Trajectories of cultural innovation from the Middle to Later Stone Age in Eastern Africa: Personal ornaments, bone artifacts, and ocher from Panga ya Saidi, Kenya

Abstract

African Middle Stone Age (MSA) populations used pigments, manufactured and wore personal ornaments, made abstract engravings, and produced fully shaped bone tools. However, ongoing research across Africa reveals variability in the emergence of cultural innovations in the MSA and their subsequent development through the Later Stone Age (LSA). When present, it appears that cultural innovations manifest regional variability, suggestive of distinct cultural traditions. In eastern Africa, several Late Pleistocene sites have produced evidence for novel activities, but the chronologies of key behavioral innovations remain unclear. The 3 m deep, well-dated, Panga ya Saidi sequence in eastern Kenya, encompassing 19 layers covering a time span of 78 kyr beginning in late Marine Isotope Stage 5, is the only known African site recording the interplay between cultural and ecological diversity in a coastal forested environment. Excavations have yielded worked and incised bones, ostrich eggshell beads (OES), beads made from seashells, worked and engraved ocher pieces, fragments of coral, and a belemnite fossil. Here, we provide, for the first time, a detailed analysis of this material. This includes a taphonomic, archeozoological, technological, and functional study of bone artifacts; a technological and morphometric analysis of personal ornaments; and a technological and geochemical analysis of ocher pieces. The interpretation of the results stemming from the analysis of OES beads is guided by an ethnoarcheological perspective and field observations. We demonstrate that key cultural innovations on the eastern African coast are evident by 67 ka and exhibit remarkable diversity through the LSA and Iron Age. We suggest the cultural trajectories evident at Panga ya Saidi were shaped by both regional traditions and cultural/demic diffusion.

Introduction

The present study aims to document the emergence of key behavioral innovations in the late Middle Stone Age (MSA) and the Later Stone Age (LSA) of eastern Africa. The emergence of innovations such as the use of mineral pigments, the wearing of ornaments, and the production of abstract engravings has traditionally been interpreted as the direct consequence of cognitive changes linked to the origin of Homo sapiens in Africa (McBrearty and Brooks, 2000, Henshilwood and Marean, 2003, Shea, 2011, Bruner, 2014, Marean, 2015, Coolidge and Wynn, 2017). A significant debate crosscutting archeology, paleoanthropology, and archeogenetics has persisted over whether complex cultural traits originated and spread uniquely with H. sapiens (Stringer and Andrews, 1988, Relethford and Harpending, 1994, Klein, 1995, Henn et al., 2011; see Rito et al., 2019 and Chan et al., 2019 for controversial updates of this view) or whether they emerged as behavioral responses by diverse hominin populations to specific environmental, geographic, and cultural settings (d'Errico, 2003, Zilhão, 2006, Gunz et al., 2009, Hublin et al., 2017, Neubauer et al., 2018).

Recent studies combining methodological advances and data from paleoanthropology, genetics, and archeology seem to question a direct cause-and-effect relationship between the speciation event that would have given rise to H. sapiens and the emergence of modern cultural traits (Scerri et al., 2018, Scerri et al., 2019). For more than two decades, current genetic variability was interpreted as evidence that all present-day humans descended from a unique ancestral population, which is to be sought by some authors in eastern or southern Africa (Henn et al., 2011, Pagani et al., 2012, Rito et al., 2019, Chan et al., 2019). Modern human remains, such as those found at Herto and Kibish, in Ethiopia, dated between 150 and 200 ka, have long been considered fossil representatives of H. sapiens (Stringer, 2002, White et al., 2003). Endowed with a modern cranial morphology and cognitive capacities, these populations were seen as having rapidly replaced, without notable genetic or cultural exchange, archaic African populations. Furthermore, these populations were understood to have left Africa, perhaps several times, and replaced with minimal interactions, after 60 ka, all archaic Eurasian hominins, such as the Neanderthals in Europe and the Near East, and the descendants of Homo erectus in Asia.

A growing body of evidence now suggests, however, that the anatomical traits that distinguish our cranial morphology from that of our archaic African ancestors developed gradually, in Africa, between 300 ka and 35 ka, with present-day human morphological variation only reached in the window of 100 ka to 35 ka (Gunz et al., 2009, Hublin et al., 2017, Scerri et al., 2018, Neubauer et al., 2018). In other words, paleoanthropology does not seem to find a before (‘archaic’) and an after (‘modern’) character but identifies, starting from 300 ka, a mix of ‘modern’ associated with ‘archaic’ characteristics in populations throughout Africa. The morphological variability, chronology, and geographical distribution of the first fossils now attributed to H. sapiens suggest that evolution has progressed partially independently in several African regions (Scerri et al., 2018, Scerri et al., 2019).

Critical analysis of the approaches used in the past by geneticists to identify population splits and infer the timing of the speciation event that would have produced an ancestral modern population shows that these computational models neglected factors such as population structure, changes in population size, and gene flow (Wakeley, 1999, Storz and Beaumont, 2002, Scerri et al., 2017, Chikhi et al., 2018). It is now increasingly accepted that these factors are capable of better explaining current genetic variability than the classic version of the Single-Origin or Out-of-Africa model. Taking into account these factors not only is more realistic in light of what is known about population dynamics but also better explains genomic evidence that indicates admixture with archaic hominins or the signature of very ancient modern human ancestry (Wolf and Akey, 2018, Durvasula and Sankararaman, 2018, Patin and Quintana-Murci, 2018, Villanea and Schraiber, 2019). A population structure model is also aligned with evidence demonstrating genetic admixture between archaic hominins and modern populations expanding out of Africa (Fu et al., 2015, Harris and Nielsen, 2016, Schlebusch et al., 2017, Browning et al., 2018, Lombard et al., 2018, Slon et al., 2018, Gunz et al., 2019).

Archeological data on the emergence of key cultural innovations corroborate a complex demographic and cultural scenario for modern human origins. In the case of a unique origin in Southern or Eastern Africa and a causal relationship between speciation and cultural ‘complexity,’ one would expect a clear correspondence between the emergence of our species and that, for example, of complex technologies and symbolic material cultures, that would come to be nearly universal hallmarks of H. sapiens. The beginning of the MSA in Africa is characterized by the gradual shift from the large cutting tools typical of the Acheulean to the production of flakes from prepared Levallois cores and the creation of moderately sized (sometimes hafted) retouched flakes (Douze and Delagnes, 2016). This innovation is also, almost at the same time and perhaps even earlier, observed in vast regions of Eurasia, without its point of origin being known (Adler et al., 2014, Hu et al., 2019). It cannot therefore be interpreted, at least without further supporting evidence, as reflecting the successful expansion of early anatomically modern populations within the African continent. It is only from 140 ka in North Africa with the Aterian (Doerschner et al., 2016, Scerri, 2017) and from 100 ka in Southern Africa with the Pietersburg and subsequently the Still Bay and the Howiesons Poort (d'Errico et al., 2017, De la Peña et al., 2018, Porraz et al., 2018) that we can easily recognize regionally distinct lithic technologies and tool types.

Unfortunately, the archeological record for manifestations of major cultural innovations remains poorly resolved overall. Current models are shaped by the higher resolution records from Europe (d'Errico and Banks, 2015, Locht et al., 2016, Giaccio et al., 2017, Zilhão et al., 2017) and increasingly from southern and northern Africa (Jacobs et al., 2008, Jacobs et al., 2011, Jacobs et al., 2012, Jacobs and Roberts, 2015, d'Errico et al., 2018). Eastern Africa—one of the assumed centers of human cultural and biological evolution—is represented by only a handful of sites with artifacts indicative of symbolic mediated behavior and complex bone technology (Ambrose, 1998, Assefa et al., 2008, Assefa et al., 2018, Miller and Willoughby, 2014, Rosso et al., 2014, Rosso et al., 2016, Rosso et al., 2017, Brooks et al., 2018). Most of these have lacked systematic analysis until only very recently (Brooks et al., 2018, Tryon et al., 2018, Miller, 2019, Tryon, 2019). These records are biased toward the Central Rift Valley of the interior, which may not reflect the full diversity of behaviors across the region. A greater limitation is that many of these important assemblages lack the chronospatial resolution to accurately gauge change over time, they are not associated with local climatic data, and/or are not associated with the remains of their hominin creators. As a result, we need a more accurate chronology for changes in human behavior in the region and data linkages to test aforementioned hypotheses regarding the drivers for the emergence of key cultural innovations such as ocher use, the production of bone tools, and personal ornaments (following McBrearty and Brooks, 2000; see also Shea, 2011).

Panga ya Saidi (PYS), a recently published archeological sequence located north of Mombasa in Kenya and spanning the last 78 kyr (Shipton et al., 2018), has yielded a large collection of personal ornaments, bone artifacts, modified ocher, and engraved objects spanning the MSA-to-LSA transition in a continuous sequence that extends into the late Iron Age of recent centuries. This sequence is well dated, linked with a high-resolution paleoecological record and rich technological assemblages (Shipton et al., 2018, Roberts et al., 2019), and contains human burials that have yielded ancient DNA (Skoglund et al., 2017). PYS is also the only site where systematic flotation and fine mesh screening permitted high rates of recovery for small symbolic artifacts. As a result, the site offers a unique insight into the emergence of key cultural innovations in eastern Africa. In the present article, we characterize these innovations with the aim of identifying regional trends and similarities with innovations recorded in neighboring regions, proposing hypotheses regarding the impact of these innovations on late MSA and early LSA eastern African societies, and discussing the implication of this record for our understanding mechanisms of behavioral change in this region.

The complex and nonlinear patterns emerging from our current state of research highlight the need to expand relevant data sets within Africa. Artifacts that suggest symbolic practices (e.g., pigments, ornaments, burials, abstract engravings and drawings, systems of notation) appear at different times in Africa, and some of them are still not found in vast areas of this continent until a few thousand years ago. Moreover, some of these innovations seem to disappear for thousands of years and then reappear in other forms, contradicting the idea of an exponential expansion of symbolic material culture linked to the sudden origin of new cognitive abilities.

The use of mineral pigments is the only innovation that coincides, to a degree, with the beginning of the MSA. The recent discovery of modified red ocher fragments at Olorgesailie, Kenya (Brooks et al., 2018), confirms previous discoveries of modified ocher in the early MSA made at Kapturin in Kenya, Twin Rivers in Zambia, and Wonderwerk, Canteen Kopje, and Kathu Pan 1 in South Africa (Barham, 2002, Watts et al., 2016). However, considering the age of these findings, 320–280 ka for Olorgesailie and earlier for the southern Africa sites, and what we know about the anatomy of the contemporary African human populations, the first users of ocher pigments probably possessed a number of archaic morphological attributes. Therefore, the first pigment use cannot be taken as reflecting the emergence of modern culture triggered by a speciation event. The first known ornaments, consisting of perforated marine gastropods covered with ocher, belonging to the species Tritia gibbosula, are found at sites in Morocco and Algeria dating back to between 120 ka and 80 ka (Bouzouggar et al., 2007, d'Errico et al., 2009, Steele et al., 2019). Perforated shells belonging to the same species and bivalves of the genus Glicymeris, found at Qafzeh and Skhul, in the Near East, date back to about the same period (Mayer et al., 2009). Afrolittorina africana and Mancinella capensis shells are used as beads at Sibudu, KwaZulu-Natal, in layers dated to between 70 ka and 46 ka (Vanhaeren et al., 2019). Perforated and ochered marine gastropods, belonging to Nassarius kraussianus, are used as ornaments at Blombos Cave, Eastern Cape, around 73 ka (d’Errico et al., 2005). At Border Cave, KwaZulu-Natal, a whole Conus ebraeus that was perforated and ochered was found in a pit dated to 74 ka, in which the body of an infant was deposited (d’Errico and Backwell, 2016).

Abstract representations, engraved on bone and ocher pieces or drawn with ocher pencils on stone, are found at Blombos Cave, Klasies River, Klein Kliphuis, Sibudu, and Pinnacle Point, South Africa, where they date to between ca. 100 and 70 ka (Mackay and Welz, 2008, Henshilwood et al., 2009, Watts, 2010, d'Errico et al., 2012a). They disappear in the following period and reappear in the same region, but engraved on ostrich eggshell (OES) used as flasks for water, ∼66–58 ka (Texier et al., 2013, Henshilwood et al., 2014), and later on Achatina shells (Bicho et al., 2018) and bone (d’Errico et al., 2012b). Abstract representations disappear later and are not found again at African sites until a few thousand years ago. The first African figurative art, discovered in the Apollo 11 cave in Namibia, is only 30,000 years old (Wendt, 1976, Rifkin et al., 2015). Although the subject of bitter discussion (Val, 2016, Dirks et al., 2016), the first African mortuary practices could be associated not with modern humans but with a population of Homo naledi. Discovered at the bottom of the cave of Rising Star in South Africa, this hominin with a small cranial volume may have survived until between 226 ka and 335 ka (Dirks et al., 2017, Hawks et al., 2017). The very few primary burials discovered in this continent are younger than burials of Neanderthals and modern humans found outside Africa (Pettitt, 2010).

What makes it difficult to interpret the first examples of complex technologies and symbolic behaviors from Africa as the direct consequence of a geographically and chronologically constrained emergence of H. sapiens is the fact that comparable cultural manifestations are now known in Eurasia before the arrival of modern populations in those regions (Soressi and d'Errico, 2007; Roebroeks and Villa, 2011, Joordens et al., 2015, Jaubert et al., 2016, Majkić et al., 2017, Aranguren et al., 2018, d'Errico et al., 2018, Hoffmann et al., 2018). In other words, the innovations that we encounter at archeological sites and in which we recognize the first traces of a cognition similar to ours, in particular symbolic manifestations, do not appear to be the direct result of a change related to the sudden emergence of a modern anatomy but rather the expression of complex and apparently nonlinear cultural trajectories (Johansson, 2015, Ackermann et al., 2016, d'Errico et al., 2017, Colagè and d'Errico, 2018, d'Errico and Colagè, 2018, Dediu and Levinson, 2018, Kissel and Fuentes, 2018, Scerri et al., 2018, Will et al., 2019). How these trajectories were, in different regions on the planet, conditioned by biological, environmental, and social factors as well as by migrations and diffusion of cultural traits remains to be explored.

The MSA-to-LSA transition in eastern Africa is marked by the development and proliferation of disk beads, particularly those made from OES. At present, eastern Africa has yielded the earliest occurrences of OES beads, with specimens directly dated to ca. 52 ka at Mumba Rockshelter in Tanzania (Gliganic et al., 2012), >50 ka at Magubike Rockshelter in southern Tanzania (Miller and Willoughby, 2014), 40.6 ka at Kisese II in central Tanzania (Tryon et al., 2018), and 39.9 ka at Enkapune ya Muto in southern Kenya (Ambrose, 1998). Variation in OES beads representation, size, and morphology through time and space has itself been the source of considerable debate in the Pleistocene through to the Iron Age. While bead sizes seem to generally decrease through the Pleistocene and early Holocene in southern Africa (Miller, 2019), sites in eastern Africa appear to show more heterogeneous patterns, and most sites in the interior appear to show little overall change through time (Miller, 2012, Miller and Willoughby, 2014, Biittner et al., 2017). Tryon et al. (2018) reported a slight reduction in disk bead dimensions through the sequence at Kisese II. In southern Africa, a sudden shift to larger bead sizes in the Late Holocene is thought to correspond to the arrival of new populations of herders with different stylistic preferences. So far, there have been very few data on any possible shifts in bead size through dynamic population changes now demonstrated through archeogenetics for the Middle to Late Holocene in eastern Africa (Skoglund et al., 2017, Prendergast et al., 2019). There are also few indications of periods of intensification of symbolic behaviors that may lend clues to the forces affecting their initial emergence.

Information on the earliest instances of bone technology in eastern Africa is scant. The bone tool and notational/ornamental assemblages documented in the MSA of southern Africa starting from 80 to 75 ka (d'Errico and Henshilwood, 2007, Backwell et al., 2008, Vogelsang et al., 2010, d'Errico et al., 2018), in North Africa at 90 ka (Bouzouggar et al., 2018), and in Central Africa at 95 ka (Brooks et al., 1995, Yellen et al., 1995) have for the moment no equivalent in eastern Africa. Bone implements first appear en masse in eastern Africa in the form of barbed bone points or ‘harpoons’ associated with aquatic resource intensification by foragers during the wet phase of the early Holocene (Yellen, 1998). Similarly, widespread bone ornament traditions are not visible until the Middle to Late Holocene (Mehlman, 1989, Helm et al., 2012, Langley et al., 2016). However, descriptions of five unserrated bone projectile points, a bone awl, and a bone notched piece from Kuumbi Cave on the Zanzibar archipelago of Tanzania hint at earlier manifestations of osseous technologies on the eastern African coast (Langley et al., 2016).

Panga ya Saidi is a large, partially unroofed cave complex located on the Nyali Coast of Kenya, 15 km from the present-day shore, at an altitude of 150 m (Fig. 1). It opens on the eastern flanks of the Dzitsoni Uplands, a ridge of Middle Jurassic limestone that separates the late Quaternary coastal plains from the large, arid Nyika Plateau. The site remained in close proximity to the coast throughout the Late Pleistocene because the sea depth at −125 m is within 5 km of the present shoreline. Ecologically, the Dzitsoni Uplands vegetation belongs to the Zanzibar-Inhambane Regional Mosaic. The site is situated at the limit between the Sokoke Forest, characterized by the association of Cynometra-Manilkara, and the Shale Savanna with Manilkara-Dalbergia. Climate models (Shipton et al., 2018) suggest that during the Last Interglacial, the Nyika Plateau received less rainfall than that received during the present interglacial, whereas the coastal area where the site is located witnessed higher precipitation in the Last Glacial Maximum (LGM) than at present. Excavations conducted at PYS between 2010 and 2013 have reached a depth of 3 m (Shipton et al., 2018). The sequence includes, from the bottom to the top, four lithological units (units I–IV) encompassing 19 archeological layers (Figure 2, Figure 3; Table 1). Although animal burrows, tree root channels, and termite galleries were recorded during the excavations, most stratigraphic boundaries could be identified along the entire profile. Twelve stratigraphically coherent accelerator mass spectrometry radiocarbon and seven optical stimulated luminescence (OSL) ages indicate that the sequence accumulated between Marine Isotope Stage (MIS) 5 (∼78 ka) and the historical era (0.4 ka). Researchers in charge of the OSL dating of the Diepkloof sequence (Tribolo et al., 2013; Jacobs and Roberts 2015) identify specific problems for which concrete solutions have been proposed. These are as follows: (a) performing dose recovery experiments on every sample, which was performed for six of the seven PYS OSL samples, and (b) the use of a standard dose rate model, which was also performed at PYS. Thus, the PYS data set tests avoid the problems identified at Diepkloof. Second, the OSL and 14C samples from layer 11 are in good agreement. Ideally, more levels would need to be dated with the two techniques, but the agreement observed in level 11 suggests that the PYS OSL ages are reliable.

Unit I (layers 19–17), featuring a relatively low occupation density, is followed, at the unit I/II interface (layer 17/16), by a possible depositional hiatus and limited human use of at least the part of the cave sampled. Anthropogenic signatures, in the form of ash and burnt bone, gradually increase within unit II (layers 16 to 14). This trend is interrupted by a sharp surface of erosional relief separating the top of this unit from the archeologically richer unit III (layers 13 to 5). This unit comprises numerous ash lenses interpreted as burning features. Stratigraphic interfaces interpreted as occupation surfaces with some degree of erosion are identified within unit III, in particular at the layer 13/12 and layer 10/9 boundaries. The overall sequence is as follows: layers 1–3 (Iron Age), layer 4 (earlier Holocene), layers 5–6 (terminal Pleistocene), layers 7–8 (Last Glacial Maximum), layer 9 (late MIS 3), layer 10 (mid-MIS 3), layers 11–12 (early MIS 3), layers 13–16 (MIS 4), and layers 17–19 (late MIS 5). A range of paleoecological indicators identify a persistence of more or less open tropical forest throughout the sequence. Within this general trend, several proxies suggest slightly drier conditions during the deposition of layers 16 (71–67 ka) to 8 (20 ka).

A significant increase in archeological finds and a change in stone artifact size, technology, raw material, and tool type are recorded after the unit I–II transition (73–67 ka). Lithic assemblages from unit I are primarily composed of typical MSA large flakes, mostly made of limestone, produced with the Levallois method, occasionally retouched into points. The shift to a largely predominant use of quartz (70–90%) in units II and III corresponds to an increased use of bipolar technology and a reduction in stone tool size. Chert use increases (40–60%) in unit IV. Backed crescents appear for the first time in layers 11 and 12 (48.5 ka) and are also found from layer 6 upward. They coexist in units II and IV with bipolar and Levallois technology, with blades becoming common in the upper part of the sequence (layers 8–3; 25–1 ka). The increase in archeological remains observed in units II–IV is tentatively interpreted by Shipton et al. (2018) as evidence for a growing human presence in the region, while the absence of unidirectional changes in lithic technology is seen as an indication that this region did not witness the sudden appearance of an LSA technological package.