ARTICLE

Rethinking the Dispersal of Homo sapiensout of AfricaHUW S. GROUCUTT, MICHAEL D. PETRAGLIA, GEOFF BAILEY, ELEANOR M. L. SCERRI, ASH PARTON,LAINE CLARK-BALZAN, RICHARD P. JENNINGS, LAURA LEWIS, JAMES BLINKHORN, NICK A. DRAKE,PAUL S. BREEZE, ROBYN H. INGLIS, MAUD H. DEV�ES, MATTHEW MEREDITH-WILLIAMS, NICOLE BOIVIN,MARK G. THOMAS, AND AYLWYN SCALLY

Current fossil, genetic, and archeological data indicate that Homo sapiens ori-ginated in Africa in the late Middle Pleistocene. By the end of the Late Pleisto-cene, our species was distributed across every continent except Antarctica,setting the foundations for the subsequent demographic and cultural changes ofthe Holocene. The intervening processes remain intensely debated and a keytheme in hominin evolutionary studies. We review archeological, fossil, environ-mental, and genetic data to evaluate the current state of knowledge on the dis-persal of Homo sapiens out of Africa. The emerging picture of the dispersalprocess suggests dynamic behavioral variability, complex interactions betweenpopulations, and an intricate genetic and cultural legacy. This evolutionary andhistorical complexity challenges simple narratives and suggests that hybridmodels and the testing of explicit hypotheses are required to understand theexpansion of Homo sapiens into Eurasia.

A variety of dispersal models(Table 1) address the period betweenthe widely accepted African origin ofHomo sapiens by around 200-150 kaand the arrival of our species at themargins of the Old World, includingAustralia, Siberia, and northwestEurope, by 50-40 ka.1–4 The evolu-tionary, demographic, and culturalprocesses between these milestones

remain unclear, but a variety ofrecent studies add important newdata.

Whereas earlier models focused onassessing the geographical origins ofour species based on fossil data, morerecent approaches seek to combinefossil, genetic, archeological, andpaleoenvironmental data to illumi-nate the nuances of dispersal into

Asia (Table 1). These models empha-size different hypotheses concerningfactors such as when dispersalsbegan, how many occurred and whichroutes were followed. Recent modelshave largely fallen into two broad cat-egories, emphasizing Marine IsotopeStage (MIS) 5 (early onset dispersalmodel) or post-MIS 5 (late dispersalmodel) time frames (Table 1). This,however, is not a rigid dichotomy. Forexample, models proposing an earlyonset to dispersal are consistent withsubsequent post-MIS 5 dispersalshaving also played an important rolein patterns of human diversity.

FOSSIL EVIDENCE

Hominin fossil remains providedthe initial foundations for the Outof Africa model.3 Future fossil dis-coveries in Southern Asia have thepotential to radically transform ourunderstanding of that dispersal. Early

Huw S. Groucutt, Eleanor M. L. Scerri, Michael D. Petraglia, Ash Parton, Laine Clark-Balzan, Richard P. Jennings, Laura Lewis, Nicole BoivinSchool of Archaeology, Research Laboratory for Archaeology and the History of Art, University of Oxford, Oxford, OX1 3QY, United Kingdom.Corresponding author: Huw Groucutt. Email: huw.groucutt@rlaha.ox.ac.ukGeoff Bailey, Robyn H. Inglis, Maud H. Deves, Matthew Meredith-Williams Department of Archaeology, University of York, York, YO1 7EP,United KingdomJames Blinkhorn McDonald Institute for Archaeological Research, University of Cambridge, Downing Street, Cambridge CB2 3DZNick A. Drake, Paul S. Breeze Department of Geography, King’s College London, WC2R 2LS, United KingdomMaud H. Deves Laboratoire Tectonique, Institut de Physique du Globe, Paris, 75252, FranceMark G. Thomas Research Department of Genetics, Evolution and Environment, University College London, London, WC1E 6BT, United KingdomAylwyn Scally Department of Genetics, University of Cambridge, Cambridge, CB2 3EH, United Kingdom

Key words: Paleolithic; Homo sapiens; dispersal; demography; genetic ancestry

VC 2015 Wiley Periodicals, Inc.DOI: 10.1002/evan.21455Published online in Wiley Online Library (wileyonlinelibrary.com).

Evolutionary Anthropology 24:149–164 (2015)

TABLE 1. Summary of Selected Key Models for the Dispersal of Homo sapiens out of Their Place of Origin, Presented in BroadlyChronological Order of Formulationa

Model Inferred timing of dispersal Description

Examples of key

references

Models focusing on Homo sapiens’ originsAsian cradle model Unclear Asia as birthplace of

H. sapiens5

The SW Asian/NE Africancradle model

Unclear SW Asia and NE Africa ascradle in which H. sapiensevolved and from which itsubsequently dispersed

6

Multiregional model Throughout the Pleistocene H. sapiens evolved simultane-ously in several parts of theworld, with species integritymaintained by recurrentgene flow

7

Recent African Origin (RAO)model

�100-40 ka H. sapiens evolved in Africa,perhaps in one region suchas East Africa, andsubsequently dispersed

3,4

RAO and hybridization model �100-40 ka Accepts RAO, but infersgreater levels of hybridiza-tion with other homininspecies in Eurasia

8–10

Assimilation model �100-40 ka H. sapiens evolved in Africa,but subsequent spreadrepresents gene flow ratherthan replacement

11,12

Variants of Late Dispersal ModelMultiple dispersals model MIS 4 (�70 ka)

then MIS 3 (�50 ka)MIS 4 dispersal by southern

route to Australia, then MIS3 dispersal of populations bynorthern route

13

Upper Paleolithic model �50-45 ka Successful out of Africaoccurred after 50 ka, withderived technology such asprojectiles

14,15

MIS 4 single coastal dispersalmodel

�75-60 ka Structure of mtDNA tree inter-preted as indicating disper-sal around MIS 4

16

Single coastal dispersal withgeometric technologymodel

�60-50 ka A single dispersal out of Africafollowed a coastal route,marked by a trail of geo-metric technologies andsymbolic artifacts

17

Variants of Early Onset Dispersal ModelEarly onset multiple dispersal

modelBeginning in MIS 5 (�130-75 ka),

also MIS 3 (�55-45 ka) keyMultiple dispersals out of

Africa, associated withclimatic “windows ofopportunity.” Early dispersalsassociated with MPtechnology

2,18

“Jebel Faya” model �130 ka Dispersal out of Africa acrosssouthern Arabia withLevallois, blade. and bifacialtechnologies in MIS 5e

19

“Nubian” model by �106 ka Dispersal out of Africa by MIS5c marked by presence of“beaked” (“Nubian”) Leval-lois technology in Arabia

20

aKey references may considerably postdate the initial formulation of particular models.

150 Groucutt et al. ARTICLE

Homo sapiens was morphologicallyvariable.4,21 Traits that characterizeHomo sapiens include: neurocranialglobularity; a divided supraorbitaltorus and central and lateral portions;the face retreated below the forepartof the brain; a bony chin, even ininfants; a gracile tympanic bone; theabsence of an iliac pillar; and a short,thickened superior pubic ramus.3 Yeteven at a single site, morphologicalvariability can be striking. Omo-Kibish 1, for instance, stronglyexpresses the derived features ofHomo sapiens, whereas Omo-Kibish2, which is believed to be of similarage, is much more archaic.3 Givensmall samples, it is difficult to tell ifthe variation of early Homo sapiensrepresents intrapopulation variationor the existence of several highlystructured populations by the laterMiddle Pleistocene. Nevertheless, thefossil record is most parsimoniouslyinterpreted as demonstrating thepiecemeal development of Homo sapi-ens in Africa during the later MiddlePleistocene.

The earliest known Homo sapiensfossils from outside Africa are foundin the Levant, one of the few relativelyintensively studied areas in Asia, atthe sites of Skhul (�120-90 ka) andQafzeh (�100-90 ka).3 These fossilsdisplay numerous derived traits, witha small number of primitive (archaic)features. Subsequently, Homo sapiensare present in the Levant from around43 or 42 ka,22 and perhaps at around55 ka (75.2–33.6 ka) at Manot Cave.23

In the latter case, however, the ageestimates come from a calcitic patinaor crust that covers the calvaria;hence, these are minimum dates. Thespecimen may be considerably olderand/or may not reflect dispersal fromAfrica. Stalagmites from the site dem-onstrate a hiatus in speleothem for-mation between late MIS 5 and MIS3. Neanderthal fossils have been dis-covered in the Levant and elsewherein Asia dating to �70-50 ka.24,25 Fossildata are consistent with archeologicaldiscoveries in suggesting that in theLate Pleistocene, reliably dated Nean-derthals are present in the Levantonly after MIS 5, when Homo sapiens,possibly aside from the Manot Cavecranium, appear to be absent.21 Thisapparently asynchronous timing may

suggest that interbreeding betweenthe species took place elsewhere, thatsmall populations of Homo sapienssurvived into MIS 4 in the Levant, orthat Homo sapiens reoccupying theLevant in MIS 3 encountered lateNeanderthals.

Vast areas of Asia have yet to pro-duce any Pleistocene hominin fos-sils.26,27 Sites further east face datingproblems and taxonomic ambiguitiesassociated with elements such asteeth and foot bones that are notstrongly diagnostic of species.25,28,29

While the relatively well understoodLevantine record may provide a nullhypothesis for demographic changeacross a wider area of SouthwestAsia,24 this must be qualified by the

atypical ecological features of theLevant as a Mediterranean biome ina region more widely characterized

by the particularities of the muchlarger Saharo-Arabian biome.

Dennell has reviewed the fossilrecord for Homo sapiens between Ara-bia and Australia.25 The oldest South

Asian fossils found, from the Sri Lan-kan cave of Fa Hien, date to �33-30ka. In Southeast Asia, the oldest fos-sils are from the cave of Tam Pa Ling

in Laos and date to � 65-45 ka.29

This age admits the possibility thatHomo sapiens either left Africa earlierthan the Upper Paleolithic model sug-

gests or that dispersal was extremelyrapid, as hypothesized by coastal dis-persal models (Table 1). Several newbut preliminary findings suggest that

Homo sapiens may have arrived inSoutheast Asia earlier than was previ-ously thought. The site of Callao Cavein the Philippines has produced a

hominin metatarsal dating to �67 ka,

which is provisionally assigned toHomo sapiens.28 Several sites inChina are claimed to demonstrate thepresence of Homo sapiens by MIS 5.However, these sites, which oftenhave poor stratigraphic and chrono-logical control, have produced taxo-nomically ambiguous specimens.25

GENETIC EVIDENCE

The first reconstructions of humangenetic ancestry were based on datafrom mitochondrial DNA (mtDNA),chromosome Y, and a small numberof nuclear loci.30–32 Much of whatthey revealed, such as evidence of arecent African origin of Homo sapi-ens, remains central to our under-standing today. However, recentdevelopments have changed thenature of the genetic evidence forhuman evolution and dramaticallyincreased its scope. New sequencingtechnologies and computationalapproaches have enabled large-scalewhole-genome analyses of humanpopulations, while the ability torecover ancient DNA sequences fromfossils has extended our view ofgenetic diversity by tens of millenniainto the past.

These developments have led toimportant revisions in how we under-stand the ancestry of humans andother hominins.33 For example, it isclear that this ancestry involves amuch greater degree of demographiccomplexity than could previously beresolved, with evidence of pervasivegene flow and admixture between pop-ulations.9,10,34,35 The relationshipbetween genetic ancestry and demo-graphic history is less straightforwardthan it was often assumed to be, andrequires more sophisticated inferentialapproaches. Inferences based on thegenealogy of a single genetic locus,such as the mtDNA tree, can be prob-lematic, particularly for older events.Such a genealogy represents one ran-dom outcome of the genealogical pro-cess, the shape of which is only weaklyconstrained by demography. More-over, while a simple tree is inadequateto describe the complexity of humanancestral demography, genealogiesare always strictly tree-like.

The timing of the dispersal ofHomo sapiens out of Africa is a case

the fossil record is mostparsimoniously inter-preted as demonstratingthe piecemeal develop-ment of Homo sapiens inAfrica during the laterMiddle Pleistocene.

ARTICLE Rethinking the Dispersal of Homo sapiens out of Africa 151

Box 1. The mtDNA Genealogy and the Chronology of Dispersal out of Africa

Human mtDNA exhibits a geneal-ogy in which all haplotypes (uniqueDNA sequences) in present-day non-Africans are placed within a clade orhaplogroup, the most recent com-mon ancestor (MRCA) of which hasbeen dated to �79 to 60 ka.36 Thishaplogroup, named L3, encompassesseveral others that are found in mod-ern Africans, predominantly in East

Africa.93 This has led to the argu-ment that the MRCA of L3 is anupper boundary on the exit fromAfrica, and that the mtDNA geneal-ogy is incompatible with an earlierpresence outside Africa for theancestors of present-day humans.

This argument, however, restslargely on the assumption thathuman demographic history has

been tree-like (Fig. 1A).13 If,instead, we allow for gene flowbetween the ancestral African andnon-African populations after anearlier initial divergence (Fig, 1B,C), then it is possible for L3 tohave arisen during this period inone or the other population andstill be found in both populationstoday. Indeed, evidence for ongoing

Box Figure 1. Alternative models of the relationship between the mtDNA genealogy and demographic history. Models B and Cillustrate the possibility of an early divergence of African and non-African ancestors to, with subsequent gene flow, potentiallycongruent with fossil and archeological evidence of a dispersal of Homo sapiens out of Africa ca. 100 ka. [Color figure can beviewed in the online issue, which is available at wileyonlinelibrary.com.]

TABLE 2. Demographic parameters and ms commands used for coalescent simulation, Representing exponential growthstarting 15 ka in all three populations from an ancestral Ne of 1,500 to a present-day Ne of 25,000 (corresponding to auto-somal Ne growth from 6,000 to 100,000, assuming equal male and female Ne); African ancestral Ne of 2,500 100ka; non-African Ne of 500 from 55–40 ka in model A and 80–40 ka in models B and C; for model B, migration of 5 individuals per

generation out of Africa 60–50 ka; for model C, migration of 1.2 individuals per generation into Africa 100–50 ka. Relativelikelihood is the proportion of simulations for which the age of the L3 node was less than or equal to 80 ka, relative to this

value for model A.

Model Relative likelihood

A Late dispersal from Africa 55 kams 600 20000 -t 1.25 -I 3 200 200 200 -T -eN 0 10 -eG 0 56.27 -eG 0.05 0 -ej 0.1167 3 2 -en 0.1333 2 0.2 -ej 0.1833 2 1 -en 0.3333 1 1 45369 44223 59953

1.0

B Early dispersal 120 ka with gene flow from Africa ca 55 kams 600 20000 -t 1.25 -I 3 200 200 200 -T -eN 0 10 -eG 0 56.27 -eG 0.05 0 -ej 0.1167 3 2 -en 0.1333 2 0.2 -em 0.1667 2 1 100 -em 0.2 2 1 0 -en 0.2667 2 0.32 -en 0.3333 1 1 -ej 0.42 1 45539 39872 63447

0.6

C Early dispersal 120 ka with subsequent gene flow back into Africams 600 20000 -t 1.25 -I 3 200 200 200 -T -eN 0 10 -eG 0 56.27 -eG 0.05 0 -ej 0.1167 3 2 -en 0.1333 2 0.2 -em 0.1667 1 2 2 -en 0.2667 2 0.32 -em 0.3333 1 2 0 -en 0.3333 1 1 -ej0.4 2 1 4984 41383 33507

1.3

152 Groucutt et al. ARTICLE

in point (Box 1). It has been arguedthat the chronology and spatial dis-tribution of branches of the human

mtDNA and Y chromosomal treesare inconsistent with dispersal anyearlier than �60-50 ka.17,36,37 How-

ever, as Box 1 shows, this argument,which assumes straightforward cor-respondence between genealogical

gene flow has been inferred ingenome-wide analyses.39,94

It has been argued that suchalternative models can be dis-counted on the grounds that if theywere true some non-L3 haplotypeswould be found in present-daynon-Africans.17 However, model Binvolves the fixation of migrant lin-eages from Africa within the ances-tral non-African populationfollowing secondary gene flow.This occurs in mtDNA, but not nec-essarily at autosomal loci due totheir much higher effective popula-tion size (Ne), and reflects the factthat mtDNA genealogies are poten-tially more susceptible to suchmigration and introgression events.Model C involves a recent coales-cence of mtDNA lineages outsideAfrica combined with migration ofone or more lineages back toAfrica. Both of these possibilitiesare made more likely by a lowancestral non-African Ne. Further,whole-genome analyses have shownthat the ancestors of present-daynon-Africans experienced a pro-found reduction in autosomal Ne

to below 3,000 for much of theperiod between MIS 5 and theHolocene.43,46

To explore these possibilities, wesimulated examples of each of themodels in Box Fig. 1 using a coales-cent approach and investigated theresulting distributions of L3 nodeage. Simulated mtDNA genealogieswere generated for three popula-tions representing present-day Afri-can, European, and Asian samples,with the European and Asian popu-lations diverging at 35 ka, and theinitial African/non-African splitoccurring at 55 ka in model A and120 ka in models B and C. Formodel B, we simulated a shortperiod of strong migration (5 indi-viduals per generation) out ofAfrica from 60–50 ka, while modelC featured an extended period ofweaker back-migration (1.2 individ-uals per generation) into Africafrom 100–50 ka. Full demographicparameters and simulation com-mands are listed in Table 2. Foreach model, we generated 20,000simulations of 600 samples usingms95; increasing the number ofsamples had a negligible effect onresults since the vast majority ofcoalescent events occur recently.Conservatively, we took a present-day tgen of 30 yr as valid for thewhole of human prehistory.96

In each simulation, the coales-cent tree was inspected and theyoungest node ancestral to all non-Africans and some Africans wasidentified as the node equivalent toL3. (Note that there will always besuch a node in any genealogy, butunder an arbitrary or unstructureddemography it will often coincidewith the global root.) Box Fig. 2shows distributions of the age ofthis node for each model. Relativeempirical likelihoods of L3�80 kaare given in Table 2.

The models simulated representplausible demographic historiesbased on genetic and other evi-dence for human prehistory. Theydemonstrate the range of variationin the age of the L3 node that onemight expect under similar scenar-ios. However, they are not formallyfitted to the observed humanmtDNA genealogy; indeed, a valueof 80 ka or less is unlikely under allthese models (only 9% of simula-tions under Model A). It would bepossible to infer parameters maxi-mizing this likelihood, but infer-ence based on one node in a singlegenealogy would not be robust.

Nevertheless, these results showthat alternative models can be con-structed involving an early disper-sal out of Africa for which a recentage for L3 is not substantially lesslikely than under an equivalentmodel of late dispersal. Withoutfurther evidence of the geographi-cal extent and structure of humanpopulations during this period, anabsence of non-L3 haplotypes out-side Africa today cannot beregarded as conclusive evidenceagainst a dispersal of Homo sapiensout of Africa beginning by MIS 5.It is also worth noting that a morerecent analysis incorporatingancient DNA estimated an age of95-62 ka for L3,43 representing asubstantial overlap with MIS 5.

Box Figure 2. Distribution of the age of the L3 node in 20,000 simulations for each ofthe models listed in Table 2. The vertical dashed line at 80 Ka indicates the maximumestimated age of L3 in the human mtDNA tree. [Color figure can be viewed in theonline issue, which is available at wileyonlinelibrary.com.]

ARTICLE Rethinking the Dispersal of Homo sapiens out of Africa 153

trees and demographic history, isnot valid under plausible alternativemodels of divergence, with gene flowbetween subpopulations over tens ofmillennia. Support for such modelscomes in part from recent evidencethat much of the population struc-ture in Africa is of surprisinglyancient provenance.38 Further sup-port comes from genome-wide infer-ences of a gradual divergence withongoing gene flow between Africanand non-African ancestors duringMIS 5.39 Indeed, with a revised esti-mate of 0.5 x 109 bp21 y21 for thenuclear genomic mutation rate,whole-genome demographic studiesfavor an older time scale and a morecomplex process of dispersal out ofAfrica.40 Evidence of this rate hascome primarily from sequencingstudies of de novo mutations. Someconcerns have been raised aboutthe influence of false negatives onsuch data.41 However, not onlyhave more than a dozen such stud-ies now arrived at similarly low val-ues,42 but independent evidencehas also come from comparingancient and modern human DNA.43

The lower rate is more consistentwith inferences for the timing ofrecent events such as the diver-gence of Native American and EastAsian populations.39

These considerations are not to dis-pute the continuing value of mtDNAas a source of information on humanevolution, particularly for more recentevents. It is still more widely sampledthan are the autosomes (chromo-somes 1–22) and more amenable toancient DNA studies. mtDNA also hasa smaller effective population size (Ne)(around one-quarter of the mean auto-somal value, depending on certaindemographic factors), meaning thatpatterns of diversity in mtDNAsequences respond more rapidly todemographic changes. Thus, mtDNAtrees can be informative about morerecent demographic history wherethere are numerous uncoalescedbranches in the tree, albeit with thecaveats previously mentioned. Forexample, the “star-like” topology ofnon-African branches of the mtDNAand Y chromosome trees around 50ka44,45 suggests an acutely reducednon-African Ne at this time, as well as

rapid population growth following it,which may or may not correspond to amajor migration event such as a popu-lation dispersal. This topology alsocoincides with the Ne minimum of thenon-African bottleneck inferred fromwhole-genome analysis.46 Future anal-yses combining widely sampledmtDNA, Y-chromosomal, and genome-wide data should provide a morepowerful means of inferring recentdemographic processes.

Returning to earlier events, theemerging picture is one in which byMIS 5 Homo sapiens existed withinvarious subpopulations, differing intheir size and degree of genetic con-tact, and perhaps distributed over awide area. At least one of theseincluded the ancestors of present-day non-Africans and was character-ized by low Ne. Based solely onpresent-day genetic data, which hasweak geographical resolution whenlooking distantly into the past, it isdifficult to say how far one or moreof these populations might haveextended into Asia before �60 ka. Amajor reason for this is the preva-lence of subsequent migration andgene flow, not only within Eurasia,but also from Eurasia back intoAfrica and within Africa.47,48 Suchevents weaken the correlationbetween present and ancestral haplo-type distributions and depend onecological and environmental factorsthat are challenging to model.

It may therefore be that some ofthese questions can be resolved onlyby a combination of archeology andancient DNA sequencing.49 AncientDNA has already been transformativein revealing interbreeding betweenHomo sapiens and other hominins,including Neanderthals, Denisovans,and perhaps other archaic popula-tions.9,10,49 In particular, one episodeof Homo sapiens-Neanderthal inter-breeding has now been dated to 60-50ka, based on the clear signature it leftin the ancient genomes of MIS 3 indi-viduals from Siberia.35,43 This hasimplications for the question ofwhen Homo sapiens left Africa.For example, if it represents theearliest episode of interbreeding,and if we expect interbreeding tohave begun as soon as humans leftAfrica, this finding would seem to

cast doubt on the hypothesis of anearlier human exodus. This findingalso suggest that genetic signalsof population divergence before60 ka relate to substructure withinAfrica. However, there are alterna-tive possibilities:

(1) Since the method used to dateintrogression assumed only a singleepisode of gene flow, earlier episodesmay have been undetected. (2) Nean-derthals may have ranged farthernorth before �70-50 ka, so that Homosapiens encountered them only some-time after leaving Africa as part of asecondary Homo sapiens migration orNeanderthal expansion southward.(3) The vast majority of post-60-kahuman lineages may descend from asecond wave of Homo sapiens out ofAfrica at that time (model B in Box 1),in which case earlier episodes ofinterbreeding may have left a mini-mal genetic legacy. (4) Contactbetween Homo sapiens and Neander-thals may have occurred earlier butunproductively, perhaps due to lowhybrid viability or fertility, or otherreproductive obstacles.50

ARCHEOLOGICAL EVIDENCE

Several dispersal models claimsupport from patterns in archeologi-cal and, particularly, lithic data(Figs. 1 and 2; Table 1). In evaluatingthe evidence underlying these claims,it must be recognized that multipleprocesses, in this case branching(cultural inheritance and spread),blending (cultural diffusion betweenpopulations), and convergence (inde-pendent reinvention), can producesimilar forms of material culture(equifinality). The first two of theseprocesses can alternatively bedescribed as homology and the latteras analogy.

There are many examples of con-vergence in lithic technology. Partic-ular care must be taken when likelydrivers of independent reinvention,such as the constraints of hafting,exist. In other ways, however, arche-ological data can provide robust sig-nals of dispersal. A key examplerelates to Australia, which, in con-trast to Asia, with its multiple homi-nin species, was peopled, as far as isknown, only by Homo sapiens.

154 Groucutt et al. ARTICLE

Available data suggest that Homosapiens had reached Australia before50 ka.51 The Australian archeologicalrecord provides a key minimum agefor dispersal out of Africa.

The Evidence of MIS 3 Dispersals

Various models cite archeologicaldata as indicating the dispersal ofHomo sapiens into Asia �70-40 ka.

Most prominently, the appearance ofnew lithic technology and otheraspects of material culture tradition-ally described as “Upper Paleolithic”in the Levant from �47/45 ka has

Figure 1. Selected lithics (stone tools) from East and North Africa for MIS 5 (1–11) and MIS 3 (12–25). 1–4: iconic MIS 5 Middle Paleolithic(Middle Stone Age) lithic types and techniques of North Africa; 1–2: tanged or pedunculated Aterian points, widely thought to havebeen hafted tools68; 3, 4: “beaked” (“Nubian”) Levallois cores62; 5–11: other common components of North (5–7, from Bir Tarfawi,Egypt)64 and East (8–11, from BNS, Omo Kibish, Ethiopia)63; African MIS 5 MP assemblages; 5, 8, 9: recurrent centripetal Levallois cores;6, 10: centripetally prepared Levallois flakes; 7, 11: retouched points. Late MP cores (12, 13) and retouched points (14, 15) and backedmicrolithic (16) from Mochena Borago, Ethiopia, �50 ka.55 17–25: Early Late Paleolithic (Later Stone Age) lithics from Enkapune YaMuto, Kenya � 50-40 ka; 17: end and side retouched flake; 18–24: backed flakes or microliths; 25: burin.103

ARTICLE Rethinking the Dispersal of Homo sapiens out of Africa 155

Figure 2. Selected lithics from Southwest and South Asia from MIS 5 (1–12) and MIS 3 (13–27). 1–4: Arabian Peninsula. 1: centripetally pre-pared preferential Levallois core, Jebel Faya, UAE, �125 ka; 2: bifacially flaked tool, Jebel Faya104; 3,4: beaked (or Nubian) Levalloiscores from TH-59, Oman, probably MIS 520; 5–8: Qafzeh Cave, Israel, �100-90 ka; 5: recurrent centripetal Levallois core; 6: centripetallyprepared preferential Levallois core; 7: Levallois flake; 8: side retouched Levallois flake.65 9–12: MIS 5 lithics from Jwalapurum 22, India,�75 ka. 9: recurrent centripetal Levallois core; 10: centripetally prepared preferential Levallois core; 11: tanged or pedunculated flake;12: retouched point66; 13–16, typical artifacts of the Levantine Late Middle Paleolithic, Dederiyeh Cave, Syria, �60 ka. 13,14: unidirec-tional convergent Levallois cores; 15,16: Levallois points with unidirectional convergent preparation.105 17–20, Late Middle Paleolithiclithics from Jwalapurum 3 and 20, India, �55-30 ka, 17: centripetally prepared preferential Levallois core; 18: recurrent centripetal Leval-lois core; 19, 20: Levallois flakes.66 21–27: Early Upper Paleolithic lithics from the Levant, � 40 ka. 21: blade core; 22–26: points; 27:blade.15

156 Groucutt et al. ARTICLE

been seen as evidence of the rearrivalof Homo sapiens from Africa (Fig. 2:21–27).15 Key aspects of this includethe hypothesis that this dispersalreflects the invention of projectiletechnology in Africa.15 Such develop-ments perhaps gave Homo sapiens aselective advantage over Neanderthalpopulations in Eurasia. However, theextent to which this model can begeneralized beyond the Levant iscurrently unclear.

Microlithic or geometric technolo-gies were variably present through-out the African Middle Paleolithic(MP). (Here we use “Middle Paleo-lithic” to include the synonymousMiddle Stone Age and “Late Paleo-lithic” as a way of describingassemblages traditionally designatedas Upper Paleolithic, Later StoneAge, or Microlithic). Mellars and col-leagues17 argue for dispersal intosouthern Asia by a coastal routebefore the origin of the Upper Paleo-lithic in the Levant. They cite the dis-tribution of microlithic andgeometric technologies around theIndian Ocean rim as providing evi-dence of a single dispersal of Homosapiens out of Africa �55-50 ka (forexample, Fig 1: 17–25). The Howie-sons Poort (HP), emphasized by Mel-lars and colleagues,17 is an earlyexample of technologies commonlydescribed as microlithic or geometricthat become temporarily a commonpart of assemblages. However, recentstudies, emphasize the diversity ofHP assemblages, for example in corereduction methods and in features ofretouched tools.52–54

In East Africa, microliths occur inlow frequencies from around 50 kaand subsequently increase in fre-quency.55 For example, a single com-plete crescent was found at MochenaBorago (Fig. 1: 16) within a patternof general technological continuity.Earlier claims face taphonomicand chronological problems.52 Theonset of the Late Paleolithic inEast Africa appears to have beeninitially characterized by a reduc-tion in the size of lithics and anincrease in bipolar reduction. InSouth Asia, there also appears to bea gradual and complex transitionfrom the Middle to Late Paleo-lithic.56–61 The early phase of the

Late Paleolithic in South Asia wasdominated by blade and microbladeproduction, with microlithic tech-nology becoming widespread from�38/35 ka. The oldest microlithicsite in Sri Lanka, Fa Hien-lena, dat-ing to �38 ka, contains only nongeo-metric forms.59

While these findings do not fit eas-ily with the notion of microlithictechnology as a unique marker ofthe dispersal out of Africa, Mellarsand colleagues emphasize the similarrange of shapes, such as crescents,lunates, and trapezoids, of some sub-sequent South Asian microliths andthe earlier African forms.17 To Mel-lars and colleagues, these similaritiesin shape are best explained bybranching and blending processes ofcultural interaction and dispersal.

Advocates of the notion of dispersalfrom Africa at around 60-50 ka alsocite the apparent distribution of“symbolic” artifacts such as beads andincised ostrich eggshell.17 Cited exam-ples include Batadomba-lena at �35 kaand Jwalapurum 9 at �20-12 ka.58,59

Mellars and colleagues argue that olderexamples, as well as early examples ofmicrolithic technology, have been con-cealed by sea-level rise.17

MIS 5 DISPERSALS

Archeological findings in southernAsia have been interpreted as indi-cating early dispersals of Homo sapi-ens out of Africa (that is, by MIS 5,�130-75 ka) (Table 1). While theMiddle Paleolithic in Europe is asso-ciated with Neanderthals, in Africa itnot only temporally overlaps withmost of the period in which Homosapiens was present, but also charac-terizes the early expansion of Homosapiens to the Levant.62–65 Given thatavailable data indicate that humanswere in Southeast Asia and Australiabefore the origin of the Late Paleo-lithic in Africa and Asia, there is astrong indication that at least anearly phase of dispersal out of Africawas associated with MP lithictechnology.66,67 Elucidating variabili-ty within MP technologies is there-fore of great importance forunderstanding dispersal.68

One recent model emphasizes thecombination of Levallois, blade, and

faconnage reduction methods foundat the Arabian site of Jebel Faya andclaims that these are similar to fea-tures of the East and Northeast Afri-can Middle Paleolithic (Fig 2: 1–2).19

An alternative model instead stressesthe discovery of “beaked” (or“Nubian”) Levallois technology, previ-ously best known from NortheastAfrica (Fig. 1: 3–4), in Arabia (Fig 2: 3–4).20 The former emphasizes a ratherbroad combination of features; the lat-ter highlights one aspect of technologythat may represent convergent (inde-pendent) evolution. The notion thatthe “Nubian complex” is a spatiallyand temporally restricted technocom-plex becomes problematic with thediscovery of similar technologies fromMauritania to the Thar Desert viaSouth Africa.67 The distinctive“beaked” or “Nubian” cores were firstdescribed by Seligman,62 who thoughtthat the shape of the median-distalridge was similar to the shape of atortoise beak. We propose use of themorphologically descriptive term“beaked” instead of “Nubian,” whichimplies an automatic association ofthis technology with Nubia; emerg-ing evidence may be consistent withthe convergent reinvention of thistechnology.

An alternative lithic Out-of-Africasignal may be the spread of East Afri-can technologies, with a trail of simi-lar assemblages linking East Africa,63

Northeast Africa,64 the Levant,65 andas far east as India by late MIS 5 (Fig.1: 5–11, Fig. 2: 5–12).51,66 The searchfor an archeological “smoking gun”for dispersal out of Africa is chal-lenged by the diversity of lithic tech-nology within Africa before dispersal.Scerri and colleagues, for instance,have demonstrated spatially struc-tured lithic variability in MIS 5 NorthAfrica,68 correlating with modeledecozones rather than traditional“industrial” nomenclatures such as“Aterian” and “Nubian complex.” Thisfinding is interpreted as indicatingstructured (subdivided) populationsby MIS 5. Such inferences representan important archeological findingthat can be factored into models usinggenetic data.

The South Asian MP shows consid-erable technological continuity fromlater MIS 5 through MIS 3 (Fig. 2: 9–

ARTICLE Rethinking the Dispersal of Homo sapiens out of Africa 157

12, 17–20),66,67 suggesting that homi-nin population continuity in themosaic environments of South Asiaoccurred through the Toba super-eruption of 75 ka. It is possible, how-ever, that there was also an earlier insitu Lower to MP transition,although this possibility requires fur-ther analysis. South Asian MPassemblages dating to MIS 5 featurebeaked (“Nubian”) Levallois technol-ogy and other components commonin the African record,66,67 but lackthe kind of technology associatedwith Neanderthals in at least theLevant (Fig. 2: 13–16) and centralAsia.52 Key reviews of Asian paleoan-thropology include those by Den-nell,69 who focused on the pre-MIS 5period, and Rabett,70 who addressesthe post-MIS 5 period.

In the case of the Levant, MP tech-nology is found alongside variousindications of complex behavior,including deliberate burials, beads,and the use of ochre, features thatarguably articulate the record moreclosely with that of Africa thanEurope.24,71 Some of the earliestexamples of such behavior, whichare actually found at the non-Africansite of Skhul (�130-100 ka),71 andmore robustly at the slightly youngersite of Qafzeh (�100-90 ka), suggestthat by at least MIS 5 Homo sapienswas capable of complex, includingsymbolic, behavior.72,73 Such be-havior was expressed in a variablemanner, perhaps in relation toenvironmental or demographic fac-tors.74 The current lack of evidenceof symbolic behaviors in areas suchas Arabia may reflect factors suchas lack of research in a given areaor the fact that most sites discov-ered consist of raw material pro-curement and early stage reductionlocalities.

ENVIRONMENTS ANDDISPERSAL ROUTES

The final major component ofdebate over the dispersal of Homosapiens out of Africa concerns theroutes taken and how they correlatemore widely with ecological condi-tions. Some have emphasized terres-trial dispersal routes.27 For example,combining archeological and envi-

ronmental data for North Africa hasshown that MP sites tend to be mosttechnologically similar to nearbysites except where they were con-nected by rivers (see Box 2).68 Analternative perspective hypothesizesthat coastal routes were key.17

Environmental variation influencesdispersal not only in terms of itseffects on factors such as net primaryproductivity, which leads to demo-graphic fluctuation, but also by open-ing and closing routes, such asthrough the generally arid Saharo-Arabian zone.75,76 The history ofresearch cautions against directly cor-relating environmental and demo-graphic processes. For example,paleoclimate data from Lake Malawisuggesting that “mega-droughts”occurred in sub-Saharan Africa hasbeen cited as a key mechanismexplaining the apparent dispersal outof Africa at around 60 ka.17 However,subsequent research has demon-strated that the MIS 5 mega-droughtended much earlier, by �85 ka.77

While numerous archives in theSaharo-Arabian belt attest to dra-matic increases in humidity duringperiods such as MIS 5,75,76 early MIS3 (�60-50 ka) also witnessed a signif-icant humid phase in Arabia.78 ThisMIS 3 wet period may have providedthe context for a renewed phase ofdispersal out of Africa and/or theexpansion of refugial populationsalready within Arabia. The evidenceof interior humidity and homininoccupations some distance from thecoast in Arabia in MIS 3 suggeststhat a coastal route need not havebeen exclusively followed (Box 2). Italso suggests that terrestrial disper-sals need not have been limited toMIS 5.

Coastal Dispersal Models

Recent years have witnessed inten-sified debate about the role of coast-lines in hominin evolution anddispersal.17,27,79–83 At one extreme,proponents of purely coastal routessee those routes as providing amechanism for fast, directional, pop-ulation expansion along an ecologi-cally uniform coastal highway.17

According to this view, MIS 3 coastalregions were highly stable and pro-

ductive environments. Field data,however, remain minimal.

South African sites demonstratethe use of molluscs, fish, and seamammals exploited from the sea-shore alongside terrestrial foods, butthey lack evidence of offshore activ-ity.83 Most of these findings relate toMIS 5/4, but earlier examples maybe concealed by sea level rise. Theonly other reported example of possi-ble early coastal subsistence outsideSouth Africa is the MIS 5 Eritreansite of Abdur,79 where shells fromshellfish apparently collected forfood have been reported alongsidestone tools and terrestrial mammalsin a beach deposit. However, withoutfurther supporting details or evi-dence of other marine resource use,Abdur’s place in the record of Pleis-tocene coastal exploitation isunclear.84 There is nothing in thesefindings to indicate that marineresources were uniquely associatedwith anatomically modern humansor that they supported marine-focused paleoeconomies. Other hom-inins, and even nonhominin prima-tes, also exploited marine foods onthe seashore.85

Only in Sahul is there clear evidenceof the conjunction of colonization,seafaring, and heavy dependence onmarine resources during the Pleisto-cene. Occurring only in the archipel-ago environments of Wallacea and theBismarck islands, this may reflect aunique combination of circumstances:abundant bamboo driftwood for rafts;an archipelago environment withfavorable winds and currents whereland is rarely out of sight; a depauper-ate island fauna and little available ter-restrial food; a rich and varied supplyof marine resources; and upliftedcoastlines preserving caves occupiedduring low sea level above the presentshoreline.86–89 Once humans reachedlandfall in Sahul, they rapidly movedinland, leaving little evidence of pro-longed coastal settlement even on thetectonically uplifted coastlines ofnorthern New Guinea.

It is possible that analogous condi-tions existed on paleoshorelines else-where around the Indian Ocean, butthat these have been submerged bysea level rise.17 Testing this hypothe-sis requires surveys both underwater

158 Groucutt et al. ARTICLE

and on land proximal to coast-lines.81,89 A broad delineation of thecontinental shelf around the IndianOcean approximating a sea levelposition of 2100 m at the maximum

Late Pleistocene regression high-lights the considerable variability inthe width and topography of thecoastal shelf, and therefore in thearea and nature of Pleistocene land-

scapes that have been submerged(Fig. 3). Intensive surveys in someareas proximal to narrow coastalshelves, such as the Dhofar coast ofOman (Fig. 3D), have failed to reveal

Box 2. Paleoclimate, Paleohydrology and the Distribution of Middle Paleolithic Sites

The distribution of MP sites in theSaharo-Arabian belt providesinsights into Late Pleistocene land-scape use, given that at least theearly phase of dispersal out of Africawas associated with MP toolkits. As

shown in Box Fig. 1, vast areas ofthe generally arid Saharo-Arabianbelt were potentially transformedinto grassland and savanna environ-ments by increased rainfall duringseveral humid periods, includingthose of MIS 5. Box Fig. 1 combinespaleohydrological data63,97–99 with

modeled 130 ka precipitation dataderived from the down-scaled Com-munity Systems Model (CCSM 3)data.100,101 Box Fig. 1 shows thatmost MP sites in this region occurin areas of increased rainfall; how-

ever, it also demonstrates the per-sistence of an arid desert belt evenduring times such as MIS 5e. Paleo-rivers formed potentially crucialcorridors (and filters) through thesearid environments, as demonstratedfor the Sahara during MIS 5e.68,102

In Arabia, paleohydrological model-

ing has shown that the number ofrecorded MPs declines considerablyas distance from major riversincreases, with 74% of MP siteswithin 30 km of large paleodrainagesystems.99 This is ongoing research,

but initial results are congruentwith hypotheses that fluvial net-works formed important dispersalcorridors, as Arabian MP sites aregenerally located much closer tomajor paleorivers than would beexpected if they were randomlydistributed.99

Box Figure 1. The distribution of Middle Paleolithic sites across East Africa, the Saharo-Arabian belt, and India, plotted on a mod-eled precipitation map for the last interglacial (MIS 5) with positions of major paleolakes (dark blue areas) and paleorivers, whichform extensive riparian corridors (blue lines). The Neanderthal range line shows the estimated extent of Neanderthal dispersalfrom the north. The map shows that Middle Paleolithic sites are commonly located in interior regions and that their presence intypically arid areas can be explained by the humid climate conditions of periods such as MIS 5, which activated paleohydrologi-cal networks and potentially transformed major deserts into savannah grasslands and shrublands (green areas) containing numer-ous freshwater lakes and rivers. The paucity of sites in Pakistan and eastern Iran almost certainly reflects research history ratherthan a real pattern. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

F3

ARTICLE Rethinking the Dispersal of Homo sapiens out of Africa 159

Late Pleistocene marine-focusedarcheology.20 The presence of LowerPaleolithic and Neolithic materialindicates that this absence is notsimply a question of preservation.Parts of the coastal margin of east-ern Arabia have experienced tectonicuplift, including during the Pleisto-cene,90 but no archeological evidenceconsistent with the expectations ofpurely coastal models has been iden-tified on these landforms either.Conversely, a large area of sub-merged coastal shelf occurs betweenEast Africa and southwest Arabia,making it a key area for further sur-vey both underwater and on neigh-boring land.89,91

Coastal regions are extremely vari-able both in space and through time.Some regions are attractive as muchbecause of favorable conditions forproximal terrestrial resources,including abundant water supplies,river estuaries, equable climate con-ditions, a long growing season, and

great ecological diversity, as for theirmarine resources. Others are barrenor inaccessible because of theencroachment of steep mountainranges, estuarine mudflats, thick for-est, or desert.

Examples of Holocene Homo sapi-ens marine specialist economies arerare. They typically occur in archipel-ago environments with highly pro-ductive marine resources accessedthrough advanced technologies,including seafaring and food storage,often at high latitudes alongsideunproductive or inaccessible terres-trial environments (notably in Nor-way, the northwest coast of NorthAmerica, and Tierra del Fuego). Thepossibility that similar economiesexisted on now-submerged Pleisto-cene shorelines cannot be ruled out,but in any case such economies arelikely to have been very patchy intheir distribution, as in the Holocene.To the extent that marine resourceswere exploited, they are likely to have

been combined with terrestrialresources and the use of the hinter-land through seasonal movements orin symbiosis with inland commun-ities, the more typical pattern in Holo-cene ethnographic and archeologicalrecords.81

Archeological sites that appearclose to the coast on large-scalemaps are often much further inland.For instance, the average distancefrom the modern coast of the 31Howiesons Poort (HP) sites we ana-lyzed is �105 km, despite a researchbias toward coastlines. HP sites areup to 350 km from the modern coast(Rose Cottage Cave). Post-HP sites inSouth Africa are typically even fur-ther from the coast than are HPsites. The East African sites thathave been argued to support acoastal dispersal �55 ka17 are, onaverage, 600 km from the coast;early South Asian Late Paleolithicsites are also typically far inland. InSri Lanka, early microlithic sites are

Figure 3. Map showing topography and bathymetry (SRTM30PLUS). Areas in yellow correspond to currently submerged land that wouldhave been exposed when sea level was �100m lower than present. Colored dots correspond to key archeological sites (HowiesonsPoort, MSA/LSA transitional, and South Asian Late Paleolithic) emphasized by the model of Mellars and coworkers.17 The inserts showthat the landscape can be very different from place to place and that there is no “typical” coastal environment. [Color figure can beviewed in the online issue, which is available at wileyonlinelibrary.com.]

160 Groucutt et al. ARTICLE

generally inland and clearly demon-

strate terrestrial subsistence, particu-

larly the hunting of monkeys.59

Much remains to be learned about

the role of coastal or marine resour-

ces and habitats in the Pleistocene.

However, evidence of a dramatic

adaptation to coastal ecologies and

dispersal along coasts at or before

�50 ka is currently lacking. It is

likely that coastlines provided patches

of favorable habitats, but that these

were discontinuous in space and

time. Rather than limiting dispersal

models to the strictly dichotomous

“interior” or “coastal,” we suggest

that, alongside a range of other Late

Pleistocene habitats, from the semi-

arid to rainforests, the use of coastal

ecozones is best seen as part of the

behavioral flexibility of Homo sapiens.

DISCUSSION ANDCONCLUSIONS

Our review of evidence relating to

the timing and routes of dispersal of

Homo sapiens out of Africa shows

that fossil, genetic, and archeological

data are currently consistent with

several different models. There is

much more uncertainty in the timing

and character of this dispersal than

is generally suggested by those who

argue that the first successful disper-

sal, occurred �50 (610) ka.1,14,17

Uncertainties remain about the ex-

tent of cultural and biological conti-

nuity in Homo sapiens populations

outside Africa from MIS 5 onward,

as well as about the meaning of the

major cultural and demographicchanges around 50 to 30 ka.

We suggest that accumulating dataincreasingly support a hybrid modelwhereby early expansions wereessentially swamped by subsequentones. In terms of dispersal routes,we suggest that populations em-ployed behavioral flexibility andadaptation to use a range of differentecologies, including interior savan-nahs and the coast. Accumulatingevidence of early population struc-ture and multiple population interac-tions indicates that simple modelsfor the dispersal process are no lon-ger sufficient. A key point is that theAsian paleoanthropological recordremains extremely poor, as illus-trated by the recent discovery ofcave paintings dating to at least 40ka in Indonesia.92

Further work is needed to under-stand what constraints genetic andarcheological data place on modelsof ancestral population dispersalacross the Middle East and SouthAsia. We have shown that inferencesfrom single-locus genetic data needto be based on an understanding ofthe relationship between demogra-phy and genealogical stochasticity,as embodied in coalescent or otherpopulation genetic models. Just asarcheology has largely, but notentirely, transcended the “culture-history” approach, according towhich pottery and tool types weresimplistically seen as direct proxiesfor populations, so genetic analysesmust avoid the “gene-history 5

population history” paradigm. Like-

wise, similarities between lithicassemblages, as well as other sorts ofarcheological data, can be explainedby different mechanisms. A strongarcheological argument for dispersalwould involve the correlated appear-ance of a package of several ele-ments of material culture. It is clearthat many key cultural featuresevolved convergently, among themLevallois and blade technology, aswell as tangling or pedunculation.We interpret available data as indi-cating the repeated and independentevolution of microlithic technology,but acknowledge that testing this, aswell as notions of early dispersalswith Middle Paleolithic technology,requires comparative analyses ofassemblages in Africa and Asia.Quantitative analysis of attributes areone method that, critically, can derivetechnological insights from typologi-cally indistinct artifacts.54,68

Different regions have differentstrengths and weaknesses in deter-mining the presence of Homo sapi-ens (Table 3). The Levant seems tofeature a genuine occupational hia-tus. The data from Arabia remainsomewhat ambiguous, with thestrongest archeological evidence con-sisting of MIS 5 lithics displayingsimilarities to African and Levantinematerial associated with Homo sapi-ens fossils, while post-MIS 5 Arabianlithics are culturally ambiguous.26 InSouth Asia, fossil data are currentlyabsent before �35 ka, while archeo-logical data provide moderate indica-tions of the presence of Homosapiens from MIS 5.66,67 In Southeast

TABLE 3. Tabulation of the Relative Strength of Fossil and Archeological Evidence of the Presence of Homo sapiens in SouthernAsia and Australia for selected time periodsa

Levant Arabia South Asia Southeast Asia Australia

Date (ka) Fossil Arch. Fossil Arch. Fossil Arch. Fossil Arch. Fossil Arch.

40 111 111 1 111 111 111 111 111

50 1 11 1 11 111 11 1 111

60 11 - 1 11 11 1

75 1 11 11 11 1 - -100 111 111 11 1 1 - -140 1 1 - -

a111 5 strong evidence, 11 5 moderate evidence, 1 5 weak evidence, - 5 relatively good evidence of absence, graycell 5 uncertain/insufficient information to assign to one of these categories. For details see text.

ARTICLE Rethinking the Dispersal of Homo sapiens out of Africa 161

Asia, the archeological data areambiguous regarding the presence ofHomo sapiens until �50 ka at sitessuch as Niah Cave.70 A series of ear-lier fossils from across SoutheastAsia provide possible support for thepresence of Homo sapiens back to�100 ka, and more securely to 70/60ka.28,29 In Australia, archeologicaldata indicate the presence of Homosapiens by �50 ka, with the earliestsecure fossil evidence dating to �45ka. Cumulatively, these data demon-strate that Homo sapiens were inSouthwest Asia by �120 ka andSoutheast Asia by �50 ka. The arche-ological and fossil data betweenthese points can currently be inter-preted in different ways and, as wehave outlined here, genetic data havebeen subject to questionable inter-pretations that require explicit mod-eling to be formally tested.

A number of predictions can bemade from various models andtested in future research. Regardingthe question of whether Homo sapi-ens successfully dispersed into Asiabefore �60 ka, several hypothesesand expectations can be posed.These include the discovery of pre-60ka Homo sapiens fossils in Asia out-side the Levant, as well as the dem-onstration of similarities in materialculture reasonably explained bybranching or blending in Africa andAsia before 60 ka. We expect furtheranalyses of the genomic divergencebetween African and non-Africanancestors to reveal signals of geneflow and population substructure atthis time and for ancestral demogra-phy inferred from genetic data out-side Africa to reflect dispersal intoAsia from MIS 5.

A difficulty in inferring the routesand chronology of Homo sapiens’ dis-persal out of Africa is that it requiresintegration of many different sourcesof information, each with its ownambiguities and assumptions. Inaddition, quite distinct processes cangenerate very similar patterns of vari-ation in both genetic and archeologi-cal data (equifinality). If dispersal outof Africa occurred in several waves,then it was neither exclusively “early”nor “late,” but both. Clarification ofthis important issue requires bettercross-disciplinary understanding and

the formulation of clear hypothesesthat make explicit predictions about

patterns in different types of data.Enhancement of the southern Asianfossil and archeological records

remains critical, alongside the appli-cation of more ancient DNA sequenc-ing. The interpretation of such futurefindings will most robustly be

achieved within the context of multi-disciplinary collaborations.

ACKNOWLEDGMENTS

We acknowledge funding support

from the European Research council(ERC) to M. D. Petraglia (AdvancedGrant 295719, “PALAEODESERTS:Climate Change and Hominin Evolu-

tion in the Arabian Desert: Life andDeath at the Cross-roads of the OldWorld”), G. Bailey/G. C. P. King(Advanced Grant 269586, “DISPERSE:

Dynamic Landscapes, Coastal Envi-ronments and Human Dispersals”),and N. Boivin (Starter Grant 206148,

“SEALINKS: Bridging ContinentsAcross the Sea”), all under the “Ideas”specific Programme of the 7th Frame-work Programme (FP7). M. G.

Thomas is supported by a WellcomeSenior Investigator Award in MedicalHumanities (Grant number: 100713/Z/

12/A). E. L. M. Scerri and J. Blinkhornthank the Fondation Fyssen. L. Lewisacknowledges the support of the Artsand Humanities Research Council

(UK, #513691), the Wenner-GrenFoundation (#8684), and the BoiseFund (University of Oxford). P. Breezeis funded by NERC studentship NE/

J500306/1. We thank C. Stringer fordiscussion on the morphological defi-nition of Homo sapiens.

REFERENCES

1 Willoughby PR. 2007. The evolution of mod-ern humans in Africa; a comprehensive guide.New York: Altimira.

2 Petraglia MD, Haslam M, Fuller DQ, et al.2010. Out of Africa: new hypotheses and evi-dence for the dispersal of Homo sapiens alongthe Indian Ocean rim. Ann Hum Biol 37: 288–311.

3 Stringer C. 2011. The origin of our species.London: Penguin.

4 Stringer C. 2014. Why we aren’t all multire-gionalists now. Trends Ecol Evol 29: 248–251.

5 McCown TD, Keith A. 1939. The Stone Age ofMt. Carmel, vol 2. The fossil human remainsfrom the Levalloiso-Mousterian. Oxford: Clar-endon Press.

6 Field H. 1932. The cradle of Homo sapiens.J Archaeol 36:426–430.

7 Thorne AG, Wolpoff MH. 2003. The multire-gional evolution of humans (revised paper). SciAm 13:46–53.

8 Br€auer G. 1992. Africa’s place in the evolu-tion of Homo sapiens. In: Br€auer G, Smith F,editors. Continuity or replacement? Controver-sies in Homo sapiens evolution. Rotterdam:Balkema. p 83–98.

9 Green RE, Krause J, Briggs AW, et al. 2010.A draft sequence of the Neandertal genome.Science 328:710–722.

10 Reich D, Patterson N, Kircher M, et al.2011. Denisova admixture and the first modernhuman dispersals into southeast Asia and Oce-ania. Am J Hum Genet 89:516–528.

11 Smith F. 1992. The role of continuity inmodern human origins. In: Br€auer G, Smith F,editors. Continuity or replacement? Controver-sies in Homo sapiens evolution. Rotterdam:Balkema. p 145–156.

12 Trinkaus E. 2005. Early modern humans.Ann Rev Anthropol 34:207–230.

13 Lahr MM, Foley RA. 1998. Towards a theoryof modern human origins: geography, demog-raphy and diversity in recent human evolution.Yearbk Phys Anthropol 41:137–176.

14 Klein RG. 2009. The human career: humanbiological and cultural origins. Chicago: Chi-cago University Press.

15 Shea JJ. 2011. The archaeology of an illu-sion: the Middle-Upper Paleolithic transition inthe Levant. In: Le Tenesor JM, Jagher R, OtteR, editors. The Lower and Middle Paleolithic inthe Middle East and neighbouring regions.ERAUL 126. Li�ege: Universit�e de Li�ege. p 169–182.

16 Oppenheimer S. 2012. A single southern exitof modern humans from Africa: before or afterToba? Quat Int 258:88–99.

17 Mellars P, Gori KC, Carr M, et al. 2013.Genetic and archaeological perspectives on theinitial modern human colonization of southernAsia. Proc Natl Acad Sci USA 110:10699–10704.

18 Reyes-Centeno H, Ghirotto S, D�etroit F,et al. 2014. Genomic and cranial phenotypedata support multiple modern human disper-sals from Africa and a southern route into Asia.Proc Natl Acad Sci USA 111:7248–7253.

19 Armitage SJ, Jasim SA, Marks AE, et al.2011. The southern route “out of Africa”: evi-dence for an early expansion of modernhumans into Arabia. Science 331:453–456.

20 Usik VI, Rose JI, Hilbert YH, et al. 2013.Nubian Complex reduction strategies in Dho-far, southern Oman. Quat Int 300:244–266.

21 Gunz P, Bookstein FL, Mitteroecker P, et al.2009. Early modern human diversity suggestssubdivided population structure and a complexout-of-Africa scenario. Proc Natl Acad Sci USA106:6094–6098.

22 Douka K. 2013. Exploring “the great wilder-ness of prehistory”: the chronology of the Mid-dle to the Upper Palaeolithic transition in theNorthern Levant. Mitt Gessell Urgeschichte 22:11–40.

23 Hershkovitz I, Marder O, Ayalon A. 2015.Levantine cranium from Manot Cave (Israel)foreshadows the first European modernhumans. Nature 520:216–219.

24 Shea JJ. 2003. The Middle Paleolithic of theEast Mediterranean Levant. J World Prehist 17:313–394.

25 Dennell R. 2014. Smoke and mirrors: thefossil record for Homo sapiens between Arabiaand Australia. In: Dennell R, Porr M, editors.

162 Groucutt et al. ARTICLE

Southern Asia, Australia and the search forhuman origins. Cambridge: Cambridge Univer-sity Press. p 33–50.

26 Groucutt HS, Petraglia MD. 2012. The pre-history of the Arabian Peninsula: deserts, dis-persals and demography. Evol Anthropol 21:113–125.

27 Boivin N, Fuller DQ, Dennell R, et al. 2013.Human dispersal across diverse environmentsof Asia during the Upper Pleistocene. Quat Int300:32–47.

28 Mijares SR, Detroit F, Piper P, et al. 2010.New evidence for a 67,000-year-old humanpresence at Callao Cave, Luzon, Philippines.J Hum Evol 59:123–132.

29 Demeter F, Shackelford L, Westaway K,et al. 2015. Early modern humans and morpho-logical variation in Southeast Asia: fossil evi-dence from Tam Pa Ling, Laos. PLoS ONE 10:e0121193.

30 Cann R, Stoneking M, Wilson A. 1987. Mito-chondrial DNA and human evolution. Nature325:31–36.

31 Hammer MF. 1995. A recent commonancestry for human Y chromosome. Nature378:376–378.

32 Mountain JL, Jin AA, Bowcock AM, et al.1993. Evolution of modern humans: evidencefrom nuclear DNA polymorphisms. In: AitkenMJ, Stringer CB, Mellars PA, editors. The originof modern humans and the impact of chrono-metric dating. Princeton University Press:Princeton.

33 Veeramah KR, Hammer MF. 2014. Theimpact of whole-genome sequencing on thereconstruction of human population history.Nat Rev Genet 15:149–162.

34 Pickrell JK, Reich D. 2014. Toward a newhistory and geography of human genesinformed by ancient DNA. Trends Genet 30:377–389.

35 Seguin-Orlando A, Korneliussen TS, SikoraM. 2014. Genomic structure in Europeans dat-ing back to least 36,000 years. Science 346:1113–1118.

36 Soares P, Alshamali F, Pereira JB, et al.2012. The expansion of mtDNA haplogroup L3within and out of Africa. Mol Biol Evol 29:915–927.

37 Pearson OM. 2013. Africa: the cradle ofmodern people. In: Smith FH, Ahern JCM, edi-tors. The origins of modern humans: biologyreconsidered. Hoboken: John Wiley & Sons. p1–43.

38 Campbell MC, Tishkoff S. 2010. The evolu-tion of human genetic and phenotypic variationin Africa. Curr Biol 20:R166–173.

39 Schiffels S, Durbin R. 2014. Inferringhuman population size and separation historyfrom multiple genome sequences. Nat Genet46:919–925.

40 Scally A, Durbin R. 2012. Revising thehuman mutation rate: implications for under-standing human evolution. Nat Rev Genet 13:745–753.

41 Fu Q, Mittnik A, Johnson PLF, et al. 2013. Arevised timescale for human evolution based onancient mitochondrial genomes. Curr Biol 23:553–559.

42 S�egurel L, Wyman J, Przeworski M. 2014.Determinants of mutation rate variables in thehuman germline. Ann Rev Genomics HumGenet 15:47–70.

43 Fu Q, Moorjani P, Jay F, et al. 2014.Genome sequencing of a 45,000-year-old mod-ern human from western Siberia. Nature 514:445–449.

44 Underwhill PA, Kivisild T. 2007. Use of Ychromosome and mitochondrial DNA popula-tion structure in tracing human migrations.Ann Rev Genet 41:539–564.

45 Wei W, Ayub Q, Chen Y, et al. 2013. A cali-brated human Y-chromosal phylogeny basedon resequencing. Genome Res 23:388–395.

46 Li H, Durbin R. 2011. Inference of humanpopulation history from individual whole-genome sequences. Nature 475:493–496.

47 Pagani L, Kivisild T, Tarekegn A, et al. 2012.Ethiopian genetic diversity reveals linguisticstratification and complex influences on theEthiopian gene pool. Am J Hum Genet 91:83–96.

48 Gurdasani D, Carstensen T, Tekola-Ayele F,et al. 2014. The African genome variation pro-ject shapes medical genetics in Africa. Nature517:327–332.

49 Pr€ufer K, Racimo F, Patterson N, et al.2014. The complete genome sequence of aNeandertal from the Altai Mountains. Nature505:43–49.

50 Currat M, Excoffier L. 2014. Strong repro-ductive isolation between humans and Nean-derthals inferred from observed patterns ofintrogression. Proc Natl Acad Sci USA 108:15129–15134.

51 Clarkson C, Smith M, Marwick B, et al. n.d.The archaeology, chronology and stratigraphyof Madjedbebe (Malakunanja II): a site innorthern Australia with early occupation.J Hum Evol. In press. doi:10.1016/j.jhevol.2015.03.014.

52 Groucutt HS, Scerri EML, Lewis L, et al.n.d. Stone tool assemblages and models for thedispersal of Homo sapiens out of Africa. QuatInt. In press. doi:10.1016/j.quaint.2015.01.039

53 Porraz G, Texier PJ, Archer W, et al. 2013.Technological successions in the Middle StoneAge sequence of Diepkloof Rock Shelter, West-ern Cape, South Africa. J Archaeol Sci 40:3376–3400.

54 Lewis L, Perera N, Petraglia M. 2014. Firsttechnological comparison of Southern AfricanHowiesons Poort and South Asian Microlithicindustries: an exploration of inter-regional vari-ability in microlithic assemblages. Quat Int350:7–25.

55 Brandt SA, Fisher EC, Hildebrand EA, et al.2012. Early MIS 3 occupation of Mochena Bor-ago rockshelter, southwest Ethiopian high-lands: implications for Late Pleistocenearchaeology, paleoenvironments and modernhuman dispersals. Quat Int 274:38–54.

56 James HVA, Petraglia M. 2005. Modernhuman origins and the evolution of behavior inthe later Pleistocene record of South Asia. CurrAnthropol 46(suppl):S4–S27.

57 Petraglia M, Clarkson C, Boivin N, et al.2009. Population increase and environmentaldeterioration correspond with microlithic inno-vations in South Asia ca. 35,000 years ago.Proc Natl Acad Sci USA 106:12261–12266.

58 Clarkson C, Petraglia MD, Korisettar R,et al. 2009. The oldest and longest enduringmicrolithic sequence in India: 35,000 years ofmodern human occupation at the JwalapuramLocality 9 rockshelter. Antiquity 83:326–348.

59 Perera N, Kourampas N, Simpson IA, et al.2011. People of the ancient rainforest: LatePleistocene foragers at the Batadomba-lenarockshelter, Sri Lanka. J Hum Evol 61: 254–269.

60 Blinkhorn, J. 2014. Late Middle Palaeolithicsurface sites occurring on dated sediment for-mations in the Thar Desert. Quat Int 350:95–104.

61 Deraniyagala SU. 1992. The prehistory ofSri Lanka: an ecological perspective. Colombo:Department of Archaeological Survey.

62 Seligman CG. 1921. The older Palaeolithicage in Egypt. J R Anthropol Inst GB Ireland 51:115–153.

63 Shea JJ. 2008. The Middle Stone Agearchaeology of the Lower Omo Valley Kibishformation: excavations, lithic assemblages, andinferred patterns of early Homo sapiens behav-ior. J Hum Evol 55:448–485.

64 Wendorf F, Schild R, Close AE, et al. 1993.Egypt during the Last Interglacial: the MiddlePaleolithic of Bir Tarfawi and Bir Sahara East.Plenum: New York.

65 Hovers E. 2009. The lithic assemblages ofQafzeh Cave. Oxford: Oxford University Press.

66 Clarkson C, Jones S, Harris C. 2012. Conti-nuity and change in the lithic industries of theJurreru Valley, India, before and after the Tobaeruption. Quat Int 258:165–179.

67 Blinkhorn J, Achyuthan H, Petraglia MD,et al. 2013. Middle Palaeolithic occupation inthe Thar Desert during the Upper Pleistocene:the signature of a modern human exit out ofAfrica? Quat Sci Rev 77:233–238.

68 Scerri EML, Drake NA, Jennings R, et al.2014. Earliest evidence for the structure ofHomo sapiens populations in Africa. Quat SciRev 101:207–216.

69 Dennell RW. 2009. The Palaeolithic settle-ment of Asia. Cambridge: Cambridge UniversityPress.

70 Rabett RJ. 2012. Human adaptation in theAsian Palaeolithic: hominin dispersal andbehaviour during the Late Quaternary. Cam-bridge: Cambridge University Press.

71 Vanhaeren M, d’Errico F, Stringer C, et al.2006. Middle Paleolithic shell beads in Israeland Algeria. Science 312:1785–1788.

72 McBrearty S, Brooks AS. 2000. The revolu-tion that wasn’t: a new interpretation of the ori-gin of modern human behavior. J Hum Evol39:453–563.

73 Shea JJ. 2011. Homo sapiens is as Homosapiens was. Curr Anthropol 52:1–35.

74 Powell A, Shennan S, Thomas M. 2009. LatePleistocene demography and the appearance ofmodern human behavior. Science 324:1298–1301.

75 Blome MA, Cohen AS, Tryon CA, et al.2012. The environmental context for the originsof modern human diversity: a synthesis ofregional variability in African climate 150,000–30,000 years ago. J Hum Evol 62:563–592.

76 Drake NA, Blench MA, Armitage SJ, et al.2011. Ancient watercourses and biogeographyof the Sahara explain the peopling of thedesert. Proc Natl Acad Sci USA 108:458–462.

77 Lane CS, Chorn BT, Johnson TC. 2013. Ashfrom Toba supereruption in Lake Malawishows no volcanic winter in East Africa at 75ka. Proc Natl Acad Sci USA 110: 8025–8029.

78 Parton A, White TS, Parker AG, et al. n.d.Orbital-scale climate variability in Arabia as apotential motor for human dispersals. Quat Int.In press. doi:10.1016/j.quaint.2015.01.005.

79 Walter RC, Buffler RT, Bruggemann JH,et al. 2000. Early human occupation of the RedSea coast of Eritrea during the Last Intergla-cial. Nature 405:65–9.

80 Erlandson JM. 2001. The archaeology ofaquatic adaptations: paradigms for a new mil-lennium. J Archaeol Res 9:287–350.

81 Bailey GN, Milner NJ. 2002. Coastal huntersand gatherers and social evolution: marginal orcentral? Before Farming 3-4:1–15.

ARTICLE Rethinking the Dispersal of Homo sapiens out of Africa 163

82 Bailey GN, Flemming N. 2008. Archaeologyof the continental shelf: marine resources, sub-merged landscapes and underwater archaeol-ogy. Quat Sci Rev 27: 2153–2165.

83 Marean CW. 2014. The origins and signifi-cance of coastal resource use in Africa andWestern Eurasia. J Hum Evol 77:17–40.

84 Bruggemann JH, Buffler RT, GuillaumeMMM, et al. 2004. Stratigraphy, palaeoenviron-ments and model for the deposition of theAbdur Reef limestone: context for an importantarchaeological site from the last interglacial onthe Red Sea. Palaeogeogr, Palaeoclimatol,Palaeoecol 20:179–206.

85 Stringer CB, Finlayson JC, Barton RNE,et al. 2008. Neanderthal exploitation of marinemammals in Gibraltar. Proc Natl Acad Sci USA108:14319–14324.

86 Irwin G. 1992. The Prehistoric explorationand colonization of the Pacific. Cambridge:Cambridge University Press.

87 O’Connor S, Ono R, Clarkson C. 2011.Pelagic fishing at 42,000 years before the pres-ent and the maritime skills of modern humans.Science 334:1117–1121.

88 O’Connell JF, Allen J, Hawkes K. 2010.Pleistocene Sahul and the origins of seafaring.In: Anderson A et al., editors. The global originsand development of seafaring. Cambridge:McDonald Institute. p 57–68.

89 Bailey GN, King GCP, Deves M, et al. 2012.DISPERSE: dynamic landscapes, coastal envi-ronments and human dispersals. Antiquity 86:334. http://antiquity.ac.uk/projgall/bailey334.

90 Kusky T, Robinson C, El-Baz F. 2005. Terti-ary-Quaternary faulting and uplift in the north-ern Oman Hajar mountains. J Geol Soc 162:871–888.

91 Bailey GN, Deves M, Inglis RH, et al. n.d.Blue Arabia: Palaeolithic and underwater sur-vey in SW Saudi Arabia and the role of coastsin Pleistocene dispersal. Quat Int. In press.doi:10.1016/j.quaint.2015.01.002.

92 Aubert M, Brumm A, Ramli M, et al. 2014.Pleistocene cave art from Sulawesi, Indonesia.Nature 514:223–227.

93 Behar DM, van Oven M, Rosset S, et al.2012. A “Copernican” reassessment of thehuman mitochondrial DNA tree from its root.Am J Hum Genet 90:675–684.

94 Gronau I, Hubisz MJ, Gulko B, et al. 2011.Bayesian inference of ancient human demogra-phy from individual genome sequences. NatGenet 43:1031–1034.

95 Hudson RR. 2002. Generating samplesunder a Wright-Fisher neutral model of geneticvariation. Bioinformatics 18:337–338.

96 Fenner JN. 2005. Cross-cultural estimationof the human generation interval for use ingenetics-based population divergence studies.Am J Phys Anthropol 128:415–423.

97 Crassard R, Petraglia MD, Drake NA, et al.2013. Middle Palaeolithic and Neolithic occupa-tions around Mundafan palaeolake, Saudi Ara-bia: implications for climate change andhuman dispersals. PLoS ONE 8:e69665.

98 Lehner B, Verdin K, Jarvis A. 2008. Newglobal hydrography derived from spaceborne

elevation data. EOS, Trans Am Geophys Union89:93–94.

99 Breeze P, Drake NA, Groucutt HS, et al. n.d.Remote sensing and GIS techniques for recon-structing Arabian palaeohydrology and identify-ing archaeological sites. Quat Int. In press.doi:10.1016/j.quaint.2015.01.022.

100 Hijmans RJ, Cameron SE, Parra JL, et al.2005. Very high resolution interpolated climatesurfaces for global land areas. Int J Climatol25:1965–1978.

101 Jennings RP, Singarayer J, Stone E, et al.n.d. The greening of Arabia: an ensemble of cli-mate model simulations infers multiple oppor-tunities for human occupation of the ArabianPeninsula during the Late Pleistocene. QuatInt. In press. doi:10.1016/j.quaint.2015.01.006.

102 Drake NA, Blench MA, Armitage SJ, et al.2011. Ancient watercourses and biogeographyof the Sahara explain the peopling of thedesert. Proc Natl Acad Sci USA 108:458–462.

103 Ambrose SH. 1998. Chronology of theLater Stone Age and food production in EastAfrica. J Archaeol Sci 25:377–392.

104 Bretzke K, Conard NJ, Uerpmann H-P.2014. Excavations at Jebel Faya: the FAY-NE1shelter sequence. Proc Sem Arabian Stud 44:69–82.

105 Nishiaki Y, Kanjo Y, Muhesen S, et al.2012. The temporal variability of Late Levan-tine Mousterian lithic assemblages from Deder-iyeh Cave, Syria. Eurasian Prehist 9:3–27.

VC 2015 Wiley Periodicals, Inc.

164 Groucutt et al. ARTICLE