and tuber samples collected from around Blombos Cave Sample Common Classification Name Family No. Name Elytopappus Dicots BBC-10 rhinocerotis

BBC-11 sp. Asteraceae

Felicia amoena BBC-01 Asteraceae Felicia latifolia

BBC-06 Pelargonium sp. Geraniaceae

Pharnaceum BBC-08 Aizoaceae Radish dichotum

Rhoicissus BBC-05 Vitaceae Wild grape tridentata

Ruschia BBC-02 Aizoaceae geminiflora

Climbing Monocots BBC-09 Bowiea vollibilis Hyacinthaceae Potato BBC-07 Lachanalia Hyacinthaceae

Ledebouria BBC-12 Liliaceae revoluta

Massonia BBC-04 Asparagaceae pustulata

BBC-03 Watsonia sp. Iridaceae

309 310 All species were collected from the immediate vicinity of Blombos Cave, with the 311 exception of the Ledebouria revoluta that came from the fynbos behind the Blombos 312 site. The particularly heavy rains during the rainy season that had washed out part of 313 the road and some of the paths around Blombos had also washed out plants, corms and 314 tubers from the rocky slopes around the cave, making collection easy and 315 identification more complex. 316

14 317 S.7 Summary of results 318 319 The total results from the samples taken from Blombos Cave can be seen at Table S.3. 320 321 Table S.3: Botanical Results from Blombos Cave (* = 50% of sample screened). Shaded samples unlikely to be 322 hearths or combustion features, possibly organic deposits of humified plant remains. Sample numbering starts at 4 323 because samples 1-3 were trial floated on site. No. of Total no. of Starchy starchy Weight Volume Sample plant plants as % Context Phase plant sample sample No. fragments of total fragments (g) (ml) in sample plants in sample Still 4 I6c CF 58 25 43 207 250 Bay G7a Still 5 24 0 0 236 200 CFB/CFC Bay G7b CGAB 6 MSA1 77 33 43 513 650 h1 7 I6c CGAC MSA1 14 6* 43 196 250 8 G7b CGAC MSA1 76 31* 41 805 825 9 H7a CGAC MSA1 26 9* 35 456 500 Still 10 F7b CFA 29 2 7 40 125 Bay 324 325 There is a high difference in density of charred parenchyma fragments between those 326 contexts that are anthropogenic combustion features and those that turned out not to be 327 combustion features. In the case of the two samples discussed in this paper had the 328 highest number of parenchyma fragments, which formed a high total percentage of 329 charred plant fragments recovered from the sample. The samples that turned out not to 330 come from anthropogenic combustion features had zero – low total parenchyma 331 fragment numbers, especially as F7b CFA, which was the actual control sample. 332 333 334 335

15 336 337 References for Supplementary Information 338 339 1 Haaland, M. M., Friesem, D. E., Miller, C. E. & Henshilwood, C. S. Heat- 340 induced alteration of glauconitic minerals in the Middle Stone Age levels of 341 Blombos Cave, South Africa: Implications for evaluating site structure and 342 burning events. Journal of Archaeological Science 86, 81-100 (2017). 343 2 Mentzer, S. M. Microarchaeological approaches to the identification and 344 interpretation of combustion features in prehistoric archaeological sites. 345 Journal of Archaeological Method and Theory 21, 616-668, 346 doi:10.1007/s10816-012-9163-2 (2012). 347 3 Mentzer, S. M. in Encyclopedia of Geoarchaeology (ed A. S. Gilber) 411- 348 424 (Springer, 2016). 349 4 Pearsall, D. M. Paleoethnobotany: A Handbook of Procedures. Second edn, 350 (Academic Press, 2000). 351 5 Knight, M., Ballantyne, R., Robinson Zeki, I. & Gibson, D. The Must Farm 352 pile-dwelling settlement. Antiquity 93, 645-663, doi:10.15184/aqy.2019.38 353 (2019). 354 6 van der Veen, M. Formation processes of desiccated and carbonized plant 355 remains - the identification of routine practice. Journal of Archaeological 356 Science 34, 968-990 (2007). 357 7 Hather, J. G. The Morphological and Anatomical Interpretation and 358 Identification of Charred Vegetative Parenchymatous Plant Remains - Volume 359 2 Ph.D thesis, University College London, (1988). 360 8 Hather, J. G. An Archaeobotanical Guide to Root and Tuber Identifiation: 361 Volume 1 Europe and South West Asia. (Oxbow, 1993). 362 9 Hather, J. G. The identification of charred archaeological remains of vegetative 363 parenchymous tissue. Journal of Archaeological Science 18, 661-675 (1991). 364 10 Hather, J. G. Archaeological Parenchyma. (Archetype Publications, 2000). 365 11 Kubiak-Martens, Brinkkemper, O. & Oudemans, T. F. M. What’s for dinner? 366 Processed food in the coastal area of the northern Netherlands in the Late 367 Neolithic. Vegetation History and Archaeobotany 24, 47-62, 368 doi:10.1007/s00334-014-0485-8 (2015). 369 12 Kubiak-Martens, L. New evidence for the use of root foods in pre-agrarian 370 subsistence recovered from the late mesolithic site at Halsskov, Denmark. 371 Vegetation History and Archaeobotany 11, 23-31 (2002). 372 13 Kubiak-Martens, L. Evidence for possible use of plant foods in Palaeolithic and 373 Mesolithic diet from the site of Calowanie in the central part of the Polish 374 Plain. Vegetation History and Archaeobotany 5, 33-38 (1996). 375 14 Kubiak-Martens, L. The plant food component of the diet at the late Mesolithic 376 (Ertebolle) settlement at Tybrind Vig, Denmark. Vegetation History and 377 Archaeobotany 8, 117-127 (1999). 378 15 Mason, S. L. R. & Hather, J. G. in Hunter-Gatherer Landscape Archaeology: 379 the southern Hebrides mesolithic project 1988-1998. Archaeological fieldwork 380 on Colonsay, computer modelling, experimental archaeology, and final 381 interpretations Vol. 2 (ed S. Mithen) 415-425 (McDonald Institute for 382 Archaeological Research, 2000). 383 16 Mason, S. L. R., Hather, J. G. & Hillman, G. C. Preliminary investigation of 384 the plant macro-remains from Dolni Vestonice. Antiquity 68, 48-57 (1994).

16 385 17 Paz, V. Rock Shelters, Caves, and Archaeobotany in Island Southeast Asia. 386 Asian Perspectives 44, 107-118 (2005). 387 18 Barker, G. et al. The 'human revolution' in lowland tropical Southeast Asia: the 388 antiquity and behavior of anatomically modern humans at Niah Cave (Sarawak, 389 Borneo). Journal of human evolution 52, 243-261, 390 doi:10.1016/j.jhevol.2006.08.011 (2007). 391 19 De Vynck, J. C., Cowling, R. M., Potts, A. J. & Marean, C. W. Seasonal 392 availability of edible underground and above ground carbohydrate resources to 393 human foragers on the Cape south coast, South Africa. PeerJ 4, e1679, 394 doi:10.7717/peerj.1679 (2016). 395

17