Research ArticlePALEONTOLOGY

Human footprints provide snapshot of last interglacial ecology in the Arabian interior

See allHide authors and affiliations

Science Advances  18 Sep 2020:
Vol. 6, no. 38, eaba8940
DOI: 10.1126/sciadv.aba8940


The nature of human dispersals out of Africa has remained elusive because of the poor resolution of paleoecological data in direct association with remains of the earliest non-African people. Here, we report hominin and non-hominin mammalian tracks from an ancient lake deposit in the Arabian Peninsula, dated within the last interglacial. The findings, it is argued, likely represent the oldest securely dated evidence for Homo sapiens in Arabia. The paleoecological evidence indicates a well-watered semi-arid grassland setting during human movements into the Nefud Desert of Saudi Arabia. We conclude that visitation to the lake was transient, likely serving as a place to drink and to forage, and that late Pleistocene human and mammalian migrations and landscape use patterns in Arabia were inexorably linked.


Southwest Asia represents the main biogeographical gateway between Africa and Eurasia (1) and is, therefore, key for understanding hominin and faunal dispersal and evolution across continents. Fossils of Homo sapiens are first recorded outside of Africa ~210 and ~180 thousand years (ka) ago in southern Greece (2) and the Levant (3), respectively, and recent fossil evidence demonstrates the arrival of our species in the Arabian interior by at least ~85 ka (4). Understanding the nature of early H. sapiens out-of-Africa dispersals, however, has remained challenging owing to the poor resolution of paleoenvironmental and paleoecological data in direct association with remains of the earliest non-African people (5).

Here, we report hominin and non-hominin mammal footprints and fossils from the Alathar lacustrine deposit in the western Nefud Desert, Saudi Arabia (Fig. 1). We argue that the footprints, dated to the last interglacial and therefore contemporaneous with an early H. sapiens out-of-Africa dispersal (Fig. 2) (6), most likely represent the earliest evidence of our species in the Arabian Peninsula. The unique setting and taphonomic factors affecting the long-term preservation of footprints mean that groups of footprints, and particularly those in similar states of preservation, can be assumed to have been generated within a very short window, typically within a few hours or days (see full discussion in text S5) (79). An experimental study of modern human footprints in mud flats found that fine details were lost within 2 days and prints were rendered unrecognizable within four (7), and similar observations have been made for other non-hominin mammal tracks (9). The findings presented here, therefore, provide a unique opportunity to examine the close ecological interplay between late Pleistocene humans, animals, and their environments as our species began to venture into Eurasia.

Fig. 1 The location of the Alathar paleolake site.

(A) Map showing the location of the site within the western Nefud Desert, Saudi Arabia. (B) Three-dimensional oblique map of the site and location of tracks, fossils, and optically stimulated luminescence (OSL) samples.

Fig. 2 Critical dates relating to fossil evidence for H. sapiens outside of Africa during MIS 5 compared with the bracketed age for the Alathar footprints.

(A) LR04 stack (top) (blue line) displaying the marine δ18O record (62). (B) Dated evidence for H. sapiens occupation of the Levant and the western Nefud Desert based on fossil evidence. Ages for occupation of the Levant are based on thermoluminescence (TL) ages from Skhul (63) and Qafzeh (64); see (65) for more detailed discussion. Al Wusta lake and fossils represent the Bayesian modeled age for units 2 and 3 (carbonate lake sediment and waterlain sands) (4). (C) OSL ages for the two lake samples from Alathar between which the hominin and faunal footprints lie (this study). (D) Summer insolation at 30°N (bottom) (orange line) (66).


Geology and geochronology

The Alathar paleolake deposit lies as an inverted relief feature within an interdunal depression in the south-western portion of the Nefud Desert sand sea (Fig. 1). The sedimentary sequence comprises a ~1.8-m-thick deposit of sandy-silt diatomite (units 2 to 7) underlain by wind-blown sands (unit 1) (Fig. 3 and text S1). Two paleolake sections were recorded at the site and a composite section formed from both (Fig. 3). The first section (units 2 to 4) is exposed along the south-western margin of the deposit, and the footprints were found in the uppermost portion of unit 4b (i.e., the surface). A small number of fossils (n = 4) were also found eroding out of the surface of this unit. The second section (units 5 to 7) is located along parts of the western edge and toward the center of the modern-day surface of the paleolake deposit and stratigraphically overlies the footprint-bearing unit. The stratigraphic relationship between the two sections was visually correlated in the field based on lithostratigraphy. Samples for optically stimulated luminescence (OSL) dating were taken from sediments directly below (unit 2; PD61) and above (unit 5; PD62) the footprint-bearing unit and yielded ages of 121 ± 11 and 112 ± 10 ka, respectively, effectively bracketing the age of the prints (Fig. 3 and text S2). Diatom paleoecology and sedimentary analysis indicate that Alathar was an oligotrophic (nutrient-poor) and shallow freshwater lake for the majority of its existence (text S1). This is consistent with similar-aged nearby freshwater paleolake deposits (4, 1011) situated in the southern reaches of a “freshwater corridor” that connected the Arabian interior to the Levant and northeast Africa at times during marine isotope stage 5 (MIS 5; ~130 to 80 ka) (12). The presence of freshwater lakes in the western Nefud Desert provided a vital resource and habitable landscape for hominins and animals.

Fig. 3 Full sedimentary sequence from Alathar with location of OSL dates shown by stars.

The first section (bottom) (units 1 to 4b) lies stratigraphically below the second section (top) (units 5 to 7). The two sections were located ~30-m apart (see “OSL samples and strat logs” in Fig. 1), and the stratigraphic relationship between them is easily traced in the field. All footprints are located on the top of the first section (unit 4b) but beneath the second section. Photo credit: Richard Clark-Wilson, Royal Holloway, University of London.

Fossil and footprint evidence

The freshwater lake at Alathar attracted a variety of large mammals, as represented by hominin, elephant, equid, and bovid tracks and fossils (Figs. 4 and 5, and text S5). The lake surface is heavily trampled, which probably reflects a dry season during which herbivores congregate around diminishing water supplies (1314), and is consistent with sedimentary evidence for the drying up of the lake at that time (text S1). A total of 376 footprints were recorded, of which 177 could be either confidently or provisionally referred to an ichnotaxon. Seven hominin footprints were confidently identified, and given the fossil and archeological evidence for the spread of H. sapiens into the Levant and Arabia during MIS 5 (4, 6, 1519) and absence of Homo neanderthalensis from the Levant at that time (20), we argue that H. sapiens was responsible for the tracks at Alathar (text S5). In addition, the size of the Alathar footprints is more consistent with those of early H. sapiens than H. neanderthalensis (fig. S9 and text S5).

Fig. 4 Digital elevation models of three selected hominin tracks (HPR001, HPR002, and HPR003).

Photo credit: Klint Janulis, University of Oxford.

Fig. 5 The Alathar paleolake sediments, footprints, and fossils.

(A) Plan view of the Alathar paleolake deposits with researchers indicated by white arrowheads. (B) First stratigraphic section (units FS1 to FS3). (C) Second stratigraphic section (units SS1 to SS3) overlying the first but located toward the center of the paleolake. (D and E) Example of an elephant track and trackway, Proboscipeda isp. (F) Camelid trackway, Lamaichnum isp. (G) Camelid forefoot (H) Camelid hindfoot. (I) Equid track, Hippipeda isp. (J) Bovid axis vertebra eroding out of the paleolake sediment. Photo credit: Gilbert Price, The University of Queensland and Richard Clark-Wilson, Royal Holloway, University of London.

Four of the human prints (HPR001 to HPR004) were found adjacent to one another along the south-western edge of the paleolake exposure (Figs. 1 and 4). Given their similar orientation, distances from one another, and differences in size, they are interpreted as two, or up to three, individuals traveling in concert. Although the human track sample is small, three important observations can be made from the trace and body fossils preserved at the site. First, the tracks are scattered across the paleolake and orientated in various directions, indicating that they were not simply traversing the lake but were engaged in nondirectional activities. Second, the human tracks, similar to many of the animal tracks, indicate movement in a mostly southward direction. Last, the body fossils recovered show no evidence for butchery, nor were any stone tools recovered at the site, although the former may be due to the poor surface preservation of the fossils (text S5). This contrasts with other paleolake deposits in the western Nefud Desert that document intensive and repeated use of lake margin habitats by late Pleistocene hominins (16, 19, 21). From these observations, it appears that the Alathar lake was only briefly visited by humans. It may have served as a stopping point and place to drink and forage during long-distance travel, perhaps initiated by the arrival of dry conditions and dwindling water resources.

Elephant (n = 43) and camel (n = 107) footprints are most abundant and document herds (i.e., adults and juveniles) moving through the landscape (Fig. 5). Animal tracks are overall randomly orientated (Rayleigh’s R = 0.142, P < 0.393), suggesting that movement in the vicinity of the lake was not geographically or geologically constrained, probably reflecting an open landscape (fig. S9). Some tracks clearly document individuals traveling to and from the lakeshore, which was situated southwest of the present-day exposure (text S1). Nevertheless, an overarching north-south trend is perhaps more consistent with movements tied to seasonal shifts in rainfall as opposed to water resource acquisition, as the latter typically involves movements perpendicular to the lakeshore among herbivores (7). This is particularly evident in the elephant tracks that are disproportionately orientated southward (Rayleigh’s R = 0.369, P < 0.001), and similar north-south seasonal movements following lakes have been observed among modern elephant populations in East Africa (22). Elephants, in particular, suggest the regional presence of freshwater sources and substantial plant biomass (23), while the size of the tracks is suggestive of a species larger than any extant taxon (text S5). Elephants are notably absent from the nearby Levant from ~400 ka onward and considering their importance in the diets of Pleistocene hominins (24), their presence in Arabia may have made the region particularly attractive to dispersing H. sapiens. Some of the ungulate prints are consistent in shape and size with a giant buffalo, possibly Syncerus, a taxon previously identified at nearby MIS 5 sites (4, 25). A single small equid track may represent wild ass, which were common in southwest Asia during the late Pleistocene (26), while a pair of ungulate prints are probably those of a medium-sized bovid (text S5).

In addition to footprints, 233 fossils were recovered and included remains of Oryx and elephant (text S5). The discovery of fossils eroding out of the footprint-bearing sediments and similar taxonomic representation across the footprints and fossils imply coeval formation of footprints and deposition of bones. However, direct U-series analysis of several fossil teeth seems to draw an unexpectedly more complex taphonomical history for some specimens of the fossil assemblage (see full discussion in text S4). Carnivores are inferred from tooth-marked bone, and it is likely that they were drawn to Alathar by the dense congregation of herbivores, as is observed in modern-day African savanna ecosystems (27). Stable carbon isotope (δ13C) analysis of fossil tooth enamel indicates substantial portions of C4 grasses in the diets of herbivores, but with fewer lush grasses than in those from the nearby middle Pleistocene site of Ti′s al Ghadah (fig. S11 and text S6) (28). Sequential isotope analysis of elephant tooth enamel (fig. S12 and text S6) indicates a persistent source of water and vegetation that may also be explained by seasonal migrations, and similar results are reported at Ti’s al Ghadah (TAG) (28).


Together, the sedimentary and ichnofossil data are consistent with a well-watered semi-arid grassland setting during a dry period. Lakes and rivers serve as focal points on the landscape for large mammals. During dry seasons, when resources are scarce and herbivores congregate around small watering holes, they are also attractive for hunters and, more often, forages (29). They may also act as effective corridors during seasonal migrations (7, 30), and archaeological data suggest that late Pleistocene Homo in Arabia was highly mobile, penetrating deeper into the Arabian interior than their middle Pleistocene predecessors (19, 21). We demonstrate a direct spatial and temporal association between late Pleistocene humans and medium and large herbivores, indicating that movements and landscape use by humans and mammals in Arabia were inexorably linked. The lack of archaeological evidence suggests that the Alathar lake was only briefly visited by people. These findings indicate that transient lakeshore use by humans during a dry period of the last interglacial was likely primarily tied to the need for potable water.


Sedimentology and diatom analysis

Macroscale sedimentology. The paleolake deposit was initially surveyed to identify the best exposed sedimentary sequence for analysis. Once sections were identified, they were logged and described from the base of the paleolake sediments upward to determine the sedimentary structure (bedding and units) and texture (sorting and grain size). Samples for laboratory analysis were collected by extracting coherent blocks from each section. In the laboratory, these blocks were subsampled, powdered, and analyzed using Bascomb calcimetry (31) and whole-rock x-ray diffraction (XRD). XRD work was carried out in the Department of Earth Sciences, Royal Holloway, on a Philips PW1830/3020 spectrometer with copper Kα x-rays. Mineral peaks were identified manually from the International Centre for Diffraction Data (ICDD) Powder Diffraction File database.

Micromophology. A total of 14 thin section slides across the sedimentary sequences (see fig. S2 for sample locations) were prepared from fresh sediment blocks subsampled from larger block samples collected in the field. Thin section preparation followed standard methods devised in the Centre for Micromorphology at Royal Holloway University of London (32). Thin sections were analyzed using an Olympus BX-50 microscope with magnifications from ×20 to ×200. Photomicrographs were captured with a Pixera Penguin 600es camera. Microfacies were qualitatively analyzed.

Diatom analysis. Three diatom samples were taken at 40 cm (unit 2), 140 cm (unit 4b), and 175 cm (unit 7) above the base level. Samples were prepared as follows. Approximately 0.1 g of dried sediment were placed into a centrifuge tube, immersed in 5 ml of 30% H2O2, and stored at room temperature until all organic material was removed. Samples were rinsed four times with deionized water and subsequently centrifuged for 4 min at 1200 rpm. The diatom suspension was diluted to a suitable concentration and then settled onto coverslips. These were left to air dry in a dust-free environment before mounting with a high-contrast medium (Naphrax). At least 315 diatom valves were counted for each sample (mean, 321 valves) under oil immersion with phase contrast illumination at ×1000. Quantitative reconstructions were carried out using the program C2 (v1.7.7) (33) (available at

Analog matching of fossil samples was performed within the combined African salinity dataset (n = 370) from the European Diatom Database (34), which includes modern samples from northern and eastern Africa (35). The closest analogs in all cases included both eastern and northern African samples (in agreement with the geographic location of the site), and hence, the combined African dataset was used for our diatom-conductivity reconstructions. Inferred conductivity (microsiemens per centimeter) was derived from fossil assemblages (percent data) using a weighted average transfer function with inverse deshrinking. The conductivity model performs well when internally validated by bootstrapping with 999 cycles (r2boot = 0.755, Root Mean Squared Error of Prediction (RMSEP)boot = 0.466 log units) and provides sample-specific errors on diatom-inferred conductivity that are significantly smaller (0.12 to 0.14 log conductivity units) than the overall predicted model errors (fig. S4).

OSL dating

Sample collection, preparation, and analysis. Samples for luminescence dating were collected from the underlying dune sands and directly from the paleolake sediments. This sampling strategy yields direct ages for lake formation and humidity. Luminescence dating was carried out at the Royal Holloway Luminescence Laboratory, Royal Holloway University of London. All luminescence measurements presented here use the quartz OSL signal on 180 to 210-μm quartz grains, and measurements were carried out using a Risø TL/OSL-DA-15 automated dating system (36). Before single-aliquot measurement, purified quartz was placed as a 5-mm circular monolayer on a 9.7-mm-diameter stainless steel discs using Silkospray oil applied via a 5-mm mask. Equivalent doses were determined using the single-aliquot regenerative-dose protocol (37). The burial dose for each sample was calculated using the central age model (38). Environmental dose rates consisted of external beta, gamma, and cosmic ray components (table S3). Beta doses were calculated using a Risø GM-25-5 low-level beta counting system (39) with MgO and Volkagem loess standards (40) standards. Gamma dose rates were measured in the field using an EG&G ORTEC digi Dart-LF gamma spectrometer using the “threshold” method. Cosmic dose was calculated on the basis of location, altitude, and present day burial depth (41). A full description of the luminescence methods is provided in text S2.

Electron spin resonance and U-series dating

Samples and sample preparation. Five fossil teeth were selected for U-series and electron spin resonance (ESR) dating purposes (table S5 and fig. S7). All sampled specimens were recovered from the surface of the paleolake deposit and exhibit weathering and fragmentation characteristic of the assemblage (text S5). None of the teeth could be identified to the species level, but all can be attributed to bovid and at least two to Oryx. The teeth were prepared following the same standard ESR dating procedure based on enamel powder used earlier for the TAG fossil specimens (42): The enamel layer was mechanically separated from the other dental tissues, and both inner and outer surfaces were removed with a dentist drill to eliminate the volume that received an external alpha dose. The dentine attached to the enamel layer was kept aside for subsequent solution bulk U-series analyses. Enamel and dentine were ground and sieved <200 μm.

ESR dose evaluation. Dose evaluation used the multiple aliquot additive dose method. The enamel powder was split into 11 aliquots and irradiated with a Gammacell 1000 Cs-137 gamma source [dose rate, 6.4 grays (Gy)/min] to the following doses: 0.0, 80.1, 150.2, 250.5, 350.7, 500.8, 701.1, 901.6, 1502.6, 3005.2, and 5008.6 Gy. Room-temperature ESR measurements were carried out at the Centro Nacional de Investigación sobre la Evolución Humana (Burgos, Spain) with an EMXmicro 6/1 Bruker ESR spectrometer coupled to a standard rectangular ER 4102ST cavity. The following procedure was used to minimize the analytical uncertainties: (i) All aliquots of a given sample were carefully weighted into their corresponding tubes, and a variation of <1 mg was tolerated between aliquots; (ii) ESR measurements were performed using a Teflon sample tube holder inserted from the bottom of the cavity to ensure that the vertical position of the tubes remains exactly the same for all aliquots. The following acquisition parameters were used: 1 scan, 1-mW microwave power, 1024-point resolution, 15-mT sweep width, 100-kHz modulation frequency, 0.1-mT modulation amplitude, 20-ms conversion time, and 5-ms time constant. All aliquots of a given sample were measured within a short time interval (<1 hour). This procedure was repeated two to three times over successive days without removing the enamel from the ESR tubes between measurements to evaluate measurement and equivalent dose (DE) precisions (table S6). The ESR intensities were extracted from T1-B2 peak-to-peak amplitudes of the ESR signal (43) and then normalized to the corresponding number of scans and aliquot mass. DE values were obtained by fitting a single saturating exponential through the pooled ESR intensities derived from the repeated measurements. Fitting was performed with Microcal OriginPro 9.1 software, which is based on a Levenberg-Marquardt algorithm by chi-square minimization. Data were weighted by the inverse of the squared ESR intensity (1/I2) (44). ESR dose-response curves are shown in fig. S8.

U-series analysis of dental tissues. Powdered enamel, dentine, and cement samples were weighed then spiked using a 229Th-233U tracer before being digested in concentrated HNO3. The solutions were then treated with H2O2 to remove trace organics, with U and Th then separated using conventional column chemistry techniques described in (45). Both U and Th were collected into the same precleaned test tube using 3 ml of 2% HNO3 mixed with a trace amount of hydrofluoric acid (HF). U-Th isotopic ratios were then measured using a Nu Plasma multicollector inductively coupled plasma mass spectrometer (MC-ICP-MS) in the Radiogenic Isotope Facility at The University of Queensland, Brisbane, Australia, following analytical protocols established by (4546). Numerical results are given in table S7.

Dose rate and age calculations. No in situ evaluation of the gamma dose rate associated with the teeth was performed. Beta and gamma dose rates were derived from the laboratory analysis of two bulk sediment samples collected from the uppermost layer of the two sedimentary sections (Fig. 3): Ala-140 (0 to 6 cm below surface of unit 4b) and Ala-SCS (0 to 4 cm below surface of unit 7). MC-ICP-MS analyses were performed by Genalysis Laboratory Services, following a four-acid digest preparation procedure. The following results were obtained: U = 1.806 ± 0.004 ppm (parts per million), Th = 5.283 ± 0.021 ppm, and K = 0.686 ± 0.005% for Ala-140 and U = 0.649 ± 0.004 ppm, Th = 0.681 ± 0.003 ppm, and K = 1.098 ± 0.009% for Ala-SCS. Given the large variability observed between the two samples, initial dose rate calculations were performed using Ala-140 sediment sample, which best related to teeth #2 and #232 (fig. S7).

The following parameters were used for the dose rate calculations: an alpha efficiency of 0.13 ± 0.02 (47), Monte-Carlo beta attenuation factors from (48), dose-rate conversion factors from (49), an estimated water content of 5 ± 3 and 5 ± 2.5 weight % in dentine and sediment, respectively. These values were used to derive the beta and gamma dose rate components. The thickness of the overburden was assumed to be 0.5 ± 0.2 m. Cosmic dose rate was calculated using (50).

Age calculations were performed with USESR, a MATLAB-based program (51) using the Uranium series (US) model defined by (52). The accelerating uptake model (53) that can take into account uranium leaching was also tested. Additional closed system U-series (CSUS)–ESR age calculations were also carried out using DATA program (54). The closed system (CS) model defined by (55) is based on the assumption of a closed system behavior after a rapid uranium uptake event in dental tissues. The CSUS-ESR age is usually considered as providing a maximum age constraint for the fossil. US and CSUS models are typically considered to encompass all possible uptake scenarios.

Fossil and ichnofossil analysis

Fossil analysis. Fossils were systematically collected during pedestrian surveys across the site and their position recorded using a Differential Global Positioning System. Pedestrian surveys were conducted by three to five people walking together in a straight line and separated by not more than 2 m. The entire exposure of the paleolake was examined, and all fossils regardless of size were collected. Each fossil specimen was identified to the lowest taxonomic level possible and facilitated by osteological collections housed at the University of New South Wales, Australia and the Smithsonian National Museum of Natural History, USA. Each specimen was examined by eye and hand lens (10× to 20×), and bone surface modifications were identified and recorded following standard methodologies [e.g., (5657)].

Footprint analysis. Footprints were systematically documented during pedestrian surveys across the site, and their position was recorded using a Leica Total Station. Each print was photographed with a scale, compass, and arrow denoting orientation of movement when discernible (e.g., Fig. 3I), and track length and width were recorded. Directionality was assessed using Rayleigh’s Z test for uniformity for circular data and a Watsons U2 test for von Mises normal distribution. Three hominin tracks (HPR001, HPR002, and HPR003) were extensively photographed to generate high-resolution, scaled three-dimensional models (AgiSoft PhotoScan Professional, Agisoft LLC, St. Petersburg, Russia). Additional total station points (~15 points for each) were taken to map the print outline, as well as high and low points in and around the prints, and these acted as ground control points during model generation. Print morphometric data were collected in the field following (fig. S9A) (58). Footprint index was calculated by dividing footprint length (FPL) (Heel-Hallux) by ball breadth (B1 and B2).

Hominin stature, body mass, and speed estimates were calculated following (59) and based on regression equations derived from footprint morphometric data of modern, habitually unshod Daasanach people (Lake Turkana region, Kenya), a group that inhabits a semi-arid grassland environment likely similar to the western Nefud Desert during the Pleistocene humid phases (4). Stature was determined from FPL, body mass from footprint area [FP area = FPL × ball width (BW)], and speed from stride length (SL; distance from the heel of one step to the heel of the next step made by the same foot) divided by average FPL (SL/avgFPL). Comparative stature and mass data for late Pleistocene H. sapiens (n = 89) and H. neanderthalensis (n = 51) were sourced from the literature and based on allometric relationships between limb bone size and body height (fig. S10). Values were averaged in instances where multiple estimates existed for a single individual (e.g., the Skhul IV hominin).

Stable isotope analysis

Eight samples, which comprise seven Bovidae teeth and one elephant tooth, were selected for stable carbon and oxygen isotope analysis of tooth enamel from the available fossil material recovered from atop modern-day paleolake deposit (table S14 and fig. S11). The elephant molar was used for additional, sequential stable carbon and oxygen analysis based on its completeness and robustness to endure additional sampling (table S15 and fig. S12). Fourier transform infrared spectroscopy was used to assess enamel preservation as per (28).

All teeth or teeth fragments were cleaned using air abrasion to remove any adhering external material. Enamel powder for bulk analysis was obtained using gentle abrasion with a diamond-tipped drill along the full length of the buccal surface to ensure a representative measurement for the entire period of enamel formation. For sequential samples, each sample was a 1- to 2-mm-wide groove perpendicular to the tooth growth axis, through the thickness of the enamel layer. The distance of the base of each sample groove from the enamel/root junction from the furthest sample margin was recorded. All enamel powder was pretreated to remove organic or secondary carbonate contaminates. This consisted of a series of washes in 1.5% sodium hypochlorite for 60 min, followed by three rinses in purified H2O and centrifuging, before 0.1 M acetic acid was added for 10 min, followed by another three rinses in purified H2O [as per (6061)].

Following reaction with 100% phosphoric acid, gases evolved from the samples were analyzed to stable carbon and oxygen isotopic composition using a Thermo Gas Bench 2 connected to a Thermo Delta V Advantage mass spectrometer at the Department of Archaeology, Max Planck Institute for the Science of Human History. Carbon and oxygen isotope values were compared against international standards (NBS 19, Merck) registered by the International Atomic Energy Agency. Replicate analysis of ostrich egg shell (OES) standards suggests that machine measurement error is c. ±0.1‰ for δ13C and ±0.2‰ for δ18O. Overall measurement precision was studied through the measurement of repeat extracts from a bovid tooth enamel standard (n = 40, ±0.2‰ for δ13C and ±0.3‰ for δ18O).


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We would like to thank the Saudi Commision for Tourism and National Heritage (SCTH) for continued support of the Palaeodeserts Project. Funding: The Palaeodeserts Project is supported by the Max Planck Society, European Research Council (no. 295719 to M.D.P.), Bristish Academy, and Leverhulme trust (PG-2017-087). M.S. would like to thank the Leakey Foundation for generous funding. R.C.-W.’s contribution to this work was supported by a Natural Environmental Research Council DRP studentship, and data relating to this contribution can be found in the texts S1 and S2. P.S.B.’s research is supported by the Leverhulme Trust (ECF-2019-538). M.D.’s research is supported by the Australian Research Council (ARC) Future Fellowship grant (FT150100215). S.J.A.’s contribution to this work was partly supported by the Research Council of Norway, through its Centres of Excellence funding scheme, SFF Centre for Early Sapiens Behaviour (SapienCE) (no. 262618). J.L.’s research is supported by the Australia Research Council (ARC) Future Fellowship Grant (FT160100450). A.M.A. acknowledges the financial support of the Researchers Supporting Project Number (RSP-2019/126), King Saud University, Riyadh, Saudi Arabia. Author contributions: M.S., R.C.-W., P.S.B., K.J., I.C., S.J.A., J.L., G.J.P., P.C., M.A.B., N.A.D., H.S.G., A.M.A., B.Z., A.A.-O., and M.D.P. conceived the project. M.S. performed the analysis of fossil material. R.C.-W., S.J.A., and I.C. conducted the sediment and OSL dating analyses. P.S.B. coordinated on-site surveys and produced site maps. G.J.P. and M.D. performed the uranium-series and ESR dating study. D.B.R. performed the diatom analysis. P.R. performed the stable isotope analysis. M.S. and R.C.-W wrote the manuscript with support from J.L., M.D., G.J.P., D.B.R., I.C., S.J.A., P.R., and H.S.G. All authors contributed to the manuscript and approved the final version, and all authors qualifying for authorship are listed. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or Supplementary Materials. Additional data may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article