Special Reviews

Going East: New Genetic and Archaeological Perspectives on the Modern Human Colonization of Eurasia

See allHide authors and affiliations

Science  11 Aug 2006:
Vol. 313, Issue 5788, pp. 796-800
DOI: 10.1126/science.1128402


The pattern of dispersal of biologically and behaviorally modern human populations from their African origins to the rest of the occupied world between ∼60,000 and 40,000 years ago is at present a topic of lively debate, centering principally on the issue of single versus multiple dispersals. Here I argue that the archaeological and genetic evidence points to a single successful dispersal event, which took genetically and culturally modern populations fairly rapidly across southern and southeastern Asia into Australasia, and with only a secondary and later dispersal into Europe.

Research over the past 20 years has provided an increasingly clear picture of the way in which our own species (Homo sapiens) emerged and subsequently spread across the rest of the occupied world. DNA evidence and fossil skeletal remains indicate that human populations that were essentially “modern” both anatomically and in their mitochondrial and Y-chromosome lineages had emerged in Africa by at least 150,000 years ago, perhaps closer to 200,000 years ago (111). Studies of present-day world populations (especially those based on the maternally inherited mitochondrial DNA) strongly suggest that a small subset of these African populations made the crossing from northeastern Africa, probably over the mouth of the Red Sea, and subsequently dispersed into Arabia and southern Asia sometime before 50,000 years before present (yr B.P.) (2, 8, 1217) (Fig. 1). Recent studies have suggested that these populations expanded rapidly along the coastlines of southern Asia, southeastern Asia, and Indonesia to arrive in both Malaysia and the Andaman Islands by at least 55,000 yr B.P., and conceivably as early as 60,000 to 65,000 yr B.P. (12, 1821)—though more recent estimates of mitochondrial DNA mutation rates (8) suggest that these figures may be overestimates. As Carl Sauer pointed out in 1962 (22), a strongly coastal pattern of dispersal would make good sense in ecological and demographic terms, because this would presumably have required only limited economic adaptations from one coastal location to another.

Fig. 1.

Map of possible dispersal routes of anatomically and genetically modern human populations from Africa to Asia and Australia according to Forster and Matsumura (12). The models assume an origin in eastern Africa, and dispersal either via the Nile Valley and Sinai Peninsula (the “northern” route) or via the mouth of the Red Sea to Arabia and Australia (the “southern” route).

The subsequent dispersal southward of these populations into New Guinea and Australia (at that time connected as an extended landmass known as “Sahul”) is currently more controversial. We know that typically anatomically modern populations were present at the Niah cave in Sarawak by at least 41,000 years ago (23), and that some of these populations had made a sea crossing of at least 90 km to reach parts of Australia by at least 45,000 yr B.P.—best represented by the typically anatomically modern skeleton from the site of Lake Mungo 3 in New South Wales (1, 2426). Some archaeological claims have been made for the initial colonization of Australia as early 50,000 to 60,000 yr B.P. (27), but these remain speculative and contested (24). Similarly, claims for the presence of behaviorally modern populations in Malaysia before the Mount Toba (Sumatra) volcanic “supereruption” around 74,000 yr B.P. (16) have yet to receive any clear support from recent archaeological research in the region (28). But in any event, it is clear that the initial dispersal of modern human populations eastward from their original African homeland along the so-called coastal express (12, 18, 20) route into Australasia occurred over a comparatively short time, amounting to at most 15,000 years, and probably less than 10,000 years (i.e., an overall dispersal rate of at least 1.0 km per year) if we take the combined DNA and archaeological evidence into account (12, 18, 21).

Single or Multiple Dispersals?

The most controversial issue at present centers on whether there could have been not one, but possibly two or more separate dispersals of anatomically and genetically modern humans from Africa into Eurasia. An earlier model advanced by Lahr and Foley in 1994 (2931) envisaged at least two separate episodes of dispersal from northeastern Africa, associated with sharply differing patterns of stone-tool technology. Of these, the “northern” dispersal extended northward via the Nile Valley and the Sinai Peninsula into southwestern Asia (and eventually Europe), associated with typically blade-dominated, “Upper Palaeolithic” or “Mode 4” technology, best represented at the sites of Boker Tachtit in southern Israel and Ksar Akil in Lebanon, both dated to around 45,000 to 50,000 yr B.P. (3234). The separate “southern” dispersal extended from the Horn of Africa across the mouth of the Red Sea (the Bab el Mandeb straits) carrying technologically simpler “Middle Palaeolithic” or “Mode 3” technology, which subsequently dispersed eastward along the coasts of southern and southeastern Asia into Australia (21, 2931). The sharp contrasts in the technology associated with these two dispersals were taken as an explicit reflection of two separate source populations in Africa, with the southern, Mode 3 dispersal occurring substantially earlier than the northern, Mode 4 dispersal event (2931).

Recently, the notion of two or more separate dispersals of anatomically and genetically modern humans has come under increasing scrutiny from molecular geneticists, based on studies of both mitochondrial and Y-chromosome patterns in African and Asian populations (2, 8, 1215, 18). These recent studies suggest that the whole of modern Asian and European populations derive from one small subset of the so-called L3 mitochondrial lineage in Africa, which subsequently diverged into the derivative M and N lineages, probably shortly after their dispersal from Africa (2, 8, 1215). The crux of the arguments advanced by Kivisild (8), Forster (2), Matsumura (12), Macaulay et al. (18), and others is that the very limited genetic diversity exhibited by modern European and Asian populations—compared to those in Africa—would be effectively impossible to reconcile with the model of two separate dispersal events, deriving from separate source populations in Africa, and hypothetically at two different dates. Essentially the same conclusion has been drawn independently by Endicott and others (13) based on studies of the paternally inherited Y-chromosome lineages. Even allowing for the current controversies surrounding the interpretation and dating of the DNA evidence (2, 5, 8, 11, 14, 3537), it is becoming increasingly difficult to reconcile the available genetic data with the hypothesis of two or more separate dispersal events from Africa into Eurasia—although this point has been debated in some earlier genetic studies (15, 38).

The Archaeological Evidence

The major challenge of this scenario now is to document the individual steps in this colonization process on the basis of the “hard” archaeological evidence. Large areas of both Arabia and India in particular are at present largely blank areas on the archaeological map over the critical time range from ∼50,000 to 60,000 yr B.P. in question (39, 40). And of course, all the coastlines of this period are now deeply submerged below the rapidly rising sea levels of the past 15,000 years (20, 21). There are, however, already some intriguing hints from south Asia of what future research may reveal. From the sites of Patne in western India (41), Jwalapuram in southeast India (42), and Batadomba-lena in Sri Lanka (43, 44), there are archaeological assemblages showing some striking resemblances to those from eastern and southern Africa that must be from very close to the period when modern humans first dispersed from Africa (Fig. 2A). These sites contain large numbers of small “crescentic” forms of stone tools (evidently parts of hafted implements, and conceivably components of archery equipment) (45, 46) that are markedly similar to those that define the so-called Howiesons Poort technology in southern and eastern Africa, dated broadly to around 55,000 to 65,000 yr B.P. (4551)—as at the sites of Mumba in Tanzania (45, 52), Norikiushin in Kenya (53), and a range of similar sites in southern Africa (4549) (Fig. 2B). Broadly similar industries including rather larger forms of backed “segment” forms are dated to between 60,000 and 40,000 yr B.P. at the site of Enkapune ya Muto in Kenya (53, 54). Although the Indian crescentic tools are generally smaller than those from the African sites, the sequences at Patne and Jwalapuram suggest that these forms become increasingly larger in the lower levels of the sequence, and correspondingly more similar to the African tools (41, 42). Even more strikingly, these Indian sites have produced carefully shaped and perforated beads manufactured from fragments of ostrich eggshell, closely similar to those found in African sites (39, 41, 45, 47), together with a further piece of ostrich eggshell incised with a distinctive crisscross motif (41), which is strongly reminiscent of designs engraved on pieces of red ochre from the later Middle Stone Age levels in the Blombos Cave in South Africa (55) (Fig. 2, A and B), together with similar designs incised on fragments of ostrich-eggshell water containers from the site of Diepkloof in the western Cape (56), dated respectively to ∼75,000 and 60,000 yr B.P. (50, 51). At present, the Indian and Sri Lankan sites in question can only be reliably dated back to around 34,000 yr B.P. (in calibrated radiocarbon terms) (34, 39, 4144), but current excavations at the Jwalapuram site in southern India suggest that similar technologies may go back to a much earlier date (42). Unless these striking similarities in material culture are entirely coincidental, they point strongly to a direct connection between the earliest modern human colonists in southern Asia and their probable ancestors in eastern and southern Africa.

Fig. 2.

(A) (Upper) Assemblages of “crescentic” and related stone tool forms from the sites of Patne in western India (41) and Batadomba-lena in Sri Lanka (43), radiocarbon dated to between 30,000 and 34,000 (calibrated) yr B.P. (Lower) Geometric design engraved on ostrich eggshell fragment from Patne (left) and specimens of ostrich eggshell beads (and preforms) and a perforated shell recovered from Patne and Batadomba-lena, respectively. (B) (Upper) Crescentic and related tool forms from the “Howieson's Poort” levels at Klasies River (South Africa) dated to ∼60,000 to 65,000 yr B.P. (48). (Lower) Geometric designs incised on pieces of red ochre from the Blombos Cave, South Africa (∼75,000 yr B.P.) (55) and ostrich eggshell beads from the site of Enkapune ya Muto, Kenya (∼40,000 yr B.P.) (56).

The Australian Archaeological Record

The greatest enigma in the current archaeological record lies in the lack of similarly “advanced” technologies in the areas to the east of the Indian subcontinent, and especially in the relatively well-explored areas of Australia and New Guinea, which were colonized by anatomically modern humans from at least 45,000 yr B.P. onward, as discussed above (2426). The earliest stone-tool technologies documented across the whole of Australasia are conspicuously lacking in any trace of distinctively “modern” or “Upper Palaeolithic,” blade-based technologies of the kind recorded from both the later African Middle Stone Age sites and the earliest modern human sites in southwest Asia and Europe (26, 57, 58). These Australian technologies consist of very simple, flake-based industries, completely lacking in typical blade forms and apparently with little or no trace of typically Upper Palaeolithic tool forms such as end scrapers, backed blades, or burins—a pattern of technology that persisted in Australia from at least 45,000 yr B.P. down to the middle of the Holocene period, around 5000 to 7000 yr B.P. (26, 57, 58). How can we reconcile this observation with the hypothesis that these technologies developed from more “advanced,” blade-based technologies in Africa (and apparently parts of India) with the initial dispersal of anatomically and genetically modern populations in their eastward migration?

The answer to this paradox might lie partly in environmental factors, and partly in the patterns of cultural and technological development that are probably inherent in the progressive dispersal of small-scale human populations across a long and environmentally complex colonization route. Three factors in particular are likely to have been significant in this context. The first, and potentially most important, factor lies in the general scarcity of high-quality, fine-grained stone for tool production in most areas of eastern and southeastern Asia (59, 60). Blade technology in particular is heavily dependent on the availability of nodules of fine-grained stone such as flint, chert, obsidian, or other fine-grained rocks, which are scarce over many areas of eastern Asia. Even if there were occasionally localized areas of better-quality stone, such as obsidian or high-quality cherts, the patterns of lithic technology would inevitably have to “adapt” to the more extensive areas in which these better-quality raw materials were lacking. Indeed, it has often been suggested that the technology over large areas of eastern Asia may have been much more dependent on wood, or even bamboo, for tool manufacture than on conventional stone tool supplies (59, 60).

The second major factor in stone tool technology lies in the specific functions for which the tools were required. If, as most of the current models suggest, the initial colonization of southeastern Asia and Australasia followed a primarily coastal route (12, 18, 20, 21, 61), then the technologies would be likely to adapt primarily to the exploitation of coastal resources, such as fish, shellfish, and marine mammals (together with tropical plant foods) with perhaps only a minor component of hunting larger land mammals, of the kind that clearly formed a major part of the human economy in both Africa and the whole of western Asia and Europe (21, 59). This would presumably have involved much less emphasis on various forms of hunting equipment (such as spears, meat-processing tools, etc.), as well as equipment involved in the manufacture of elaborate skin clothing, or the construction of tents and other living structures, that were essential to survival in much colder, more northerly environments (59, 62).

Finally, there is the more fundamental demographic and evolutionary issue of the repeated founder effects, and associated cultural drift, as relatively small population units expanded progressively eastward along the southern and southeastern Asian coasts (21, 63). These repeated, successive, and cumulative small-scale founder effects would inevitably operate not only on the biological and genetic features of the populations (6, 7, 11) but also on their repertoires of cultural and technological behavior—probably leading to a progressive loss in the complexity and diversity of cultural and technological patterns with increasing distance from their demographic point of origin. Arguably the most striking illustration of this kind of founder effect and technological drift process can be seen in the loss of several technological features (such as fishing, bone tools, and other cultural elements) associated with the settlement and cultural development of Tasmanian populations, following their initial colonization of the island around 35,000 yr B.P. (6466). All of these cultural and technological processes could be seen as a direct parallel to the progressive loss in the genetic diversity of the dispersing modern human populations over geographical trajectories extending from their putative African origin progressively eastward and westward—as recently documented in studies of present-day genetic patterns by Prugnolle et al. (6), Liu et al. (7), and others (Fig. 3).

Fig. 3.

Decline in the genetic diversity of present-day human populations with increasing distance from the presumed point of dispersal of anatomically and genetically modern populations in East Africa. [Reprinted from Prugnolle et al. (6), with permission from Elsevier.]

When viewed in these terms, the relative “simplicity” of the technology associated with the initial modern human settlement of southeast Asia and Australia becomes not merely plausible but arguably largely predictable, in demographic and cultural terms. Although these early Australian technologies are strikingly different from those of the early Upper Palaeolithic, blade-dominated industries in southwestern Asia and Europe, there is nothing distinctively Middle Palaeolithic (or Mode 3) about their character (2931). The whole of the Australian technology is conspicuously lacking in anything resembling typically Middle Palaeolithic “Levallois” or similar “prepared core” techniques (26, 57, 58), and the kinds of simple retouched flake tools encountered in these industries could be paralleled just as easily in some of the early Upper Palaeolithic industries in Europe (67) as in the Eurasian and African Middle Palaeolithic/Middle Stone Age sites. More specifically, it could be argued that the curious, single-platform “horse-hoof” cores that are such a distinctive feature of the earliest Australian industries (such as those from Lake Mungo in New South Wales) (26, 57, 58, 68) have more in common with simplified forms of single-platform blade cores (Fig. 4) in their basic conceptual and flaking strategies than with the distinctively “radial” or “centripetal” patterns of flaking, which are the hallmark of the African and Eurasian Middle Palaeolithic prepared core techniques (59, 69). Any suggestion that the earliest colonists of southeast Asia and Australia carried with them distinctively Middle Palaeolithic, Mode 3 technologies, as suggested in the “multiple dispersals” model (2931, 39), would seem to have little support in the documented archaeological record from Australia. The totally “modern” character of the burial rituals, personal ornaments, abundant use of red ochre, and elaborate ground and shaped stone axes, documented from effectively the earliest stages of colonization of Australia (26, 57, 58, 70), should also be kept in mind in this context. As several authors have pointed out in relation to the 100,000-year-old ritualistic burials from the sites of Skhul and Qafzeh in Israel (17, 7173), there is clearly much more to the emergence of cognitively “modern,” symbolically constructed behavior than the production of typically Upper Palaeolithic stone tools.

Fig. 4.

“Horse hoof” core from Lake Mungo, Australia, dated to ∼40,000 to 45,000 B.P. The basic flaking strategy of these cores is analogous to that of simplified forms of single-platform blade cores (26).

The Colonization of Western Asia and Europe

One important implication of this single-dispersal, “southern route” colonization of Eurasia from eastern Africa is that the modern human colonization of the Near East and Europe must have reached these areas via western or southern Asia, rather than via the Nile Valley and Sinai Peninsula, as implied in the “northern” dispersal route. This conclusion receives further support from the relatively late dating of the arrival of modern humans in Europe (∼40,000 to 45,000 yr B.P.) implied by both the archaeological and DNA evidence (9, 34, 74, 75), as compared to the earlier colonization of southern and southeastern Asia implied by both the DNA evidence and the early archaeological and skeletal evidence from Australia. Some archaeological support for the northern route could perhaps be seen in the presence of early blade industries at sites such as Taramsa in the Nile Valley (76), but these sites are only tentatively dated at present, and the industries show at best only tenuous morphological similarities to those from the 45,000- to 50,000-year-old sites of Boker Tachtit and Ksar Akil in the Near East (32, 34, 76).

The alternative possibility would be to see the origins of the Near Eastern industries as related directly to the single, southern dispersal event, which carried modern humans first across the Bab el Mandeb straits to Arabia, followed by a splitting of the migration routes, one moving eastward toward India, and the other moving northward, either through central Arabia or the eastern coast of the Red Sea into the Near East (8, 13, 14, 16, 21)—with an associated divergence of technological patterns during the course of these dispersals. Alternatively, this split could have occurred rather further to the east in, say, eastern Arabia or Iran (14, 16, 21). The character of the industry from the basal (“Endingi”) levels in the Enkapune ya Muto rock shelter in Kenya, dated to ∼50,000 to 60,000 yr B.P. (comprising elongated Levallois points and blade forms, reminiscent of those in the earliest Near Eastern industries) (54), could perhaps provide some support for this model.

But until the character of the archaeological evidence from both the Nile Valley and Arabia is better documented (40, 76), the precise source of the earliest modern human colonists in the Near Eastern zone will be difficult to pin down. And the evidence for the earlier, unsuccessful dispersal of anatomically modern humans from North Africa into the Near East around 100,000 years ago (as represented by the finds from Skhul and Qafzeh in Israel) must make us aware that not all modern human dispersal events were necessarily successful in either the long or short term, or in terms of their continuity in the subsequent human DNA record (1, 2, 17, 29, 32). Even if the broad outlines of the modern human dispersal from Africa into Eurasia are now becoming much clearer, in terms of both the genetic and archaeological evidence, the finer details of this colonization process have still to be worked out.

References and Notes

View Abstract

Navigate This Article