Introduction
The number of research projects into the Middle Palaeolithic of southern Arabia (200–40 kya) has increased in recent decades, providing diverse archaeological data from surface scatters of stone tools and from stratified sites (Rose et al. Reference Rose2011; Delagnes et al. Reference Delagnes2012; Bretzke et al. Reference Bretzke2022). Palaeoenvironmental and archaeological data (Parton et al. Reference Parton2015) suggest an intensification and northward shift of low-latitude rain-bearing systems in Marine Isotope Stage 5 (MIS-5, c. 130–71ka). Middle Palaeolithic occupation of southern Arabia was likely widespread, but isolation events may have occurred later during the more arid MIS-4 (c. 71–57 ka).
The Huqf anticline in central Oman (Figure 1) has remained a blank spot with regard to Middle Palaeolithic occupation, even after a series of intensive surveys (Jagher & Pümpin Reference Jagher and Pümpin2010). Despite the abundance of high-quality chert, natural springs and freshwater lakes in this area during the Late Pleistocene (129–11.7 kya) (Rosenberg et al. Reference Rosenberg2011), Middle Palaeolithic finds are elusive. This study contributes to the infilling of this knowledge gap by using a multi-scalar approach to analyse locations not previously surveyed.
Map of the Duqm area showing the location of WB3 and WB4 and other relevant archaeological sites (base map (a) GEBCO 2022 grid (http://gebco.net); (b) SRTM 1 Arc-Second, generated using QGIS v.3.26 Buenos Aires; figure by authors).
Research questions and method
Our project aims to assess the timing and nature of human occupation in the Late Pleistocene of southern Arabia to establish whether it was episodic and refugial or widespread and interconnected. Survey methods used to locate Palaeolithic sites build on desk-based assessment of potential locations derived from geological maps, field reports and geomorphological assessment of the landscapes of south/central Oman. Following the identification of Middle Palaeolithic finds, we employed a targeted collection of cores and diagnostic finds. Detailed techno-typological analysis of the lithic sample allowed comparison with other Middle Palaeolithic sites in the region and surrounding regions (East Africa, Levant, Middle East).
Results
The expedition recorded two new sites exhibiting Middle Palaeolithic characteristics in the area known as Wadi Baw East, south-west of Duqm, southern Huqf. Wadi Baw 3 (WB3) is a very low-density archaeological site on top of an inselberg without raw material outcropping (Figure 2A & B). The collected assemblage (Table 1) includes one heavily weathered preferential centripetal Levallois core, two unidirectional flat cores and debitage. Wadi Baw 4 (WB4) is a large (length >100m) and relatively dense (>30 artefacts/m2) lithic scatter located on a slightly elevated limestone ridge with outcropping chert nodules at its base and flanks (Figure 2C & D). Both sites are elevated and are most likely deflation surfaces caused by wind erosion of fine material. There may have been potential movement of coarser material in the past wet phases.
Images of the archaeological sites of WB3 (A & B) and WB4 (C & D) (figure by authors).
Number of finds and weathering patterns.
Lithics at both sites exhibit specific weathering patterns indicating two different post-depositional trajectories (Table 1). Artefacts showing a high degree of ridge rounding, edge damage, surface erosion, potlids and generally dark manganese coating fall within the Middle Palaeolithic technological spectrum. Artefacts presenting an advanced desert varnish, with moderately sharp ridges and moderate edge damage, can be classified as blades and bifaces (Figure 3). The raw material used to manufacture both sets of lithics is the same and—while not in situ as evident from the edge damage and rounding observed on the specimens—the Middle Palaeolithic finds are clearly discernible from the later, less-weathered blade/biface assemblage. Recycled, double-patinated artefacts (Figure 3E) further support the suggested relative chronology.
Weathering patterns on artefacts from WB3 and WB4: A–D) Middle Palaeolithic finds with advanced ridge and edge abrasion, pits and surface dissolution; E) Middle Palaeolithic find with traces of reuse; F–H) Early Holocene material (figure by authors).
Most of the Middle Palaeolithic cores might be categorised as unidirectional parallel and orthogonal flat cores for flakes without or with only minor preparation of striking platforms and convexities of exploited surfaces (Table 2). Five cores might be classified as Levallois (sensu stricto) and these include three preferential centripetal Levallois cores, one preferential bidirectional core and one small Nubian Levallois Point core (Figure 4). Besides the core reduction strategies, the assemblage contains one heavily weathered, small (up to 100mm) possible biface (Figure 4, no. 9) and three bifacial thinning flakes.
Wadi Baw 4 core types.
WB4 artefact sample: 1–3) unidirectional parallel cores; 4) orthogonal core; 5–7) preferential Levallois cores; 8) Nubian Levallois core; 9) possible biface (figure by authors).
At both sites, the lithics exhibiting typical desert varnish weathering relate to the production of blades from unidirectional single platform cores and small to medium-sized biconvex and bi-pointed bifaces. Similar industries are well attested in the area (Jagher & Pümpin Reference Jagher and Pümpin2010) and might be tentatively dated to the terminal Pleistocene/mid-Holocene (14–6 kya) based on analogous sites in Dhofar (Hilbert Reference Hilbert2014).
Discussion
WB3 and WB4 represent the first evidence of Middle Palaeolithic activity in Huqf. The lithic assemblages exhibit technological variability and weathering heterogeneity, indicating a palimpsest of Pleistocene and Holocene occupation phases. Surface erosion has affected both sites and subjected the lithics to mechanical weathering since their deposition. Subsequent exploitation of the site as a raw-material quarry, possibly by Holocene groups, has resulted in the recycling of larger Middle Palaeolithc artefacts, introducing bias to the assemblage.
Despite this, some preliminary observations can be made. The small Nubian Levallois core from WB4 might indicate a tenuous link to the Nubian technocomplex in Arabia, which needs to be explored further in the coming seasons; the rest of the assemblage, however, is dissimilar from Nubian sites in adjacent Dhofar. Centripetal Levallois cores are rare within Dhofar Nubian assemblages (Rose et al. Reference Rose2011) but are reported from south-western and central Arabia (e.g. Bretzke Reference Bretzke, Conard and Delagnes2015; Crassard & Hilbert Reference Crassard and Hilbert2020). Additionally, the majority of the Middle Palaeolithic WB4 assemblage shows simple, unidirectional, occasionally orthogonal flake production. Weathering of the Levallois components suggests similar exposure periods to post-depositional agents as the non-Levallois (flat) cores; these technologies, however, are seldom reported except at Jebel Faya (Bretzke Reference Bretzke, Conard and Delagnes2015).
The lithics from WB3 and WB4 may indicate a cultural separation of Huqf and possibly a larger part of eastern Arabia, with only infrequent Nubian production sites (Beshkani et al. Reference Beshkani, Beuzen-Waller, Bonilauri and Gernez2017). This separation could be attributed to the geographical barrier of the Jiddat al-Harasis desert, as suggested elsewhere (e.g. Hilbert Reference Hilbert2014; Brezke Reference Bretzke, Conard and Delagnes2015), allowing only sporadic visits rather than continuous occupation. Work is in progress to establish a robust chronostratigraphic framework for these sites through a multitechnique dating approach.
Acknowledgements
We are deeply grateful to the Ministry of Heritage and Tourism of the Sultanate of Oman for providing permission and support for fieldwork. The research was conducted in collaboration with the Czech expedition ARDUQ (Archaeological Landscapes of Duqm and Dhofar) active in Duqm. We thank the reviewers for their insight and comments that have greatly improved this manuscript.
Funding statement
Funded by Australian Research Council project DP230100152 and Praemium Academiae of the Czech Academy of Sciences. Yamandú H. Hilbert is financed in the frame of the REVIVE project, European Research Council grant #101001889. Mathieu Duval is supported by the Spanish Ramón y Cajal Fellowship RYC2018-025221-I, funded by MCIN/AEI/ 10.13039/501100011033 and ‘ESF Investing in your future’.
Open access
