Timing and causes of North African wet phases during the last glacial period and implications for modern human migration

We present the first speleothem-derived central North Africa rainfall record for the last glacial period. The record reveals three main wet periods at 65-61 ka, 52.5-50.5 ka and 37.5-33 ka that lead obliquity maxima and precession minima. We find additional minor wet episodes that are synchronous with Greenland interstadials. Our results demonstrate that sub-tropical hydrology is forced by both orbital cyclicity and North Atlantic moisture sources. The record shows that after the end of a Saharan wet phase around 70 ka ago, North Africa continued to intermittently receive substantially more rainfall than today, resulting in favourable environmental conditions for modern human expansion. The encounter and subsequent mixture of Neanderthals and modern humans – which, on genetic evidence, is considered to have occurred between 60 and 50 ka – occurred synchronously with the wet phase between 52.5 and 50.5 ka. Based on genetic evidence the dispersal of modern humans into Eurasia started less than 55 ka ago. This may have been initiated by dry conditions that prevailed in North Africa after 50.5 ka. The timing of a migration reversal of modern humans from Eurasia into North Africa is suggested to be coincident with the wet period between 37.5 and 33 ka.


Stalagmite SC-06-01
The sample used for this study is a 93 cm-long stalagmite (SC-06-10), which was found broken on site. The three broken pieces fit together and the bottom also fits onto a stump found in the cave. Thus it represents the complete specimen with no missing parts. The diameter varies along the growth axis between 11 and 16 cm. The stalagmite consists of sections of dense clear calcite and milky white calcite. There are several thin red layers similar in colour to the surface dust layer, indicating dust deposition during frequent drying of the stalagmite surface. The main growth axis also shows some slight changes in growth direction. Over the top 30 cm a twin stalagmite formed. We defined three main growth sections and twelve minor growth sections based on the crystal structure and visual layering, positions of red layers in the stalagmite and chronological results (see below). Half of the red layers are associated with a hiatus identified by U-series chronology data and mark the start / end of a growth phase. The other red layers are probably also associated with (short) hiatuses, which cannot be constrained within uncertainty of Useries results. The main sections are all longer than 16 cm, and cover 75 % of the total length of growth without detectable hiatuses. Section I is between 5.6 and 23.7 cm from the bottom, section II from 34.2 to 50.2 cm and section III from 55. 4

U-series dating and age model
Stalagmite SC-06-01 consists of three broken pieces, which were cut along the growth axis and polished. Some sections were further cut into slabs along the growth axis for subsampling by micromill or with a wire saw. Sub samples for U-Th dating were either drilled using a hand held microdrill (powders) or small pieces were cut from slabs using a diamond wire saw (pieces). Sample masses for U-Th analyses varied between 10 and 180 mg.
The concentration of 238 U varies between 134 and 756 ng/g, with an average of 388 ng/g. U concentrations are found to be slightly higher in the bottom 300 mm and lower in the top part of the stalagmite. Initial 234 U/ 238 U activity ratios are quite consistent around 1.28. The calcite is clean and almost free of detrital components. All 230 Th/ 232 Th activity ratios are greater than 240 and the mean ratio is 6200, thus, a correction for detrital Th has no significant effect on the measured activity ratios and calculated ages (Table S1). However, a correction is usually applied where 232 Th is measurable, and a bulk earth value of the upper crust ( 238 U/ 232 Th act = 0.8 ± 0.4) 8 is used for detrital correction.
The stalagmite started forming around 66.5 ka and grew episodically until about 30.8 ka. The detailed chronology for SC-06-01 is based on 116 dating results along the growth axis (104 on the main axis and 12 on the twin axis at the top) and typical age uncertainties are in the range of 0.7 %. All uncertainties are at the 95 % confidence interval. Table S1 shows all analytical results and calculated ages.     Dating results of the parallel growth section at top of stalagmite are presented in Fig. S6. Here distances are shown from the top of the section. One dating result obtained on a sample taken from a boundary area between two sections including a red layer (between III p and 11 p ) is not in stratigraphic order.

Distance -age model
A distance-age model was generated for SC-06-01 using StalAge 9 . Distance age modeling was done sectionwise. The start and end of the sections / growth episodes are defined by the red layers associated with a detectable hiatus in the dating results. The distance-age model is shown in Fig. S7 a (main growth axis) and Fig. S7b (parallel growth axis at the top). The red lines indicate upper and lower 95% confidence levels of the age model.

Comparison of timing of wet phases to Northern Hemisphere insolation changes
Growth phases of SC-06-01 are compared to insolation at 30°N in June and December as well as to the insolation gradient between 60°N and 30°N (the difference is calculated as insolation at 60°N minus insolation at 30°N). Insolation values are taken from Berger and Loutre (1991) 12 . Fig. S11    The degree of detrital 230 Th contamination is indicated by the measured 230 Th/ 232 Th activity ratio and corrections were calculated using a 238 U/ 232 Th activity ratio of 0.8 ± 0.4. 234 U/ 238 U ini is the initial 234 U/ 238 U activity ratio.
Ages are calculated based on corrected activity ratios. For all samples, correction is less than 0.1 ka and corrected and uncorrected results are identical within uncertainties.