Geography and Environments

Represented by the modern nations of Bangladesh, India, Nepal, Pakistan, Sri Lanka, and the Maldives, South Asia is a region both characterized and constrained by major geographical features (Robinson 1989). Bordered by the Himalayas to the north and the Arabian Sea and the Bay of Bengal to the west and east respectively, the Indian subcontinent is subdivided by numerous mountain ranges and plateaus, including the Western Ghats, the Eastern Ghats, and the Deccan Plateau. Numerous major river valleys, including the Indus, Narmada, Ganges, Godavari, and Krishna, cross the region. In the northwest, the Thar Desert (or Great Indian Desert) represents the easternmost extension of the midlatitude desert belts of Africa and Central Asia (Deotare et al. 2004). The major part of the Indian subcontinent has been a monsoonal environment since the Miocene, although fluctuations and shifts in its intensity, perhaps related to HimalayanTibetan uplifts, are registered through time (e.g., An et al. 2001, Retallack 1995). Monsoonal shifts during the Pleistocene and marked seasonal changes in wet and dry periods are thought to have structured hominin settlement behaviors (Korisettar and Rajaguru 1998, Korisettar and Ramesh 2002, Paddayya 1982).

In the Later Pleistocene as today, South Asias variable ecology and landscape provided a wide range of potential settings for hominin adaptations. Archaeological evidence clearly indicates Later Pleistocene occupation throughout the subcontinent (fig. 1), including the settlement of both coastal and estuarine environments such as at the Ramayogi Agraharam locality (Rath, Thimma Reddy, and Vijaya Prakash 1997) and in the Hiran Valley (Marathe 1981). Most sites have been identified in interior river valleys (Pappu and Deo 1994, Raju 1988), though this distribution may be partly a reflection of survey bias rather than a reliable record of hominin landscape use. The analysis of a number of ancient basins indicates that hominins sought areas where water (in the form of lakes, streams, and springs), lithic resources, and animal communities converged (Korisettar 2004). Archaeological sites are found in a range of topographic settings, including lowlands, uplands, and submountainous zones (Pappu 1995). Resource exploitation strategies involved both openair contexts, such as Hokra and Gurha in the Thar Desert (Allchin, Goudie, and Hegde 1978), and caves and rockshelters, such as Borra (Vijaya Prakash, Rath, and Krishna Rao 1995) and Adamgarh (Joshi 1978).

Terrestrial environments were subject to ecological changes as part of climatic oscillations throughout the Pleistocene. Paleoenvironmental data from both continental and oceanic records appear to indicate cycles of arid and humid conditions coupled with a trend toward increasing aridity as the Upper Pleistocene progressed. Increasingly open environments supplanted the mixed woodland and grassland ecosystems that characterized the Middle Pleistocene, with brackish swamps replacing plentiful freshwater sources (Misra 2001). With the exception of the ancient basins, most inland areas (including the Ganga Plain) were grasslands.

Within the Thar Desert, phases of aridity (and dune formation) are interspersed with periods of wetter, ameliorated climate (Andrews et al. 1998, Deotare et al. 2004, Kar et al. 2001). Such climatic and environmental fluctuations would have influenced the demographic profile of South Asias ancient populations, as well as requiring cultural innovations. During periods of favorable climate (i.e., interglacial phases), increased resource availability would have enabled populations to expand. Unstable or less favorable conditions and their impact on the carrying capacity of the environment would have led to the reduction and fragmentation of populations. Various types of evidence support such patterns of demographic change. The ameliorated climate of oxygen isotope stage 3 coincides with wetter, more stable conditions in the Thar Desert (Andrews et al. 1998, Deotare et al. 2004) and may correspond to a demographic expansion proposed on the basis of mitochondrial DNA analysis (Kivisild et al. 1999a). Occupation in the Thar Desert becomes increasingly sparse and isolated after ca. 25,000 years ago, reflecting the heightened aridity and loss of available water sources (e.g., Deotare et al. 2004, Misra 2001). Extreme aridity characterizes the Thar Desert during the Last Glacial Maximum, a time period that saw the weakening of the southwest monsoon (Deotare et al. 2004). The effects of climatic oscillations on population size and dispersal within South Asia are likely to have been as dramatic as those seen elsewhere (Forster 2004).

Tectonic and volcanic events may have had significant effects on Later Pleistocene hominin populations. Recent research indicates that the Ganga Plain of northern India remained tectonically active throughout the last 80,000 years (Srivastava et al. 2003). Tephra deposits from the Toba supereruption, ca. 75,000 years ago, have been reported from river valleys within peninsular India (Westgate et al. 1998). It has been proposed that this volcanic event caused paleoenvironmental changes and a major genetic bottleneck of hominin populations in the Late Pleistocene, resulting in populationlevel extinctions (Ambrose 1998). The effects of the Toba supereruption on hominin populations in India are currently being examined through ongoing field studies in the Kurnool District.

Climatic changes during the Later Pleistocene have significant implications for understanding the size and isolation of hominin populations and changes in adaptive strategies. The fragmentation of viable ecological niches and the resulting isolation of populations from one another must have affected the transmission of cultural and technological practices. At certain time periods, including oxygen isotope stage 4, it is likely that the Thar Desert, the Himalayas, and possibly the Ganges Delta would have provided barriers to hominin population movement (Field and Lahr 2005, Field, Petraglia, and Lahr 2005). Not only would this have affected the transmission of cultural information within the region but also it would have reduced the number of routes into the region for populations dispersing from elsewhere.

Current State of Knowledge about the South Asian Record

Though the fossil record of the Indian subcontinent is sparse, two hominin species are known from the Later Pleistocene. A partial cranium recovered from the Narmada River Basin, India, and dated by faunal correlation to ca. 300,000250,000 years ago (Kennedy 2000) is currently attributed to H. heidelbergensis (Rightmire 2001). Although the Narmada calvarium is associated with Late Acheulean artifacts, no hominin remains have been found with Middle Paleolithic industries. Excavations in Sri Lanka have recovered remains of H. sapiens dated to ca. 31,000 years ago at Fa Hien Cave and ca. 28,500 years ago at Batadombalena (Deraniyagala 1992). The association of these remains with microlithic industries provides the earliest conclusive evidence for humans that were both anatomically and behaviorally modern within the Indian subcontinent. Terminal Pleistocene to midHolocene human remains have been discovered throughout South Asia, providing a rich source of information on paleodemography and biocultural adaptations (Kennedy 2000).

Phylogenetic patterns from both mitochondrial DNA and the Y chromosome support the colonization of South Asia by modern humans originating in Africa (Kivisild et al. 1999a, 1999b, 2003; Metspalu et al. 2004; QuintanaMurci et al. 1999). As in the majority of mitochondrial DNA variation outside of Africa, South Asian lineages belong to haplogroups M and N (groups U and R being major subclades of N), thought to be descended from the L3 haplogroup that arose in Africa ca. 85,000 years ago (Forster and Matsumura 2005, Metspalu et al. 2004). While coalescence dates for haplogroup M, which is shared by most nonEuropean populations, average between 73,000 and 55,000 years before present (Kivisild et al. 2000), the geographic origin of the M lineages remains uncertain. The high numbers of Indiaspecific M lineages have led to the suggestion of a Southwest Asian origin (e.g., Richards et al. 2003; Roychoudhury et al. 2000, 2001), although the presence of the M1 lineage in some African populations means that an African origin cannot be rejected (Metspalu et al. 2004). Currently it is unclear whether the African M1 lineage represents an ancestral M population or a backward migration. The fact that African M is nearly exclusively found in AfroAsiaticspeakers may suggest a younger presence in Africa (Forster 2004). These issues illustrate some of the difficulties in reconstructing past population movements from the genetic data. Understanding the location of the origin of haplogroup M is, however, crucial to dating the initial dispersal of anatomically modern humans from Africa.

A number of mtDNA lineages (specifically U2i, M2, and R5) share coalescence dates of 50,00070,000 years ago (Kivisild et al. 2000, Metspalu et al. 2004) and may represent an Indiaspecific subclade related to the initial dispersal of modern humans into the peninsula. Such early coalescence dates are supported by those recently obtained for the Andamanese M31 and M32 and Malaysian M21 and M22 lineages (Macaulay et al. 2005, Thangaraj et al. 2005) and may support the possibility that modern humans arrived in South Asia during the Middle Paleolithic. Coalescence dates can give only upper estimates for the timing of such dispersals, as the dispersing population may have contained DNA sequences that had already diverged (Nei and Kumar 2000). Strictly speaking, the initial dispersal of modern humans into South Asia may have occurred at any point within the past 70,000 years. Crucially, however, such early coalescence dates raise the possibility that the earliest Homo sapiens fossils recovered from the region may not have been the first anatomically modern humans to have reached the subcontinent. Given the lack of any fossils from the Indian subcontinent that date to between 250,000 and 31,000 years ago, the use of the early fossils from Sri Lanka as evidence for a definitive date for such a dispersal is a misuse of the paleoanthropological record. The dates of the Sri Lanka fossils provide the latest date for the colonization of the region by H. sapiens.

Whatever the timing of the dispersal of anatomically modern populations into the subcontinent, the route or routes of this population movement remain controversial (Forster and Matsumura 2005). The great time depths apparent in the Andamanese and Malaysian mtDNA have been argued to support the rapid colonization of the region as part of a southern, coastal route to Australia (Endicott et al. 2003b, Macaulay et al. 2005, Thangaraj et al. 2005), though some of this evidence has been disputed (Cordaux and Stoneking 2003). The phylogeography of mtDNA haplogroups has also been argued to support a southern, coastal route to South Asia (Metspalu et al. 2004). Genetic research has indicated dispersals from the Indian subcontinent towards Southeast Asia (Majumder 2001, Roychoudhury et al. 2000). A much later, terminalPleistocenetoearlyHolocene dispersal from South Asia to Australia has been inferred from the analysis of Ychromosome data (Redd et al. 2002). It has been suggested that a single dispersal event and hence Asian origins for the initial modernhuman colonization of Europe is the most parsimonious explanation for the majority of the genetic evidence. Some aspects of the Y chromosome and mitochondrial genome may suggest a later Upper Paleolithic dispersal to Europe via the Levant, but genetic support for an earlier northern route to Europe seems to be lacking (Kivisild et al. 2000, Metspalu et al. 2004). The contradiction between the most parsimonious explanation for the genetic evidence and that for the morphometric evidence is interesting and may suggest that further work is needed for an understanding of the correlation between phenotypic and genetic change with regard to modern human diversity.

Proponents of the southern dispersal route have argued for a rapid initial dispersal around the South Asian coastline (e.g., Oppenheimer 2003), but the mitochondrial DNA evidence suggests expansions of modern human populations within South Asia close to its initial colonization (Kivisild et al. 1999b, Metspalu et al. 2004). A further demographic expansion is suggested by the coalescence of a number of Indiaspecific M lineages to 20,00030,000 years ago (Kivisild et al. 1999b, Metspalu et al. 2004) and may be related to other demographic expansions within the Old World and the more stable climate of oxygen isotope stage 3 (Forster 2004). Intriguingly, recent work has indicated a genetic continuum between the Near East and India coalescencedated to between 50,000 and 30,000 years ago. By 20,000 years ago the window of this continuum had closed, perhaps representing the isolation of populations due to the increased aridity of western India and Iran as the Last Glacial Maximum approached (Metspalu et al. 2004).

Archaeology offers insights into the population history of South Asia, as more than a century of research has revealed a rich Later Pleistocene record. While most attention has been focused on the Lower Paleolithic of the region (e.g., Petraglia 2001), the Later Pleistocene record is abundant, revealing information about hominin adaptations through time (e.g., Misra 1989, 2001; Paddayya 1984). Later Pleistocene sites are known mainly from surface surveys (e.g., Gudzer 1980, Joshi et al. 197980), some of them providing vital information on settlement patterns and landscape use (e.g., Pappu and Deo 1994, Raju 1988). Several significant excavations of Later Pleistocene sites have provided important information on chronology and changes in paleoenvironments, settlement, and technology (e.g., Joshi 1978; Misra 1985, 1989). Serious methodological problems do pervade archaeological investigations in South Asia, where little attention has been paid to highprecision fieldwork and artifact analyses. Unfortunately, most studies have been conducted without consideration of the role of postdepositional processes in contributing to site formation, and few studies have examined the spatial distribution of material remains to identify hominin activities across living surfaces (Petraglia 1995). Archaeological investigations carried out in the Kortallayar Basin are a notable exception (Pappu 1999, 2001a, b; Pappu et al. 2003).

Chronology

Dating by relative and absolute methods has helped clarify the development of the series of different industrial complexes produced during the Later Pleistocene. While some dates are available, the precise boundaries of the lithic industries are not well known. The 16R Dune in Rajasthan supplies valuable information about the dating of archaeological assemblages (fig. 2). Thermoluminescence, uraniumseries, and radiocarbon dates have been gathered from a 19meter sequence. Here a typologically nondiagnostic sample of artifacts has been dated to > 390,000 years ago. Archaeological assemblages interpreted as Middle Paleolithic industries are dated to ca. 150,000 years ago (fig. 3), whereas those identified as Upper Paleolithic are dated to 26,210 years ago (Misra 1995a).

Most other dates are from single samples or from contexts that are not part of long stratigraphic profiles. Assemblages identified as Middle Paleolithic have been difficult to date on account of their contexts and the limitations of the chosen chronometric methods. A Middle Paleolithic scraperbased industry from Patpara in the Middle Son Valley is dated to < 103,000 years ago (Blumenschine, Brandt, and Clark 1983, Williams and Clarke 1995), while dates of 75,000 and > 60,000 years ago are associated with artifacts recovered from Samnapur (Narmada Valley) and Balotra (Luni River valley) (Mishra et al. 1999, Misra et al. 1990). The dating of miolites in the Hiran Valley places assemblages classed as Middle Paleolithic at 69,00056,000 years ago (Baskaran et al. 1986). The earliest assemblage classified as Upper Paleolithic is currently Site 55, Pakistan, where the loess overlying the occupational horizon has been dated to ca. 45,000 years before present (Dennell et al. 1992).

In India, a number of assemblages identified as Upper Paleolithic have been dated to between 40,000 and 20,000 years ago, including those from the sites of Mehtakheri, Inamgoan, Chandrasal, Dharamouri, and Nandipalle (Mishra 1995). The microlithic assemblages from Batadombalena, Sri Lanka, date from 28,500 years ago, and two other microlithic assemblages from Sri Lanka, those of Site 49 and Site 50, date to 28,000 years before present (Deraniyagala 1992). Overall, there is a paucity of chronometric information, especially in comparison with other regions (e.g., Western Europe), which have thousands of available dates. Moreover, there is little confidence in some of the chronometric results, as dates have not been verified by independent methods or by robust sampling procedures.

Relative sequences of assemblages have been obtained from stratigraphic contexts in individual sites and across landforms such as river valleys. Excavated sites in the Thar Desert (Singi Talav and IndolakiDhani) (Misra et al. 1982) and in the Kaladgi Basin (Lakhmapur East and West) (Petraglia, Schuldenrein, and Korisettar 2003) indicate the presence of early Middle Paleolithic industries overlying the Late Acheulean. Initial work in the Kortallayar Basin has produced a relative sequence of sites from the Acheulean to Upper Paleolithic (e.g., Pappu 2001a). A relative sequence of excavated AcheuleantoMesolithic sites has been compiled for the Middle Son and Belan Valleys (Clark and Williams 1990, Sharma and Clark 1983). The most important site sequence is that of Bhimbetka III F23, where 3.8 meters of excavated deposit have produced artifacts from the Late Acheulean to the Mesolithic (Misra 1985). The excavated sequence at Patne contains late Middle PaleolithictoMesolithic assemblages in a 10meter section (Sali 1989) (fig. 4).

Later Pleistocene Technology and Industry Characteristics

South Asian hominin populations of the Later Pleistocene produced two distinct stone tool industries. These assemblages were originally classified as Middle Stone Age (MSA) and Late Stone Age (LSA) (e.g., Allchin 1959; Sankalia 1964a, b), apparently as a consequence of their technological affinity with African assemblages. Although these assemblages were later termed Middle Paleolithic and Upper Paleolithic, the Upper Paleolithic designation remains problematic, as, despite claims of close similarities, it is not coincident with its European counterpart. Flakebased artifact assemblages consisting of prepared cores, retouched flakes, and diminutive bifaces generally characterize the Middle Paleolithic of the subcontinent (e.g., Jayaswal 1978, Paddayya 1984). Following the Middle Paleolithic, bladebased and microlithic industries appear to become increasingly important. The socalled Upper Paleolithic industries are variable in composition but demonstrate an increase in the production of burins and backed tools at some sites (Murty 1979, Paddayya 1984). There is considerable temporal and spatial variation in the appearance and frequency of these flake and bladebased industries, and the technological shift does not appear to have been rapid.

The Middle Paleolithic

The majority of South Asian Middle Paleolithic industries are produced on flakes struck from prepared cores. Technological studies in the Kaladgi Basin (fig. 5) suggest that the use of preparedcore methods developed from the preceding Acheulean (Petraglia, Schuldenrein, and Korisettar 2003). It therefore forms part of a growing body of global evidence (e.g., Petraglia and Alsharekh 2003, White and Ashton 2003) which supports the local, convergent evolution of preparedcore technology. During the Middle Paleolithic there does not appear to be a single or favored technique of core preparation throughout the subcontinent. Levallois and discoidal techniques are the most commonly identified, but other core types have been described (e.g., cylindrical) (Misra 1967, 1968). Technological diversity is also indicated by the presence of unprepared cores in Middle Paleolithic assemblages. Numerous sites show evidence of the use of a wide range of core reduction techniques, including Hajiakheri (Misra 1968), Lahchura 2 (Pant 1982), and Attirampakkam (Pappu 2001a). Flakes may also be derived from natural spalls and from amorphous cores (Pappu 2001a). Many of these coretype designations have yet to be supported by detailed reductionsequence analysis, but studies have suggested that there is variability in core reduction sequences within the region (Jayaswal 1978). The frequency of different methods of core preparation shows spatial variation across the subcontinent (James 2003). In the river valleys of Uttar Pradesh (Pant 1982) and the Kortallayar Basin (Pappu 2001a) the Levallois technique dominates, but in the Wagan and Kadmali River basins (Misra 1967, 1968) the use of the discoidal technique is far more common.

Middle Paleolithic industries from Sri Lanka and Nepal appear to differ from those recovered in other parts of South Asia. Evidence for premicrolithic industries in Sri Lanka is sparse (Deraniyagala 1992). In Nepal, the site of Arjun 3 (DangDeokhuri Valley) has produced a Levalloisbased industry containing scrapers, points, and blades that is older than ca. 30,000 years (Corvinus 1994, 1995, 2002). This industry is succeeded by industries consisting of unifacial choppers produced on large cobbles. These later industries lack any evidence for core preparation (Corvinus 1994).

Excluding the later sites in the Nepalese sequence, there appears to be a notable blade and flakeblade component to the South Asian Middle Paleolithic. Blade and flakeblade cores are documented at localities such as Patpara (Blumenschine, Brandt, and Clark 1983), the Kortallayar Basin (Pappu 2001a, b), and Bhimbetka (Misra 1985) (fig. 6). The unidirectional cores recovered from the Thar Desert region at sites such as Hokra 1a and Gurha (Allchin, Goudie, and Hegde 1978) appear to indicate the production of blades as blanks. The blades are not struck from prismatic cores but appear to be the intended product. Typically a single flake is removed from a core, providing a platform from which a small number of blades and flakes can be struck. The cores do not appear to be systematically reduced, with only a small number of flakes and blades being removed. The blades and bladelike flakes represent the majority of the blanks struck from the core, suggesting that the intention of the cores was to produce narrow, elongated flakes. Given the paucity of chronological data for the period, determining changes in core reduction strategies over time is difficult. Initial comparisons suggest, however, that the technique may have become more developed as the Middle Paleolithic progressed. Cores from the advanced Middle Paleolithic assemblage at Patne indicate a similar process of blade production. They differ from those of earlier sites by reduction intensity, with more blades (and indeed flakes) removed from each core (fig. 7). While they typically form a small proportion of total blank production, the blades and bladelike flakes produced by these methods are used in the manufacture of a number of finished tools, including various scraper forms. Such bladebased tools are noted from the Middle Paleolithic industries of Chancha Baluch (Allchin, Goudie, and Hegde 1978), the Panchmahals (Sonawane 1984), the Godavari Valley (Joshi et al. 197980), Bhagi Mohari (Paddayya 198283), and other localities.

In terms of assemblage composition, the South Asian Middle Paleolithic exhibits variation in the presence or absence of different tool types and their relative frequencies (James 2003). While scrapers are the dominant tool form in much of the Indian subcontinent, there are sites in, for example, the Kurnool District and the Gunjana Valley where points have been identified as increasingly important (Raju 1988). This seems to fit into a general pattern of variation in which points are much less common in northwestern and northcentral regions than they are in the southeast. Points, for example, are particularly poorly represented at Bhimbetka III F23 (Misra 1985), Patpara (Blumenschine, Brandt, and Clark 1983), and the Upper Son (Ahmed 1984). Similarly variable is the presence of diminutive handaxes within Middle Paleolithic assemblages, though they are distributed throughout the subcontinent. No clear geographical patterning can be observed in their distribution, and the chronological resolution is currently lacking to test the assumption that they are present only in early Middle Paleolithic assemblages (Paddayya 1984). Tool types identified as knives and borers are rarer in Middle Paleolithic assemblages, though their presence in Bhimbetka indicates that they are a component of at least some Middle Paleolithic industries (Misra 1985). Denticulates, notches, and, rarely, burins have been noted from a small number of sites, including Parsidhia (Uttar Pradesh) (Pant 1982), Mangalpura (Didwana) (Misra et al. 1982), and Bhimbetka (Misra 1985). Chronological change may explain some of the variation in the occurrence of these rarer tool forms. Polyhedrons are found in earlier Middle Paleolithic sites, such as IndolakiDhani and Singi Talav, Didwana (Misra et al. 1982), but are lacking from the assemblage in Patpara (Blumenschine, Brandt, and Clark 1983). Tanged points, though rare, have been noted from a number of assemblages, including the Upper Son Valley (Ahmed 1984) and Ramayogi Agraharam (Rath, Thimma Reddy, and Vijaya Prakash 1997). Formalized retouch toward the production of particular tool types is not a general characteristic, and therefore the use of a generalized Middle Paleolithic terminology is appropriate.

The Late Paleolithic

From approximately 45,000 years ago, increases in blade production and variability in assemblage composition characterize the South Asian archaeological record. Traditionally classified as Upper Paleolithic and divided into flakeblade, bladebased, and blade and burin industries (Murty 1979), such assemblages are fewer than those identified as belonging to the preceding Middle Paleolithic. In addition, South Asian microlithic industries, often referred to as Mesolithic, date to 28,500 years ago, indicating that they may be part of a diverse Later Pleistocene package. We term this package the Late Paleolithic in order to emphasize its differences from both the LSA of Africa and the Upper Paleolithic of Europe.

Large, thick blades (Paddayya 1984:353) associated with scrapers, borers, and points produced mainly on flakes characterize industries such as those from Singhbum and Watru Abri (Murty 1979, Paddayya 1984). Scrapers on flakes are associated with small backed blades at the excavated site of Mehtakheri (Mishra n.d.). Both macro and microblades and cores are reported from Visadi (Allchin 1973). At Site 55, Pakistan, flakeblades are associated with blades small enough to be classed as microliths (Dennell et al. 1992). In contrast, the Sri Lankan sites of Batadombalena, Site 50, and Site 49, dating from 28,500 years ago, contain industries that are based on the production of geometric microliths (Deraniyagala 1992) (fig. 8). In sites such as Patne, geometric microlith technology develops from an industry characterized by a few backed blades and burins (Sali 1989). However, Patnes microlithic industries date to ca. 24,500 years ago, slightly postdating their early appearance in Sri Lanka. In addition, contemporary sites from the Indian subcontinent such as Inamgaon (ca. 25,00021,000 years ago) have produced industries consisting of scrapers, blades, points, and fluted cores in which the production of backed blades is minimal (Murty 1979). With the exception of the geometric microliths, the standardization of retouched artifact forms is not comparable to that seen in the Aurignacian and later industries of Upper Paleolithic Europe.

Research is continuing into the differences in core reduction strategies that may characterize these different industries, but initial comparisons between cores within the Patne sequence suggests that there is no sudden shift to classic prismatic cores at the onset of the Late Paleolithic. Cores from the early Upper Paleolithic strata are comparable in technique to those from the advanced Middle Paleolithic from the same site but exhibit a greater number of flake and blade removals (fig. 7). Prismatic blade cores are present in the Late Paleolithic assemblages from Patne and increasingly dominate the assemblages as the period progresses. These prismatic cores are small and seem to have been used in the production of blades, microlithic blades, and bladelets. It is interesting that these levels also include geometric microliths such as lunates and triangles.

There is no clear sequence of industrial subdivisions within the Late Paleolithic of South Asia. There is, however, contemporaneity between flake and bladebased technology and microlithic industries after ca. 28,500 years ago. The microlithic industries in South Asia are earlier than those seen in Europe, though they postdate the appearance of microlithic technology in Africa. At the moment, however, the finds at sites such as Patne and the presence of microliths in other Late Paleolithic assemblages from the subcontinent suggest that at least some of these early microlithic industries developed regionally rather than resulting from a dispersal from elsewhere. The similarity in core reduction techniques between the Middle and Late Paleolithic industries at Patne is intriguing. Transitional Middle to Late Paleolithic industries have been reported at sites such as Chancha Baluch (Allchin, Goudie, and Hegde 1978). This evidence suggests that at least some Late Paleolithic industries developed from the Middle Paleolithic.

Symbolism, Structures, and Modern Human Behavior

The majority of the evidence for modern behaviors and symbolism in South Asia dates to no earlier than 28,500 years ago, although there are several notable exceptions. There is possible evidence for early ochre use in the form of a number of fragments of haematite in Acheulean sites in the Hunsgi Valley, including a striated crayon (Bednarik 1990, Paddayya 1982). The early use of ochre within South Asia may not necessarily be symbolic. Another claim for nonutilitarian use of materials are six quartz crystals associated with Acheulean materials at Singi Talav (dErrico, Gaillard, and Misra 1989). Though petroglyphs have been identified on cave and rockshelter walls and boulders, including a cupule and groove on a boulder in Acheulean levels of Auditorium Cave, Bhimbetka (Bednarik 2003, Kumar 1996), their interpretation as some of the earliest rock art remains controversial.

Although the earliest evidence for symbolism is sparse, it is clear that by the terminal Pleistocene South Asian populations were selecting and using various materials to manufacture objects. Bone tools have been recovered from a number of Sri Lankan microlithic contexts, of which the earliest is the 28,500yearold assemblage at Batadombalena (Deraniyagala 1992). Though bone artifacts, including possible pendants, have been reported from the Late Paleolithic contexts of the Kurnool region (Murty and Reddy 1975, Thimma Reddy 1977), the artifactual nature of some of these objects has been called into question (Petraglia 1995). A number of Later Pleistocene sites are associated with ostricheggshell fragments (Kumar et al. 1988, 1990). The earliest evidence of adornment is the ostricheggshell beads recovered from the 28,500yearold horizon at Batadombalena (Deraniyagala 1992) and the 25,000 200yearold strata at Patne (Sali 1989). The same level at Patne also produced a geometrically incised fragment of ostrich eggshell, the earliest deliberate artifact decoration from peninsular India (fig. 4). Recent excavations have discovered a Late Paleolithic bead production site in Madhya Pradesh (Mishra, Ota, and Naik 2004). One other possible example of Late Paleolithic art is a carved and polished goddess figurine from the Belan Valley of Uttar Pradesh (Misra 1977), which has recently been reinterpreted as part of a bone harpoon point (Bednarik 2003). A burial of a late modern human at Bhimbetka contains two ostricheggshell beads found near the neck (Bednarik 2003).

Evidence for the deliberate construction of structures is rare in South Asia. The earliest evidence consists of the stonelined pit and low wall reported from Site 55, Pakistan, and dated to 45,000 years ago (Dennell et al. 1992). Structures have also been reported from Bhimbetka (Misra 1989). A sandstone platform and curiously patterned rock discovered at the Late Paleolithic site of Baghor I have been interpreted (via the use of ethnographic comparisons) as the earliest mother goddess shrine in the subcontinent (Kenoyer et al. 1983).

The evidence for symbolic thought and, indeed, for the majority of the traits cited as evidence for modern human behavior is relatively sparse. The population of the region during the Late Paleolithic was clearly capable of explicitly symbolic behavior, leaving little doubt that the populations that produced these artifacts and sites were behaviorally modern, but there is no symbolic revolution of the kind that accompanies the Aurignacian in Europe. While differential survival (and, indeed, recovery) of artifacts may well play a role in the discrepancy, this cannot be the whole story. Given the considerable archaeological research that has occurred in South Asia during the past century, it seems reasonable to argue that the Late Paleolithic of the Indian subcontinent is unique. Despite the current Upper Paleolithic label, it is as different from the Upper Paleolithic of Europe as it is from the LSA of Africa.

In sum, modern human behavioral traits such as symbolic thought and the construction of structures appear to date to after ca. 45,000 years ago. A range of novel technologies characterizes the Late Paleolithic, including the manufacture of bone tools and geometric microliths by 28,500 years ago. The technological variability within this period needs to be further investigated, but the evidence suggests a diversification of modern human behaviors during the climatic oscillations of the Terminal Pleistocene. Other technological innovations, such as deliberate blade production, variability in flake reduction strategies, and geographical and chronological variability in the presence of retouched tool types, date to the Middle Paleolithic. The relationship between assemblage composition and resource availability in this period remains to be fully investigated. Chronological, geographical, and possible stylistic variation between lithic assemblages has been interpreted as signifying the presence of behavioral modernity within the MSA of Africa (McBrearty and Brooks 2000). The presence of similar variation within South Asia (James 2003) as well as Europe and the Near East suggests that such behaviors may not be uniquely human. The South Asian evidence for early deliberate blade production is consistent with the growing evidence for blade manufacture by hominin populations other than H. sapiens (BarYosef and Kuhn 1999). The ability to produce a flexible technology that could be adapted to different ecological niches and tasks (which is presumably what such chronological and geographical variation would represent from an evolutionary perspective) was not limited to early anatomically modern humans. Instead, it formed part of the behavioral repertoire of related species such as the Neanderthals and the archaic hominin populations of South Asia.

Hominin Dispersals

The evidence of local evolution of preparedcore technology from Late Acheulean assemblages in South Asia is inconsistent with the hypothesis that preparedcore technology arose in Africa and spread to the rest of the world with dispersing Middle Pleistocene populations. While the archaeological evidence from the Indian subcontinent cannot be used to argue that such population dispersals did not occur, if populations (such as H. helmei or early H. sapiens) with Middle Paleolithic technologies dispersed toward South Asia from the west (Lahr and Foley 1998, 2001) they would have encountered local populations (H. heidelbergensis or an unknown species) using Middle Paleolithic technologies.

A similar local development of technology seems to be indicated when the transition from the late Middle Paleolithic to the Late Paleolithic is considered, at least for some of the assemblages of the subcontinent. Given the date at which this transition appears to have occurred (ca. 45,00028,500 years ago), it seems logical that the replacement of archaic hominin populations by behaviorally modern H. sapiens should be marked by some distinct archaeological signal. Yet the archaeological record currently suggests no abrupt technological changes from the Late Acheulean through to the microlithic industries of the Terminal Pleistocene.

Multiple dispersals of anatomically modern humans from Africa remain the most parsimonious explanation for modern human diversity (e.g., Lahr and Foley 1994). H. sapiens colonized Australia at least 45,00042,000 years ago (e.g., OConnell and Allen 2004) and possibly earlier (e.g., Bowler et al. 2003). Archaeologists have also begun to document coastal occupations along the African (Walter et al. 2000) and Arabian (Petraglia and Alsharekh 2000) coasts, and, as indicated, South Asia has coastal and nearcoastal sites that may be related to such dispersal events. This evidence increases the plausibility of the hypothesized dispersal of modern human populations from Africa and the initial colonization of South Asia and Australia via a southern or coastal route (e.g., Cann 2001, Lahr and Foley 1994, Stringer 2000). The genetic evidence is consistent with such an early dispersal into South Asia, though it cannot directly support it (e.g., Kivisild et al. 2003, QuintanaMurci et al. 2001, Redd and Stoneking 1999). Our hypothesis is that this initial dispersal reached South Asia during the Middle Paleolithic. Given that the earliest modern humans outside of Africa, at Qafzeh and Skhl Caves in the Levant, were undoubtedly using a Middle Paleolithic technology, it is perhaps not surprising that the archaeological signal for such a dispersal is difficult to detect. Given the large size of the landmass and the use of similar technology, it is possible that the replacement event took longer in South Asia than elsewhere.

The Later Pleistocene archaeological record of the Indian subcontinent indicates increasing technological diversity. A number of innovations occur in the Later Pleistocene, such as the shift to preparedcore technology, an increasing dependence on blade production, microliths, specialized tools, and selfadornment, but these appear gradually and sporadically. They do not provide a marker for the dispersal of anatomically modern humans. There is no Upper Paleolithic revolution like that seen in Europe (e.g., Mellars 2005) or even a clear change in technology coincident with the arrival of modern humans like that seen in East Asia (Gao and Norton 2002). The timing of some of these behavioral changes does, however, correspond with genetic evidence of demographic expansions in South Asia. The first production of explicitly symbolic artifacts (i.e., beads and art), the use of bone, and the first microlithic assemblages appear between 30,000 and 20,000 years ago, coinciding with a major expansion of human populations in the Indian subcontinent (Kivisild et al. 1999b).

Conclusion

As currently understood, the archaeological record from Later Pleistocene South Asia has much to contribute to our understanding not only of what constitutes the material evidence for behavioral modernity but also of the ways in which culture is transformed by demographic processes. The hominins that occupied South Asia during the Later Pleistocene left behind a technologically diverse archaeological record that undoubtedly represents a palimpsest of different, flexible adaptive strategies to variable ecological niches. Evidence for the early intentional production of blades, early microlithic industries, and technology that varies over both time and space provides an intriguing glimpse of the way in which populations coped with fluctuating and often challenging environmental conditions. By the Last Glacial Maximum these populations were undoubtedly both anatomically and behaviorally modern, but the modern human behaviors exhibited were expressed in a way unique to the region. Overt symbolism, in terms of art and selfadornment, is relatively rare and appears relatively late in the record, coincident with evidence for population expansion within the subcontinent. Other aspects of modern behavior, such as structured site use, appear somewhat earlier. The roots of some of the more functional aspects of modern behavior may date to the Middle Paleolithic. These behaviors are also represented in the Levant and in Europe during this period and are plausibly interpreted as something that H. sapiens shared with closely related species. Because they are not unique to our species they should not, in our opinion, be considered as markers for behavioral modernity.

Modern humans colonized South Asia as part of an as yet undated expansion of H. sapiens from Africa. If the modern behavioral package can be taken as a marker for such dispersals, then the South Asian record should logically show similarities to the archaeological record of Europe rather than to Africa. But in contrast to that of Europe, this record contains no dramatic appearance of a technology such as the Aurignacian that fulfills the fully modern behavior criterion and could be linked to the expansion of H. sapiens. There is no clear evidence in the South Asian Middle Paleolithic for an early dispersal such as that suggested by some discussants of the genetic data (e.g., Oppenheimer 2003). Yet neither do the Late Paleolithic and the Indian subcontinents precocious microlithic industries constitute a sudden break. The mosaic of industrial components suggests a gradual shift to intensive blade production, of which the development of microlithic technology seems to be a part. The increasing predominance of blades over flakes is related to the intensification of methods with their technological roots in the Middle Paleolithic. Until the appearance of geometric microliths, the degree of artifact standardization remains debatable, and, indeed, explicitly symbolic artifacts are rare.

A current theme in debates regarding the origin of modern human behavior is that explicit symbolism and complicated resource acquisition (expressed in the archaeological record as multicomponent tools, evidence of food not exploited before, and longdistance exchange networks) are perhaps the best indicators of behavioral modernity (e.g., Henshilwood and Marean 2003). The evidence for complex resource acquisition or at least the date of its origin remains controversial, even within Africa (Klein 2000 contra McBrearty and Brooks 2000), and further research is needed before answers can be ascertained from the South Asian record. Symbolic thought or at least evidence for explicit symbolism appears later in the South Asian record than in Africa or Europe, and when it does it appears gradually. This situation is much more like the African MSA, where Blombos Cave (e.g., Henshilwood et al. 2002, 2004) constitutes the best and some would argue the only evidence for symbolic thought associated with early anatomical moderns, than the sudden explosion that is seen with the arrival of the Aurignacian in Europe. This gradual appearance in South Asia has significant implications for the way we define behavioral modernity.

If symbolic thought is indeed the best indicator of the presence of a modern brain, then the appearance of explicitly symbolic artifacts should serve as a proxy signal or marker for the arrival of H. sapiens in a given region. Such artifacts appeared gradually in Africa because this is where the modern mind likely evolved. The European record, with its sudden technological and symbolic revolution (Mellars 2005), represents the dispersal into the region of hominins with fully modern minds. But, as we have seen, the evidence for symbolic thought does not appear suddenly within South Asia. There are two possible reasons that this is the case.

First, it is possible that the members of our species that first colonized South Asia were not behaviorally modernthat they were incapable of fully symbolic thought. If behavioral modernity is indeed the result of a reasonably late neurological change (e.g., Klein 2000), then this is precisely what would be expected if the initial dispersal of modern humans into the area occurred prior to ca. 50,000 years ago. An early dispersal of modern humans into the Indian subcontinent is, in our view, the most parsimonious explanation of the available evidence, but on its own it cannot explain the eventual presence of behaviorally modern humans in the region. Either more cognitively able members of the same species replaced such initial colonizers after ca. 50,000 years ago (in which case we might expect some kind of archaeological marker) or such genetic/mental change occurred synchronously within H. sapiens populations in Africa and South Asia. Both of these explanations appear highly problematic. In our view it makes more sense to see the Middle Paleolithic colonizers of South Asia as both anatomically and cognitively modern.

The second possibility is equally applicable to an early dispersal into South Asia or a later one. Ethnographic studies have indicated that symbolic thought is one of the suite of behaviors that may well be unique to humans, as reflected by our need to create and manipulate identities and our incorporation of material culture into that process (e.g., Hodder 1982). But they also show that this behavior is not always expressed in the same way. To oversimplify, different cultures use different aspects of material culture in identity construction. Symbolic thought and, by extension, modern human behavior are not just represented by art or beads. Thus, while the appearance of such objects in the archaeological record indicates the presence of a modern mind, the lack of such objects cannot be taken to mean the absence of such a mind. The most parsimonious explanation for the absence of a correlation between the arrival of the modern humans in South Asia and the explosion of symbolic thought is that the latter is not being expressed in the form that we are expecting.

If this is indeed the case, the Upper Paleolithic revolution in Europe and the sporadic use of overtly symbolic artifacts in MSA Africa and, later, in South Asia are the result of something other than neurological change. Instead, it is probable that they are the result of particular demographic situations. The specific nature of these situations remains to be elaborated, but it is clear that population size plays a role in cultural change. By affecting the relative importance of natural selection or cultural drift, changes in population size will affect changes in material culture (Shennan 2000, 2001). It may well be that signaling identity (whether individual or group) through the use of ornaments or other forms of art is adaptive in situations in which there is competition for resources. But if cultural drift plays a bigger role in determining the contents of material culture (as it may well do in a small population) than natural selection, the behavior may not be expressed or may be expressed in another, less adaptive way (Shennan 2000).

The Later Pleistocene is characterized by climatic oscillations that must have influenced both the skills hominins needed to survive and their population size, but the hominin populations that inhabited South Asia were able to endure them. It is likely that marginal populations were forced to disperse or decrease in size during periods of resource scarcity. Especially harsh conditions may have contributed to the extinction of such populations. Crucially, however, such processes of extinction, growth, and dispersal must be viewed at a population rather than a species level. While climatic change affected population movements within the region, it would also have influenced the likelihood of population movement between South Asia and the rest of the Old World. Such demographic fluctuations were not restricted to South Asia, as the genetic evidence clearly shows (Forster 2004). Given the effect of demographic processes on cultural evolution and innovation, archaeologists have to be careful in selecting the traits considered representative of behavioral modernity. Until we understand how these processes influence material culture, creating a trait list of modern behaviors extrapolated from presentday huntergatherers or based on any one regional record is fraught with difficulty. To ascertain what makes modern humans unique from a behavioral perspective we need to understand the cognitive and behavioral capabilities of our closest relatives, including the archaic hominins that inhabited South Asia.

Further highquality investigations are needed if such hypotheses are to be tested, and it is vital that this research continue not only in Europe and Africa but also in other areas of the world. This synthesis has indicated the enormous potential of South Asia for contributing to investigations into modern human origins and the evolution of behavior and the need to place the fossil, genetic, and archaeological records in a global framework. Ongoing research in South Asia may help to unravel the processes that are part of the modern behavioral package.