Stone and water governed early human settlement of the desert Rudolf W. Dellmour, Emily Coco, Osama Samawi, Mohammed Al Kindi, Ali Al Mahrooqi, Michelle R. Bebber, Metin I. Eren, Mathias Harzhauser, Roman Garba & Jeffrey I. Rose

https://www.nature.com/articles/s41586-026-11180-5 Models of early human dispersal rely on climate to predict where populations could settle, with freshwater assumed to govern the timing and geography1,2, particularly across the Saharo-Arabian arid belt, where humid windows episodically opened corridors into the desert interior3. Yet Pleistocene dispersals into ‘Green Arabia’ were brief and geographically constrained4,5, a pattern not explained by water availability alone. Here we show, from an approximately 1,500-km geoarchaeological survey through the Rub’ al Khali of Arabia, the largest sand sea in the world, that the availability of stone for tool manufacture operated as an independent factor on early human settlement. Palaeolake deposits were recorded at 55 of 125 localities investigated, yet no lithic artefacts and no knappable raw materials were found along the transect, with ground-surface visibility exceeding 95%, despite abundant freshwater and contemporary occupation at the desert margins. Spatial modelling of the Peninsula supports the inference that settlement tracked the co-occurrence of water and lithic raw materials rather than water alone. Across the Peninsula, 57% of water-accessible terrain lies on formations that lack knappable stone and adding raw-material availability raises discrimination of known site locations from an area under the curve of 0.695 to 0.768. Convergent patterns from Africa, North America and Asia indicate that the constraint is not peculiar to Arabia. Dispersal models that omit raw-material geology may systematically overestimate the habitable range of stone-tool-dependent populations.