Bab-al-Mandab Strait reconstruction
The southern dispersal route hypothesis suggests that H. sapienscrossed into Arabia via the Bab al Mandab Strait (Stringer, 2000; Mellars, 2006; Armitage et al., 2011; Rohling et al., 2013). With a current width of ~26 km, the Bab-al-Mandab Strait likely represented a significant challenge to Homo sapiens dispersals along the southern route. Thus, hypotheses favouring this route propose that maritime dispersal occurred during punctuated sea-level low-stands between 150,00 to 50 ka BP (Armitage et al., 2011; Rohling et al., 2013), when the width of the (BaMwidth) Strait was substantially reduced. Previous analysis of composite Globigerinoides ruber δ18O values from Red Sea marine cores KL-09, KL-11 and MD92-1017 have been used to reconstruct relative sea-levels over the past 500 kyrs (Rohling et al., 2009; Grant et al., 2012, 2014). The narrow Strait limits water to flow into the Red Sea creating a basin isolation effect, meaning that – unlike ocean core sea-level reconstructions which are based on deep-sea δ18Obenthic values – the Red Sea Relative Sea Level curve (RSL) is independent from temperature forcing and instead records residence times of water in the basin, which are based on evaporation rates as determined by sea-level. Additionally, isostatic effects were accounted for in the creation of the RSL curve (Rohling et al., 2009), allowing us to assess sea-levels which control the sill depth at the Bab-al-Mandab strait. The Red Sea Relative Sea-Level (RSL) curve has been tuned to the δ18OG. ruber curve of LC21 (Mediterranean), another record to which the basin isolation concept has been applied (Grant et al., 2012). The chronology of LC21 has been derived from correlations with the revised chronology of the Soreq Cave δ18O stalagmite record for the last 150 kyrs, and includes the chronology of the Sanbao Cave δ18O stalagmite records and KL09 Ca/Ti up to 500 ka BP (Grant et al., 2012, 2014). Thus, the RSL is an excellent record for studying sea-level variation in the Red Sea. We therefore use the RSL curve and freely available elevation and bathymetry data at 15 arc-second interval resolution (~450 metres at 14oN) (GEBCO Compilation Group, 2020) to reconstruct BaMwidthover the past 150 kyrs.
We measured the minimum distance between contour lines at 10 m intervals between 20 to -120 m amsl on both sides of the strait using a variable-point method. Unlike a straight-point method, in which distances between equal contour lines are measured across a straight point A to point B transect, a variable-point method takes into consideration that the minimum width of the Strait shifted in space (Fig. S3). The variable model therefore provides an approximation of minimum possible crossing distances for H. sapiens in the Bab-al-Mandab area. We do not include islands in our analysis, until they are connected to the greater African and Arabian landmasses. While islands may have provided benefits to maritime crossings (i.e., breaking the crossing into smaller journeys), our method provides a record of the minimum distance of a crossing achieved in one journey.