Discriminating between the end products of the earlier Middle Stone Age sub-stages at Klasies River using biplot methodology
Introduction
The African Middle Stone Age has become of increased interest because under the Out-of-Africa hypothesis it was during this stage in the Late Pleistocene that modern humans may have migrated to Eurasia. Sub-Saharan Africa may have been the biogeographic centre for the evolution and dispersal of modern humans [26] and this contention has support in various molecular biological studies [15]. The evidence from Middle Stone Age archaeological sites is thus relevant to the study of the emergence of modern humans. In the main, it is sites dating to the first half of the Late Pleistocene that are well represented, notably in southern Africa, and this time was probably a time of continent-wide population expansion [10]. This time range also provides some of the earliest evidence for regional distributions of typologically distinctive artefact types in horizon markers, like the Howiesons Poort, Lupemban and the Aterian [8], [32], suggesting patterning different from the Earlier Stone Age.
For the most part, Middle Stone Age artefact assemblages include few distinctive retouched artefact types which makes their typological description less informative. Assemblages give the impression of belonging to a culture stratigraphic stage of long duration with little change in the artefacts produced. In the classic South African sequence established by reference to the Klasies River main site [39], five culture-stratigraphic sub-stages have been recognised as subdividing the Middle Stone Age. In stratigraphic order from oldest to youngest, Singer and Wymer labelled these sub-stages MSA I, MSA II, Howiesons Poort, MSA III and MSA IV. An extension of this scheme proposed by Volman [49], [50] recognises an older sub-stage as MSA I and labels the original MSA I and MSA II, 2a and 2b to indicate that these two sub-stages are very similar to each other. The sub-stages of the Middle Stone Age, with the exception of the Howiesons Poort, have not been viewed as discrete industries [3], [39], [50], [42], [43]. The reasons posited include the lack of clear-cut technological and typological divisions, and continuities in metric parameters between the sub-stages. These assessments are mainly based on typological considerations, which for such assemblages may not be an entirely appropriate approach. This is because the end products are predetermined by preparation of the core rather than through retouch and therefore, technological analyses are more informative.
Technological analyses of the Middle Stone Age at Klasies Rivers has led to the conclusion that the MSA I and MSA II are more than entities of convenience and that they represent culture-stratigraphic divisions at the same level as the Howiesons Poort [54], [55]. In this paper, quantitative measurements of the end products, points and blades, from the MSA I and MSA II sub-stages have been subjected to extensive univariate and multivariate statistical analyses in order to explore this proposition further. The MSA II sub-stage has been divided into two classes, MSA II lower (MSA II L) and MSA II upper (MSA II U) to investigate the degree to which the end products of the two sub-stages differ in terms of seven continuous variables. A further goal has been to determine whether the differences between MSA I and MSA II are of the same order as the differences between the two MSA II groups. Ultimately then, the purpose is to develop a methodology that would not only explore the statistical basis of the sub-stage divisions in the Klasies River sequence but also allow comparisons to be drawn with other Middle Stone Age assemblages in southern Africa and elsewhere.
Graphical representations of results to accompany any statistical analysis are strongly recommended by Chambers et al. [7] in the statement “there is no single statistical tool that is as powerful as a well-chosen graph”. In particular, this is true for multivariate statistical analysis, where by nature multidimensional data are explored. Biplot methodology is innovatively used in the present study, not only to provide graphical representations of multidimensional data accompanying well-established multivariate statistical methods, but also to direct the multivariate statistical analysis itself. The biplot method is of particular importance firstly, as it summarises the multivariate analyses, secondly, because it is an exploratory tool and thirdly, because it provides a single informative quantitative display of all observations from the different sub-stages together with all the variables at a glance. Gardner [19] provides a detailed exposition of biplot methodology for discriminant analysis. She proposed several novel extensions in biplot methodology for describing and quantifying multidimensional separation and overlap of predefined groups. In the present paper some of these proposals are implemented for displaying the observations from the different sub-stages together with all variables simultaneously. The biplot techniques can be extended to cluster analysis situations where groups are not predefined as they are here. This extension of the technique may have application in biological taxonomy and archaeology.
Section snippets
Site and sample
Klasies River main site is situated on the southern coast of South Africa (33.06°S, 23.24°E). Main site is a single depository against a cliff face with several openings into the cliff, described as caves 1, 1B, 1A and 2 (Fig. 1). The deposits, which were continuous between the caves before truncation by erosion, have a thickness of more than 18 m. They consist of numerous layers of hearths, ash, stone and bone intercalated by sandy layers. This suggests intermittent occupation by groups of
Statistical analysis of the end products
Refitting and experimentation have indicated that end products can be recognised by their symmetry, prepared platforms and a regular disposition of dorsal scars [46]. These characteristics were used to select the end products of the MSA I and MSA II. In this analysis end products were labelled blades (the term flake-blade is used by some researchers) and points (termed pointed flake-blades by Singer and Wymer [39] and convergent flake-blades by Thackeray and Kelly [44]). In the process of
Blades
The properties of the blades data are summarised in the boxplots appearing in Fig. 2. (In the electronic version of this paper, Fig. 2, Fig. 3, Fig. 4, Fig. 5, Fig. 6, Fig. 7 appear in colour. This substantially enhances the interpretative value of the graphical displays). The boxplots (Fig. 2) display a comparison of the observations from the three groups in the two sub-stages in terms of the symmetry/skewness, variation, location, range and outlying values for each variable recorded. Each box
Relationship between the statistical results and technological behaviour
From the statistical procedures carried out it is concluded that the MSA I differs from the two MSA II groups to a greater degree than the two MSA II groups differ from each other. The procedures show that the ratio of length to platform thickness is the variable that discriminates most clearly between the MSA I and MSA II blades and points. The MSA I has larger ratio of length to platform thickness values while the MSA II U and MSA II L are very similar in terms of this ratio. Further, it is
Classification using CVA biplot classification procedure
The above discriminant analyses not only provide a description of the overlap and separation among the various groups, but offer the potential for blades and points of uncertain provenance to be classified asbelonging to one or other of the sub-stages. In order to evaluate the results obtained by the CVA biplot classification procedure a resubstitution method (cf. [17]) was used. Using the data above, it is found that this classification procedure does not classify the end products of the MSA
Concluding discussion
This paper has applied a suite of multivariate statistical techniques to an archaeological problem, the validity of the separation of two proposed sub-stages of the Late Pleistocene Middle Stone Age in the stratified sequence at Klasies River main site. The focus has been on end products of core reduction, in this case regular blades and point blanks that are rarely retouched. The form of the end products is predetermined by the shaping of the core and in the process flaking debris is produced
Acknowledgements
The Katholieke University of Leuven, Belgium and the University of Stellenbosch, are thanked for supporting this study through post-doctoral research fellowships. This paper is based on D.Phil and Ph.D studies for which financial assistance was provided by the NRF (Centre for Scientific Development). The Struwig Trust facilitated research in the field and the Iziko Museums of Cape Town made collections available for study. Hannes van der Merwe is thanked for valuable comments and Liesl van
References (55)
Luminescence dating in less than ideal conditions: case studies from Klasies River main site and Duinefontein, South Africa
Journal of Archaeological Science
(2002)- et al.
The use of a multivariate graphic display technique as an exploratory tool in the analysis of inter-assemblage lithic variability: a case study from Qafzeh Cave, Israel
Journal of Archaeological Science
(2000) Middle Stone Age artefacts from the 1993 and 1995 excavations of Die Kelders Cave 1, South Africa
Journal of Human Evolution
(2000)Variability in the Middle Stone Age lithic sequence, 115 000–60 000 years ago at Klasies River, South Africa
Journal of Archaeological Science
(2002)- et al.
The Definition and Interpretation of Levallois Technology
(1995) Exploratory Multivariate Analysis in Archaeology
(1994)- Beaumont, P. Border Cave. Masters Thesis. University of Cape Town,...
- et al.
Modern man in Sub-Saharan Africa prior to 49 000 years B.P.: a review and evaluation with particular reference to Border Cave
SouthAfrican Journal of Science
(1978) Trees or chains, links or branches: conceptual alternatives for consideration of stone tool production and other sequential activities
Journal of Archaeological Method and Theory
(2001)Levallois: a volumetric construction, methods, a technique
Graphical Methods for Data Analysis
African and Asian perspectives on the origins of modern humans
Philosophical Transactions of the Royal Society, London B
Two Late Pleistocene-Holocene archaeological depositories from the southern Cape, South Africa
South African Archaeological Bulletin
Modern human emergence: an African archaeological perspective
The origins of modern people: the evidence from Klasies River
The stratigraphy and sedimentology of the main site sequence, Klasies River, South Africa
South African Archaeological Bulletin
Middle Pleistocene populations of southern Africa and the emergence of modern behaviour
Biological and cultural development of Pleistocene people in an Old World southern continent
Estimating relative population sizes from simulated data sets and questions of greater African effective size
American Journal of Physical Anthropology
A First Course in Multivariate Statistics
The biplot graphical display of matrices with application to principal component analysis
Biometrika
Biplot methodology for discriminant analysis based upon robust methods and principal curves
A commentary on the history and present position of South African prehistory with full bibliography
Bantu Studies
Biplots
Electron-spin resonance dating of tooth enamel from Klasies River Mouth Cave
Current Anthropology
Cited by (39)
Early or Middle Stone Age? The lithic assemblage of Capangombe – Santo António, Namibe Province (Angola)
2023, Anthropologie (France)Human manual distal phalanges from the Middle Stone Age deposits of Klasies River Main Site, Western Cape Province, South Africa
2020, Journal of Human EvolutionThe palaeoecology of Klasies River, South Africa: An analysis of the large mammal remains from the 1984–1995 excavations of Cave 1 and 1A
2020, Quaternary Science ReviewsCitation Excerpt :There is evidence for more intensive reduction of cores in the MSA l (Wurz, 2010). This techno-complex is, for example, distinguished by intensive platform preparation in the form of step flaking and rubbing, and a statistically significant difference in platform size to length, compared to the other phases at KRM (Wurz et al., 2003). The MSA l fauna analysed here are from Cave 1/1A, Squares Z44-AA43, Layers SCB3S- SBS (Table 1; Fig. 3, see also Wurz, 2000 Table 11).
Connections, culture and environments around 100 000 years ago at Klasies River main site
2018, Quaternary InternationalCitation Excerpt :For example, analyses of the recovered human remains have contributed to our understanding of the evolution of modern human anatomy (e.g., Rightmire and Deacon, 1991; Churchill et al., 1996; Grine et al., 1998; Rightmire and Deacon, 2001; Grine et al., 2017). The shellfish and fish remains from the site provided notable information on early exploitation of coastal resources (Thackeray, 1988; Von den Driesch, 2004; Van Niekerk, 2011; Langejans et al., 2012, 2017), and lithic analyses contributed towards understanding of MSA technological strategies (Thackeray, 1989; Wurz, 1999, 2002; 2012, 2013; Wurz et al., 2003; Villa et al., 2010). The faunal data have been discussed in the context of subsistence behavior (Klein, 1976, 1989; Milo, 1998) and palaeoenvironmental reconstructions (e.g. Klein, 1976; Avery, 1987; Van Pletzen, 2000; Hillestad-Nel et al. 2018).
The Middle Stone Age human fossil record from Klasies River Main Site
2017, Journal of Human Evolution