Challenges of Ancient DNA

The difficulty in recovering and analyzing authentic ancient DNA fragments from degraded organic material has been well described in the literature (e.g. (Briggs et al. Reference Briggs, Stenzel and Johnson2007; Dabney et al. Reference Dabney, Meyer and Paabo2013; Kistler et al. Reference Kistler, Ware, Smith, Collins and Allaby2017; Pruvost et al. Reference Pruvost, Schwarz and Correia2007)). These and other studies have revealed that the preservation of DNA molecules over time is a result of postmortem putrefaction phases followed by the effects that environmental conditions such as temperature, humidity, salinity, and pH have on the frequency of random processes leading to strand breakage, oxidation damage, and cytosine deamination. In addition, the element of time allows the accumulation of these events, resulting in increasingly shorter molecules and an increasing amount of base-altering lesions (Allentoft et al. Reference Allentoft, Collins and Harker2012; Deagle et al. Reference Deagle, Eveson and Jarman2006). The hot and humid environment in southern East Asia in particular has been described as especially unfavorable for the recovery of ancient DNA (Kistler et al. Reference Kistler, Ware, Smith, Collins and Allaby2017; Wang et al. Reference Wang, Yeh and Popov2021a). The high level of exogenous environmental DNA relative to ancient endogenous DNA typically co-recovered from ancient material, which can often exceed 99%, also decreases the overall efficiency and substantially raises the expense of ancient DNA recovery. Discoveries in the last decade of the exceptional DNA preservation often encountered within the dense inner part of the petrous bone (Gamba et al. Reference Gamba, Jones and Teasdale2014) and, to some extent, the cementum layer of the teeth (Hansen et al. Reference Hansen, Damgaard and Margaryan2017) have increased the efficiency of experiments and the reliability of results, and have allowed some recent progress in retrieving ancient DNA information from difficult regions. Still, cultural factors such as cremation, the earliest East Asian example of which has been reported in Neolithic Jiangsu (Yan et al. Reference Yan, Ge and Wang2022), and the prospect of selective burials (Fernández-Crespo & de-la-Rúa Reference Fernández-Crespo and de-la-Rúa2015) remind us that our understanding of the past through what remains has its own survivorship biases, and must be cautiously interpreted with an awareness that some records will be forever missing.

The genetic information recovered from fossil material consists of short sequences generated from fragments of DNA molecules which are likely to have acquired sequence-altering genetic lesions over the years. The most common of these is cytosine deamination (Briggs et al. Reference Briggs, Stenzel and Johnson2007), or the conversion of cytosine to uracil, which manifests in recovered sequences as a conversion of cytosine to thymine. Protocols have been developed to remove or correct this damage using enzymatic or in silico approaches at various points along the chain of analysis (Briggs et al. Reference Briggs, Stenzel and Meyer2010; Prüfer Reference Prüfer2018; Rohland et al. Reference Rohland, Harney, Mallick, Nordenfelt and Reich2015). Although the ubiquity of this form of damage decreases the reliability of these bases at lower coverages, this damage can serve a useful purpose in separating authentic ancient DNA from contaminating modern human DNA (Fu et al. Reference Fu, Hajdinjak and Moldovan2015). Due to the low endogenous DNA content of ancient material, most common ancient genomics studies consider only an array of specific genomic loci, SNPs (single nucleotide polymorphisms), which have been predetermined to be informative between populations, such as the “Human Origins” SNP array consisting of approximately 600,000 SNPs (Patterson et al. Reference Patterson, Moorjani and Luo2012). The technique of targeted sequence capture (Ávila-Arcos et al. Reference Ávila-Arcos, Cappellini and Romero-Navarro2011; Fu et al. Reference Fu, Meyer and Gao2013) can increase the efficiency of ancient DNA recovery by targeting only those DNA fragments containing bases required for comparison between genomes. This approach uses synthetically created “bait” oligonucleotides to hybridize to ancient DNA fragments that contain sequences similar to those of the targeting baits, and is particularly useful when a high background of environmental DNA in a sample makes a non-targeted, “shotgun” sequencing approach infeasible.

The EDAR Variant

An adaptive genetic variant of the EDAR gene occurred in the ancestors of East Asian and Native American populations, where it has risen through positive selection to high frequencies in these groups (Kamberov et al. Reference Kamberov, Wang and Tan2013), with an average 91.6% frequency among several East Asian populations. The variant, EDAR_V370A, codes for a cell surface receptor related to ectodermal development resulting in a phenotype of shovel-shaped incisors, thick hair shafts, and increased sweat glands (Bryk et al. Reference Bryk, Hardouin and Pugach2008). There have been several theories as to the specific selective pressure through which positive selection had acted on this variant, and where and when this process began, but the answers have yet to be satisfactorily determined. Age estimates for this mutation using various models have been between 3,000 and 22,000 years ago, and possibly as old as 43,000 (Bryk et al. Reference Bryk, Hardouin and Pugach2008; Peter et al. Reference Peter, Huerta-Sanchez and Nielsen2012; Smith et al. Reference Smith, Coop, Stephens and Novembre2018). Suggested selective advantages have considered that the allele may have increased vitamin-D intake from milk in low UV environments during the LGM 23,000 years ago, or alternatively, the allele may have aided heat dispersal during a warm period of high humidity beginning 40,000 years ago (Kamberov et al. Reference Kamberov, Wang and Tan2013). The oldest individual yet identified with the EDAR_V370A variant has been AR19K in the Amur Region, which places the allele at the latest toward the end of the LGM. It is common in subsequent northeastern East Asia individuals but absent in pre-LGM AR33K and Tianyuan (Mao et al. Reference Mao, Zhang and Qiao2021). It has also been reported to be absent in ancient Jōmon individuals and Papuans (Wang et al. Reference Wang, Yeh and Popov2021a). More information is needed to better understand the origin and selective advantages of this variant, but current data suggest the allele originated along an inland East Asian lineage, perhaps during the LGM, which would imply selective pressure occurred during a challenging cold climatic period.

Ancient Northeast Asian Ancestry

Several Neolithic groups around the Amur River Basin dating from between 14,000 and 2,000 years ago have been analyzed genetically and have been found to cluster together with a common ancestry described as Ancient Northeast Asian (ANA) (Mao et al. Reference Mao, Zhang and Qiao2021; Ning et al. Reference Ning, Li and Wang2020b; Robbeets et al. Reference Robbeets, Bouckaert and Conte2021; Siska et al. Reference Siska, Jones and Jeon2017; Wang et al. Reference Wang, Yeh and Popov2021a). ANA ancestry could be modeled as branching from a common lineage as that of AR19K, although with some genetic distance separating these two ancestries (Mao et al. Reference Mao, Zhang and Qiao2021). ANA ancestry has been found in Neolithic hunter-gatherer populations living in the Amur region and Primorye to the east, as well as Baikal in Siberia, the Inner Mongolian Plateau, and the West Liao River. During the Neolithic it was present at the Songhua River, the Houtaomuga 后套木嘎 site near Da’an 大安 in Jilin province of China, the site of Jalainur 扎赉诺尔 in eastern Inner Mongolian, as well as the coastal populations of Devil’s Gate Cave (Chertovy Vorota) and the Boisman-2 site in the Primorye region of Siberia, and appears to have been indigenous to these regions since at least the Paleolithic era. These and other contemporary sites in northeast East Asia containing relevant archeological material indicate that these groups practiced a hunter-gatherer subsistence strategy supplemented by intensified fishing and limited horticulture (Kuzmin Reference Kuzmin2013), but did not practice large-scale organized agriculture at this time. In contrast to the common interpretation of the Neolithic of Europe, the Neolithic of eastern Eurasia is marked not by the beginning of structured farming practices but by a broader spectrum of human behavioral changes including the appearance of pottery, where it predated the Holocene (Popov et al. Reference Popov, Tabarev and Mikishin2014). In northern East Asia, pottery has been documented more than 14,000 years ago (Kuzmin Reference Kuzmin2017; Shoda et al. Reference Shoda, Lucquin and Yanshina2020), while organized millet cultivation does not appear in the Amur River Valley until after 6,000 years ago (Li et al. Reference Li, Ning and Zhushchikhovskaya2020). An additional east-west genetic cline has been reported for ANA populations, with higher similarity to Neolithic Mongolians to the west and higher ancestry related to Jōmon 绳文 populations from Paleolithic Japan to the east (Robbeets et al. Reference Robbeets, Bouckaert and Conte2021).

Compared to present-day populations, ANA ancestry is most similar to Tungusic and Nivkh language speakers such as the Ulchi, Oroqen, Hazhen, and Nivkh, all of whom still live in the same geographic region as the Neolithic ANA populations (Ning et al. Reference Ning, Li and Wang2020b), and thus represent a genetic and geographic continuity going back at least 14,000 years. Possible reasons for the persistence of ANA populations for so long may have been related to the late adoption of agriculture, having not arrived in the Amur Region until the Middle to Late Neolithic (Cui et al. Reference Cui, Zhang and Ma2020). The difficulty of transferring agriculture from warmer to colder climates and the reliance on alternative food sources may have protected this region from the severe regional climatic variations that can impact agriculture-based societies (Lutaenko et al. Reference Lutaenko, Zhushchikhovskaya, Mikishin, Popov, Anderson, Maasch and Sandweiss2007) and helped them to resist population expansions from the south tied to the spread of agriculture (Mao et al. Reference Mao, Zhang and Qiao2021; Ning et al. Reference Ning, Li and Wang2020b). A steady decrease in genetic relatedness in the Amur Region over the Neolithic period as measured by Runs of Homozygosity analysis, which can be used to derive population demography and size by looking at the length of identical genomic sections, has been linked to an expanding population indicative of a continuous successful exploitation of the resources of the area (Mao et al. Reference Mao, Zhang and Qiao2021).

West Liao River Farmers and the Origins of the Transeurasian Language Family

The West Liao River flows from its tributaries in the steppes of the Inner Mongolian Plateau where it joins the East Liao River and continues into the Liaodong 辽东 Bay in the Yellow Sea. Unlike the Neolithic subsistence practices of the Amur region to the northeast, the Hongshan 红山 culture in the West Liao River valley had been cultivating Broomcorn millet as a staple crop by the Middle Neolithic, beginning 6,500 years ago (Li et al. Reference Li, Ning and Zhushchikhovskaya2020; Sun & Zhao Reference Sun and Zhao2013). Several genetic studies of these early farmers in the West Liao River region have concluded both Middle Neolithic Hongshan and Early Bronze Age Lower Xiajiadian cultures of the West Liao River contained ANA ancestry mixed with ancestry from farming societies found along the Yellow River to the south in Shandong and Henan, and associated with the Neolithic Longshan 龙山 culture (Ning et al. Reference Ning, Li and Wang2020b). The cultural transition from the Lower to the Upper Xiajiadian 夏家店 beginning roughly 3,000 years ago saw the trend reverse, and an increase in ANA ancestry over Yellow River ancestry (YR) occurred. This population shift coincided with mid-Holocene climatic changes to a colder, dryer period at the end of the Holocene Optimum and the adoption of a more pastoral lifestyle around the West Liao River, which appears to have been related to population expansions from the north (Jia et al. Reference Jia, Sun and Wang2016; Ning et al. Reference Ning, Li and Wang2020b).

Transeurasian Language Dispersal

The discovery that ANA ancestry is carried by all present-day speakers of Transeurasian languages, a macro-family largely encompassing the proto-languages of Mongolic, Turkic, Tungusic, Koreanic, and Japonic, has led to genetic and multidisciplinary investigations of the origins of the Transeurasian language family (Ning et al. Reference Ning, Li and Wang2020b; Robbeets et al. Reference Robbeets, Bouckaert and Conte2021; Wang et al. Reference Wang, Yeh and Popov2021a). These studies leverage information of the dispersal of ancient genetic ancestral components to test the validity of the Transeurasian hypothesis, which posits that these language groups arose from a proto-language that expanded from the West Liao River Basin along with the spread of millet farmers during the Neolithic. There has been substantial geographical and temporal overlap found between ANA ancestry and the linguistic histories of these languages. Some of these results agree with an Early Neolithic eastward movement of separate West Liao River populations into Primorye and Korea, which may correspond to the proto-Altaic and proto-Japano-Koreanic, language groups, respectively, the second group being linked to admixed ANA and YR ancestry. Bronze Age migrations corresponding to the Upper Xiajiadian-related ancestry arrived in Japan from Liaodong via the Korean Penninsula bearing an expanded agricultural package including rice, barley, and wheat. The westward spread from West Liao River populations into the eastern Eurasian Steppes during the Bronze Age, where they admixed with arriving western Eurasian ancestry, may account for noted linguistic borrowings concerning agriculture and pastoralism found in proto-Mongolic and proto-Turkic language groups (Robbeets et al. Reference Robbeets, Bouckaert and Conte2021). One complication of these findings is that in addition to ANA, the earliest genomes from the West Liao River also contain YR ancestry, which is not found in Amur or Primorye (Ning et al. Reference Ning, Li and Wang2020b). It has been argued that the West Liao River genomes that have been characterized so far date only to the Middle Neolithic, and older genomes from this region should predate these YR ancestry admixture events (Robbeets et al. Reference Robbeets, Bouckaert and Conte2021). Another complication is that the similarities between Neolithic ANA ancestries originating from the West Liao River with those native to the Amur region make such movements into this region difficult to track genetically.

Contributions to Native Americans

The exploration of the genetic history of northern East Asia has also had a major impact in our understanding of the origins of Native American groups. A 10,000-year-old genome from a site near the Kolyma River in northeastern Siberia was found to contain a distinct genetic ancestry, Ancient Paleo-Siberian (APS) (Sikora et al. Reference Sikora, Pitulko and Sousa2019), which can be modeled as an admixture between west Eurasian branching ANS-like ancestry, such as that found in Yana, and an East Asian lineage. Although it was known that Ancient North Eurasian (ANE) ancestry, such as that of the 25,000-year-old MA-1 individual from Mal’ta near Lake Baikal in Siberia, shared ancestry with Native Americans, it was clear that ANE ancestry required additional East Asian gene flow in order to approximate the Native American founding population (Moreno-Mayar et al. Reference Moreno-Mayar, Potter and Vinner2018; Raghavan et al. Reference Raghavan, Skoglund and Graf2014). APS ancestry found in Kolyma1 was the closest match to this founding population, as represented by genomes recovered from two children at the 11,500-year-old Alaskan site Upward Sun River in North America, USR1 (Moreno-Mayar et al. Reference Moreno-Mayar, Potter and Vinner2018), yet in attempts to best model the relationship between APS ancestry and USR1 using statistical genetics, Kolyma1 still lacked additional ancestry from an unidentified East Asian source. Among the currently available ancient sources, AR14K from the Amur Region was determined to be the closest candidate to this East Asian source, permitting the Native American USR1 to be modeled as an admixture of 20–30% AR14K-related East Asian ancestry and 70–80% APS ancestry (Mao et al. Reference Mao, Zhang and Qiao2021). Demographic models estimate the split between APS and populations ancestral to Native Americans to have occurred approximately 24,000 years ago (Sikora et al. Reference Sikora, Pitulko and Sousa2019). Other interpretations have questioned the simple model of a single early founding population and propose several migration waves across Beringia with different levels of ANA ancestry (Ning et al. Reference Ning, Fernandes and Changmai2020a). Additional research and samples will be needed to better identify potential source populations closer to these separation events, and the areas where they were likely to have occurred, as well as to determine the complex population dynamics of Upper Paleolithic northeast Eurasia and Beringia with relation to Native American founding groups.

Fujian Coastal Populations

Fujian is a mountainous province located in southeastern China along the coast of the East China Sea that contains a narrow, low, coastal plain facing the island of Taiwan across the Taiwan Strait. The Fujian coast is characterized by numerous small islands, including Liangdao 亮岛, where 8,300-year-old remains were recovered beneath a shell mound containing pottery and tools made from both stone and bone. Genomic analysis of the individual, Liangdao1, along with Liangdao2, a second individual who lived approximately 600 years later, found they clustered closely together with two individuals who were excavated from the Qihe 奇和 cave site in mainland Fujian, an 8,400-year-old individual, Qihe2, and a 12,000-year-old Qihe3. This ancient southern coastal East Asian ancestry, later termed ancient Fujian ancestry, had close genetic similarities to Neolithic Southeast Asians and Austronesian-related islanders from the Southwest Pacific, and have all been shown in statistical tests to share more similarities with each other than with northern ancient East Asian groups (Wang et al. Reference Wang, Wang and Xie2021b; Yang et al. Reference Yang, Fan and Sun2020).

In the Late Neolithic, the Tanshishan 昙石山 culture arose after 5,000 years ago in the lower Min River 岷江 valley in Fujian. The Tanshishan site shows evidence of mixed rice and millet farming from at least 5,000 years ago (Dai et al. Reference Dai, Cai and Jin2021), and the culture shared characteristics with the Longshan culture in the Yellow River valley (Chang Reference Chang1989 p. 89). To characterize the long-term demographic changes of the southeastern Chinese coastal populations, 4,500-year-old populations from two Tanshishan sites, Tanshishan and Xitoucun 溪头村, were investigated along with the contemporary site of Suogang 锁港 on the southeastern coast of the Penghu 澎湖 archipelago in the Taiwan Strait. These younger southeastern coastal East Asia populations also clustered with the older Fujian ancestry from the Early Neolithic, showing the persistence of the regional populations for over 8,000 years, however these southern populations were not equally differentiated from coastal and inland populations from northern East Asia. Some Late Neolithic Fujian (Fujian_LN) ancestry was genetically closer to Early Neolithic northern coastal East Asians than to inland groups, a pattern less pronounced but still observed in Early Neolithic Fujian (Fujian_EN) ancestry (Yang et al. Reference Yang, Fan and Sun2020). These connections, which are also supported by maternal genetics (Liu et al. Reference Liu, Zeng and Sun2021), could be described by increasing gene flow between coastal populations, although some deeper population structure cannot also be ruled out.

Austronesians

Cultural links have been previously noted between Neolithic southern mainland coastal populations and the Austronesian groups who populated Taiwan 6,000 years ago (Wu Reference Wu and Wu2021). Austronesians would later colonize islands of the South West Pacific and Indian Oceans beginning some 2,000 years later, eventually reaching as far as the African coast and distant islands of the Pacific, but their mainland origins have been difficult to identify. Three models have been proposed for the origin of the Neolithic in Taiwan: (1) the Mainland East Asian Interaction Sphere proposes a local expansion of the mainland Neolithic across the Taiwan strait and is supported by similarities in pre-agriculture ceramics; (2) the Northeastern Seaboard theory describes an origin from the coast of northeastern China, especially around the Shandong Peninsula, and is supported by a proposed Sino-Tibetan linguistic origin of Austronesian languages (Sagart et al. Reference Sagart, Blench and Sanchez-Mazas2005); and (3) a Lower Yangtze (LY) origin from the expansion of Hemudu 河姆渡 culture-linked rice agriculturalists along the lower Yangtze River Basin (Sagart et al. Reference Sagart, Hsu and Tsai2018).

Ancient genomic studies can shed light on this debate by directly comparing early Austronesians with their ancient potential source populations. In doing this, several 3,000-year-old Austronesian-related southwestern Pacific Islanders from Vanuatu (Skoglund et al. Reference Skoglund, Posth and Sirak2016) were found to be a sister group to Fujian_LN populations from the southern East Asian mainland (Yang et al. Reference Yang, Fan and Sun2020). This signal is obscured when comparing the ancient Austronesians to present-day and even contemporary southern East Asian populations due to an elevated proportion of northern East Asian ancestry in Fujian, a continuation of a trend already observed in the Neolithic. From Taiwan, 3,000- to 1,000-year-old Austronesians (Ko et al. Reference Ko, Chen and Fu2014) also have been found to contain some northern ancestry, and can be modeled as having approximately 25% ancient northern East Asian ancestry similar to that found in the West Liao or Yellow Rivers, near Shandong (nEA, described herein), and 75% southern East Asian ancestry matching Fujian_EN, as represented by the 8,300-year-old remains from Liangdao, an ancestry also present in Tai-Kadai, Austroasiatic, and Austronesian language speakers (Wang et al. Reference Wang, Yeh and Popov2021a). The presence of northern East Asian ancestry makes the sequenced Austronesian genomes from Taiwan poor sources for the ancient southwest Pacific Austronesians who lack this northern component, but the relationship to mainland groups becomes clearer with the Neolithic data from the mainland. Additionally, both Fujian_EN and _LN share more alleles with present-day indegenous groups from Taiwan than with other present-day southern East Asian groups (Yang et al. Reference Yang, Fan and Sun2020). Mitochondrial sequences also support this relationship. The mitochondrial haplogroup, E, recovered from Liangdao1 is rare in present-day mainland East Asia, but is found at high frequencies among Austronesian language speakers. The Liangdao mitochondrial sequence occupies a position basal to those found in indigenous Atayal and Ami groups in Taiwan (Ko et al. Reference Ko, Chen and Fu2014). Altogether, the shared ancestry between ancient Fujian populations and ancient Vanuatu, as well as the gene flow shared with populations in Taiwan, points to a geographic origin of Austronesians somewhere along the southern coast of China, although a specific population of the proto-Austronesians that can be modeled as their direct ancestors has not yet been found. These results appear to favor perhaps a combination of Mainland East Asian Interaction Sphere, and, depending on the date of appearance of northern East Asian component, the Northeastern Seaboard model, but Neolithic Yangtze populations have yet to be fully examined genetically, and future work may uncover additional connections.

Ancient Guangxi Ancestry and Admixture

Guangxi Province, characterized by a mountainous, karst topography and numerous caves, is located in southern China where it borders Vietnam and the Gulf of Tonkin. A 1979 excavation at Laomocao 老磨槽 Cave in Longlin 隆林 County, Guangxi, uncovered human remains including an unusually shaped skull that combined both archaic and modern features. Originally proposed to have belonged to a late-surviving archaic population, the skull was radiocarbon dated to only 11,000 years old. Despite its appearance, the genetic ancestry of the Longlin skull was found, however, to fall completely within modern human variation (Wang et al. Reference Wang, Wang and Xie2021b). Characterizing the ancestry present in ancient Guangxi represented by Longlin was a challenge. Both Northern and Southern Neolithic East Asians shared more ancestry with each other than either did with Longlin, yet compared to more deeply branching East Asian lineages, such as Tianyuan, Papuans, Andaman Islanders, and a deeply diverged ancestry found in an 8,000-year-old hunter-gatherer from Southeast Asia, Hòabìnhian (McColl et al. Reference McColl, Racimo and Vinner2018), Longlin was genetically closer to both northern and southern East Asian groups. This suggested the ancestry found in Longlin Cave, termed ancient Guangxi ancestry, represented a previously uncharacterized deeply branching East Asian lineage (Wang et al. Reference Wang, Wang and Xie2021b) (Figure 1). The coalescence time of the mitochondrial haplogroup identified in Longlin, M71d, and the geographic distribution of related lineages, suggested a possible early migration between southern East Asia and mainland Southeast Asia by at least 22,000 years ago. Interestingly, ancient Guangxi ancestry shares a closer genetic relationship with the Paleolithic Jōmon in Japan than with either of the two major language groups found in present-day Guangxi, Tai-Kadai, and Hmong-Mien. Jōmon and ancient Guangxi ancestry are both similarly related to both northern and southern branches of mainland East Asian ancestry, which is indicative of a similar separation time, yet each shares different specific relationships with members of these groups. Jōmon and Guangxi ancestry thus appear to have separated early from the common East Asian ancestry prior to the diversification of the widespread northern and southern East Asian lineages, but some additional complexity among them remains to be clarified before the history of these two deeply branching ancestries can be more fully reconstructed (Wang et al. Reference Wang, Wang and Xie2021b).

Ancient Guangxi ancestry seems to have left no descendants among modern populations, but a 9,000-year-old individual found in Dushan 独山 Cave, 400 kilometers from Longlin Cave, could be modeled as having 83% ancestry similar to ancient Fujian and a 17% ancestral contribution from a Longlin-related population, showing that ancient Guangxi ancestry persisted in the area in admixed form into at least the Early Neolithic. This could be explained by incoming groups related to southern coastal populations in Fujian mixing with, rather than completely replacing, an indigenous Guangxi population at some point prior to 9,000 years ago. Further analysis showed the ancient Fujian component of Dushan to be more closely derived from Fujian_LN populations such as Tanshishan, and even 2,000-year-old Taiwan islanders, than to older Fujian_EN populations such as Qihe and Liangdao (Wang et al. Reference Wang, Wang and Xie2021b). The admixed Fujian-Guangxi ancestry of Dushan was also identified in two related individuals from Baojianshan 宝剑山 Cave near the Vietnam border dating from between 8,300 and 6,400 years ago. The Baojianshan ancestry, however, had an additional Southeast Asian element not seen before in East Asia, and could be modeled as 72% Dushan-related ancestry and 28% ancestry similar to Hòabìnhian (Wang et al. Reference Wang, Wang and Xie2021b). Hòabìnhian ancestry was first identified from remains associated with the Hòabìnhian Cultural complex in Laos and Malaysia, the examples of which are mainly concentrated in mainland Southeast Asia, where Hòabìnhian ancestry has been presumed to represent indigenous Southeast Asian hunter-gatherer groups. Hòabìnhian ancestry is also closely associated with present-day groups speaking Austroasiatic languages (McColl et al. Reference McColl, Racimo and Vinner2018) (Figure 1). The findings from Baojianshan expand the range of groups carrying Southeast Asian Hòabìnhian ancestry into East Asia, where they interacted with the admixed descendants of ancient Fujian and local Guangxi populations. This interpretation is further supported by proposed Hòabìnhian cultural material at other southern Chinese archeological sites (although not at Baojianshan) (Ji et al. Reference Ji, Kuman and Clarke2016). The genetic results of these studies in ancient Guangxi highlight both the long-term isolation preserving the deeply branching Longlin-related population and the frequent interactions between disparate populations that have occurred in southern East Asia and northern Southeast Asia. The surprising amount of admixture identified that predated agricultural-associated expansions speaks to the broad mobility and communication networks of hunter-gatherer communities in this region.

Lower Yellow River: Shandong

Shandong Province is a coastal region situated on the eastern North China Plain and includes the Yellow River Estuary and the Shandong Peninsula, which separates the Bohai and Yellow seas. Shandong has been home to several important archeological sites, such as the 9,500-year-old Bianbian 扁扁 Cave, and has been home to several successive Neolithic cultures. The earliest of these was the Houli culture (8,250–7,350 BP), which produced evidence of millet and rice grains, although isotopic analysis concluded that these had yet to play a significant dietary role (Hu et al. Reference Hu, Wang, Luan, Wang and Richards2008). The Beixin 扁扁 culture (7,350–6,150 BP) practiced small-scale farming with evidence of dietary diversification, including domestic chickens (Xiang et al. Reference Xiang, Gao and Yu2014). Two major archeological cultures followed, Dawenkou 大汶口 and Shandong Longshan. The Dawenkou culture (6,000–4,600 BP), which co-existed with the larger Yangshao culture centered in the middle reaches of the Yellow River, had clear signs of social stratification, and possessed a broad agricultural suite of domestic pigs, chickens, and cattle, as well as rice and millet farming (Jin et al. Reference Jin, Wagner, Tarasov, Wang and Liu2016). The Songze 崧泽 and Liangzhu 良渚 cultures of the Lower Yangtze River to the South also interacted with the Dawenkou culture, and connections between the Dawenkou culture and West Liao River Basin also occurred through the Jiaodong Peninsula 胶东半岛 (Ren & Wu Reference Ren and Wu2010). The Shandong Longshan culture (4,600–4,000 BP), which coincides with the Late Longshan period of Henan to the west, is known by its settlements on the plains near Mount Tai 泰山. The Shandong Longshan period saw an increased cultural and social sophistication, with the skilled production of distinctive black pottery, worked jade, and small-scale silkworm cultivation. These settlements also oversaw an agricultural shift that intensified millet production over rice, with rice appearing to be reserved primarily for human consumption, while livestock were fed millet (Weisskopf et al. Reference Weisskopf, Deng, Qin and Fuller2015). Beginning after 4,000 years ago, historical records suggest that the Xia dynasty was established, and Chinese history entered the dynastic era (Ren & Wu Reference Ren and Wu2010).

The earliest sequenced genomes from Shandong date to between 9,500 and 8,000 years ago, and have been described as having ancestry that is distinct, yet branching, from that found further inland, as represented by the contemporary 8,000-year-old Yumin 裕民 individual from Inner Mongolia, and similar to that found in the majority of present-day northern East Asians. This ancestry, known broadly as northern East Asian ancestry (nEA), likely finds its present-day wide distribution due to its association with Yellow River agricultural (YR) populations that expanded into neighboring regions during the Neolithic (Ning et al. Reference Ning, Li and Wang2020b; Yang et al. Reference Yang, Fan and Sun2020) as marked by the distribution of Sino-Tibetan languages (Sagart et al. Reference Sagart, Jacques, Lai and List2019; Wang et al. Reference Wang, Wang and Xie2021b), but also may describe the northern East Asian ancestral substrates prior to these movements. In Early Neolithic Shandong, this ancestry is found admixed to a small degree with southern coastal ancient Fujian ancestry, in agreement with archeological evidence of communication among coastal populations increasing from the Early to Late Neolithic as expanded interactions between northern and southern coastal groups appear to have led to more homogenized populations (Yang et al. Reference Yang, Fan and Sun2020).

Although large genomic studies of Middle Neolithic and later Shandong populations have yet to appear, maternally inherited mitochondrial ancient DNA studies have given us insights into the changing demographic structure of this region and its association with important cultural shifts. These have shown a maternal continuity in Shandong connecting the earliest Shandong individual sequenced to date, the 9,500-year-old BianBian with the populations associated with the 4,600-year-old Dawenkou culture at the Fujia 傅家 and Beiqian 北阡 sites, notably in possessing basal examples of the B5b2 haplogroup, which is still found in present-day northern East Asians. Shandong populations from this period were found having mitochondrial haplogroups that occur at high frequencies in both present-day northern (haplogroup D) and southern (haplogroups B and F) East Asians (Liu et al. Reference Liu, Zeng and Sun2021). This is consistent with the genomic data from older Shandong individuals showing a southern coastal East Asian ancestry component, and corresponds to archeological evidence linking the Beixin and Dawenkou cultures with the coastal Yangtze River Basin cultures of Maijabang and Liangzhu to the south, in particular at the Huating 华亭 site in Jiangsu (Zhang Reference Zhang2015). The first observation in Shandong of mitochondrial haplogroups from inland populations, such as haplogroups M8 and A, occurs 3,100 years ago, which may correspond to the expanding influence of the Longshan culture from the Middle Yellow River toward the coast, and the cultural shift to the local Shandong Longshan that began 1,500 years earlier (there are no individuals sampled from the period between 4,600 and 3,100 years ago). The connections between the beginning of the local Longshan culture and this influx of new haplogroups still need to be explored, but at some point after 4,600 years ago, we see an enriched diversity of the matrilineal structure of Shandong corresponding to a reduction of the genetic differences between coastal and inland populations. As a result of this process, present-day East Asians have a greater genetic proximity to younger rather than older Shandong populations (Liu et al. Reference Liu, Zeng and Sun2021).

Middle Yellow River: Henan

The Middle Yellow River region is located further inland from Shandong, in present-day Henan, and has had a major impact on the development of past societies in China as far back as the Neolithic period. The region’s strategic location on the fertile Central Plain surrounded by rivers made it an ideal place to facilitate movement and cultural exchange, and it has long been considered the predominant center for the early beginnings of Chinese civilization (Wang & Zhao Reference Wang and Zhao2022; Zhang Reference Zhang1991). Some of the earliest evidence of agriculture has been found along the alluvial plains of the Middle Yellow River, and from the Early Neolithic a continuous sequence of influential and complex cultures emerged, including the Peiligang 裴李岗 (9,000–7,000 BP), Yangshao (7,000–5,000 BP), and Longshan cultures (5,000–4,000 BP), followed by the Bronze Age cultures of Erlitou 二里头 and Erligang 二里岗, and the Shang and Zhou dynasties.

Yellow River farmers began building semi-permanent settlements in Henan during the Yangshao period and expanded rapidly. Although Early Neolithic genomic information from Yellow River farmers is still lacking, genome-wide analysis from the Middle Neolithic Yangshao period showed the occupants of the Yangshao Wangou 汪沟 and Xiaowu 晓坞 sites had a high affinity with each other, demonstrating a genetic homogeneity among Yangshao settlements that clusters with many northern East Asian populations (Ning et al. Reference Ning, Li and Wang2020b). The appearance of Yangshao pottery in the western Gan-Qing region and Tibet (Han Reference Han2012), and Yangshao rammed-earth construction techniques in the Songze culture near the mouth of the Yangtze River (Kim & Park Reference Kim and Park2017), demonstrates the broad influence the Yangshao culture may have had over surrounding areas. Yangshao-associated YR ancestry from the Middle Neolithic (YR_MN) has likewise been found to be widespread among Middle Neolithic populations from the West Liao River to Inner Mongolia (Ning et al. Reference Ning, Li and Wang2020b) and Tibet (Wang et al. Reference Wang, Yang and Wangdue2023, Reference Wang, Wang and Xie2021b). Mitochondrial haplogroups distributed throughout present-day northern East Asia (e.g. haplogroups A, D4, D5, F2, and G) have been found in the majority of inhabitants of the Yangshao sites studied (Li Reference Li2015; Miao et al. Reference Miao, Liu and Gu2021). The appearance and distribution pattern of haplogroup D5a2a1 in particular are indicative of a rapid expansion (Miao et al. Reference Miao, Liu and Gu2021). This corresponds to the broad appearance of the Yangshao settlements over a relatively short period of time, possibly linked to the warm Mid-Holocene Climatic Optimum (Hou et al. Reference Hou, Xu, Lu, Chen and Lancuo2019; Liu & Chen Reference Liu and Chen2017).

During the Longshan period that followed (5,000–2,900 BP), Yellow River settlement sites are marked by an increase in the consumption of domestic animals and intensified agriculture, along with a distinctive change of pottery style, but the demographic transition between the Yangshao and Longshan cultures is not clearly understood. Longshan populations display a general genetic continuity with the previous Yangshao populations. However, compared with Yangshao populations, Longshan-associated (YR_LN) populations have a notably higher southern East Asian component, reflecting a northward gene flow from southern populations after 4,800 years ago (Li Reference Li2019; Ning et al. Reference Ning, Li and Wang2020b). This increasing influx of southern East Asian ancestry parallels a growth in population density and settlements during the Late Longshan period (Li et al. Reference Li, Gao and Wu2021). The genomic ancestries of the Middle Yellow River populations observed during the Longshan period persist into the Bronze and Iron Ages (Ning et al. Reference Ning, Li and Wang2020b), as do mitochondrial haplogroups D and F, having been present there since at least the Middle Neolithic (Li Reference Li2019). Compared to present-day Chinese populations, YR_MN ancestry is most similar to Naxi and Yi populations centered near the Himalayan foothills of western China, while present-day Han Chinese show additional affinities to southern coastal East Asians than Middle Neolithic or even Iron Age Yellow River populations (Ning et al. Reference Ning, Li and Wang2020b).

Yellow River Origins for Sino-Tibetan Languages

The Sino-Tibetan language family encompasses a large and complex assemblage of languages spoken throughout present-day China (Sinitic branch) and Myanmar, northeastern India, Malaysia, and the Tibetan Plateau (Tibeto-Burma branch). The earliest written attestations date back over 3,000 years to the Shang Dynasty in Henan (and perhaps 800 years earlier at the Middle Yellow River site of Taosi 陶寺 (Demattè Reference Demattè2010)). Proposals for the origin and dispersal route of this language group have included a series of outward radiations from several possible homelands: in Sichuan in southwest China (van Driem Reference van Driem, Blench and Sagart2005), a western origin from the Tibetan Plateau or northeastern India, where the greatest linguistic diversity exists (Blench & Post Reference Blench, Post, Hill and Owen-Smith2014), and an origin among the early agricultural societies of the Middle or Upper Yellow River (Matisoff Reference Matisoff1991). Recent linguistic and genetic studies increasingly point to an origin for the Sino-Tibetan language family to languages spoken by the Yellow River populations of the Early to Middle Neolithic. Time-calibrated Bayesian language phylogenies estimated the earliest split between Tibeto-Burman and Sinitic language groups to have occurred 7,200–8,000 years ago, with a later dispersal accompanying the introduction of millet agriculture (Sagart et al. Reference Sagart, Jacques, Lai and List2019; Zhang et al. Reference Zhang, Yan, Pan and Jin2019). This scenario is in congruence with identification through genetics of YR ancestry universally present among speakers of Sinitic and Tibeto-Burman languages (Ning et al. Reference Ning, Li and Wang2020b; Wang et al. Reference Wang, Wang and Xie2021b). Together, these results suggest a first separation of language groups at some point after 8,000 years ago, with a westward expansion of millet agriculturists speaking proto-Tibeto-Burman languages from the Middle or Upper Yellow River Basin arriving at the Tibetan Plateau at some point around 5,000 years ago, while proto-Sinetic speakers would spread east and southward across the Central Plains.

East Asian Influences on Tibetan Plateau Population Structure

Several genomic analyses present evidence for an early split time between Han and Tibetan lineages predating the introduction of agriculture to the region, at either 12,600 to 6,600 (Qi et al. Reference Qi, Cui and Peng2013), or 15,000 to 9,000 (Lu et al. Reference Lu, Lou and Yuan2016) years ago, with some uniparental markers pointing to even earlier (Qi et al. Reference Qi, Cui and Peng2013). This raises the possibility of contacts between native Tibetans and Upper Yellow River groups prior to the Holocene, although lineage split times do not reflect actual population separation dates and do not specify where these may have occurred. Subsequent contacts or multiple migrations can also reduce these times, but additional evidence such as Sherpa and Han split times of 16,000 to 11,000 years ago and the appearance of Yellow River ancestry along the Himalayan arc far from the northeastern fringe well before the evidence of high-altitude millet farming 5,600 years ago (Lu Reference Lu2023) argues for pre-agricultural interactions, perhaps at lower altitudes, with native Tibetans and Upper Yellow River foragers from Gansu, Qinghai or northern Sichuan. Such contacts would coincide with the warming climate at the close of the glacial period that may have allowed early westward migration into higher altitudes (Lu et al. Reference Lu, Lou and Yuan2016). In this scenario, agricultural-based expansions during the Majiayao period may have encountered a Tibetan substrate already admixed to some extent with northern East Asian ancestry. The early local domestication of yak dating to 7,300 years ago suggests that pastoral nomads may have preceded agricultural societies at high elevations (Qiu et al. Reference Qiu, Wang and Wang2015), who may have traded with lowland agricultural populations.

The oldest genome from the Tibetan Plateau, Zongri5.1K, dates to 5,100 years ago and was recovered from the Zongri site in the Gonghe 共和 Basin, 3,000 meters above sea level in the northeastern plateau (Wang et al. Reference Wang, Yang and Wangdue2023). The Zongri culture is a local Tibetan culture with strong influence from the contemporary Maijiayao culture of the nearby Upper Yellow River region (Lancuo et al. Reference Lancuo, Hou and Xu2023). Genomic information from several younger individuals dating from 4,700 to 3,900 years ago also at Zongri, as well as genomic population data from a nearby younger and higher-elevation site, the 2,900-year-old Pukagongma 普卡贡玛 site in Qinghai, offers a time transect to evaluate the Middle-to-Late Neolithic demographic patterns between the Tibetan Plateau and the Upper Yellow River populations (Figure 3). Zongri5.1K ancestry is distinct from more recent individuals from Zongri in that it lacks a southern East Asian component in addition to having reduced Yellow River ancestry, and has been used to represent Early Ancient Tibetan (EAT) ancestry. When compared with present-day individuals, EAT ancestry is highest among the Qiang, Tibetan, and Sherpa, and thus still can be found in admixed form on the Tibetan Plateau (Wang et al. Reference Wang, Yang and Wangdue2023). EAT ancestry has also been identified in an individual from the 3,400-year-old site of Lubrak in the Himalayan arc (Liu et al. Reference Liu, Witonsky and Gosling2022). More recent Zongri individuals can be modeled as having 40–74% EAT ancestry in addition to ancestry from Early to Middle Neolithic Yellow River populations, with some additional ancestry similar to Yumin in Inner Mongolia. This influx of lowland East Asian ancestry prior to 4,700 years ago does not appear at the higher-elevation Pukagongma site nearly 2,000 years later, which could signal that Neolithic migrations into the plateau may have been sporadic and localized (Wang et al. Reference Wang, Yang and Wangdue2023).

Defining the sources of East Asian ancestry found throughout the Tibetan Plateau can allow us to better understand the timing, routes, and cultural contacts underlying the Neolithic demographic shifts of the region. In general, this incoming ancestry is shared by Early and Middle Neolithic groups from the Yellow River to Inner Mongolia (Wang et al. Reference Wang, Yang and Wangdue2023). In the Himalayan arc this ancestry has been shown to more closely resemble Upper Yellow River Late Neolithic populations over those belonging to the Yellow River Middle Neolithic (Liu et al. Reference Liu, Witonsky and Gosling2022). In addition to the majority component of Yellow River–related ancestry in Tibetans, several other sources appear at different times and places across the plateau, attesting to the multiple migrations and interactions occurring between the surrounding regions over the past few thousand years. In the eastern plateau, ancestry closely shared with an Iron Age Upper Yellow River individual from Dacaozi 大槽子 begins appearing only in the past 700 years. In the southeastern plateau, southern East Asian ancestry appears by at least 2,800 years ago, Central Asian ancestry was found dating back to 1,500 years ago, which appears to be widespread throughout the Himalayan arc, but at lower levels than in present-day Sherpa. South Asian ancestry, as represented by a Bronze Age genome from Turkmenistan, was present at levels between 6% and 14% along the western plateau in individuals dating from 2,300 years ago to recent times (Wang et al. Reference Wang, Yang and Wangdue2023).

Plateau-wide population structures have also been identified. These fall into three geographic clusters of shared ancestry: a “northeast” cluster, a “southeast-central” cluster, and a “south-southwestern” cluster including the Himalayan arc in Nepal (Figure 3). Each of these showed a greater genetic diversity between them in the past than in more recent times, due to a gradual homogenization of trans-plateau populations. The south-southwestern cluster was found to closely follow the Yarlung Tsangpo River Valley, today the most populous area of the plateau, showing the importance of the river to early (before 3,400 years ago) settlement patterns and the maintenance of genetic homogeneity (Wang et al. Reference Wang, Yang and Wangdue2023).

Neolithic, Late Upper Paleolithic, and Deep Paleolithic Roots

Although the admixture date of the earliest East Asian component of Tibetan genomes is estimated to the Early or pre-Neolithic (Lu et al. Reference Lu, Lou and Yuan2016; Qi et al. Reference Qi, Cui and Peng2013), attempts to identify the divergence time of the core Tibetan founding population with the other East Asians using the coalescence times of Tibetan Y chromosome and mitochondrial haplogroups have revealed a curious pattern of Early Neolithic East Asian haplogroups mixed with much deeper branching groups either specific to Tibetans or found also in more distant populations. Over half of Tibetans belong to the East Asian Y chromosome haplogroup D-M174, a haplogroup with origins dating 60,000 to 32,000 years ago and with sister lineages found in both the Andaman Islands and the Japanese archipelago (Liu et al. Reference Liu, Witonsky and Gosling2022; Qi et al. Reference Qi, Cui and Peng2013; Shi et al. Reference Shi, Zhong and Peng2008). In contrast, the lower frequency Y haplogroup O-M117, dates up to 7,000 years ago and is common in the Upper Yellow River from the Middle to Late Neolithic where it is associated with Yangshao and Qijia agricultural cultures, respectively (Liu et al. Reference Liu, Witonsky and Gosling2022). Common Tibetan mitochondrial lineages also show both deeply and more recently branching haplogroups. Of the mitochondrial haplogroups found in the Tibetan Plateau, haplogroup M62 (formerly M16) dates to approximately 28,000 years ago and is rarely found outside of the region (Qi et al. Reference Qi, Cui and Peng2013; Zhao et al. Reference Zhao, Kong and Wang2009), while M9a, found at high frequency in southern East Asians, dates to 9,500 years ago, and the Tibetan-specific haplogroup A dates from 13,000 to 7,000 years ago (Qi et al. Reference Qi, Cui and Peng2013).

This pattern of both deep Upper Paleolithic and Early Neolithic uniparental chromosomal lineages is borne out in nuclear genomic analysis. Although the dominant ancestry across the plateau can be traced to northern East Asian sources primarily from the Yellow River Basin and Inner Mongolia dating from 9,500 to 4,000 years ago, between 6% and 26% of Tibetan genomes are derived from a deeply branching modern human ancestry of unknown origin exclusive to the Tibetan Plateau (Liu et al. Reference Liu, Witonsky and Gosling2022; Lu et al. Reference Lu, Lou and Yuan2016; Wang et al. Reference Wang, Yang and Wangdue2023). No ancestral reference corresponding to this deeply branching “ghost” population has yet been discovered, but it has been loosely modeled as branching from an Asian lineage prior to the lineage leading to the 40,000-year-old Tianyuan, and is similarly distant from the enigmatic 45,000-year-old Siberian Ust-Ishim and the deeply branching southern East Asian branch leading to the Andaman Islanders (Önge), which has also been reported appearing at low levels along the East Asian coast, including in Paleolithic Japanese Jōmon (Wang et al. Reference Wang, Wang and Xie2021b). There may be some link to ancient ANA ancestry as well, since statistical tests show a possible affinity to Amur River individuals from 19,000 to 14,000 years old (Wang et al. Reference Wang, Yang and Wangdue2023). A competing hypothesis has proposed that this highly divergent Tibetan ancestry may have come from an extremely early modern human population from Siberia which split from other modern human lineages shortly after the Out-of-Africa event, and admixed with the remnants of admixed archaic Tibetan Plateau populations. The descendants of this population would have remained in the region until East Asians arrived on the plateau in the Early Neolithic (Lu et al. Reference Lu, Lou and Yuan2016); however, this explanation appears to be implausible since in later analyses, Tibetans appear to lack a corresponding increase in archaic ancestry (Liu et al. Reference Liu, Witonsky and Gosling2022; Wang et al. Reference Wang, Yang and Wangdue2023). The similar proportion of this ancestry throughout the Tibetan Plateau populations signifies that it likely derives from a common source population and, thus, may represent the autochthonous Plateau population prior to the arrival of post-LGM East Asians. Given the depth of the lineages, the origin of this population may have first arrived on the Plateau prior to the LGM, and may have admixed with remnant Denisovan populations and acquired the EPAS1 gene advantageous to high-altitude habitation (Huerta-Sánchez et al. Reference Huerta-Sánchez, Jin and Asan2014; Lu et al. Reference Lu, Lou and Yuan2016). This time period would be close to the most recent estimated introgression date of this allele of around 48,000 (16,000–59,500) years ago (Zhang et al. Reference Zhang, Witt and Bañuelos2021c). A warmer humid period in the Upper Paleolithic between 40,000 and 30,000 years ago would have made the Tibetan Plateau more habitable than at the present, and may have facilitated the pre-LGM entry of early modern humans (Yang et al. Reference Yang, Shi, Braeuning and Wang2004). However, selective pressure on this allele is thought not to have begun until much later (9,000 years ago) on standing variation, so the allele may have alternatively been introduced at lower altitudes (Zhang et al. Reference Zhang, Witt and Bañuelos2021c), not rising in frequency until the receiving population later settled on the plateau. A similarity reported between Paleolithic Tibetan Plateau archeology sites and contemporary sites of East Asia may give some clues as to the geographic origin of this ghost population (Gao et al. Reference Gao, Zhou and Guan2008).

From our current understanding from ancient genetic data, the formation of the present-day Tibetan population was likely to have involved an initial migration into the Tibetan Plateau during a mild climatic period prior to the Holocene by a deeply divergent modern human ‘ghost’ population that has only partially been identified. No evidence of migrations into the plateau during the LGM has been found, but interactions at the eastern edge may have occurred after the climate improved. Later, increasing gene flow from surrounding lowland populations may have reached its maximum with the introduction of agriculture, followed by more sporadic migrations into historical times. Further research will be needed to more fully characterize the Upper Paleolithic substrate, as well as to elucidate the mechanisms for the formation of the observed Late Neolithic and Bronze Age population structures.

High-Altitude Adaptations: EPAS1 and ELGN1

The ability of humans to live at the hypoxic high-altitudes of the Tibetan Plateau has been greatly facilitated by genetic adaptations that allow the more efficient use of oxygen by the blood in low oxygen environments. Among these, EPAS1 and its negative regulator ELGN1 have been identified by the strength of the selective sweeps of their diverged genetic variants (Peng et al. Reference Peng, Yang and Zhang2011). The introgressed Denisovan variant of EPAS1 is correlated with low hemoglobin and a reduced physiological response to hypoxia. An additional beneficial ELGN1 variant is found at frequencies of 64.3–75.8% among present-day Tibetans, with the frequency correlated to elevation. The onset of selection of the ELGN1_D4E allele has been estimated at 8,400 years ago. It is believed to play a role dependent upon the earlier EPAS1 variant, improving its activity, and thus has more recent selection dynamics (Xiang et al. Reference Xiang, Ouzhuluobu and Peng2013). Observing the occurrence of EPAS1 across the Plateau over the past 5,100 years revealed the oldest sample, Zongri5.1K, to have been homozygous for the derived EPAS1 allele, and an increasing allele frequency over time could be observed (Wang et al. Reference Wang, Yang and Wangdue2023). The derived EPAS1 variant had an allele frequency among plateau populations of 36% between 5,100 and 2,500 years ago, 47% 2,400 to 1,900 years ago, and 59% 1,600 to 700 years ago (Wang et al. Reference Wang, Yang and Wangdue2023), rising until the present-day frequency for the derived EPAS1 allele among Tibetans of 86%. The rate of increase observed for the EPAS1 allele frequency appears to support the recent selective pressure relative to the much older introgression time reported for this variant (Zhang et al. Reference Zhang, Witt and Bañuelos2021c).

Neolithic and Early Bronze Age

The earliest post-LGM genomes characterized in Mongolia date to the Late Neolithic and Early Bronze Age between roughly 7,000 and 5,000 years ago, and show that at this early date, the region was already home to diverse populations occupying different geographical areas. In northeastern Mongolia, Neolithic sites around the Kerulen River, associated with a hunting and fishing-based culture practicing small-scale agricultural activities, were found to have a similar ANA ancestry to that found among contemporary populations around the Amur River Basin and Primorye further east (Jeong et al. Reference Jeong, Wang and Wilkin2020; Wang et al. Reference Wang, Yeh and Popov2021a). In fact ANA ancestry appears to have been the primary ancestral component across much of northern and eastern Mongolia during the Late Neolithic and Early Bronze Ages. It existed unadmixed in eastern Mongolia from the Neolithic to the Middle and Late Bronze Age (MLBA), where it is also found in MLBA Ulaanzuukh culture-associated burials there. Outside of eastern Mongolia, ANA ancestry is found admixed with other minor ancestral components. A Neolithic burial near the Egiyn-Gol River Valley in northern Mongolia had ANA ancestry admixed with 17% ancestry associated with ANE-related sources. This ancestry profile matches that found at Early Neolithic Fofonovo culture burial sites further to the north surrounding Lake Baikal in Siberia, where individuals were found with ANA ancestry at 83–87% and the remainder belonging to an ANE-related component. This particular admixed ancestry persisted in the Baikal region from the Neolithic (Kitoi culture) through the Early Bronze Age (Glazkovo culture) and later, during which time the ANE component gradually rose from 6% to 20% (Barros Damgaard et al. Reference Damgaard, Marchi and Rasmussen2018; Jeong et al. Reference Jeong, Wang and Wilkin2020).

The first appearance of western steppe cultures in Mongolia is found in the Early Bronze Age, possibly entering the region through the Upper Yenisei and Sayan Mountains, or the Altai Mountain range (Janz et al. Reference Janz, Odsuren and Bukhchuluun2017). Five-thousand-year-old Afanasievo burials in Shatar Chuluu near the Khangai Mountains in central Mongolia are genetically identical to the Western Steppe Herder (WSH) ancestry found in Afanasievo groups from Yenisei in the Altai (Allentoft et al. Reference Allentoft, Sikora and Sjögren2015; Jeong et al. Reference Jeong, Wang and Wilkin2020; Narasimhan et al. Reference Narasimhan, Patterson and Moorjani2019). This links the introduction of pastoralism in Mongolia to the eastward expansion of western steppe culture. Chemurchek burials in western Mongolia, which have been archeologically linked to sites further west (Kovalev Reference Kovalev and Bobrov2016), have shown greater genetic diversity. The ancestry found at Chemurchek burials from the southern Altai site of Yagshiin Huduu could be modeled as western steppe ancestry, but with an ANE component similar to Botai groups from Kazakhstan. A smaller component was similar to ancient Iranian ancestry associated with the Central Asian Bronze Age Bactria-Margiana Archaeological Complex (BMAC), previously identified in Kazakhstan (Narasimhan et al. Reference Narasimhan, Patterson and Moorjani2019). Chemurchek burials from the northern Altai, however, were characterized as 80% ANA ancestry and the rest being that of the southern Altai Chemurchek individuals, which may be explained by incoming western Chemurchek-associated groups admixing locally with ANA ancestry, perhaps in the areas around the Altai or western Sayan Mountains (Jeong et al. Reference Jeong, Wang and Wilkin2020). ANA ancestry also appears in Inner Mongolia to the south in the 8,400-year-old individual from Yumin village site near Wulanchabu, although at Yumin, the ANA ancestry appears admixed with East Asian ancestry along the Tibetan cline (Yang et al. Reference Yang, Fan and Sun2020). As discussed earlier, an east-west cline among the ANA ancestry was identified, with western Mongolian sites showing an increase in ANE admixed ancestry, and in the Amur River region to the east, a component with affinity to ancestry found in Paleolithic Jōmon groups from Japan was present up to 13% (Wang et al. Reference Wang, Yeh and Popov2021a). The distribution pattern of ANA ancestry in Mongolia and surrounding regions suggests a common northern East Asian ancestry was dispersed over a wide geographical region from west Baikal in Siberia to the northeastern Asian coastal Primorye and the Amur River Basin, and south into Inner-Mongolia, admixed with ANE groups in Siberia, and occasionally with incoming Bronze Age populations from the western steppes in northwestern Mongolia. The association of this ancestry with present-day Transeurasian language speakers, as mentioned earlier, may give some indication of the geographical origins of the Transeurasian language family (Robbeets et al. Reference Robbeets, Bouckaert and Conte2021).

Three Bronze Age Populations

Large ancient genomic studies of over 300 individuals beginning with the MLBA revealed the emergence of a geographic-genetic structure across the region consisting of three distinct populations (Jeong et al. Reference Jeong, Wang and Wilkin2020). These ancestries have been termed Khövsgöl_LBA, Altai_MLBA, and Ulaanzuukha_SlabGrave, and are found in northern central, western, and eastern Mongolia, respectively. Khövsgöl_LBA defines the previously described ancestry with a majority ANA admixed with ANE found chiefly between the Kanghai Mountains and Lake Baikal. A genetic a continuity was found including Late Bronze Age (LBA) pastoralists and both Neolithic occupants of this region and hunter-gatherers of Lake Baikal to the north, demonstrating a persistence of this ancestry over 3,000 years in the area (Figure 4).

Altai_MLBA ancestry is defined as Khövsgöl_LBA admixed with a newly appearing western steppe ancestry matching that of Sintashta culture-associated steppe herders originating from east of the Urals (Allentoft et al. Reference Allentoft, Sikora and Sjögren2015; Narasimhan et al. Reference Narasimhan, Patterson and Moorjani2019). This incoming WSH population, which had genetic links to the Corded Ware cultures of central and eastern Europe (Mathieson et al. Reference Mathieson, Lazaridis and Rohland2015), likely arrived as part of an eastward expansion of Sintashta groups from the western steppe, and is found predominantly in western and northern Mongolia. Analysis of the age of admixture of this Sintashta WSH component in MLBA burials in western Mongolia estimates an arrival roughly 300 years prior to when the individuals lived, or about 3,500 years ago, and was found to be distinct from the earlier WSH ancestry associated with the Afanasievo (Jeong et al. Reference Jeong, Wang and Wilkin2020). This time period coincides with a rising use of horses as transport animals on the Mongolian steppe, and includes the earliest known appearance of horse milking (Wilkin et al. Reference Wilkin, Ventresca Miller and Taylor2020). Horses associated with the Sintashta expansion are ancestral to modern domestic horses, and the intensified use of horses of this period can be linked to these incoming steppe pastoralists (Librado et al. Reference Librado, Khan and Fages2021). Several individuals within the Altai_MLBA group required additional ancestral contributions associated with the BMAC, similar to the ancestry appearing in earlier Chemurchek individuals (Jeong et al. Reference Jeong, Wang and Wilkin2020). An interesting aspect of the MLBA population survey of Mongolia was the apparent replacement of much of the EBA ancestry profile found in Afanasievo and Chemurchek populations, although analysis of MLBA populations across a broader region will be required to better address the fate of these EBA populations. Altai_MLBA ancestry persisted into the Iron Age and is found along with BMAC ancestry in the succeeding Scytho-Siberian Sagly/Uyuk culture in northwestern Mongolia. This Sagly/Uyuk genetic profile, Altai_MLBA admixed with Iranian-related BMAC-like ancestry, is also represented in other Scythian-complex cultures, being similar to that of the Tagar north of the Sayan Mountains and the Saka to the west (Jeong et al. Reference Jeong, Wang and Wilkin2020). The evolving ancestry signatures in western Mongolia attest to the ongoing exchange and mobility between western steppe populations and inhabitants of western Mongolia since at least the Early Bronze Age.

The third ancestry identified, Ulaanzuukh_SlabGrave, was restricted to central, eastern, and southeastern Mongolia. Like contemporary groups in central and western Mongolia, the Ulaanzuukh culture also practiced a pastoralist economy, but Ulaanzuukh burials differ from those further to the west, with bodies being placed in prone position in a rectangular structure composed of flat stones, sometimes marked with undecorated standing stones (Dashtseveg et al. Reference Dashtseveg, Dorjpurev and Myagmar2013). Although dating to the Late Bronze Age, Ulaanzuukh cultural material shows little use of metal, consisting primarily of stone tools and ceramics (Wright et al. Reference Wright, Ganbaatar, Honeychurch, Byambatseren and Rosen2019). The Ulaanzuukh_SlabGrave ancestry associated with Ulaanzuukh graves was determined to contain an average of 75% ANA, similar to the pre–Bronze Age eastern Mongolian ancestry found throughout the Amur River Basin, but with a 25% component derived from Khovsgol_LBA ancestry from northern Mongolia, with one Ulaanzuukh individual having as high as 63.5% Khovsgol_LBA. Central and eastern Mongolian individuals from the succeeding Slab Grave culture also carried this ancestry, some with even greater proportions of Khovsgol_LBA (Lee et al. Reference Lee, Miller and Bayarsaikhan2023).

Xiongnu and the Formation of the Present-Day Mongolians

The genetic separation between the populations living in eastern and western Mongolia from the Neolithic to the Iron Age is remarkable, especially since the same nomadic pastoralism was practiced across the whole of Mongolia dating back to the Bronze Age, and despite the broad use of domestic animals originating from the west. This structure is indicative of a transfer of economic practices from the west to the east without a large detectable genetic impact on eastern populations. This order changes during the Xiongnu period in the Iron Age. The Xiongnu confederation, a nomadic equestrian steppe culture, is the first historically attested empire of the eastern Eurasian steppe, eventually extending from Mongolia into present-day Inner Mongolia, Russia, Kazakhstan, Kyrgyzstan, and parts of northwestern China (Miller Reference Miller2024). Sixty Xiongnu burials were genetically analyzed dating from the beginning and end of the Xiongnu period, which began roughly 2,200 years ago and lasted for nearly 300 years. Although some early Xiongnu period individuals from northern central Mongolia derived the majority of their ancestry from populations similar to the Early Iron Age Scythian-complex-related Sagly/Uyuk in the Altai region of western Mongolia, admixed with approximately 8% additional Iranian BMAC ancestry, others from the same region were admixed between Sagly/Uyuk ancestry and ancestry similar to that of Ulaanzuukh_SlabGrave, still others had unadmixed Ulaanzuukh_SlabGrave ancestry (Jeong et al. Reference Jeong, Wang and Wilkin2020). The heterogeneity present in the early Xiongnu, which involved admixed and unadmixed Early Iron Age ancestries from across Mongolia, increased in later Xiongnu period individuals, with additional ancestral elements. In addition to more extensive admixtures involving early Xiongnu components as well as ancestries present in Mongolia during the MLBA, such as the Khövsgöl_LBA, late Xiongnu also contained ancestry related to another WSH Scythian-related culture distinct from the Sagly/Uyuk, the Sarmatians, who originated from the western Steppe of Central Asia. These results point to the continuation of interactions between western steppe cultures in Mongolia that began in the Bronze Age. An East Asian ancestry separate from ANA was also represented, close to modern-day Han Chinese, which may reflect the increasing interactions with the Han Empire, involving both military disputes and the use of marriage alliances to secure diplomatic relationships (Jeong et al. Reference Jeong, Wang and Wilkin2020).

After the eventual defeat of the Xiongnu by the Han dynasty, several confederations rose to power over the Mongolian Plateau. Genomic information from the medieval period Turkic and Uyghur Khagnates document a changing demography with persistently heterogeneous populations. An additional western steppe ancestry closer to the Alans than to the Sarmatians appears in several Uyghur burials, and Iranian-related and Ulaanzuukh_SlabGrave ancestry is found in both Turkic- and Uyghur-period individuals (Jeong et al. Reference Jeong, Wang and Wilkin2020). The Mongol Empire, which dates to the thirteenth and fourteenth centuries and eventually encompassed much of the Eurasian continent, began by uniting several diverse nomadic confederations across Mongolia and neighboring regions. Sixty-one Mongol Empire–period individuals were analyzed and exhibited a greater genetic homogeneity than the preceding populations of the region, demonstrating a close continuity with the present-day population of Mongolia. This profile is marked by an increase in East Asian genetic affinities, as well as the disappearance of ANE-related components present in earlier ancestries of the northern central and western regions. Overall, the population dating to the Mongol period consisted of three components, with the majority being represented by Ulaanzuukh_SlabGrave ancestry, and included a Han-related component and a western Eurasian component most closely matching that of the Alans (Jeong et al. Reference Jeong, Wang and Wilkin2020). Comparisons of Mongol-era individuals with present-day populations showed more than half of them to be closely related to present-day Mongolic language speakers. These analyses illustrate how the genetic homogenization process across the Mongolian Plateau that began with the rise of nomadic pastoralist confederations such as the Xiongnu had led to the early profile of the present-day Mongolic population by the time of the Mongol Empire.

The Tarim Basin Mummies

Numerous burial sites around the Tarim Basin have been excavated containing the naturally mummified remains of Bronze and Iron Age individuals. These sites date from 4,100 to 2,000 years ago, placing the older burials among the earliest Holocene archeological cultures in the region. Both western and eastern Eurasian physical characteristics have been reported among these mummies, and the grave goods associated with them, such as fermented cheese, as well as their woolen clothing, suggest a western Eurasian steppe origin (Abuduresule et al. Reference Abuduresule, Li and Hu2007; Mallory & Mair Reference Mallory and Mair2000; Thornton & Schurr Reference Thornton and Schurr2004; Yang et al. Reference Yang, Shevchenko and Knaust2014b). Domestic grains of both western wheat and eastern millet have also been recovered from some of the graves (Yang et al. Reference Yang, Yang and Li2014a). Given the western Eurasian features of these burials, it has been proposed that these populations may have been descended from Indo-European groups who brought Tocharian languages into the region (Hemphill & Mallory Reference Hemphill and Mallory2004; Peyrot Reference Peyrot, Kelder, de Jong and Mouret2017). Among the oldest of these was found near Xiaohe 小河 in the eastern Lop Nur region of the Tarim Basin. Mitochondrial and Y-chromosome analyses of this site had identified a mixture of western Eurasian haplogroup lineages, such as H, U5, and K, as well as eastern haplogroups such as D and C4, similar to those found in southern Siberia, demonstrating that both western and eastern Eurasians inhabited the Tarim Basin in the Bronze Age. Individuals from more recent layers also carried haplogroups more commonly found in South and Central Asia, such as U7 and M5, which may perhaps represent a later arrival of BMAC-related ancestry. The Y-chromosome haplogroup R1a1a, associated with the MLBA Andronovo western steppe culture, was found among several males (Li et al. Reference Li, Li and Cui2010, Reference Li, Ning and Hagelberg2015). Together, these results were the first genetic confirmations of eastern and western origins of the Tarim Basin mummies.

Genomic analysis of Xiaohe, Gumugou 古墓沟 burial sites in the eastern, and Beifang 克里雅河北方 in the southern Tarim Basin gave a fuller picture of the ancestral origins of the Tarim Basin mummies, and how these populations have changed over time. In contrast to the earlier mitochondrial and Y-chromosome findings, individuals from the oldest layers of the eastern sites were found to lack any western steppe ancestry, but instead showed 72% of their ancestry being enriched in an ANE component that closely matched the 16,000-year-old Mesolithic individual Afontova Gora 3 from the Yenisei River in Siberia, and the remainder being similar to Early Bronze Age ancestry from the Baikal region. This ancestral profile, known as Tarim_EMBA1 ancestry, is believed to have represented the ancestry present in the region prior to the arrival of groups from the west. The Beifang site further to the south had a similar ancestry profile, but with slightly more Baikal_EBA ancestry. The identification of Tarim_EMBA1 ancestry allowed Baikal_EBA ancestry to be alternatively described as a mixture of 75% ANA ancestry and 25% Tarim_EMBA1. Furthermore, the Tarim ancestries were relatively homogeneous with low levels of diversity, pointing to a possible population bottleneck, and were found to have been genetically isolated for an extended period, with the estimated date of the admixture of the ANA and Baikal_EBA ancestries to have been roughly 9,000 years ago (Zhang et al. Reference Zhang, Ning and Scott2021b). Proteomic analysis of the dental calculus of several mummies confirmed the adoption of dairy consummation in this population; even though similar to other pastoralists of Mongolia, they lacked the allelic variants conferring lactase persistence (Zhang et al. Reference Zhang, Ning and Scott2021b). This also indicates an early adoption of pastoralism, and other western influences, in the Tarim Basin independent of genetic exchange. While the language spoken by the EMBA Tarim Basin populations is unknown, these results cast doubt that Tocharian was spoken among them at this time, at least prior to interactions with Afanasievo-related groups.

Bronze Age Xinjiang

The timing and means by which western Eurasian and other ancestries arrived to the Tarim Basin populations were addressed by larger ancient genomic studies of Bronze Age populations across the broader Xinjiang region. Genomic analysis of twenty individuals from six sites in northern, western, and southern Xinjiang identified four ancestries that were already admixed to various degrees during the Bronze Age in the northern and northwestern regions. In the Ili River Valley and northern Xinjiang near the Altai Mountains, the dominant Bronze Age ancestries were ANE (similar to Tarim_EMBA1), and Afanasievo-related ancestry, with a minor East Asian ANA ancestry modeled by Neolithic Baikal populations, and a rare Central Asian BMAC-related component (Kumar et al. Reference Kumar, Wang and Zhang2022). Most of the individuals had a high affinity for each other, showing some degree of local homogeneity, but overall ancestry components were highly dynamic between populations and individuals, with several outliers being relatively unadmixed. Afanasievo-related ancestry demonstrates the establishment of Indo-European western steppe herders in Xinjiang, and was further supported by the appearance of Afansievo-associated Y-haplogroup R1b1a in 67% of the individuals with Afanasievo ancestry. One 5,000-year-old individual of the Songshugou site in the north was found to contain 92% East Asian-related ancestry, more similar to that found at the contemporary Shamanka site in the Baikal region than to Mongolian ANA sources, with the rest of this ancestry being Tarim_EMBA1. Two Chemurchek-associated individuals contained a BMAC component, and showed a similar genetic profile to Chemurchek populations from Mongolia discussed previously, confirming a trans-Altai genetic continuity among Chemurchek culture. The date of the integration of this BMAC component was estimated to have occurred between 5,300 and 4,600 years ago, more recent than the admixture date of ANA ancestry in Xinjiang populations, which was approximately 6,400 years ago. The BMAC component overlaps slightly with the admixture date calculated for admixture with Afanasievo-related ancestry of between 4,900 and 4,600 years ago. Together, these dates can give a broad idea of the relative sequence of the interactions of different ancestries in the region. Although a larger population sampling would be needed to better refine these estimated dates of admixture, the Afanasievo admixture date does support the linguistic evidence of the timing of the splitting of Tocharian languages from the rest of the Indo-European family (Kumar et al. Reference Kumar, Wang and Zhang2022). Another study centered around the Dzungarian Basin of five individuals from three 5,000-year-old sites associated with the Afanasievo culture showed similar genetic profiles, having ancestral components of 50–70% Afanasievo-related, 19–36% Tarim_EMBA1 (ANE), and 9–21% Baikal_EBA (ANA). This study found that this profile, Dzungharian_EBA, could further be used to model Chemurchek-associated ancestry as a mixture of Dzungarian_EBA, Tarim_EMBA1, and BMAC, giving a clearer picture of the formation of Chemurchek populations (Zhang et al. Reference Zhang, Ning and Scott2021b).

Archeological evidence records the appearance of Andronovo and Sintashta cultures in the Tianshan Mountains by the beginning of the MLBA at least 3,000 years ago (Betts et al. Reference Betts, Vicziany, Jia and Castro2019b), sometimes referred to as the Eastern Fedorovo variant (Jia et al. Reference Jia, Betts, Cong, Jia and Dupuy2017). Although genetically similar, individuals associated with the Andronovo and the preceding Sintashta WSH cultures can be differentiated from earlier Afanasievo-related WSH individuals by the presence of Anatolian farmer ancestry, which is assumed to have been derived from interactions with European agriculturalists of the Corded Ware culture (Allentoft et al. Reference Allentoft, Sikora and Sjögren2015; Damgaard et al. Reference Damgaard, Marchi and Rasmussen2018). The transition to the MLBA in north and northwest Xinjiang is indeed marked by an increase of Anatolian farmer ancestry, with one individual from the Ili Valley having unadmixed Sintashta ancestry (Kumar et al. Reference Kumar, Wang and Zhang2022). A separate genomic study of an individual from the Baigetuobie 拜格托别 site of the eastern Tianshan Mountains documents an Andronovo-associated ancestry in Xinjiang as early as 3,600 years ago (Zhu et al. Reference Zhu, Ma and Zhang2021). Combined with the archeological data, these results together confirm that from the onset, the MLBA expansion of Andronovo steppe cultures into Xinjiang involved the migration of populations associated with these cultures rather than the horizontal spreading of Andronovo cultural material.

The question of the origin of the earliest Holocene Xinjiang populations thus becomes more complex than the prevailing hypotheses. We note the presence of groups bearing ancestry from the steppes, Central and South Asia, as well as East Asian ancestry from Siberia are all established in Xinjiang at the beginning of the Bronze Age. While the sources of the BMAC component can be explained through contact with the Chermurchek-related populations in the Altai region, arrival of this ancestry into Xinjiang through the IAMC route cannot be discounted. Surprisingly, the earliest human remains from the Tarim Basin were found to belong to neither of these groups, but to represent a deeply branching, seemingly earlier substrate of the most recently surviving carriers of ANE ancestry from Paleolithic Siberia, the unadmixed form of which has vanished from present-day populations (Raghavan et al. Reference Raghavan, Skoglund and Graf2014). These people appeared to have practiced an agro-pastoralist culture that adopted material and domestic animals from the western steppe and Central Asian traditions, which, along with their associated ancestries, have been active in northern Xinjiang at least 1,000 years earlier, before eventual genetic exchange reached Xiaohe. Around the Dzungarian Basin, the Tarim_EMBA1 ancestral component found admixed with WSH and ANA groups indicates that perhaps they once occupied a larger territory, but more genetic information from earlier, pre–Bronze Age populations will need to be recovered before the population history of the Bronze Age people of the Dzungarian Basin can be more fully reconstructed.

Iron Age and Present-Day Populations

As in Mongolia, the onset of the Iron Age saw an increase in mobility linked to nomadic equestrian expansions, which is marked by a new influx of ancestry from surrounding regions into Xinjiang. Along with the continued increase in Andronovo-associated steppe MLBA ancestry, an increase in both BMAC ancestry from Central Asia and diverse East Asian ancestries is observed. Despite the growing spread of MLBA steppe ancestry, distinct Afanasievo-related EBA steppe ancestry still remained in the region. An analysis of the Iron Age site of Shirenzigou 石人子沟 in the Dzungarian Basin found the inhabitants to be composed of Saka and Afanasievo steppe ancestry without any evidence of Anatolian farmer components that would indicate Andronovo gene flow (Ning et al. Reference Ning, Wang and Gao2019). In other sites, these two WSH ancestries were found to co-exist, with both Afanasievo-associated R1b and Andronovo-associated R1a1a Y-chromosome haplogroups being observed during this period. Populations at four sites were found to be admixed between the two steppe ancestries, presenting examples of preexisting Afanasievo-related descendants integrating with, rather than being replaced by, more recently arriving Andronovo populations (Kumar et al. Reference Kumar, Wang and Zhang2022).

The increasing BMAC ancestry in the Iron Age has been linked to incoming migrations of the Saka, an Indo-Iranian-speaking tribe of the broader Scythian Cultural group who appeared in the eastern Kazakh Steppe and Tianshan mountains in the first millennium BCE, before later settling in the Tarim Basin (Kuzʹmina & Mallory Reference Kuzʹmina and Mallory2007). In the Tarim Basin, the Saka confederacy controlled a large region centered on the city of Khotan until their defeat by the Turkic Kara-Khanid Khanate in 1006 CE (Millward Reference Millward2007). Saka ancestry from Tianshan has been modeled broadly as a largely MLBA steppe ancestry admixed with East Asian ancestry from southern Siberia, and Iranian ancestry related to the BMAC component (Damgaard et al. Reference Damgaard, Marchi and Rasmussen2018; Jeong et al. Reference Jeong, Wang and Wilkin2020), which may have increased throughout the Iron Age (Gnecchi-Ruscone et al. Reference Gnecchi-Ruscone, Khussainova and Kahbatkyzy2021). In some individuals, a new South Asian component related to Andaman Islanders (Önge) also appears, possibly reflecting changing population dynamics and expanded contacts between Central and South Asia through the IAMC (Kumar et al. Reference Kumar, Wang and Zhang2022).

Not only was the increase in East Asian ancestry signifying greater contacts between Xinjiang and East Asia, but the sources of East Asian ancestry are also seen to expand. The ANA-associated ancestry from southern Siberia and Mongolia that was present during the Bronze Age continued to increase, which may in part have been due to the diffusion of the Scythian-related Pazyryk culture of the Altai region (Kumar et al. Reference Kumar, Wang and Zhang2022). Additionally, ancestry more closely related to the Xiongnu, as well as YR and present-day Han ancestry, began to appear. These newer ancestries mirror the westward expansion of the Xiongnu as they attempted to establish control of the region (Millward Reference Millward2007). Expanding trade networks and military contacts with the Qin and Han Dynasties through the Hexi corridor are also a likely source for incoming YR-related ancestries (Allen et al. Reference Allen, Yu and Yang2022). New East Asian Y-chromosome haplogroups such as O3a2c make their first appearance in southern and eastern Xinjiang during the Iron Age (Kumar et al. Reference Kumar, Wang and Zhang2022). Population profiles of the following historical period of Xinjiang closely resembled those of the preceding Iron Age, clustering mostly with the Saka and Steppe populations, but also representing individuals with elevated East Asian or BMAC-related ancestry. Population continuity with these groups can still be detected among some Xinjiang populations. Iron Age Xinjiang ancestral components are found in present-day Uyghur and Central Asian genomes, with increased steppe ancestry in Iron Age individuals corresponding to increased affinity to present-day Uyghur populations (Kumar et al. Reference Kumar, Wang and Zhang2022). The intricacy of population and cultural interactions as shown through paleogenomic analysis of Xinjiang, arguably the most demographically complex East Asian region yet studied, has given valuable insights to previous ideas of population migrations and the communication networks through which both cultural material and ideas flowed as they transfer between the eastern and western divisions of the continent. There are still mysteries, however, regarding the role this region may have played prior to the Bronze Age, although the paucity of human remains from this time period promise further research in this area to be challenging.

Jōmon Origins

The origin of the Paleolithic population of the Japanese islands has been of high interest due, on one hand, to the evidence of extended cultural continuity throughout this period (Habu Reference Habu2004), and on the other, to the stark morphological differences observed between Paleolithic remains and contemporary mainland East Asians and Japanese. A degree of morphological diversity has also been identified among Jōmon populations, and skeletal and dental features of different populations were found to resemble either Southeast Asians (Matsumura et al. Reference Matsumura, Hung and Higham2019; Turner Reference Turner1987), Upper Paleolithic Asians (Yamaguchi Reference Yamaguchi1982), or Northeastern Asians (Hanihara & Ishida Reference Hanihara and Ishida2009), as well as present-day indigenous populations of the Ainu of Hokkaido and Ryukyu Islanders (Jinam et al. Reference Jinam, Kanzawa-Kiriyama and Saitou2015; Yamaguchi Reference Yamaguchi1982). In contrast to this, the similarity of many present-day Japanese with mainland East Asians has led to the proposal of a dual-layer model of Japanese population origins. In this model, a morphologically distinct founding population, represented by the Paleolithic Jōmon, which may be ancestral to Ainu and Ryukyu Islanders, is later admixed with groups arriving from the Korean Peninsula during the Neolithic (Hanihara Reference Hanihara1991). Early attempts to identify the genetic origin of the Jōmon people used mitochondrial, Y-chromosome, and low-coverage genomic information. These studies confirmed the uniqueness of Jōmon populations, and proposed various influences from Tibetans, Southern Siberians (Adachi et al. Reference Adachi, Shinoda and Umetsu2011; Kanzawa-Kiriyama et al. Reference Kanzawa-Kiriyama, Saso, Suwa and Saitou2013), or Andaman Island-related Hòabìnhian from Southeast Asia (McColl et al. Reference McColl, Racimo and Vinner2018). A Tibetan link was argued based on the presence of the deeply branching Y-chromosome haplogroup D-M55, which is present in a majority of Ainu (Hammer et al. Reference Hammer, Karafet and Park2006; Watanabe et al. Reference Watanabe, Naka and Khor2019) and is a sub-branch of the rare D-M174 found at high frequency in Tibet (Qi et al. Reference Qi, Cui and Peng2013; Shi et al. Reference Shi, Zhong and Peng2008). Proposed gene flow from southern Siberia relied on mitochondrial haplogroups predominant among Hokkaido and Honshu Jōmon-associated individuals and also found among southern Siberians, but rare in present-day Japanese and other East Asian groups, such as N9b, a lineage dating to 22,000–20,000 years ago (Adachi et al. Reference Adachi, Shinoda and Umetsu2011; Jinam et al. Reference Jinam, Kanzawa-Kiriyama and Saitou2015; Kanzawa-Kiriyama et al. Reference Kanzawa-Kiriyama, Saso, Suwa and Saitou2013). Some clarity was brought to these early findings by more recent studies involving higher coverage ancient genomic DNA from additional ancient Jōmon individuals. Despite the shared rare Y-chromosome haplogroup connection, no special genomic affinities were detected between Jōmon and either ancient or modern Tibetans (Boer et al. Reference Boer, Yang, Kawagoe and Barnes2020; Gakuhari et al. Reference Gakuhari, Nakagome and Rasmussen2020; Kanzawa-Kiriyama et al. Reference Kanzawa-Kiriyama, Jinam and Kawai2019). Nor was any specific gene flow from Hòabìnhian found (Boer et al. Reference Boer, Yang, Kawagoe and Barnes2020; Cooke et al. Reference Cooke, Mattiangeli and Cassidy2021; Gakuhari et al. Reference Gakuhari, Nakagome and Rasmussen2020; Yang et al. Reference Yang, Fan and Sun2020), although it should be noted that one group was able to model Jōmon as containing 44% of a component related to Andaman Islanders (Önge), which is also found in Hòabìnhian. Since traces of this deeply branching ancestry have been found along coastal Southeast Asian populations, this was proposed to be a remnant of an initial coastal migration route populating East Asia (Wang et al. Reference Wang, Wang and Xie2021b). These higher-resolution analyses did identify elements of shared genetic ancestry between Jōmon and coastal, but not inland, populations from both northern and southern East Asia, especially from eastern Siberia, such as Okhotsk-Primorye region populations of coastal Siberia and the Ami and Ayatal Austronesians from the island of Taiwan, as well as coastal southern East Asians (Boer et al. Reference Boer, Yang, Kawagoe and Barnes2020; Gakuhari et al. Reference Gakuhari, Nakagome and Rasmussen2020; Kanzawa-Kiriyama et al. Reference Kanzawa-Kiriyama, Jinam and Kawai2019; Wang et al. Reference Wang, Wang and Xie2021b; Yang et al. Reference Yang, Fan and Sun2020). Whether this signal represents a deep ancestral current indicative of an early coastal route into East Asia or more recent interactions among coastal populations, or perhaps both, has yet to be fully investigated. An additional mystery comes from studies reporting some degree of gene flow between Jōmon and ANS ancestry represented by the Upper Paleolithic Yana of northern Siberia (Cooke et al. Reference Cooke, Mattiangeli and Cassidy2021; Osada & Kawai Reference Osada and Kawai2021). One explanation may be that Tianyuan-like early ancestors of the Jōmon interacted with groups that entered Siberia through a northern migration route (Osada & Kawai Reference Osada and Kawai2021), but this possible connection will require a richer dataset than currently exists to better understand.

The current understanding is that the Jōmon represent a distinct East Asian lineage that separated from the basal East Asian lineage between 38,000 and 25,000 years ago, after the divergence of Tianyuan-related and Önge-related lineages, but prior to the separation of northern and eastern Asians and groups that would contribute to Native Americans (Kanzawa-Kiriyama et al. Reference Kanzawa-Kiriyama, Jinam and Kawai2019) (Figure 1). They appear to have remained in relative isolation, although possibly periodically interacting with neighboring mainland coastal groups. Runs of Homozygosity analysis estimates that the Jōmon were likely to have gone through a population bottleneck and have maintained a relatively small population size. Another analysis proposes a more recent population split time of between 20,000 and 15,000 years, which may have been related to the closing of the last land route to the islands (Cooke et al. Reference Cooke, Mattiangeli and Cassidy2021; Kanzawa-Kiriyama et al. Reference Kanzawa-Kiriyama, Jinam and Kawai2019), and is more in line with coalescence estimates for the deeply branching mitochondrial and Y-chromosome haplogroups found at high frequency in Jōmon populations, N1b and D-M55, respectively (Kanzawa-Kiriyama et al. Reference Kanzawa-Kiriyama, Jinam and Kawai2019).

The Jōmon contribution to indigenous ethnic populations of the Japanese islands, the Ainu of the northern islands and the Ryukyuans of the Ryukyu island chain to the south, has also been examined using both modern and ancient genetics. Anthropologists have long proposed a common origin connecting the two aboriginal groups inhabiting the extreme ends of the Japanese islands (Baelz Reference Baelz1911), which has underlain the development of the dual-layer model. This was demonstrated by genomic analysis of modern populations that found Ainu and Ryukyuans clustering tightly together, followed by the present-day Japanese population (Japanese Archipelago Human Population Genetics Consortium 2012). Although both of these groups showed more recent admixture from mainland sources, gene flow from Austronesian sources was minimal in Ryukyuans, despite an Austronesian cultural influence observed in the southern Ryukyu islands dating to the Neolithic (Sato et al. Reference Sato, Nakagome and Watanabe2014). The availability of the first ancient Jōmon genome established the genetic continuity of the Jōmon with the Ainu and Ryukyuans (Kanzawa-Kiriyama et al. Reference Kanzawa-Kiriyama, Kryukov and Jinam2017), with the Ainu exhibiting additional admixture from eastern, but not central Siberia (Adachi et al. Reference Adachi, Kakuda, Takahashi, Kanzawa-Kiriyama and Shinoda2018; Jeong et al. Reference Jeong, Nakagome and Di Rienzo2016). The common Jōmon source connecting geographically dispersed present-day indigenous groups provides genomic support to the dual-layer model. Whether the most ancient remains found in the Japanese archipelago also belong to Jōmon populations has not yet been convincingly determined. The oldest Jōmon genomic data presently dates to 9,000 years ago (Cooke et al. Reference Cooke, Mattiangeli and Cassidy2021), and earlier human remains have yet to be analyzed genomically. The mitochondrial haplogroup of the 20,000-year-old Minatogawa individual from Okinawa belonged to an extinct lineage of haplogroup M, which occupied a phylogenetic position basal to both Jōmon and mainland East Asian mitochondrial genomes, confounding a precise ancestral assignment (Mizuno et al. Reference Mizuno, Gojobori and Kumagai2021). Successful recovery of genomic data from Upper Paleolithic sites found throughout the Japanese islands will bring more clarity to early Jōmon origins and the peopling of the archipelago.

Three-Source Origin of the Modern Japanese Population

Ancient genomes from the Jōmon, Yayoi, and Kofun periods have allowed a more thorough testing of the dual-model explanation of the population structure of present-day Japan, and have resulted in the recent inclusion of an additional third layer. Genomes from the Yayoi Period attest that the expansion of large-scale agricultural practices introducing wet-rice and cereal cultivation across Japan was accompanied by a substantial amount of admixture with Northern East Asian ancestry from the mainland. This mainland ancestry was found not to derive from the YR-related farmer dispersal from the Yellow River Valley, but instead to share the most alleles with West Liao River millet farmers from the Middle Neolithic to the Bronze Age. High affinities with ancient Baikal and eastern Siberian populations could be explained by the large component of ANA ancestry in these populations, and West Liao River farmers with greater YR content shared fewer alleles with Yayoi (Cooke et al. Reference Cooke, Mattiangeli and Cassidy2021). This narrows the window of possible migration times from the West Liao River region, since YR ancestry increased in the West Liao region during the Neolithic, before decreasing in the Bronze Age (Ning et al. Reference Ning, Li and Wang2020b). The incoming agriculturalists were likely descendants of Middle Neolithic West Liao River populations containing a low YR ancestry component. The source and timing of this migration also support the agricultural-linked expansions associated with the spreading of Transeurasian languages, and it is thus likely that the Yayoi Period introduced proto-Japonic languages to the Japanese islands by 2,300 years ago (Robbeets et al. Reference Robbeets, Bouckaert and Conte2021). Although rice was not cultivated in the West Liao River, the population migrating into Japan from the southern Korean Peninsula was assumed to have belonged to the Mumun culture 无纹陶器时代, whose origins, based on shared features of ceramic styles, were linked to the Pianpu 偏堡 culture from the Liangdong region south of the Liao River (Miyamoto Reference Miyamoto2022) (Figure 6). Rice farming had previously spread from Shandong to the Liangdong Peninsula after 4,500 years ago (Whitman Reference Whitman2011), where wet-rice farming was widely adopted by the Middle Mumun Period (Crawford & Lee Reference Crawford and Lee2003). Two Yayoi-associated genomes were found to have contained roughly similar admixed amounts of Middle Neolithic West Liao River and Jōmon ancestry, although it should be noted the skeletons of the individuals analyzed showed greater morphological similarities with Jōmon than typical Yayoi-associated remains (Cooke et al. Reference Cooke, Mattiangeli and Cassidy2021). Given the present-day persistence of Jōmon-related Y-chromosome and mitochondrial haplogroups in Japan, the Yayoi Period contacts between the two populations may have been more equitable than is seen with other insular agricultural expansions, for example, in Britain, where preexisting hunter-gatherer Y-chromosome haplogroups were largely replaced (Brace et al. Reference Brace, Diekmann and Booth2019).

Figure 6

The origins of modern Japanese. Middle Neolithic movements of farmers from the West Liao River region (WLR_MN) may have brought wet-rice farming practices originating from Shandong first to the Korean Peninsula and later, during the Yayoi period, to the Japanese Archipelago. Han ancestry migrated from the mainland to Japan during the Kofun Period (~1.7 kya). Kya = kiloyears ago.

Kofun Period genomes include the ancestry present during the preceding Yayoi Period, but add to this an additional component of mainland East Asian ancestry, best modeled as Han Chinese. This newer component, which accounts for approximately 60% of Kofun genomes, likely represents new migrations accompanying the expanding economic and political coordination between China, Korea, and Japan during the Kofun Period (Cooke et al. Reference Cooke, Mattiangeli and Cassidy2021). The Kofun burials studied belonged to individuals from the same site, and were not associated with the keyhole-shaped elite tombs the period was known for. Additional Kofun genomes may better explain the status and role these newer East Asian immigrants played in the Kofun society. The genomic profile describing the Kofun-era individuals largely reflects the current genomic profile of Mainland Japan, as present-day Mainland Japanese have been modeled as having an average of 70% Han-related and 10% Jōmon-related ancestry, with the rest being derived from northeast East Asians (Cooke et al. Reference Cooke, Mattiangeli and Cassidy2021).