A review of A Pleistocene Passage to India: The Paleoanthropology of Early Human Settlement in Coastal Western India, by August Gerald Costa
A commonly held premise in archeological research posits that the further back in time that one looks, the bigger the geographic gaps in our knowledge tend to be. When it comes to the first settlement of new continents as modern Homo sapiens spread around the world, replacing ‘archaic’ antecedents, Neanderthals, etc., the settlement of South Asia remains, as Costa notes, “among the most critical unresolved problems in South Asian anthropology.” (p. xi). The so-called Southern Dispersal Hypothesis posits that initial human colonization of South Asia occurred from Africa along coastal routes c. 50,000 BP, but contextualized paleoanthropological evidence remains sparse on the ground, a problem compounded by the distinct possibility that sites of appropriate age and location have not been discovered yet. In a dissertation composed of seven parts, Costa thoroughly reviews the material cultural and faunal evidence recovered from each site identified and sampled, and uses this new information to provide the first comprehensive picture of the chronology, technology and environment of this key Late Pleistocene coastal corridor.
Part I (three chapters) provides the theoretical, historical, geographic and climatological framework needed to understand the overall goals The Gujarat Paleoanthropology Project (hereafter the GPP). The chapters are together primarily concerned with defining and discussing the controversy surrounding the Southern Dispersal Hypothesis (hereafter SDH) as it relates to the Late Pleistocene settlement of South Asia c. 125,000-40,000 BP. In the broader context of clarifying the debate between multiregional and recent “out of Africa” dispersal models of recent human evolution, Costas notes the potential of South Asia as a “proving grounds” for the Southern Dispersal hypothesis. Several variants of the Southern Dispersal hypothesis are evaluated, differentiated by dispersal scenarios before or after the Toba super-eruption c. 74 Ka, or via the nature of associated lithic technology. If the disappearance of miniature bifaces c. 100 Ka indicates the extinction of AMHS (anatomically modern Homo sapiens) during India’s Late Acheulean-early Middle Paleolithic transition, as some have suggested (p. 4), pre-Toba fossil hominid evidence in India or elsewhere is lacking.
Within approximately 196,000km, Gujarat is divisible into distinct coastal, desert and upland environmental zones, dependence on the SWIM (Southwest India Monsoon) today as it would have been over the last approximately 15 million years (despite periods of fluctuation and disruption that would have affected hominid and human dispersal and survival). The Kutch is an extremely arid saline wasteland with associated highlands, and previous work has revealed several Harappan and Paleolithic sites from alluvial settings. Mainland Gujarat consists of a northern arid alluvial plain and a southern sub-humid north-south coastal belt with ample rainfall, bounded by eastern highlands. The rich mangrove dominated ecosystems of the region’s Makran coast would have provided a challenging environments to dispersing hominids. Only in Gujarat have archeological investigations of this area proved possible for Western researchers. Peninsula Gujarat (Saurashtra) is the largest sub-region of Gujarat. It is bordered by the Gulf of Cambay, Gulf of Kutch, extensive salt flats and the Indus Delta. The Gulf of Cambay can be considered its own separate region, given its size and tidal extent.
The final chapter in Part I discusses the history of prehistoric research in Gujarat, noting the paucity of Palaeolithic research as opposed to later Mesolithic, Chalcolithic and Harappan investigations, especially after the 1970s, early attempts at collaboration with Western scholars yet a continued reluctance to view even their Paleolithic prehistory in terms of “invasion” (as one iteration of the SDH suggests), problems associated with inconsistent definitions and nomenclature used to describe the SDH, and the need to “ground truth” the broad genetic underpinnings of the SDH with new archeological fieldwork and reanalysis.
Part II (four chapters) discusses the field localities and sampling strategies Costa utilized during GPP fieldwork from 2010-2011. For his research, Gujarat was divided into four distinct zones; the Bhadar Valley (western Saurashtra), Bhavnagar District (eastern Saurashtra), the Coastal Plain (southern Saurashtra) and Mainland Gujarat. The geographical and geological and history of quaternary research for each zone are described in more detail, before survey and excavation strategy is covered. In all instances, site survey was very broad and more geological in approach than typical archeological surveys due to the need to locate exposed fluvial sections of appropriate age; importantly noted to not represent discrete spatial and temporal loci of Paleolithic hominin behaviour, but rather aggregates (p. 48). In the Bhadar Valley, a dozen sites were identified; some new to science and some known from previous surface collection or excavation, reanalysed or re-excavated as part of the GPP.
In the Bhavnagar district area of southeast Saurashtra, two riverine locations were examined; one (Shetrunji) representing new exploration, while the other (Kalubhar) including a site known since the 1970s. Exploration of these sites was opportunistic on the way to and from Bhavnagar District and MSU Baroda, the university from which local collaborators derived. The primary sites within the Coastal Plain study area include Badalpur, Madhuban and Gopnath, while Mainland Gujarat is represented by a peculiar quartz assemblage from the paleodune site of Visadi, a small assemblage from the Chelawada rockshelter, and an alluvial bank surveyed at Tejpur. All of these sites (or their vicinities) have seen previous research, so their selection was primarily to reassess past conclusions in light of new data.
Part III (six chapters) reviews the history of Pleistocene lithic technology research in South Asia and the categories and raw material diversity of tools collected during GPP survey and excavation. His attempt to locate and reanalyse previously collected assemblages is also discussed (with mixed results). To this day, it is suggested that lithic research in South Asia is still heavily weighted towards typology, making inter-regional comparisons difficult, and that most advances in understanding prehistoric manufacture and tool source come from Chalcolithic and Harappan sites (pg. 89). There is a dearth of experimental archeology in the region, let alone studies of reduction sequences, geochemical sourcing, etc. Much remains to be done.
Costa classifies all Pleistocene sites into three categories; those buried too deep, those buried in reworked contexts, and those not buried at all. Very clear diagrams and figures (here and throughout the dissertation) illustrate this concept. Thus, many important sites are submerged or washed out, leaving surface scatters or decontextualized finds in alluvial contexts. In these instances, equifinality is a concern; numerous populations could have contributed to the assemblage over time. Even the harsh arid environment can destroy many types of tool stone through dissolution.
The goal of the GPP lithic analyses was primarily to compare the characteristics of Gujarati assemblages to those from Africa and the Near East, compared to later Holocene assemblages from South Asia. Lithic “Modes” 3, 4, and 5 are focused on, covering pre and post-Toba facies. How similar to African examples are the Gujarat assemblages? The unique Visadi “Upper” Paleolithic assemblage was also reassessed. The discussion of field methodology clearly explains how lithic specimens were recovered from in-situ deposits or (when unique or diagnostic) from the surface, how each was logged with GPS and thoroughly recorded. Classification following standard Middle Stone Age typologies was conducted (according to core type, flake size and length, number of faces, shape, etc.) Debitage was treated separately, as were the vein quartz assemblages from sites such as Visadi dunes, due to their amorphous nature.
Following on from this, discussion turns to the successes and challenges of analysing previously collected assemblages, not least among them the problem of actually locating them. Unfortunately, the ~ 800 alleged lithics from five previous field expeditions do not derive from excavated contexts, with the small subsample able to be located proving minimally useful to project goals. Those able to be located consist of dubious “Middle” and “Upper Paleolithic” tools of various typologies from the Hiran Valley (igneous jasper and rhyolite), the Bhadar Valley (rhyolite), and part of the quartz assemblage collected by the Cambridge-Baroda expedition in the Visadi dunes. Reanalysis of this assemblage was done to determine if it exhibit stronger affinities with Mesolithic than Upper Paleolithic technology. A Mesolithic assemblage from Dhadar was compared to those from Visadi, due to similarity in provenience.
Results of this reanalysis were mixed. Most of the quartz assemblages were dominated by debitage, with many forms of breakage not previously noted. Unbroken flakes were rare and cores are primarily small and unifacial. No diagnostic Upper Paleolithic artifacts were confirmed. A reanalysis of cumulative weights of the Visadi flakes compared to known Mesolithic quartz samples demonstrates that they are actually smaller than expected, and overall, the Gujarat assemblages consist of smaller flakes than those from Upper Paleolithic Pakistan. The reanalysis of the Visadi assemblage itself showed marked reclassification from previous attempts, primarily due to more recognized cores, elimination of the blade category, and reduction in number of identifiable flakes.
A total of 393 flakes were collected during the course of GPP fieldwork along three river basins and from 15 localities. The majority derive from open-air sites in the Bhadar Valley, but a small crude quartz assemblage was recovered from Chelawada rock shelter, similar to those from Visadi. The Bhadar Valley assemblage underwent the most extensive analysis, with the majority of artifacts classified as whole flakes, and the assemblage overall notable for its simplicity, with few handaxes, scrapers or cores. Interestingly, Costa argues that assemblages of this nature are not indicative of AMHS manufacture, but many more nominally Late Pleistocene assemblages are needed for comparison. The Kalubhar assemblage, although smaller, contained more Acheulean-type handaxes, likely brought there from elsewhere. The remaining chapters discuss the diversity and origins of stone sources available across Gujarat, noting the lack of obsidian as a tool source in light of several known outcrops. Overall, the lithic component of the GPP represents a shift to researching South Asian Paleolithic site formation processes, given that many assemblages are recovered ex-situ. Costa finds no evidence for Middle or Upper Paleolithic cultures in Saurashtra (pg. 153), while the Late Paleolithic of the Saurashtra peninsula, unlike the mainland, and appears to lack miniature bifaces. The most promising future research directions identified are geochemical analyses to characterize key outcrops, further survey to identify an undisturbed primary reduction site, and more work to understand how and why quartz was utilized.
Part IV (nine chapters) discusses the faunal assemblage recovered from Gopnath, descriptively and taxonomically. After a brief description of the history of Quaternary paleontology in South Asia and a demonstration of the paucity of previous published work (and thus the import of the Gopnath assemblage), the sampling location and strategy is then discussed. Gujarat is noted to have few sites with significant fossil remains, although site numbers increase from the Cenozoic onwards. Prior to the GPP project only a single Pleistocene vertebrate fossil was known from Gujarat (pg. 160). This equid fossil from Tejpur was collected in the 1970s, but two return trips by Costa yielded additional bovid and equus. Only a few ex-situ teeth possibly from the Bhadar Valley area were noted by the author outside of the Gopnath excavations.
From March 2010-October 2011, the coastal site of Gopnath was excavated, with a large fragmentary faunal assemblage excavated in four large blocks after the specimens were noted to be eroding out of a cliff face. After a brief discussion of cleaning and reconstruction methods, the composition of the assemblages is discussed. Species identification, evolutionary history of each family, known duration in India; minimum number of individuals, etc. Families represented include: Subulinidae (Zootecus insularis molluscs); Canidae (MNI=1; Canis lupus familiaris); Equidae (MNI=3-4; Equus cf. sivalensis or hemionus); Rhinocerotidae (MNI=2; Rhinoceros cf. unicornis); Bovidae (MNI=4; Bos, Bubalus and Reduncini tribes). Importantly, the Gopnath assemblage contains very rare specimens of nilgai and the first known fossil of a bovid species from the Reduncini tribe known from Paleolithic Eurasia. The order Testudines (family Testudinae; turtles) is represented by remains from two individuals.
Overall, the Gopnath assemblage has become one of the most important in South Asia; containing 95 identifiable specimens from at least 13 individuals representing 10 taxa and 6 families. The remainder of this section compares this assemblage to the few others known from South Asia, such as the Late Pleistocene-Early Holocene cave assemblage from Kurnool District, Andhra Pradesh (an assemblage poorly curated and not yet published), yet one which contains numerous unique taxa such as hippos, crocodiles and ostriches (pg. 234). No cervids, pigs, hippos or crocodiles are present in the Gopnath assemblage however, despite the general nature of the assemblage being indicative of a grassland/marsh ecosystem. The remainder of the analysis discusses assemblage taphonomy, indicating a lack of cut marks or tooth marks, but signs of trampling and evidence for quick burial after slight dispersal by water. Using a paleosol chronology, the age of the assemblage is suggested to be ~70 Ka, but this is tentative.
Part V (two chapters) presents an overview of the timeline for hominin occupation in South Asia, the history of geochronology in paleoanthropology in the region and the objectives, methods and results of the GPP dating program. Paleoanthropological chronologies in South Asia are very sparse compared to Eurasia and Africa, with the one known pre-modern hominin’s age escaping exact resolution to date. The first chapter presents what is known about regional chronologies both pre and post-Toba eruption, and where in the sequence the South Asian Acheulean fits within the Middle Paleolithic (c. 780-100 Ka), the timing of the Middle to Late Pleistocene transition (c. 125 Ka), the earliest dated “Upper Paleolithic” (Site 55, N. Pakistan c. 45 Ka) and the first South Asian AMHS human remains (Sri Lanka c. 30 Ka). Importantly, Costa notes that no securely dated Upper or Late Paleolithic sites exist in Gujarat, neither from previous work or his dissertation.
Thus, a serious need exists to obtain much more thorough chronologies beyond the limits of C14 dating. Only two partial sequences exist from rock shelters excavated 40 years ago, and no sequence exists of all Paleolithic tool types in situ and correct depositional sequence. Handaxe seriation is argued to be fundamentally unreliable (pg. 257), but discussion of other attempts to date various sequences using fluorine dating, Electron Spin Resonance, cosmogenic nucleotide (26Al/10Be) and Th/U methods, but each come with age or sample type caveats that make them only semi-useful to South Asian Paleolithic contexts.
The goals of the dating component of the GPP project were to evaluate the reliability of previously published Middle Paleolithic sequences in Saurashtra, identify and date new localities relevant to the SDH, and refine chronologies using U-series and OSL methods. As is explained, the error ranges inherent in previous Th/U dating sequences make them unreliable, and the use of the U-series method was discarded from the GPP research design. This left OSL and AMS dating methods, attempted on sediments from Bhadar Valley alluvium, Gopnath carbonates, and Visadi dunes (in order to confirm the age of its in-situ lithic assemblage) and two mammalian tooth samples from Gopnath and Tejpur.
Samples were sent to the Luminesce Dating Laboratory at the University of Illinois, Chicago, and Beta Analytic, respectively. Only the Tejpur dentin sample produced viable collagen to proceed further. Unfortunately, Costa reports that only two of the eight samples collected yielded favourable results. Two OSL samples date to ~ 12 Ka, suggesting the quartz lithic collection from the Visadi dune surface post-dates ~11 Ka; a humid, stable period along the Thar Desert margins. Four additional OSL dates yield results inconsistent with stratigraphic sequences, especially those from the Bhadar Valley. No comparative published results exists for Pleistocene Saurashtra. Furthermore, Costa notes several issues with burial or exposure histories that can negatively affect OSL results, as well as sampling issues encountered when a trained geochronologist is not at hand to collect samples. The Tejpur AMS sample dates to ~ 600 BP, indicating the site’s faunal assemblage is too recent to warrant further exploration.
Part VI (two chapters) specifically review the history and current understanding of fluctuating landscapes and seascapes of Late Pleistocene Gujarat (pg. 272). The stated goals of this section is to simulate the environments that would’ve existed throughout the period to contextualize existing evidence of Paleolithic occupation, explain why hoped-for evidence of occupation may be absent, and develop hypotheses regarding human adaptions to environmental dynamism. Costa identifies a bias that has existed in paleoenvironmental research in Gujarat since the 1970s, in which generalist geomorphological data is favoured over spatially and temporally variant palynological, phytolith or ‘micro-paleontology.’ The first chapter reviews the history of these studies, beginning with attempts to identify diagnostic landforms that showed evidence of sea level fluctuation and carbonate or sand dune intrusion. This foundational research was fuelled by early concerns to understand past water flow and desert expansion to contextualize recent droughts. Following this came research concerned with understanding marine transgression (the “miliolite” debate), dune chronology, fluvial cycles related to SWIM onset, and initial sub-surface investigations of lake and off-shore pollen and clay.
As Costa notes, for hominin migrants that crossed the Indus river, “Gujarat would have been a major intersection where either 1) the eastward coastal trek continued, or 2) migrants found a more direct passage along the Narmada or Tapti river valleys through peninsular India and beyond” (p. 275). Given numerous recent disturbances to the landscape through marine transgressions, damming, etc., the need for modelling of deep-time environmental change remains high, especially in relation to determining exactly how dry it got. In general, “the lower the paleoseascape, the drier the paleolandscape” (pg. 278). On the other hand, the lowered seascape would have supported numerous ecosystems not present in Gujarat today but exploitable to colonising AMHS, until a ~50 m transgression occurred, flooding the coast and creating the river deltas that today empty into the Gulf of Cambay. During very arid glacial phases, the Pleistocene alluvial valleys would’ve created “ice-age spring” refugia; grassland havens for diverse fauna such as those recovered from Gopnath.
In regards to hominin occupation of the landscape, Costa makes it clear that the lack of precise chronologies hinders clear understanding of Paleolithic human adaptions to the landscapes encountered. Most of the Pleistocene artifacts reported by others or Costa himself come from secondary deposits only, not necessarily corresponding to alluvial sediment deposition. Three phases of prehistoric human occupation are subsequently defined, corresponding to changes in tool technology within environmental context. Most surprising is the long hiatus between late Pleistocene and Mesolithic/Chalcolithic tool traditions, entirely postdating the last glacial maximum. Costa postulates that it is possible that LGM aridity and marine regression encouraged both populations to inhabit the now-drowned lowlands. The remainder of part VI discusses possible ethnographic analogies to understand Pleistocene behaviour along these submerged coasts.
In Part VII, the final chapter, the numerous strands of evidence used to place Gujarat on the map of paleoanthropological research in India and globally are synthesized, and variants of the Southern Dispersal hypothesis reviewed. Evidence for human dispersal “out of Africa” between 70-50k BP is now well established by genetic and archeological evidence in Southeast Asia and elsewhere and revised OSL dates of Late Paleolithic sediments that refute the validity of the long (pre-Toba Middle Paleolithic) version of the SDH. The LSD model further postulates that Late Paleolithic microliths were also locally invented in South Asia due to demographic expansion after 35 Ka (p. 295). The numerous findings of the GPP overall, made in light of the difficulties of conducting fieldwork and located previous assemblages, go a long way to ground truthing these models.
Costa’s work not only helps to clarify previous misconceptions about the Paleolithic of this region (for example, the investigations in Saurashtra reported above), but also revealed brand new evidence, such as the Gopnath faunal assemblage. Those few ASM and OSL dates reported have also put to rest the notion of Holocene lithic manufacture in the region. The significant caveat underlying this new research is that Saurashtra’s role in early human migration into India is “constrained by an imperfect Paleolithic record” (pg. 298). Another important issue raised is the continuity of Late Pleistocene bifacial technology and the definition of a “Mode 2.5” industry consisting of distinctive small bifaces and miniature cutting tools, allegedly dating to c. 140-120 Ka and made by a pre-modern or archaic species, with connections to Paleolithic traditions across Late Pleistocene India. The distribution of these tools, hybrids between Acheulean and Middle Paleolithic technologies, need further investigation across India.
A later Pleistocene hiatus from c. 50-10 Ka is also argued for, given that the fieldwork and new chronologies presented here lead Costa to suggest that no true Middle Paleolithic or Late Paleolithic industries are known from Gujarat (p. 302). In contrast to abundant evidence representing this time period from Sri Lanka and Southern India, their absence suggests to the author that latest Pleistocene evidence for human settlement was wiped from the landscape by paleoclimatic events. It’s postulated that further exploration of the Cambay uplands will reveal signs of “incursions” by Middle-Late Paleolithic populations 50-30 Ka. The dissertation ends by putting the overall findings of the GPP into the broader regional context of the MSA (Middle Stone Age) of the Afro-Asiatic (Saharosindian) interaction sphere, concluding importantly that a “conspicuous lack of clear standardized MSA like points in Arabia and South Asia suggests either this technology was lost in the dispersal process or more likely AMHS dispersed into southern Asia prior to the profusion of MSA point types in Africa (i.e. before ~100 Ka)” (pg. 209).
Overall, Costa’s work helps to fill in significant gaps in what paleoanthropologists understand about early human settlement in South Asia and the region as a transit corridor between continents. Little evidence was found to support either the long or short versions of the SDH, nor any Late Pleistocene “invasive technology” indicative of AMHS arrival. Although no microlithic sites are known from Gujarat, the possibility that they exist on the submerged Cambay shelf remains possible. Researchers still need to answer the question of why such tools appeared as late as 35 Ka if AMHS were in Australia by at least 45 Ka. Are the “Mode 2.5” tools merely another example of independent invention and does this indicate human survival after the Toba eruption? The new “BSD” (bifacial southern dispersal) hypothesis is proposed by the author as an explanation, with further techno-morphological studies of associated bifaces diachronically and synchronically c. 180-70 Ka seen as the best means to test it. The foundations laid by Costa’s thorough research provides ample argument for increased survey of new areas of Gujarat in future, for which Costa is well-suited to lead.
Museum Conservation Institute and Dept. of Physical Anthropology
National Museum of Natural History
Fieldwork in India
Peer-Reviewed Published Sources
Indiana University, 2012. 349 pp. Primary Advisor: Jeanne Sept.
Image: View of Fields at Biccavolu, Eastern coastal plains, India, https://en.wikipedia.org/wiki/Eastern_coastal_plains