2014 年 122 巻 3 号 p. 137-148
This study quantifies the population history of Iron Age nomads of southern Siberia by analyzing craniofacial diversity among contemporaneous Bronze and Iron Age (7th–2nd centuries BC) groups and compares them to a larger geographic sample of modern Siberian and Central Asian populations. In our analyses, we focus on peoples of the Tagar and Pazyryk cultures, and Iron Age peoples of the Tuva region. Twenty-six cranial landmarks of the vault and facial skeleton were analyzed on a total of 461 ancient and modern individuals using geometric morphometric techniques. Male and female crania were separated to assess potential sex-biased migration patterns. We explore southern Siberian population history by including Turkic-speaking peoples, a Xiongnu Iron Age sample from Mongolia, and a Bronze Age sample from Xinjiang. Results show that male Pazyryk cluster closer to Iron Age Tuvans, while Pazyryk females are more isolated. Conversely, Tagar males seem more isolated, while Tagar females cluster amongst an Early Iron Age southern Siberian sample. When additional modern Siberian samples are included, Tagar and Pazyryk males cluster more closely with each other than females, suggesting possible sex-biased migration amongst different Siberian groups. This is evident in modern female Tuva, who cluster with modern female Kalmyk, while modern Tuvan males do not. Male and female Iron Age Tuvans are not closely related to modern Tuvan peoples living in the region today, resulting from the influx of the Xiongnu beginning in the Late Iron Age. Both male and female Pazyryk and Tagar crania appear more similar to Central Asia groups, especially the Kazakh and Uzbek samples. However, there is evidence that Tagar females have a common origin with the Yakut, a modern nomadic population that resides in northeastern Siberia. These results would suggest variable genetic contributions for both sexes from Central and East Asia.
The cultures of southern Siberia, and specifically the people of the Altai during the Bronze and Iron Ages, have previously been studied by physical anthropologists using traditional craniometric data (Chikisheva, 2000a, b, 2008), however, they have not been examined using geometric morphometrics (GMM). This study investigates two of the better-known cultures of the Altai during this time period, the Tagar and Pazyryk cultures (Figure 1). Although each cultural tradition has been examined archaeologically, paleogenetically, and anthropologically, their relationship to one another has yet to be fully explored. Here, we present results from a craniofacial study that examines population history and structure by using exploratory multivariate analyses, including canonical variate analysis (CVA) and Mahalanobis distances between ancient and modern groups from southern, western, and eastern Siberia, as well as comparative ancient and modern samples from Central Asia, Mongolia, and western China.
Map showing the extent of of the cultural influence of Pazyryk, Tagar and some adjacent groups in southern Siberia in the 6th–3rd centuries BC.
The Tagar people are associated with the Late Bronze Age/Early Iron Age archaeological culture of the same name, which flourished in the Minusinsk Basin along the Upper Yenisei River from the 7th–2nd centuries BC (Chikisheva, 2000a). The Minusinsk Basin is located in Khakassia, which is bordered by the Republic of Altai and the Republic of Tuva in the Russian Federation. The Tagar people were thought to have been semi-nomadic pastoralists who raised livestock, such as horses, goats, and sheep. However, based on recent archaeological evidence discovered in places such as Mongolia and Kazakhstan, the development of nomadic pastoralism on the eastern steppe was highly complex and perhaps included agriculture, in addition to the pastoral economy that epitomizes peoples of the Altai (Di Cosmo, 1994; Spengler et al., 2014). The production of cereal has been speculated among Tagar peoples due to a large number of bronze sickles found in associated archaeological sites (Murphy et al., 2013; Svyatko et al., 2013). The Tagar people also had one of the largest bronze-smelting centers in ancient Eurasia.
The Pazyryk culture is contemporaneous with the Tagar culture, which flourished in the Altai Mountains of southern Siberia and eastern Kazakhstan from the 6th–3rd centuries BC (Rudenko, 1970). More recently, Pazyryk burial sites have been found in the Mongolian Altai (Jordana et al., 2009). The Pazyryk people are well known due to the exceptionally well-preserved mummified remains found in stone tumuli of the Ukok Plateau. These burial mounds were intended for high-ranking members of society, such as chiefs, priests, and elders, and included concubines, horse remains, and precious artifacts. Archaeologists know that extensive trading networks existed among these nomads as textiles, mirrors, and silk from places such as China, and even Iraq, have been found among the burials (Chikisheva, 2000a, b). Some of the more famous remains include a man and woman with extensive tattooing on their bodies, which depicted elaborate animal motifs. The woman, known as the “Siberian Ice Maiden,” is considered a priestess by archaeologists because of the items with which she was buried.
Some Iron Age nomadic peoples of the Altai, especially Tagar people, have been described as being part of the Scythian tradition, though Scythian is a general moniker applied to all those tribes who shared economic, cultural, and perhaps linguistic traditions. These mobile equestrian tribes migrated east from the European steppe beginning in the 7th century (Ricaut et al., 2004a). Other known tribes that belonged to the Scythian tradition include Sauromatians of the Lower Volga and southern Urals and the Saka of the Kazakhstanian steppes and the valleys of Tian Shan and Pamir mountain ranges of Central Asia. The Tagar people produced artwork of animal motifs that were similar in nature to works produced by the Scythian tribes. The Scythians are well known in historical texts and are one of a large number of tribes to have emerged from the Pontic-Caspian steppe and migrated eastward to the Altai from around 2000 BC. The end of the Scythian period was probably the result of the westward expansion of Turko-Mongolic nomadic peoples from East Asia known as the Xiongnu (Lalueza-Fox et al., 2004). AlThough the Scythians may have played a role in Tagar development, there is abundant evidence to suggest that peoples of the Tuva region and peoples of the Pazyryk culture formed via a largely autochthonous component (Chikisheva, 2008).
Therefore, the Altai region has been one of contact, conflict, and trade since early times. The genetic diversity of the region suggests extensive contact over the course of the last several millennia (Gonzalez-Ruiz et al., 2012). Recent ancient DNA studies indicate the Scythian peoples to have a mix of both Western and Eastern Eurasian DNA types (Voevoda et al., 2000; Clisson et al., 2002; Ricaut et al., 2004b; Chikisheva et al., 2007; Keyser et al., 2009; Pilipenko et al., 2010). Nevertheless, the history and origins of these people remain relatively obscure. Physical anthropologists have suggested the appearance of these peoples, based on craniofacial traits, to be unlike others in the Altai region during the Bronze and Iron Ages (Rudenko, 1970; Kozintsev et al., 1999; Moiseyev, 2006). Most of the populations residing in the Altai during that time possess a more European-like appearance relative to populations now residing in the region of southern Siberia (Kozintsev, 2009). However, the Tagar and Pazyryk, when compared to other Bronze Age and Early Iron Age samples from the Altai, evidence greater admixture from Eastern Eurasian populations (Chikisheva, 2000b). Therefore, morphological features more representative of Western Eurasian populations may have appeared as early as the Neolithic and the Bronze Age with more recent admixture occurring during the Late Iron Age in southern Siberia (Chikisheva, 2000b). These data have raised questions about the origins and migration routes possibly taken by the Altai Iron Age nomads. In this study, we use newly acquired craniofacial data analyzed in a GMM framework in an attempt to tease out the possible origins and interactions of these Early Iron Age nomads.
The analysis of ancient populations through the investigation of craniofacial variation is an effective and informative way to understand modern population structure and infer relationships in the past (Howells, 1973, 1989; Hanihara, 1996). Although morphological variation is subject to developmental and environmental processes that might confound population history, studies have suggested that the majority of cranial traits for populations not residing in extreme climates (such as the Arctic) fall under a model of neutral evolution and are not greatly influenced by selection (Relethford, 2004, 2010; Weaver et al., 2008; von Cramon-Taubadel, 2009a; Betti et al., 2010). Therefore, phenotypic variability can be used as a proxy to study biological distances between groups.
In this study, we use craniofacial diversity as a proxy for genetic distance between populations. We quantify the distance between the ancient male and female Pazyryk and Tagar peoples using methods developed in the GMM analysis. Although recent studies have used ancient DNA to study this region’s population history, sample sizes are often small and usually only contain information about the maternal ancestry (mtDNA) of the individuals. In addition, sampling for ancient DNA is destructive. GMM methods allow researchers to study population history in a non-destructive context. Therefore, sample size is increased and a broader geographic area can be included for comparison. GMM can also be used to corroborate existing ancient DNA studies.
All analyses in this study include both Tagar and Pazyryk samples to understand how they are related to each other. That is, we do not analyze each group separately against our comparative dataset. If their close relationship persists when compared to geographically and temporally diverse samples, then we can infer a close population history for these two groups. If either group appears more similar to other samples used in the analysis, then we can infer something about between group population structure and possibly external and independent sources of gene flow. We suggest the Iron Age Tagar and Pazyryk peoples will be more closely related to each other than they are to other peoples in the region based on craniofacial variation. As phenotypic variation is akin to genetic distance, we should expect to see strong similarity between these two ancient Siberian groups.
Our results show differential patterns of sex-biased migration amongst Iron Age nomadic peoples of southern Siberia. We show that different patterns emerge between male and female Pazyryk peoples. Male Pazyryk appear more closely related to Iron Age Tuvans and female Pazyryk are more isolated and diverse. Both male and female Pazyryk show similarity to the Kazakh sample. Iron Age peoples of the Tuva region are not closely related to modern Tuvans as a result of large migrations of Iron Age Mongolians known as Xiongnu. We show that Tagar males are more isolated from other ancient southern Siberian groups and may represent more recent migrants during the Iron Age, perhaps influenced by groups residing in present day Central Asia, such as Kazakhstan or Uzbekistan. Tagar females seem to share a population history with the Yakut of northeastern Siberia, who speak a branch of the Turkic language.
Phenotypic variability of crania was assessed using a suite of 26 homologous landmarks that are common amongst craniometric studies and are used to assess overall shape and size differences among individuals (Table 1). This suite of landmarks includes traits associated with the upper face, orbital areas, cranial vault, and occipital region. Cranial coordinate data were collected using a Microscribe portable digitizer by the first author. Male and female adult crania from 201 ancient crania and 260 modern crania are included in the analyses. All analyses separate sex to assess patterns of migration and differential classification. Ancient crania sample names, male and female sample size, site information, and periods are displayed in Table 2. Examples of Tagar, Iron Age Tuva and Pazyryk male cranial morphology are shown in Figure 2. All comparative cranial samples are shown in Table 3. Tagar and Pazyryk samples contain individuals from several different archaeological sites. In order to increase sample size, we pooled Tagar sites and Pazyryk sites independently. All Tagar and Pazyryk samples date to the Late Bronze Age or Early Iron Age (based on artifacts found with the skeletal remains or carbon-14 dating of objects found in the graves), and derive from locations in southern Siberia (Alekseev et al., 2002; van Geel et al., 2004). Pazyryk remains were not sampled from other locations that have associated remains, such as the Mongolian Altai (Jordana et al., 2009).
JIDA, Research Center for Chinese Frontier Archaeology, Jilin University, Changchun, China;
NUM, National University Mongolia, Ulaanbaatar, Mongolia;
RASN, Institute of Archaeology and Ethnography, Russian Academy of Sciences, Siberian Branch, Novosibirsk, Russian Federation.
Photographs of representative skulls for Pazyryk (A), Tagar (B) and Iron Age Tuva (C) peoples. All skulls are male. Pazyryk and Iron Age Tuva skull photos taken from Chikisheva (2008) and reprinted with permission from Elsevier.
LMSU, Lomonosov Moscow State University, Moscow, Russian Federation;
MUSE, Musée d l’Homme, Paris, France.
In order to assess population history and structure, we include both modern crania from Siberia and Central Asia (18th–20th centuries) and pooled samples of ancient crania from Siberia, Mongolia, and China. We assume the provenance of our modern samples to derive from the last few centuries. Most of our ‘modern’ samples have documented provenience. However, several samples from the Musée de l’Homme in Paris contain little or no provenance. The only information available to us was the geographic location from which the skulls were collected.
For instance, the linguist and ethnographer N.N. Pantusov, who excavated Kurgan remains dating from the Bronze Age to modern times in Kazakhstan, also excavated our sample labeled ‘Kazakhs’ during the late 1800s and early 1900s, now curated at the Museé de l’Homme. These Kazakh crania were collected from a region historically known as Semirechye, now known as Zhetysu in Almaty Province. However, records indicate that the Semirechye Oblast, a former administrative locale of the Russian Empire, also included lands now part of northern Kyrgyzstan and adjacent provinces of Kazakhstan. Modern Kazakh peoples share similar biological features with modern ethnic Kyrgyz. In our analysis, the Kazakh and Kyrgyz crania appear very different. Therefore, we have, as a caveat, tentatively labeled the Kazakh sample ‘modern,’ though they could most certainly date to the medieval period (1100–1500 AD) or perhaps even earlier.
We also use one sample, the Kalmyks, as representing the northern Caucasus and Lower Volga region. The reason we include Kalmyks in the analysis is their recent western Mongolian origin: they began migrating through the steppes of present days Kazakhstan and Central Asia at the end of 16th century AD and settled in the northern Caucasus steppes and Volga basin in the 17th century AD (Nasidze et al., 2005).
All of our ‘ancient’ samples have sufficient provenance (Keyser-Tracqui et al., 2003; Tumen, 2006; Chikisheva, 2008; Zubova, 2008, 2013). Some of the samples used in the present study have been used in previous studies of ancient population history and structure (Pozdnyakov, 2004; Moiseyev, 2006; Tumen, 2006; Chikisheva, 2008); however, our methodology (GMM coupled with multivariate analyses) has not been used previously. In addition, the sample comparisons we make have not been previously analyzed, with the exception of Chikisheva (2008), who included a more diverse dataset with different aims (specifically uncovering the population history of the Tuvan peoples), and Moiseyev (2006) who analyzed non-metric traits of the skull.
The medieval Siberian sample (1100–1500 AD) contains pooled remains from several sites located in western and southern Siberia (Pozdnyakov, 2004). The other Early Iron Age Siberian sample (5th–3rd centuries BC) comes from one site (Bystrovka-2) in southern Siberia (Moiseyev, 2006). The Iron Age Tuva sample (7th–3rd centuries BC) is pooled from two sites (Arzhan-2 and Dogehe-Baary II) (Chikisheva, 2008). In addition, we include a Turkic-Mongolian sample (Xiongnu, pooled series from several sites in Central and Western Mongolia) that dates to the Iron Age (3rd century BC–2nd century AD) (Keyser-Tracqui et al., 2003; Schmidt, 2012). Lastly, we include a Bronze Age sample (Tianshanbeilu, Hami City) from the eastern Xinjiang Province (4th– 3rd centuries BC) (Schmidt, 2012). We include this ancient sample as previous studies indicated some biological affinity among ancient Siberians and Bronze Age Xinjiang peoples of the Tian Shan mountain range. This sample also included distinct pottery closely linked to types seen in southern Siberia and Western Mongolia (Liu and Cheng, 2012).
The raw data were analyzed using the software program MorphoJ (Klingenberg, 2011). Landmark configurations were processed by means of GMM (Klingenberg, 2010). Original configurations were superimposed according to the generalized Procrustes analysis (GPA) procedure in MorphoJ using the total covariance matrix. Sample variance was assessed using canonical variates analysis (CVA) and Mahalanobis distances generated from the total covariance matrix. In this study we do not attempt to interpret morphological shape changes between samples, rather our goal is overall classification.
Three separate CVA analyses (CVA1, CVA2, CVA3) were performed to assess ancient Siberian classification (for samples included in each analysis, see Table 1 and Table 2). In CVA1, we chose only representative samples from southern and western Siberia that date from the Iron Age to the medieval period. This was done in order to assess male and female distances among ancient Siberian groups. Differential patterns of sex-biased distances may indicate whether the local autochthonous component in southern Siberia was due to female or male groups. In CVA2, we chose both ancient and modern Siberian groups. This analysis was performed in order to assess the apparent differences of the Iron Age groups to modern northeastern Siberian groups. That is, we included these groups intentionally to show the significant differences among southern and eastern Siberian groups. Based on previous research, there have been suggestions of the influence of a Turkic-speaking people in southern Siberia. Therefore, in CVA3, we included both male and female Pazyryk and Tagar samples and compared them with modern and ancient Turkic groups located in southern Siberia, Central Asia, and Western China.
Male and female Tagar and Pazyryk craniofacial diversity was initially examined against several other contemporaneous samples from the Iron Age in southern Siberia (CVA1), in addition to a pooled sample from the Siberian medieval period (~1100–1500 AD). The CVA results show differential patterns for Iron Age males and females. The plot of the first two canonical variates accounts for 86.4% of the total variance for males (Figure 3A) and 71.7% for females. It is clear from this plot (Figure 3A) that along CV1 Pazyryk males cluster closer to Iron Age Tuvans, while the Tagar sample and Iron Age males from the Bystrovka-2 site are more isolated. On the other hand, female Pazyryk and Iron Age Tuvans are separate along CV1, while the female Tagar are closely related to the Bystrovka-2 Iron Age females (Figure 3B). In both males and females, the pooled medieval series is separated along CV2.
CVA plot for ancient male and female Southern/Western Siberians (CVA1). Closed circles are individual crania. (A) Male crania (n = 77; CV1 = 71.8%, CV2 = 14.6%). (B) Female crania (n = 75; CV1 = 47.9%, CV2 = 23.8%).
The Tagar and Pazyryk males and females were then compared to modern Siberian crania (CVA2), including groups from northeastern Siberia (Evenks, Orochi, Ulchi) and the Caspian Sea region (Kalmyks); and ancient Siberian crania, mainly from southern and western Siberia (pooled medieval sample, Iron Age Bystrovka-2, pooled Iron Age Tuva). The plot of the first two canonical variates accounts for 56.2% of the total variance for males, and 53.6% for females (Figure 4). The ellipses drawn around the population mean are 75% frequency ellipses. The CVA plots for both males (Figure 4A) and females (Figure 4B) indicate a general separation of eastern Siberian groups and the Kalmyk sample compared to southern Siberian and ancient Siberian groups. However, female crania appear more diverse. For example, although there is still a general separation among ancient and modern groups along CV1, the modern Tuvan females cluster tightly with the Kalmyk females. This is not the case for modern male Tuvans, who appear in an intermediate position; perhaps reflecting a mixed ancestry related to greater gene flow from ancient groups compared with modern Tuvan females. In this analysis, the Tagar and Pazyryk male samples are more closely related due to the introduction of more highly differentiated groups. However, the females for both groups do not show a similar pattern. Although female Tagar and Pazyryk are fairly close in Figure 4B, they do not overlap. The female Iron Age Tuvans are now more closely related to both female Tagar and female Pazyryk. Also, the Tagar males have become less isolated in this analysis.
CVA plot of the first two canonical variates of ancient and modern Siberian samples (CVA2). (A) Male crania (n = 142; CV1 = 40.2%, CV2 = 16.0%). (B) Female crania (n = 152; CV1 = 42.5%, CV2 = 11.1%). Closed circles are individual crania, ellipses are drawn as equal frequency ellipses (75%) around group mean.
It has been suggested that Central Asian Turkic (or Turkish speaking) groups have played a role in shaping the genetic diversity seen today in southern Siberia. Therefore, the Tagar and Pazyryk were analyzed against samples collected from Central Asia (Kazakh, Kyrgyz, Uighur, Uzbek, and Turkmen), Siberia (modern Tuvans, a Turkic ethnic group), a Mongol-Turkic Iron Age group from Mongolia (Xiongnu), and a Bronze Age sample from Xinjiang (Xinjiang Bronze) (CVA3). For male crania, the first two canonical variates account for 48.8% of the variance (Figure 5A), while for females, the first two canonical variates account for 53.8% of the variance (Figure 5B). Similar to Figure 4, the ellipses are 75% frequency ellipses. In Figure 5A, the Xinjiang males are clear outliers along CV1. The Tagar males are also slight outliers, clustering loosely with the Pazyryk, modern Uzbeks, and the Kazakhstan sample. Interestingly, the female CV plot shows a tight cluster of female Pazyryk, Tagar, and Yakut to the exclusion of the other samples, which is not shown among male crania (Figure 5B). Mahalanobis distances among modern Siberian and Central Asian groups indicate more similarity of the Tagar and Pazyryk peoples to some Central Asian samples as opposed to modern Siberian samples (Table 4, Table 5, Table 6, Table 7). Although the result of NE Siberian groups separating from southern Siberian groups is not surprising given the history of the region, the relative closeness of Kazakh male and female crania (that could date to as early as the medieval period) and male Xiongnu sample to Pazyryk and Tagar could indicate ancient admixture or gene flow among these groups.
CVA plot of the first two canonical variates for Pazyryk, Tagar and modern ethnic Turkic or ancient Turkic-speaking groups (CVA3). (A) Male crania (n = 140; CV1 = 31.9%, CV2 = 16.9%). (B) Female crania (n = 125; CV1 = 36.9%, CV2 = 16.9%). Closed circles are individual crania, ellipses are drawn as equal frequency ellipses (75%) around group mean.
Southern Siberian groups, both modern and ancient, have been investigated extensively, owing to a rich and diverse cultural hybrid zone that has seen significant contact among peoples originating from both Western and Eastern Eurasia since as early as the Upper Paleolithic. Importantly, this region has been hypothesized to be the origin of modern-day Native American peoples, while also maintaining high levels of genetic and cultural diversity (Quintana-Murci et al., 2004; Dulik et al., 2012; Raghavan et al., 2013). This study has attempted to reconcile the population history of southern Siberia by examining a small temporal slice during the Iron Age. Notably, two important cultural groups with questionable origins were investigated craniometrically to elucidate questions owing to divergent morphological appearances and mtDNA haplotype composition (Voevoda et al., 1998; Chikisheva et al., 2007). These cultural groups, the Tagar and Pazyryk, are important to understanding the broader history of the region. Our results have shown these two groups to be outliers when compared to many modern living peoples of Siberia and Central Asia (Figure 5), while maintaining a connection to some peoples from the Iron Ages of southern and western Siberia (Figure 3, Figure 4, Figure 5). These findings are similar to previous research done in ancient DNA studies and physical anthropological studies (Chikisheva, 2000a; Moiseyev, 2006; Chikisheva et al., 2007; Pilipenko et al., 2010).
It is well known that the modern genetic diversity seen in the region of southern Siberia today stems from extensive contacts among diverse peoples (Comas et al., 1998, 2004). Mitochondrial DNA and Y chromosome studies have been conducted on a range of living peoples in Northern Asia, including Tuvinans, Buryats, Khakassians, Sojots, Todjins, Tofalars, Kalmyks, Kazakhs, Kizhi, Mongols, Evenks, and Yakuts, among others (Kolman et al., 1996; Wells et al., 2001; Zerjal et al., 2002; Derenko et al., 2003, 2006, 2007; Pakendorf et al., 2003, 2006; Nasidze et al., 2005; Starikovskaya et al., 2005; Phillips-Krawczak et al., 2006; Gokcumen et al., 2008). These groups range in linguistic diversity as well, covering Turkic, Mongolic, and Tungusic language families. The majority of these studies include analyses of haplogroup reconstruction for the uniparentally inherited mtDNA and Y chromosome. The general consensus among these studies is that the mountain belt zone of southern Siberia is where populations began to expand into Eastern and Northern Europe following the Last Glacial Maximum. From the Mesolithic, the region of southern Siberia has witnessed extensive migrations, most notably during the Bronze and Iron Ages.
Paleogenetic studies of the southern Siberian region during the Bronze and Iron Ages have included mtDNA and Y chromosome analyses of Pazyryk, Xiongnu, Scythian, Tagar, and Kazakhstan peoples (Clisson et al., 2002; Keyser-Tracqui et al., 2003; Lalueza-Fox et al., 2004; Ricaut et al., 2004a, b; Chikisheva et al., 2007; Keyser et al., 2009; Pilipenko et al., 2010; Gonzalez-Ruiz et al., 2012). The general consensus among these studies is that the region of southern Siberia and Central Asia is quite diverse. However, studies have shown that Pazyryk and Tagar peoples possessed haplogroups that are rare or absent in southern Siberia today, such as mtDNA haplogroups U and U5a1 (Pilipenko et al., 2010). Interestingly, haplogroup U is present in high frequencies in ancient hunter-gatherers of Europe (Bramanti et al., 2009; Malmstrom et al., 2009) and was also found in a 24000 year old Upper Paleolithic boy from the Lake Baikal region (Mal’ta) of south-central Siberia (Raghavan et al., 2013). Haplogroup U5a1 has also been observed in eastern Kazakhstan from the Bronze Age (Lalueza-Fox et al., 2004). Therefore, the connection between pre-agricultural European peoples and the Pazyryk peoples may have been established from as early as the pre-Neolithic.
Our craniofacial study did not include peoples of modern-day Europe with high frequencies of haplogroup U or other typical Western Eurasian haplogroups such as R (Quintana-Murci et al., 2004). We did, however, observe stark differences between the Iron Age Siberians and modern Siberians that contain higher frequencies of common Eastern Eurasian haplogroups, such as M (Derenko et al., 2012; Gonzalez-Ruiz et al., 2012). Our results would therefore suggest the Pazyryk and Tagar peoples either had greater contact with West Eurasian peoples, or their ancestors originated or had close connections with the hunter-gatherers of Paleolithic Europe. This observation is reinforced through ancient DNA analysis. Keyser et al. (2009) analyzed mtDNA and Y chromosome haplotypes of Middle and Late Bronze Age, and Iron Age southern Siberian samples, including samples from the Tagar culture. Their results revealed that all of their samples contained the Y haplogroup R1a1, which is widely distributed on the Eurasian continent (Karafet et al., 2008). This haplogroup most likely reflects the expansion of peoples after the Last Glacial Maximum (~20–12 kya) and therefore the Bronze and Iron Age peoples of Siberia were part of a continuation of West Eurasian peoples that may have reached as far as Lake Baikal during the Upper Paleolithic (Raghavan et al., 2013).
While population genetic studies point mostly to Western (European) direction of relationships of Tagar and Pazyryk peoples, both archaeological and craniological data suggest fairly strong influence of Central or even East Asian populations on the Iron Age tribes of southern Siberia (Chikisheva, 2000a, 2008). It is known from ancient DNA studies (Keyser et al., 2009; Pilipenko et al., 2010) that western and southern Siberian Bronze Age samples harbored higher frequencies of Western Eurasian than Eastern Eurasian mtDNA haplogroups prior to the Iron Age. In the Keyser et al. (2009) study, the frequency of Western Eurasian mtDNA haplogroups reached 90%, but lowered to 67% during the Iron Age. This is similar to the results of Lalueza-Fox et al. (2004), who found that lineages in Kazakhstan before the Iron Age all belonged to Western Eurasian lineages. During the Late Iron Age, the influx of Turko-Mongolian Xiongnu increased the Eastern Eurasian lineages present in Kazakhstan (Lalueza-Fox et al., 2004). Therefore, the Tagar male outlier position in Figure 3 may be the result of incoming East Eurasian peoples during the Late Iron Age.
More generally, our results show some population differences between the Pazyryk and Tagar peoples. As seen in Figure 3, the two Iron Age groups do not overlap for the first two canonical variates when compared with a small number of contemporaneous groups. In fact, the Tagar males seem to be greater outliers than the Pazyryk while we can see the opposite situation for females of both groups. The Pazyryk in Figure 3 overlap with, and are close to the Iron Age Tuvans, and to a lesser degree to other Siberian Iron Age samples. Therefore, the connection between the Pazyryk peoples in Siberia may be stronger than for Tagar peoples.
The situation for female groups looks to be otherwise: the Pazyryk females seem to be outliers while Tagar females are similar to the Iron Age Siberia sample from Bystrovka-2. Taken together, these results point to a high probability of sex-biased admixture in both groups where migrating Tagar males assimilated some previous Iron Age population while Pazyryk males are more similar to neighboring Iron Age Tuvan peoples who may have had other sources of marital partners. Chikisheva (2008) showed that the nomads of Iron Age Tuva formed a subcluster with the pooled Pazyryk sample, but not the Tagar peoples. This result is clearly seen in our CVA shown in Figure 3, but only for the male sample while the females obviously differ from the Iron Age Tuvan group, at least on CV1. It should be noted that the Iron Age Tuvan sample is the closest group for both males and females of Tagar. However, this result should not be overemphasized as the comparison here is done among very few groups of generally common origin and cranial morphology.
After introducing several northeastern Siberian groups into the analysis (Figure 4), the Tagar and Pazyryk males do not differ greatly and strongly overlap, while females of both groups are still fairly distinguishable and do not overlap, indicating greater heterogeneity among ancient southern Siberian females. Overall, females of the Iron Age and medieval southern Siberian groups look more diverse compared to males of the same groups even against the background of such morphologically different groups such as Northeast Asians (Figure 4B).
However, when we compared the Pazyryk and Tagar samples to crania collected from Central Asian groups (Kazakhs, Turkmen, Uzbeks, Uighurs) with the addition of ancient Xiongnu, who originated in Mongolia and date to the Iron Age, and a Bronze Age sample from Xinjiang (Figure 5), we found the Pazyryk males to be more closely related to Central Asian groups than Tagar males, who show an outlier position though fairly close to Pazyryk males (Figure 5A). On average, both the Tagar and Pazyryk peoples are more closely related to Central Asian groups, namely the Uzbek sample and Pantusov’s collection from Kazakhstan (see Materials and Methods section for details of this sample provenance) than modern northern Siberian groups. Interestingly, when the ancient Xiongnu and the Bronze Age Xinjiang are included in the analysis, neither the Tagar nor Pazyryk peoples overlap with individuals from the Xiongnu sample (Figure 5). Thus, a direct genetic influx from the Xiongnu population does not look very probable in either the Tagar or Pazyryk peoples.
Since the Pazyryk and Tagar samples are pooled from different sites and different time periods, their affinity to the Central Asian Kazakh sample could stem from later incursions of Eastern Eurasian peoples toward the end of the Iron Age. These results have not been demonstrated previously, though Chikisheva (2000a) did observe for both male and female Pazyryk crania a close biological similarity to the ancient Saka and Wusun groups of eastern Kazakhstan and Xinjiang. Unfortunately, to our knowledge, we were unable to include ancient samples from Kazakhstan, however, we do show a similar pattern for our pooled Pazyryk sample, which could indicate that our Kazakh sample dates from much earlier than the last few centuries. Importantly, the Kazakh sample is the only one showing similarity to both sexes of (Tagar and Pazyryk) cultures. Pazyryk females, unlike males, do not show close affinity to modern Uzbek sample, and this fact again points to possible different origins of both sexes for this group. Tagar females, unlike males, overlap with Yakut females, which is very interesting when one accounts for the close geographical position of Tagar culture to the Pazyryk culture. Both Tagar and Pazyryk females are even less similar to Xiongnu than males of these groups.
The other interesting result in our data suggests the Pazyryk sample to be very closely related to a pooled sample from Iron Age Tuva. Chikisheva (2008) showed that the nomads of Iron Age Tuva formed a subcluster with the pooled Pazyryk sample, but not the Tagar peoples. This result is clearly seen in Figure 3. We have shown that the Tagar and Pazyryk males are closely related when multiple comparisons are made. This is not the case for females, who show similarity in only one of the analyses performed, and Tagar females display similarity to Yakut females, which may point to their possible origin.
These results suggest that the peoples of the Tagar and Pazyryk cultures each have a shared population history and contributed to the diversity of the southern Russian Altai. To better understand their population structure, definitive Iron and Bronze Age samples from Central Eurasia need to be included in the analyses. To get a clearer picture of these two important Iron Age nomadic groups, more Bronze and Iron Age samples from Siberia, Central Asia, and perhaps even Europe, should be analyzed using various physical anthropological, archaeological, and molecular methods in an effort to understand the early Iron Age of southern Siberia.
We would like to thank the museum curators and individuals from the institutions where skeletal material is curated. This includes Alisa Zubova at the Institute of Archaeology and Ethnography, Russian Academy of Sciences, Siberian Branch, Novosibirsk, Russian Federation, Zhu Hong and Wei Dong at the Research Center for Chinese Frontier Archaeology of Jilin University, Changchun, People’s Republic of China, Philippe Mennecier at the Musée de l’Homme, Paris, France, and Tumen Dashtseveg at the National University of Mongolia, Ulaabaatar. The first author would also like to thank Mary Margaret-Murphy at the University of Montana for help with data formatting, Pauline Sebillard at Jilin University for translating in Changchun, and Natasha Kharlamova at the Russian Academy of Sciences for help in Moscow. We would also like to thank the three anonymous reviewers and the Editor’s comments in helping to greatly improve the final version of this manuscript. Research was partially supported through a National Science Foundation Doctoral Dissertation Improvement Grant (BCS no. 1028773) (R.W.S.).