Takarkori rock shelter (SW Libya): an archive of Holocene climate and environmental changes in the central Sahara

Quaternary Science Reviews 101 (2014) 36e60

Contents lists available at ScienceDirect

Quaternary Science Reviews journal homepage: www.elsevier.com/locate/quascirev

Takarkori rock shelter (SW Libya): an archive of Holocene climate and environmental changes in the central Sahara Mauro Cremaschi a, Andrea Zerboni a, *, Anna Maria Mercuri b, Linda Olmi b, Stefano Biagetti c, d, 1, Savino di Lernia e, f
degli Studi di Milano, Via L. Mangiagalli 34, I-20133 Milano, Italy Dipartimento di Scienze della Terra “A. Desio”, Universita
di Modena e Reggio Emilia e Viale Caduti in Guerra 127, Laboratorio di Palinologia e Paleobotanica, Dipartimento di Scienze della Vita, Universita I-41121 Modena, Italy c The Italian Society for Ethnoarchaeology, Via dei Duchi di Castro 1, I-00135 Roma, Italy d Institute for Applied Archaeology and Sustainability, Via San Quintino 47, I-00185 Roma, Italy e
, Sapienza Universita
di Roma, Via dei Volsci 122, I-00185 Roma, Italy Dipartimento di Scienze dell'Antichita f School of Geography, Archaeology and Environmental Studies, University of the Witwatersrand, Private Bag 3, Johannesburg 2050, South Africa a

b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 26 December 2013 Received in revised form 17 May 2014 Accepted 3 July 2014 Available online

Rock shelters in the central Saharan massifs preserve anthropogenic stratigraphic sequences that represent both a precious archive for the prehistory of the region and a powerful proxy data for Holocene palaeoenvironments. The geoarchaeological (micromorphology) and archaeobotanical (pollen analysis) approaches were integrated to investigate the anthropogenic sedimentary sequence preserved within the Takarkori rock shelter, a Holocene archaeological site located in the Libyan central Sahara (southern Tadrart Acacus massif). The site was occupied throughout the Early and Middle Holocene (African Humid Period) by groups of hunteregatherers before and by pastoral communities later. The investigation on the inner part of the sequence allows to recognize the anthropogenic contribution to sedimentation process, and to reconstruct the major changes in the Holocene climate. At the bottom of the stratigraphic sequence, evidence for the earliest frequentation of the site by hunters and gatherers has been recognized; it is dated to c. 10,170 cal yr BP and is characterized by high availability of water, freshwater habitats and sparsely wooded savannah vegetation. A second Early Holocene occupation ended at c. 8180 cal yr BP; this phase is marked by increased aridity: sediments progressively richer in organics, testifying to a more intense occupation of the site, and pollen spectra indicating a decrease of grassland and the spreading of cattails, which followed a general lowering of lake level or widening of shallowwater marginal habitats near the site. After this period, a new occupational phase is dated between c. 8180 and 5610 cal yr BP; this period saw the beginning of the frequentation of pastoral groups and is marked by an important change in the forming processes of the sequence. Sediments and pollen spectra confirm a new increase in water availability, which led to a change in the landscape surrounding the Takarkori rock shelter with the spreading of water bodies. The upper part of the sequence, dating between c. 5700 and 4650 cal yr BP records a significant environmental instability towards dryer climatic conditions, consistent with the end of the African Humid Period. Though some freshwater habitats were still present, increasing aridity pushed the expansion of the dry savannah. The final transition to arid conditions is indicated by the preservation of ovicaprines dung layers at the top of the sequence together with sandstone blocks collapsed from the shelter's vault. On the contrary, the outer part of the sequence preserves a significantly different palaeoenvironmental signal; in fact, the surface was exposed to rainfall and a complex pedogenetic evolution of the sequence occurred, encompassing the formation of an argillic laminar horizon at the topsoil, the evolution of a desert pavement, and the deposition of Mn-rich rock varnish on stones. These processes are an effect of the general environmental instability that occurred in the central Sahara since the Middle Holocene transition. Finally, the local palaeoclimatic significance of the sequence fits well with Holocene regional and continental environmental changes

Keywords: Rock shelter site Site formation processes Climate changes EarlyeMiddle Holocene Micromorphology Palynology Hunteregatherers Pastoralists Tadrart Acacus Sahara

* Corresponding author. Tel.: þ39 02 50315292; fax: þ39 02 50315494. E-mail address: [email protected] (A. Zerboni). 1 www.ethnoarchaeology.org. http://dx.doi.org/10.1016/j.quascirev.2014.07.004 0277-3791/© 2014 Elsevier Ltd. All rights reserved.

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

37

recorded by many palaeohydrological records from North Africa. This highlights the potential of geoarchaeological and archaeobotanical investigations in interpreting the palaeoenvironmental significance of anthropogenic cave sediments in arid lands. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction Key to understand Holocene palaeoclimatic and palaeoenvironmental oscillations is disentangling and isolating climatic and human impacts on the landscape and environment and distinguishing the relative contribution of each (e.g., Cullen et al., 2000; Messerli et al., 2000; Bolle, 2003; Ruddiman, 2003, 2007; € pelin, Butzer, 2005; Coombes and Barber, 2005; Kuper and Kro 2006; Roberts et al., 2008; Zanchetta et al., 2013). Within this framework, the geoarchaeological and archaeobotanical study of archaeological sequences, defined as human-disturbed deposits, is of crucial importance to understand the human overprint on climate-triggered palaeoenvironmental changes and geomorphic processes (e.g., Anderson et al., 2007; Cremaschi and Zerboni, 2011; Mercuri et al., 2011; Roberts et al., 2011b). However, to infer palaeoenvironmental and palaeoclimatic information from humandisturbed deposits, a multidisciplinary approach integrating the study of sediments, biological remains and material culture is required. Cave sediments formed during the human occupation of rock shelters (e.g., Goldberg and Macphail, 2006; Weiner, 2010 and references cited therein) are an example of anthropogenic deposits with great potential for palaeoclimatic studies. Among such samples, the Holocene infillings of the rock shelters of the Tadrart Acacus Massif (Central Sahara, South-West Libya) play an important role in palaeoenvironmental reconstructions, as they resulted from peculiar depositional and post-depositional processes controlled by both natural and anthropic factors (e.g., Cremaschi, 1998; Cremaschi and di Lernia, 1999a; Mercuri, 2008a; Cremaschi and Zerboni, 2011; Biagetti and di Lernia, 2013). These deposits date back to a pivotal period for human development as they include the transition from hunteregatherer subsistence to food production (e.g., Barich, 1987; Cremaschi and di Lernia, 1998; di Lernia, 1999, 2001, 2002). This event occurred in the Early Holocene, during the African Humid Period (AHP), c. 11,000e6000 cal yr BP, when the Sahara, different from its present aridity, was fed by monsoon rain zine, 1989; Gasse and and covered by patches of savannah (e.g., Le Van Campo, 1994; Gasse, 2000; Hoelzmann et al., 2004; Kuper € pelin, 2006; Le zine et al., 2011). and Kro Numerous caves and rock shelters dot the walls flanking the fossil drainage network that dissect the Tadrart Acacus massif. They formed under a Tertiary warm and humid (tropical) climate thanks to solutional processes (Cremaschi, 1998; Zerboni, 2011). Also, most of them are renowned for their rock art galleries (e.g., Mori, 1965; di Lernia and Zampetti, 2008), placed on the UNESCO World Heritage List in 1985. Since the 1960s, archaeological excavation carried out at several cave sites in Libya and Algeria (for instance, at Ti-nHanakaten, Ti-n-Torha, wadi Athal, Uan Telocat, Uan Tabu, Uan Afuda, Uan Muhuggiah, Fozzigiaren) has illustrated that the infillings of rock shelters preserve sequences of utmost archaeological and biological relevance, most of which include the Upper Pleistocene and the Early and Middle Holocene (Pasa and Pasa Durante, 1962; Barich and Mori, 1970; Aumassip and Delibrias, 1982; Hachi, 1983; Aumassip, 1984; Barich, 1987; Schulz, 1987; Wasylikowa, 1992; Cremaschi, 1998; Mercuri et al., 1998; Cremaschi and di Lernia, 1998, 1999a; Cremaschi and Trombino, 1999; di Lernia, 1999; Garcea, 2001; Mercuri, 2008b; Linseele et al., 2010;

Cremaschi and Zerboni, 2011; Biagetti and di Lernia, 2013). The peculiarity of the sequences is the surprising preservation of organic matter, especially for the Holocene contexts; this has made these deposits high-quality archives in which archaeological evidence can be studied in its environmental context. Specifically, they offer the possibility to understand (if and) how human groups coped locally with environmental variations and changes in natural resource availability over the course of the Holocene. In many cases, the human overprint on sediments, in combination with local processes (e.g., human-driven processes and micro-environmental factors), has increased the complexity of stratigraphic records. This makes it difficult to interpret such records from the perspective of palaeoenvironmental reconstruction. Taking this into consideration, the sequence of the Takarkori rock shelter (Fig. 1) was selected to perform an interdisciplinary study of the natural and anthropic depositional and post-depositional processes affecting a central Saharan cave filling. Due to a thick and composite stratigraphic sequence that spans several millennia and a rich archaeological context, the Takarkori site might be considered representative of the central Saharan massifs, being one of the few locales in the central Sahara preserving the transition from hunting and gathering to food production. Some of the archaeological and bio-anthropological aspects have been already published (Biagetti et al., 2004, 2009; Tafuri et al., 2006; Biagetti and di Lernia, 2007, 2013; Olmi et al., 2011; di Lernia et al., 2012; Dunne et al., 2012; Biagetti and di Lernia, 2013; Cherkinsky and di Lernia, 2013; di Lernia and Tafuri, 2013), confirming that the site represents an outstanding laboratory for multidisciplinary archaeological research. This paper focuses on processes that contributed to the formation of the stratigraphic sequence, mostly on the basis of geoarchaeological and palynological evidence. Data are interpreted from a palaeoclimatic and palaeoenvironmental perspective. In fact, besides human activities, global environmental factors and local environmental (or micro-environmental) settings are interpreted as actors in the formation processes. Furthermore, data from the site are compared with the regional and continental archives for Holocene environmental modifications, demonstrating the high sensitivity of Saharan cave sediments to global climate changes. 2. Palaeoclimate and past environments of the central Sahara From the Early to the Middle Holocene, the central Sahara enjoyed a period of high rainfall, as did the entire region (e.g., Cremaschi, 1998, 2002; Gasse, 2000; deMenocal et al., 2000; Hoelzmann et al., 2004; Mayewski et al., 2004; Kuper and €pelin, 2006; Wendorf et al., 2007; Arbuszewski et al., 2013). Kro Data from interdune lake deposits (Cremaschi, 1998; Zerboni, 2006; Cremaschi and Zerboni, 2009; Zerboni and Cremaschi, 2012), spring tufa (Cremaschi et al., 2010) and anthropogenic sequences inside rock shelters (Cremaschi, 1998; Mercuri, 2008b; Cremaschi and Zerboni, 2011) indicate that the renewal of water reservoirs and the expansion of the savannah vegetation date to the beginning of the Holocene. The recharge of the aquifers during the AHP was driven by the extension of the summer monsoon from the Gulf of Guinea and the migration of the ITCZ (Intertropical Convergence Zone) to northern positions (Gasse, 2000; deMenocal et al., 2000),

38

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

Fig. 1. (A) Landsat satellite imagery showing the position of the Takarkori rock shelter in the southern Tadrart Acacus massif; the grey transect refers to the area of the Wadi Takarkori Project, while the star indicates the position of the rock shelter. The insert (B) indicates the study site in its regional context and illustrates the extant position of the ITCZ (dotted line) and main regional winds (arrows). (C) A detail of the Takarkori area (Google Earth™) encompassing the rock shelter and the depression formerly occupied by a lake (indicated by the arrow).

impacted by a maximum solar heating of the Earth. In the region, a dramatic interruption of the AHP is dated to c. 8000 cal yr BP (Cremaschi et al., 2010; Zerboni and Cremaschi, 2012), while its termination, preserved in palaeohydrological records and pollen spectra, occurred after the Middle Holocene transition (Cremaschi, 1998; Cremaschi et al., 2006; Mercuri, 2008b; Cremaschi and Zerboni, 2009). At that time, water resources decreased and the desert expanded up to its current limit. This led to the present desert conditions following different ways as the varying physiographic and biological features responded differently to aridification (Cremaschi and Zerboni, 2009, 2011; Mercuri et al., 2011). Pollen and plant macro-remains from the anthropogenic sediments of rock shelters give further information about the environmental conditions during the Early and Middle Holocene and the adaptive strategies of human groups living in the area. Permanent wet habitats (with Typha and aquatics such as Potamogeton) and a fairly continuous grassland (abundant caryopses of Brachiaria, Urochloa and other Paniceae) were featured in the plant landscape during the earlier phases of human occupation, which would have provided an abundance of natural resources to gatherers and foragers until c. 8600 cal yr BP (Mercuri, 2001 and references therein; Mercuri, 2008b). Wet conditions permitted the maintenance of moist environments and development of a CyperaceaeePoaceae savannah cover until c. 6300 cal yr BP. The green

cover was suitable to sustain domestic animals of pastoralists and withstand their browsing. However, the perennial vegetation of the first phase gave way to semi-arid seasonal savannah in the second. The decrease of freshwater habitats, increase of bushlands and the addition of new species, floristic changes in grass cover and the spread of desert vegetation were fairly gradual under increasing seasonality. The psammophilous vegetation (e.g., pollen of Cornulaca monacantha and Moltkiopsis ciliata), indicating the spreading of dunes and beginning of a hyperarid environment, expanded after c. 6300 cal yr BP (Mercuri, 2008a,b). During the AHP, caves and rock shelters of the central Sahara were frequented regularly (Mori, 1965; Barich, 1987; Cremaschi and di Lernia, 1998; di Lernia, 1999; Garcea, 2001). They were occupied first by groups of Early Holocene hunteregatherers (between c. 11,100 and 8200 cal yr BP), and later by cattle and sheep/goat herders (between c. 8000 and 4500 cal yr BP); during historical times rock shelters were occasionally occupied by the Garamantes, while in recent times they have been exploited by Tuareg. 3. The Takarkori rock shelter in its regional setting The Takarkori rock shelter is located on the left bank of wadi Takarkori (Fig. 1), which is the largest pass that separates the Tadrart Acacus in Libya from the Algerian Tadrart (Desio, 1937; El-

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

ghali, 2005) and was identified during the geoarchaeological survey of the area in 1999 by one of the authors (MC). A detailed geomorphological survey of the Takarkori area (in the field for the Libyan territory, on the basis of remote sensing analyses for the Algerian bank) elucidated the physiographic settings of the region. The rock shelter is close to an endorheic basin (Figs. 1 and 2), fed by a complex fluviatile system originating from the Algerian Tassili and, on the basis of geoarchaeological evidence, active until the onset of desert conditions after the Middle Holocene transition. During the AHP the depression was occupied by a lake fed by several influents that descended from the Algerian heights. The shores of the lake (Fig. 2) are at present indicated by the occurrence of a silty to sandy sediment rich in organic matter; they are quite evident and are dotted by the vestiges of Holocene archaeological sites (fireplaces, pits, grinding equipments, lithics and pottery), which record the life of the lake up to the Middle Holocene. The present climate of the region is hyperarid. The mean annual temperature is approximately 30  C (at the Ghat weather station), and the mean annual rainfall is between 0 and 20 mm, mostly distributed in spring and summer (Walther and Lieth, 1960; ElTantawi, 2005). The surrounding vegetation is sparse and limited to AcaciaePanicum communities (White, 1983; Schmidt, 2003; Mercuri, 2008a). The rock shelter is positioned on a structural terrace of the Tadrart Acacus sandstone massif, approximately 100 m above the

39

wadi floor (Figs. 1 and 2). It opens westwards onto a large shelf (ca 2200 m2) and, through the process of undersapping, developed in the shape of alcove-type morphology (Fig. 3) at the interlayer of the local sandstone (Young et al., 2009). Most of the archaeological deposit has been eroded, as attested by artefacts' distribution on the surface of a stone-dominated desert pavement surrounding the site (Biagetti et al., 2004). However, the deposit survives (ca 200 m2) in the most recessed part of the shelter (Figs. 3 and 4): its preservation is also due to the presence of boulders that fell from the vault as well as arranged stones, which constitute part of the same filling. This part of the deposit was excavated under the direction of one of us (SdL) between 2003 and 2006 (Biagetti and di Lernia, 2013) in four sectors over 143 m2 of investigated surface (Fig. 4): (i) the Northern Sector (TK-NS), a 4  2 m trench; (ii) the Main Sector (TK-MS), the largest area of excavation (117 m2), near the rock wall and better protected by wind erosion (it includes a stone cairn found empty); and (iii) the Western Sector (TK-WS), a 3  3 m trench located beyond the drip line of the rock shelter, where the surface potsherds were present in the highest concentration. 4. Material and methods The strategies employed during the archaeological excavation of the deposit and the main archaeological finds are described in

Fig. 2. Geomorphological map of the wadi Takarkori area illustrating the main geomorphological units and their age; the star indicates the position of the Takarkori rock shelter. Key: 1) Palaeozoic sandstone bedrock; 2) slope deposits (Tertiary/Quaternary); 3) red dunes (Tertiary/Early Quaternary); 4) yellow dunes (Holocene); 5) fluvial gravel (Holocene); 6) swamp deposit (Holocene); 7) shore deposit (Holocene); 8) recent alluvium (Late Holocene).

40

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

Fig. 3. (A) General view of the Takarkori rock shelter; an archaeologist for scale (arrow). (B) Panoramic view of the landscape surrounding the site as seen from the outer part of the rock shelter; note in the lower part of (B) the position of the ancient swamp and in the background the Algerian Tassili.

Biagetti and di Lernia (2013 and references therein). This study reports sedimentological, micromorphological and palynological analyses carried out on samples collected from the TK-NS and along a section opened in the TK-WS. The deposits found in the inner part of the rock shelter (TK-NS), due to their position, suffered limited wind erosion; on the other hand, sediments from TK-WS, which are located outside the drip line, suffered more severe weathering. Comparison of the results from the two deposit areas (TK-NS and TK-WS) offers the opportunity to understand the depositional and post-depositional processes acting in two different topographic settings. Besides archaeological excavation, the area surrounding the rock shelter was surveyed to check the geomorphological context of the range, which humans crossed to find natural resources (Biagetti and di Lernia, 2013). The survey was mostly dedicated to the area at the bottom of the wadi Takarkori, where satellite imagery evidenced the existence of an ancient lake (Figs. 1 and 2). Archaeological sites are very common along the shores of the lake, and this is a further proof that a strict relationship existed between the area of the lake and the rock shelter. From a palaeoecological point of view, in the Holocene the flat area at the basis of the site, which also

Fig. 4. Simplified representation of the areas investigated within the Takarkori rock shelter. Key: 1) limit of the rock shelter; 2) drip line; 3) margin of the terrace; 4) position of the stratigraphic sections described in the text and in Tabs. 1 and 2, and reported in Figs. 5 and 6 (AA0 : TK-NS; BB0 : TK-WS).

included the swamp, was exploited for a long time and represented a composite unit of the archaeological landscape, where the inhabitants of Takarkori acted during their occupation of the rock shelter. 4.1. Chronology of the stratigraphic sequence The periodization of the Holocene human occupation of the rock shelter is crucial for a comprehensive chronological assessment of the strata formation and to interpret their palaeoenvironmental significance. A first chronology of the stratigraphic sequence was provided by archaeological finds (mostly pottery and lithics), classified according to the cultural phases of the Holocene Prehistory in the Sahara as discussed in Cremaschi and di Lernia (1998) and di Lernia (1999). The archaeological stratigraphy was also supported by more than 40 conventional and AMS (Accelerator Mass Spectrometry) radiocarbon dates performed on organic matter-rich sediment, charcoal, dried plant, human and faunal remains. 14C results are published in Dunne et al. (2012), di Lernia et al. (2012), di Lernia and Tafuri (2013), Biagetti and di Lernia (2013) and Cherkinsky and di Lernia (2013): the latter paper includes a full discussion of calibration and statistical distribution of radiocarbon dates from the Takarkori rock shelter and the time constraints of archaeological horizons. Calibrated dates, expressed as cal yr BP, were obtained using OxCal online version 4.1 (Bronk Ramsey, 2009); they are summarized in Table 3. According to the results discussed by Cherkinsky and di Lernia (2013), the Takarkori rock shelter was occupied during several phases, which overlap with the main cultural horizons identified for the Holocene exploitation of the Tadrart Acacus massif (Cremaschi and di Lernia, 1998). The cultural horizons are: Late Acacus (LA) from 10,170 to 8180 cal yr BP; Early Pastoral Neolithic (EP), 8000 to 6890 cal yr BP; Middle Pastoral Neolithic (MP), 7160 to 5610 cal yr BP; and Late Pastoral Neolithic (LP), 5700 to 4650 cal yr BP. On the basis of archaeological data (Biagetti and di Lernia, 2013), it is possible to distinguish between an exploitation of the rock shelter by groups of hunteregatherers in the early Holocene (LA phase) and a longer Pastoral Neolithic occupation (EP, MP, LP phases) characterized by cattle and ovicaprine herders. Stone structures, huts, and fireplaces together with large quantities of grinding stones and potsherds suggest a prolonged, semiresidential use of the area, which started with the occupation of the LA foragers from c. 10,170 to 8180 cal yr BP (Biagetti et al., 2004, 2009; Biagetti and di Lernia, 2013). The EP occupation is represented by some fireplaces and strips of organic layers together with several human graves (Tafuri et al., 2006; di Lernia and Tafuri, 2013). These finds are the remains of thicker deposits

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

41

Table 1 Description of the section of TK-NS, southern wall (chronology according to Biagetti and di Lernia, 2013). Unit

Description

Cultural phase

Chronology (cal yr BP)

0 I

Loose, pink (7.5YR 7/4) aeolian sand Organic loose sand, light grey (2.5Y 7/2); concave and planar lamination; common vegetal remains, charcoal and coprolites; large stones from vault collapse; abrupt erosive boundary sloping to the west Organic sand, greyish brown to light olive brown or olive brown (2.5Y 5/2, 2.5Y 5/3, 2.5Y 4/4); from loose to compact; planar bedding gently sloping to the west; frequent charred and uncharred vegetal remains; lenses of white ash (10Y 8/1) and charred material; clear boundary Loose thin laminated sand, very pale brown (2.5Y 7/2); frequent uncharred vegetal remains and coprolites; abrupt linear boundary Organic sand, greyish brown to light olive brown or olive brown (2.5Y 5/2, 2.5Y 5/3, 2.5Y 4/4); from loose to moderately hard; planar bedding gently sloping to the west; frequent charred and uncharred vegetal remains. Lenses of white (10YR 8/1) ash and black charred material. Some large structural stones. clear (erosive?) concave boundary Massive organic loose sand including two dark greyish black (2.5Y 4/2) thin layers of laminated sand cemented by organic matter; clear boundary gently sloping to the west Loose sand, light olive brown (2.5Y 5/3); scarce vegetal fragments, often charred; planar bedding slightly sloping to the west; clear planar boundary, interrupted to the west by a large stone structure Loose organic sand, olive brown (2.5Y 4/4); planar discontinuous bedding gently sloping to W; rare vegetal remains distributed in thin layers; rare weathered sandstone fragments; in the east side planar clear boundary to the bedrock, to the west affected by hardened filling of large stone structure

e Middle Pastoral 2

e 6300e5750

Middle Pastoral 1

6950e6300

Early Pastoral 1

8250e7800

Late Acacus 3

8950e8450

Late Acacus 2

9450e8950

Late Acacus 2

9450e8950

Late Acacus 1

9950e9450

II

III IV

V VI VII

possibly removed by erosion and indicate the use of the site by cattle herders from c. 8180 to 6890 cal yr BP. The MP occupation of the rock shelter, radiocarbon dated between c. 7160 and 5610 cal yr BP, is a phase characterized by a fully cattle-based economy, which included dairying as confirmed by the presence of bones of Bos taurus (F. Alhaique, pers. comm.) and organic residues on the potsherds (Dunne et al., 2012). Finally, strips of ovicaprine dung and organic sands with cuvette hearths are the evidence of specialized LP goat herders who used the site probably during the winter season between approximately 5700 and 4650 cal yr BP. Four more dates were obtained from samples collected directly along the section of the North Sector (Fig. 5) which is studied in detail in the present paper: one date, 5170 ± 25 uncal yr BP (ovicaprid dung: 5990e5900 cal yr BP), comes from Unit I; Unit IV dates to 7910 ± 30 (seeds: 8800e8600 cal yr BP), Unit VI to 8410 ± 30 (charcoal: 9520e9400 cal yr BP), and Unit VII to 8820 ± 60 (charcoal: 10,170e9670 cal yr BP). The archaeological content ensures a strict correlation with the deposits inside the rock shelter. 4.2. Description of the sequences and sampling sites 4.2.1. Northern Sector (TK-NS) The TK-NS filling is 1.6 m thick and lies upon the sandstone bedrock of the rock shelter. The stratigraphic sequence has been macroscopically grouped into main Units based on macroscopic features, main unconformities and archaeological content (Biagetti and di Lernia, 2013). The sampling of the sequence for laboratory

analyses was performed on the southern trench wall (Fig. 5, Table 1), with additions from the northern wall. From the top, there is Unit 0-NS, a thin layer of aeolian sands. Unit I-NS (Middle Pastoral 2 or MP2) includes organic sand mixed with dried and charred plant remains and several hearths. Unit IINS (Middle Pastoral 1 or MP1) is similar to Unit I-NS, though the organic fraction is less recurrent and hearths and ash lenses increase. Early Pastoral layers are represented in Unit III-NS, where unhumified plant remains and coprolites have been observed within an organic sand matrix. Late Acacus 3 (LA3) archaeological contexts are included in Unit IV-NS. These are similar to Unit III-NS, but they are characterized by larger hearths and ash dumps. Unit VNS (Late Acacus 2 or LA2) features lenses of undecomposed plant remains, but planar and almost continuous slurry levels characterize this part of the section. Unit VI-NS (Late Acacus 1 or LA1) comprises loose sand with reducing plant remains. Unit VII-NS (LA1) contains coarse light yellow sand, intermixed with wellpreserved plant remains and charcoal. A vertical set of 26 pollen samples (at 5e10 cm intervals) plus 5 samples from lateral positions were collected from the southern wall. Seven samples for micromorphological analyses were taken to check specific contexts throughout the sequence (Fig. 5). Samples for sedimentological analyses were collected in the vicinity of pollen samples. To better appreciate lateral variations in the deposit and their environmental meanings, five other pollen samples were also collected from the northern wall (two from Unit V-NS, samples 35 and 36; three from Unit VII-NS, samples 28e30), with one sample for micromorphological analysis (no. 8, Unit V-NS).

Table 2 Description of the section of TK-WS, eastern wall (chronology according to Biagetti and di Lernia, 2013). Unit

Description

Cultural phase

Chronology (cal yr BP)

0

Desert pavement veneered by pink (7.5Y 7/4) aeolian sand; isolated stones, their exposed part is coated by a Mn-rich rock varnish; clear planar boundary B21t horizon (cm 0e15); sandy loam, reddish yellow (7.5Y 7/6), few stones, moderate discontinuously platy structure, moderately hard, undulated clear boundary B22t horizon (cm 16e40); sandy loam, yellowish brown (10Y 5/4), rare to common stones, weak platy to subangular blocky structure, slightly hard, clear boundary 2A horizon (cm 41e70); sandy loam, dark grayish brown (10Y 4/2), rare stones, massive, slightly hard, gradual boundary to 2AC horizon (cm 71e80), sandy loamy sand, grayish brown (2.5Y 5/2), slightly hard to soft, lower boundary not exposed

e

e

Middle Pastoral 2

6300e5750

Late Acacus 3

8950e8450

Late Acacus 3 (?)

8950e8450 (?)

I II III

42

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

Table 3 Radiocarbon dating results and calibration (according to Bronk Ramsey, 2009) from the Takarkori rock shelter (modified, Cherkinsky and di Lernia, 2013). Laboratory number

Stratigraphic number

Description of the context

Cultural attribution

Material

Uncalibrated age

LTL670A GX-31071 GX-30325 LTL908A UGAMS#8707 LTL907A LTL362A UGAMS#10149 UGAMS#01843 UGAMS#01841 GX-31077 LTL367A UGAMS#2852 GX-30324-AMS UGAMS#01842* GX-31074-AMS GX-31073-AMS LTL1585A GX-31075-AMS LTL911A GX-30326 GX-31064 GX-31076-AMS LTL1586A LTL914A GX-31066 GX-31069 LTL369A GX-31070 LTL364A LTL365A GX-31068 UGAMS#10148 UGAMS#8708 LTL910A LTL368A GX-31065 LTL366A GX-31067 GX-31072 UGAMS#10150 UGAMS#10147 UGAMS#01844

H13 76 (pit) 6 24 25 22 25 330 374 25 H9 25 25 H1 41 H5 H4 H11 H6 H10 41 93 H7 H12 H14 96 108 38 123 150 96/103 103 337 359 H8 96/103 129 101 103 136 351 344 408

Burial Item Dung Hearth Organic sand Hearth Item Organic sand Formal stone structure Organic sand Burial Organic sand Organic sand Burial Organic sand Burial Burial Burial Burial Burial Organic sand Floor Burial Burial Burial Organic sand Hearth Humified organic sands Ash dump Hearth Organic sand Organic sand Ash dump Organic sand Burial Organic sand Hearth Hearth Organic sand Hearth Hearth Floor Hearth

Late Pastoral Late Pastoral Late Pastoral Late Pastoral Middle Pastoral Middle Pastoral Middle Pastoral Middle Pastoral Middle Pastoral Middle Pastoral Middle Pastoral Middle Pastoral Middle Pastoral Middle Pastoral Early Pastoral Early Pastoral Early Pastoral Early Pastoral Early Pastoral Early Pastoral Early Pastoral Early Pastoral Early Pastoral Early Pastoral Early Pastoral Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus Late Acacus

Human bone Skin (sheep/goat) Dung Coprolite Seeds Charcoal Skin (sheep/goat) Dung strip Collagen Collagen Bone collagen Dung strip Enamel bioapatite Human bone Collagen Human bone Human bone Human bone Human bone Human bone Dung strip Soil Human bone Human bone Human bone Soil Soil Charcoal Charcoal Charcoal Dung strip Soil Seeds Seeds Human bone Charcoal Soil Charcoal Soil Charcoal Charcoal Charcoal Charcoal

4291 4590 4800 4841 4970 5064 5070 5170 5280 5340 5600 5980 5980 6090 6230 6540 6740 6763 6900 7068 7070 7130 7130 7155 7327 7470 7580 7694 7730 7801 7820 7890 7910 7930 7973 8031 8040 8049 8080 8290 8410 8410 8820

4.2.2. Western Sector (TK-WS) The TK-WS deposit is rather homogeneous and almost massive; a stone line at the depth of c. 30 cm is the sole evidence for stratification in this part of the deposit. For that reason, the differentiation of archaeological sub-phases is only considered tentative and, lacking radiocarbon dates, given the absence of organic substance, was determined mostly on the basis of the poor archaeological evidence. The sequence was excavated to a depth of c. 90 cm without reaching bedrock (Fig. 6, Table 2). It is characterized, from the top, by Unit I-WS, which is a layer displaying a lamellar aggregation, with archaeological materials of Middle Pastoral age (MP2). Below it, a layer with a lamellar aggregation (Unit II-WS) including materials of Late Acacus age (LA3) is present. Unit III-WS is a thick layer of weathered aeolian sand with bioturbation pedotubules and an archaeological discontinuity signalled by the presence of Late Acacus materials in its upper part. The deposit was systematically sampled: 12 pollen samples were taken at 5e10 cm intervals, and four undisturbed samples of sediment for thin sections were taken from each pedological horizon distinguished in the three main Units. 4.3. Pedosedimentary and micromorphological analyses Thin sections from undisturbed blocks, integrated with grain size and routine chemicalephysical analyses on bulk samples, have

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

50 80 70 50 25 55 35 25 50 50 70 50 70 60 90 70 70 55 70 100 100 100 70 65 65 100 110 60 100 35 50 110 30 30 45 65 110 40 110 140 30 30 60

Calibrated years BC (95%)

Calibrated years BP (95%)

3090e2700 3630e3030 3710e3370 3750e3510 3800e3660 3970e3710 3960e3780 4003e3951 4240e3980 4330e4040 4600e4330 5000e4720 5050e4710 5210e4840 5470e4940 5630e5370 5760e5520 5750e5560 5980e5660 6210e5730 6210e5730 6230e5800 6210e5840 6210e5890 6370e6060 6490e6080 6650e6230 6640e6440 7010e6390 6700e6510 6820e6500 7060e6500 6842e6653 7030e6680 7050e6690 7140e6690 7310e6650 7140e6820 7360e6670 7600e6850 7553e7452 7553e7452 8220e7720

5040e4650 5580e4970 5660e5320 5700e5460 5750e5610 5920e5660 5910e5730 5990e5900 6190e5920 6280e5990 6550e6280 6950e6670 7000e6660 7160e6790 7420e6890 7570e7310 7710e7470 7700e7510 7930e7610 8160e7670 8160e7680 8180e7750 8160e7790 8160e7840 8320e8000 8440e8030 8590e8180 8590e8390 8950e8340 8650e8450 8770e8450 9010e8450 8800e8600 8980e8630 9000e8640 9090e8640 9260e8600 9080e8770 9310e8620 9550e8800 9520e9400 9520e9400 10,170e9670

been used to identify the stratigraphic sequence-forming processes and infer the environmental and anthropogenic factors for accumulation and post-depositional changes (Courty, 2001; Goldberg and Macphail, 2006; Goldberg and Berna, 2010). Oriented and undisturbed sediment blocks from selected Units were collected as described above. Thin sections (5  9 cm) were manufactured after consolidation according to standard methods (Murphy, 1986). Micromorphological observation under plane-polarized light (PPL) and cross-polarized light (XPL) of thin sections employed an optical petrographic microscope Olympus BX41 with a digital camera (Olympus E420). For the description and interpretation of thin sections, the reader should consider the terminology and concepts established by Bullock et al. (1985), Stoops (2003) and Stoops et al. (2010). Properties of each sample detected by thin section analysis are summarized in Table 4. Oriented samples for thin sections were collected from TK-NS and TK-WS (Figs. 5 and 6). Bulk samples from the TK-NS were also collected for textural and chemicalephysical analyses, the results of which are reported in Figs. 7 and 8. Applied methods are summarized as follows. (i) Humified organic carbon was identified following the Walkley and Black (1934) method, using chromic acid to measure the oxidizable organic carbon (titration). (ii) Total organic carbon was estimated by loss on ignition (LOI; Heiri et al., 2001); samples were air-dried and organic matter was oxidized

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

43

Fig. 5. Schematic representation of the EasteWest profile of the southern wall of the TK-NS; the position of radiocarbon-dated samples and results (uncal BP) are also indicated (profile drawn by A. Monaco and S. di Lernia). Key: a); aeolian sand; b) sand rich in organic matter; c) lenses of undecomposed plaint remains; d) ash; e) charcoal; f) slurry deposit; g) eroded sand from the wall; j) bedrock; location of undisturbed block for k) micromorphology and j) palynological samples.

at 500e550  C to carbon dioxide and ash, then the weight lost during the reaction was measured by weighing the samples before and after heating. (iii) Conductivity, reflecting the concentration of soluble salts, was performed by dispersing sediment in pure water and was measured through a conducimeter. (iv) Grain size was determined (diameter from 2000 to 63 mm) through wet sieving after removing organics by hydrogen peroxide (130 vol) treatment (Gale and Hoare, 1991).

4.4. Pollen analysis The 48 pollen samples collected from the TK-NS and TK-WS (Figs. 5 and 6) were arranged into two main pollen sequences. About 10e20 g of sediment (dry weight) were treated with tetra-sodium pyrophosphate, sieved with a nylon 7 mm mesh, and treated with HCl 10%, HF 40%, acetolysis and heavy liquid separation (sodium metatungstate hydrate; Florenzano et al., 2012).

Fig. 6. Schematic representation of the NortheSouth profile of the eastern wall of the TK-WS (profile drawn by A. Monaco and S. di Lernia). Key: a) lamellar structures; b) lamellar aggregation; c) weathered sand including bioturbation pedotubules; d) micromorphological and e) palynological samples.

44

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

Table 4 Summary of the micromorphological properties of the samples from the TK-NS (N), TK-WS (W), and TK-MS (M) sectors of the excavation. Unit

Sample

Sector

Microstructure

Biogenic constituents

Groundmass

Related distribution

Pedofeatures

S1

S2

S3

S4

Coarse minerals M1

M2

M3

M4

B1

B2

B3

B4

G1

G2

G3

G4

R1

R2

R3

P1

P2

P3

P4

   þ          

R   R   R

      

 C   F R 

      





R



   

C   

   

C C R**** 

I II III IV IV VI V

TK1 TK2 TK3 TK4 TK5 TK6 TK7

Northern Northern Northern Northern Northern Northern Northern

þ  þ þ  þ 

 þ   þ  þ

 þ þ  þ  þ

      

R R R R R R 

F F C R F F F

 R     

      R

C R F R C* R C

C C C R C C R

C R R C R R R

R R   R R 

þ þ þ   þ þ

F F C F  R C

      

    þ  

þ  þ   þ þ

 þ   þ  þ

AC58

TK8

Main



þ

þ



R

F



R

C

R

C

R

þ

R





þ



Bt21 Bt22 2A 2AC

W1 W2 W3 W4

Western Western Western Western

 þ þ þ

þ   

   

þ þ  

C C R R

F F F F

R   

   

   

R** R**  

   

R  R R***

   

   

þ þ  

   

þ þ þ þ

þ þ  

Microstructure: S1, bridged grain structure/single grain structure; S2, intergrain microaggregate structure; S3, platy structure (trampling); S4, laminated structure (sedimentary, colluvial). Coarse mineral components (>250 microns): M1, lithorelicts (sandstone fragments); M2, medium and coarse rounded and subrounded quartz grains; M3, pedorelicts; M4, pottery fragments. Biogenic components: B1, vegetal fragments; B2, charcoal (>500 microns); B3, coprolites; B4, bone fragments. Groundmass: G1, undifferentiated to poorly b-spekled b fabric; G2, punctuated organic fabric or organic pellets; G3, birefringent fabric; G4, crystallitic fabric. C/F related distribution: R1, chitonic; R2, close porphyric; R3, monic. Pedofeatures: P1, pathches of phosphate (apatite); P2, impregnative features: calcitic concretions and pendents; P3, impregnative features: pseudomorphic calcite aggregates, ash connected; P4, texturale pedofeatures: dusty lamimated clay coatings. *Charred vegetal fragments; **finely subdivided charcoal; ***fish bones; ****isotic. Estimated concentration: R, rare (<15%); C, common (16e30%); F, Frequent (31e>50%).

Lycopodium spores were added to calculate pollen concentration, expressed as pollen per gram (p/g). Permanent slides were mounted with glycerol jelly. Pollen analyses were made at 400 and 1000 with immersion oil. About 340 (TK-NS) and 140 (TK-WS) pollen grains per sample were counted on average. Pollen identification was based on the reference pollen collection and relevant literature (Bonnefille and Riollet, 1980; Reille, 1992, 1995, 1998; Ayyad and Moore, 1995). Some types were distinguished within the Poaceae, i.e., the >40 mm and larger pollen types, and the 18e26 mm pollen of Phragmites with imperceptible punctae in a scabratae exine (Hall, 1981). Several floras are used here to determine habitus, distribution, ecology and phytogeographical affinity of genera and species (Corti, 1942; Turril and Milne-Redhead, 1952; Maley, 1980, 2004; White, 1983; Ozenda, 2000; Watrin et al., 2009). The pollen nomenclature follows the African Pollen Database. Percentages are calculated on a pollen sum, which includes all pollen grains. Sums of the most abundant taxa are calculated as ‘Dry’ and ‘Wet’ sums. The first sum includes Asteraceae and Chenopodiaceae/Amaranthaceae, hereafter Chenopodiaceae, that are typical of dry environments; the second sum includes Poaceae and Cyperaceae that are more abundant under comparatively moist conditions (Fowell et al., 2003; Herzschuh et al., 2004). The D/W ratio was used by Hooghiemstra (1996) for vegetational shifts in marine cores, and similar indexes are useful as crude indicators of main vegetational changes: e.g., the ratio Artemisia to Cyperaceae is used to distinguish steppe to meadow vegetation in lake sediments (Li et al., 2011); Cyperaceae to Poaceae is calculated in order to distinguish the lake marginal from terrestrial source of pollen (Gillson, 2006); Artemisia to Chenopodiaceae ratios help to discriminate between steppe-like vegetation and desert-like environments and to infer changes in moisture conditions (El-Moslimany, 1990). Some of these indexes are useful tools for qualitative and semi-quantitative palaeoenvironmental reconstruction (Herzschuh, 2007). In this paper, the D/W ratio is used as an indicator of relative moisture conditions based on the assumption that grassland vegetation needs more humidity, whilst Chenopodiaceae and Asteraceae are today the plant families that host the most common representatives of desert vegetation in the Sahara and its fringes (Ozenda, 2000).

The sum of plants associated with wet environments includes telmatophytes growing in the reed-bed belt (cattails: Typha domingensis-type, Typha latifolia-type, Typha minima-type, and reedsPhragmites-type) and a few floating aquatics (Lemna, Potamogeton). Pollen diagrams, including the calculation of the zonation by cluster analysis, were drawn with Tilia 2.0 and TGView (Grimm, 1991e1993). 5. Results 5.1. The pedosedimentary sequence 5.1.1. TK-NS Grain-size cumulative curves (Figs. 7 and 8) indicate homogeneity in the mineral fraction of the deposits, as the grain size distribution along the studied sequence is almost identical to that recorded at its base (Unit VII-NS). The latter deposit, too poorly sorted to be of aeolian origin, originated from the disaggregation of the friable Palaeozoic sandstone (fluvial and deltaic facies; El-ghali, 2005) in which this part of the rock shelter was excavated. A local origin of Unit VII-NS sand is further supported by its olive-brown colour, similar to the sandstone bedrock, but in contrast to the dominant pink-reddish colour of the aeolian sand in the Takarkori dune field and aeolian sand formations in the rock shelter's surroundings. In the thin section, the sand granules in Unit VII-NS and along the sequence are inhomogeneous in shape, associating angular to rounded habits, and are related to angular lithorelicts; this also excludes an aeolian origin of the deposits. Moreover, quartz grains are not stained by a film of iron oxides, which is common in local aeolian sand (Zerboni et al., 2011). However, the percentage of the silt þ clay fraction changes significantly along the sequence (Figs. 7 and 8), reaching a peak in coincidence at the bases of Units I-NS and II-NS (25% and 50%, respectively). Along the sequence coarse sand is around 20%, but it decreases significantly (c. 10e15%) in Units III and II and at the base of Unit I, while at its top it increases again to c. 20e25%. Humified organic carbon, loss on ignition and conductivity (Fig. 7) display similar, parallel trends. This suggests that the amount of organic matter may control the distribution of soluble salts and therefore play a main role in determining the geochemical properties of the sequence. Organic

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

45

Fig. 7. Results of sedimentological analyses (grain size, humified carbon, loss on ignition, conductivity). The main stratigraphic Units are also indicated, while dots represent the sampling spots for sedimentological analyses.

matter content and chemical parameters (Fig. 7) show higher concentration in the upper part of the sequence and a significant peak in correspondence in Unit II-NS; they then decrease, showing a saw-tooth pattern in Units IV-NS and V-NS and a more regular trend in Unit VI-NS, reaching the lowest values in Unit VII-NS. From the micromorphological point of view (Table 4), undecomposed plant fragments and coprolites dominate the whole sequence, whose matrix is made of coarse mineral components (Fig. 9). They are associated with many spherulites and phytolith chains (these elements are almost ubiquitous and therefore not specifically mentioned in Table 4). In one case (Unit IV-NS), plant fragments suffered sin-depositional biological activity, and undecomposed pellets largely dominate the thin section. Humified organic matter in the shapes of intergranular isotic fillings and punctuations occur in several cases, and it is particularly developed in sample 7 (Unit V-NS), which was collected in correspondence with a planar organic-rich layer described in the field as a ‘slurry’ deposit. Calcite micro- and macro-crystals, providing some evidence for re-crystallization, are related to ashes and often occur in

Fig. 8. Selected cumulative grain size curves for samples collected along the section of the TK-NS.

association with macroscopic evidence for hearths. Bioturbation is particularly evident in the inner part of the rock shelter, at the interface between its vault and the deposit; thin section AC58 from Unit V of the Main Sector showed a large number of voids related to invertebrate bioturbation. In this part of the excavation (at the interface with the sandstone wall), sand grains are distributed in discontinuous upward-finning laminae and a larger redistribution of calcite and soluble salts also occurred (Fig. 10), as confirmed by a diffuse crystallization (impregnation) of the groundmass. Qualitative X-ray diffraction analysis highlights the occurrence of dominant potassium nitrate (KNO3), quartz, and, in very limited quantity of calcium carbonate. Mineral and biological components of different sizes are frequently redistributed by discontinuous planar lamination due to occupation trampling (Courty et al., 1989; Matthews et al., 1997). However, the sand grains in Units VI-NS and VII-NS are distributed in discontinuous upward-finning laminae, testifying for water transportation originating from colluvial processes (Fig. 10). These features are in accordance with the discontinuous planar nonparallel bedding that is macroscopically observed. 5.1.2. TK-WS This sector is located beyond the drip line. Lateral continuity and archaeological content permit a strict correlation between the NS and WS sequences, but their pedosedimentary properties differ. No plant macro-remains were observed in the WS sequence, either in the field or thin sections. The dark colour of the deposit confirms the presence of highly humified organic matter. The sequence may thus be interpreted as a soil profile (Fig. 6), composed of two superposed laminar horizons (B21t, B22t, which represent Units I-WS and II-WS, respectively) below the desert pavement, covering organic horizons (2A and 2AC, which constitute Unit III-WS). At the micromorphological level, the microstructure of the B21t and B22t horizons (Units I-WS and II-WS) consists of discontinuous planar gradated laminae (Fig. 11), indicating a contribution of colluvial processes and water washing in to form the parent material of these horizons. Furthermore, clay illuviation in

46

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

Fig. 9. Photomicrographs of herbivore coprolites and vegetal fibres from the upper part of the stratigraphic sequence. A) Detail of a well preserved coprolite (Unit I-NS, PPL) and a detail (B) of spherulites (XPL); C) fragmented coprolites from Unit III-NS mixed to vegetal fibres (PPL); D), detail of (C) showing the platy distribution of fibres and spherulites (XPL).

intergranular voids is represented and displays a lamellar distribution pattern (Fig. 11); in the surface horizon calcite concretions also occur. In the 2A and 2AC horizons (Unit III-WS), former anthropogenic organic content is indicated by rare charcoal and

weathered bone fragments (possibly including fish bones). No plant fragments were observed, and the humified organic matter formed the isotic groundmass distributed in the intergranular spaces (Fig. 11).

Fig. 10. Photomicrographs of selected micro-features from TK-MS and TK-NS sectors. A) Sample AC 58 Unit V-MS (PPL), note the crescentic distribution of organics; B) sample AC 58 Unit V-MS soluble salts (niter) and calcite impregnations in the groundmass (XPL); C) sample 6 Unit VI-NS showing gradated distribution of the coarse mineral fractions (PPL); D) sample 7 Unit V-NS (slurry layer), intergranular organic matter concentrations with planar distribution due to trampling (PPL).

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

47

Fig. 11. Photomicrographs from TK-WS. A) Sample 1, Unit I-WS (B21t horizon), gradated distribution of coarse mineral fraction due to colluvial process (PPL); B) sample 2, Unit II-WS (B22t horizon), laminated clay coating related to intergranular voids (XPL); C) sample 3, Unit III-WS (2A horizon), minute fragments of bones in the mineral groundmass (PPL); D) sample 3, Unit IV-WS (2A horizon), enaulic related distribution of organic matter (PPL).

5.2. Palynological analysis Pollen preservation and floristic richness are different in the sequences of the TK-NS and TK-WS sectors. Spectra (Figs. 12e14) are characterized by low tree pollen values, which give evidence for sparse tree cover and grass dominance near the rock shelter. The co-presence of pollen from plants living in ecologically wet and in arid environments (Fig. 15) indicates that local and regional signals of different vegetation types, also simultaneously living in the area, are represented in the spectra. Human and animal plant transport inside the rock shelter is mainly inferred by two evidences: i) at the macroscopic level: an impressive quantity of undecomposed plant remains was observed in the layers, including high quantity of seeds/fruits and plant remains that were sorted by dry sieving or picked up collection (Olmi et al., 2011; Biagetti and di Lernia, 2013); further, thanks to ii) the microscopic analysis very high percentages and concentrations of pollen from useful plants (e.g., Typha and Artemisia) were observed in slides from similar deposits of the Teshuinat area (Mercuri, 2008a). Since it is known that pollen spectra from archaeological sites (on-site) are strongly influenced by human activities, reconstruction of plant cover and inferences regarding climate are problematic. In such contexts, humans and their animals continuously transported the pollen produced by plants living in the 'range of influence of the site' (Mercuri et al., 2012); this occurred through plant harvesting or by involuntary collection of dust by feet and skin, or by coprolites (Dimbleby, 1985; Faegri et al., 1989). Therefore, pollen accumulated by anthropogenic transport does not reflect only edaphically restricted plant communities, as typically occurred in natural sites near wadis (Mercuri, 2008b). Though over-representation of some pollen types is evident in single samples of the Takarkori sequence, the analysis of many samples from one layer/Unit helps to obtain reliable palaeoenvironmental

interpretations based on average data per phase (zone). The anthropogenic pollen, discussed within a multidisciplinary perspective, is useful for environmental reconstructions because humans must have collected plants in the wild, and changes in plant selection must have relied on changes in the environment shared by plants and humans. In the following subsections, pollen zones and sub-zones of the two sequences are reported with the corresponding stratigraphic Units and archaeological cultural phase, obtained from the interdisciplinary investigations on the site. If not differently reported, pollen percentages are mean values of the zone or sub-zone, and interpretation relies on the mean average changes per zone/subzone to avoid the one-sample over-representation bias. 5.2.1. The TK-NS pollen sequence Folded grains are common, but pollen is prevalently well preserved, especially in samples of older age. Only sample 15, corresponding to an ash layer, is sterile. Pollen concentrations are approximately 120,000 p/g on average, a very high value that mirrors the human action of on-site plant accumulation (Mercuri, 2008a). Pollen clumps are frequent, particularly in samples with very high concentrations. They indicate the deposition of blooming plants or the presence of dung or faeces (Fægri and Iversen, 1989). The floristic list contains 92 pollen taxa, including 28 woody taxa. Poaceae-grasses are dominant (63%, including 9% of >40 mm pollen). The Asteroideae-daisy family (11%, mainly Artemisia 4%), Cyperaceae-sedges (4%), Chenopodiaceae-chenopods (3%) and Brassicaceae-cabbage family (1%) are common. Only Tamarixtamarisk (4%) has a significant value among the trees (8%). Hygrophilous plants (8%) include the telmatophytes Typha-cattails (3%) and Phragmites-type-reeds (2%) while aquatics include waterfloating plants such as Potamogeton and Lemna.

48 M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

Fig. 12. Percentage pollen diagram of the profile TK-NS, selected taxa.

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

49

Fig. 13. Percentage pollen diagram of the profile TK-NS, pollen sums and concentrations. Zonation is calculated with Tilia (CONISS ¼ Constrained Incremental Sum of Squares).

Two main pollen zones and six subzones are recognized, starting from the bottom (Figs. 11 and 12). 5.2.1.1. Zone Tk 1 (18 samples, from c. 135 to 71 cm; Units VII-NS to IV-NS; archaeological context Late Acacus). Pollen concentration is c. 51,000 p/g, matching the low organic matter content found in sedimentological analyses. The tree percentage is low (11.5%; 5.0% excluding Tamarix). The more abundant or exclusive taxa to this zone are Tamarix, Capparaceae (Capparis, Maerua-type, Bosciatype), Hyphaene, Maytenus and Salvadora among trees, and Apiaceae, Asteraceae (Ambrosia-type maritima, Artemisia, Centaureatype, Cichorieae), Commelina-type, Tribulus terrestris-type, Zygophyllum. Most are part of a sparsely wooded savannah vegetation that was present together with grassland habitats. The D/W ratio is very low (0.3; 0.1 excluding Artemisia), suggesting that savannah is more represented than xerophilous vegetation in pollen spectra. The wet environments are represented by herbaceous telmatophytes and aquatics (3.7%) as well as tamarisk (6.6%). 5.2.1.1.1. Subzone Tk 1a (5 samples, from c. 135 to 120 cm; Unit VII-NS). Pollen concentrations are the lowest of the sequence. Potamogeton-pondweed, with its highest value in the sequence, marks the presence of wet environments, with reeds and cattails (especially Typha domigensis-type) growing in marginal zones of lakes, rivers and ponds. Considering the low pollen production of pondweeds that have wind or water pollination (Zhang et al., 2010), this aquatic plant is a good indicator of local permanent freshwater habitats. The psammophilous shrub Cornulaca monacantha-type is absent. There are significant values of Tamarix (up to 31%), Capparis, and scent plants like Artemisia, Mentha-type and other Lamiaceae. As they typically have been recorded in fairly coeval layers from other rock shelters and caves of the Tadrart Acacus, they are

anthropogenic pollen indicators representing medicinal and food plants harvested by hunteregatherers (Mercuri, 1999, 2008a). 5.2.1.1.2. Subzone Tk 1b (6 samples, from c. 119 to 101 cm; bottom of Unit VI-NS). Mean values of plants from wet environments increase, especially those form marginal zones like T. latifolia-type. This could be regarded as a widening of shallow-water marginal zones or general water level lowering. Grassland also decreases, while xerophilous plants increase, and Cornulaca monacantha-type is present with Urtica-type. The tropical-Sahelian Acacia, Boscia-type and Maerua-type are present, and the Capparaceae have maximum values. Tamarix has a second peak (46%), and scent-medicinal plants are still significant (Myrica besides Artemisia and other Asteraceae). 5.2.1.1.3. Subzone Tk 1c (7 samples, from c. 100 to 71 cm; top of Units VI-NS, and V-NS to IV-NS). Evidence for wet environments is low: the lack of pondweeds suggests that this plant was not so available in the area as it was previously, and therefore the further reduction of permanent water bodies had occurred. Oscillations of xerophilous plants are complementary to those of grasslands (Fig. 12). Cyperaceae pollen reaches values, on average higher than previously, that remain fairly steady until the top of the diagram. Tamarix nearly disappears. In this subzone, pollen spectra possibly mirror a period of environmental instability (bottom part), and the exploitation of different natural resources. The cultural shift toward new plants is visible in the second part of this subzone through the abandonment of Artemisia and especially through the change in the harvesting of wild cereals. This is evident in the spectra from the notable increase of large pollen types, belonging to some Panicum as well as Echinochloa and Setaria species, as visible in coeval sites (Mercuri, 2008a). The rapid oscillation of D vs. W sums in the bottom part may also be explained by an increased seasonality that made available different plant resources in different seasons (Fig. 13).

50 M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

Fig. 14. Percentage pollen diagram of the profile TK-WS, selected taxa, pollen sums and concentrations. Zonation is calculated with Tilia (CONISS).

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

51

Fig. 15. Well-preserved pollen grains from the TK-NS sequence. Key: A and B) Poaceae, large pollen >40 mm (pollen zone Tk1); C) Poaceae and phytolits (Tk1); D) Ficus (Tk2); E) Cyperaceae (Tk2); F) Typha latifolia type (Tk1); G) Cichorieae (left) and Brassicaceae (right) (Tk1).

5.2.1.2. Zone Tk2 (14 samples, from c. 70 to 6 cm; Units III-NS to I-NS; archaeological context Early and Middle Pastoral). Pollen concentrations quadruplicate to c. 207,000 p/g, suggesting high plant accumulation in this zone. This matches the peak of organic matter found in the upper part of the sequence (Figs. 6 and 12). The anthropogenic transport (by humans and animals) of plants inside the rock shelter is especially visible in the increase in grasses (food and pasture plants). In these spectra, the tree percentage notably decreases (3.3%; 2.6% excluding Tamarix). Acacia (0.6% vs. 0.1% in Tk1) and Ficus among trees, together with Fabaceae including

Astragalus, and Heliotropium, Paronychia and Plantago are more abundant or exclusive to this zone. Capparaceae, Tamarix, Artemisia and Zygophyllum became insignificant, signalling a general reduction in the dry savannah vegetation. The D/W ratio (0.1) shows that xerophytes were fairly negligible in spectra, while plants reflecting wet conditions are better represented. Hygrophilous trees are low (tamarisk and fig tree amount to 0.8%). The hygro-hydrophilous pollen sum is somewhat better represented than previously (5.6%; Fig. 13), with higher T. latifolia-type and Lemna added to Potamogeton among floating plants; its value is largely due to the

52

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

53

Phragmites-reeds that grow in the shallow-water marginal zones of ponds or rivers. The Cyperaceae percentage is also higher (5.2% vs. 3.0% in Tk1). 5.2.1.2.1. Subzone Tk 2a (2 samples, from c. 70 to 60 cm; bottom of Unit III-NS). Pollen concentrations are very high, suggesting that there was an intensive plant accumulation. However, the collected wild cereals do not include such abundant large pollen species as the previous sub-zone (Tk 1c). As pollen of plants from wet environments (mainly Phragmites) prevails over pollen of xerophytes, this event probably occurred during a wet phase favouring the spreading of grasslands. 5.2.1.2.2. Subzone Tk 2b (7 samples, from c. 59 to 30 cm; top of Units III-NS, II-NS, and bottom of Unit I-NS). Pollen concentrations are unevenly high. The increase of Potamogeton suggests that permanent freshwater bodies spread again. As xeric environments also spread, a mosaic of diverse habitats was present during this phase. Poaceae include a high quantity of different large pollen grasses in the archaeological deposit, due to increased natural availability and the cultural selection of ‘new’ wild cereals used as fodder. These were either transported and consumed within the rock shelter or consumed elsewhere and excreted at the site by domestic flocks. 5.2.1.2.3. Subzone Tk 2c (5 samples, from c. 29 to 6 cm; top part of Unit I-NS). Pollen concentrations decrease notably. A significant change toward environmentally dry conditions is evident. Two acacias (Acacia, Acacia ehrenbergiana-type), with a slight increase of Tamarix and Maerua-type, mark the diffusion of dry savannah. Accordingly, xerophytes (D sum) increase while grasses decline. The tendency has a reverse trend only in the top sample. Though the recovery of pollen of hygro-hydrophytes indicates that wet environments were still present, they must have decreased in size. In fact, there is an isolated presence of Lemna, and telmatophytes also tend to diminish. Astragalus, with other Fabaceae, Echium, some Asteraceae, Brassicaceae and Caryophyllaceae, represent the plants brought into the rock shelter as fodder or deposited with furs, hooves and excrements. In particular, milk vetches and viper's bugloss herbs have been found as common parts of animal diets and are prevalent in dung layers as evident in other coeval sites (Trevisan Grandi et al., 1998). Plantago and Urtica-type come from trampled areas and places enriched by faeces/dung/coprolites by pastoralists and their animals (Giraudi et al., 2013).

of the vegetation cover rather than plant harvesting: this, however, is evident because anthers (i.e., flowers) of grasses were observed. The Asteroideae-daisy family (3%, with traces of Artemisia 0.1%) is relatively low, and the Chenopodiaceae-chenopods (0.7%) and Brassicaceae-cabbage family (0.3%) are insignificant. Tamarixtamarisk (0.1%) is also rare, while Acacia-acacia is only found overrepresented in the top sample (see below). Hygrophilous plants (12%) are mainly represented by Typha-cattails (10%). Though there are few samples, 3 pollen zones and 4 sub-zones may be distinguished from the bottom on the base of their stratigraphic distribution and archaeological content (Fig. 14).

5.2.2. The TK-WS pollen sequence Due to the intense humification, pollen is preserved in low quantities in this sector; thinned exines are common, and some pollen grains are folded or crumpled. Post-depositional disturbances, including hydratationedehydratation cycles confirmed by the geoarchaeological analyses, may have determined the deterioration of most pollen grains. Four samples are sterile (2, 6, 11, and 12), and concentrations are decidedly lower than in TK-NS (2300 p/g on average, with five samples <1000 p/g). In TK-WS, the floristic list is reduced by half to 50 pollen taxa, including 13 woody taxa (Fig. 14). Poaceae-grasses are dominant (58%) as in the TK-NS sequence, but a lower percentage of pollen >40 mm (1%) was found. With Cyperaceae-sedges (10%), they are most likely more representative

6. Discussion

5.2.2.1. Zone Tkw 1 (4 samples, from c. 65 to 35 cm; Unit III-WS; AC 109, Late Acacus). The zone has relatively high taxa diversity, including Chenopodiaceae (1%), Asteraceae (4%) and Typha (7%). The two bottom samples (Tkw 1a) show low pollen concentrations, higher amounts of Aster-type and less Poaceae than the other two samples (Tkw 1b). Xerophilous plants and a drier environment are especially represented at the bottom of the sequence. A mosaic of habitats coexisted at this phase, attesting to a trend from relatively drier to wetter environmental conditions. 5.2.2.2. Zone Tkw 2 (3 samples, from 30 to 12 cm; Unit II-WS; AC 89, Late Acacus 3). Chenopods are absent, and there are traces of Asteraceae (0.4%). A very high percentage of Typha (43%) with high pollen concentration was found in the bottom sample (Tkw 2a), which is evidence that this plant was transported to the site. Without this over-representation, the percentages of this sample fit those of other samples in the zone (Tkw 2b). Poaceae (69%) and Cyperaceae (15%) represent grassland as the main vegetation of this phase. 5.2.2.3. Zone Tkw 3 (1 sample, 5 cm; Unit I-WS; AC 83; Middle Pastoral 2). The top sample has a very low pollen concentration, and plants from wet environments are not represented. The spectrum is dominated by Acacia ehrenbergiana-type, suggesting the presence of dry savannah during the deposition of this layer. The low pollen content reflects both a low plant cover and the spread of low pollen-producing species that grew in the area during an arid climatic phase.

6.1. Natural and anthropic factors in formation processes Stratigraphic sections in the Takarkori fill differ mainly because of their location (e.g., under or outside the rock shelter vault) and subsequent exposure to rainfall. Water percolation through the deposit, though climatically limited since the Middle Holocene transition, primarily controlled the state of preservation of the organic components of the deposit (including humified groundmass, finely subdivided remains, and plant fragments) and drove their transformations. The organic coarse material was almost completely removed from the TK-WS deposits, and it only survived

Fig. 16. Comparison between the signals of regional proxies for Holocene climate changes in North Africa. The vertical grey bars indicate the climate anomaly at c. 8200 cal yr BP (according to Thomas et al., 2007) and the end of the AHP respectively. The chronology is expressed in cal yr BP. (A) Cultural phases at Takarkori (di Lernia and Tafuri, 2013; Biagetti and di Lernia, 2013); (B) pollen record from the Takarkori rock shelter: the solid curve indicates desert taxa (Asteraceae þ Chenopodiaceae), the dashed one hydro-hygrophilous taxa; (C) pollen record for the Wadi Teshuinat area (central Tadrart Acacus): the solid curve indicates desert communities and psammophilous, the dashed one taxa of wet environments (Mercuri, 2008b); (D) lake-level fluctuations in the central Sahara (Zerboni, 2006; Zerboni and Cremaschi, 2012); (E) calcareous tufa sedimentation in the Tadrart Acacus massif (Cremaschi et al., 2010); (F) width of tree rings of Cupressus dupreziana from the central Sahara (Cremaschi et al., 2006); (G) phases of high stand of Lake Gureinat in Sudan (Hoelzmann et al., 2010); (H) activity of Nubian lakes (Hoelzmann, 2002); (I) deposition of Sapropel S1 in the eastern Mediterranean (Ariztegui et al., 2000); (J) sum of tropical € pelin et al., 2008); (K) Lake Qarun level changes in the Fayum Depression (Hassan, 1986); (L) Lake Chad lake-level changes (Servant, 1983); (M) pollen types, lake Yoa Chad (Kro  level changes in terrigenous (Ti) input to Lake Tana (Marshall et al., 2011); (N) Bahr El-Ghazal depression lake-level changes (Servant and Servant-Vildary, 1980); (O) Lake Abhe eastern Africa (Gasse, 1977); (P) evolution of Lake Bosumtwi in Ghana (Shanahan et al., 2006); (Q) Sahara dust record off Mauritania (deMenocal et al., 2000); (R) the d18O record of  the Greenland NGRIP ice core (North Greenland Ice Core Project Members, 2004); (S) mean summer insolation at 20 N (Berger and Loutre, 1991).

54

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

in the lower Unit III-WS (2A and 2AC horizons) in the form of very small fragments of charcoal and bones. The redistribution of humified organic matter along the profile, determined by the isohumic process (Duchaufour, 1983), which required wet environmental conditions, began in the LA period according to the archaeological material present in the sequence. Pollen analysis indicates humid habitats and sparse shrubby grassland vegetation during this phase. Wet conditions are also suggested by the occurrence of fish bone fragments found at the sequence base, which are related to the lake and other ponds as indicated by geomorphological and archaeological survey of the region and that was active in the depression close to the Takarkori rock shelter at that time (Fig. 2). The colluvial features observed in the B21t horizon (Unit I-WS) required that vegetation cover provided weak protection for the soil surface and a water flux, even not regular throughout the year, on the topsoil. The removal of the organic matter in the B21t (Unit IWS) and B22t (Unit II-WS) horizons promoted clay translocation along the profile and lamellar argillic horizon formations. These processes are related to the desert pavement formation at the soil surface. This is not due to wind erosion and consequent deflation, but instead depends on continuously incorporating dust and the migration of the fine fraction inside the soil (Wells et al., 1995; Amit et al., 2011). This process predates the onset of the current hyperarid conditions in the area. The exposed parts of the stones included in the desert pavement are covered by black varnish, whose initial formation dates to c. 5500 cal yr BP (Cremaschi, 1996; Zerboni, 2008), by which time the pavement thus already existed. Its formation and the development of the underlying soil occurred due to the persistence of weak precipitation in a scenario of general decline of vegetation cover. The processes involved in the formation of the stratigraphic sequence inside the rock shelter (within the limit of the drip line) are rather different. The mineral fraction with coarse and medium sand is derived from the degradation of the rock shelter vault and wall and may be ascribed to slight thermoclastism and alternating dry and wet cycles. No significant aeolian input was recorded at any level of the sequence. Oxidizable carbon content versus loss in ignition indicates a continuous supply of organic matter throughout the section and a slow degradation rate. Due to the limited presence of water inside the rock shelter, the high amount of organic matter generated by the anthropogenic activities acted as a buffer for preserving plant remains and other features related to anthropogenic activities, including buried corpses, coprolites, food residues, ecofacts, and artefacts on perishable materials (Tafuri et al., 2006; di Lernia et al., 2012; Dunne et al., 2012; di Lernia and Tafuri, 2013), and even nucleic acids of plants (Olmi et al., 2011). Along the sequence, post-depositional bioturbation is rare, as confirmed by thin section analysis, indicating that the microenvironment of the deposit was barely suitable for invertebrate fauna activity. Evidence of bioturbation was only observed in sample AC 58, which was collected at the interface between the deposit and the sandstone wall where fissures and drip provided oxygenation and water availability, allowing for biological activity and solute precipitation (e.g., niter e KNO3). Human activities must therefore be considered as the dominant formative factor for the fill inside the rock shelter (with the supply of the coarse mineral fraction from the disaggregation of the cave roof and walls). It is likely that processes related to human activities played a main role in obscuring possible environmentecontrolled sedimentary and post-sedimentary processes. The colluvial features recorded in Units VI-NS and VII-NS, at the base of the sequence, constitute the sole exception to this conclusion, and they indicate high water availability in this period, which is consistent with pollen analyses. Similar evidence for colluvial

processes was also observed in the deposits at the base of Uan Afuda dating to 11,245 and 10,740 cal yr BP (Cremaschi and di Lernia, 1999b). Conversely, the Early Pastoral deposits at the Takarkori rock shelter are less humified than the nearly contemporaneous anthropogenic layers at the base of the sequence at the Uan Muhuggiah rock shelter (Cremaschi, 1998), underlying the role of stational factors in determining the characteristics of rock shelter fillings in the area. Furthermore, the increase of the coarse sand fraction at the top of Unit I fits well with pollen data, which indicates for the same unit dry environmental conditions. This should be attributed to a relative decrease in the organic fraction and not to an aeolian input, given the grain size distribution. 6.2. Palaeovegetational reconstruction Despite the strong anthropogenic influence on deposit formation, taphonomic features and the pollen record of the Takarkori rock shelter are highly informative about the local and regional palaeovegetational and palaeoenvironmental conditions (Fig. 16). The pollen zones, in fact, describe primary and varied habitat conditions. By considering the archaeological stratigraphy and radiocarbon dates, the zones may be attributed to subsequent chrono-cultural phases and vegetational events ranging from c. 10,000 to 4500 cal yr BP. The Tk-NS pollen sequence shows that in the LA phase, at c. 10,000e8100 cal yr BP, the plant cover was largely constituted by grassland and less so by sparsely wooded savannah vegetation. Permanent freshwater habitats with floating pondweeds, and reeds and cattails on the marginal zones, were common. A fairly diversified set of food and other useful plants were available for gatherers to harvest (Tk1a). A widening of shallow-water marginal zones or a general lowering of the water level was recognized during the LA2 phase (Tk1b), while the xerophilous plant presence began to expand. A further reduction of permanent water bodies is visible at the end of the LA3 phase (c. 8300e7900 cal yr BP; Tk1c), when environmental instability, seasonality and changes in plant exploitation are evident. At c. 8000e7000 cal yr BP (in the EP phase, Tk2), a significant increase in accumulation of both food and fodder/pasture plants is registered in pollen spectra. Although wet conditions were still present, increased seasonality made the environment more articulated or changeable than was the case previously. Wet habitats became smaller or were seasonally reduced, while Asteraceae and Chenopodiaceae spread. The selection of ‘new’ wild cereals used as fodder suggests that new resources were available and substituted the previous ones. Acacias and other xerophytes spread, matching a significant change toward dryness at c. 6900e5500 cal yr BP (MP phase, TK2c). A variety of habitats coexisted, while the establishment of very dry environmental conditions and desert savannah was evident at the time of the Middle Holocene transition (Tkw3). 6.3. Palaeoclimatic inferences and correlations 6.3.1. Palaeoclimate at Takarkori Most of the palaeoenvironmental information comes from the deposits preserved inside the rock shelter. Comparing the inner and the outer sequence it is clear that in the latter post-depositional processes strongly affected the state of the organic matter, which is well preserved in the inner part. Also, the concentration of pollen grains is substantial inside the rock shelter (TK-NS), while poor outside of the shelter vault (TK-WS). Besides the local palaeoclimatic significance, this evidence is further proof that pollen data from Saharan archaeological sites are more informative in palaeoclimatic studies when they are preserved within well protected sequences, while the interpretation of pollen spectra obtained from open air archaeological contexts in dry conditions may

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

be less informative or much more problematic (Horowitz, 1992; Mercuri, 2008b). In the inner part of the rock shelter, the significant mean value of plants from permanent freshwater habitats and the comparatively low values of chenopods and psammophilous shrubs clearly highlights that layers were deposited under environmental conditions marked by the occurrence of widespread freshwater ecosystems. In this area, these pre-dated the current Late Holocene hyperarid phase (e.g., Gasse, 2000; Mercuri, 2008b; Cremaschi and Zerboni, 2009; Watrin et al., 2009). Unit VII and part of Unit VI, corresponding to pollen zone TK-NS, includes the Tk1a, which lies upon the disaggregated and friable sandstone bedrock; the Units preserve evidence for the first human frequentation of the Takarkori rock shelter. During this phase, at around 10,000 cal yr BP, the LA hunteregatherers occupied an empty rock shelter. The micromorphological evidence for redistribution of sediments after runoff indicates a high availability of water. In fact, in the region at that time monsoonal precipitation recharged the surface aquifers in the Tadrart Acacus and surroundings (Cremaschi et al., 2010; Zerboni and Cremaschi, 2012). Sparsely wooded savannah vegetation with grassland habitats spread. In the first phases of occupation of the site, human groups lived near wet habitats, where pondweeds floated in the water surface of freshwater ponds or rivers, and cattails grew at the margins and in wet soils. Harvesting of plants for food, medicine and other purposes was centred not only on grasses, but also on a wide range of vegetal species including tamarisks, capers, and scent plants such as Artemisia and Mentha-type; plant remains were carried to and processed in the rock shelter. Units VI (upper part), V and IV (pollen zones TK-NS: Tk1b, Tk1c) were formed during the LA2 frequentation. Sediments dating to this period are slightly richer in organics, attesting to a more intense occupation of the site, while pollen spectra show a decrease of grassland, acacias and other tropical trees. The spread of cattails can be explained by a general lowering of lake levels or widening of shallow-water marginal habitats near the site, which is a confirmation of a general trend towards decreasing monsoonal water supply to the central Sahara recorded by carbon and oxygen stable isotopes of the calcareous tufa of the Tadrart Acacus (Cremaschi et al., 2010). There is evidence of increasing aridity at the end of this sub-phase, characterized by the local reduction of permanent water bodies and mixed vegetation cover, with a relevant increase of xerophilous plants to the detriment of grassland. Even though the Takarkori sequences do not display a decisive interruption in sedimentation and human occupation is rather continous, this phase might correspond to the environmental crisis related to the 8.2 ka BP event (Alley et al., 1997), which led to the desiccation of most of the springs in the Acacus (Cremaschi et al., 2010) and the shrinkage of the level of interdune lakes in the ergs around it (Zerboni and Cremaschi, 2012). A more detailed discussion of this topic is presented in the following subsection. The upper part of Unit IV, formed during the LA3 occupation of the site, corresponds to an increase of the content of organic matter content and can be correlated to the pollen zone TK-WS (sub zone TK 1c) on the basis of pollen content. Pollen indicates a drier environment compared to the lower stratigraphic Units and strong seasonal variations characterized by a variety of habitats with xerophilous plants alternating with grassland. LA3 foragers coped with this variability through a series of changes in intrasite organization (see Biagetti and di Lernia, 2013), and resource exploitation; for instance, new wild cereals, including millets with large pollen grains, were collected. Besides a strong reduction in water availability, residual freshwater bodies continued to provide food and raw materials, which is evident from Typha transported inside the rock shelter. The fairly rapid increase of Poaceae pollen in

55

spectra highlights the relevance of wild cereal exploitation: intensively harvested, possibly in the late summer, they were stored using both kettles and basketry containers (di Lernia et al., 2012). Moreover, seed and fruit accumulations are common at the site and, as reported by Olmi et al. (2011), sometimes they suggest special selection. Deposit properties and the pollen assemblage suggest an important change in the formation processes of the sequence throughout Units III, II and I (pollen zones 2 a, b, c), which accumulated during the EP and MP periods and attest to a new increase in water availability. These indicate a modification in the landscape surrounding the Takarkori rock shelter with the spread of permanent water bodies and the coexistence of different habitats. Xeric environments were still present representing the echoes of the previous arid spell. A cultural shift in plant exploitation is also evident in the pollen record: Poaceae include many large pollen grasses, which belong to the wild cereals harvested or browsed and brought into the site. During this phase, human groups contributed to the formation of the deposit with a larger input of plant organics within the rock shelter. The significant increase in fine sediments (clay þ silt) evident in this part of the sequence, in fact, can be hardly explained by climatic-driven processes, such as wind activity or in situ weathering of quartz grains. Instead, it was probably promoted by an anthropogenic contribution to sedimentation by means of the introduction of exotic grains during domestic activities. The upper part of Unit I, which was occupied during a second phase of the MP period (pollen zone TK-NS: Tk2c; correlated to TK-WS: Tkw3), records significant environmental instability in a shift towards drier climatic conditions. A general increase of the sandy fraction in the deposit, with sand grains displaying a moderate sorting in thin section, and a reduction of the organic content confirm this evidence. Moreover, the decrease in pollen concentrations is due to low input of pollen in sediments. This mirrors the spread of low pollen-producing plants such as acacias and other entomophilous species, and the decrease of highproducing plants such as grasses. Though some freshwater habitats were still present, increasing aridity pushed the expansion of the dry savannah. Inside the rock shelter the final transition to arid conditions is indicated by well preserved layers of ovicaprine dung and sandstone blocks collapsed from its vault. The latter acted as protection for the stratigraphic sequence by later enhanced wind erosion. The collapse of the vault (and the protection of cave sediments from erosion) is a widespread MiddleeLate Holocene phenomenon in the Tadrart Acacus region (Cremaschi, 1998; Cremaschi and Zerboni, 2011) and was related to slope instability under progressively more arid environmental conditions. Outside the rock shelter, where the surface was exposed to residual rainfall, a complex pedogenetic evolution of the sequence occurred. From the beginning of drier conditions, this included the formation of an argillic laminar horizon at the topsoil, the formation of a desert pavement, and the deposition (between c. 5500 and 4000 cal yr BP) of Mn-rich rock varnish on the outcropping part of rocks and stones lying on the topographic surface (Cremaschi, 1996; Zerboni, 2008).

6.3.2. Local and regional palaeoenvironmental correlations The palaeoenvironmental record inferred from the stratigraphic sequence at Takarkori reflects local modifications in environmental settings and changes in the availability of natural resources. Moreover, data from this site can be compared with other regional and continental archives for proxy data, confirming the global importance of anthropogenic sequences within rock shelters in arid lands. A general assessment of the palaeoclimatic significance of

56

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

this case study is given by correlating it with data from other Saharan and African localities (Fig. 16). Firstly, new evidence from Takarkori can be compared and validated with other palaeobotanical and palaeohydrological records available for the Tadrart Acacus and the Libyan central Sahara. Data from Takarkori illustrate that the first occupation of the site (LA phase, 10,170e8180 cal yr BP) was characterized by a continuous plant cover and possibly ample water availability; these observations well fit with palaeohydrological data from the Tadrart Acacus and the surrounding lowlands, which confirm a general recharge of surface aquifers after the increase in monsoonal rainfall supply. This event, consistently dated across the entire region of North Africa (e.g., Damnati, 2000; Gasse, 2000; Nicoll, 2004; Zerboni, 2013), occurred since the beginning of the Holocene. Radiocarbon chronology of interdune lake basins in the erg Uan Kasa (c. 50 km off the eastern fringe of the Tadrart Acacus) and edeyen of Murzuq (c. 200 km E of the Tadrart Acacus) indicates the outcrop of the water table at c. 10,450e9500 cal yr BP (Cremaschi, 1998, 2002; Zerboni, 2006; Cremaschi and Zerboni, 2009; Zerboni and Cremaschi, 2012). The activation of springs within the Tadrart Acacus massif is recorded in the same phase, indicated by the deposition of calcareous tufa since c. 9500 cal yr BP (Cremaschi et al., 2010), and the activation of the main river systems (Pachur, 1980; Perego et al., 2007; Cremaschi and Zerboni, 2011). A number of other proxy data confirm a general increase of water availability along the first two millennia of the Holocene (Cremaschi, 1998, 2002); moreover, during this phase the region is systematically exploited by hunteregatherers groups (di Lernia, 1999; Cancellieri and di Lernia, 2014) and pollen spectra from the Wadi Teshuinat region (central Tadrart Acacus) indicate a wide distribution of grasslands with Poaceae and Cyperaceae (Mercuri, 2008b; Fig. 16). At a continental scale, the increase in water availability in the Libyan central Sahara is in agreement with the onset of the AHP, controlled by the migration of the ITCZ, which penetrated about 500e800 km north of its present position (e.g., Petit-Maire et al., 1995; Gasse, 2000; Maley and Vernet, 2013), thanks to the expansion of the SW African Monsoon domain. It is widely accepted that the strengthening of the African Monsoon was primarily controlled by an insolation maximum (Berger and Loutre, 1991; RossignolStrick, 1999; Gasse, 2000; deMenocal et al., 2000; Garcin et al., 2007). In this phase the recharge of the North African deep aquifers began (Zuppi and Sacchi, 2004) and the Saharan drainage systems became active (Pachur, 1980; Williams and Adamson, 1980; Becker and Fürst, 1991), contributing to the discharge of freshwater into the Mediterranean Sea and allowing the deposition of the organicrich sapropel S1 (Ariztegui et al., 2000; Rohling et al., 2002). Moreover, the Sahara and Sahel saw a consistent phase of lake activities: increased monsoonal activity contributed to high stands of the main lake basins (e.g., Gasse, 1977, 2000; Servant and ServantVildary, 1980; Maley, 2004; Shanahan et al., 2006; Marshall et al., 2011), while a number of piezometric ponds came to light from Sudan to Mali (e.g., Pachur and Hoelzmann, 1991; Ritchie, 1994; €pelin, 2006; Hoelzmann et al., 2001, 2010; Kuper and Kro Williams, 2009; Hassan et al., 2012). According to pollen data and the degree of organic preservation, the subsequent phases of LA occupation of the rock shelter (LA2 and LA3) occurred under progressively more arid environmental conditions, which led to a marginalization of freshwater environments. In the Tadrart Acacus, for almost the same period, stable isotope (C and O) values from calcareous tufa suggest a substantial reduction of precipitation (Cremaschi et al., 2010), signalling a more consistent drought event, which in the region is well attested by a generalized shrinking of interdune lakes level and interruption of spring activity (Zerboni and Cremaschi, 2012). Also, the pollen record from the

Teshuinat area highlights a decrease in water availability thanks to a rapid expansion of desert taxa (Mercuri, 2008b). Palaeohydrological data indicate that the conditions suitable for the outcrop of the water table were substantially reduced around 8200 cal yr BP, largely coincident with the well-known event recorded at 8.2 ka BP (Alley et al., 1997; Barber et al., 1999; Mayewski et al., 2004; Kobashi et al., 2007; Thomas et al., 2007). The identification of the environmental effects of a rapid climate change, as in the Saharan region the 8.2 ka BP event, is still matter of discussion, as in most of the studied localities the palaeoclimatic signature of this phase is poorly or not preserved and many dating uncertainties still exist (Zerboni, 2013). But the attribution of the Early Holocene dry phase interrupting the AHP, leading to a shutdown of the monsoonal water supply is, for instance, attested by a palaeoclimatic model (Wiersma and Renssen, 2006) supported by independent field data. This confirms the strong influence of an abrupt drainage of ice-dammed lakes in North America on the circulation in the Atlantic Ocean; its main effect was a significant reduction in sea-surface temperature in the northern and tropical Atlantic, resulting in a temperature reduction in the Guinea Gulf, strong decrease in evaporation and decreased strength of the SW African monsoon. Furthermore, a general reduction in the intensity of the SW African monsoon in the northern Sahel and in the Sahara was observed in records from freshwater ecosystems at several localities. This dry interval was reported from sites including Lake Tigalmamine (Lamb et al., 1995), Sebkha Mellala (Gasse et al., 1990), Tin Ouaffadene depression (Gasse, 2000), Bahr El-Ghazal depression (Servant and Servant Vildary, 1980), Lake Bosumtwi (Talbot et al., 1984), Lake Abhe (Gasse, 1977), Lake Tanganyika, and Lake Malawi (Gasse, 2000). Some African summer monsoon-fed lakes also show a reduction in the intensity and penetration of the African Monsoon between c. 8500 and 7800 cal yr BP (Gasse and Van Campo, 1994; Gasse, 2000). The occurrence of a dry period during the AHP is evident also in the eastern African monsoon domain, where a high aerosol content of the Kilimanjaro ice core, associated with rapid fluctuations of lake levels (Thompson et al., 2002), and a major arid phase in the Tigray region are registered (Dramis et al., 2003). Apart from this interpretation of published data, the most significant identification of the 8.2 ka BP event in the Saharan region is from lake Gureinat, in western Nubia. This lacustrine sequence was found in a basin isolated from large-scale artesian systems and thus it preserves hydrological changes resulting from any modification in local rainfall (Hoelzmann et al., 2010). This long-term lake trend was interrupted by a shorter-lived regression event at 8.2/7.6 cal ka BP, which possibly led to complete lake desiccation. It is noteworthy that at the Takarkori rock shelter this climatic event is not recorded by any change in the sedimentary sequence, which is continuous and dominated by anthropogenic input. However, the climatic change is clearly seen in the pollen assemblage, which in this period indicates an increase of desert taxa and decreased environmental humidity. The effects of drought at around 8.2/7.6 cal ka BP were much more effective in the lowlands surrounding the massifs, where the interdune lakes dried out completely (Cremaschi and Zerboni, 2011; Zerboni and Cremaschi, 2012), than inside the mountain system, where residual water persisted in the phreatic network (Cremaschi et al., 2010). Water availability gave human communities the opportunity to overcome the drought by settling the highest reach of the Tadrart Acacus massif. This may explain why the anthropogenic contribution to the sedimentation of the Takarkori rock shelter was not interrupted. It must also be emphasized that, as discussed elsewhere (di Lernia, 2013), the 8.2 ka BP dry event seems also to correspond in the archaeological record of the Tadrart Acacus to a main transition from the last hunteregatherer activity to the first introduction of domestic cattle by Early Pastoral Neolithic groups.

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

In the Takarkori sequence, the second part of AHP (Pastoral phases, Units III, II, and I) is marked by a greater input of plant remains and the pollen assemblage reflects an expansion of freshwater conditions in the surroundings, with several patches of xeric environments. Finally, the upper part of Unit I indicates a climatic transition towards a more unsteady climate, with marked seasonality and aridity increasing progressively. The harshening phase culminates at the end of the occupation of the site, when the onset of desert environmental conditions led to the collapse of the roof of the rock shelter and the formation of rock varnish. Also in the Teshuinat region, the pollen record encompassing the Pastoral period suggests the restoration of wetter environmental conditions at its beginning, followed after the Middle Holocene by a progressive increase of taxa belonging to desert communities (Mercuri, 2008b). Between the dunes, the water table cropped out again but sedimentological and palaeontological data from lake sediments indicate an enhanced seasonality, which was characterized by monsoonal precipitations in summer and winter season marked by the drop (or at least the desiccation) of lake basins (Zerboni, 2006; Cremaschi and Zerboni, 2011; Zerboni and Cremaschi, 2012). These processes are a local expression, together with the tree-ring record of Cupressus dupreziana (Cremaschi et al., 2006), of the general environmental instability that occurred in the central Sahara since the Middle Holocene transition (Cremaschi, 1998; Mercuri, 2008a,b; Cremaschi and Zerboni, 2009, 2011; di Lernia et al., 2013). In many parts of North Africa the Middle Holocene is characterized by a change in the environment, announcing the gradual decline of monsoon rainfall tuned by the modifications of the orbital forcing (weakening of incoming insolation) and the termination of the AHP (e.g., Gasse, 2000; Mayewski et al., 2004; Nicoll, 2004). Even if the withdrawal of the monsoon systems was regulated by an orbital change in summer insolation, the steps toward aridity appear to have been shaped by local hydrological and geomorphological conditions, and at some locations, water availability was reduced more rapidly than elsewhere. Freshwater ecosystems responded to drought differently, mostly on the basis of € pelin et al., 2008; the size of their hydrological reservoir (Kro zine, 2009). Vegetation diminCremaschi and Zerboni, 2009; Le ished and the savannah environment was disappearing in the Middle Holocene, whereas desert species substituted more waterdependent plant taxa (Neumann, 1989). The onset of aridity is recorded in lakes from many localities between 6 and 5 cal ka BP, and isotopic data indicate a progressive decrease in the precipitation/evaporation balance, confirming a transition towards desert environmental conditions (Abell and Hoelzmann, 2000; Hoelzmann et al., 2001; Nicoll, 2004; Zerboni, 2013). 7. Conclusions The deposit of the Takarkori rock shelter may be regarded mainly as anthropogenic and testifies to an almost continuous human occupation since the onset of the AHP up to its termination, which corresponded with the Middle Holocene transition (Roberts et al., 2011a). While the pollen content highlights a reduction of water availability in the region at c. 8200 cal yr BP, archaeological, chronometric, and sedimentary characteristics of the sequence exclude any interruption in the human frequentation, even during the dry period. This is interpreted as being attributable to a possible persistence of water resources inside the central Saharan massifs, in contrast with surrounding lowlands that dried out completely. As a consequence these areas acted as a refuge for human groups, which in the same period modified their subsistence strategies, also exploiting domesticated animals. A wet Middle Holocene phase is attested both by sediments and pollen spectra, corresponding to an intensified use of the rock shelter and to an increased accumulation

57

therein of grass, including wild cereals. Climate deterioration at the end of the AHP period began to influence the vegetal cover, promoting the spread of dry species from the base of Unit I. At this period the site was occupied only seasonally by Late Pastoral groups, namely during the winter season. The sedimentary record of this period inside the rock shelter is affected by erosion and mostly preserved by collapsed blocks from the vault. However, the contemporary horizons of the western sequence outside the shelter point to the complexity of the processes conducive to the onset of dominant dry conditions in the area. The stratigraphic sequence of Takarkori is representative of the high potential for palaeoclimatic investigation of the central Saharan rock shelter deposits, which have preserved the organic matter accumulated in the EarlyeMid-Holocene due to a longlasting occupation. The potential of this kind of sediments as proxy data for environmental changes is not only local, but global, as it was influenced by both long and short-term fluctuations of Earth's climate. Acknowledgements The research has been carried out under the aegis of the ItalianLibyan Archaeological Mission in the Acacus and Messak, Sapienza University of Rome and Department of Archaeology, Tripoli, directed by Savino di Lernia. Main funds come from Sapienza
di Roma through Grandi Scavi di Ateneo and Italian Universita Ministry of Foreign Affairs (DGSP), from 2003 to 2011 entrusted to SDL, together with additional funds coming from Italian Ministry of University and Research, entrusted to MC, AMM, and SdL. All necessary permits were obtained for the field studies and laboratory analyses (including destructive processes) presented here. SdL designed the research and directed the fieldwork, providing the chronological and archaeological framework. MC and AZ performed geomorphological and micromophological analysis. AMM and LO performed pollen analysis. SB contributed to the analysis of the stratigraphy and archaeological. Our warmest thanks to the excavation staff: R. Castelli (also for taking part to the survey of the area), L. Cavorsi, E. Cancellieri, F. Del Fattore, T. Latini, M. Massussi, F. Merighi, A. Monaco, C. Pizzi, G. Poggi, F. Ricci, and M. Tarantini. We are indebted with M. Gallinaro for her advice, assistance and help. Many thanks to S. Giovannetti for her help in the field and lab. We wish to thank all the colleagues of the Libyan Department of Archaeology for their help and support: in Tripoli A. Khaddouri, G. Anag, S. Agab, M. Turjman, A. Jamali, B. Galgam. A special thought to the late E. Azzebi, co-director of the project before his sudden and premature death. Many thanks to N. Bergamaschi and E. Ferrari for helping with sedimentological analyses. E. Modrall is kindly acknowledged for the skilful revision of English language. We wish to thank the people in Ghat, in particular M. Denda and B. Baba; the several workers who helped during excavation; the people and staff at Dar Sahara Co., that provided logistics and camp: S. Scarpa, A. Ravenna, Dawd, Malik, Bilad, Ameneknek, and Hassan. We are grateful to CO.NI.COS. staff, in particular A. Becchio, A. Bottero and F. Castigliola for their support and hospitality in Tripoli and Tahala. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.quascirev.2014.07.004. References Abell, P.I., Hoelzmann, P., 2000. Holocene palaeoclimates in north-western Sudan: stable isotope studies on mollusks. Global Planet. Change 26, 1e12.

58

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

Alley, R.B., Mayewski, P.A., Sowers, T., Stuiver, M., Taylor, K.C., Clark, P.U., 1997. Holocene climatic instability: a prominent, widespread event 8200 years ago. Geology 25, 483e486. Amit, R., Simhai, O., Ayalon, A., Enzel, Y., Matmon, A., Crouvi, O., Porat, N., McDonald, E., 2011. Transition from arid to hyper-arid environment in the southern Levant deserts as recorded by early Pleistocene cummulic Aridisols. Quat. Sci. Rev. 30, 312e323. Anderson, D.G., Maasch, K.A., Sandweiss, D.H., 2007. Climate Change and Cultural Dynamics. Academic Press, London. roux, C., Bradtmiller, L., Mix, A., 2013. Arbuszewski, J.A., deMenocal, P.B., Cle Meridional shifts of the Atlantic intertropical convergence zone since the Last Glacial Maximum. Nat. Geosci. 6, 959e962. http://dx.doi.org/10.1038/ngeo1961. Ariztegui, D., Asioli, A., Lowe, J.J., Trincardi, F., Vigliotti, L., Tamburini, F., Chondrogianni, C., Accorsi, C.A., Bandini Mazzanti, M., Mercuri, A.M., Van der Kaars, S., McKenzie, J.A., Oldfield, F., 2000. Palaeoclimate and the formation of sapropel S1: inferences from Late Quaternary lacustrine and marine sequences in the central Mediterranean region. Palaeogeogr. Palaeoclimatol. Palaeoecol. 158, 215e240. Aumassip, G., 1984. Le site de T-in-Hanankaten et la neolithisationsur le marges orientales du Sahara central. Chaiers ORSTOM, Ser. Geol. 14, 201e203. Aumassip, G., Delibrias, G., 1982. Ages des depots neolithiques du gisement de Ti-nHanakaten (Tassili-n-Ajjer. Algelrie). Libyca XXXeXXXI, 207e211. Ayyad, S.M., Moore, P.D., 1995. Morphological studies of the pollen grains of the semi-arid region of Egypt. Flora 190, 115e133. Barber, D.C., Dyke, A., Hillaire-Marcel, C., Jennings, A.E., Andrews, J.T., Kerwin, M.W., Bilodeau, G., McNeely, R., Southon, J., Morehead, M.D., Gagnon, J.-M., 1999. Forcing of the cold event of 8,200 years ago by catastrophic drainage of Laurentide lakes. Nature 400, 344e348. Barich, B.E., 1987. Archaeology and Environment in the Libyan Sahara. The Excavations in the Tadrart Acacus, 1978e1983. In: British Archaeological Reports International Series, vol. 368. Oxford. Barich, B.E., Mori, F., 1970. Missione Paletnologica Italiana nel Sahara Libico. Risultati della campagna 1969. Origini 4, 79e142. Becker, R.E., Fürst, M., 1991. Sedimentological time markers and groundwater dating: a study of the Quaternary evolution of Al Kufrah area. In: Salem, M.J., Belaid, M.N. (Eds.), The Geology of Libya, vol. V. Elsevier, Amsterdam, pp. 2017e2026. Berger, A., Loutre, M.-F., 1991. Insolation values for the climate of the last 10 million years. Quat. Sci. Rev. 10, 297e317. Biagetti, S., Merighi, F., di Lernia, S., 2004. Decoding an Early Holocene Saharan stratified site: ceramic dispersion and site formation processes in the Takarkori rockshelter (Acacus Mountains, Libyan). J. Afr. Archeol. 2, 3e21. Biagetti, S., di Lernia, S., Poggi, G., 2009. Unearthing the hearths. Preliminary results on the Takarkori rockshelter fireplaces (Acacus Mts., Libya). In: Cavulli, F., ^ncio, M.I., Dias, M.I. (Eds.), Proceedings of the XV UISPP Conference, LisPrude bon, 4e9 September 2006. Archaeopress, Oxford, UK, pp. 23e29. Biagetti, S., di Lernia, S., 2007. Reflections on the Takarkori rockshelter (Fezzan, Libyan Sahara). In: Kornfeld, M., Vasilev, S., Miotti, L. (Eds.), Proceedings of the XV World Congress UISPP (Lisbon, 4e9 September 2006), British Archaeological Report S1655. Archaeopress, Oxford, pp. 125e132. Biagetti, S., di Lernia, S., 2013. Holocene fillings of Saharan rock shelters: the case of Takarkori and other sites from the Tadrart Acacus Mts. (SW Libya). Afr. Archaeol. Rev. 30, 305e338. Bolle, H.J., 2003. Mediterranean Climate. Variability and Trends. Springer, Berlin. Bonnefille, R., Riollet, G., 1980. Pollens des savanes d'Afrique orientale. CNRS, Paris. Bronk Ramsey, C., 2009. Dealing with outliers and offsets in radiocarbon dating. Radiocarbon 51, 1023e1045. Bullock, P., Fedoroff, N., Jongerius, A., Stoops, G., Tursina, T., Babel, U., 1985. Handbook for Soil Thin Section Description. Waine Research Publication, Albrighton, ST, USA. Butzer, K.W., 2005. Environmental history in the Mediterranean world: crossdisciplinary investigation of cause-and-effect for degradation and soil erosion. J. Archaeol. Sci. 32, 1773e1800. Cancellieri, E., di Lernia, S., 2014. Re-entering the central Sahara at the onset of the Holocene: a territorial approach to Early Acacus hunter-gatherers (SW Libya). Quat. Int. 320, 40e62. http://dx.doi.org/10.1016/j.quaint.2013.08.030. Cherkinsky, A., di Lernia, S., 2013. Bayesian approach to 14C dates in the estimation of long-term archaeological sequences in arid environment: the Holocene site of Takarkori rock shelter, Southwest Libya. Radiocarbon 55, 771e782. Coombes, P., Barber, K., 2005. Environmental determinism in Holocene research: causality or coincidence. Area 37, 303e311. Corti, R., 1942. Flora e vegetazione del Fezzan e della Regione di Gat. Tipografia Editrice Mariano Ricci, Firenze, Italy. Courty, M.A., 2001. Microfacies analysis assisting archaeological stratigraphy. In: Goldberg, P., Holliday, V.T., Ferring, C.R. (Eds.), Earth Sciences and Archaeology. Kluver Academic/Plenum Publishers, New York, pp. 205e239. Courty, M.-A., Goldberg, P., Macphail, R., 1989. Soil and Micromorphology in Archaeology. Cambridge University Press, Cambridge. Cremaschi, M., 1996. The rock varnish in the Messak Settafet (Fezzan, Libyan Sahara), age, archaeological context, and paleo-environmental implication. Geoarchaeol. Int. J. 11, 393e421. Cremaschi, M., 1998. Late Quaternary geological evidence for environmental changes in south-western Fezzan. In: Cremaschi, M., di Lernia, S. (Eds.), Wadi Teshuinat e Palaeoenvironment and Prehistory in South-western Fezzan (Libyan Sahara), Quaderni di Geodinamica Alpina e Quaternaria, vol. 7. C.N.R., Roma-Milano, Italy, pp. 13e48.

Cremaschi, M., 2002. Late Pleistocene and Holocene climatic changes in the central Sahara: the case study of the southwestern Fezzan Libya. In: Hassan, F.A. (Ed.), Droughts. Food and Culture. Kluwer Academic, New York, pp. 65e82. Cremaschi, M., di Lernia, S., 1998. The geoarchaeological survey in the central Tadrart Acacus and surroundings (Libyan Sahara). Environment and cultures. In: Wadi Teshuinat. Palaeoenvironment and Prehistory in South-Western Fezzan (Libyan Sahara). Quaderni di Geodinamica Alpina e Quaternaria, vol. 7. C.N.R., Roma-Milano, Italy, pp. 243e325. Cremaschi, M., di Lernia, S., 1999a. Holocene climatic changes and cultural dynamics in the Libyan Sahara. Afr. Archaeol. Rev. 16, 211e238. Cremaschi, M., di Lernia, S., 1999b. The 1993 and 1994 excavations. Geomorphology, stratigraphic context and dates. In: di Lernia, S. (Ed.), The Uan Afuda Cave (Tadrart Acacus, Libyan Sahara). Archaeological Investigations of Late Pleistocene and Early Holocene Human Occupations, Arid Zone Archaeology, vol. 1. Edizioni All'Insegna del Giglio, Firenze, Italy, pp. 9e26. Cremaschi, M., Pelfini, M., Santilli, M., 2006. Cupressus dupreziana: a dendroclimatic record for Middle-Late Holocene in the central Sahara. Holocene 16, 293e303. Cremaschi, M., Trombino, L., 1999. Forming processes in the Uan Afuda sequence. Palaeoenvironment and human activities: a micromorphological approach. In: di Lernia, S. (Ed.), The Uan Afuda Cave (Tadrart Acacus, Libyan Sahara). Archaeological Investigations of Late Pleistocene and Early Holocene Human Occupations, Arid Zone Archaeology, vol. 1. Edizioni All'Insegna del Giglio, Firenze, Italy, pp. 27e38. Cremaschi, M., Zerboni, A., 2009. Early to Middle Holocene landscape exploitation in a drying environment: two case studies compared from the central Sahara (SW Fezzan, Libya). C. R. Geosci. 341, 689e702. Cremaschi, M., Zerboni, A., 2011. Human communities in a drying landscape. Holocene climate change and cultural response in the central Sahara. In: Martini, I.P., Chesworth, W. (Eds.), Landscape and Societies: Selected Cases. Springer Science, Dordrecht, pp. 67e89. €tl, C., Felletti, F., 2010. The calcareous tufa in the Cremaschi, M., Zerboni, A., Spo Tadrart Acacus Mt. (SW Fezzan, Libya). An early Holocene palaeoclimate archive in the central Sahara. Palaeogeogr. Palaeoclimatol. Palaeoecol. 287, 81e94. Cullen, H.M., deMenocal, P.B., Hemming, S., Hemming, G., Brown, F.H., Guilderson, T., Sirocko, F., 2000. Climate change and the collapse of the Akkadian empire: evidence from the deep sea. Geology 28, 379e382. Damnati, B., 2000. Holocene lake records in the Northern Hemisphere of Africa. J. Afr. Earth Sci. 31, 253e262. deMenocal, P., Ortiz, J., Guilderson, T., Adkins, J., Sarnthein, M., Baker, L., Yarusinsky, M., 2000. Abrupt onset and termination of the African Humid Period: rapid climate responses to gradual insolation forcing. Quat. Sci. Rev. 19, 347e361.
Geografica Italiana (Ed.), Il Desio, A., 1937. Geologia e Morfologia. In: Regia Societa
Italiana Arti Grafiche, Sahara Italiano. Parte I: Fezzan e Oasi di Ghat. Societa Roma, Italy, pp. 39e94. di Lernia, S., 1999. The Uan Afuda Cave: Hunter-Gatherers Societies of Central Sahara. In: Arid Zone Archaeology Monographs. All'Insegna del Giglio, Firenze, Italy. di Lernia, S., 2001. Dismantling dung: delayed use of food resources among Early Holocene foragers of the Libyan Sahara. J. Anthropol. Archaeol. 20, 408e441. di Lernia, S., 2002. Dry climatic events and cultural trajectories: adjusting Middle Holocene pastoral economy of the Libyan Sahara. In: Hassan, F.A. (Ed.), Droughts, Food and Culture. Kluver Academic/Plenum Publishers, New York, pp. 225e250. di Lernia, S., 2013. The emergence and spread of herding in Northern Africa: a critical reappraisal. In: Mitchell, P.J., Lane, P.J. (Eds.), Oxford Handbook of African Archaeology. Oxford University Press, Oxford, pp. 527e540. di Lernia, S., Massamba N'siala, I., Mercuri, A.M., 2012. Saharan prehistoric basketry. Archaeological and archaeobotanical analysis of the Early-Middle Holocene assemblage from Takarkori (Acacus Mts., SW Libya). J. Archaeol. Sci. 39, 1837e1853. di Lernia, S., Tafuri, M.A., Gallinaro, M., Alhaique, F., Balasse, M., Cavorsi, L., Fullagar, P., Mercuri, A.M., Monaco, A., Perego, A., Zerboni, A., 2013. Inside the ‘African cattle complex’: animal burials in the Holocene Central Sahara. PLoS ONE 8 (2), e56879. di Lernia, S., Zampetti, D., 2008. La Memoria dell'Arte. Le Pitture Rupestri dell'Acacus tra Passato e Futuro. Edizioni All'Insegna del Giglio, Firenze, Italy. di Lernia, S., Tafuri, M.A., 2013. Persistent deathplaces and mobile landmarks. The Holocene mortuary and isotopic record from Wadi Takarkori (SW Libya). J. Anthropol. Archaeol. 32, 1e15. Dimbleby, G.W., 1985. The Palynology of Archaeological Sites. Academic Press, London. Dramis, F., Umer, M., Calderoni, G., Haile, M., 2003. Holocene climate phase from buried soils in Tigray (northern Ethiopia): comparison with lake level fluctuations in the Main Ethiopian Rift. Quat. Res. 60, 274e283. Duchaufour, P., 1983. Pedologie: 1. Pedogenese et classification. Masson, Paris. Dunne, J., Evershed, R.P., Salque, M., Cramp, L., Bruni, S., Ryan, K., Biagetti, S., di Lernia, S., 2012. First dairying in ‘green’ Saharan Africa in the 5th millennium BC. Nature 486, 390e394. El-ghali, M.A.K., 2005. Depositional environments and sequence stratigraphy of paralic glacial, paraglacial and postglacial Upper Ordovician siliciclastic deposits in the Murzuq basin, SW Libya. Sediment. Geol. 177, 145e173. El-Moslimany, A.P., 1990. Ecological significance of common non-arboreal pollen: examples from drylands of the Middle East. Rev. Palaeobot. Palynol. 64, 343e350.

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60 El-Tantawi, A.M.M., 2005. Climate Change in Libya and Desertification of Jifara Plain Using Geographical Information System and Remote Sensing Techniques. Johannes Gutenberg-Universit€ at, Mainz. Fægri, K., Iversen, J., 1989. In: Fægri, K., Kaland, P.E., Krzywinski, K. (Eds.), Textbook of Pollen Analysis, fourth ed. John Wiley & Sons, Chichester, UK. Faegri, K., Kaland, P.E., Krzywinski, K., 1989. Textbook of Pollen Analysis, fourth ed. Wyley & Son, London. Florenzano, A., Mercuri, A.M., Pederzoli, A., Torri, P., Bosi, G., Olmi, L., Rinaldi, R., Bandini Mazzanti, M., 2012. The significance of intestinal parasite remains in pollen samples from Mediaeval pits in the Piazza Garibaldi of Parma, Emilia Romagna, Northern Italy. Geoarchaeol. Int. J. 27, 34e47. Fowell, S.J., Hansen, B.C.S., Peck, J.A., Khosbayar, P., Ganbolde, E., 2003. Mid to late Holocene climate evolution of the Lake Telmen Basin, North Central Mongolia, based on palynological data. Quat. Res. 59, 353e363. Gale, S.J., Hoare, P.G., 1991. Quaternary Sediments. Belhaven Press, New York. Garcea, E., 2001. Uan Tabu: In the Settlement History of the Libyan Sahara. In: Arid Zone Archaeology Monographs, vol. 2. Edizioni All'Insegna del Giglio, Firenze, Italy. Garcin, Y., Vincens, A., Williamson, D., Buchet, G., Guiot, J., 2007. Abrupt resumption of the African Monsoon at the Younger DryaseHolocene climatic transition. Quat. Sci. Rev. 26, 690e704. . Nature 256, 42e45. Gasse, F., 1977. Evolution of lake Abhe Gasse, F., 2000. Hydrological changes in the African tropics since the Last Glacial Maximum. Quat. Sci. Rev. 19, 189e211. Gasse, F., Van Campo, E., 1994. Abrupt post-glacial climate events in West Asia and North Africa monsoon domains. Earth Planet. Sci. Lett. 126, 435e456. Gasse, F., Tehet, R., Durand, A., Gibert, F., Fontes, J.C., 1990. The aridehumid transition in the Sahara and the Sahel during the last deglaciation. Nature 346, 141e156. Gillson, L., 2006. A ‘large infrequent disturbance’ in an East African savanna. Afr. J. Ecol. 44, 458e467. Giraudi, C., Mercuri, A.M., Esu, D., 2013. Holocene palaeoclimate in the northern Sahara margin (Jefara Plain, Northwestern Libya). Holocene 23, 339e352. Goldberg, P., Berna, F., 2010. Micromorphology and context. Quat. Int. 214, 56e62. Goldberg, P., Macphail, R.I., 2006. Practical and Theoretical Geoarchaeology. Blackwell Publishing, Oxford. Grimm, E.C., 1991e1993. Tilia v. 2.0. Illinois State Museum, Research and Collections Center, Springfield, Illinois. Hachi, S., 1983. Place du gisment de T-in-Hanakaten (Tassili-N-Ajjer, Algerie) dans le contexte climatique Saharien. In: Paleoecologie des regions sahariennes. Actes du colloquie international, Beni-Abbes, pp. 115e121. Hall, S.A., 1981. Deteriorated pollen grains and the interpretation of Quaternary pollen diagrams. Rev. Palaeobot. Palynol. 32, 193e206. Hassan, F.A., 1986. Holocene lakes and settlements of the Western Faiyum. J. Archaeol. Sci. 13, 483e501. Hassan, F.A., Hamdan, M.A., Flower, R.J., Keatings, K., 2012. Oxygen and carbon isotopic records in Holocene freshwater mollusc shells from the Faiyum palaeolakes, Egypt: palaeoenvironmental and palaeoclimatic implications. Quat. Int. 266, 175e187. Heiri, O., Lotter, A.F., Lemcke, G., 2001. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. J. Paleolimnol. 25, 101e110. Herzschuh, U., 2007. Reliability of pollen ratios for environmental reconstructions on the Tibetan Plateau. J. Biogeogr. 34, 1265e1273. Herzschuh, U., Tarasov, P., Wünnemann, B., Hartmann, K., 2004. Holocene vegetation and climate of the Alashan Plateau, NW China, reconstructed from pollen data. Palaeogeogr. Palaeoclimatol. Palaeoecol 211, 1e17. Hoelzmann, P., 2002. Lacustrine sediments as indicators of climate change during the Late Quaternary in western Nubia (eastern Sahara). In: Lenssen-Erz, T., € pelin, S. (Eds.), Tides of the Desert d Contributions to the Tegtmeier, U., Kro Archaeology and Environmental History of Africa in Honour of Rudolph Kuper. Heinrich Barth Institut, Cologne, pp. 375e398. € pelin, S., Kruse, H.-J., 2001. Environmental Hoelzmann, P., Keding, B., Berke, H., Kro change and archaeology: lake evolution and human occupation in the Eastern Sahara during the Holocene. Palaeogeogr. Palaeoclimatol. Palaeoecol. 169, 193e217. Hoelzmann, P., Gasse, F., Dupont, L.M., Salzmann, U., Staubasser, M., Leuschner, D.C., Sirocko, F., 2004. Palaeoenvironmental changes in the arid and subarid belt (SaharaeSaheleArabian Peninsula) from 150 kyr to present. In: Battarbee, R.W., Gasse, F., Stickley, C.E. (Eds.), Past Climate Variability Through Europe and Africa, Developments in Palaeoenvironmental Research Series, vol. 6. Springer, pp. 219e256. Hoelzmann, P., Schwalb, A., Roberts, N., Cooper, P., Burgess, A., 2010. Hydrological response of an easteSaharan palaeolake (NW Sudan) to earlyeHolocene climate. Holocene 20, 537e549. Hooghiemstra, H., 1996. Aspects of Neogene-Quaternary environmental and climatic change in equatorial and Saharan Africa. Palaeoecol. Afr. 24, 115e132. Horowitz, A., 1992. Palynology of Arid Lands. Elsevier, Amsterdam. Kobashi, T., Severinghaus, J.F., Brook, E.J., Barnola, J.-M., Grachev, A.M., 2007. Precise timing and characterization of abrupt climate change 8200 years ago from air trapped in polar ice. Quat. Sci. Rev. 26, 1212e1222. €pelin, S., Verschuren, D., Le zine, A.-M., Eggermont, H., Cocquyt, C., Francus, P., Kro Cazet, J.P., Fagot, M., Rumes, B., Russell, J.M., Darius, F., Conley, D.J., Schuster, M., von Suchodoletz, H., Engstrom, D.R., 2008. Climate-driven ecosystem succession in the Sahara: the past 6000 years. Science 320, 765e768.

59

€ pelin, S., 2006. Climate-controlled Holocene occupation in the Sahara: Kuper, R., Kro motor of Africa's evolution. Science 313, 803e807. Lamb, H.F., Gasse, F., Benkaddour, A., El Hamouti, N., van der Kaars, S., Perkins, W.T., Pearce, N.J., Roberts, C.N., 1995. Relation between century-scale Holocene arid intervals in tropical and temperate zones. Nature 373, 134e137. zine, A.-M., 1989. Late Quaternary vegetation and climate of the Sahel. Quat. Res. Le 32, 317e334. zine, A.-M., 2009. Climatic history of the African and Arabian deserts. C. R. Geosci. Le 341, 569e574. zine, A.-M., He ly, C., Grenier, C., Braconnot, P., Krinner, G., 2011. Sahara and Sahel Le vulnerability to climate changes, lessons from Holocene hydrological data. Quat. Sci. Rev., 3001e3012. Li, Q., Lu, H., Zhu, L., Wu, N., Wang, J., Lu, X., 2011. Pollen-inferred climate changes and vertical shifts of alpine vegetation belts on the northern slope of the Nyainqentanglha Mountains (central Tibetan Plateau) since 8.4 kyr BP. Holocene 21, 939e950. Linseele, V., Marinova, E., Van Neer, W., Vermeersch, P.M., 2010. Sites with Holocene dung deposits in the Eastern Desert of Egypt: visited by herders? J. Arid Environ. 74, 818e828. Maley, J., 1980. Les changements climatiques de la fin du Tertiaire en Afrique: leur quence sur l'apparition du Sahara et de sa ve ge tation. In: Williams, M.A.J., conse Faure, H. (Eds.), The Sahara and the Nile: Quaternary Environments and Prehistoric Occupation in Northern Africa. Balkema, Rotterdam, pp. 63e86. cent: formations Maley, J., 2004. Le bassin du lac Tchad au Quaternaire re dimentaires, pale oenvironnements et pre histoire. In: Renault-Miskovsky, J., se otchads. L'Evolution de la Ve g e tation Semah, A.M. (Eds.), La question des Pale es. Publication Errance, Paris, pp. 179e217. depuis Deux Millions d'Anne Maley, J., Vernet, R., 2013. Peuples et evolution climatique en climatique en Afrique nord-ropicale, de la fin du Neolithique a l'aube de l'epoque moderne. Afriques 4. http://dx.doi.org/10.4000/afriques.1209. Marshall, M.H., Lamb, H.F., Huws, D., Davies, S.J., Bates, R., Bloemendal, J., Boyle, J., Leng, M.J., Umer, M., Bryant, C., 2011. Late Pleistocene and Holocene drought events at Lake Tana, the source of the Blue Nile. Global Planet. Change 78, 147e161. Matthews, W., French, C.A.I., Lawrence, T., Cutler, D.F., Jones, M.K., 1997. Microstratigraphic traces of site formation processes and human activities. World Archaeol. 29, 281e308. Mayewski, P.A., Rohling, E.E., Stager, J.C., Karlen, W.K., Maasch, K.A., Meeker, L.D., Meyerson, E.A., Gasse, F., Van Kreveld, S., Holmgren, K., Lee-Thorp, J., Rosqvist, G., Rack, F., Staubwasser, M., Schneider, R.R., Steig, E.J., 2004. Holocene climate variability. Quat. Res. 62, 243e255. Mercuri, A.M., 1999. Palynological analysis of the Early Holocene sequence. In: di Lernia, S. (Ed.), The Uan Afuda Cave, Arid Zone Archaeology Monographs, vol. 1. Edizioni All'Insegna del Giglio, Firenze, pp. 149e181. Mercuri, A.M., 2001. Preliminary analyses of fruits, seeds and few plant macrofossils from the Early Holocene sequence. In: Garcea, E.A.A. (Ed.), Uan Tabu in the Settlement History of the Libyan Sahara. Edizioni All’Insegna del Giglio, Firenze, Italy, pp. 189e210. Mercuri, A.M., 2008a. Plant exploitation and ethnopalynological evidence from the Wadi Teshuinat area (Tadrart Acacus, Libyan Sahara). J. Archaeol. Sci. 35, 1619e1642. Mercuri, A.M., 2008b. Human influence, plant landscape evolution and climate inferences from the archaeobotanical records of the Wadi Teshuinat area (Libyan Sahara). J. Arid Environ. 72, 1950e1967. Mercuri, A.M., Bandini Mazzanti, M., Torri, P., Vigliotti, L., Bosi, G., Florenzano, A., Olmi, L., Massamba N'siala, I., 2012. A marine/terrestrial integration for mid-late Holocene vegetation history and the development of the cultural landscape in the Po valley as a result of human impact and climate change. Veg. Hist. Archaeobot. 21, 353e372. Mercuri, A.M., Sadori, L., Uzquiamo Ollero, P., 2011. Mediterranean and northAfrican cultural adaptations to mid-Holocene environmental and climatic changes. Holocene 21, 189e206. Mercuri, A.M., Trevisan Grandi, G., Mariotti Lippi, M., Cremaschi, M., 1998. New pollen data from the Uan Muhuggiag rockshelter (Libyan Sahara, VIIeIV millennia BP). In: Cremaschi, M., di Lernia, S. (Eds.), Wadi Teshuinat e Palaeoenvironment and Prehistory in South-western Fezzan (Libyan Sahara), Quaderni di Geodinamica Alpina e Quaternaria, vol. 7. C.N.R., Roma-Milano, Italy, pp. 107e122. Messerli, B., Grosjean, N., Hoefer, T., Pfister, C., 2000. From nature-dominated to human-dominated environmental changes. Quat. Sci. Rev. 19, 459e479. Mori, F., 1965. Tadrart Acacus. Arte rupestre e culture del Sahara preistorico. Einaudi, Torino, Italy. Murphy, C.P., 1986. Thin Section Preparation of Soils and Sediments. AB Academic Publishers, Berkhamsted, Herts. Neumann, K., 1989. Holocene vegetation of Eastern Sahara: charcoal from prehistoric sites. Afr. Archaeol. Rev. 7, 97e116. Nicoll, K., 2004. Recent environmental change and prehistoric human activity in Egypt and Northern Sudan. Quat. Sci. Rev. 23, 561e580. North Greenland Ice Core Project Members, 2004. High-resolution record of Northern Hemisphere climate extending into the last interglacial period. Nature 431, 147e151. Olmi, L., Mercuri, A.M., Gilbert, M.T.P., Biagetti, S., Fordyce, S., Cappellini, E., Massamba N'siala, I., di Lernia, S., 2011. Morphological and geneticanalyses of earlymid Holocene wild cereals from the Takarkori rockshelter (central Sahara, Libya): first results and prospects. In: Fahmy, A.G., Kahlheber, S., D'Andrea, A.C.

60

M. Cremaschi et al. / Quaternary Science Reviews 101 (2014) 36e60

(Eds.), Windows on the African Past: Contemporary Approaches to African Archaeobotany. Africa Magna Verlag, Frankfurt, pp. 175e184. ge tation du Sahara. CNRS, Paris. Ozenda, P., 2000. Flore et ve Pachur, H.-J., 1980. Climatic history in the Late Quaternary in southern Libya and western Libyan Desert. In: Salem, M.J., Busrewil, M.T. (Eds.), The Geology of Libya, vol. III. Academic Press, London, pp. 781e788. Pachur, H.-J., Hoelzmann, P., 1991. Palaeoclimatic implications of Late Quaternary lacustrine sediments in Western Nubia. Quat. Res. 36, 257e276. Pasa, A., Pasa Durante, M.V., 1962. Analisi paleoclimatiche nel deposito di Uan Muhuggiag, nel massiccio dell'Acacus (Fezzan meridionale). Mem. Mus. Civ. St. Nat. Verona 10, 251e255. Perego, A., Cremaschi, M., Zerboni, A., 2007. Il telerilevamento nella ricostruzione della paleo idrografia olocenica in zone aride. Il caso di studio di Wadi Tanezzuft, Libia SO. Rend. Soc. Geol. Ital. 4, 97e99. Petit-Maire, N., Arnold, X., Aucour, A.M., Carbonel, P., Delibrias, G., Erlenkeuser, H.,
-Marsche, I., Thinon, M., 1995. Holocene Lakes Fabre, M., Goetz, M., Riser, J., Soulie in Northern Mali (23 N). Abstract INQUA Meeting, August 1995, Berlin, 216. Reille, M., 1992. Pollen et spores d'Europe et d'Afrique du Nord. URA CNRS, Marseille. Reille, M., 1995. Pollen et spores d'Europe et d'Afrique du Nord (Suppl. 1). URA CNRS, Marseille. Reille, M., 1998. Pollen et spores d'Europe et d'Afrique du Nord (Suppl. 2). URA CNRS, Marseille. Ritchie, J.C., 1994. Holocene pollen spectra from Oyo, northwestern Sudan: problems of interpretation in a hyperarid environment. Holocene 4, 9e15. Rohling, E.J., Casford, J., Abu-Zied, R., Cooke, S., Mercone, D., Thomson, J., Croudace, I., Jorissen, F.J., Brinkhuis, H., Kallmeyer, J., Wefer, G., 2002. Rapid Holocene climate change in the Eastern Mediterranean. In: Hassan, F.A. (Ed.), Droughts, Food and Culture. Kluwer Academic/Plenum Publishers, New York, pp. 35e46. Roberts, N., Jones, M.D., Benkaddur, A., Eastwood, W.J., Filippi, M.L., Frogley, M.R., Lamb, H.F., Leng, M.J., Reed, J.M., Stein, M., Stevens, L., Valero- Garce, B., Zanchetta, G., 2008. Stable isotope records of Late Quaternary climate and hydrology from Mediterranean lakes: the ISOMED synthesis. Quat. Sci. Rev. 27, 2426e2441. Roberts, N., Brayshaw, D., Kuzucuoglu, C., Perez, R., Sadori, L., 2011a. The midHolocene climatic transition in the Mediterranean: causes and consequences. Holocene 21, 3e13. Roberts, N., Eastwood, W.J., Kuzucuoglu, C., Fiorentino, G., Caracuta, V., 2011b. Climatic, vegetation and cultural change in the eastern Mediterranean during the mid-Holocene environmental transition. Holocene 21, 147e162. Rossignol-Strick, M., 1999. The Holocene climatic optimum and pollen records of sapropel in the eastern Mediterranean, 9000e6000 BP. Quat. Sci. Rev. 18, 515e530. Ruddiman, W.F., 2003. The anthropogenic greenhouse era began thousands of years ago. Clim. Change 61, 261e293. Ruddiman, W.F., 2007. The early anthropogenic hypothesis: challenges and responses. Rev. Geophys. 45, 2006RG000207. Schmidt, M., 2003. Development of a Fuzzy Expert System for Detailed Land Cover Mapping in the Dra Catchment (Morocco) Using High Resolution Satellite Images (PhD. thesis). University of Bonn. Schulz, E., 1987. Holocene vegetation in the Tadrart Acacus: the pollen record of two early ceramic sites. In: Barich, B.E. (Ed.), Archaeology and Environment in the Libyan Sahara. The Excavations in the Tadrart Acacus, 1978e1983. British Archaeological Reports, Oxford, pp. 313e326.  quences Continentals et Variations Climatiques: Evolution Servant, M., 1983. Se du Basin du Tchad au Cenozoique Superieur, Travaux et Documents, 159. ORSTOM, Paris. Servant, M., Servant-Vildary, S., 1980. L'environnement quaternaire du bassin du Tchad. In: Williams, M.A.J., Faure, H. (Eds.), The Sahara and the Nile. A.A. Balkema, Rotterdam, pp. 133e163. Shanahan, T.M., Overpeck, J.T., Wheeler, C.W., Beck, J.W., Pigati, J.S., Talbot, M.R., Scholz, C.A., Peck, J., King, J.W., 2006. Paleoclimatic variations in West Africa from a record of late Pleistocene and Holocene lake level stands of Lake Bosumtwi, Ghana. Palaeogeogr. Palaeoclimatol. Palaeoecol. 242, 287e302. Stoops, G., 2003. Guidelines for Analysis and Description of Soil and Regolith Thin Sections. Soil Science Society of America, Madison, WI. Stoops, G., Marcellino, V., Mees, F., 2010. Interpretation of Micromorphological Features of Soil and Regoliths. Elsevier, Amsterdam. Tafuri, M.A., Bentley, R.A., Manzi, G., di Lernia, S., 2006. Mobility and kinship in the prehistoric Sahara: strontium isotope analysis of Holocene human skeletons from the Acacus Mts (southwestern Libya). J. Anthropol. Archaeol. 25, 390e402.

Talbot, M.R., Livingstone, D.A., Palmer, D.G., Maley, J., Melack, J.M., Delibrias, G., Gulliksen, J., 1984. Preliminary results from sediments core from lake Bosumti, Ghana. Palaeoecol. Afr. 16, 173e192. Thomas, E.R., Wolff, E.W., Mulvaney, R., Steffensen, J.P., Johnsen, S.J., Arrowsmith, C., White, J.W.C., Vaughn, B., Popp, T., 2007. The 8.2 ka event from Greenland ice cores. Quat. Sci. Rev. 26, 70e81. Thompson, L.G., Mosley-Thompson, E., Davis, M.E., Henderson, K.A., Brecher, H.H., Zagorodnov, V.S., Mashiotta, T.A., Lin, P.-N., Mikhalenko, V.N., Hardy, D.R., Beer, J., 2002. Kilimanjaro ice core records: evidence of Holocene climate change in tropical Africa. Science 298, 589e593. Trevisan Grandi, G., Mariotti Lippi, M., Mercuri, A.M., 1998. Pollen in dung layers from rockshelters and caves of Wadi Teshuinat (Libyan Sahara). In: Cremaschi, M., di Lernia, S. (Eds.), Wadi Teshuinat e Palaeoenvironment and Prehistory in South-western Fezzan (Libyan Sahara), Quaderni di Geodinamica Alpina e Quaternaria, vol. 7. C.N.R, Roma-Milano, Italy, pp. 95e106. Turril, W.B., Milne-Redhead, E., 1952. Flora of Tropical East Africa. Crown Agents, London. Walkley, A., Black, I.A., 1934. An examination of Degtjareff method for determining soil organic matter and a proposed modification of the chromic acid titration method. Soil Sci. 37, 29e38. Walther, H., Lieth, H., 1960. Klimadiagramm-Weltatlas. G. Fischer Verlag, Jena. Wasylikowa, K., 1992. Holocene flora of the Tadrart Acacus area, SW Libya, based on plants macrofossils from Uan Muhuggiag and Ti-n-Torha, two caves archaeological sites. Origini XVI, 125e159. zine, A.-M., He ly, C., 2009. Plant migration and plant communities at Watrin, J., Le the time of the ‘green Sahara’. C. R. Geosci. 341, 656e670. Weiner, S., 2010. Microarchaeology. Beyond the Visible Archaeological Record. Cambridge University Press, Cambridge, UK. Wells, S.G., McFadden, L.D., Poths, J., Olinger, C.T., 1995. Cosmogenic 3He surface exposure dating of stone pavements: implications for landscape evolution in deserts. Geology 23, 613e616. Wendorf, F., Karlen, W., Schild, R., 2007. Middle Holocene environments of north and east Africa, with special emphasis on the African Sahara. In: Anderson, D.G., Maasch, K.A., Sandweiss, D.H. (Eds.), Climate Change and Cultural Dynamics. Academic Press, London, pp. 189e227. White, F., 1983. The Vegetation of Africa. UNESCO, Paris. Williams, M.A.J., 2009. Late Pleistocene and Holocene environments in the Nile basin. Global Planet. Change 69, 1e15. Williams, M.A.J., Adamson, D.A., 1980. Late Quaternary depositional history of the Blue and White Nile rivers in central Sudan. In: Williams, M.A.J., Faure, H. (Eds.), The Sahara and the Nile. A.A. Balkema, Rotterdam, pp. 281e362. Wiersma, A.P., Renssen, H., 2006. Modeledata comparison for the 8.2 ka BP event: confirmation of a forcing mechanism by catastrophic drainage of Laurentide Lakes. Quat. Sci. Rev. 25, 63e88. Young, R.W., Wray, R.A.L., Young, A.R.M., 2009. Sandstone landforms. Cambridge University Press, Cambridge, UK. Zanchetta, G., Bini, M., Cremaschi, M., Magny, M., Sadori, L., 2013. The transition from natural to anthropogenic-dominated environmental change in Italy and the surrounding regions since the Neolithic: an introduction. Quat. Int 303, 1e9. Zerboni, A., 2006. Cambiamenti climatici olocenici nel Sahara centrale: nuovi archivi paleoambientali. Tesi di Dottorato in Scienze Naturalistiche ed Ambi
degli Studi, Milano, p. 232. entali, XVIII ciclo. Universita Zerboni, A., 2008. Holocene rock varnish on the Messak plateau (Libyan Sahara): chronology of weathering processes. Geomorphology 102, 640e651. Zerboni, A., 2013. Early Holocene palaeoclimates in northern Africa: an overview. In: Shirai, N. (Ed.), Neolithisation of Northeastern Africa, Studies in Early Near Eastern Production, Subsistence, and Environment, vol. 16. ex Oriente, Berlin, pp. 65e82. Zerboni, A., 2011. Geomorfologia delle arenarie nel Sahara centrale (SW Libia). In: Fredi, P., Guglielmin, M. (Eds.), Atti della IV Giornata Nazionale dei Giovani Geomorfologi, Como, pp. 35e38. Zerboni, A., Trombino, L., Cremaschi, M., 2011. Micromorphological approach to polycyclic pedogenesis on the Messak Settafet plateau (central Sahara): formative processes and palaeoenvironmental significance. Geomorphology 125, 319e335. Zerboni, A., Cremaschi, M., 2012. Il significato paleoclimatico dei carbonati continentali olocenici centro-sahariani: memoria delle forzanti orbitali e di eventi climatici rapidi. Rend. Online Soc. Geol. Ital. 18, 36e39. Zhang, X., Gituru, R., Yang, C., Guo, Y., 2010. Exposure to water increased pollen longevity of pondweed (Potamogeton spp.) indicates different mechanisms ensuring success of anfiosperms in aquatic habitat. Evol. Ecol. 24, 939e953. Zuppi, G.M., Sacchi, E., 2004. Hydrogeology as a climate recorder: SaharaeSahel (North Africa) and the Po Plain (Northern Italy). Global Planet. Sci. 40, 79e91.