Central Asia is positioned at a crossroads linking several zones important to hominin dispersal during the Middle Pleistocene. However, the scarcity of stratified and dated archaeological material and paleoclimate records makes it difficult to understand dispersal and occupation dynamics during this time period, especially in arid zones. Here we compile and analyze paleoclimatic and archaeological data from Pleistocene Central Asia, including examination of a new layer-counted speleothem-based multiproxy record of hydrological changes in southern Uzbekistan at the end of MIS 11. Our findings indicate that Lower Palaeolithic sites in the steppe, semi-arid, and desert zones of Central Asia may have served as key areas for the dispersal of hominins into Eurasia during the Middle Pleistocene. In agreement with previous studies, we find that bifaces occur across these zones at higher latitudes and in lower altitudes relative to the other Paleolithic assemblages. We argue that arid Central Asia would have been intermittently habitable during the Middle Pleistocene when long warm interglacial phases coincided with periods when the Caspian Sea was experiencing consistently high water levels, resulting in greater moisture availability and more temperate conditions in otherwise arid regions. During periodic intervals in the Middle Pleistocene, the local environment of arid Central Asia was likely a favorable habitat for paleolithic hominins and was frequented by Lower Paleolithic toolmakers producing bifaces.

Funding: This work was funded by the Max Planck Society: https://www.shh.mpg.de/en . Research by PSB was funded by the Leverhulme Trust (ECF-2019-538). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2022 Finestone et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Introduction

Central Asia is situated at a crossroad that links east and north Asia with Europe and the Levant. This region is fundamental to questions of early hominin dispersals because of its position at the gateway between key regions where at least two Middle Pleistocene hominin species are known to have interacted [1]. Subsequently our own species also moved through these regions [2, 3], with the routes and timings of its initial dispersals remaining debated. Despite the importance of Central Asia for understanding the spatial and temporal patterning of hominin occupations in Eurasia, however, our knowledge of hominin activity in this vast and diverse landscape is disproportionately limited when compared with other regions on the continent.

Because of the absence of dated and stratified Lower Paleolithic sites in Central Asia, most studies have focused on the region’s Middle and Upper Paleolithic occupation. Currently, available evidence indicates that the Pamir, Tian Shan, and Altai mountains served as corridors of occupation and movement for populations of multiple hominin species through the Late Middle and Late Pleistocene, including Denisovans, Neanderthals, and modern humans [4]. Hominins occupied Central Asia consistently through the Late Pleistocene, even throughout periods of climatic downturn during the coldest episodes of the Last Glacial Period [5–8]. However, the initial occupation of Central Asia and the role the low and mid-altitude plains played in dispersal and occupation remain poorly understood and contentious.

The early colonization of Central Asia has been previously reviewed most notably by Ranov and Davis [9], Davis and Ranov [10], Vishnyatsky [11], Derevianko [12], and Glantz [13]. However, understanding the environmental dynamics of these regions is made difficult by the lack of well-dated paleoclimate records and the paucity of stratified lithic assemblages from Pleistocene arid Central Asia. These limitations remain the most important hurdle for the systematic study of the Lower Paleolithic in this region.

Climate of arid Central Asia Central Asia encompasses the former Soviet republics of Kazakhstan, Uzbekistan, Turkmenistan, Tajikistan, and Kyrgyzstan (Fig 1). The interior of Central Asia sits at mid- and low-altitudes and is geographically diverse, being characterized by both tectonic activity and continentality. The vast majority of the region consists of a relatively flat desert and semi-desert plain (i.e., the Turan Depression) bounded in the south by the Tien Shan, Pamir, and Hindu Kush mountains, to the southwest by the Kopetdag, Alborz and Zagros mountains, to the northwest by the Ural Mountains, and to the east by the Altai Mountains. Its western border is in part defined by the Caspian Sea and its northern one by the Siberian plains (Fig 1). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. The study region. A map of Central Asia derived by the authors using data for annual precipitation (Worldclim [14]), surface water [15], and rivers (Hydrosheds [16]). The study region (encircled with black line) is shown with the region of Caspian catchment (enclosed with red line). Key mountain ranges, deserts, and bodies of water are labeled. The location of Amir Timur Cave that hosted stalagmite S-12-4 analyzed in this study is shown as red circle. https://doi.org/10.1371/journal.pone.0273984.g001 The climate of the mountains is largely controlled by the westerlies bringing moisture to this region [17]. However, the large distance from the global ocean and the topography of the surrounding mountains isolate the lowlands, resulting in dry continental climate conditions, which have produced the Karakum and the Kyzylkum Deserts. Consequently, the water supply in the region is largely delivered by winter precipitation (i.e., rain and snow) at high altitudes, and snow/glacier melt during the summer. Drainage from the mountains into the Aral and Caspian Seas provides the primary water supply throughout the arid Central Asian lowlands. However, the paleoclimatic history of the Aral Sea shows it as a transient water body, only forming a large lake when the Amu Darya River flows into it, as it does today. Yet this river has switched between the Caspian and Aral Seas numerous times during the Pleistocene, causing the lake to fluctuate substantially [18]. The size of the Caspian Sea has also fluctuated greatly during the Pleistocene [19, 20]. This is partly due to it gaining and losing waters from the switching of the Amu Darya River. However, given the fact that the rivers feeding the Caspian Sea span so many different climates and altitudinal zones (Fig 1), it is clear that a multitude of factors would have caused changes to the size of the lake during the Pleistocene. The main river feeding the Caspian Sea today is the Volga. It has an extensive catchment feeding in from the north, with its headwaters as far north as Moscow. Thus, changes in moisture availability and ice sheet fluctuations in this region will cause lake level variations over time, with particularly high discharge occurring at the termination of glacial periods when the Scandinavian Ice sheet rapidly melts [19]. Though the mountains of Central Asia provide water to the Caspian and Aral Seas, evaporation from these waterbodies in turn provides an important moisture source for the precipitation in the mountains. For example, isotopic studies of ice cores from the Tien Shan and Altai Mountains suggest that about a third of the Altai precipitation is derived from the Aral and Caspian Seas, whereas 87% of the precipitation in the Tien Shan Mountains is derived from the Aral, Caspian, Mediterranean and Black Seas [21]. Thus, the fluctuations in the size of the Pontocaspian seas will cause fluctuations in the amount of precipitation within Central Asia. The climatic trends that resulted in the modern arid climate of Central Asia took root in the Early Pleistocene [22, 23]. At the beginning of the Pleistocene, the climate was semi-arid, but warmer and wetter than later periods [22]. Annual rainfall and temperature steadily decreased through the Middle Pleistocene [23]. This trend is tied in part to the Middle Pleistocene Transition (MPT, 1250–700 ka), during which the high amplitude periodic 100-kyr glacial-interglacial cycles emerged (Fig 2) [24, 25], promoting dryland environments throughout the lowland in the mid-altitude of Central Asia, particularly during glacial periods [25]. Notwithstanding this, the expansion of the Karakum and Kyzylkum deserts likely took place gradually in the Middle Pleistocene, as indicated by a progressive increase in loess deposits [25–28]. However, after the MPT these loess records show significantly increased changes in grain size that are largely coincident with glacial interglacial cycles, indicating that there were significant variations in aridity throughout this period, with interglacials generally being wetter and glacials being dryer. Thus overall, the intensification of glacial-interglacial cycles contributed to the aridification of Central Asia through the Middle Pleistocene. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Climate variables and evidence of hominin occupation in Central Asia through time. Global oxygen isotope records with numbered MIS interglacials (CENOGRID dataset) [29], the Tajik loess record [26], and relative Caspian Sea level variation [19]. Blue shading indicates wet intervals in the Tajik loess record and pink shading indicates intervals of Caspian outflow into the Black Sea [19]. All known hominin taxa from Central Asia during the Pleistocene (Homo erectus [H.e.], Denisovans, Homo neanderthalensis [H.n.], Homo sapiens [H.s.]) are displayed alongside the suggested chronology for Lower (green) and Middle (red) Paleolithic assemblages. Symbols indicate the basis of age estimations and assemblages without symbol represent dates previously proposed in the literature but are unverified/not tested by dating methods. https://doi.org/10.1371/journal.pone.0273984.g002 While loess profiles inform on glacial-interglacial environmental changes with (at best) centennial to multi-decadal resolution, speleothem-based proxy records can allow detailed insights into past hydrology and environmental variability even at seasonal scale. Well-dated and highly resolved reconstructions of Pleistocene regional and local environmental changes are extremely scarce in Central Asia, however. Regional reconstructions of Pleistocene climatic conditions rely mostly on loess sequences [26] and only two speleothem records from Tonnel’naya and Kesang caves [30–32]. This dearth of high-resolution data on past hydrological and thermal conditions severely limits our understanding of human activity in this highly dynamic environment.

The timing of the earliest occupation of Central Asia Hominins first dispersed out of Africa and into Asia approximately two million years ago. This is evidenced by the existence of Lower Paleolithic toolkits in China by 2.1 Ma [33] and fossil remains in Dmanisi, Georgia, dated to around (ca.) 1.8 Ma [34–36]. Hominin remains are present prior to 1 Ma in Georgia (i.e., Dmanisi ca. 1.8 Ma), Turkey (Kocabaş, ca. 1.3–1.1 Ma) [37], China (Gongwangling, ca. 1.63 Ma) [38], and Indonesia (Java, ca. 1.6–1.5 Ma) [39]. The lack of hominin remains or stone tools prior to 1 Ma in Central Asia may suggest that early hominin dispersals into East Asia bypassed Central Asia completely and later entered Central Asia either through an east-to-west route, or in subsequent migration events from Africa north-eastward. However, Early Pleistocene stratified deposits are sparce and largely unexplored in Central Asia. It is possible that hominins initially colonialized and dispersed across Central Asia during Early Pleistocene migrations to northern China, but that evidence is currently lacking because of an absence of systematic investigations of this time interval. Although the archaeological record is sparse and dating difficult, some evidence suggests that between 1.0–0.8 Ma hominins inhabited Central Asia and produced Lower Paleolithic toolkits. The oldest age for any Paleolithic locality in the region has been proposed for an assemblage of Mode 1 lithics from eastern Kopetdag, in the valley of the River Keshefrud [40], in southwestern Central Asia. While the proposed geological age for this assemblage is between 1 and 0.8 Ma [40], the basis for this date is unknown and unsubstantiated [11]. Another Lower Paleolithic assemblage identified in Yangadzha, Turkmenistan, may also date to the Middle Pleistocene based on correlation with similar technocomplexes in the Caucasus [10]. However, the lack of stratified, datable archaeological or fossil material from Turkmenistan makes it impossible to verify the arrival of hominins in the Kopetdag region in the Early or Middle Pleistocene. More convincing evidence comes with the oldest known stratified localities in Central Asia, identified in loess deposits in Tajikistan. A reliable date is associated with stone tools from the Kul’dara location in the Tajik Depression [41, 42]. Artifacts from Kul’dara are associated with pedocomplexes (PC) 12 and 11 [41, 42] and were excavated from strata between the Brunhes/Matuyama boundary and the Jaramillo subchron, indicating an age between 900–800 Ka (reviewed in [11]). Kul’dara is the oldest assemblage in this region, although other Middle Pleistocene localities with stratigraphic contexts are known from the Tajik loess deposits [10]. Isolated artifacts recovered between PC 11 and 6 [42], including finds from Karamaidan (PC 6; [43]) are likely older than 600 ka [11]. Larger collections of artifacts resembling similar but later forms of Kul’dara industries were recovered at Karatau (PC 6), Lakhuti (PC 5), Obi-Mazar (PC 4), and Khonako (PC 4; [44–47]). Dating frameworks indicated that Karatau and Lakhuti were formed approximately 500 to 600 ka [23, 48]. Lower Paleolithic toolkits found in Kazakhstan have also been attributed to the Middle Pleistocene. Two excavated localities from the South-Kazakhstan Province in the foothills of the Karatau mountains region (Koshkurgan and Shoktas) have each yielded hundreds of artifacts. The oldest of these deposits were dated by electron spin resonance (ESR) to ca. 500 ka [49]. However, these dates are accepted with caution and it is possible that the Koshkurgan-Shoktas industry could be attributed to the Middle Paleolithic. In eastern Kazakhstan, Lower Paleolithic artifacts are mixed into deflated assemblages along the Zaisan Basin (e.g., Kuruchum) [50, 51]. At several locations, artifacts have been found eroding from what are estimated to be Early/Middle Pleistocene deposits, and the patination on the most archaic of these artifacts indicates warm and relatively humid conditions [52]. Aside from the Tajik loess deposits and the Koshkurgan-Shoktas complex, Lower Paleolithic finds in Central Asia provide very little basis for dating. However, it is worth noting that the handful of Lower Paleolithic sites with established dates fall within the Middle Pleistocene or older, suggesting that the majority of Lower Paleolithic artifacts may have been produced during this time period. In any case, by the Middle Pleistocene, hominins were clearly inhabiting Central Asia (Fig 2). However, areas of Lower Paleolithic occupation were likely inhabited intermittently and primarily during warm and relatively humid interglacial periods [44, 53].

Features of the lower paleolithic of Central Asia The Lower Paleolithic of Central Asia presents a great deal of variability in tool form and assemblage composition. Lithic assemblages are generally attributed to the Lower Paleolithic based on the presence of core-and-flake industries or large Acheulean bifaces. Glantz [13] divides these diverse assemblages into four industrial complexes: micro artifacts, core-and-flake industries, pebble culture, and Acheulean-like bifaces. The Koshkurgan and Shoktas sites preserve the only known assemblages from a unique microlith industry [49]. Tools were primarily made from raw materials available in the vicinity, which were almost entirely small pebbles with an average size of 4 to 5 cm [54, 55]. Nearby Lower Paleolithic assemblages from the Karatau region such as Kyzyltau 1–2 [56], Borykazgan, Tanirkazgan, and Akkol [57] resemble a different technocomplex distinct from the small artifact industry of Koshkurgan and Shoktas. These assemblages contain general core-and-flake industries characterized by large chopping tools and proto-Levallois techniques [58, 59]. Pebble cultures are found in the Tajik depression and can also be considered a subset of core-and-flake industries [11]. Pebble technology is most prominent at Karatau but also identified at other nearby assemblages such as at Lakhuti/Obi-Mazar, Karamaidan, and Kul’dara. Reduction strategy is based on metamorphic pebble flaking, which produces a large portion of cortical flakes, fragments, and chips [60]. This industry is characteristic of the earliest dated material in Central Asia and is suggested to associate with the first migration wave out of Africa [47]. Finally, the desert plateaus and steppe regions of Kazakhstan, Uzbekistan, and Turkmenistan have yielded surface scatters of bifacially worked tools resembling the Acheulean. The clearest examples of these occur in the Krasnovodsk Peninsula on the Caspian Sea in Kazakhstan [11]. Surface scatters containing bifaces are uncommon but present throughout the Central Asian steppe. This includes eastern Kazakhstan in the Lake Balkash region (i.e., the sites of Semizbugu and Bale [61, 62]), and northeastern Kazakhstan (Kudaikol, Vishnevka-3 [63, 64]). It has been suggested that a second wave of Out-Of-Africa expansion brought an Acheulean bifacial industry by ca. 450–350 ka [47]. However, the chronology for this arrival remains uncertain because all Central Asian localities yielding bifaces lack stratigraphic context and sound chronologies. In Central Asia, bifaces often occur alongside assemblages characteristic of Middle Paleolithic industries. For example, the Paleolithic sites at Kyzylnura ([65], reviewed in [11, 13]) are diagnostic of the Middle Paleolithic, but additionally contain large bifaces. The lack of secure dates and stratified contexts hinders strict assessment. It remains unknown whether these assemblages are truly representative of an Acheulean industry, a mix of multiple industries, a late developed Acheulean that occurred alongside Middle Paleolithic Industries, or perhaps even more recent bifacial traditions [66]. The absence of any evidence for dating associated with this typology makes it impossible to evaluate. Vishnyatsky [11] reviewed and mapped the location of assemblages characterized by varieties of flake-and-core industries, and those containing bifaces. He noted that bifaces are known in the west and north, but are largely absent in the south, and generally associated with the plains. In contrast, core-and-flake industries occur almost exclusively in southern zones. At the time, Vishnyatsky [11] noted that one possible exception to this trend was Kulbulak, Uzbekistan, where handaxes and bifaces appear to be present in mountain foothills and at a more southern latitude than expected [67]. However, according to Vishnyatsky [68] these bifaces are more likely Middle Paleolithic tools and more recently, Kolobova et al. [69] re-evaluated the Kulbulak assemblage and concluded that Acheulean tools are absent at Kulbulak altogether. Additional surveys have also recovered Acheulean-like bifaces in northern Kazakhstan at the Mugodzhar hills [70], in northeastern Kazakhstan at Ekibastuz 4 [71], and to the east of the Aral Sea [72, 73] with some of these possibly belonging to the Lower Paleolithic. These findings further support the notion that bifaces frequently occur in northern arid zones and are uncommon in mountainous, but hydrologically better supplied, areas.