Journal of Anthropological Archaeology 26 (2007) 88–108 www.elsevier.com/locate/jaa

The hypothesis and the South Asian archaeological record: Examination of dispersal routes through GIS analysis

Julie S. Field ¤, Michael D. Petraglia, Marta Mirazón Lahr

Leverhulme Centre for Evolutionary Studies, University of Cambridge, Downing Street, Cambridge CB2 3DZ, UK

Received 6 October 2005; revision received 5 June 2006 Available online 27 July 2006

Abstract

This research advances a model for coastal-based dispersals into South Asia during oxygen isotope stage (OIS) 4. A series of GIS-based analyses are included that assess the potential for expansions into the interior of South Asia, and these results are compared with known archaeological signatures from that time period. The results suggest that modern sapiens could have traversed both the interior and coastlines using a number of routes, and colonized South Asia rel- atively rapidly. Use of these routes also implies a scenario in which modern H. sapiens, by either increased population growth or competitive ability, may have replaced indigenous South Asian hominin populations. © 2006 Elsevier Inc. All rights reserved.

Keywords: South Asia; Human dispersals; Coastal routes; GIS; Modeling

Over the last decade, archaeological research has maps of modern human expansions. Typically, the identiWed South Asia as a crucial new frontier in the route is represented by a broad arrow that sweeps study of human evolution. The presence of both fos- north-eastwards out of Africa ca. 45,000 years ago sil and lithic components throughout the region (kya), which then bifurcates into sub-branches that attests to Pleistocene colonization and occupation enter Europe, Siberia, and the northern portion of of the subcontinent by hominin populations (Ken- South Asia (Cavalli-Sforza et al., 1993; Foley, 1987). nedy, 2000; Misra, 2001). In terms of the region’s This route is purely illustrative, and is not the prod- colonization by modern , the general uct of either paleogeographical or paleoenviron- assumption is that Homo sapiens came from the mental analysis. In contrast, it has been suggested West, ostensibly from a Eurasian population that that earlier dispersals of modern humans may have had its roots in the Levant (Klein, 1999; Straus and taken place between ca. 70 and 45 kya, or during the Bar-Yosef, 2001). This premise is largely implicit in glacial conditions of oxygen isotope stage (OIS) 4 and the interstadial Stage 3 (Lahr and Foley, 1994, 1998; Oppenheimer, 2003; Stringer, 2000). These * Corresponding author. Fax: +44 1223 335 460. would have traced the coastlines of the Arabian E-mail address: [email protected] (J.S. Field). Peninsula and the , and the

0278-4165/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jaa.2006.06.001 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 89 descendents of these people would have ultimately demographic expansions, and the movement of pop- colonized Australasia and Island by ca. ulations within and beyond South Asia’s bound- 50 kya. The presence of sites dating to 42–47 kya in aries. Therefore, the paleoenvironment of the region (O’Connell and Allen, 2004) provide the and its relation to human occupation represents an calibration point for this ‘Southern Dispersal’, important source of information on the factors implying that modern humans must have left Africa aVecting hominin evolutionary patterns in South prior to this date. The large numbers of sites across Asia. East Africa, the Arabian Peninsula, and South Asia Over a century of archaeological research in that are broadly attributable to the Middle Paleo- South Asia has revealed a wealth of late Pleistocene lithic also hint at the potential for human dispersals sites (Misra, 1989, 2001; Paddayya, 1984; Pappu and through these regions (James and Petraglia, 2005; Deo, 1994; Raju and Venkatasubbaiah, 2002). Petraglia, 2003; Petraglia et al., 2003). Unfortunately, fossils are extremely rare. In the Genetic studies support the colonization of South Narmada Valley, a hominin calvarium has been Asia by modern humans originating in Africa, and classiWed as Homo heidelbergensis (Cameron et al., more speciWc studies highlight the importance of 2004; Kennedy, 2000; Rightmire, 2001), and dated this region in the global expansion of particular lin- via associated fauna to 250–300 kya. For the rest of eages 50–70 kya (Forster, 2004; Kivisild et al., 2000; prehistory in South Asia, lithic assemblages and a Macaulay et al., 2005). Recent GIS-based analyses scattering of other artifacts have been used to deWne have produced the Wrst geographically explicit temporal units on the basis of technological sophis- model of this ‘Southern Dispersal Route’ along the tication, with reWnement contributed by relative and rim, and also examined the environ- absolute dating methods. Generally, Middle Palaeo- mental conditions that may have aVected subsis- lithic industries (assemblages that are predomi- tence, migration speed, and demographic expansion nantly based on Xakes and prepared cores) date to had this route been employed in prehistory (Field as early as 150 kya (Misra, 1995), occur in abun- and Lahr, 2005). This work highlights the attractive- dance throughout South Asia, and in a variety of ness of South Asia to coastal foragers, and also sug- topographic contexts (Jayaswal, 1978; Paddayya, gests that portions of the route through this region 1984; Pal, 2002). industries, which would have compelled populations to move inland. contain a greater number of blades and microliths, Field and Lahr also suggest that the presence of the appear as early as 45 kya (Dennell et al., 1992). Indus and Ganges–Brahmaputra deltas would have However, it must be noted that the Middle Paleo- served as full or partial barriers, keeping popula- lithic lithic industries of South Asia are exception- tions more or less within the conWnes of the South ally diverse. A recent assessment by James and Asian subcontinent, and encouraging expansions Petraglia (2005) outlines the regional diversity of into the interior. Recent genetic studies that indicate such industries in South Asia, as well as general reduced gene Xow between South Asia and the sur- trends towards the inclusion of Xake, blade and bla- rounding regions lend additional support to this delet industries at a few key sites. Elsewhere, these hypothesis (Metspalu et al., 2004). characteristics are commonly associated with mod- Moreover, preliminary studies of the genetic and ern H. sapiens, however, James and Petraglia note phenotypic signatures of South Asian populations that their sporadic occurrence in South Asia is more suggest that this region holds clues to understanding indicative of localized trends and independent the evolution and structure of human diversity out- invention, rather than the inXux of a new population side Africa. These signatures vary from the persis- (or species) with a distinctive tool kit. The potential tence of physical traits, such as skin color and body for a disassociation between H. sapiens and what is size, that may have originally developed in Africa, to typically identiWed as Upper Paleolithic/microlithic unique genetic sequences that may have evolved in raises a number of important issues for human evo- South Asia (Kivisild et al., 1999, 2003; Metspalu lution (Foley and Lahr, 1997). In particular, the et al., 2004; Vishwanathan et al., 2004). Although it behaviors that scholars usually identify as ‘modern’, is likely that demographic factors shaped much of the timing of dispersals out of Africa, the demogra- this diversity, the environment of the subcontinent, phy of hominins and modern humans, and the util- in particular geographical and ecological con- ity of ‘Middle/Upper Paleolithic’ typologies, become straints, would also have inXuenced diVerential rates open to re-interpretation. Additional archaeological of population growth, the source and extent of research in South Asia will undoubtedly clarify 90 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 these issues, but at the moment it is impossible to through other regions is not prohibited, but varies discern a uniquely modern H. sapiens signature according to relative measures of cost. In this study, from any deposits prior to 28 kya. However, the topography is the primary controlling factor for dis- hypothesis of a modern human occupation of the persals, although the extent of other geographic fea- region in the early Upper Pleistocene remains tures such as extents of sand seas or water gaps, also entirely feasible. serve as barriers. The present study, thus, aims to compare the Our analysis of dispersal routes is followed by the archaeological record with a investigation of locations known to have been colo- hypothetical model of human dispersal and later nized by hominins (but not necessarily by H. sapi- demographic expansion, and discuss the implica- ens) in relation to dispersal routes and particular tions of the model in relation to chronologies and environmental conditions. The model anticipates prehistoric trends from particular areas. Although that modern human populations would have we will not be able to discuss the Wner details of encountered other hominins, most likely H. heidel- genetic and archaeological evidence in relation to bergensis or their descendants. Dispersals into the the model, the broad patterns of convergence and interior zones of the subcontinent undoubtedly put mismatch will be used to formulate new hypotheses these two populations in direct competition with about South Asian prehistory that may be tested in one another, perhaps in a similar fashion to the future studies. SpeciWcally, this paper uses GIS- expansion of modern humans into - based analyses to assess the potential for entry into dominated Europe ca. 45 kya. Overall, our model South Asia via a northern terrestrial or southern suggests the manner in which modern human popu- coastal route. A second set of analyses suggests the lations may have circumscribed and inWltrated most parsimonious routes around and through the South Asia, and ultimately expanded demographi- subcontinent. A simple set of assumptions and con- cally to Wll the subcontinent. Regions that would straints are employed in these analyses, and the have served as corridors or barriers, or areas that resulting routes are those that demonstrate the most would have been attractive in terms of fauna, water, adherence to the model. GIS is used as an analytical or lithic resources are identiWed and discussed in engine in this respect, and we must stress that the relation to archaeological chronologies, artifact results are a single outcome, but would not have assemblages, and modern human diversity. been the only possible outcome had we chosen to include diVerent or additional variables. We do not Paleoenvironmental reconstruction mean to imply that dispersals occurred in the precise way that our spatial maps portray, but rather that In order to understand ancient dispersal patterns, these maps serve as a hypothesis, and a heuristic it is essential to consider the role of palaeoenviron- device, for the discussion of South Asian prehistory. mental conditions. A full examination of the eVects of Fundamentally, we assume that the Wrst modern climatic and tectonic change on South Asian prehis- humans to enter South Asia were mobile hunter- tory is beyond the scope of the research summarized gatherers, and in addition to hunting large game in this paper. However, a broad survey of some of the animals in a variety of terrestrial environments, they major paleoenvironmental trends provides a frame- were also familiar with coastal resources and marine work for our model of dispersals and population foods. This assumption stems from research in the expansions. South Asia is a large and complex region, African middle stone age (Henshilwood et al., 2001; and the modern climate and environments bear only Klein et al., 2004) that documents a generalized sub- partial resemblance to the conditions of OIS 4. Major sistence system that includes coastal organisms such terrestrial features, such as the Indus and Ganges Riv- as shellWsh, Wsh, sea lions, and small rodents, as well ers and tributaries, have shifted their locations across as larger prey such as bovids and antelope. These the landscape (Chattopadhyaya, 1996; Schuldenrein, Wndings imply that coastal resources, and perhaps 2002), and produced extensive Holocene-aged depos- more speciWcally aquatic resources such as Wsh and its that overlie earlier Pleistocene-aged ones. Uplift shellWsh, may have encouraged dispersals via the and regional aridity have changed the activity pat- coasts or along river valleys. To this end, we have terns of fresh water sources, and either increased or weighted particular variables in our analyses (e.g., decreased sediment loads in major drainages (Bhan- river corridors) in order to emphasize the attractive- dari et al., 2005; Enzel et al., 1999; Kar et al., 2001; ness of these environments. However, movement Srivastava et al., 2003). These and other changes J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 91 suggest that environmental dynamism and localized the eastern portion of the region, in addition to a transitions are the norm for South Asia, and modern thorny scrubland and mangrove environment reconstructions must take these paleogeographic fea- around the Ganges–Brahmaputra Delta (Adams tures into account as much as possible. and Faure, 1997). More recent palynological investi- South Asia formed as a subcontinent after the gations suggest the presence of a wet evergreen for- collision of the India and Eurasian plates ca. 55 mya. est refugium in the extreme southern end of South Consequently, the region is Xanked on the north by Asia (i.e., the southern tip of India, and also Sri the Central Makran, Sulaiman, Hindu Kush, Pun- Lanka), perhaps in combination with tropical wood- jab, Himalaya, and Arakan Mountain ranges lands in the southern ranges of the Western and (Fig. 1). The ranges are dissected by numerous Eastern Ghats (Prabhu et al., 2004). Mangroves streams and rivers, and can be crossed at speciWc would have been present at times along the Western points through high mountain passes. However, it is coastal shelf, with increases and decreases in man- unlikely that these mountainous passes would have grove expanses signalling variations in monsoonal been open for much of OIS 4 due to the prevalence rains and discharges of outXow (Kumaran et al., of extremely cold conditions and glaciers (Han, 2005, 2004; Narayana et al., 2002). 1991). In broad outline, the conditions of OIS 4 are The dynamic Xuctuations of the monsoon system thought to approximate (although to a somewhat are thought to have had a major impact on hominin lesser extent in terms of extremes in temperature and populations throughout South Asian prehistory (An aridity) what is known for OIS 2 (the last glacial et al., 2001). Seasonal and spatial Xuctuations are maximum, 24–13 kya) (Adams and Faure, 1997). known to have occurred episodically in the past, These studies indicate glacial conditions in northern opening up some of the more extreme environments latitudes (e.g., extremely dry and cool) during OIS 4, to habitation during brief periods (Korisettar and which resulted in the expansion of deserts in North Ramesh, 2002; Von Rad et al., 1999). Archaeological Africa, Arabia, and Western South Asia (the proto sites in the Thar Desert demonstrate periods of wet Thar Desert) (Goudie, 1983), and the suppression of conditions (Andrews et al., 1998; Deotare et al., the monsoon system that normally results in sea- 2004; Kar et al., 2001; Misra, 1995). Episodic mon- sonal rainfall patterns (Andrews et al., 1998; Over- soon activity had a variety of eVects on the river sys- peck et al., 1996). Analyses of river deposits from tems of the subcontinent. The drainages that this period indicate that stream Xow would have originate in the Himalayan Ranges would have been reduced in all South Asian drainage systems, experienced more instances of lateral shift and aban- resulting in stream incision and thalwegs that doned channels due to increased sediment load and reached depths of ¡80 m (R. Korisettar, pers. comm. post-monsoon discharge, whereas the rivers of the to M. Petraglia). Global sea-level fall also lowered south (Peninsular India) would have maintained coastlines to approximately 80–88 m below modern broad shallow valleys on a rocky landscape, with lit- levels ( Cutler et al., 2003; Siddall et al., 2003). This tle instance of braiding (R. Korisettar, pers. comm. would have exposed a considerable portion of the to M. Petraglia). However, discharge in these regions coastal shelf, in particular the Western side of the was enough to form large lakes in the broad shallow subcontinent. valleys, which transitioned into dry basins as aridity Increased aridity suppressed vegetation in the returned (Korisettar, 2004). higher elevations, resulting in a topographic mosaic Lastly, a period of punctuated change that of polar deserts and montane tundra throughout the occurred within the time frame of the Southern Dis- Himalayas and Hindu Kush ranges (Adams and persal may also have had an impact on hominin Faure, 1997). Palynological cores and faunal populations and the expansion of modern human remains from the south coast of India indicate that populations into South Asia. Ash and tephra depos- much of the interior of South Asia was covered with its located throughout the region are known to be dry savannahs and grasslands during glacial periods the product of a volcanic eruption, speciWcally the (Erdelen and Preu, 1990; Kotlia and Sanwal, 2004). super-eruption of the Toba volcano in Indonesia These conditions would have extended longitudi- between 71 and 74 kya (Acharyya and Basu, 1993; nally from the Deccan Plateau to the northern Gan- Shane et al., 1993; Westgate et al., 1998). It has getic Plain, changing into dry scrublands at the been suggested that the after-eVects of this foothills of the eastern and Western Ghats. Dry eruption resulted in a population crash for all tropical woodlands are thought to have existed in terrestrial organisms, including hominin popula- 92 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108

Fig. 1. Palaeoenvironmental model for South Asia during OIS 4. Based on reconstructions summarized in Adams and Faure (1997) (2005 website), and also personal comments from Dorian Fuller in 2005. Extent of Thar Desert derived from Glennie and Singhvi (2002). J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 93 tions (Ambrose, 1998). The magnitude of this event lized a rasterized model of the landscape of South and its eVects on hominin populations in South Asia (i.e., the features of the landscape were repre- Asian prehistory is currently under scrutiny (Jones, sented as a cellular grid). Each of the cells in the in preparation). model was coded to a particular value, thus creating a ‘friction surface’ that could be used to analyze por- A GIS-based model of dispersals across South Asia tions of the landscape for the least costly routes between particular points. In this case, rasterized A geographic information system (GIS) was slope data from the HYDRO1K and ETOPO2 employed to better understand dispersal routes (LPDAAC, 2003; NGDC, 2004) data sets supplied across South Asia. Although there is good knowl- the founding layers for the friction surface. These edge about the location of archaeological sites data were merged to produce a composite terrain for thanks to numerous surveys, the actual routes that the region, with a resolution of 1 km for the hominins may have undertaken has not been a sub- HYDRO1K terrestrial surfaces, and 3.7 km for the ject of serious study. The current study employs GIS ETOPO2 bathymetric surfaces (i.e., portions of methodologies in order to examine the likeliest South Asia exposed by the 80–88 m sea-level fall). routes for dispersals. While GIS has become a stan- The act of merging these rasters produced a resolu- dard tool in archaeology (Church et al., 2000; tion ‘edge’ along the coastlines. This edge was not Wheatley and Gillings, 2002), there have been rela- mosaiced (i.e., the diVerence between resolution of tively few studies which attempt the kind of analyses the rasters was not generalized) in order to preserve portrayed here (for an exception, see Anderson and the quality of both data sets in the coastal areas. The Gillam, 2000). However, we must stress that our presence of the edge did present some problems, as methodology is derived from a very simple model route segments that were plotted through the ETO- and set of assumptions, some of which have been PO2 data were of lower resolution than those of the discussed elsewhere in studies of the colonization of HYDRO1K dataset. However, an ‘edge eVect’ in the New World (Dixon, 1999, 2001; Grühn, 1994; which the routes followed the merged edges of the Mandryk et al., 2001; Mosimann and Martin, 1975) data sets occurred only rarely, due to the use of the and Australia (Birdsell, 1977; Bowdler, 1977; wandering method outlined below. O’Connell and James, 1998; O’Connor and Chap- The merged surface was then reclassiWed, and the pell, 2003; Wild, 1985). In our model, we assume that values of individual increments of slope recalibrated the Wrst modern humans to colonize South Asia using a scale that has been adjusted to reXect human were mobile hunter-gatherers, and that these groups energy expenditure (Bell and Lock, 2000; GorenXo may have relied upon, or at least been familiar with, and Gale, 1990; Llobera, 2000; van Leusen, 2002) subsistence strategies that included coastal (Table 1). In this scaling, the relative energetic cost resources. We also assume that regions that were of crossing slopes increases with slope grade. We mountainous, at high elevation, or with steep slopes acknowledge that slope is only one aspect of topog- would not have been included as part of major raphy that aVects biogeographical processes, how- routes of colonization. Montane environments and ever, it allows for a relative assessment of energy resources were undoubtedly important to early colo- expenditure for movement across long distances, nizers; however, when considering human dispersals and is fairly sensitive to topographic constraints these regions are less likely to have served as major such as cliVs or rugged topography, which may have routes, as they are more costly to hunter-gatherers in directed population movements in one direction or terms of energy expenditure, and less attractive in another. Equating slope grade with travel cost also terms of the availability of game and collectable parallels a basic assumption of biological theory, foods. Thus, our focus is on coastlines, plains, and which proposes that routes up and across steep river valleys, and more generally on corridors that slopes are more costly in terms of energy expendi- connect the interior of South Asia with the coasts. ture and foraging time than across Xat landscapes, and are less often pursued by mobile organisms Methodology I: friction surface (Krebs and Davies, 1993). Our friction surface also utilized barriers, which The generation of routes into and through a we deWned as features of the landscape that would paleoenvironmentally adjusted model of South Asia eVectively block or re-direct travel, such as lakes, was performed with ArcGIS 8.3 software, and uti- sand seas, wide rivers, coastlines, and continental ice 94 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108

Table 1 ers that were wider than 5 k were set as impermeable Variables and weights used in GIS calculations of least-cost barriers, but rivers and streams of lesser size were routes presumed to have been crossable. The location of Variable Weight the coastline at OIS 4 was projected as an imperme- Slope (0°–25°)a 0–25 able boundary line on top of the friction surface Slope 26°b 27.95 using a topographic algorithm within ArcGIS. We Slope 27° 29.19 employed ¡88 m as the coastline for our model, Slope 28° 30.47 W Slope 29° 31.76 which exposed a signi cant portion of the coastal Slope 30° 33.08 shelf. However, this boundary designation is only a Slope 31° 34.43 rough estimate of the coastline at OIS 4, as this Slope 32° 35.76 method does not take into account other factors Slope 33° 37.21 that would have aVected sea levels at that time, such Slope 34° 38.64 Slope 35° 40.12 as continental uplift. Slope 36° 41.63 The location of sand seas during OIS 4 was esti- Slope 37° 43.18 mated for South Asia using models generated by Slope 38° 44.77 geologists (Glennie and Singhvi, 2002) and the mod- Slope 39° 46.40 ern extent of the Thar Desert from the Global GIS Slope 40° 48.08 Slope 41° 49.81 Database (USGS, 2000). Sand seas were not classi- Slope 42° 51.59 Wed as impermeable boundaries, but were given a Slope 43° 53.43 value that was high in relation to other slope costs in Major riversc 0 the study (15°), but not high enough to completely d Sand seas 15 preclude the possibility of crossing the narrower Sea levele Impermeable portions (Table 1). Other GIS-based studies that uti- a Slope calculations were obtained from ancillary slope data lized friction surfaces have identiWed the importance sets of HYDRO1K and calculations based on the topographic coverage of ETOPO2. of partial barriers such as sand seas, as they can also b Slope energy values were derived from the following calcula- re-direct movement, and slow down traverse rates tion: tan slope x/tan (1°) (Bell and Lock, 2000, pp. 88–89). (Ray et al., 1999; Steele et al., 1998; Surovell, 2003; c The location of major rivers likely to have been present dur- Young and Bettinger, 1995). ing OIS 4 was calculated from the HYDRO1K stream data set. Lastly, we also assume that during the glacial Stream order (Strahler, 1964) was used to select streams of order 3 or higher, as these drainages are of signiWcant size and volume, conditions of OIS 4, riparian areas that would have and more likely to have persisted in their present location during provided water and habitat for a variety of animals glacial periods. They are also more likely to have provided per- would have been most attractive to hunter-gather- manent water Xow. ers. To this end, riparian areas that Xank major river d The extent of aeolian sand deposits is derived from the courses were assigned relatively low values (or no Global GIS Database (USGS, 2000) and published data by Glen- nie and Singhvi (2002) for the extent of desert sands during gla- cost) in order to emphasize their attractiveness on cial episodes in Arabia and South Asia. The value of 15 prohibits the rasterized landscape (Table 1). We acknowledge crossing by least-cost pathways in nearly all circumstances, but that privileging these features in the landscape will does allow for crossing of short stretches (i.e., 90 km). have the eVect of driving route generation towards a e As discussed previously, global sea levels for OIS 4 were given particular outcome; however, in the context of a an average of ¡80 m. The location of this level on the ETOPO2 data set was calculated for the study area, and this line was then slope-based friction surface, riparian areas along reclassiWed as an impermeable barrier. river courses were already of extremely low cost. By emphasizing these areas, we are attempting to con- struct a model that is relevant to the needs of sheets. In this study, the location of major rivers hunter-gatherers at OIS 4. Other variables that likely to have been present during OIS 4 was calcu- relate to human subsistence and mobility (e.g., vege- lated from the HYDRO1K stream data set. Stream tation types, the density of game animals, and rain- order data included in the data set was used to select fall averages) were not included in analyses that streams of order 3 or higher (Strahler, 1964), as result in route generation, as precise empirical data these drainages are of signiWcant size and volume, concerning their presence and character is lacking and more likely to have persisted in their present for this time period. Rather, the eVect these condi- location during glacial periods. They are also more tions may have had on populations as they moved likely to have provided permanent water Xow. Riv- along the topographically derived routes will be J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 95 discussed in terms of their impact on foraging, tion, the search radius was moved to the newest demography, and settlement patterns. endpoint, and the path was calculated from that point onwards. Analyses of routes by segment leg Methodology II: least cost routes continued until the culminated route reached a cul de sac, such as a high mountain valley, or joined Our model assumes that migrating populations with a least-cost wandering route that had already generally follow the “easier”, or least costly, route been plotted. At this point, the route was started across a terrain. A route may be deWned as a linear again from a new origin. It is important to note that course of travel through a physical landscape, which the wandering route technique is distinctly diVerent includes the crossing of Xat plains, river banks, or at from the ‘origin’ and ‘destination’ driven routes times crossing of mountain passes. This study exempliWed by Anderson and Gillam’s (2000) analy- employed two discrete analyses: a ‘direct route’ ses of colonization routes into the New World. Their analysis and a ‘wandering route’ analysis. The direct technique plots the least costly routes from origin route analysis sought to examine the most parsimo- points to sites in the interior that are known to have nious route into South Asia from points to the west; been occupied by Paleoindians. This does not the wandering route analysis aimed to produce emphasize any particular environment as more hypothetical routes within South Asia that were not attractive or habitable, but indicates a potential driven by a destination, but would have led to the sequence for colonization of the continent via a distribution of modern human populations in diVer- direct route. The wandering method employed in ent regions, and perhaps at diVerent times. Both of this study eliminates the problem of having a desti- these analyses rely upon a least-cost algorithm nation, and more closely mimics the movement of housed within the GIS. This analysis determines the colonizing populations, who would not have known pathway of least-cost between two points by mea- what lay before them. However, it should be noted suring the value of adjacent and intervening cells in that these GIS routes remain hypothetical, and the analysis Weld (Madry and Rakos, 1996; van Leu- should be recognized as predicted outcomes from sen, 2002). In the direct route analysis, an expansive the range of possibilities identiWed above in the dis- ‘origin’ that lies to the west of South Asia was used cussion of the friction surface. Additional analyses to simulate the entry into South Asia from either a that incorporated other variables, such as the attrac- northern terrestrial route, or a Western coastal tiveness of lithic resources, or seasonal migration route. This origin was a line that stretched 2200 km routes for herds of large mammals, would have pro- across nearly all of southwest Asia, from the Pamir duced alternative outcomes in the wandering analy- Mountain Range that lies to the west of the Tibetan sis. Plateau, to the banks of the Euphrates River in Eastern Arabia (Fig. 2). This area was chosen as the Result of the direct routes analysis origin based upon the consensus view that modern humans dispersed out of Africa, rather than a source Approximately 2200 separate sub-routes were from the northeast, such as Eastern Asia. Points generated from the points along the origin line, were set up along this line at an interval of every which overlapped into three major routes that pro- kilometre. Using the friction surface described ceeded around or through the Zagros Mountains above, and a destination point set on the southern- (Fig. 2). These sub-routes united near the Straits of most tip of South Asia, the least costly ‘direct’ route Hormuz, and then headed east along the Makran from each of the points along the line was calcu- Coast into South Asia. The linear origin analysis lated. suggests that the most economical routes into South The wandering route method employed a 60 km Asia cross through the lower elevations of the search radius and a starting point on the friction Zagros Mountains, or follow a predominantly ripar- surface (the location of the starting point is ian corridor along the Euphrates River and onto the described below). Analysis of the values of the cells Makran Coast. This indicates that populations orig- surrounding the starting point determined the least inating in the Levant, Northern Africa, or other costly path, which could proceed in any direction points to the west would have been more likely to towards the boundary of the search radius. Thus, enter South Asia if they took a southern route. This each wandering route was generated as a culmina- also holds true for dispersals originating along the tion of 60 km path segments; following each calcula- southeast edge of the Eurasian Steppes (the farthest 96 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108

Fig. 2. Location of linear origin and the results of the least-cost route analysis into South Asia. Least-cost route is indicated by the grey line. Costly areas, including sand seas and areas of high slope are indicated by the color grade. Inset shows detail of Zagros Mountains. eastern point of the line). Analyses of least-cost ward direction. Of note, the least-cost routes do not routes from this region encounter the steep terrain cross the Hindu Kush Mountains into the northern- of the Hindu Kush and Pamir Mountains, and when most headwaters of the Indus Valley. Not only were they cannot proceed south they turn back, in a west- these regions rugged and steep, but during OIS 4 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 97 they would have been partially ice-bound (Paterson, tres to the northeast generated a route that paral- 2003), and less likely to have attracted hunter-gath- leled the Western bank of the Indus. This route erers that were moving eastwards. This does not pre- proceeded northwards along the Indus Valley, even- clude the possibility that populations entered and tually reaching a cul de sac at the foothills of the occupied these regions, but implies that these areas Sulaiman Range. Additional analyses from a start- were not major routes into South Asia for the bulk ing point further upriver allowed for this route to of the population. continue onto the Potwar Plateau, and end in the The potential sub-routes also oVered markedly foothills of the Hindu Kush Mountains. This route diVerent resources. The routes through the Zagros also suggests that entry into the Gangetic Plains Mountains cross the arid Naomid Plain and Lut could have been gained via a northern route around Basin, which during OIS 4 consisted of windswept the Thar Desert. yardangs and expansive salt Xats (Selivanov, 1982) Independent of the Indus Valley diversion, place- (Fig. 2, inset). Although wet phases could have sup- ment of a third origin on the southern side of the ported small populations of ungulates and birds, Indus Delta (which presumes that hominin popula- during dry periods this region would have been a tions were able to cross both the water and marsh very harsh environment. In contrast, the Tigris– barriers within the delta), generated a wandering Euphrates route would have crossed through ripar- path that continued down the coastline to the south. ian areas and marshes, providing an expansive envi- Subsequent wandering down the coast led the route ronment with both water and game (Aqrawi, 2001). to skirt the southern fringe of the sand seas associ- It is more likely that this region could have sup- ated with the Thar Desert. The route then encoun- ported a dispersing population during OIS 4, and tered the delta of the Narmada River, which Xows thus led them directly towards the Straits of Hor- into the Indian Ocean just south of the boundary for muz and the Makran Coast. Once in this region, the Pleistocene Thar Desert. The least cost analysis migrating populations would have encountered an predicts entry into the interior at this point, using arid coastline that provided a variety of marine the Marble Rocks gorge as a corridor into the Nar- foods (e.g., molluscs, crustaceans, birds, and turtles) mada Valley. The route continued to wander along (Hopner et al., 2000), and with further travel east- the Narmada Valley for approximately 1500 km, wards would have reached the rich ecosystem of the and then jumped into the drainage of the Son River. Indus Delta (Meadows and Meadows, 1999). This leg of the wandering route was allowed to ter- minate at the conjunction with the Ganges River. Result of wandering routes analysis Back on the coast, the placement of a fourth ori- gin on the southern side of the Narmada Delta The initiation point for the wandering routes resulted in a route that wandered southwards. This begins on the Western bank of the Indus Delta route followed a particularly irregular coastline, (Fig. 3). This location was selected from the results with numerous diversions around bays and onto of the direct route analysis, which passed through headlands. The landscape here consisted of a rocky, this region. We also realize that caution needs to be faulted coastline, with numerous promontories (R. exercised in placing the wandering origin here, given Korisesttar, pers. comm. to M. Petraglia). On the that the bed of the Indus and its tributaries has way, a third interior diversion was encountered at repeatedly changed its course over the Holocene. Cape Rama, a headland that lies approximately The Indus Valley is an ancient riparian landscape 80 km south of the modern city of Panaji. In that has been carrying outXow from the Himalayas repeated analyses, the headland and the surrounding to the Indian Ocean throughout the Pleistocene hills forced the route inland through a low pass in (Schuldenrein, 2002). It may have been one of the the Western Ghats. The route then ranged widely Wrst of several diversions into the interior, as the through the interior, Wrst travelling north along trib- river itself could have proved a barrier during OIS 4, utaries of the Krishna River, and then swinging and the Xat alluvial terraces of the valley bottom south to skirt the northern boundary of the Deccan would have drawn hunter-gatherer populations Plateau. The great degree of wandering in this northwards. Indeed, this was the result of the Wrst region is largely due to the extremely Xat nature of analysis; the wandering route reached a cul de sac the topography; it is far more likely that a more between the coastline and the wide Indus Delta, and focused dispersal occurred along the river valleys of the placement of a new starting point a few kilome- this region, which would have provided both water 98 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108

Fig. 3. Results of the wandering path analyses into and through South Asia. Least-cost routes are indicated by the grey lines. Costly areas, including sand seas and areas of high slope are indicated by the color grade. Inset shows detail of Cape Rama diversion. J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 99 and game. Archaeological survey in this region has, levels after the last glacial maximum. This is appar- in fact, indicated an abundance of sites (Pappu and ent in a comparison of Fig. 4 with a modern map of Deo, 1994), although at least a proportion of this South Asia. Our study suggests that submerged sites abundance is due to the ease of surveying in this should exist along both coasts of India as well as region. Additional wandering indicated a diversion between India and Sri Lanka, although to date no northwards into the Krishna Basin, which eventu- detailed examination of potential underwater depos- ally turned southeast and followed the Penner River its has been attempted (Flemming, 2004). A few sites Valley to the Coromandel Coast. are known that are adjacent to coastal regions, such However, if a Wfth origin was placed on the as Ramayogi Agraharam (Rath et al., 1997), Borra southern side of Cape Rama, a wandering route was (Vijaya Prakash et al., 1995), Attirampakkam predicted that ran southwards towards the tip of (Pappu, 2001), the Konkan complex (Gudzer, 1980), peninsular South Asia. At Cape Comorin, the wan- and the Hiran Valley (Baskaran et al., 1986; Mar- dering route headed northwards, and crossed the Sri athe, 1981), and these sites contain assemblages that Lankan isthmus. The route continued to Xank the are categorized as Middle Paleolithic. The Hiran coastline for the next 2500 km, occasionally dipping Valley deposits have associated dates within the into low lying river deltas. The route ultimately range we suggest for the Southern Dispersal (56– entered the Penner Delta, and then upon reaching 69 kya), although which hominin species made the the Ganges–Brahmaputra Delta crossed into the artefacts found remains unknown. Damodar Valley. As with the Narmada and the If the lack of sites along the coast is not a product Indus, the size of the Ganges–Brahmaputra river of survey eVort, another explanation could be the system, as well as the expanse of mangroves and speed of dispersal. It is possible that coastal move- marsh around the delta, implies a region that would ments occurred relatively rapidly, and left little in the have impeded colonization to the east. Based on way of a recognizable archaeological signature. Such this, and also the attractiveness of the riparian corri- a scenario has been suggested by wave of advance dor connecting the Damodar River with the coast- models of colonizing populations, such as Hazelwood line, our analyses suggest that populations would and Steele’s (2004) and Steele et al.’s (1998) analyses have been diverted, at least initially, into the Ganges for pioneer expansions into continents. Their work Valley, rather than continuing eastwards into the outlines the actual mechanisms of wave dispersals in Pleistocene Sundarbans. The remainder of the wan- detail, suggesting that ‘broad’ wave proWles, in which dering route analysis predicted travel westward, population density is low and dispersals are rapid, Xanking the southern side of the Ganges–Brah- leave only remnant deposits and no clear pattern of maputra Delta. Continued wandering along the decreasing age from the ‘origin’ to the outer edges of south bank of the Ganges River produced four Wnal the expansion. We suggest a similar scenario for bifurcations: two routes crossed the river at a nar- South Asia, with the added twist that its coastlines row point and headed northwards into the valleys of would have initially funnelled the dispersals south the Arun and Sun Kosi Rivers, which Xow out of the and eastwards. The Makran Coast and the coastline Himalayas. Two other routes continued westward of the Thar Desert, which appears as the most likely and followed the Ganges and the Ghaggar Rivers. entry point into South Asia in our analyses, were The routes were allowed to terminate at this point. extremely arid during OIS 4, and consisted of a nar- row coastal corridor backed by rugged desert hills Proposed routes, colonization issues, and the (Singhvi and Kar, 2004). The intertidal zones would archaeological record of South Asia have provided a variety of marine foods, such as mol- luscs, sand dwelling shellWsh, nesting sea turtles, and The least-cost routes suggest that the coastlines crabs (Hopner et al., 2000), although the lack of water could have provided major migratory corridors into and narrowness of this corridor implies that popula- and within South Asia, and these regions should tions would have had to move through quickly (Field bear evidence for human occupation. However, an and Lahr, 2005). Further south, coastal dispersals archaeological signature that could be linked to may have encountered brackish marshes and salt coastlines remains elusive. Perhaps one of the obvi- Xats, and perhaps also expanses of mangroves (Kum- ous, but salient, explanations for this is that many aran et al., 2004). Subsistence in this region may also Pleistocene sites along the proposed coastal routes have relied exclusively upon marine resources, as would have been submerged following the rise in sea these environments provide few terrestrial ones. 100 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108

Fig. 4. Summary of proposed routes, with location of hominin sites of known Pleistocene age. Site numbers: (1) Didwana Complex, (2) Budha Pushkar, (3) Hockra, (4) Luni Valley, (5) Hiran Valley, (6) Beas-Berach, (7) Bhimbetka, (8) Adamgarh, (9) Samnapur, (10) Patne, (11) Konkan Complex, (12) Upper Son, (13) Belan Valley, (14) Middle Son, (15) Dang-Deokhuri Complex, (16) Singbhum, (17) Ramayogi Agraharam, (18) Bora, (19) Kaladgi Basin, (20) Shorapur Doab, (21) Kurnool, (22) Attirampakkam, (23) Batadomba-Lena, (24) Fa Hien Cave, (25) Site 50, (26) Site 49, (27) Site 55. J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 101

This scenario of an initial coastal occupation for Sulaiman and Hindu Kush ranges implies a degree South Asia has also been suggested for the Ameri- of isolation for these populations, although net- cas, where several early dates for human occupation works of exchange and interaction may have episod- appear in South America, rather than near the pre- ically moved people and materials out of the region. sumed origin point in Beringia (Dillehay, 2000; A similar environment would have been associ- Dixon, 1999; Roosvelt et al., 1996). Dixon has sug- ated with the Narmada Delta. If, as hypothesized, gested that this pattern relates to an initial expan- populations of mobile foragers were moving down sion along the coastlines of the Americas by the Western coastline of South Asia, the expanses of populations utilizing coastal resources and water- riparian areas associated with the Narmada Delta craft. He also hypothesizes that a generalized coastal would have provided a welcome refuge from the foraging strategy is advantageous in colonizing sce- aridity of the desert coastlines to the north. In addi- narios, as it does not require a complex tool kit or tion, the Narmada Valley would have provided a access to lithic resources, nor large populations of rich terrestrial habitat for ungulates, birds, and terrestrial mammals. Rather, such dispersals may other wildlife, and mobile foragers that used this have relied on simple gathering of intertidal animals, valley as a route into the interior would have such as shellWsh, which require few tools. The initial encountered savannahs and grasslands. These land- colonization of South Asia by modern H. sapiens scapes would have supported terrestrial fauna of could have followed a similar pattern, with less similar composition to sub-Saharan Africa, includ- investment in complex lithics that are associated ing horse, rhino, boar, deer, antelope, wild cattle, with terrestrial hunting (e.g., blades and composite pygmy hippo, elephant, stegodon, and buValo tools), and a highly developed suite of tools associ- (Badam, 1979). The archaeological sites of Bhim- ated with coastal foraging, such as wooden spears, betka rockshelter (III F-23) and Adamgarh occur shell and bone hooks, Wber nets and baskets. along this route, occupying either side of the Nar- In terms of dispersals into the interior of South mada River. The deep deposits (over 3 m, and rang- Asia, the route analyses identify deltas as the initial ing from the Late Acheulean to the Mesolithic) of portals inland. The Indus River is the Wrst encoun- both sites indicate the presence of populations in the tered in these analyses, and although increased arid- region in the Late Pleistocene (Joshi, 1978; Misra, ity during OIS 4 would have signiWcantly reduced 1985). Aridity in this environment would have the size of the river, in the midst of the surrounding required populations to be mobile, yet tied to crucial desert the estuary would have provided a crucial environmental zones that contained water and lithic habitat for migrating birds, as well as contained Wsh resources. However, if a relative abundance of ter- and shellWsh. Larger animals that were common to restrial fauna is assumed, then there is the potential the savannahs of South Asia, such as elephant, for demographic expansion in this region, with addi- buValo, wild cattle, horse, deer, and rhinoceros may tional geographic expansions to the grasslands of have been present (Badam, 1979). These conditions the north and south. suggest that populations could have remained in this A third interior dispersal is hypothesized by the region for longer periods, and probably expanded crossing of the Western Ghats at Cape Rama. Popu- upriver as indicated by the wandering route analyses lations that entered this portion of the subcontinent (Fig. 3). The occurrence of Site 55 in the foothills of would have Wrst encountered the hills of the Western the Hindu Kush Mountains, which is dated to Ghats, which during OIS 4 supported dry wood- 45 kya, is the only Pleistocene site in this part of lands and scrublands. Once in the interior of penin- South Asia that may correspond with this dispersal sular South Asia, migrants would have encountered route. Dennell and colleagues document the pres- grassy country in the Deccan Plateau, which was ence of stone blades at the site, and also suggest that broken occasionally into hilly terrain with a vegeta- the activities included habitation and lithic procure- tion of scrub and dry deciduous forest. This region ment in the high foothills (Dennell et al., 1992). also contains a number of basins, such as the Kala- Although the high rate of sedimentation has dgi and Cuddapah. Archaeological investigations in resulted in a poorly known Paleolithic archaeologi- this region indicate a high density of sites and a cal record for the rest of the Indus, we hypothesize range of lithic assemblages (Murty and Reddy, 1975; that other sites are likely to be found in abundance Pappu and Deo, 1994; Petraglia et al., 2003; Raju, in future. In addition, the fact that the region is 1988). Excavations in the Kurnool Caves have Xanked by the Thar Desert and the mountains of the revealed Upper Pleistocene deposits, which include 102 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 lithic artifacts and the remains of horse, rhino, boar, by the intermittent streams that Xow southwards gazelle, deer, antelope, and wild cattle (Murty, 1979; out of the desert. Also missed by the predicted Prasad, 1996). As with the grasslands to the north, routes are the late Pleistocene sites of Badatomba- human populations are expected to have expanded lena and Fa Hien on Sri Lanka (Deraniyagala, demographically in this environment, and eventu- 1992). The earliest known remains for modern ally Wlled the various tributaries with small, mobile H. sapiens and shell beads in South Asia come from groups. this region, dating to ca. 31–28 kya (Kumar et al., If the least-cost routes are viewed as a hypotheti- 1988). If these dates indeed indicate the earliest cal “distribution scheme” for South Asian popula- occupation of Sri Lanka by modern humans, then a tions, it is important to note the instances where late settlement date for the isthmus (and later island) routes seem to result in convergences. For example, is predicted. continued migrations via the Narmada Valley would eventually reach the Ganges Plain, which Discussion could also be reached through separate movements across the Potwar Plateau, and along the Western Current understanding of our evolutionary his- side of the Ganges–Brahmaputra Delta. This sug- tory suggests that South Asia must have played a gests that the eastern end of the Gangetic Plain pivotal role in the expansion of modern humans out could have experienced a higher frequency of dis- of Africa (Lahr and Foley, 1994; Stringer, 2000). As persals, and perhaps also accumulated a more the mid-way point between Africa and Australia, diverse population as a result. This region may have and as a large region that supported a savannah also acted as a partial cul-de-sac, as further expan- environment similar to that of Sub-Saharan Africa, sions eastwards would have been slowed down by it seems a prime candidate for early colonization by the expanse of the Ganges–Brahmaputra Delta act- modern H. sapiens, as well as a potentially major ing as a potential spatial ‘bottleneck’. A number of locus for subsequent expansions of human popula- sites exist in this region, with sequences that range tions. Such a pivotal role is also suggested by the from the Acheulean to the Mesolithic. The Son pattern of non-African mitochondrial DNA River Valley and the Belan Valley are particularly (mtDNA) diversity, showing that populations in the rich areas with stratigraphic sequences and paleo- region today carry lineages that are part of one of ecological information (Ahmed, 1984; Blumens- the two Wrst clades to derive from African diversity chine et al., 1983; Clark and Williams, 1990; Sharma (namely M* lineages). More recent genetic analyses and Clark, 1983). Dispersals in the south suggest the indicate that certain geographically isolated popula- convergence of routes in and around the Coroman- tions in retain unique mtDNA lin- del Coast and the isthmus of Pleistocene Sri Lanka. eages with time-depths of 44,000–63,000 years, Increased interaction in this region may have also suggesting that lineage divergence occurred very resulted in a more diverse population, and higher early in the region (Macaulay et al., 2005; Thangaraj frequency of exchange concerning technology and et al., 2005). ideas. The extensive evidence of a lack of genetic diver- There are a number of instances in which the sity of living humans (including South Asians) also archaeological record seems unrelated to the least- strongly suggests that modern humans replaced, cost routes emerging from this study. For example, a rather than intermingled with the archaic popula- series of important sites in the Thar Desert; the sites tions that already resided outside of Africa (Cann, of Didwana (Misra et al., 1982), Luni Valley 2001; Cann et al., 1987; Currat and ExcoYer, 2004; (Mishra et al., 1999), and Hockra (Allchin et al., Eswaran et al., 2005; Forster, 2004; Relethford and 1978). Although the simulations did not suggest dis- John, 1995; Watson et al., 1997). Currently, the lack persals into the Thar Desert under the palaeoenvi- of hominin fossils makes it diYcult to determine ronmental conditions stipulated, the area would exactly what these populations were in South Asia— have been attractive during wet periods, even if of either Homo erectus, H. heidelbergensis, Homo nean- short duration. Indeed, archaeological sites in the derthalensis, or a yet undescribed species of Homo. area are often associated with riverine and lacustrine From extensive archaeological deposits it appears settings (Mishra et al., 1999; Misra, 1995). In such that they enjoyed a lengthy history, with their own circumstances, this region may have been reached trends towards more diverse and sophisticated lithic via dispersals along the Ghaggar River, or perhaps assemblages (James and Petraglia, 2005). However, J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 103 the fact that these populations went extinct with the Godavari drainages. These regions appear to have inXux of modern humans suggests that H. sapiens been major loci for hominin populations in antiq- had some competitive advantage over pre-existing uity, as evinced by the abundant Middle Paleolithic South Asian hominins, perhaps in terms of and Late Acheulean assemblages. Further popula- increased rates of population growth, more Xexible tion growth would probably have spread modern subsistence strategies, or more extensive social net- humans north and south, to the limits of the savann- works. Moreover, the absence of a distinctive lithic ahs. Such demographic growth and expansion assemblage that can be linked to modern H. sapiens across South Asia may be reXected in the increased may be more telling than previously realized. If these complexity of the archaeological record after 45 kya populations relied upon generalized coastal hunting (James and Petraglia, 2005). and gathering strategies similar to those that have Additional aspects of South Asian prehistory are been documented for Blombos Cave and Ysterfon- not well articulated in the model of South Asian dis- tein (Henshilwood et al., 2001; Klein et al., 2004), persals presented here. For one, the importance of then the search for microliths or points which are lithic sources in the hominin choice of habitats was usually associated with specialized terrestrial hunt- not included in the analyses. Discovery of quarries ing and equated with modern H. sapiens, may be a or other lithic sources may have provided the impe- false trail. A generalized tool kit, and an accompa- tus for additional expansions into parts of South nying subsistence system that could rapidly traverse Asia, while the absence of these resources, such as in coastal and riparian environments may have pro- the Ganges Valley, may also have played a role in vided one of the crucial advantages that allowed the lack of sites in this region. The impact of the modern humans to move into regions that may have Toba eruption on South Asian hominins is also already been inhabited by South Asian hominins. unknown. If, as has been suggested (Ambrose, 1998), Moreover, this Xexibility may have meant that they the eruption and its associated ashfall had a signiW- were less constrained by broader ecological regions, cant and negative impact on hominin populations in and could move freely between them without a South Asia, modern H. sapiens may have led a rela- lengthy period of technological adjustment. We also tively unchallenged invasion of the region after stress that the search for a parallel to the European 70 kya. As the dating for the initial entry into Aus- Upper Paleolithic (e.g., blades, carved bone points, tralasia is at a maximum 50 kya, this leaves plenty of beads, eYgies, etc.) in South Asia enforces a particu- time for populations to have dispersed into and lar model of human behavioural evolution (Klein, through South Asia in the millennia following the 1999), and precludes the usage of broader deWni- Toba eruption. tions of modernity. Research of modernity in Africa (Foley and Lahr, 1997; McBrearty and Brooks, Conclusion 2000) has shown that behavior appears to follow multiple development trajectories, The results of GIS-based analyses of least-cost rather than a single progression. routes predicts that entry into Asia during OIS 4 is In addition, the least-cost routes suggested by the more likely to have used a coastal corridor that orig- analyses presented in this paper indicate that mod- inated to the west, as opposed to a northern mon- ern humans would have been able to follow natural tane corridor through the Hindu Kush Mountains. corridors deep into the countryside, and establish Once in South Asia, populations may have followed populations within the heart of South Asia. At a a number of routes, which included both coastal and minimum, if modern humans were able to maintain terrestrial contexts. Diversions into the interior are a higher rate of population growth after their entry suggested via the Indus River and Narmada River into the interior, then outward population expan- valleys, and also through a break in the Western sions from these regions may have placed additional Ghats. The analyses also indicate that dispersals pressure on indigenous South Asian Homo by forc- along the coasts of South Asia would have eventu- ing them out of the most productive savannahs and ally turned inland at the Ganges–Brahmaputra into the marginal areas. Based on the results of least- Delta, and that this feature may have blocked or cost route analysis, the most likely loci for the slow populations attempting to move eastwards. expansion of anatomically modern human popula- The identiWcation of archaeological sites and site tions within South Asia were the Son, Belan and complexes along these corridors indicates some con- Upper Narmada Valleys, as well as the Krishna and currence with these hypotheses in several cases, 104 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 although their absence is notable for the coastal Acknowledgments regions. The rise in sea levels since the Last Glacial Maximum may explain this absence to some degree, This research was funded by the UK NERC- although we also suggest that the subsistence tech- EFCHED project ‘Searching for Traces of the nologies employed by coastal foragers, such as per- Southern Dispersal’, which had as its goal the exam- ishable nets or collecting equipment, or expedient ination of the environmental factors aVecting tools used for collecting shellWsh, may not have pre- human evolution and dispersal in prehistory. Addi- served into the Holocene. Moreover, dispersals tional funding was provided by the Leverhulme through arid coastal regions may have been very Trust. We wish to acknowledge and thank the input rapid, and left little in the form of archaeological of Ravi Korisettar, Shanti Pappu, and Dorian Ful- deposits. In terms of the Southern Dispersal ler, who provided useful suggestions on this manu- Hypothesis, the analysis of routes described here script. The input of two anonymous reviewers was suggests that dispersals into South Asia could have also greatly appreciated. happened very rapidly, and either expanded through the interior or along the coastal perimeter. However, References there is also the potential for a relative deceleration out of South Asia (i.e., south and eastwards into Acharyya, S., Basu, P.K., 1993. Toba ash on the Indian subconti- Asia) as populations expanded demographically nent and its implications for correlation of Late Pleistocene into the varied environments of the region, and also alluvium. Quaternary Research 40, 10–19. Adams, J.M., Faure, H., 1997. Review and Atlas of Palaeovegeta- met the barrier of the Ganges–Brahmaputra Delta. tion: preliminary land ecosystem maps of the world since the What is known for the paleoenvironment of Last Glacial Maximum. In: QEN Website. South Asia suggests that both the coasts and the Ahmed, N., 1984. The stone age cultures of the Upper Son Valley, interior would have been attractive to mobile parties Madhya Pradesh. Agam Kala Prakashan, Delhi. of modern hunter-gatherers. If the use of coasts and Allchin, B., Goudie, A.S., Hegde, K., 1978. The prehistory and palaeogeography of the Great Indian Desert. Academic Press, riparian corridors allowed for rapid dispersals into London. South Asia, and movement between these environ- Ambrose, S., 1998. Late Pleistocene human population bottle- ments was not a problem in terms of subsistence and necks, volcanic winter, and diVerentiation in modern humans. subsistence related technologies, our analyses sug- Journal of Human Evolution 34, 623–651. gest that populations could have expanded rapidly An, Z., Kutzbach, J.E., Prell, W., Porter, S.C., 2001. Evolution of Asian monsoons and phased uplift of the Himalaya–Tibetan into the heartland of South Asia. Reasons for the plateau since Late Miocene times. Nature 4, 62–66. extinction of non-modern hominins, if still present Anderson, D.G., Gillam, J.C., 2000. Paleoindian colonization of in these regions, are discussed. Later expansions that the Americas: implications from an examination of physiog- led to the extinction of indigenous hominins are also raphy, demography, and artefact distribution. American surmised. This hypothesis has further support from Antiquity 65, 43–66. Andrews, J., Singvi, A., Kuhn, P., 1998. Do stable isotope data genetic data that suggests lineage divergence in from calcrete record late Pleistocene monsoonal climate varia- South Asia ca. 44,000–63,000 years ago, as well as tion in the Thar Desert of India? Quaternary Research 50, archaeological evidence for increasingly modern 240–251. behaviors in South Asia after this time period. Aqrawi, A., 2001. Stratigraphic signatures of climate change during In conclusion, the prehistory of South Asia will the Holocene evolution of the Tigris-Euphrates delta, lower Mesopotamia. Global and Planetary Change 28, 267–283. be further uncovered in future years through study Badam, G.L., 1979. Pleistocene Fauna of India. Deccan College of the archaeological and fossil record, and the Postgraduate and Research Insititute, Pune. genetics of modern populations. The assessment of Baskaran, M., Marathe, R., Rajaguru, S., Somayajulu, B., 1986. the potential routes of dispersal through the region Geochronology of Paleolithic cultures in the Hiran Valley, allows for an examination of the paleoenvironmen- Saurashtra, India. Journal of Archaeological Science, 13. Bell, T., Lock, G., 2000. Topographic and cultural inXuences on tal factors that may have shaped or directed prehis- walking the Ridgeway in later prehistoric times. In: Lock, G. toric use of the South Asian landscape, and also (Ed.), Beyond the Map: Archaeology and Spatial Technolo- provides a model and a series of hypotheses that gies. IOS Press, Amsterdam, pp. 85–100. may be tested against others. Undoubtedly, future Bhandari, S., Maurya, D.M., Chamyal, L.S., 2005. Late Pleisto- work and exploration in the region will further cene alluvial plain sedimentation in Lower Narmada Valley, W Western India: Palaeoenvironmental implications. Journal of develop and re ne this model, and lead to new Asian Earth Sciences 24, 433–444. hypotheses concerning the dispersal of modern Birdsell, J.B., 1977. The recalibration of a paradigm for the Wrst humans throughout the world. peopling of Greater Australia. In: Sunda and Sahul. J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 105

Blumenschine, R., Brandt, S., Clark, J., 1983. Excavations and and hypotheses. In: Thornes, J.B. (Ed.), Vegetation and Ero- analysis of Middle Palaeolithic artifacts from Patpara, Mad- sion. John Wiley and Sons, New York, pp. 491–520. hya Pradesh. In: Sharma, G., Clark, J. (Eds.), Palaeoenviron- Eswaran, V., Harpending, H., Rogers, A.R., 2005. Genomics ments and Prehistory in The Middle Son Valley. Abinash refutes an exclusively African origin of humans. Journal of Prakashan, Allahabad, pp. 39–114. Human Evolution 49, 1–18. Bowdler, S., 1977. The coastal colonisation of Australia. In: Allen, Field, J.S., Lahr, M., 2005. Assessment of the southern dispersal: J., Golson, J., Jones, R. (Eds.), Sunda and Sahul. Academic GIS based analyses of potential routes at oxygen isotope stage Press, London, pp. 205–246. 4. Journal of World Prehistory 19, 1–45. Cameron, D., Patnaik, R., Sahni, A., 2004. The phylogenetic sig- Flemming, N.C., 2004. Submarine prehistoric archaeology of the niWcance of the Middle Pleistocene Narmada cranium from Indian continental shelf: a potential resource. Current Science Central India. International Journal of Osteoarchaeology 14, 86, 1225–1230. 419–447. Foley, R., 1987. Another Unique Species: Patterns in Human Cann, R., 2001. Genetic clues to dispersal in human populations: Evolutionary Ecology. Wiley and Sons, New York. retracing the past from the present. Science 291, 1742–1748. Foley, R.A., Lahr, M.M., 1997. Mode 3 technologies and the evo- Cann, R., Stoneking, M., Wilson, A., 1987. Mitochondrial DNA lution of modern humans. Cambridge Archaeological Journal and human evolution. Nature 325, 31–36. 7, 3–36. Cavalli-Sforza, L.L., Piazza, A., Menozzi, P., 1993. Demic expan- Forster, P., 2004. Ice ages and the mitochondrial DNA chronol- sions and human evolution. Science 259, 639–646. ogy of human dispersals: a review. Philosophical Transactions Chattopadhyaya, U., 1996. Settlement pattern and the spatial of the Royal Society of London, 255–264. organization of subsistence and mortuary practices in the Glennie, K.W., Singhvi, A.K., 2002. Event stratigraphy, palaeoen- Mesolithic Ganges Valley, North-Central India. World vironment and chronology of SE Arabian deserts. Quaternary Archaeology 27, 461–476. Science Reviews 21, 853–869. Church, T., Brandon, R.J., Burgett, G.R., 2000. GIS applications GorenXo, L.J., Gale, N., 1990. Mapping regional settlement in in archaeology, method in search of theory. In: Westcott, K.L., information space. Journal of Anthropological Archaeology Brandon, R.J. (Eds.), Practical Applications of GIS for 9, 240–274. Archaeologists, A Predictive Modeling Kit. Taylor and Fran- Goudie, A.S., 1983. Dust storms in space and time. Progress in cis, London, pp. 135–155. Physical Geography 7, 502–530. Clark, J.D., Williams, M.A., 1990. Prehistoric ecology, resource Grühn, R., 1994. The PaciWc coast route of initial entry: an over- strategies, and culture change in the Son Valley, Northern view. In: Bonnischsen, R., Steele, D.G. (Eds.), Method and Madhya Pradesh, Central India. Man and Environment, 13– Theory for Investigating the Peopling of the Americas. Center 24. for the Study of the First Americans, Corvallis, pp. 249–256. Currat, M., ExcoYer, L., 2004. Modern humans did not admix Gudzer, S., 1980. Quaternary Environment and Stone Age Cul- with during their range expansion into Europe. tures of The Konkan, Coastal Maharashtra, India. Deccan PLoS Biology 2, 1–11. College Postgraduate and Research Institute, Pune. Cutler, K.B., Edwards, R.L., Taylor, F.W., Cheng, H., Adkins, J., Han, T., 1991. Quaternary geology of Qinghai–Tibet (Xizang) Gallup, C.D., Cutler, P.M., Burr, G.S., Bloom, A.L., 2003. plateau. In: Zhang, Z. (Ed.), The Quaternary of China. China Rapid sea-level fall and deep-ocean temperature change since Ocean Press, Beijing, pp. 405–440. the last interglacial period. Earth and Planetary Science Let- Hazelwood, L., Steele, J., 2004. Spatial dynamics of human dis- ters 206, 253–271. persals: constraints on modelling and archaeological valida- Dennell, R., Rendell, H., Halim, M., Moth, E., 1992. A 45,000- tion. Journal of Archaeological Science 31, 669–679. year-old open-air Paleolithic site at Riwat, Northern Pakistan. Henshilwood, C., Sealy, J., Yates, R., Cruz-Uribe, K., Goldberg, Journal of Field Archaeology 19, 17–33. P., Grine, F., Klein, R.G., Poggenpoel, C., van Niekerk, K., Deotare, B., Kajale, M., Rajaguru, S., Basavaiah, N., 2004. Late Watts, I., 2001. Blombos Cave, Southern Cape, South Africa: Quaternary geomorphology, palynology, and magnetic sus- Preliminary Report on the 1992–1999, Excavations of the ceptibility of playas in western margin of the Indian Thar Middle Stone Age Levels. 28:421–448. Desert. Indian Geophysical Union 8, 15–25. Hopner, T., Ebrahimipour, H., Maraschi, S., 2000. Five intertidal Deraniyagala, S.U., 1992. The Prehistory of Sri Lanka: An Eco- areas of the Persian Gulf. Wadden Sea Newsletter 2, 30–33. logical Perspective. Department of the Archaeological Survey, James, H., Petraglia, M., 2005. Modern human origins and the Government of Sri Lanka, Colombo. evolution of behaviour in the later Pleistocene record of South Dillehay, T.D., 2000. The Settlement of The Americas. Basic Asia. Current Anthropology 46, S3–S27. Books, Perseus Book Group, New York. Jayaswal, V., 1978. Palaeohistory of India: A Study of teh Pre- Dixon, E.J., 1999. Bones, Boats, and Bison: Archaeology and The pared Core Technique. Agam Kala Prakashan, Delhi. First Colonization of Western North America. University of Jones, S.C., The Toba supervolcanic eruption: tephra-fall deposits New Mexico Press, Albuquerque. in India and paleoanthropological implications. In prepara- Dixon, E.J., 2001. Human colonization of the Americas: timing, tion. technology, and process. Quaternary Science Reviews 20, 277– Joshi, R.V., 1978. Stone-Age Cultures of Central India (Report on 299. The Excavations of Roch Shelters at Adamgarh, Madhya Enzel, Y., Ely, L.L., Mishra, S., 1999. High-resolution Holocene Pradesh). Deccan College Post-Graduate and Research Insti- environmental changes in the Thar Desert, Northwestern tute, Poona. India. Science 284, 125–128. Kar, A., Singhvi, A.K., Rajaguru, S.N., 2001. Reconstruction of Erdelen, W., Preu, C., 1990. Quaternary coastal and vegetation the late Quaternary environment of the lower Luni Plains, dynamics in the Palk Strait region, South Asia: the evidence Thar Desert, India. Journal of Quaternary Science 16, 61–68. 106 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108

Kennedy, K.A.R., 2000. God-apes and Fossil Men: Paleoanthro- LPDAAC 2003 HYDRO1K Elevation Derivative Database. pology in South Asia. The University of Michigan Press, Ann Macaulay, V., Hill, C., Achilli, A., Rengo, C., Clarke, D., Meehan, Arbor. W., Blackburn, J., Semino, O., Scozzari, R., Cruciani, F., Taha, Kivisild, T., Kaldma, K., Metspalu, M., Parik, J., Papiha, S., Vil- A., Kassim Shaari, N., Maripa Raja, J., Ismail, P., Zainuddin, lems, R., 1999. The place of the Indian mtDNA variants in the Z., Goodwin, W., Bulbeck, D., Bandelt, H.-J., Oppenheimer, S., global network of maternal lineages and the peopling of the Torroni, A., Richards, M., 2005. Single, rapid coastal settle- old world. In: Dekha, R., Papiha, S. (Eds.), Genomic Diversity. ment of Asia revealed by analysis of complete mitochondrial Kluwer/Academic/Plenum, New York, pp. 135–152. genomes. Science 308, 1034–1036. Kivisild, T., Papiha, S., Rootsi, S., Parik, J., Kaldma, K., Reidla, Madry, S.L., Rakos, L., 1996. Line-of-sight and cost-surface tech- M., Laos, S., Metspalu, M., Pielberg, M., Adojaan, M., Metsp- niques for regional research in the Arroux river valley. In: alu, E., Mastana, S., Wang, Y., Golge, M., Demirtas, H., Maschner, H. (Ed.), New methods, Old Problems: Geographic Schnakenberg, E., De Stefano, G., Geberhiwot, T., Claustres, Information Systems in Modern Archaeological Research. M., Villems, R., 2000. An Indian ancestry: a key for under- Center for Archaeological Investigations, Southern Illinois standing human diversity in Europe and beyond. In: Renfrew, University, Carbondale, pp. 104–126. C., Boyle, K. (Eds.), Archaeogenetics: DNA and The Popula- Mandryk, C., Josenhans, H., Fedje, D.W., Mathewes, R., 2001. tion Prehistory of Europe. McDonald Institute for Archaeo- Late Quaternary paleoenvironments of Northwestern North logical Research, Cambridge, pp. 267–275. America: Implications for inland versus coastal migration Kivisild, T.S., Rootsi, M., Metspalu, S., 2003. The genetic heritage routes. Quaternary Science Reviews 20, 301–314. of the earliest settlers persists both in Indian tribal and caste Marathe, A., 1981. Geoarchaeology of the Hiran Valley, Saurash- populations. American Journal of Human Genetics 72, 313– tra, India. Deccan College Postgraduate and Research Insti- 332. tute, Poona. Klein, R.G., 1999. The Human Career. University Press, Chicago. McBrearty, S., Brooks, A., 2000. The revolution that wasn’t: a Klein, R.G., Avery, G., Cruz-Uribe, K., Halkett, D., Parkington, J., new interpretation of the origin of modern human behavior. Steele, T., Volman, T., Yates, R., 2004. The Ysterfontein 1 Mid- Journal of Human Evolution 39, 453–563. dle Stone Age site, South Africa, and early human exploitation Meadows, A., Meadows, P. (Eds.), 1999. The Indus River: Biodi- of coastal resources. Proceedings of the National Academy of versity, Resources, Humankind. Oxford University Press, Sciences of the United States of America 101, 5708–5715. Delhi. Korisettar, R., 2004. Geoarchaeology of the Purana and Gondw- Metspalu, M., Kivisild, T., Metspalu, E., Parik, J., Hudjashov, G., ana Basins of Peninsular India: Peripheral or paramount? In Kaldma, K., Serk, P., Karmin, M., Behar, D., Thomas, M., Gil- Presidential Address, Sixty-Wfth Session of the Indian History bert, P., Endicott, P., Mastana, S., Papiha, S., Skorecki, K., Congress. Torroni, A., Villems, R., 2004. Most of the extant mtDNA Korisettar, R., Ramesh, R., 2002. The Indian monsoon: roots, boundaries in South and Southwest Asia were likely shaped relations, and relevance. In: Indian Archaeology In Retro- during the initial settlement of Eurasia by anatomically mod- spect: Archaeology and Interactive Disciplines. Indian Coun- ern humans. BMC Genetics, 25. cil of Historical Research and Manohar Publishers, New Mishra, S., Jain, M., Tandon, S.K., Singvi, A., Joglekar, P., Bhatt, Delhi, pp. 23–59. S., Kshirsagar, A., Maik, S., Deshpande-Muhkerje, A., 1999. Kotlia, B.S., Sanwal, J., 2004. Fauna and palaeoenvironment of a Prehistoric Cultures and Late Quaternary Environments in Late Quaternary Xuvio-lacustrine basin in Central Kumaun the Luni Basin around Balotra. Man and Environment XXIV, Himalaya. Current Science 87, 1295–1299. 39–49. Krebs, J.R., Davies, N.B., 1993. An Introduction of Behavioral Misra, V.N., 1985. The Acheulean Succession at Bhimbetka, Cen- Ecology. The Alden Press, Oxford. tral India. In: Misra, V.N., Bellwood, P. (Eds.), Recent Kumar, G., Narvare, G., Pancholi, R.K., 1988. Engraved ostrich Advances in Indo-PaciWc Prehistory. Oxford I-B-H, New eggshell objects: new evidence of Upper Palaeolithic art in Delhi, pp. 35–47. India. Rock Art Research, 5. Misra, V.N., 1989. Stone Age India: an ecological perspective. Kumaran, K.P.N., Nair, K.M., Shindikar, M., Limaye, R., Pad- Man and Environment 14, 17–64. malal, D., 2005. Stratigraphical and palynological appraisal of Misra, V.N., 1995. Geoarchaeology of the Thar Desert, North- the Late Quaternary mangrove deposits of the west coast of west India. In: Wadia, S., Korisettar, R., Kale, V.S. (Eds.), India. Quaternary Research 64, 418–431. Quaternary Environments and Geoarchaeology of India. Kumaran, K.P.N., Shindikar, M.R., Mudgal, T.R., 2004. Floristic Memoirs Geological Society of India, Bangalore, pp. 210–230. composition, palynology and sedimentary facies of Hadi man- Misra, V.N., 2001. Prehistoric human colonization of India. Jour- grove swamp (Maharashtra). Journal of the Indian Geophysi- nal of Biosciences 26, 491–531. cal Union 8, 63–66. Misra, V.N., Rajaguru, S., Raju, D., Raghavan, H., Gaillard, C., Lahr, M., Foley, R., 1994. Multiple dispersals and modern human 1982. Acheulian occupation and evolving landscape around origins. Evolutionary Anthropology 3, 48–60. Didwana in the Thar Desert, India. Man and Environment 6, Lahr, M., Foley, R., 1998. Towards a theory of modern human 72–86. origins: geography, demography, and diversity in recent Mosimann, J.F., Martin, P.S., 1975. Simulating overkill by Paleo- human evolution. Yearbook of Physical Anthropology 41, indians. American Scientist 63, 304–313. 137–176. Murty, M., 1979. Recent research on the Upper Palaeolithic Llobera, M., 2000. Understanding movement: a pilot model phase in India. Journal of Field Archaeology 6, 301–320. towards a sociology of movement. In: Lock, G. (Ed.), Beyond Murty, M.L.K., Reddy, K.T., 1975. The signiWcance of lithic Wnds the Map: Archaeology and Spatial Technologies. IOS Press, in the cave areas of Kurnool, India. Asian Perspectives 18, Amsterdam, pp. 65–84. 214–226. J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108 107

Narayana, A.C., Priju, C.P., Rajagopalan, G., 2002. Late Quater- Relethford and John, H., 1995. Genetics and modern human ori- nary peat deposits from Vembanad Lake (lagoon), Kerala, gins. Evolution Anthropology 4, 53–63. SW coast of India. Current Science 83, 318–321. Rightmire, G.P., 2001. Comparison of Middle Pleistocene homi- NGDC 2004 ETOPO2 CD-ROM. nids from Africa and Asia. In: Robson-Brown, L., Barham, K. O’Connell, J., Allen, J., 2004. Dating the colonization of Sahul (Eds.), Human Roots: Africa and Asia in the Middle Pleisto- (Pleistocene Australia-): a review of recent cene. Western Academic Press, Bristol, pp. 123–135. research. Journal of Archaeological Science 31, 835–853. Roosvelt, A.M., Lima da Costa, C., Lopes Machado, M., Michab, O’Connell James, F., 1998. When did humans Wrst arrive in Australia N., Mertier, H., 1996. Paleoindian cave dwellers in the Ama- and why is it important to. Evolutionary Anthropology 6, 132. zon: the peopling of the Americas. Science 272, 373–384. O’Connor, S., Chappell, J., 2003. Colonisation and coastal subsis- Schuldenrein, J., 2002. Geoarchaeological perspectives on the tence in Australia and Papua New Guinea: DiVerent timing, Harappan Sites of South Asia. In: Settar, S., Korisettar, R. diVerent modes? In: Sand, C. (Ed.), PaciWc Archaeology: (Eds.), Indian Archaeology in Retrospect: Protohistory: Assessments and Prospects. Service des Musees et du Patrimo- Archaeology of the Harappan Civilization. Indian Council of ine, Noumea, pp. 17–32. Historical Research, Manohar, pp. 47–80. Oppenheimer, S., 2003. Out of Eden: The Peopling of the World. Selivanov, E.I., 1982. The Dasht-E Lut desert of Iran. Problemy Constable, London. Osvoeniya Pustyn, 14–19. Overpeck, J., Anderson, D.G., Trumbore, S., Prell, W., 1996. The Shane, P., Westgate, J., Williams, M., Korisettar, R., 1993. New southwest Indian monsoon over the last 18,000. Climate geochemical evidence of the youngest Toba tuV in India. Qua- Dynamics 12, 213–225. ternary Research 44, 200–204. Paddayya, K., 1984. India. In: Bar-Yosef, O., Corvinus, G., Hahn, Sharma, G., Clark, J., 1983. Palaeoenvironments and Prehistory J. (Eds.), Neue Forschungen zur Altsteinzeit.. Verlag CH in the Middle Son Valley. Abinash Prakashan, Allahabad. Beck, Munich, pp. 345–403. Siddall, M., Rohling, E.J., Almogi-Labin, A., Hemleben, C., Mei- Pal, J.N., 2002. The Middle Palaeolithic Culture of South Asia. In: schner, D., Schmelzer, I., Smeed, D.A., 2003. Sea-level Xuctua- Settar, S., Korisettar, R. (Eds.), Indian Archaeology in Retro- tions during the last glacial cycle. Nature 423, 853–858. spect: Prehistory: Archaeology of South Asia. Indian Council Singhvi, A.K., Kar, A., 2004. The aolian sedimentation record of of Historical Research, Manohar, pp. 67–83. the Thar Desert. Proceedings of Indian Academy of Science Pappu, S., 2001. A re-examination of the Palaeolithic archaeolog- (Earth and Planetary Sciences) 113, 371–401. ical record of northern Tamil Nadu, South India. Srivastava, P., Singh, I.B., Sharma, M., 2003. Luminescence chro- Pappu, S., Deo, S., 1994. Man-Land Relationships During Palae- nometry and Late Quaternary geomorphic history of the olithic Times in The Kaladgi Basin, Karnataka. Deccan Col- Ganga Plain, India. Palaeogeography, Palaeoclimatology, lege, Prune. Palaeoecology 197, 15–41. Paterson, T., 2003. The glaciation on the Himalayan slope. In: De Steele, J., Adams, J., Sluckin, T., 1998. Modelling paleoindian dis- Terra, H., Paterson, T. (Eds.), The Ice Age in the Indian Sub- persals. World Archaeology 30, 286–305. continent and Associated Human Cultures: With Special Ref- Strahler, A.N., 1964. Quantitative geomorphology of drainage erence to Jammu, Kashmir, Ladakh, Sind, Liddar & Central basins and channel networks. In: Chow, V.T. (Ed.), Handbook and Peninsular India. Aryan Books, New Delhi. of Applied Hydrology. McGraw-Hill, New York, pp. 39–76. Petraglia, M., 2003. The of the Arabian Penin- Straus, L.G., Bar-Yosef, O., 2001. Out of Africa in the Pleistocene. sula: occupations, adaptations, and dispersal. Journal of Quaternary International 75, 1–3. World Prehistory, 17. Stringer, C.B., 2000. Coasting out of Africa. Nature 405, 24–27. Petraglia, M., Schuldenrein, J., Korisettar, R., 2003. Landscapes, Surovell, T.A., 2003. Simulating coastal migration in New World activity, and the Acheulean to middle Paleolithic transition in colonization. Current Anthropology 44, 581–591. the Kaladgi Basin, India. Eurasian Prehistory 1, 3–24. Thangaraj, K., Chaubey, G., Kivisild, T., Reddy, A., Kumar Prabhu, C.N., Shankar, R., Anupama, A., 2004. A 200-ka pollen Singh, V., Rasalkar, A., Singh, L., 2005. Reconstructing the and oxygen-isotopic record from two sediment cores from the origin of Andaman Islanders. Science 308, 996. eastern Arabian Sea. Palaeogeography, Palaeoclimatology, USGS, 2000. Global GIS Database, Digital Data Series DDS-62- Palaeoecology 214, 309–321. B. United States Geological Survey, FlagstaV. Prasad, K., 1996. Pleistocene cave fauna from peninsular India. van Leusen, P.M., 2002. Pattern to Process: Methodological Journal of Caves and Karst Studies 58, 30–34. Investigations into The Formation and Interpretation of Spa- Raju, D., 1988. Stone Age Hunter-Gatherers: An Ethnoarchaeology tial Patterns in Archaeological Landscapes. University of Cuddapah Region, Southeast India. Ravish Publishers, Pune. Library Groningen, Groningen. Raju, D., Venkatasubbaiah, P., 2002. The archaeology of the Vijaya Prakash, P., Rath, A., Krishna Rao, S., 1995. Prehistoric upper Palaeolithic phase in India. In: Settar, S., Korisettar, R. cultural evidence from Borra Limestone Caves, Eastern (Eds.), Indian Archaeology in Retrospect: Prehistory. Indian Ghats. Man in India 75, 163–173. Council of Historical Research, New Delhi, pp. 85–109. Vishwanathan, H., Deepa, E., Cordaux, R., Stoneking, M., Usha Rath, A., Thimma Reddy, K., Vijaya Prakash, P., 1997. Middle Rani, V., Majumder, P., 2004. Genetic structure and aYnities Palaeolithic assemblage from Ramayogi Agraharam in the among tribal populations of southern India: a study of 24 auto- red sediments in the Visakhapatnam coast. Man and Environ- somal DNA markers. Annals of Human Genetics 68, 128–138. ment 22, 31–38. Von Rad, U., Schulz, H.V., Reich, V., 1999. Multiple monsoon- Ray, N., Schneider, S., ExcoYer, L., 1999. Potential early modern controlled breakdown of the oxygen-minimum conditions humans dispersals in Africa: a GIS model based on past envi- during the past 30,000 years documented in the laminated sed- ronmental constraints. In: XV INQUA Congress, South iments oV Pakistan. Palaeogeography, Palaeoclimatology, Africa. Palaeoecology 152, 129–161. 108 J.S. Field et al. / Journal of Anthropological Archaeology 26 (2007) 88–108

Watson, E., Forster, P., Richards, M., Bandelt, H.J., 1997. Mito- Wheatley, D., Gillings, M., 2002. Spatial Technology and Archae- chondrial footprints of human expansions in Africa. Ameri- ology: The Archaeological Applications of GIS. Taylor and can Journal of Human Genetics 61, 691–704. Francis, London. Westgate, J., Shane, R., Pearce, N., Perkins, W., Korisettar, R., Wild, S., 1985. Voyaging to Australia: 30,000 years ago., Brisbane. Chesner, C., Williams, M., Acharyya, S., 1998. All Toba tephra Young, D.A., Bettinger, R.L., 1995. Simulating the global human occurrences across peninsular India belong to the 75,000 yr expansion in the late pleistocene. Journal of Archaeological BP eruption. Quaternary Research 50, 107–112. Science 22, 89–92.