The Scythians: Who Were They?

DISCUSSION
Our analysis of more than 100 ancient individuals from Central Asia shows that IA nomad populations of the Kazakh Steppe formed through extensive admixture, resulting from complex interactions between preceding MLBA populations from the steppe and the neighboring regions (Figs. 2A, 3, A and C, and 4A). Our findings shed new light onto the debate about the origins of the Scythian cultures. We do not find support for a western Pontic-Caspian steppe origin, which is, in fact, highly questioned by more recent historical/archeological work (1, 2). The Kazakh Steppe origin hypothesis finds instead a better correspondence with our results, but rather than finding support for one of the two extreme hypotheses, i.e., single origin with population diffusion versus multiple independent origins with only cultural transmission, we found evidence for at least two independent origins as well as population diffusion and admixture (Fig. 4B). In particular, the eastern groups are consistent with descending from a gene pool that formed as a result of a mixture between preceding local steppe_MLBA sources (which could be associated with different cultures such as Sintashta, Srubnaya, and Andronovo that are genetically homogeneous) and a specific eastern Eurasian source that was already present during the LBA in the neighboring northern Mongolia region (27). The genetic structure of the Early IA Tasmola culture of central and northern Kazakhstan is mostly composed of an equal mixture of these two ancestries, although smaller amounts of gene flow from an Iranian-related source are also required (Figs. 3A and 4A and data file S4). We found that overall BMAC-related populations from Turan provide the best fit to our models while Iranian-related sources further to the west, such as the BA groups from the northern Caucasus, fail (data file S4). These results corroborate the historical/archaeological hypotheses of a cultural connection between the southern civilizations and the northern steppe people (3). This BMAC influx continues in the later fourth- to first-century BCE-CE Scythian groups from the northeastern Pazyryk site of Berel and becomes increasingly higher and nonuniformly distributed in the southeastern Saka individuals from the Tian Shan mountains (Figs. 2, B and D, 3, A and C, and 4B; fig. S4; table S3; and data file S4).




FIG. 4 Summary maps describing the major genetic turnovers that occurred at the turn and throughout the first millennium BCE.(A) Formation of a three-way LBA admixture cline from which (B) eastern Scythian and western Sarmatian gene pools arose and spread throughout the Steppe and (C) a new source of eastern Eurasian ancestry influx admixing with the Scythian gene pools started in the IA and becoming predominant and widespread at northern latitudes during the Xianbei-Hun period. On the very southern tips of the Steppe, a very different ancestry shift occurred, likely linked with the expansion of the Persian world. The arrows represent the demographic processes analyzed in the present study and are numbered from 1 to 6 to connect them to the main results shown in Fig. 3 from which these inferences have been drawn.

The two previously published individuals from the Aldy-Bel culture of the Arzhan 2 site in the Tuva region fall within the main eastern Scythian genetic cluster, confirming that it was present also in the same site where the earliest Scythian burials are found (Fig. 2A). These data, coupled with recent findings from the IA transition in Mongolia (28), seem to point to an origin in the Altai area of a main genetic substratum that formed all the eastern Scythians (Fig. 4B). The western Sarmatians from the southern Ural region also formed as a result of admixture between the same three ancestral sources as the eastern Scythians (Fig. 3A). Nevertheless, the eastern Eurasian ancestry is present only in a small amount in Sarmatians (Fig. 3A). In addition, their early admixture dates (Fig. 3C) and the absence of an admixture cline between the Sarmatians and the eastern groups (Fig. 2, A, B, and D) suggest that the Sarmatians descend from a related but different LBA gene pool compared with the one that contributed to the eastern Scythians (likely differently located along an LBA admixture cline). Given the geographic location of the earliest Sarmatian sites found so far, we hypothesize that this gene pool originated in the LBA southern Ural area (Fig. 4B). More data from the later and westernmost Scythian cultures of the Caucasus and eastern Europe will provide a better understanding of their genetic affinities with the earlier Scythians from the Kazakh Steppe analyzed in the current study. Furthermore, our results show that the northern sedentary Sargat-related cultures show a close genetic proximity with the Scythians especially with the eastern nomad groups (Fig. 2B). The Sargats show additional affinity not found in the Scythian groups ultimately related to a northern Siberian lineage (Figs. 2D and 3A). This is consistent with the historical hypothesis that the Sargat people formed as a result of admixture between incoming Scythian groups and an unsampled local or neighboring population that possibly carried this extra Siberian ancestry (3, 18).
From the second half of the first millennium BCE, we detect a major genetic shift in a number of outliers that are interestingly linked with the emergence of the Korgantas culture that replaced the Tasmola in central Kazakhstan. In particular, we observe an influx from an eastern Eurasian source that is different from the one that contributed to the shift in the LBA (Figs. 3A and 4C and table S2). At the turn of the first millennium CE, this mixed genetic profile became widespread among the northeastern individuals associated with the Xianbei-Hun cultures and the later medieval individuals (Figs. 2, C and D, 3B, and 4C, and table S2). The highly variable admixture proportions and dates obtained for those individuals suggest that this was an ongoing process that characterized the first centuries CE (first to fifth century at least; Fig. 3C, fig. S4, and table S3). Additional genetic data from the first millennium CE will allow a more comprehensive understanding of the nature and the extent of this heterogeneity. Instead, in the southern Kazakhstan region, the individuals from the Konyr Tobe site located in the ancient city of Otyrar Oasis show a different genetic turnover mostly characterized by an increase in Iranian-related genetic ancestry, most likely reflecting the influence of the Persian empires (Fig. 4C) (20, 29). Outliers, with high eastern Eurasian admixture or with gene flow from South Asia, suggest that the population of this city at that time was heterogeneous (Fig. 2C and data file S4). During this period, Otyrar was a main center of the Kangju kingdom and a crossroad along the Silk Road (29). In the neighboring region of the Tian Shan mountains, in the third century CE site of Alai Nura, a genetic profile typical of the much earlier IA Tian Shan Sakas can still be found (Fig. 3B and data file S4).
The heterogeneity and geographic structuring observed during the IA, the Xianbei-Hun, and the medieval periods in Kazakhstan come in strong contrast with the genetic homogeneity observed among present-day Kazakhs (fig. S5). Fine-scale haplotype-based analyses confirmed this homogeneity and showed, in line with previous findings (26), that the Kazakh gene pool is a mixture of different western and eastern Eurasian sources (table S4). Our results on the ancient populations revealed that this was a result of the very complex demographic history, with multiple layers of western and eastern Eurasian ancestries mixing through time. The admixture dates obtained for present-day Kazakhs overlap with the period when the Kazakh Khanate was established (~15th century CE; table S5). Furthermore, the gene pool of present-day Kazakhs cannot be fully modeled as a mixture of post-IA northern Xianbei-Hun and southern Kangju-related gene pools (data file S4). These findings suggest that recent events, likely enfolding during the second millennium CE, were associated with more demographic turnovers in this region that ultimately lead to the homogenization of the Kazakh gene pool as a consequence of the establishment of the Kazakh Khanate with its strict exogamic rules (21).