The MSA phytolith assemblage in this study represents early human occupation from the Pre- Still Bay to the post-Howiesons Poort techno-complexes that span ca. > 70,000 to 55,000 yr BP at Sibudu cave. Of the 28 samples taken from 28 layers, only two were taken from a clearly recognisable hearth. Phytoliths are abundant and extremely diverse, in agreement with previous studies, and a variety of plants were utilised at Sibudu (Schiegl et al., 2004; Allott, 2004). In all samples, many phytoliths appear heat-altered, similar to Schiegl et al. (2004) who found heat- altered phytoliths in samples taken from hearths and in layers with no visible hearths. Whereas phytoliths are abundant and identifiable in the assemblage, in several samples they appear destroyed and unidentifiable providing evidence for the use of extensive fire at Sibudu as previously suggested (Schiegl et al., 2004). Phytoliths that are characteristic of woody taxa and
147 grasses are well represented at Sibudu and the abundance of non-grass phytoliths confirms that wood ash is an important component of sediments at Sibudu and that they also often contain some grass phytoliths (Schiegl et al., 2004; Goldberg et al., 2009). It has been suggested that at archaeological sites where wood was the main plant utilised, eudicot plants might be underrepresented in the archaeological record because they are low phytolith producers especially in their wood (Albert, 2000; Collura and Neumann, 2017). This study found abundant phytoliths and although many are redundant morphotypes, they are not typical for grasses and are types that are often associated with wood, bark and leaves. More so, although grass short cell phytoliths can reach up to 50% of the phytoliths in few samples, their numbers can be considered lower than expected compared to non-grass phytoliths because of the high phytolith production in grasses compared to woody taxa that are very low phytolith producers (Piperno, 2006). Grasses are reported to produce up to about 20 times more phytoliths than wood/bark and that wood/bark are likely to be under represented compared to grasses in the archaeological record (Albert and Weiner, 2001).
This study set out to count approximately 300 phytoliths per sample, and this number was achieved, with ease in nearly all samples. The issue of how phytoliths entered the site and their preservation have already been discussed for Sibudu (Schiegl et al., 2004; Schiegl and Conard, 2006; Goldberg et al., 2009). Phytoliths were grouped into grass and dicotyledonous types in 19 sediment samples representing the post-Howiesons to the final MSA period (ca. 55,000 to 39,000 yr BP) (Schiegl et al., 2004). Their focus was mainly characterising the wood ash at Sibudu. This current study focused on discussing phytoliths in relation to early human-plant interaction, vegetation and climate during the MSA of South Africa. To achieve the latter, there was need to identify the grass phytoliths at Sibudu to a higher taxonomic resolution. Although grass phytoliths are preserved in the MSA assemblage at Sibudu, the assemblage is mainly dominated by the problematic rondel morphotypes that are often associated with C3 cold climate grasses of the subfamily Pooideae (Piperno, 2006; Cordova, 2013). Rondels are problematic because they have been found to be an important component of the modern phytolith assemblages of other subfamilies in Africa (e.g. Bamford et al., 2006; Cordova, 2013).
148 From the results in this present study, rondels are an important component of all grass subfamilies studied from the SRZ of South Africa, a region where grasses of the subfamily Pooideae are rare (Cordova, 2013). Rondels were found to also be the dominant morphotype in a potentially new MSA site in Limpopo also within the SRZ (Wadley et al., 2016). Rondels may therefore be an important part of grasses from this region. Their dominance in the archaeological record might also point to their high resistance to dissolution and fragmentation compared to other morphotypes. It has been found that smaller phytolith forms like rondels and trapezoids have less surface pitting than elongates or bulliforms due to less exposed surface (Osterrieth et al., 2009). It has also been suggested that phytolith with a more pure silica composition and less impurities from minerals like Calcicum and Iron tend to experience less dissolution (Osterrieth et al., 2009). Interpretations of the phytolith archaeological record in the current study are therefore based on combinations of rondels with another more diagnositic dominant grass short cell phytolith as suggested by the modern reference. The three modern soil samples analysed also suggest that rondels are the dominant grass short cell.
This study also set out to identify the dicotyledonous phytoliths to any highest resolution possible for specific plants that produced diagnostic phytoliths. Because sedges have been identified at Sibudu (Sievers, 2006, 2011, 2013), this study also set out to identify sedge phytoliths in the MSA assemblage. To enable identification and interpretation of the MSA assemblage, a modern plant reference collection was made to understand the phytolith morphotypes that can be expected for taxa from this region and that are likely to be found in the MSA assemblage. Important to note here is that phytoliths from various plants studied where found in the archaeological record but some important types from woody taxa do not seem to be preserved in the archaeological record. Hair bases that were common in woody taxa were hardly encountered in the archaeological record and the morphotypes that are diagnostic of Celtis, the only genus that was identifiable to that level, were not observed in the phytolith record. Piperno (2006) reported that hair bases may not be commonly preserved in sediments making them of less taxonomic value in phytolith analysis. In contrast, hair bases were indicated in the archaeological phytolith study at Sibudu by Schiegl et al. (2004).
Although plants at Sibudu were brought to the site by human activity and the sediments are largely anthropogenic (Goldberg et al., 2009), it is possible that some phytoliths found in the
149 sediment layers may represent grasses that were not brought in intentionally. The modern plant reference collection showed that some grass phytoliths could have been deposited at the site through woody taxa and sedges. During analysis of woody taxa for their phytolith content, some species were found to contain some grass phytoliths in their assemblage i.e. Baphia racemosa,
Eugenia capensis, Syzygium cordatum and Diospyros natalensis. A conical shaped phytolith was
observed in Ficus polita but it appeared somewhat different from the sedge type to the eye. Although it was treated as contamination in this study, it is important to mention that Hart (1990) presented evidence for the redundancy of the sedge diagnostic cone shaped phytoliths as they were observed in three species of dicotyledonous families: Acacia schinoides, Banksia
oblongifolia and Casuarina distyla and argues against their use as diagnostic to Cyperaceae at
least for Australian vegetation. For the sedges, both the Bulbostylis spp. produced some grass phytoliths while two species produced a copious amount of grass short cell phytoliths mainly bilobates i.e. Schoenoxiphium sparteum and Cyperus obtusifloris spp. flavissimus. However, these four sedge species were not collected from the main study area (Sibudu). This study supports previous studies that suggest that this kind of grass phytolith deposition from other plants in the sediment is possible in archaeological sites although this kind of contamination is reported to be minimal (Albert and Weiner, 2001) and recently also in South African eudicots (Esteban, 2016). More so, a few phytoliths at Sibudu may also have been transported by wind, water or animals as suggested by Schiegl et al. (2004).
What follows here is how phytoliths have provided evidence for early human-plant interactions, and the prevailing vegetation and climate in which the early humans lived at Sibudu during the MSA. It is important to note that the phytolith data at Sibudu is most likely very local with forest taxa obtained from the forest around Sibudu and grasses subfamily types and sedges were influenced by the Tongati River. The phytolith morphotypes used to make these inferences and their abundance are shown in Figure 7.1 in Chapter seven. Lobates, saddles and rondels including trapeziform short cells represent grasses, while blocky parallelepiped and blocky other are associated with sedges and woody taxa, globular phytoliths are associated with woody taxa, cones and achenes are associated with sedges while elongate and tabular phytoliths are generally of little taxonomic value. Those grouped under ‘other’ are often associated with eudicot plants but can also occur in monocots and they were: tracheids, multicellular polygonal cells, sclereids,
150 facetate bodies, bulbous structures and other various morphologies that did not occur in the modern grass and sedge phytolith assemblage.