Inside the Denisova Cave in Siberia

Figure 3

Summary stratigraphic logs of the sequences exposed in (a) DCM and (b) DCE, showing the locations of the micromorphological samples and key microstratigraphic features. To the right of each log, optical ages (in ka, with uncertainties at 95.4% probability) are shown for the major boundaries between lithological units in the thin sections, together with the associated archaeological phases (from ref.9).

In DCE (Fig. 3b), we record trace quantities of fine charcoal fragments and flecks closely associated with crushed charcoal and bone fragments in layer 16, and more commonly in layer 15—the earliest layer containing artefacts (early Middle Palaeolithic)9 and also Denisovan DNA10. This indicates fire use and trampling occurring tentatively from 259 ± 28 ka and certainly from 203 ± 14 ka, but we cannot unequivocally and directly link the manufacturers of the stone tools with fire-use because these fine combustion products are highly mobile. In layer 14, which yielded Neanderthal DNA and early Middle Palaeolithic artefacts and was deposited between 193 ± 12 and 187 ± 14 ka, there is a marked increase in micro-charcoal that imparts a dark colouration, and also a small piece (3–4 mm) of angular chert debitage in our sample.

We identify large charcoal fragments (>4–5 mm) in layers 11.4 and 11.3 in DCE, which were deposited between 120 ± 11 and 70 ± 8 ka and contain middle Middle Palaeolithic artefacts. A Neanderthal toe phalanx (Denisova 5) was recovered from layer 11.4, but we cannot confidently associate this and other similarly small and isolated hominin fossils with elements of the sedimentary matrix, given the possibility of displacement9,12. In thin section, we observe sediment movement in the form of micro-faulting and slippage features in these layers, presumably associated with the aforementioned post-depositional subsidence. However, such deformation processes do not necessarily promote the translocation or mixing of fine material across lithological boundaries. Our sample location was close to the rear of DCE, where the chamber tapers to a narrow slot (~1 m wide), a locale unlikely to have been conducive to human occupation—especially the lighting of a fire—given the confined space. Evidence of sediment compaction does suggest compressive forces, however, so animals that are represented in the faunal record may have been present in this restricted space.

Ashes are present in very low quantities in the Denisova Cave microstratigraphy. Nevertheless, we cannot rule out ash dissolution as the biasing factor, given the mobility of calcite and decalcification recorded locally in some layers. Could fire have been used at Denisova Cave more extensively by hominins, but with the associated evidence subsequently removed from the stratigraphic record? Bearing in mind the rate of cave sedimentation, erosional (chronological) gaps and the evidence for bioturbation in some parts of the sequence—mostly parts of the upper layers of DCE—reworking and redistribution of combustion bi-products may have occurred, although it is unlikely that all micro-traces would have been completely removed. The reworking of previously in situ fire residues is supported by the absence of structured combustion features that would signify the presence of intact hearths. Furthermore, stone tools do not exhibit signs of thermal alteration24, which might be expected should fire-use have been common—or even present—in these confined spaces, and other indicators of fire, such as thermally altered clays, were not evident in our samples. Although fire may not have been used extensively within the sampled areas of the cave, the lack of an obvious pyrotechnology need not preclude the use of a site by hominins, even during glacial periods25. Elsewhere in the Altai, the site of Kara-Bom contains well-preserved hearths in the Initial Upper Palaeolithic deposits26, but no clear evidence of fire-use has been found in the region beyond about 50–40 ka27.

Overall, the microstratigraphic record for Denisova Cave indicates that human activity was intermittent over the past three glacial–interglacial cycles represented by the Pleistocene sedimentary infill (>300 ka to ~20 ka). The stone artefact assemblages indicate long-term hominin occupation of the site during both warm climates and cold conditions, when the foothills of the Altai Mountains likely served as a refugium28.

Other animal users of Denisova Cave: the fossil coprolites
Coprolites are common biogenic components of the cave sediments, often present in dense concentrations, suggesting that animals visited the site for much of its depositional history. The coprolites can be grouped into a number of recurring types throughout the sequence, presumably reflecting the use of the site by a variety of animals, and potentially associated with a range of preservation states. We recognise four main coprolite types (CT-1 to CT-4), described in Table 1 and shown in Fig. 4. Although we cannot confidently attribute all of these droppings to a specific animal, we assign CT-1 to Crocuta crocuta spelaens (cave hyena), based on consistency with published results describing the morphology and optical properties of this material in thin section29,30,31; this supports the faunal evidence of regular use of the cave by these animals6,7,8,9. We tentatively attribute CT-2 to wolf (Canis lupus), based on the similarity between these coprolite fragments and dog coprolites recorded at Vanguard Cave, Gibraltar32, as well as other published data31. The coprolite fragments (CT-3) in our thin section of layer 12.2 in DCM are consistently larger and darker than CT-1 and CT-2. This layer contains very high proportions of these coprolites, and the chaotic arrangement of the coarse limestone gravel, with long axes in a vertical to sub-vertical alignment, suggests disturbance of these sediments, possibly by a large animal such as a cave bear. We cannot assign CT-3 or CT-4 to a specific species.

Figure 4



Examples of coprolites identified in the Denisova Cave microstratigraphic record (see Table 1 for coprolite descriptions). (a–c) Type CT-1 originating from cave hyena occurs through much of the sampled sequence; (d–f) Type CT-2 has a much darker matrix, possibly related to wolf; (g,h) Type CT-3 is highly weathered; (i) Type CT-4 has a distinctive red matrix. Types CT-3 and CT-4 cannot be linked to specific animals. Scale bars: red, 800 µm; green, 500 µm; yellow, 1 mm; blue, 2 mm.

Coprolite fragments commonly occur in layers that also contain stone artefacts. Given that hominins and hyenas will not cohabitate7,33, this indicates that specific occupation events will be extremely challenging—if not impossible—to tease out at Denisova Cave, with the slow sedimentation rates effectively precluding the identification of alternating hominin–carnivore occupations, should they exist. The lack of defined stratigraphy within layers (e.g., buried surfaces) may be due, at least in part, to carnivore denning or other animal burrowing activities in parts of the deposit. At Bois Roche in France, for example, stone tools accumulated by local movement (e.g., by gravity) in areas that functioned primarily as carnivore dens34. The DCE faunal record includes the remains of a number of large cave-dwelling mammals recovered in relatively high numbers, including hyena, wolf, red fox and, to a lesser extent, bear9,35. As these animals are unlikely to have cohabitated, either with each other or with hominins, the co-occurrence of their remains likely reflects the scale of resolution (time averaging) of the sedimentological, chronological and hominin occupation records at Denisova Cave.

The presence of coprolites in layers from which hominin remains and aDNA have been recovered implies that large carnivores might be an accumulating agent for these materials, particularly in areas of the site where evidence for hominin activity is scarce (e.g., the farthest recesses of DCE). Specific areas of the site might have been designated as waste dumps for lithic debitage and food detritus, for example, which in turn attracted scavengers such as the cave hyena when hominins were absent from the site. Interestingly, we record in thin section only a few examples of bone fragments that exhibit characteristic etching related to digestion in the gut of a carnivore, although etched bones are common in the faunal record35.

In the field, rodent burrows (krotovinas36) are clearly visible in the Holocene deposits of DCE, and in fewer numbers in DCM. Disturbance of the sediments by bioturbation is also evident in thin section. Parts of layer 12 and much of layer 13 in DCM display a chaotic arrangement of limestone clasts within finer material, consistent with disturbance by large animals such as bears, wolves or hyenas. This accords with field observations of layer 13 being a hyena lair9. Thin sections of layers 9.2 and 9.3 in DCM and layers 9.1, 11.3 and 11.4 in DCE display abundant, loosely arranged aggregates and irregular vughs typical of bioturbation37. These small, mm-size features are typical of smaller soil fauna, such as worms, spring-tails (Collemboles) and isopods. We note that these fine crumb structures occur essentially in the uppermost Pleistocene strata in DCE (i.e., layer 9), which accumulated after 38 ± 9 ka and may represent milder conditions that enabled these fauna to flourish.