Xinjiang, located in the northwest region of China and known as the heart of the ancient Silk Road, has a rich history of population admixture and cultural exchanges (Mair, 2021). The constant influx of western and eastern Eurasian immigrants has contributed to the diverse genetic landscapes of this area (Mair, 2021; Kumar et al., 2022). Among these populations, the Kazakhs, cross-border nomads organized into a hierarchical patrilineal clan system, have complex ancestral sources and ethnic histories. The majority of Xinjiang Kazakhs reside in Altay Prefecture, along the southern Altai Mountains, and in Ili Prefecture along the Ili River Valley. However, there is currently limited genetic information regarding the impact of past population migration on the paternal genetic structure of present-day Xinjiang Kazakhs.

The origins of the Xinjiang Kazakhs can be traced back to the Saka Culture of the Early Iron Age (Di Cosmo, 1996; Chi and Festa, 2020; Gnecchi-Ruscone et al., 2021; Khussainova et al., 2021). During the 3rd-2nd centuries BCE, Kangju and Wusun tribal confederations emerged in this region, contributing to cultural and ethnic continuity (Di Cosmo, 1996; Otarbaeva, 1998; Bi, 2015; Kumar et al., 2022). The Turkic period brought about significant changes in regional ethnic composition, as various steppe nomadic empires like the Xiongnu, Xianbei, Turk, Rouran, Khitan, and others expanded their dominance (Otarbaeva, 1998; Rogers, 2012). The Mongol invasion in the 13th century had a profound influence on the genetic makeup of Xinjiang (Otarbaeva, 1998). In the mid-fifteenth century, the Kazakh Khanate was established as a tribal and political entity, emerging as a new regional power in Xinjiang (Otarbaeva, 1998; Sun, 2022). However, the Kazakhs faced attacks from the Dzungars in the early 18th century, resulting in the loss of their pasturelands (Otarbaeva, 1998).

Genetic investigations have offered new insights into the complex history of Kazakhs, helping us understand how admixture events have shaped the population structure and genetic diversity of modern Kazakhs. Notably, the higher frequencies of C2 haplogroups along the southern and western borders of Kazakhstan align with the historical path of the Mongolian invasion (Khussainova et al., 2021). Moreover, the discovery of the haplogroup C2a1a2-M48 among Western Kazakhs has suggested a founding population from the Middle Ages (Zhabagin et al., 2021). Paternal lineage analysis further supported a common origin of South Uissun clans from Niru’un Mongols, rather than early Wusun peoples (Zhabagin et al., 2020). In China, the paternal origin of Kazakhs residing in Gansu province has been confirmed as Kerey-Abakh (Wen et al., 2020). Similar mtDNA lineage has been observed among Kazakhs from Altai, East Kazakhstan, and Xinjiang (Tarlykov et al., 2013). Moreover, comprehensive genome-wide analyses of Xinjiang Kazakhs have indicated two significant population admixture events: one occurring approximately 120 generations ago and another 20-40 generations ago (Pan et al., 2023).

Our knowledge of the paternal history of the Xinjiang Kazakhs, however, remains limited. Previous studies, using only Y-chromosomal STR data (Shan et al., 2014a; Shan et al., 2014b; Mei et al., 2016; Li et al., 2020; Zhang et al., 2021), have primarily focused on genetic polymorphisms, population comparisons, and forensic applications. Xinjiang Kazakhs have been demonstrated to be genetically close to Eastern Kazakhstan groups but distinct from Northern Kazakhstan groups based on Y-STR loci polymorphism (Ashirbekov et al., 2022; Ashirbekov et al., 2023). In this study, we aimed to gain a deeper understanding of the paternal genetic structure of modern Kazakhs in northwest China by analyzing 110 Y-SNPs and 24 Y-STRs from 410 male Kazakhs in the Altay and Ili Prefecture of Xinjiang. By doing so, our hope was to shed light on how historical events influenced the genetic makeup of this population.

This study included a total of 410 Kazakh males from the Xinjiang province in China, as previously reported (Mei et al., 2016; Li et al., 2020). Among the participants, 209 were from Altay and 201 were from Ili. Informed consent forms were signed by each participant, and the study was approved by the ethics committee at the School of Life Sciences in Fudan University.

DNA samples were extracted using the QIAamp DNA Blood Mini Kit (QIAGEN, Germany). Hierarchical genotyping was performed using the SNaPshot method with the ABI SNaPshot Multiplex Kit, as previously described (Wang et al., 2014; Xu and Li, 2017). Haplogroups were classified according to the ISOGG Y-DNA Haplogroup Tree 2019 (Version: 15.73; Date: 11 July 2020). Y-STR loci data from the 24 analyzed participants had been previously reported (Mei et al., 2016; Li et al., 2020). Haplogroup diversity were calculated using the Arlequin package.

To investigate the genetic relationships between Altay and Ili Kazakhs and other published Eurasian populations, we conducted a Y-SNP-based principal component analysis (PCA) using RStudio software (v. 2021.09.1 + 372). Additionally, we constructed median-joining networks of Y-STR haplotypes C2a1a3-F1918 and C2a1a2-M48 to examine individual-level relationships using the program NETWORK (v. 10.2) and 15 Y-STR data (DYS19, DYS389I, DYS389II, DYS390, DYS391, DYS392, DYS393, DYS437, DYS438, DYS439, DYS448, DYS456, DYS458, DYS635, GATA_H4). We calculated the time to the most recent common ancestor (TMRCA) for each cluster using the average squared distance (ASD) method, with a generation time set to 25 years, as previously described (Wang et al., 2014; Xu and Li, 2017; Zhabagin et al., 2017).

We utilized bcftools mpileup and bcftools call (Li, 2011; Poznik et al., 2013) to analyze the Y-chromosomal SNPs from published ancient DNA data. The haplogroup identification was performed using Yleaf.py script with Yleaf (Ralf et al., 2018). Furthermore, we conducted a thorough verification of the SNPs by visually inspecting them using the IGV software (Helga et al., 2013).

A total of 34 different haplogroups were identified in Altay Kazakhs, and 27 different haplogroups in Ili Kazakhs ( Table S1 ), based on analysis of 110 Y-SNP markers. Altay Kazakhs exhibited the highest haplogroup diversity (HD) among all Kazakh populations in Kazakhstan, Russia, and China, with a value of 0.94 ( Figure 1 ). This finding aligns with a previous study that compared Y-STR diversity among Kazakhs from Xinjiang, Kazakhstan, and Russia (Shan et al., 2014b).

The widely distributed haplogroup C2-M217 is prevalent among all Kazakh populations, including Chinese Kazakh populations. In Xinjiang Province, it represents 64.1% among Altay Kazakhs and 55.2% among Ili Kazakhs ( Figure 1 ). In Gansu Province, it accounts for 85% among the Aksai Kazakhs (previously reported) (Wen et al., 2020). Within this haplogroup, there are two major lineages: C2a1a3-F1918, which makes up 49.8% in Altay Kazakhs and 50.7% in Ili Kazakhs, and C2a1a2-M48, which accounts for 9.6% in Altay Kazakhs and 2.5% in Ili Kazakhs ( Table S1 ). Similarly, Gansu Aksai Kazakhs have prominent lineages of C2a1a3a-F3796 (79.6%) and C2a1a2-M48 (0.3%) (Wen et al., 2020).

Haplogroup O2-M122, the most common haplogroup in China, is found at high frequencies in several Kazakh populations. In Ili Kazakhs, it is observed at a frequency of 29.4%, while in Southeast Altai Kazakhs of Russia and East Kazakhstan Kazakhs, it is observed at frequencies of 33.7% and 23.9% respectively. Additionally, Altay Kazakhs and South Kazakhstan Kazakhs have moderate frequencies of 5.7% and 9.4% respectively ( Figure 1 ). The most significant subclade of Haplogroup O2-M122 is O2a2b1a2a-F444, which is present in 25.9% of Ili Kazakhs and 4.8% of Altay Kazakhs ( Table S1 ).

Haplogroup G-M201, primarily found in the Caucasus region (Rootsi et al., 2012), is completely absent in three Chinese Kazakhs. However, it is observed with high frequencies in Central, Northern, and Eastern Kazakhstan Kazakhs ( Figure 1 ).

The PCA was performed by projecting two Xinjiang Kazakh populations onto 156 published modern Eurasian populations ( Figure 2 , Table S2 ). The analysis revealed a comprehensive population structure across Eurasia, consisting of nine distinct genetic clusters: Indo-European, Inner Mongolia, Mongolia, Northern Han, Southern Han, Tibeto-Burman, Tungusic, Turkic, and Uralic populations. These clusters aligned well with ethnic and linguistic classifications. Notably, our study found that the Kazakh populations from Altay and Ili Prefecture in Xinjiang province occupied an intermediate position between the Turkic, Mongolia, and Tibeto-Burman clusters, indicating its close genetic affinities with these geographically neighboring populations. Specifically, the Altay Kazakhs shared genetic affinities with the Mongols from Northwest Mongolia, while the Ili Kazakhs showed closer genetic similarities to the Chamdo Tibetans on the eastern Tibetan Plateau.

The network of haplogroup C2a1a3-F1918 has been constructed from 15 Y-STR profiles of 614 individuals from 7 populations: Alaty Kazakhs (n=71), Ili Kazakhs (n=102) and Aksay Kazakhs (n=74) from China, Kazakhs from southern Kazakhstan (n=102), Mongolians from Mongolia (n=19) and Inner Mongolia, China (n=89), and Kyrgyz from Xinjiang, China (n=157) ( Figure 3 ; Table S3 ).

Among them, nearly half of Ili Kazaks (n=52) shared the haplotypes with South Kazakhstan Kazakhs (n=21), Xinjiang Kyrgyz (n=10), Altay Kazakhs (n=4) and Inner Mongolians (n=3), suggesting a potential patrilineal relationship ( Table S3 ). Ili Kazakhs share haplotypes mostly with Kazakhs from southern Kazakhstan and partly with Altay and Aksay Kazakhs from China. This implies a significant influence from southern Kazakhstan on the genetic makeup of Ili Kazakhs.

In the network, a left-side cluster was determined by allele DYS448-23 ( Figure 3 ). This cluster is mainly composed of Altay Kazakhs (34.4%) and Aksay Kazakhs (56.8%) from Northwest China ( Table S3 ). Previous reports have shown that the DYS448-23 subcluster of Kazakhs corresponds to Kereys-Abakh (Abilev et al., 2012). This specific STR haplotype accounts for 20.6% of the Altay Kazakhs’ paternal pool and as much as 76.3% of the Aksay Kazakh pool. These individuals all belong to lineage C2a1a3-F1918* (58.4%) and its downstream lineage C2a1a3a-F3796 (41.6%). A distinct central haplotype suggests that C2a1a3-F1918 may have experienced a series of founder effects or severe bottlenecks in Altay and Aksay. This also reflects the close paternal genetic relationship between these two Kazakh C2a1a3 lineages. TMRCA estimates indicate that this special haplotype cluster is approximately 289.4 ± 202.65 years old, which coincides with historical migration events of Kazakhs after the fall of the Dzungar Kingdom. The Dzungar Kingdom was defeated by the Qing dynasty during the mid-18th century.

The network of haplogroup C2a1a2-M48 reveals distinct genetic patterns among different populations ( Figure 4 , Table S4 ). Specifically, the left star cluster is unique to the south and west Kazakhstan Kazakhs, clearly distinguishing them from the Mongolic-speaking populations on the right. Interestingly, the distribution of the haplogroup C2a1a2-M48 among Ili and Altay Kazakhs is more scattered and does not form a distinct cluster. However, some Altay Kazakhs are centrally located within the network, acting as a bridge between Mongolic-speaking populations and Kazakhstan Kazakhs. This suggests that the Xinjiang Kazakh C2a1a2-M48 lineage has multiple paternal ancestral sources.

The ethnogenesis of the Kazakhs can be categorized into three distinct phases: the pre-Turkic, Turkic, and Turko-Mongolian periods. In this study, we utilized 110 Y-SNPs and 15 Y-STRs to investigate the genetic structures and individual relationships within the paternal gene pool of two Kazakh populations in Xinjiang. It is important to highlight that the majority of Y chromosome haplogroups observed in Altay Kazakhs were C2a1a3-F1918 and C2a1a2-M48, while Ili Kazakhs exhibited C2a1a3-F1918 and O2a2b1a2a-F444 as the predominant haplogroups.

The published ancient DNA data revealed the presence of haplogroup C2a1a3-F1918 in East Asia during the middle Neolithic. This haplogroup was primarily found in the Songnen Plain of northeastern China (n=1) (Mao et al., 2021) and the Boisman Culture of the Russian Far East (n=3) (Wang et al., 2021). During the Iron Age, a related lineage was identified in the Xianbei population of China (n=1) (Ning et al., 2020). Subsequently, this lineage was discovered in the Site of Kayalyk (n=1) (Gnecchi-Ruscone et al., 2021) and the Golden Horde (n=1) (Damgaard et al., 2018) in Kazakhstan, as well as in the Mongol empire (n=5) (Jeong et al., 2020) during the Medieval Age. The age estimation of modern east Eurasia Y-chromosome STR data belonging to the C2a1a3-F1918 was determined to be 6327 ± 1736 years (Wei et al., 2018). Based on these findings, we propose that haplogroup C2a1a3-F1918 originated in ancient northeast Asia and expanded to the Mongolia Plateau, later spreading westward to Central Asia and Europe, following the route of its subbranch C2a1a3a-F3796 (Wei et al., 2018). Hence, we speculate that haplogroup C2a1a3-F1918 among Xinjiang Kazakhs may have originated in ancient northeast Asia.

Haplogroup C2a1a2-M48 is widespread in central and northern Asia, with particularly high frequencies in Tungusic and Paleosiberian speaking populations (Zerjal et al., 2002), such as Manchu groups (Xue et al., 2005). It is believed to have originated around 150,000 years ago (Karmin et al., 2015). Ancient DNA data supported the presence of haplogroup C2a1a2-M48 in the Songnen Plain of northeastern China at the end of the last glacial period (n=1) and during the early Neolithic Age (n=1) (Mao et al., 2021). It has also been found in the Boisman Culture of the Russian Far East during the middle Neolithic Age (n=1) (Wang et al., 2021). From the Iron Age onwards, haplogroup C2a1a2-M48 has mainly been identified in Central Asia, including the Xiongnu empire (n=1) and the Mongol empire (n=1) in Mongolia (Jeong et al., 2020), as well as among central steppe nomads (n=1) in the Tian Shan area of Kyrgyzstan (Damgaard et al., 2018). The westernmost ancient individual with haplogroup C2a1a2-M48 was a Hungarian from the Carpathian Basin, dated to the end of the ninth century AD (Fothi et al., 2020). In summary, the spatial and temporal distribution of haplogroup C2a1a2-M48 closely aligned with that of haplogroup C2a1a3-F1918.

Our network analysis and TMRCA estimate of the special branch DYS448-23, a Kerey-Abakh-specific cluster as previously identified (Abilev et al., 2012), indicates a rapid and recent expansion of the Kerey clan in Altay, located at the northern end of Xinjiang province. The Kazakhs had long been under pressure from the Dzungars. In 1757, Ablai Sultan, a Kazakh khan from the Middle zhuz of the Kazakh Khanate, chose to reconcile and submit to the Qing dynasty. As a result, the Kazakhs were able to migrate back to northern Xinjiang and reclaim their pasturelands in the Altay Mountains, which had been taken by the Dzungars (Perdue, 2005).

Our study has revealed the remarkable genetic heterogeneity of Altay Kazakhs, confirming their diverse paternal origins and the population’s genetic admixture in this region. While they possess ancient northeast Asian lineages, our findings also suggested that the present-day Altay Kazakhs and Ili Kazakhs in Xinjiang are primarily descendants of recent waves of migration to the Xinjiang region. Specifically, the Altay Kazakhs are believed to have originated from the Kerey-Abakh subclan, while the Ili Kazakhs trace their ancestry back to the Kerey-Ashmaily subclan.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

The studies involving humans were approved by the ethics committee at the School of Life Sciences in Fudan University. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

BW: Conceptualization, Data curation, Formal Analysis, Software, Visualization, Writing – original draft. JL: Investigation, Visualization, Data curation, Validation, Writing – original draft. EA: Investigation, Writing – review & editing, Project administration. XC: Formal Analysis, Data curation, Writing – review & editing. ZJ: Data curation, Software, Formal Analysis, Writing – original draft. YY: Data curation, Formal Analysis, Software, Visualization, Writing – review & editing. MS: Conceptualization, Project administration, Resources, Validation, Writing – review & editing. SW: Conceptualization, Supervision, Writing – review & editing, Project administration, Validation, Writing – original draft.