The Pliocene-Pleistocene cave deposits in the Cradle of Humankind World Heritage Site (South Africa) preserve a diversity of hominin fossils in a varied set of contexts (Hughes and Tobias, 1977; Clarke, 1998; Partridge et al., 2003; Berger et al., 2010; Bruxelles et al., 2014; Berger et al., 2015). Hominin remains in the area are generally encased in lithified clastic deposits in caves that are situated in stromatolite-rich dolomite of the Malmani Subgroup (Eriksson et al., 2006) (Figure 1). Sedimentological and taphonomic descriptions of notable fossil sites (Brain, 1981; de Ruiter et al., 2009; Dirks et al., 2010; Pickering and Kramers, 2010; Pickering et al., 2011a, 2011b) indicate that fossils were trapped and preserved in caves as a result of a range of processes including death traps, scavenging, mud flows and predation. Distribution patterns of fossiliferous caves in the area suggest that fossil deposition occurred in caves that are close to critical resources such as water (Reynolds et al., 2011; Dirks and Berger, 2013).
The Rising Star cave system lies in the Bloubank River valley, 2.2 km west of Sterkfontein cave. It comprises an area of 250 × 150 m of mapped passageways situated in the core of a gently west dipping (17°) open fold, and is stratigraphically bound to a 15–20 m thick, stromatolitic dolomite horizon in the lower parts of the Monte Christo Formation (Eriksson et al., 2006; Figures 1B and 2A). This dolomite horizon is largely chert-free, but contains five thin (<10 cm) chert marker horizons that have been used to evaluate the relative position of chambers within the system (Figure 2B). The upper contact is marked by a 1–1.3 m-thick, capping chert unit that forms the roof of several large cave chambers. Surface mapping indicates that cave openings and flowstone-filled fractures do not penetrate this capping chert unit, except where a dextral fault truncates the stratigraphy to the south of the system (Figure 2). The network of cavities is developed along west-northwest, north and northwest trending fractures and joints.
The Dinaledi Chamber is exclusively filled by flowstone and fine-grained sediment involving two depositional facies distributed across three stratigraphic units that filled the chamber over time. Stratigraphic units are separated by erosional unconformities, or laterally continuous flowstone intercalations. Erosion remnants of the units occur in a variety of stratigraphic positions, and there is extensive evidence of reworking with older units being re-deposited into younger units. Each of the facies and units, and each of the flowstone phases that are interlayered with the various units, are described below (Figures 3, 4).
Facies 1 has been sub-divided into two sub-facies. Facies 1a consists of unlithified, horizontally laminated, orange mud, with very low sand content (Figure 4A,B). The composition is dominated by fine sericite clay with subordinate amounts of silt-sized chert and dolomite grains (Figure 5). Facies 1a is generally unconsolidated, but contains secondary Mn- and Fe- oxide phases that locally form weakly cemented concretions. Facies 1a has a patchy distribution, occurring both in undisturbed, isolated areas as accumulations atop blocks and in fissures (e.g., Figure 4A), and more commonly as erosional remnants of formerly more extensive deposits that filled the Dinaledi Chamber and side passages. Outcrops of Facies 1a (e.g., near the entry point at the top of the chamber, Figure 4F) show evidence of in situ auto-brecciation of orange mudstone around exposed margins, due to desiccation, and/or formation of Fe-Mn concretions.
Upon initial discovery of the Dinaledi Chamber, Unit 3 contained dozens of hominin bones exposed along its surface and partially buried within it, in most portions of the chamber where Unit 3 has formed (Figure 6). A series of minor test pits (a few cm deep) were dug into Unit 3 at various areas of the cave floor, which revealed that abundant additional hominin bones are buried at shallow depths within Unit 3 throughout much of the chamber. The abundance and density of H. naledi bones in the chamber is demonstrated in the single excavation, where most (1250 out of 1550 elements) of the fossil material documented from the Dinaledi Chamber was collected (Figure 7). Unit 3 also contains rare disarticulated rodent remains that are undiagnostic of age, and that are possibly derived as an erosional product from Unit 1 (Facies 1b) or that were deposited directly into Unit 3 as it accumulated.
A detailed taphonomic study of the hominin remains is ongoing, and initial results have been provided in this study to illustrate the broader context of the fossil assemblage. Analyses of taphonomic processes affecting the Dinaledi hominin fossil assemblage have been conducted to describe decomposition, weathering and fracture patterns, surface modifications as a result of invertebrate–bone interactions (as well as the lack of vertebrate–bone interactions), and spatial context including skeletal distribution patterns (Behrensmeyer et al., 1986; Galloway, 1999; Straus and Porada, 2003; Loe, 2009; Symes et al., 2013; Figures 7, 9–12). Summary tables of results are presented in Table 1 and Supplementary file 2.
Using the bone elements recovered to date, we have analysed the composition of the H. naledi assemblage, based on the observed frequency of various elements as a function of the expected sequence of disarticulation of various skeletal regions, as well as intrinsic survival characteristics of elements (Lyman, 1994; Pokines and Symes, 2013). Table 1 presents the Minimum Number of Individual and Identifiable Elements (MNIE) for the hominin skeletal material from the Dinaledi Chamber. This table is conservative and only lists elements that are clearly duplicated. Dental remains have not been included.
The hominin assemblage is homogeneous in terms of surface preservation and condition (Figures 10 and 11), suggesting that the remains share a similar depositional history (i.e., the disarticulated vs articulated material does not vary significantly in terms of surface preservation). The structural state of the material is classified as good, and surface morphology is retained for many of the specimens (i.e., no plastic deformation has been observed), even though they were water logged and friable at recovery. The bones are generally, partially mineralised; there is no evidence of calcite crystal formation in or on bones, but some bones and teeth are dotted with black iron-manganese oxi-hydroxide deposits and coatings (e.g., Figure 10), and an orange-colour residue of iron oxide (e.g., Figure 10B). Colouration of the bone underlying surficial mineral deposits ranges from light grey to red-brown. The internal structure of bones is bright white in colour.
The skeletal assemblage of H. naledi displays little variation in surface structure and condition, indicating that the hominin material has been exposed to a limited range of environmental fluctuation during its depositional history. Analysis of the weathering patterns of the Dinaledi assemblage indicates that most of the bones show weathering stages 1 or 2 (after Behrensmeyer, 1978); while 18% have an etched appearance and 11% exhibit dissolution (Supplementary file 2). No element exceeds weathering stage 3 with the majority of elements displaying stage 1 weathering patterns. Most elements show no signs of cracking or flaking, although some deeper cracks may be present due to post-depositional effects. Stage 1 weathering is evidenced in most specimens of long bones, rib fragments, and mandibular fragments as fine longitudinal cracks (Figure 10D), without concomitant flaking and the formation of a fibrous texture.
Considering the geological and taphonomic context of the Dinaledi Chamber, the occupation, predator accumulation and water transport hypotheses cannot adequately explain the fossil assemblage. Both the mass mortality or death trap scenario (although possibly not involving a single event) and deliberate disposal hypothesis are considered plausible interpretations and require additional investigation. Based on current evidence, our preferred explanation for the accumulation of H. naledi fossils in the Dinaledi Chamber is deliberate body disposal, in which bodies of the individuals found in the cave would either have entered the chamber, or were dropped through an entrance similar to, if not the same as, the one presently used to enter the Dinaledi Chamber. Reconstructions of the cave environment indicate that reaching even the entrance of the Dinaledi Chamber would always have been difficult, particularly in the absence of artificial light.
Our interpretation of events raises questions about the meaning of deliberate and repeated body disposal to this ancient group of individuals. Recent evidence has extended the record of complex behaviour from archaic and modern humans (Haglund, 1993; Arsuaga et al., 1997; Carbonell and Mosquera, 2006) to earlier hominins (McBreaty and Brooks, 2000; Douka and Spinapolice, 2012; Joordens et al., 2015). Deliberate disposal of bodies in the Dinaledi Chamber implies that morphologically primitive hominins like H. naledi may have had their own distinctive patterns of behavioral complexity, even though the reason why H. naledi may have ventured deep into the cave system remains unresolved. This leaves the important question of how old the H. naledi remains are. At this point we do not want to speculate on the age of the deposit considering the reworked nature of the sediments resulting in mixed stratigraphic signatures that impede faunal dating of the fossil rodent remains, and the limited amount of clean flowstone suitable for U-Pb dating (Pickering et al., 2011a, 2011b). Further method development is underway to circumvent this problem.
DOI:
dx.doi.org/10.7554/eLife.09561.001 elifesciences.org/content/4/e09561