Fig 1.
Panga ya Saidi site location and excavation areas.
a) Regional map showing the location of PYS in relation to other sites mentioned in the text. Author created map using base map data from Natural Earth (public domain); b) Local elevation map, elevation data derived from the JAXA ALOS World 3D 30 m Digital Elevation Model (AW3D30), Version 4.1 [14], and shoreline data [15]. Map created by the authors complying with JAXA Terms for Use. b) Photograph of main excavation area, under shelter overhang.
Fig 2.
Plan of the southern part of the PYS cave system, including the excavated areas (red rectangles), and auger test.
a–c) Cross section of selected locations, with human scale (180 cm) for reference. Cave entrances (to the south and northwest) are marked with arrows. The cave system continues north of the map edge.
Fig 3.
3D models of selected wall sections within PYS.
Northern (pink, a–c) and central (yellow, d–f) eastern walls. Each section was represented using x-ray style (a, d) and radiance scaling rendered shaders (b–c, e–f). The northern section shows a cave chamber with three separate openings and a roof consisting of a series of cupolas. The central east section displays similar cupolas with sporadic stalagmites and stalactites.
Fig 4.
Chamber and wall morphology in PYS.
a) Unroofing resulting from ceiling collapse. b) A cluster of deep, tapering-up cupolas with densely scalloped surfaces in one of the rear alcoves.The cupola at the center of the photograph has been breached and is open to the overcave surface. c) Meter-scale cupolas. d) Meter-scale cusps on alcove wall. e) Hemispherical hollows on bedrock limestone, perhaps resulting from condensation corrosion. f) Centimeter-scale grooves ca. 3.5 m above the northwestern entrance to the cave.
Table 1.
Contexts from the 2020 excavation. Description and correlation with previously identified layers, in stratigraphic order.
Fig 5.
Stratigraphic profile of the main excavation area in 2020.
Top left: North-facing view of newly excavated area (2020), find density per 8 cm² estimated using the Kernel Density method in ArcMap 10.5.1, and plotted finds by material, recorded in situ with a Leica Viva TS16 robotic total station, (visualized in ArcGIS). Top right: relative abundance (find density per 8 cm2) of ceramics, lithics, shell, and bones, visualized using the Kernel Density method in ArcMap 10.5. Bottom: Western and northern wall sections, and correlation with previously identified layers. For detailed description of contexts refer to Table 1. All scale bar segments are 20 cm.
Fig 6.
Composition and structure of correlative deposits from Trench 4 (for their spatial relationship with the 2020 excavation, see Fig 2). a–c: Flatbed scans of uncovered thin sections; d–g: photomicrographs (PPL). a) Transition between Layer 8 (correlative with Context 011: the base of the 2020 excavation) and the underlying Layer 9 (not excavated in 2020). Yellow triangle indicates fining-up domains. In Trench 4, the line of ash intraclasts (AI) and clusters of closely-packed coarse particles at mid-section (yellow lines) could be traced laterally to ash deposits and coarse limestone clasts (spall) at the top of Layer 9. b) Boundary between coarser Layer 6 (Context 090) above and Layer 7 (Context 010) – finer-grained sandy loam (SL) and even finer sandy loam with reworked ash (SLa) – below. At the lower part of the section (Layer 7: SLa), a large intraclast of laminated ash (AI) with inclusions of charcoal (black) and heat-reddened sediment (rip-up clasts: reddish-brown) adheres on a large bone fragment (Bn). The linear arrangement of (locally imbricated) platy and ovoid particles (yellow lines) and the fining-up domains (yellow triangles) are interpreted as primary depositional structures. The rounded ash intraclasts (AI) at mid-section were cemented, possibly due to exposure on the cave floor, before their erosion and redeposition, presumably from further upslope. Lt, an angular, pointy fragment of coarse-crystalline quartz, is interpreted as stone-knapping refuse. c) Transition between Layer 5 (Context 007) above and Layer 6 (Context 008) below. Note the poor sorting, linear arrangement of many platy particles (yellow lines), and fining-up domains in parts of the section (yellow triangles). Pores Pic (the larger one highlighted) are bioturbation channels filled with illuvium (see Fig 6g). The lining of pore Plc (arrows) is compositionally and microstructurally similar to that of Pic and is therefore also intepreted as illuvial in origin. Pore Pmv has been deformed by wetting and drying, but lacks illuvial filling. d) Ash intraclast at the boundary between layers 8 and 9: laminated, partly phosphatised and wood ash with inclusions of burned bone (Bn) and mineral grains (mainly quartz and ferricrete), and dark organic or organomineral (iron-manganese?) impregnation spots (Is). (Arrows: recrystallisation of ash to clear calcite microspar) e) Fining-up domain in Layer 6: Well-rounded coarser particles (bedrock limestone, intraclasts, pisolites) set in a poorly-sorted finer matrix (porphyric c/f-related distribution). The coarser particles become progressively smaller, less frequent, and more openly-spaced up-section. f) Detail of inferred stone-knapping refuse in Layer 6. g) Very poorly-sorted fill of bioturbation channel in Layer 5: grains of quartz (Qz), limestone (LS) and ferricrete ‘float’ in a matrix of ferruginised silty clay. The channel fill resulted from illuvial deposition by relatively high-energy percolating water, and may include particles translocated from overlying horizons.Key: PL: iron (+manganese?) pisolite; FC: ferricrete; PC: pedoclast (mainly red clay with quartz silt); Qz: quartz; IC: intraclast; RF: rock fragment (cave-wall limestone); LS: limestone; ST: speleothem; Lt: stone-knapping refuse; Bn: bone; Sh: shell; Ch: charcoal; IS: impregnation spot; P: pore (ic: infilled channel; lc: lined channel; mv: multiconcave vugh).
Table 2.
Radiocarbon ages from the 2013, 2017 and 2020 excavations at PYS.
Fig 7.
Bayesian modelled ages for PYS Layers 1-8, and transitions.
See S1 File for OxCal code.
Fig 8.
Relationship between the evolving cave chamber and floor morphology, sediment architecture and occupation intensity at PYS over the last ca.
25.5 ka.