The Arabian Corridor Migration Model: archaeological evidence for hominin dispersals into Oman during the Middle and Upper Pleistocene

Proceedings of the Seminar for Arabian Studies 37 (2007): The Arabian Corridor Migration Model: archaeological evidence for hominin dispersals into Oman during the Middle and Upper Pleistocene J. ROSE Summary Studies of mitochondrial DNA haplogroup M suggest the Eurasian lineage of modern humans branched from an ancestral African population sometime between c. 70,000 and 40,000 years ago. The primary route of migration out of Africa was across the Bab al-Mandeb Strait at the southern end of the Red Sea. Until now, Palaeolithic archaeology in South Arabia has been terra incognita, and therefore these genetic data cannot be verified. Fieldwork conducted by the Central Oman Pleistocene Research programme between 2002 and 2006 has documented and sampled several new Palaeolithic sites. Lithic assemblages from these find spots show techno-typological affinities to industries in the Horn of Africa, the Levant, and India, underscoring Arabia's role as a nexus between continents. Archaeological material from Oman suggests there were a series of hunter-gatherer range expansions into southern Arabia from all three refugia over the last quarter of a million years. Some of these assemblages may correlate with the predicted divergence of haplogroup M; however, the technology and distribution of these sites imply the M0 founder population came from somewhere east of the Arabian Peninsula, contrary to the traditional out-of-Africa model. Given the palaeo-environmental and genetic data, early humans with the mtDNA haplogroup M marker may have originated in the Arabo-Persian Gulf basin or Indian subcontinent. Keywords: South Arabia, Palaeolithic archaeology, modern human origins, genetics, climate change "A cloud gathers, the rain falls, men live; the cloud disperses without rain, and men and animals die. In the deserts of southern Arabia there is no rhythm of the seasons, no rise and fall of sap, but empty wastes where only the changing temperature marks the passage of the years. It is a bitter, desiccated land which knows nothing of gentleness or ease. Yet men have lived there since earliest times." (Thesiger 1959: 1) Have people lived in southern Arabia since earliest times? Within the last decade, several mitochondrial DNA (mtDNA) analyses have been conducted to track the origin and spread of Homo sapiens; their conclusions unanimously support Thesiger's prophetic claim (e.g. Quintana-Murci et al. 1999; Forster & Matsumura 2005; Macaulay et al. 2005; Thangaraj et al. 2005). These genetic data suggest that all humans are, to varying degrees, derived from a single ancestral population that originated in East Africa around 200 kya. Sometime after their emergence, a branch of early humans bearing the mtDNA haplogroup M marker expanded out of Africa by way of the Arabian Peninsula. Anatomically modern human fossils recovered in Africa and the Near East partially corroborate the sce- nario proposed by geneticists. The earliest reliably dated Homo sapiens remains have been found in East Africa (albeit with some archaic features): 195–150 kya at Omo-Kibish in Ethiopia (McDougall, Brown & Fleagle 2005); 160–150 kya for the Herto specimen from Ethiopia (Clark et al. 2003); 150–140 kya at Singa, Sudan (McDermott et al. 1996); and 130–110 kya on the teeth discovered at Mumba Cave, Tanzania (Bräuer & Mehlman 1988). Modern human remains from Skhul and Qafzeh, dated between 110–90 kya (Mercier et al. 1995), provide evidence for human expansion during oxygen isotope stage 5 (128–72 kya); "populations tracking the fluctuations of their native habitats" (Lahr & Foley 1998: 155–57). Arabia must have played an important role in the development and dissemination of these early humans. The Peninsula connects Africa, Europe, and Asia. The heat-blasted gravel plains and mountainous dune fields of the Arabian interior, which Thesiger (1959: 1) described as "empty wastes where only the changing temperature marks the passage of the years", are but one link in a chain of deserts that stretch from north-western Africa to northern India, called the Afro-Asiatic Arid Belt. Based on palaeoclimatic reconstructions, we now 2 J. Rose FIGURE 1. Initial spread of anatomically modern humans represented by mtDNA haplogroup branches M, N, and R, after Metspalu (1994: fig. 5) know that the "changing" temperature of which Thesiger spoke was more pronounced than he could have ever imagined. Over the last quarter of a million years there were episodic fluctuations in the south-west Indian Ocean Monsoon, at which times the arid wastelands of Arabia were transformed into fertile, habitable ecosystems. The proposed model of hunter-gatherer range expansion into Arabia — the Arabian Corridor Migration Model (ACMM) — synthesizes the following lines of evidence to predict the spread and development of early humans onto the Peninsula: 1) the distribution and purported coalescence of mtDNA haplogroup M; 2) the chronology of climate change during the latter half of the Middle Pleistocene (340–128 kya) and Upper Pleistocene (128–12 kya); 3) the position of the Arabian Peninsula at the nexus of three continents and proximity to the locus of human origins; and 4) the record of marine transgression and regression in the Red Sea and Arabian Gulf. This model posits that, during wet phases, the Peninsula acted as a pump, drawing in populations that spilled out from surrounding environmental refugia; eastward from East Africa, southward from the Levant, and/or westward from the Gulf basin. Conversely, dry periods led to widespread desiccation in Arabia and wiped out pre-existing human groups, called tabula rasa events. Because of these tabula rasa events, the possibility of autochthonous development during hyperarid phases of the Pleistocene can be discounted, thereby avoiding Galton's Problem of migration versus diffusion. The Arabian Corridor Migration Model is tested using archaeological data collected during the Central Oman Pleistocene Research (COPR) fieldwork campaigns conducted between 2002 and 2006. Based on a synthesis of these data with previously published finds, The Arabian Corridor Migration Model three new lithic units, sensu lato, are described: the Sibakhan, the Nejd Leptolithic, and the Khasfian (Rose 2006). Technological and typological characteristics of these lithic entities suggest affinities with all three surrounding refugia at different times, and at the very least indicate that there was some degree of human occupation in southern Arabia over the last quarter of a million years. 3 The Bab al-Mandeb Strait The logical passage from Africa to Arabia is at the southern end of the Red Sea, where there is a narrow crossing between Yemen and Djibouti called the Bab alMandeb Strait. Presently, the narrowest point is an 18 km crossing at the Perim Narrows (Murray & Johns 1997). Scholars agree that the strait was never fully exposed at any time in the past quarter of a million years; there was never a land bridge. The bathymetric profile of the rift floor is such that it has a shallow depth of around 50 m for several kilometres, before plunging into a deep, narrow trough in the central channel (Fig. 1). So, when sea levels were 120 m lower during glacial maxima, Africa and Arabia was separated by a strait 17 m deep and 11 km wide at the Perim Narrows. Therefore, one of the first questions raised by ACMM is whether the Bab al-Mandeb was navigable by early humans, or if those 11 km of water represented an insurmountable geographic boundary. The name itself gives the impression of a daunting passage: Bab alMandeb translates from Arabic as "the gate of tears", or alternately, "the cemetery gate". Perhaps this is because from November to June there is a rapid current that flows southward into the Gulf of Aden, known as the hindi current. These fast-moving waters would have threatened to wash intrepid Pleistocene seafarers out into the open waters of the Indian Ocean. When the Indian Ocean Monsoon is active, upwelling in the Gulf of Aden forces the current to flip directions by pushing water into the Red Sea (typically between June and October); this is called the shami current (Siddall et al. 2003). Presumably, navigating the straits during the shami season would be much safer than crossing during the hindi season. Shami currents were more frequent during periods of reduced sea level, according to Siddall et al. (2003), when lower global eustatic sea levels forced a greater volume of water per annum to enter the Red Sea from the Gulf of Aden. At both maximum and minimum sea levels, the shores of Arabia were visible to groups living across the Red Sea. Far away on the Indonesian island of Flores, there is archaeological evidence attesting to an early colonization by Homo erectus in the Middle Pleistocene (Morwood et al. 1998). The island is far from the continental shelf and well beyond line of sight. Thus, the voyage to Flores, which may have occurred over half a million years ago, required considerably more sophisticated nautical technology than the Bab al-Mandeb crossing. In this light, traversing the Bab al-Mandeb, whether 10 or 20 km, was probably not a problem for Upper Pleistocene groups. Bednarik (2003) demonstrated the feasibility of prehistoric seafaring by constructing a raft capable of carrying out extended ocean voyages using only the lithic toolkit and perishable raw materials available to Middle Pleistocene hominids. Bottlenecks and range expansions If they came from the Horn, the expanding populations were moving in concert with the spread of an ecosystem to which they were already adapted. Scholars have long noted floral and faunal links between East Africa and southern Arabia: "This term [Levantine Corridor] is occasionally used by palaeontologists to designate the geographic route by which Miocene fauna moved from Africa into Eurasia. A similar dispersal route has been noted by botanists attempting to explain the presence of tropical Sudanese or East African species in Israel. Both faunal and floral researches have shown that the main passage was from Ethiopia and the Afar region into the southwestern corner of the Arabian Peninsula, and then into the Levant" (Bar-Yosef 1987: 30–31). There have been several studies of Afro-Arabian biogeographic diversity that indicate the movement of terrestrial species between these two regions (e.g. Harrison & Bates 1991; Tchernov 1992; Wildman et al. 2004; Fernandes, Rohling & Siddall 2006). At least nine mammalian orders comprising sixty-two species, made up primarily of small and medium-sized carnivores (e.g. mongooses, desert foxes, honey badger, caracal, jungle cat, golden jackal), occur on both sides of the Red Sea. Based on mtDNA analyses, Fernandes, Rohling and Siddall (2006) report a genetic divergence between African and Arabian genets sometime in the Middle or Upper Pleistocene. It is clear that at least one species of primate managed to cross the Bab al-Mandeb in the Upper Pleistocene: Papio hamadryas (Sacred Baboon). Presently, this desert-adapted species is indigenous to the rocky hill country of Somalia, Ethiopia, and Yemen (Nowak 4 J. Rose FIGURE 2. The chronology of Pleistocene lithic industries in southern Arabia. 1991). Genetic mtDNA analyses of P. hamadryas lineages on both sides of the Red Sea show that baboon groups crossed into Arabia several times during the Pleistocene, beginning as early as 400 kya (Wildman et al. 2004). The authors note that the most recent divergence between African and Arabian haplotypes occurred between 74–37 kya, which is the same span of time that witnessed the branching of mtDNA haplogroup M. So, it is germane to consider hominid range expansions into southern Arabia in conjunction with these other mammalian taxa. The ACMM predicts a continuous cycle of demographic expansions and contractions — genetic bottlenecks and releases — with every pluvial phase. Palaeo-environmental conditions in the Middle Pleistocene (781–128 kya) are characterized by gradually increasing aridity, marked by occasional episodes of fluvial activity caused by spikes in monsoon intensity (Anton 1984). Burns et al. (2001) report uraniumthorium dates from speleothems in northern Oman, which show periods of increased growth (i.e. wet phases) coinciding with OIS 9 (325–300 kya), OIS 7 (200–180 kya), OIS 5e (128–120 kya), and OIS 5a (82– 72 kya). Of particular importance to the ACMM are climatic conditions during OIS 6 (180–128 kya), when the first anatomically modern humans purportedly emerged in East Africa. Parker (personal communication) reports optically stimulated luminescence dates from lacustrine deposits in the north-eastern Rub al-Khali, suggesting there may have been occasional periods when Arabia was habitable between 180–128 kya. If this is indeed the case, then the ACMM must consider the possibility that early Upper Pleistocene hominids in southern Arabia may, in part, be derived from an indigenous Middle Pleistocene population. There is a plethora of signals attesting to a pronounced wet phase during OIS 5e (128–120 kya), at the onset of the Last Interglacial. Radiometric dates from speleothems (Burns et al. 2001), palaeosols (Anton 1984; Sanlaville 1992), and aeolianites (Glennie & Singhvi 2002) all point to a rapid and intense pluvial phase. These data suggest conditions were generally warmer and wetter throughout the duration of OIS 5, punctuated by another significant pluvial during OIS 5a (85–72 kya). Ambrose (2003) posits that the first genetic bottleneck release of anatomically modern humans occurred as a result of these ameliorated conditions, represented by the human fossil remains from Skhul and Qafzeh in the Levant (Mercier et al. 1995). There was a sudden and marked snap back to glacial conditions at the beginning of OIS 4. Some scholars correlate the sudden and drastic population decline at The Arabian Corridor Migration Model 5 FIGURE 3. Archaeological sites discovered by COPR in the ad-Dakhliyah province of north-central Oman. this time with the volcanic eruption (ranked category 8 or "mega-colossal" on the volcanic explosivity index) of Toba in northern Sumatra, which ejected over 800 km3 of ash and millions of tons of sulphur gas into the atmosphere, and may have triggered a 1000 year volcanic winter, at which time most of the earth was dominated by severely cold and dry conditions (Rampino & Ambrose 2000). They speculate that the female population in Africa was culled to approximately 5000 individuals during this bottleneck. ACMM predicts a second bottleneck release of modern humans associated with OIS 3 (60–24 kya), when environmental conditions improved and sprawling savannas, playa lakes, and seasonal streams returned to the interior (McClure 1984). This pluvial phase overlaps with the timeline of modern human emergence put forth by mtDNA analyses, which indicate that all lineages of haplogroup M coalesce back to a founder population between 70 and 50 kya. Figure 1, from Metspalu (2004) depicts the coalescence of the world’s three primary mitochondrial DNA branches (haplogroups M, N, and R) from the haplogroup L3 trunk in East Africa. The locus of the initial separation from an African gene pool (represented by M0), has not yet been identified by geneticists. Given the date range overlap of the African M1 divergence (48,000 ± 15,000 BP) and Indian M2 divergence (56,000 ± 7,000 BP), there is no a priori reason to assume one group is older than the other. So, it is equally plausible to model the location of origin in South Asia, South Arabia, the Arabo-Persian Gulf Basin, or perhaps even one linked gene pool occupying this entire geographic span. Palaeo-environmental signals indicate pronounced 6 J. Rose FIGURE 4. A Sibakhan bifacial hand axe (a) and a broken preform (b) from the A7–A10 complex of find spots. The Arabian Corridor Migration Model 7 FIGURE 5. Unidirectional-parallel (a) and unidirectional-convergent (b) Sibakhan cores from the A7–A10 complex. 8 J. Rose FIGURE 6. Marginally retouched Sibakhan blades from the A7–A10 complex. The Arabian Corridor Migration Model 9 FIGURE 7. Archaeological sites discovered by COPR in the Dhofar province of southern Oman. desiccation during OIS 2 (24–12 kya) (Anton 1984). Dating of the dune formations in the Rub al-‘Khali (McClure 1984), an-Nefud (Anton 1984), and the Wahiba Sands (Glennie & Singhvi 2002) attest an intense phase of aeolian deposition between approximately 17,000 and 9000 BP. This is corroborated by Late Pleistocene calcite fractures in Oman that formed at approximately the same time (Clark & Fontes 1990). Given these harsh conditions, ACMM predicts a tabula rasa episode that would have terminated Pleistocene human occupation on the Peninsula and herded remaining groups toward the exposed Ur-Schatt river valley that runs through the Arabo-Persian Gulf basin (Alsharhan and Kendall 2003; Kennett and Kennett 2006). Upland regions such as Dhofar or the Yemeni highlands may have also served as minor refugia. South Arabian lithic industries The description and development of lithic technologies are used to test ACMM. If episodic pluvial periods triggered a series of hunter-gatherer range expansions into South Arabia, lithic technologies in South Arabia must derive from populations in one or more contiguous zones that served as refugia during arid periods, when the Peninsula was uninhabitable. Five possibilities are considered: 1) expansion(s) from East Africa; 2) expansion(s) from the Levant; 3) expansion(s) from South Asia, specifically the Arabo-Persian Gulf Basin; 4) an amalgamation of technological traditions from all three regions; or 5) autochthonous development. In each case (except the last) the model predicts techno-typological affinities with lithic industries located in or near these refugia, since expanding hunter-gatherers will bring 10 J. Rose nology in the sense of volumetric, prismatic reduction. Platforms are primarily straight, and débordant blades are common. Like the Sibakhan assemblages in Oman, Wadi Muqqah find spots yielded slightly asymmetrical flat bifaces, limande in shape, with fine retouch on one or multiple edges. Biagi (1994) posited that the material from Saiwan was Upper Acheulean, dating to around 200,000 years ago. Citing techno-typological correlates in Europe, Inizan and Ortlieb (1987) also speculated the material from Wadi Muqqah fell somewhere around the late Lower Palaeolithic or Middle Palaeolithic. With so little data, dating of the Sibakhan must remain ambiguous, ranging from approximately 340 to 73 kya (between OIS 9 and OIS 5; it is assumed Arabia was uninhabitable during OIS 4). The Nejd Leptolithic1 is named as such for its ubiquity throughout the Nejd Plateau in northern Dhofar (Fig. 7), although sites fitting the description have been reported several hundred kilometres westward in Mahra and ДaΡramawt (Amirkhanov 1994; Rose 2002; Zarins, personal communication), well beyond the confines of the Nejd. The COPR 2006 expedition documented additional Nejd Leptolithic find spots in the Maradi Hills of northern Oman, situated on the south-eastern margin of the Rub al-‘Khali basin. Four Nejd Leptolithic assemblages were analysed: A22, T21, T27.I, and T29. In each case, blades comprise around 25 % of the debitage. This is not a true blade industry in the prismatic sense; the cores are lineal, with unidirectional-parallel (Fig. 8/a) or bidirectional (Fig. 8/b) removals across the working surface. The distal platforms on the bidirectional cores are supplementary, and seem to be intended for convexity maintenance rather than obtaining usable blanks. Platform faceting is rare (less than 10 %), although it is visible on a few of the cores. Regarding the toolkit, there are neither bifacial implements, nor bifacial thinning flakes among the debitage. A variety of sidescrapers, denticulates, notches, and retouched pieces comprise the dominant tool types. Unfortunately, there are few clues as to the dating of this industry. Amirkhanov (1991) produced a radiocarbon date of 31,000 ± 2000 BP for a similar blade assemblage in Yemen. These dates are given little weight, however, as they were obtained from disturbed, secondarily deposited sediments. Considering the position of Nejd Leptolithic sites on the landscape (particularly T29) and morphogenesis of associated geographic features, they appear to be relatively older than Khasfian find spots, and somewhat younger than Sibakhan. Therefore, the Nejd Leptolithic is placed generally be- with them the technology from whence they came. During the 2002 and 2004 COPR campaigns, lithic assemblages from find spots throughout Oman were collected and analysed, culminating in the definition of three Pleistocene lithic units, sensu lato (Fig. 2): the Sibakhan, the Nejd Leptolithic, and the Khasfian. The "industries" (not in the true sense) were then placed within a rough chronological scheme based on diagnostic technological and typological characteristics, as well as site setting and related landscape morphogenesis (Rose 2006). The earliest entity is the Sibakhan. Assemblages of this type have all been found in proximity to sibakh (indicative of ancient playas), therefore, in the absence of a stratified, in situ type site, the name Sibakhan has been chosen. Biagi (1994) reported the first occurrence at Saiwan in central Oman, and hypothesized that the material was late Lower Palaeolithic. COPR discovered a series of technologically and typologically similar assemblages approximately 30 km north of Saiwan in Wadi Qilfah, including a large lithic workshop at A7– A10 and an ephemeral scatter at A14 (Fig. 3). The industry is characterized by large, thin, asymmetrical, biconvex hand axes, which sometimes exhibit marginal retouch on one side (Fig. 4). Core technologies include hard hammer unidirectional-parallel blades from flat cores with naturally elongated working surfaces (Fig. 5/a), infrequent unidirectional-convergent Levallois cores with unfaceted platforms (Fig. 5/b), and, most rarely, discoids. Tools with nibbling retouch, often on the lateral edge of blades, are prominent (Fig. 6). There have been reports of possible Sibakhan occurrences in other parts of southern Arabia. Zarins et al. (1980) and Zarins, Murad and al-Yish (1981) mention Upper Acheulean bifaces from surface sites in Saudi Arabia and Yemen, which they classify as Mousterian of Acheulean Tradition due to their association with Levallois cores and discoids. The Group Two assemblage from Jebel Barakah is comprised of radial cores and the tip of a large biface with flat, invasive, hard hammer retouch. McBrearty (1999) notes similarities between this assemblage and material from the Middle Stone Age in East Africa. Inizan and Ortlieb (1987) discovered assemblages in their work around Shabwa, western ДaΡramawt, which have Sibakhan characteristics. Two occurrences were found on the plateau above Wadi Muqqah, which displayed a specific method of unidirectional-convergent Levallois resembling Middle Palaeolithic Nubian Type I cores (Guichard & Guichard 1965). There were also blade cores at Wadi Muqqah, which, like A7–A10 and Saiwan, were not considered part of a true blade tech- The Arabian Corridor Migration Model 11 FIGURE 8. Nejd Leptolithic blade cores from T29 (a) and T27 (b). 12 J. Rose FIGURE 9. Khasfian foliates (a, b) and a high-backed biconical discoid (c) from SH22–SH23. tween OIS 6 and OIS 3 (180–24 kya). The Khasfian "industry" is characterized by the Khasfian foliate (Fig. 9/a–b). This tool is a small (mean length=5.8 cm, mean width=3.5 cm) leaf-shaped or ovoid biface, with flat, soft hammer retouch and a lenticular cross-section. They were first recorded by J. Pullar at the site of Bir Khasfa in the northern reaches of Wadi Arah (1974), and have since been reported from surface collections throughout Oman (Smith GH 1977; Villiers-Petocz 1989; Rose 2004). Upon her initial examination of the material, Pullar recognized potential affinities with artefacts from the Horn, drawing comparisons with the Somaliland Doian industry (1974: 45). Since that time, similar foliates have been recognized within the Rub al-‘Khali assemblages collected by ARAMCO workers and studied by C. Edens (1982; 1988). The tools are classified as Type 5 in the Rub al‘Khali Neolithic scheme, although it should be noted that Type 5 foliates have never been found with any pressure-flaked Neolithic implement in a secure stratigraphic context. There are other soft hammer bifacial implements known from Saruq-Facies (D-Group) assemblages along the coast (Kapel 1967; Uerpmann 1992; Charpentier 1999; Uerpmann & Uerpmann 2003). Some Arabian scholars have posited the Khasfian, DGroup, Saruq-Facies, and Type 5 foliates all belong to a single industry falling somewhere between the sixth and fourth millennia BP (e.g. Charpentier 1999). It should not be assumed that the production of all bifacial foliates in southern Arabia is related to one time period or one lithic techno-complex. Morphologically similar specimens are reported from various industries The Arabian Corridor Migration Model around the world as early as the Middle Pleistocene, such as Korolevo level 5a in Transcarpathia (Kulakovska 2004) or Galeria Pesada in Portugal (Marks et al. 2002). Foliates appear in the Micoquian (Bosinski 1967; Soressi 2002) and Solutrean Industries (Smith 1966), which occur in the European Middle and Upper Palaeolithic, respectively. Nearly identical bifacial foliates (albeit fluted) are a characteristic of the early Holocene in North America (Wormington 1964), and are even known from Middle Holocene sites in central Australia (Valoch 1979). In addition, there is some degree of variability among foliate technologies in southern Arabia: the Saruq-Facies types are typically thick biconvex or planoconvex in cross-section, with sub-triangular shapes, while Khasfian foliates are lenticular in cross-section and exhibit a range of shapes from ovoid to limande. Furthermore, the related core technologies are markedly different; Khasfian assemblages are associated with centripetally prepared cores (Fig. 9/c), while Holocene industries are most often diminutive, single, or multiple platform orthogonal cores (Rose 2006). Finally, other archaeological materials that are often found with Holocene surface scatters such as ground stone implements, ostrich eggshell, marine shell, ochre, beads, hearths, and structures are absent from the Khasfian localities. Dating of the Khasfian is based on the assemblages' affinities with OIS 3 material from East Africa, as well as their repeated association with interior playas and drainage systems that empty into the Rub al-‘Khali basin. Given the date of the two most recent Rub al-‘Khali lake phases, 60–24 kya and 10–5 kya, it is proposed that the foliates fall within the earlier wet period. This association with OIS 3 must remain speculative in the absence of a securely dated, stratified site. 13 Discussion In light of these new data from Oman, the proposed model of hunter-gatherer range expansion(s) into Arabia can be tested by investigating inter-regional affinities with surrounding refugia: the Levant, South Asia (including the Arabo-Persian Gulf basin), and East Africa. Contrary to expectations, the Sibakhan appears closest to late Lower Palaeolithic material in the Near East. The Sibakhan techno-typological package is, for the most part, absent in East Africa during the Middle and Upper Pleistocene, while it is commonly found at Levantine sites clustering between OIS 9 and OIS 6 (340–128 kya). Levantine assemblages from this phase belong to the Mugharan tradition, which encompasses the Acheulo-Yabrudian, Yabrudian, and Amudian facies (Hours, Copeland & Aurenche 1973; Copeland 2000; Jelinek 1990). The predominant Sibakhan core reduction strategy is the same as the Amudian facies — the production of large hard hammer blades with prominent bulbs of percussion (Garrod 1956; Jelinek 1990). Like the Sibakhan, Amudian blade cores are single platform parallel, single platform convergent, and partially prismatic. Furthermore, the Amudian exhibits a significant number of retouched blades with marginal, nibbling retouch along one lateral edge, which are ubiquitous in Sibakhan assemblages. The façonnage component of the Sibakhan is closest to specimens from the Acheulo-Yabrudian facies, typified by large, flat, asymmetrical bifaces (usually considered bifacial scrapers) sometimes exhibiting Quina retouch along one or both edges. Connections between the Mugharan and Sibakhan intimate north–south demographic movements across Arabia during the late Middle Pleistocene, which, in turn, has implications for the dispersal of archaic Homo sapiens (i.e. Homo helmei). Accepting Zeitoun's (2001) taxonomic designation of the Levantine Zuttiyeh skull as an early member of the Homo sapiens family (found in association with Acheulo-Yabrudian material), then it is possible that the shift from the Acheulean (> OIS 9) to the Mugharan (OIS 9–OIS 6) was associated with the introduction (or local development) of archaic Homo sapiens. While it is impossible to speculate about the directionality of this expansion, the evidence suggests hunter-gatherers in Arabia and the Levant were linked to the same geographic range at this time, or expanded from the same refugium. The ensuing Nejd Leptolithic industry, posited to fall between OIS 6 and OIS 3, displays a very different pattern of regional affinities. Unlike the Sibakhan, there are few similarities with the Levant. While blades are often a significant component of Levantine Mousterian assemblages, these leptoliths were produced by a radically different chaîne opératoire in which they were often just the by-products of unidirectional-convergent Levallois reduction (Monigal 2002). Archaeological sites in the Horn of Africa dated to OIS 6–OIS 3 are organized into a techno-typological unit called the East African MSA or Somaliland Stillbay. Some sites include Gorgora Rockshelter (Leakey 1943), Porc Epic (Pleurdeau 2005), K’One (Kurashina 1978), and Midhishi 2 (Gresham 1984). The Somaliland Stillbay is distinguished by the presence of well-made bifacial foliates and unifacial points with flat, invasive retouch. In a few cases, Somaliland Stillbay assemblages exhibit unidirectional-parallel blades made from flat/partially volumetric cores (e.g. Porc Epic), which 14 J. Rose try in this analysis with potential African connections. The material resembles the Upper Stillbay reported at Gorgora (Leakey 1943) or early Somaliland Magosian from Layer D at Gure Warbei (Clark 1954). Both undated African assemblages are roughly attributed to OIS 3, and exhibit the same well-made, diminutive foliates and radial cores that are characteristic of the Khasfian. That is not to say it is associated with the Late Stone Age, as the Khasfian does not have any of the classic LSA tool types such as backed pieces, thumbnail scrapers, or microliths. Like Gure Warbei Layer D, there are no bipolar cores or blades, only discoids. Therefore, if the Khasfian industry originated in the Horn of Africa, it was brought by late MSA hunter-gatherers, prior to the development of LSA tools and technologies. The Nejd Leptolithic and Khasfian both potentially fall within OIS 3 (c. 60–20 kya), when genetic and archaeological evidence predicts anatomically modern humans first began their journey toward global domination. In both cases, they are attributed to the Middle Stone Age/Middle Palaeolithic, preceding the Upper Palaeolithic/Late Stone Age transition. Therefore, it appears this stage of dispersal had little to do with systemic behavioural or technological changes that facilitated their colonization (contra Klein 1998; Ambrose 2003; Mellars 2006). Rather, expansion into Arabia was initiated by opportunistic hunter-gatherers following episodic pulses of their known niche. In an alternate model, the genetically purported rapid expansion at the onset of OIS 3 may be due to marine transgression into the Arabo-Persian Gulf Basin, which displaced a sizeable seafaring population that had nucleated within the refugium. Regardless of interregional affinities and local technological facies, seemingly Upper Pleistocene blade assemblages are ubiquitous throughout the entire plateau spanning the province of Dhofar in Oman westward to the ДaΡramawt in Yemen. This area promises to yield important clues for understanding human origins. Surviving the wildly oscillating Upper Pleistocene climate may have been a trial of fire for early humans; episodic desiccation triggered a negative feedback loop that caused them to innovate technologies beyond just greater efficiency in lithic reduction. It has been suggested that the rapid spread of humans around 50,000 years ago was due to innovations in coastal seafaring (Stringer 2000). If this is the case, it implies the development of complex tools, the exploitation of a broad array of resources, and possibly even navigational capabilities. Given the paucity of Pleistocene archaeological sites in Arabia and the absence of absolute dates, the scenar- typify the Nejd Leptolithic. Yet there are no bifacial foliates, nor indications of any kind of façonnage technology in the Nejd Leptolithic, thus discounting a relationship between these two lithic complexes. So, the Nejd Leptolithic bears minimal evidence for techno-typological affinities with African or Levantine industries. There are parallels, however, with Upper Pleistocene sites in the Thar Desert of north-western India, which also has a high frequency of unidirectionalparallel blades (James & Petraglia 2005). Like the Nejd Leptolithic, the blades are not prismatic, but rather from single platform, lineal cores. The toolkit comprises sidescrapers, denticulates, notches, and burins, with a complete absence of bifacial tools. Given these lithic techno-typological affinities with India, and that the locus of haplogroup M0 (the earliest founder population) is unknown, it is germane to entertain the possibility that the founder population originated in south or south-west Asia, expanding first into Arabia and then back into Africa. One indication of this is the presence of a single mtDNA haplogroup M1 in Ethiopia, while most other M branches (as well as immediately derived N and R branches) coalesce in populations distributed throughout south-west and South Asia (Metspalu et al. 2004). Y-chromosome DNA markers also point to a locus of expansion from the same general area, where haplogroup F arose some 45,000 years ago and is presently found in 90 % of the world's male population (e.g. Cruciani et al. 2002; Regueiro et al. 2006). As most lineages derived from F cluster around the Arabo-Persian Gulf, it is likely future archaeological research there will be quite fruitful. There is also a possibility that the Nejd Leptolithic is an autochthonous development stemming from the Sibakhan. While bifacial implements are absent from Nejd Leptolithic assemblages, there is considerable overlap in their methods of core reduction (i.e. unidirectionalparallel/convergent flat blade cores). This scenario is feasible if one takes into account recent findings that suggest there were occasional ameliorated phases during OIS 6 (Parker, personal communication) that may have allowed some Middle Pleistocene hominids to expand episodically into southern Arabia during the Penultimate Glaciation and into the Last Interglacial. Khasfian inter-regional affinities must be approached with caution, because the data are from a single find spot and therefore, the full range of technotypological variability is unknown. It is clear, however, that there is no possible connection with India or the Levant, since both regions are completely devoid of bifacial foliates (the Khasfian-type fossil). The Khasfian is the only South Arabian lithic indus- The Arabian Corridor Migration Model ios proposed in this paper are speculative. The ACMM is intended to provide a hypothetical, testable model with which to frame future research. Only after more vigorous investigation in southern Arabia can we come to know those "weary bands of travellers, in some shady haunt, among Arabian sands" (Wordsworth 1807: 599). 15 Acknowledgements First and foremost, I thank Dr Anthony Marks for his faith in this research and indescribable generosity of time and resources. I would also like to acknowledge the COPR team, comprised of Dr Vitaly Usik (whose lithic illustrations are featured throughout this article), Mr Yves Guichard, Mrs Teresa Medici, Dr Daniel Richter, Mr Ali al-Mahrooqi, Mr Yaqoub al-Busaidi, Mr Ahmed al-Mukhaini, Dr Said al-Sukry, and Dr Diego Angelucci. Finally, I am grateful to the countless number of individuals in the Sultanate of Oman who have offered their time and energy to assist our project. Financial support was provided by the US National Science Foundation, the Institute for the Study of Earth and Man at Southern Methodist University, and from the Anonymous Donor's Fund at Southern Methodist University. Notes 1 Leptolithic is derived from the Latin word "leptos", meaning slender or narrow. It is used to describe industries that produce a significant amount of bladeproportionate debitage and blanks. References Alsharhan A. S. & Kendall C. G. St. C. 2003. Holocene coastal carbonates and evaporites of the southern Arabian Gulf and their ancient analogues. Earth Science Reviews 61: 191–243. Ambrose S. 2003. Population bottleneck. Pages 167–171 in R. Robinson (ed.), Genetics. iii. New York: Macmillan Reference. Amirkhanov H.A. 1991. The Paleolithic of South Arabia. Moscow: Academy of Sciences. 1994. Research on the Palaeolithic and Neolithic of Hadramaut and Mahra. Arabian Archaeology and Epigraphy 5: 217–228. Anton D. 1984. Aspects of geomorphological evolution: paleosols and dunes in Saudi Arabia. Pages 275–296 in A.R. Jado & J.G. Zötl (eds), Quaternary Period in Saudi Arabia. ii. Sedimentological, hydrogeological, hydrochemical, geomorphological, and climatological investigations of western Saudi Arabia. Vienna: Springer-Verlag. Bar-Yosef O. 1987. Pleistocene connections between Africa and Southwest Asia: an archaeological perspective. African Archaeological Review 5: 29–38. Bednarik R.G. 2003. Seafaring in the Pleistocene. Cambridge Archaeological Journal 13: 41–66. Biagi P. 1994. An Early Palaeolithic site near Saiwan (Sultanate of Oman). Arabian Archaeology and Epigraphy 5: 81–88. Bosinski G. 1967. Die Mittlepaläolithischen Funde im Westlichen Mittleuropa. Cologne/Graz: Böhlau Verlag. Bräuer G. & Mehlman M.J. 1988. Hominid molars from a Middle Stone Age level at the Mumba Rock Shelter, Tanzania. American Journal of Physical Anthropology 75: 69–76. Burns S.J., Fleitmann D., Matter A., Neff U. & Mangini A. 2001. Speleothem evidence from Oman for continental pluvial events during interglacial periods. Geology 29/7: 623–626. 16 J. Rose Charpentier V. 1999. Industries bifaciales holocènes d'Arabie orientale, un exemple: Ra’s al-Jinz. Proceedings of the Seminar for Arabian Studies 29: 29–44. Clark J.D. 1954. The Prehistoric Cultures of the Horn of Africa. Cambridge: Cambridge University Press. Clark I. & Fontes J-C. 1990. Paleoclimatic reconstruction of northern Oman based on carbonates from hyperalkaline groundwaters. Quaternary Research 33: 320–336. Clark J.D., Beyene Y., WoldeGabriel G., Hart W.K. & Renne P.R. 2003. Stratigraphic, chronological and behavioural contexts of Pleistocene Homo sapiens from Middle Awash, Ethiopia. Nature 432: 747–752. Copeland L. 2000. Yabrudian and related industries: the state of research in 1996. Pages 97–118 in A. Ronen & M. Weinstein-Evron (eds), Toward Modern Humans: the Yabrudian and Micoquian, Proceedings of a Congress held at the University of Haifa, November 3–9, 1996. (British Archaeological Reports, International Series, 850). Oxford: Archaeopress. Cruciani F., Santolamazza P., Shen P., Macaulay V., Moral P., Olckers A., Modiano D., Holmes S., Destro-Bisol G., Coia V., Wallace D.C., Oefner P.J., Torroni A., Cavalli-Sforza L.L., Scozzari R. & Underhill P.A. 2002. A Back Migration from Asia to Sub-Saharan Africa Is Supported by High-Resolution Analysis of Human Y-Chromosome Haplotypes. American Journal of Human Genetics 70: 1197–1214. Edens C. 1982. Towards a definition of the Rub al Khali "Neolithic". Atlal 6: 109–124. 1988. The Rub al Khali "Neolithic" revisited: the view from Nadqan. Pages 15–43 in D. Potts (ed.), Araby the Blest: Studies in Arabian Archaeology. Copenhagen: Tusculanum Press. Fernandes C.A., Rohling E.J. & Siddall M. 2006. Afro-Arabian biogeography and the Quaternary Red Sea: the origin of the Arabian populations of the Hamadryas baboon (Papio hamadryas) as a case study. Journal of Biogeography 33/6: 961– 966. Forster P. & Matsumura S. 2005. Did early humans go north or south? Science 308: 965–966. Garrod D.A.E. 1956. Acheuléo-Jabroudien et "Pré-Aurignacien" de la Grotte du Taboun (Mount Carmel); étude stratigraphique et chronologique. Quaternaria 3: 39–58. Glennie K.W. & Singhvi A.K. 2002. Event stratigraphy, palaeoenvironment and chronology of SE Arabian deserts. Quaternary Science Reviews 21: 853–869. Gresham T. 1984. An Investigation of an Upper Pleistocene Archaeological Site in Northern Somalia. MA Thesis, University of Georgia, Athens. [Unpublished]. Guichard J. & Guichard G. 1965. The Early and Middle Paleolithic of Nubia: a preliminary report. Pages 57–116 in F. Wendorf (ed.), Contributions to the prehistory of Nubia. Dallas, TX: Southern Methodist University Press. Harrison D.L. & Bates J.J. 1991. The Mammals of Arabia. (Second edition). Kent: Harrison Zoological Museum. Hours F., Copeland L. & Aurenche O. 1973. Les industries Paléolithiques du Proche-Orient, essai de corrélation. L'Anthropologie 77/3–4.): 229–280. Inizan M.L. & Ortlieb L. 1987. Préhistoire dans la région de Shabwa au Yemen du sud. Paléorient 13: 5–22. James H.V.A. & Petraglia M.D. 2005. Modern human origins and the evolution of behavior in the later Pleistocene record of South Asia. Current Anthropology 46 Supplement: 1–27. The Arabian Corridor Migration Model 17 Jelinek A.J. 1990. The Amudian in the context of the Mugharan tradition at the Tabun Cave (Mt. Carmel), Israel. Pages 81–90 in P. Mellars (ed.), The emergence of modern humans. Edinburgh: Edinburgh University Press. Kapel H. 1967. Atlas of the Stone-Age Cultures of Qatar. Aarhus: Aarhus University Press. Karmin M. 2005. Human Mitochondrial DNA Haplogroup R in India: dissecting the phylogenetic tree of southAsian specific lineages. MSc. thesis, University of Tartu, Estonia. [Unpublished]. Kennett D. J. & and Kennett J. P. 2006. Early State Formation in Southern Mesopotamia: Sea Levels, Shorelines, and Climate Change. Journal of Island and Coastal Archaeology 1: 67-99. Klein R. 1998. Why anatomically modern people did not disperse from Africa 100,000 years ago. Pages 509–521 in T. Akazawa, K. Aoki & O. Bar-Yosef (eds), Neanderthals and Modern Humans in Western Asia. New York: Plenum Press. Kulakovska L. 2004. Le Paléolithique moyen de Korolevo (Ukraine). Pages 141–148 in Acts of the XIVth UISPP Congress. (British Archaeological Reports, International Series, 1239). Oxford: Archaeopress. Kurashina H. 1978. An Examination of Prehistoric Lithic Technology in East-Central Ethiopia. PhD thesis, University of California, Berkeley. [Unpublished]. Lahr M. & Foley R. 1998. Towards a theory of modern human origins: geography, demography, and diversity in recent human evolution. Yearbook of Physical Anthropology 41: 137–176. Leakey L.S.B. 1943. The industries of Gorgora Rockshelter, Lake Tana. Journal of the East Africa and Uganda Natural History Society 17: 199–203. Macaulay V., Hill C., Achilli A., Rengo C., Clarke D., Meehan W., Blackburn J., Semino O., Scozzari R., Cruciani F., Taha A., Shaari N., Raja J., Ismail P., Zainuddin Z., Goodwin W., Bulbeck D., Bandelt H-J., Oppenheimer S., Torroni A. & Richards M. 2005. Single, rapid coastal settlement of Asia revealed by analysis of complete mitochondrial genomes. Science 308: 1034–1036. McBrearty S. 1999. Earliest stone tools from the Emirate of Abu Dhabi, United Arab Emirates. Pages 373–88 in P.J. Whybrow & A. Hill (eds), Fossil Vertebrates of Arabia. New Haven, CT: Yale University Press. McClure H.A. 1984. Late Quaternary Palaeoenvironments of the Rub’ al Khali. PhD thesis, University of London. [Unpublished]. McDermott F., Grün R., Williams C.T. & Din V.K. 1996. New late-Pleistocene Uranium-Thorium and ESR dates for the Singa hominid (Sudan). Journal of Human Evolution 31: 507–516. McDougall I., Brown F.H. & Fleagle J.G. 2005. Stratigraphic placement and age of modern humans from Kibish, Ethiopia. Nature 433: 733–736. Marks A.E., Burgal J-P., Chabai V.P., Monigal K., Goldberg P., Hockett B., Peman E., Elorza M. & Mallol C. 2002. Le gisement Pléistocène moyen de Galeria Pesada (Estrémadure, Portugal): premiers résultats. Paleo 14: 77–99. Mellars P. 2006. Why did modern humans populations disperse from Africa ca. 60,000 year ago? A new model. Proceedings of the National Academy of Sciences 103/25: 9381–9386. 18 J. Rose Mercier N., Valladas H., Valladas G., Reyss J-L., Jelinek A., Meignen L. & Joron J-L. 1995. TL dates of burnt flints from Jelinek's excavations at Tabun and their implications. Journal of Archaeological Science 22: 495–509. Metspalu M., Kivisild T., Metspalu E., Parik J., Hudjashov G., Kaldma K., Serk P., Karmin M., Behar D.M., Thomas M., Gilbert P., Endicott P., Mastana S., Papiha S.S., Skorecki K., Torroni A. & Villems R. 2004. Most of the extant mtDNA boundaries in South and Southwest Asia were likely shaped during the initial settlement of Eurasia by anatomically modern humans. BMC Genetics 5: 26. Monigal K. 2002. The Levantine Leptolithic: Blade Technology from the Lower Palaeolithic to the Dawn of the Upper Palaeolithic. PhD thesis, Southern Methodist University, Dallas, TX. [Unpublished]. Morwood M.J., O’Sullivan P.B., Aziz F. & Raza A. 1998. Fission track ages of stone tools and fossils on the east Indonesian island of Flores. Nature 392: 173–176. Murray S.P. & Johns W. 1997. Direct observations of seasonal exchanges through the Bab el Mandab Strait. Geophysical Research Letters 24/21: 2557–2560. Nowak R.M. 1991. Walker's Mammals of the World. (Fifth edition). Baltimore, MD: Johns Hopkins University Press. Pleurdeau D. 2005. Human technical behavior in the African Middle Stone Age: the lithic assemblage from Porc-Epic Dave (Dire Dawa, Ethiopia). African Archaeological Review 22/4: 177–197. Pullar J. 1974. Harvard archaeological survey in Oman, 1973: flint sites in Oman. Arabian Seminar 4: 33–48. Quintana-Murci L., Semino O., Bandelt H., Passarino G., McElreavey K. & Santachiara-Benerecetti A.S. 1999. Genetic evidence of an early exit of Homo sapiens sapiens from Africa through eastern Africa. Nature Genetics 23: 437–441. Rampino M. & Ambrose S. 2000. Volcanic winter in the Garden of Eden: the Toba supereruption and the late Pleistocene human population crash. Geological Society of America, special paper 345: 71–82. Regueiro M., Cadenas A.M., Gayden T., Underhill P.A. & Herrera R.J. 2006. Iran: Tricontinental Nexus for Y-Chromosome Driven Migration. Human Heredity 61:132–143. Rose J.I. 2002. Survey of prehistoric sites in Mahra, eastern Yemen. Adumatu 6: 7–20. 2004. The question of Upper Pleistocene connections between East Africa and South Arabia. Current Anthropology 45/4: 551–555. 2006. Among Arabian Sands: defining the Palaeolithic of southern Arabia. PhD thesis, Southern Methodist University, Dallas, TX. [Unpublished]. Sanlaville P. 1992. Changements climatiques dans la péninsule Arabique durant le Pléistocène supérieur et l'Holocène. Paléorient 18: 5–25. Siddall M., Rohling E.J., Almogi-Labin A., Hemleben C., Meischner D., Schmelzer I. & Smeed D.A. 2003. Sea level fluctuations during the last glacial cycle. Nature 423: 853–858. Smith G.H. 1977. New prehistoric sites in Oman. Journal of Oman Studies 3: 71–81. Smith P. 1966. Le Solutréen en France. Bordeaux: Delmas. Soressi M. 2002. Le Moustérien de tradition acheuléenne du sud-ouest de la France. Discussion sur la signification du faciès à partir de l'étude comparée de quatre sites: Pech-de-l'Azé I, Le Moustier, La Rochette et la Grotte XVI. PhD thesis, University of Bordeaux. [Unpublished]. The Arabian Corridor Migration Model 19 Stringer C.B. 2000. Palaeoanthropology: coasting out of Africa. Nature 405: 24–26. Tchernov E. 1992. Eurasian-African biotic exchanges through the Levantine corridor during the Neogene and Quaternary. Courier Forsch.-Institute Senckenberg 153: 103–123. Thangaraj K., Chaubey G., Kivisild T., Reddy A.G., Singh V.K., Rasalkar A. & Singh L. 2005. Reconstructing the origin of Andaman Islanders. Science 308: 996. Thesiger W. 1959. Arabian Sands. New York: E.P. Dutton and Company, Inc. Uerpmann M. 1992. Structuring the Late Stone Age of southeastern Arabia. Arabian Archaeology and Epigraphy 3: 65–109. Uerpmann H-P. & Uerpmann M. 2003. Stone Age Sites and their Natural Environment, the capital area of northern Oman, part III. Wiesbaden: Dr Ludwig Reichert Verlag. Valoch K. 1979. Stone tool assemblages from Arnhem Land. Pages 115–157 in J. Jelínek, Anthropology of the Rembrranga People, A Contribution of the Czechoslovak Anthropos Expedition to Arnhem Land N.T. Australia. Czechoslovakia: Anthropos Institute. Villiers-Petocz L.E. 1989. Some notes on the lithic collections of the Oman Department of Antiquities. Journal of Oman Studies 10: 51–59. Wildman D.E., Bergman T.J., al-Aghbari A., Sterner K.N., Newman T.K., Phillips-Conroy J.E., Jolly C.J. & Disotell T.R. 2004. Mitochondrial DNA evidence for the origin of Hamadryas Baboons. Molecular Phylogenetics and Evolution 32: 287–296. Wordsworth H. 1807. The Solitary Reaper. Poems in Two Volumes. London: British Library. Wormington H.M. 1964. Ancient Man in North America. Denver, CO: Denver Museum of Natural History. Zarins J., Murad A. & al-Yish K. 1981. Comprehensive Archaeological Survey Program: the second preliminary report on the southwestern province. Atlal 5: 9–42. Zarins J., Whalen N., Ibrahim M., Mursi A. & Khan M. 1980. Comprehensive Archaeological Survey Program: preliminary report on the central and southwestern province survey: 1979. Atlal 4: 9–36. Zeitoun V. 2001. The taxonomical position of the skull of Zuttiyeh. Comptes Rendus de l'Académie des Sciences 332: 521–525. Author's address Dr Jeffrey I. Rose, Department of Anthropology and Geography, Oxford Brookes University, Headington, Oxford, OX3 0BP, UK. e-mail rose@uta.edu