Mapping Middle Stone Age human mobility in the Makgadikgadi Pans (Botswana) through multi-site geochemical provenancing of silcrete artefacts

Analyses of the distance over which lithic raw materials were transported for use in stone tool production provide important insights into early human mobility through prehistoric landscapes. This study combines the use of geochemical provenancing, chaîne opératoire analysis and geochronology to examine patterns of lithic raw material procurement at five single-use open-air Middle Stone Age (MSA) archaeological sites in Ntwetwe Pan, part of the Makgadikgadi Pans complex in north-central Botswana. Maximum ages of the five sites ranged from 106 ± 3 to 69 ± 7 ka, with site formation thought to have occurred before and after a lake high stand dated to c. 72-57 ka. Tool manufacture at all sites was largely confined to the production of MSA points, with silcrete used exclusively as the raw material. Geochemical provenancing investigations aimed to identify the specific silcrete outcrops used as lithic raw material sources. Analysis of contemporary pan floor deposits show that sediment (and hence silcrete) geochemistry in Ntwetwe is determined by the proportional sediment input into the pan from fluvial systems with different catchment geologies. Immobile trace element signatures for 46 waste manufacturing flakes were compared against equivalent data for 321 silcrete samples collected at outcrops within and beyond Ntwetwe Pan. Waste flakes were chosen to be representative of the main silcrete raw material types present within the artefact assemblage at each site. Fifteen waste flakes were shown to match specific outcrops on the basis of their geochemical signatures and petrographic properties; all matching outcrops were within Ntwetwe Pan, at distances ranging from 7 to 55 km from individual sites. Multi-site analysis of procurement patterns identifies common silcrete source areas, revealing a preference for silcrete from particular locations within Ntwetwe Pan. Given that the five archaeological sites were likely occupied at different times, this resource preference may have been a longer-term behavioural feature of MSA populations in the Ntwetwe region. The distances over which silcrete was transported in Ntwetwe Pan are smaller than identified in investigations at similar-aged MSA sites in northwest Botswana. The reasons for the different silcrete procurement ranges in the two regions are likely related to silcrete availability and/or raw material preference, but this requires further investigation.