State of knowledge and initial science questions during the second proposal
The Sahara was the inevitable passageway for any migration from tropical Africa to Europe during the Late Quaternary. By trying to extend the data set that now exists for the entire Holocene from Lake Yoa (Chad) to the preceding late Pleistocene climatic cycle, project A2 has continued to compile terrestrial data in the central Sahara to provide a palaeoclimatic standard for the CRC’s archaeological studies in northern Africa. This objective was approached by a regional extension of fieldwork to locations featuring Holocene to pre-Holocene deposits in Northern Chad and their multiproxy analysis.
Methods and Results
Geology and Archaeology
Two research expeditions in the Tibesti Mountains in February/March 2015 and 2016 opened a new chapter in the reconstruction of the last climatic cycles in the central Sahara and their palaeolacustrine environments. The Tibesti, the Sahara’s largest and highest mountains, may be considered the lighthouse and water tower for migration from sub-Saharan Africa to the Mediterranean coast and beyond from the Palaeolithic to the present. Lacustrine deposits in its varied craters are expected to extend the Holocene data set to the preceding late Pleistocene humid period.
In winter 2015, complete lacustrine sections were sampled for the very first time at the flanks of the 900 m deep crater of the Trou au Natron at the foot of Pic Toussidé, and in the 800 m deep crater of Era Kohor, the major sub-caldera of Emi Koussi, the Sahara’s highest peak (3,445 m a.s.l.). The sampled diatomites are located 360 and 125 m above the present-day bottom of the calderas. Studies in the 1960s suggested that lake levels in the Trou au Natron were 300-500 m high at 12,400-14,970 uncal. yrs BP. However, such lake depths and resulting water volumes, are hardly conceivable in view of the limited intake area and precipitation-evaporation ratios, which would have required local rainfall by far surpassing estimates for latitudes 19-21° N from the Holocene record of Lake Yoa that is situated from 220 km to 460 km southeast and 1,550 m to 2,450 m lower.
Interpretation of the sedimentological and geochemical analyses and first biomarker studies (diatoms, pollen, plant remains) of the section W99 in the Trou au Natron show a typical lake succession with an initial phase of varying water levels at ~8,700 cal BP. During this phase the sediments are dominated by fine sand (quartz/SiO2) and low CaCO3 contents accompanied by different types of efflorescence (Thenardite; Na2SO4). The upwards following carbonates (calcite, aragonite, dolomite, magnesite) indicate increasingly stable freshwater conditions. At the top, calcitic sediments with slightly increasing Sr/Ca and Mg/Ca values accompanied by aragonite formation suggest intensifying salinisation of the lake water towards the uppermost layer which is tentatively dated to ~4,400 cal BP. The final lake stage is not recognizable because of truncation by aeolian erosion. Our new charcoal AMS dates disprove the previous bulk carbonate 14C ages and hence any terminal Pleistocene humid conditions in the Tibesti 4,000-7,000 years earlier than in the surrounding Saharan regions.
In October 2015, the Chadian counterpart entirely covered another field trip to the Ennedi plateau and provided airlifts from N’Djaména to Fada, the region’s main town, and back, and also a helicopter to reach yet unexplored localities. Inspections and sampling sites included the hidden Guelta Maya that was sampled for isotopic studies and Lake Ajous, a yet unstudied saltwater lake in the Ennedi’s northwestern foreland. New archaeological evidence comprised MSA lithic artefacts and Neolithic pottery; presumably Iron Age to recent iron smelting sites; and hitherto unreported rock art friezes that provided clues to the interpretation of enigmatic motifs known from other Saharan regions. Geological and geomorphological studies were conducted at geological type localities with basement-sandstone boundaries and outstanding landmarks such as the canyons of Bechikelé, Archei and Nou Chero, or the planet’s second highest natural rock arch, Aloba. Various laboratory analyses and radiometric age determinations of the geological, archaeological, biological and botanical material are underway.
The second field campaign in the Tibesti in February/March 2016 included volcanologist Clive Oppenheimer (U Cambridge, UK) and documentary filmmaker Srdan Keca (U Stanford, Palo Alto). Fieldwork focused on Pic Toussidé, the Tibesti’s second highest (3,315 m a.s.l.) and most recent volcano, and the very first scientific exploration of the Tieroko volcanic complex, the Tibesti’s central massif. Sampling in the Trou au Natron focussed on another complete section of Holocene diatomites, this time at the bottom of the caldera, for comparison with the diatomite sequence sampled 300 m above at the slope in 2015. Yet undated basal layers of an terrace outcrop in Enneri Tabi were sampled for thermoluminescence dating within project F2. This section has served for the construction of the still occasionally quoted “Climate curve for the central Sahara” (Jäkel 1979). Presumably late Miocene to late Holocene volcanic rocks were systematically sampled all along the expedition track and are being analysed and dated at Cambridge University, UK (C. Oppenheimer).
Erosional remains of carbonate tufa occurring high at the flanks of the main caldera, even topographically above the most elevated diatomites, and also on top of the isolated secondary volcanic cones on the crater floor, are puzzling. Because of their facies, hardness, topographical location and stratigraphical position that shows a distinct disconformity at the boundary with the occasionally overlying early Holocene diatomites, they were tentatively put into the Eemian (MIS 5) at ~ 130,000 years ago. This presumably penultimate humid period is crucial for the “Out of Africa 2” climatic window, a focal issue of CRC 806. Petrographic and geochemical examinations were performed by Dr. Pierre Deschamps at the CEREGE (Aix en Provence) to assess possible diagenetic effects, and volcanic or later lacustrine contamination of the initial Thorium content. A series of 238U/232Th age determinations are being presently run at his specialised geochronological laboratory. Parallel thermoluminescence dating is currently being undertaken at the University of Cologne’s TL laboratory (H. Brückner, project F2).
The ongoing laboratory and biomarker studies, and 14C, OSL, U/Th and K/Ar age determinations are expected to raise the knowledge on the late Quaternary climate and volcanic evolution of the Tibesti to a new level and to elucidate its role in the spread of modern humans along this Saharan key region. Considering the difficult security situation in the Saharan countries, the acquired samples provide a presently unrivalled material basis for palaeoclimate studies of the eastern central Sahara.
First stratigraphic and chronological results already reveal the supraregional validity of the Lake Yoa record and promise to provide pristine environmental and climate clues as to the penultimate “Green Sahara” period. Talks and posters that were presented at AGU 2013 and 2015, EGU 2016, AGU 2016 and elsewhere have raised much interest in different scientific communities and science media such as Nature or New Scientist. This interest has resulted in various requests for analytical cooperation (U Harvard, U Yale, U Cambridge, CEREGE Aix en Provence, UNITAR Genève, MPI for Meteorology Hamburg, TU and FU Berlin, GFZ Potsdam and others). They build the groundwork of forthcoming manuscripts on specific issues. A geoarchaeological paper in print in Antiquity (Riemer, Kröpelin, Zboray 2016) amalgamates first-hand information in the fields of climate and environmental change, rock art, archaeology and settlement history to establish a novel absolute chronology for the Holocene rock art in the Eastern Sahara with a good level of confidence. Based on long-term investigations, it combines primary environmental and climatic evidence, 14C dates, geological and archaeological stratigraphies and other chronologically relevant archaeological indicators with a systematic analysis of the relative sequence of rock art styles derived from superimpositions and weathering. The results are thought to be welcomed in the much discussed but little substantiated field of dating of Saharan rock art.
The unparalleled varved archive of Lake Yoa with its +20,000 single laminae that was cored during A2’s first project phase is finally close to a long-awaited publication as a solicited Nature paper following the announcement in a Nature feature (Schiermeier 2012). Detailed microscopic investigation has revealed the sedimentary processes responsible for the formation of the fine laminations, identified the season during which they were formed, and their annually rhythmic nature (Francus et al. 2013). High-resolution XRF core scanning allowed distinction of each individual lamination over the entire record, opening new perspectives for the study of finely laminated sediment sequences. The completion of the manuscript on the long core has been delayed due to limited staffing (one PhD position) and extremely time-consuming multiproxy analyses such as varve counts or the enormous effort for diatom and pollen determinations (1-2 samples per day) accomplished by the most experienced international experts. The resulting diatom and pollen diagrams will close the ~ 5000-year gap between the onset of the formation of lake deposits at Lake Yoa and the already published phytoplanktonic and palynological conclusions for the mid-Holocene to the present (Kröpelin et al. 2008, Lézine et al. 2011).
The Yoa data set represents the Sahara’s (and Africa’s) most complete and detailed Holocene climate sequence. Among other highly controversial issues such as the question of an abrupt or gradual onset of the “Green Sahara” or the magnitude of the 8.2 event in a hypercontinental position, it will help to validate and improve numeric climate models. A recently published hydrogeological map of the Ounianga lakes (UNITAR 2015) greatly draw on results produced within the A2 project. Specialised papers on the biomarkers are in advanced preparation, e.g. on the diatom results including a transfer function by F. Sylvestre et al. or the pollen analyses by M. Dinies et al. A comprehensive multiauthored thematic volume will be published as an invited book/eBook in the Springer series.
A proposal submitted to the DFG in September 2016 is aiming at a chronologically more elaborate high-resolution reference pollen diagram of Lake Yoa for the early and middle Holocene time slices that will further contribute to the verification of unsolved research questions and hypotheses. The pollen and diatomite sequences from the Trou au Natron and Era Kohor craters in the Tibesti will be used to check conformities and differences with the Lake Yoa data set.
To supplement the geological fieldwork, archaeological investigations were undertaken with a focus on the Middle Stone Age (MSA). However, given the extremely remote location of the Tibesti Mountains and the general poor state of research in that region, archaeological features and finds of all epochs, from Palaeolithic to recent times, were sampled and recorded.
The journeys to and from the remote geological sampling locations provided ample opportunities to conduct archaeological surveys and even a few excavations. They covered a major geographical area and include various topographical zones from the foothills and terraces along major wadis to the high plateaus (Tarsos) and the peaks of the two highest volcanoes of the Tibesti, Emi Koussi and Pic Toussidé. The surveys were performed by car or on foot. The archaeological sites were recorded by standardised survey forms and digital photography. Site coordinates and altitude were taken by hand-held GPS receivers. At some sites, lithic artefacts and pottery sherds were collected, and on two occasions, test trenches were excavated.
As part of the survey, 352 points of archaeological evidence were recorded. They comprise stray finds, open air, rock shelter and rock art sites as well as evident features like graves, huts and other stone structures. Diagnostic finds were preliminarily attributed to the MSA, the mid- and late Holocene period and (sub-) recent times. Without excavations, the evident features were hard to date. However, an impressive, approx. 50 m long, pear-shaped grave mound, discovered on Emi Koussi, corresponds in size and shape to a grave excavated in southern Libya, which has been radiocarbon dated to 4,585 ± 280 cal BP (Gabriel 1999). In general, archaeological findings and finds were recorded in all topographical zones, although the majority was found below 2000 m a.s.l. Artefacts unambiguously attributable to the MSA, like Levallois cores and flakes, were discovered only in elevations up to 1135 m a.s.l. While the archaeological survey conducted was fragmentary of necessity, a continuous stretch of approx. 50 km in length was surveyed on foot during the ascent and descent of Emi Koussi. Altogether 74 sites, from stray finds to camp and burial sites, were recorded in the course of this survey. It is evident that the occurrence of archaeological remains decreases with increasing elevation.
Another archaeological study was conducted at a site near the Tibesti’s central settlement Bardai because of its outstanding significance for the early Neolithic in the Sahara. Gabrong 66/20 is a rock shelter containing semi-lacustrine sediments and rich archaeological finds originally excavated in 1966 by B. Gabriel, a member of the Free University of Berlin’s former research base at Bardai (Gabriel 1978). His geoarchaeological work resulted in the discovery of early Holocene pottery, which was conventionally radiocarbon-dated to 6999 ± 169 cal BCE (Hv 2748) on charcoal, putting this site among the earliest ceramic-bearing sites in the Sahara. As the in situ position of the potsherds has been questioned, a stratigraphical reinvestigation of the section was carried out that confirmed the original interpretation. AMS dating of charred Tamarix remains from the pottery-bearing layer provided an age of 7966 ± 141 BCE (Beta-441717). This date outmatches the previous date by almost 1000 years and turns Gabrong 66/20 into one of the most ancient pottery-bearing sites in the Sahara, along with the sites in the Acacus Mountains (Libya) and Nabta Playa (Egypt). Further work will check the implications for the early Holocene prehistoric reoccupation of the Sahara.
The ongoing PhD research of Jan Kuper seeks to correlate the palaeoclimate results obtained in project A2 with the archaeological record of Northeast Africa. His thesis primarily focusses on the Epipalaeolithic reoccupation of the Eastern Sahara, which is a convenient case study to examine the migration process of colonising hunter-gatherers at relatively high temporal resolution. Besides generic questions of origin and time, the PhD project aims at investigating how and finally why people spread to new and unfamiliar tracts of land. Tackling these questions will not only illuminate the prehistory of the Eastern Sahara, but will also help to better understand general mechanisms of migration, in particular the colonisation of uninhabited landscapes.
Initial results of lithic analyses of assemblages from Egypt, Sudan and Libya as well as a review of the relevant literature indicate that (1) the early Holocene reoccupation of the Eastern Sahara proceeded as a slow, more than a millennium long process, which is in line with the environmental reconstructions compiled in project A2. (2) This process had of a step-like character, as in an early phase sites are more ephemeral and occupation was only seasonal and restricted to certain landscape units. During a second phase, sites are larger, regional variation is higher and areas of the Eastern Sahara were presumably occupied for more than one season. (3) Finally, the correlation of the Epipalaeolithic sites with the topography of this area reveals that the early Holocene migration did not occur as a spatial spread, instead Epipalaeolithic colonisers followed natural corridors like palaeodrainage systems or escarpments. These results may eventually help to model previous expansions into the Sahara, like that of anatomically modern humans during the MSA.
In July 2016, 18 years after an initiative taken by S. Kröpelin in 1999, UNESCO’s World Heritage Committee has unanimously inscribed the Ennedi Massif in north-eastern Chad as one of the rare mixed natural and cultural sites. This achievement would not have been possible without the geological, archaeological, botanical, and ethnographical research completed within CRCs 389 (ACACIA) and 806 since 2003. Together with Chad’s first World Heritage site in 2012, the Lakes of Ounianga, it is another example of sustainable German-African scientific cooperation with a host country.
Geophysics
According to the funding proposal of project in the 2nd Phase, the aim of the geophysical investigation was to apply methods of multi-dimensional transient electromagnetics (TEM) and direct current resistivity (DC) on calcareous palaeolakes and dry salty lakes. Through these investigations, sediment thicknesses and lithographic layers were investigated with respect to the determination of subsequent trenching locations and the discovery of future borehole locations. Subsequently, the recorded data was subject to multi-dimensional and joint interpretation accompanied by 2D/3D modelling of synthetic data.
Finally, the recorded data of both field campaigns were interpreted by means of 1D, joint and constrained inversion techniques. Multi-dimensional modelling of synthetic data came along to analyse the degree of reliance of the achieved results.
Located on the border between Ethiopia and Kenya and as part of the East African Rift System (EARS), Lake Chew Bahir (Amharic for “Salty Lake”) is an approximately 30 x 70 km² dry saline mudflat that episodically fills with water during the rainy season, 500 m asl. Former geophysical investigations, e.g., airborne gravity and reflexion seismics, indicate sedimentary deposits of several km depths. Therefore, the basin potentially provides sedimentary archives that extend far beyond the Quaternary.
We recorded two-dimensional TEM data along seven profiles, totalling 65 soundings. All profiles were oriented either east-west or north-south. To facilitate a best possible comparison of our results with the coring analyses of the drillings of F. Schäbitz’s group, four of these profiles were positioned directly on top or in the vicinity of the two deep drilling locations.
The transmitter consisted of a 100 x 100 m² single square loop utilising a Zonge NT-20 transmitting device. Two Zonge TEM/3 coils served as receivers for the vertical and for one horizontal component of the time derivative of the secondary magnetic field. For every station, the horizontal component was aligned in east-west direction, perpendicular to the nearby western mountain ridges. This alignment was chosen to study possible 2D effects of either the mountains and their underlying crystalline bedrock or alluvial fans intruding from the basins western boundaries. The results of measuring the vertical component give insights in the distribution of the electrical resistivity with depth.
The source area of the Bisare River is located within the Hobitcha Caldera approximately 20 km east of Wolaita Sodo, southern Ethiopia. Due to its position as a basin infill, a continuous sedimentation rate is expected. First sedimentological investigations refer to varying environmental conditions. Tephra layers may give the potentiality for the correlation with data from other palaeoclimatic archives.
In February 2015, seven parallel TEM profiles were measured that covered an area of approximately 300 x 500 m², with 35 TEM stations. For each sounding, we again recorded the vertical and one horizontal component. This time, we used a 50 x 50 m² single square loop as a transmitter. For the receivers, we utilised a 20 x 20 m² single square loop for the vertical component and a Zonge TEM/3 coil for the horizontal component. The direction of the horizontal component again was east-west aligned and therefore quasi perpendicular to the surrounding Caldera. Additionally, two DC soundings were taken to increase the quality and reliability of our final results.
The double crater system of the Dendi Lakes (3,250 m a.s.l.) is located approximately 80 km west of Addis Ababa. It is situated inside an 8 km wide volcano caldera. In the framework of the CRC 806, the group of Prof. F. Schäbitz drilled two sediment cores in the eastern lake at about 50 m water depth in early 2012, revealing Holocene deposits.
We measured one TEM profile containing 23 TEM stations and 2 DC soundings. The length of the profile was 1,150 m and it was aligned south-west to northeast, starting near the eastern lake and reaching as close as possible to the beginning slopes of the caldera. Again, we used a 50 x 50 m² single square loop as a transmitter. For the receivers, we utilised a 20 x 20 m² single square loop for the vertical component and a Zonge TEM/3 coil for the horizontal component. The horizontal component was measured along the profile direction, as much as possible perpendicular to the eastern lakeshore and the surrounding caldera.
The focus for the field surveys in Ethiopia was on two-dimensional time-domain electromagnetics (TEM). In total, we measured approximately 125 TEM stations at three locations. For each station, we recorded the vertical and one horizontal component of the derivative of the secondary magnetic field. To enhance the quality of the results and to compare the results of both methods, we additionally measured four Direct Current (DC) soundings. In terms of the depth of investigations, TEM can deliver information up to several hundreds of metres, depending of the device configurations and the resistivity of the subsurface. In Ethiopia, we reached depths of 100 m to 200 m. The depth of investigation of the DC measurements was several tens of metres.
In a first step, all recorded data (TEM and DC) were solely inverted by means of 1D, utilising different 1D inversion techniques such as Occam and Marquardt. Secondly, the data of the stations on which we deployed TEM and DC simultaneously were inverted jointly by means of 1D. This means that the information of both datasets was used to find a subsurface model that describes both datasets to the best possible degree. Both methods, TEM and DC, complement one another in terms of different depths of investigation or sensitivity, respectively. An additional advantage of inverting these data sets jointly is that TEM is more sensitive to conductive structures while DC is more sensitive to resistive structures. All these aforementioned 1D inversions were conducted using the software EMUPLUS, a multi-method 1D inversion package developed at the Institute of Geophysics and Meteorology, University of Cologne, Germany.
Secondly, the recorded TEM data sets were inverted using the software AarhusInv from the Hydrogeophysics Group of the University of Aarhus, Denmark. Here, all data and models are inverted as one system, producing layered solutions with laterally smooth transitions. The models are regularised through lateral constraints that tie interface depths or thicknesses and resistivities of adjacent layers together (Auken and Christiansen 2004). When the data is recorded along a single profile, the method is referred to as Laterally Constrained Inversion (LCI). When deploying a 2D surface grid of measurements, the method is called Spatially Constrained Inversion (SCI; Viezzoli et al. 2008). According to the Hydrogeophysics Group of Aarhus, these inversion schemes are well suited for data taken in sedimentary environments.
In order to interpret the measured horizontal TEM component, modellings was conducted with a 3D time domain forward code (SLDMem3t) using the Finite Differences approach and utilising Krylov subspaces and the Lanczos spectral decomposition method (Druskin and Knizhnerman 1994).
Our geophysical measurements in the East-African Rift Valley aimed at the investigation of sediment thicknesses and stratigraphic layering in three sedimentary deposits located in Ethiopia. The physical parameter of interest was the electrical resistivity of the subsurface. The applied methods of 1D TEM and 1D DC are widely accepted within the geophysical community to meet the scientific questions raised in this project. To define the sediment thickness, we made use of the fact that usually the electrical resistivity of crystalline bedrock is much greater than any sedimentary overburden. Regarding possible different layers within the sediments, such as alluvial fans or tephra layers, they can be distinguished through significant resistivity contrasts. Additionally, 2D TEM measurements can assist in gaining information about the geological structure of the investigated sedimentary basins.
The 1D inversion results using laterally constrained spatially constrained inversions as well as the conventional Marquardt and Occam inversion results can be summarised as follows:
Due to the hyper-saline subsurface, resistivities barely exceeded 1 Ωm in Chew Bahir. Neither in the results of laterally constrained inversion, nor in the results from the Marquardt inversion could any bedrock signature be detected. The slightly higher resistivities at depths of approximately 20 m between beneath the profiles close to the borehole may refer to an alluvial fan intruding from the western boundary of the basin. The comparison between the results from the core analyses and geophysical results is still work in progress.
The resistivities of different layers derived from the spatially constrained inversion of the complete TEM array recorded at the sites of Bisare River vary in the range of 10 to 100 Ωm. At depths around 100 m, the inversion results indicate a well-resistive layer, which is supposed to be crystalline bedrock. A 3D image of the SCI result from Bisare River could be derived. The first layer boundary within the sediments is located between 5 and 10 m. The second layer is found at a depth of approximately 20 m. The third layer indicates what we interpreted as the crystalline bedrock layer. It is located between 100 and 150 m in depth and generally dips towards the north-west. Therefore, sediment thicknesses are believed to be at least of up to 100 m. The greatest sediment thicknesses for possible future drilling projects are supposed to be in the north-west region of the recorded array. Here, the sediment thickness may extend to depths greater than 100 m.
The results of the laterally constrained inversion of TEM data from Lake Dendi indicate a 5-layer subsurface containing a middle layer of relatively high resistivities around 250 Ωm. This layer is not to be interpreted as crystalline rock because it is followed by a better conducting layer with resistivities of approximately 10 Ωm. This resistive third layer is believed to represent one or more tephra layers. The resistivities of the bottom layer (at depths of ~150 m) around 100 Ωm may indicate something like bedrock but the degree of resolution and sensitivity of the recorded TEM data at these depths is not sensitive enough to allow a reliable interpretation.
In addition to 1 D inversion results of observed TEM data in Ethiopia 2D modelling studies were also carried out to explain the time dependence of the horizontal component of the magnetic field. In case of a 1D subsurface and when placing the horizontal receiver coil exactly in the centre of the transmitter loop, any recorded horizontal component should be zero. However, we observed sign reversals at early times. The 2D modelling results indicated that these sign reversals could result from an off-centre position of the receiver. A possible second sign reversal at times around 10 ms might be also due to lateral inhomogeneities in the subsurface.
