Complex Projectile Technology and Homo sapiens Dispersal into Western Eurasia

This paper proposes that complex projectile weaponry was a key strategic innovation driving Late Pleistocene human dispersal into western Eurasia after 50 Ka. It argues that complex projectile weapons of the kind used by ethnographic hunter-gatherers, such as the bow and arrow, and spearthrower and dart, enabled Homo sapiens to overcome obstacles that constrained previous human dispersal from Africa to temperate western Eurasia. In the East Mediterranean Levant, the only permanent land bridge between Africa and Eurasia, stone and bone projectile armatures like those used in the complex weapon systems of recent humans appear abruptly ca 45–35 Ka in early Upper Paleolithic contexts associated with Homo sapiens fossils. Such artifacts are absent from Middle Paleolithic contexts associated with Homo sapiens and Neandertals. Hypotheses concerning the indigenous vs. exogenous origins of complex projectile weaponry in the Levant are reviewed. Current evidence favors the hypothesis that complex projectile technology developed as an aid to ecological niche broadening strategies among African populations between 50–100 Ka. It most likely spread to western Eurasia along with dispersing Homo sapiens populations. Neandertals did not routinely deploy projectile weapons as subsistence aids. This puzzling gap in their otherwise impressive record for survival in some of the harshest environments ever occupied by primates may reflect energetic constraints and time-budgeting factors associated with complex technology. INTRODUCTION I evolution, only differences matter—differences in genes, differences in morphology, differences in behavior, differences in reproductive success, and differences in geographic dispersal. With respect to dispersal, there appear to have been two strikingly-different phases in Homo sapiens’ Pleistocene evolutionary history. From our earliest appearance in the fossil record ca 195 Ka (Fleagle et al. 2008) to around 50 Ka, Homo sapiens remained largely an endemic African species. Homo sapiens’ only demonstrable excursion from Africa prior to 50 Ka involved a short-lived dispersal into the East Mediterranean Levant between 75–130 Ka (Shea and Bar-Yosef 2005). After 50 Ka there is clear and unambiguous evidence for Homo sapiens dispersal from Africa to the rest of the Old World (Grine et al. 2007; Tishkoff and Verrelli 2003; Trinkaus, 2005; see papers in Mellars et al. 2007). Evidence for this dispersal is stronger in western Eurasia than in southern or eastern Asia and Sahul (Pleistocene New Guinea and Australia). Yet, nothing that archaeologists have found in these regions refutes the hypothesis that Homo sapiens underwent a vast geographic dispersal ca. 50 Ka (Dennell 2009; O’Connell and Allen 2007). The rich array of euphemisms bestowed on the 50 Ka dispersal event include “The Human Revolution” (Mellars and Stringer 1989), “The Upper Paleolithic Revolution” (Bar-Yosef 2002; Mellars 1994), and even “The Great Leap Forward” (Shreeve 1995). These terms reflect a widelyPaleoAnthropology 2010: 100−122. © 2010 PaleoAnthropology Society. All rights reserved. ISSN 1545-0031 doi:10.4207/PA.2010.ART36 JOHN J. SHEA Department of Anthropology and Turkana Basin Institute, Stony Brook University, Stony Brook, NY 11794-4364, USA; [email protected] MATTHEW L. SISK Interdepartmental Doctoral Program in Anthropological Science, Stony Brook University, Stony Brook, NY 11794-4364, USA; [email protected] shared belief that the human populations who dispersed ca 50 Ka differed behaviorally from earlier African humans. Complex behaviors previously only seen episodically in Africa and Southwest Asia became consistent features in the archaeological record wherever Homo sapiens fossils were deposited (Hovers and Belfer-Cohen 2006; McBrearty 2007; McBrearty and Brooks 2000; Mellars 2007; Shea 2006a, 2007a; Willoughby 2007). These behaviors include long-distance material transfers, the labor-intensive production of tools from stone and osseous tissues, elaborate exosomatic symbolic behavior (e.g., personal adornment), and complex subsistence strategies ranging from specialized biggame hunting to broad-spectrum foraging. In those parts of the Old World near Africa, first appearance dates for Homo sapiens outside of Africa frequently correspond closely with local and regional last appearance dates for other hominin species, such as the Neandertals (Mellars 2006a). In regions further afield, such as Sahul and the Americas, there is a suspiciously close chronological correlation between the appearance of Homo sapiens and large mammal extinctions (Martin 1984). Homo sapiens’ role (if any) in the extinction of the Neandertals, of penecontemporaneous Asian hominins, and of other large mammals remains uncertain and hotly disputed (Grayson 2001; Zilhao 2006). In searching for explanations for the success of the 50 Ka Homo sapiens dispersal, uniformitarian principles suggest we should start by examining modern-day evolutionComplex Projectile Technology and Human Dispersal • 101 ary processes whose outcomes are structurally analogous to those patterns in the paleoanthropological record that we are trying to explain. Since the end of the Pleistocene Homo sapiens has undergone rapid population growth and an extension of our geographic range as the result of technologically-assisted strategies for niche-broadening. The most recent and consequential of these strategies are agriculture and pastoralism. Yet, agriculture and pastoralism are strategic innovations of Holocene antiquity, possibly because wide Pleistocene paleoenvironmental variability militated against their development in earlier times (Richerson et al. 2001). The use of complex projectile weaponry is a universal human technological/subsistence strategy thought to have a Pleistocene antiquity of at least 50 Ka (Shea 2006b, 2009b). As such, it is a plausible factor in the 50 Ka human dispersal. Projectile technology is niche-broadening technology. Its universal distribution among recent humans almost certainly indicates that it confers a significant ecological advantage. Complex projectile weapons, like the bow and arrow, and spearthrower and dart, enable humans to exploit a far greater range of potential animal prey than our near primate relatives do. Recent humans use complex projectile weapons against prey ranging from elephants to rodents, against terrestrial, avian, and aquatic species (Churchill 1993). Compared to heavy hand-cast spears, complex projectile weapons have a greater effective range, are more readily transportable, and allow multiple shots at a target (Yu 2006). The bow and arrow, in particular, offers the further advantage of being usable in three dimensions. One can launch arrows up, down, or horizontally with equal effect. All these factors significantly reduce post-encounter energetic costs of predation to humans equipped with complex projectile weapons. Theoretically, the long-term consequences of such reduced costs ought to have included stabilized foraging returns, population growth, and geographic dispersal. This paper considers the role of complex projectile technology in Homo sapiens’ Late Pleistocene dispersal. We focus on the dispersal into western Eurasia through the Levant, because this region offers a rich set of data with which to test the hypothesis that complex projectile technology was a major factor in human dispersal against other hypotheses. EXPLAINING THE 50 KA HOMO SAPIENS DISPERSAL The principal hypotheses currently invoked to explain the 50 Ka dispersal into western Eurasia focus on either: (1) cognitive changes leading to greater use of symbolism and, by implication, language (Coolidge and Wynn 2009; Deacon 1997; Deacon and Deacon 1999; Gamble 2007; 2008; Henshilwood and Marean 2003; Klein 1995; Mithen 1998; Wadley 2001); (2) population increases culminating in the formation of extensive alliance networks, (Ambrose 1998; McBrearty and Brooks 2000; Powell et al. 2009; Stiner and Kuhn 2006); and, (3) some combination of both factors (Chase 2006; Clark 2002; d’Errico 2007; Mellars 2006b; Soffer 1994). Cognitive changes and population pressure are plausible factors in dispersal, but both are difficult to investigate in ways that allow hypotheses about them to be conclusively refuted with archaeological evidence. Measuring variation in human cognition and characterizing symbol use is problematical even among living populations (Gould 1981; Hodder 1982). In assessing prehistoric cognition, the best that most archaeological investigations can do is to assert a hominin population possesses “modern” human abilities because their archaeological record contains artifacts that resemble exosomatic symbols used by recent humans. There have been more nuanced efforts to analyze prehistoric cognition from the archaeological and paleontological evidence (Coolidge and Wynn 2009; Mithen 1996), but in actual practice, archaeologists usually rate prehistoric cognition as “modern,” or not. Symbol use is often the bellwether of such modern cognition (Henshilwood and Marean 2003); yet, much human symbol use involves things that do not fossilize, such as spoken language, gesture, and perishable media. This raises the possibility that archaeologists might mistakenly infer the absence of symbol use simply from an absence of preserved evidence. A shift to rendering symbolic artifacts in durable media might appear to be the beginning of symbol use rather than a change in a long-extant habit. Thus, in a comparison of symbolic evidence from two contexts, similar evidence can legitimately be taken as evidence of similar abilities, but differences remain subject to multiple interpretations. Questions about population size are some of the most difficult ones for archaeologists to answer with any degree of precision. If modern nation-states have to spend millions of dollars and complex statistical tools to accurately estimate their own populations through census, one has to expect that archaeologists’ casual efforts to estimate prehistoric population size variation are likely to be fraught with even greater problems. Most such studies are based on recent human hunter-gatherer demography (Binford 2001; Powell et al. 2009; Wobst 1974; Zubrow 1991). Yet hunter-gatherer population sizes and densities are widely-variable, even among populations living under similar ecogeographic conditions (Kelly 1995). Some of the ambiguity in these estimates reflects the widely variable methods used to gather these demographic data. But, it is also the case that such variability is an intrinsic quality of human hunter-gatherer adaptation. Forager societies often cope with resource shortfalls and conflicts by altering their distribution and density on the landscape. Moreover, they can do this quickly, situationally, and recursively. Because hunter-gatherers adapt to landscapes, while archaeologists sample locations in incompletely-preserved landscapes, it is difficult to incorporate these qualities of landscape-level variability into estimates of Pleistocene human population size and density. For example, evidence for population increase at one or more sites may not so much reflect region-wide population increases but rather aggregations of humans around those localities, or “population packing” (Binford 2001). 102 • PaleoAnthropology 2010 These considerations do not refute hypotheses that cognitive and/or demographic changes played a role in Late Pleistocene human dispersal; but, they do suggest we should continue to search for other factors leading to that dispersal. Projectile technology has been named among the derived features of the 50 Ka dispersal, but it has usually been portrayed as of secondary importance to cognitive and social-demographic processes (Bar-Yosef 2002; Binford 1970; Dennell 1983; Gilman 1984; Henshilwood and Marean 2003; Klein 1998; Klein and Edgar 2002; Mellars 1973; White 1982). Recent improvements in archaeologists’ ability to detect projectile weapon use in the Paleolithic suggest the role of projectile technology in human dispersal needs to be re-evaluated (Brooks et al. 2006; Churchill and Rhodes 2009; Lombard and Pargeter 2008; Shea 2006b; 2007a; 2008; 2009b). Archaeologists use a variety of terms to describe projectile technology (Knecht 1997). It is important to parse these terms before discussing the prehistoric record. We use the term “complex projectile technology” to refer to weapons systems that use energy stored exosomatically to propel relatively low mass projectiles at delivery speeds that are high enough to allow their user to inflict a lethal puncture wound on a target from a “safe” distance. The most widespread such weapons systems are the bow and arrow, and the spearthrower and dart. The bow and arrow stores energy in the flexion of the bow. The spearthrower stores energy in the flexion of the dart. Some prehistorians also include hand cast spears, stones, throwing sticks, and other missiles among “projectile weapons.” We refer to these as “simple” projectile weapons because they do not involve exosomatic energy storage. They are launched at their targets with unassisted bodily force. Simple projectile weapons are relatively heavy, slow-moving, and must be used at close quarters to their intended targets. It is true that some highly-trained athletes can throw spears and rocks impressive distances, but such weapons lose kinetic energy at a rapid rate. They are notoriously difficult to aim at such distances, and they offer few tactical advantages against small mobile prey. Against larger uninjured prey with “proactive” antipredator defenses, they offer few advantages at all. A healthy adult American bison or Cape buffalo can charge 50 meters in a matter of seconds. Ethnographic throwing spears are usually delivered from around 8–10 meters on average and used to dispatch prey already incapacitated by other means (Churchill 1993). Thus, the 2.5 meter-long wooden spears recovered from Middle Pleistocene contexts at Lehringen and Schöningen (Germany) can be viewed as simple projectile weapons, but not complex projectile weapons (Schmitt et al. 2003). Clubs, thrusting spears, and other weapons held in the hand during delivery are, by definition, not projectile weapons, nevertheless, some prehistoric projectile weapons may have been used in both projectile and non-projectile tasks, much as ethnographic projectile weapons are used today (Greaves 1997). EVIDENCE FROM THE LEVANT The “Levant” is a that part of Southwest Asia and the eastern Mediterranean encompassed by the modern states of Lebanon, Syria, Israel, Jordan, the Palestinian National Authority, and adjacent parts of Turkey, Iraq, Egypt, and Saudi Arabia. Ecologically, the Levant is defined by the Mediterranean oak-terebinth woodland and its ecotone with the Irano-Turanian steppe (Blondel and Aronson 1999). The Levant is a kind of transition zone between Africa and Eurasia. The flora and fauna of the Middle-Late Pleistocene Levant were overwhelmingly Eurasian (i.e., Palearctic) (Tchernov 1988); nevertheless fauna now endemic to Africa, including hippopotamus, rhinoceros, warthog, zebra, ostrich, and other Afro-Arabian fauna once ranged freely in this region (Kingdon 1990). For much of the Pleistocene, the Levantine “corridor” has been the only permanent land bridge linking Africa and Eurasia. It also straddles the boundary between the Mediterranean Basin and the Indian Ocean. As such, it is a convenient place in which to monitor the movement of hominin species and diffusion of ideas around the Old World. LEVANTINE LATE PLEISTOCENE PREHISTORY The archaeological record for the Late Pleistocene Levant is divisible into three major phases, “Interglacial Middle Paleolithic” (IMP), the Later Middle Paleolithic (LMP), and the “Early Upper Paleolithic” (EUP). The IMP lasted between 80–130 Ka and is roughly coterminous with Marine Oxygen Isotope Stage 5, or the Last Interglacial, sensu lato. Dated IMP archaeological contexts include Tabun Cave Unit I (Garrod’s Levels B–C), Skhul Cave Level B, Qafzeh Caves Units XVII–XXIV, Douara Cave Unit IIIB, Hayonim Cave Layer E (Unit 2), Nahr Ibrahim, and the Enféan II and Naaméan beaches at Naamé (Figure 1). Shea (2003a) provides a recent overview of the evidence from these and other IMP sites (see also Bar-Yosef 2000). Levantine prehistorians group IMP stone tool assemblages together into the “Tabun C-Type” Levantine Mousterian, which is named after major geological units in Garrod’s stratigraphy of Tabun Cave on Mount Carmel. Core reduction strategies in these assemblages emphasize radial-centripetal preparation and the production of large oval Levallois flakes. Prismatic blades are relatively rare, and carved bone/antler tools are unknown. Hominin fossils from IMP contexts consist primarily of the Homo sapiens fossils from Skhul and Qafzeh (Rak 1998). Neandertal fossils also are known from the upper part of Tabun Cave Unit I (Levels B and C), but the stratigraphic provenance of these fossils are problematical (Bar-Yosef and Callendar 1999). The LMP spans the period from 75–45 Ka, encompassing MIS 4 and the early part of MIS 3. Archaeological contexts dating to this period and containing hominin remains include Amud Cave Levels B1–4, Dederiyeh Levels 3–11, Geula Cave B Level B1/B2, Kebara Cave Units VI–XII, Shovakh Cave “Lower Cave Earth,” and Shukhbah Cave Level D (Figure 2). LMP archaeological assemblages not associated with hominin fossils include Boker Tachtit Level 1, Biqat Complex Projectile Technology and Human Dispersal • 103 Quneitra, Far’ah II, Jerf Ajla Cave Level C, Ksar Akil Rockshelter Level XXVI, Tor Faraj Level C, and Tor Sabiha Level C. LMP lithic assemblages show frequent use of recurrent unidirectional-convergent core preparation, as well as variable amounts of radial-centripetal preparation. There is evidence for prismatic blade production, but it is rare and follows a different set of procedures than those methods predominating among EUP assemblages. All LMP hominin fossils thus far discovered are either Neandertals, or they are too fragmentary for the morphological affinities to be conclusively assessed. The Upper Paleolithic period began around 47 Ka and lasted to around 19 Ka. This paper focuses on the evidence from 47–28 Ka, or the EUP, that part of the Upper Paleolithic that is prior to the Last Glacial Maximum. The EUP has a relatively rich archaeological record, one recently reviewed in detail by Belfer-Cohen and Goring-Morris (2003; see also Shea 2007a). Well-documented EUP contexts include Boker Figure 1. Map showing IMP sites mentioned in the text. 104 • PaleoAnthropology 2010 A Level 1, Boker BE Levels I–III, Boker Tachtit Level 4, Hayonim Layer D, Kebara Cave Units I–IV/Levels D–E, Ksar Akil Levels IV–XXV (phase III–VII), Lagama IIID, VII, and VIII, Qafzeh Cave Levels C–E/ 4–11, Üçagizli Cave Layers B–H, Umm el Tlel 2 Levels V–XI, Umm el Tlel III2a (late MP), and Wadi Abu Noshra I, II, and VI (Figure 3). Several dozen additional assemblages can be assigned to the EUP on the basis of stone tool typology. Most lithic assemblages from the EUP feature prismatic blade and bladelet cores, and laminar debitage is common. Levantine prehistorians subdivide EUP assemblages into named industries, including the Initial Upper Paleolithic/Emiran, the Ahmarian, the Levantine Aurignacian, and a fourth unnamed flake-based industry, on the basis of variation in retouched tool types and the relative frequencies of blades and bladelets. A variety of carved bone/antler implements have been recovered from EUP contexts. EUP human fossils include the burials from Ksar Akil, two sets of cranial remains from Qafzeh, Figure 2. Map showing LMP sites mentioned in the text. Complex Projectile Technology and Human Dispersal • 105 among humans possessing similar needs for shelter, food, and tools, and living in broadly comparable environments. They do not necessarily indicate a shared social identity or a close evolutionary relationship. Indeed, only Neandertal fossils are known from Levantine contexts dating to between IMP and EUP periods (45–75 Ka), raising the possibility of an interruption in Homo sapiens occupation of the Levant (Shea 2008). Table 1 summarizes the main differences between the and fragmentary remains from Hayonim Level D. The Ksar Akil and Qafzeh fossils are remains of Homo sapiens (Smith 1995). Ten isolated teeth recently recovered from EUP levels of Üçagızli Cave also are identified as Homo sapiens (Gulec et al. 2007). SIMILARITIES AND DIFFERENCES The many similarities among the evidence for IMP, LMP, and EUP behavior match what one would expect to see Figure 3. Map showing EUP sites mentioned in the text. 106 • PaleoAnthropology 2010 where. Improved radiocarbon chronology for the earliest Upper Paleolithic contexts in Europe increasingly point to a dispersal of Ahmarian-affiliated “Fumanian” assemblages from the Levant to southern Europe beginning shortly after 45 Ka (Mellars 2006a). Artifactual similarities between Levantine EUP assemblages and North Africa “Dabban” ones further hint at a possible EUP dispersal from the Levant to North Africa during EUP times (Iovita 2009). SYMBOL USE AND POPULATION DENSITY That Levantine IMP humans possessed a symbolic capacity similar to that of post-50 Ka humans seems clear from the numerous burials, perforated shells, and fragments of mineral pigment recovered from Skhul, Qafzeh, and other sites (Bar-Yosef Mayer et al. 2009; Kuhn et al. 2001; Vanhaeren et al. 2006). Nevertheless, there are a few differences. IMP burials incorporated nonhuman skeletal remains as mortuary furniture, while EUP burials apparently did not (admittedly the sample is small). EUP humans manufactured perforated ornaments from human and animal teeth, a practice that is not known from IMP contexts. Such differences are comparable in scale to variation in symbolic behavior among ethnographic human societies. They are not compelling evidence for differences in cognitive and symbolic capacities. The duration of the EUP was half that of the IMP; yet, dated EUP contexts outnumber IMP ones by more than two to one, and they are far more broadly distributed in the Levant (compare Figures 1 and 2). This would seem to argue for larger EUP populations, but this is not necessarily true. Much less of the IMP and LMP landscape is preserved simply as a function of erosion. Moreover, the recognition criteria for IMP, LMP, and EUP assemblages differ widely. EUP assemblages contain diagnostic artifact-types, such as Ksar Akil, El Wad and Emireh points, chanfrein scrapers, and various other tools that enable even small surface ocLevantine IMP, LMP, and EUP records (see also Shea 2007a, 2008, 2009a). EUP and LMP humans appear to have settled in arid, low-productivity habitats to a greater extent than their IMP counterparts (Henry 1998). Levantine EUP subsistence differs from the LMP and IMP in more systematic exploitation of smaller animals and sessile prey (rodents, lagomorphs, birds, tortoises)(Rabinovich 2003; Stiner 2006). EUP lithic assemblages differ from IMP and LMP ones in the use of microlithic technology and in the production of tools from osseous materials (Shea 2006a). Burials of adults and children are known from all periods, but clear examples of exosomatic symbolic artifacts, such as red ochre and perforated shells are known only from the IMP and EUP (Hovers et al. 2003; Kuhn et al. 2001; Shea and Bar-Yosef 2005). Whereas stone tools from IMP and LMP contexts are similar to contemporaneous tools made throughout much of Europe and Africa, stone tools from EUP contexts differ from those made at the same time in adjacent regions. EUP lithic assemblages also preserve evidence for internal stylistic variation that is not paralleled by variation among IMP and LMP contexts (Belfer-Cohen and Goring-Morris 2003; Shea 2007a). Dispersals of varying degrees of success mark the Levantine Late Pleistocene paleoanthropological record. Homo sapiens was present in the Levant 80–130 Ka, but the archaeological record for this period does not preserve evidence for dispersals further northwards into temperate western Eurasia. The presence of Neandertals in LMP contexts is seen by some researchers as reflecting movements of Neandertals into the Levant from montane western Asia, either as the result of climatic forcing during the rapid onset of glacial conditions ca. 75 Ka (Bar-Yosef 1988; Serangeli and Bolus 2008), or as the result of their own adaptive innovations. Neandertals did not demonstrably disperse further south into the Arabian Peninsula or into Africa. Only EUP humans dispersed successfully through the Levant to elseTABLE 1. BEHAVIORAL DIFFERENCES INFERRED FROM IMP VS. EUP ARCHAEOLOGICAL EVIDENCE. Behavioral Variable IMP