A Bronze Age Woman from Bohemia: Unetice Culture

Fig. 3. WHG ancestry in pre-CW Bohemia.
(A) The proportion of WHG ancestry through time in pre-CW individuals from Bohemia modeled as a two-stage linear process (table S8). The gray area indicates 95% confidence interval. (B) Left: Proportion of ancestry ascribable to Anatolia_Neolithic, Loschbour, and Körös_HG in pre-CW cultural groups from Bohemia in chronological order from bottom to top (sample size of each cultural group in brackets and P value of three-way qpAdm model indicated within orange bars; table S10). Right: Inferred dates (2 SE; table S11) of HG admixture (gray interval) relative to culture’s chronology (black interval). (C) Zoomed-in PCA showing (with the exception of TUC003) segregation between Bohemian GAC and Řivnáč individuals along with position of early CW females without steppe ancestry (green circles). Black dots represent previously published GAC individuals from Poland and Ukraine.
To gain insight into the process(es) by which HG ancestry was incorporated into the gene pool of pre-CW individuals from Bohemia, we used qpAdm to model each pre-CW cultural group as a three-way mixture of Anatolia_Neolithic, Loschbour, and Körös_HG as well as DATES to estimate the introgression date of incorporated HG ancestry (Fig. 3B and tables S10 and S11). Under a scenario of population continuity with sequential incorporation of HG ancestry, the mean date of introgression, as indicated by the gray intervals in Fig. 3B (right), for succeeding cultures is expected to become more recent through time. Conversely, under population continuity without incorporation of further HG ancestry, the admixture date should be similar for successive cultural groups who have similar HG proportions.

Our results indicate two cultural transitions for which either of these expectations is not met. First, although Bohemian Jordanów and Funnelbeaker have similar amounts of WHG ancestry [f4(Mbuti.DG, WHG; Jordanów, Funnelbeaker) ~0; z score, 0.96], the estimated date of WHG introgression for Funnelbeaker is significantly earlier (5079 to 4748 BCE) than for Jordanów (4636 to 4310 BCE) (Fig. 3B and table S11), consistent with Bohemian Funnelbeaker individuals being derived from a different population (whose HG ancestry was incorporated further back in time), which superseded the Jordanów population in Bohemia. This transition between Jordanów and Funnelbeaker is corroborated by three additional observations. First, an f4 statistic of the form f4(Mbuti.DG, Bohemia-Funnelbeaker; Bohemia-Jordanów, Germany-Funnelbeaker) is positive (z score, 3.14), revealing significantly greater genetic affinity of Bohemian Funnelbeaker to Funnelbeaker (Baalberge and Salzmünde) individuals from Saxony-Anhalt than to the preceding local Jordanów individuals. Conversely, f4(Mbuti.DG, Bohemia-Jordanów; Germany-Funnelbeaker, Bohemia-Funnelbeaker) is consistent with 0 (z score, 1.03), suggesting phylogenetic cladality between Bohemian and German Funnelbeaker with respect to Bohemia-Jordanów individuals. Second, Bohemia-Jordanów individuals can be modeled as a two-way mixture of Anatolia_Neolithic and Körös_HG but not Anatolia_Neolithic and Loschbour, while the opposite is true for Bohemia-Funnelbeaker (table S12). This suggests different affinities, in addition to the different introgression dates, of the HG ancestries in Bohemian Jordanów and Funnelbeaker cultural groups. Third, qpWave does not support cladality between Bohemian Jordanów and Funnelbeaker (P = 0.00679), while cladality between Bohemian and German Funnelbeaker cannot be rejected (P = 0.88; table S13). Together, these results indicate a largely (significantly more than 50%) nonlocal genetic origin of Bohemian Funnelbeaker individuals.

The second such case can be seen in the Řivnáč to GAC cultural transition. GAC individuals carry the most HG ancestry among pre-CW cultural groups from Bohemia (25.7%, ±1.4), significantly more than Řivnáč individuals [f4(Mbuti.DG, WHG; Řivnáč, GAC) >> 0; z score, 4.46]. However, the estimated date of HG admixture in GAC is not later than in Řivnáč individuals (Fig. 3B and table S11), suggesting that GAC individuals do not descend from a recent mixture of Řivnáč and an HG source but instead constituted a recent, nonlocal incursion in Bohemia from a region that received more HG gene flow [e.g., Poland (13, 63)], in accordance with interpretations of archaeological evidence (56).

A distinct genetic origin for Řivnáč and GAC individuals is further supported by PCA and qpAdm modeling. From PCA, we find that with the exception of TUC003, Řivnáč and GAC individuals form distinct clouds (Fig. 3C). This is confirmed by qpAdm modeling where GAC individuals can be modeled as a mixture of Anatolia_Neolithic and Loschbour but not Anatolia_Neolithic and Körös_HG, while the opposite is true for Řivnáč individuals (table S14). Consequently, Řivnáč and GAC individuals are distinguishable based on the amount and source of HG ancestry, suggesting that Bohemia was inhabited by genetically differentiated groups of Řivnáč and GAC individuals at the time of CW appearance. The Řivnáč outlier (TUC003) also raises the interesting possibility of an individual born into a GAC but buried in a Řivnáč cultural context.

Among the 16 Řivnáč and GAC individuals who are contemporaneous with or postdate the appearance of CW in Bohemia (Fig. 1B), we find no detectable traces of steppe ancestry (Fig. 2A and table S8), suggesting that biological exchange from CW/Yamnaya into culturally pre-CW people (e.g., Řivnáč and GAC) was low, possibly nonexistent. Steppe ancestry coappears with CW individuals in early third millennium BCE Bohemia.

In addition to autosomal genetic changes through time, we observe a sharp reduction in Y-chromosomal diversity going from five different lineages in early CW to a dominant (single) lineage in late CW (Fig. 4A). We used forward simulations to explore the demographic scenarios that could account for the observed reduction in Y-chromosomal diversity. Performing 1 million simulations of a population with a starting frequency of R1a-M417(xZ645) centered around the observed starting frequency in Bohemia_CW_Early (3 of 11, 0.27), we assessed the plausibility of this lineage reaching the observed frequency in Bohemia_CW_Late (10 of 11, 0.91) in the time frame of 500 years under a model of a closed population and random mating (Materials and Methods). We reject the “neutral” hypothesis, i.e., that this change in frequency occurred by chance, given a wide range of plausible population sizes. Instead, our results suggest that R1a-M417(xZ645) was subject to a nonrandom increase in frequency, resulting in these males having 15.79% (4.12 to 44.42%) more surviving offspring per generation relative to males of other Y-haplogroups. We also find that this change in Y chromosome frequency is extreme compared to the changes in allele frequencies at fully covered autosomal 1240k sites (P < 0.0003) within the same males, suggesting a process that disproportionately affected Y-chromosomal compared to autosomal genetic diversity, ruling out a population bottleneck as the likely cause. Our results suggest that the Y-lineage diversity in early CW males was supplanted by a nonrandom process [selection, social structure, or influx of nonlocal R1a-M417(xZ645) lineages] that drove the collapse in Y-chromosomal diversity. A simultaneous decline of Y-chromosomal diversity dating to the Neolithic has been observed across most extant Y-haplogroups (64), possibly due to increased conflict between male-mediated patrilines (65). We view that changes in social structure (e.g., an isolated mating network with strictly exclusive social norms) could be an alternative cause but would be difficult to distinguish in the underlying model parameters.



Fig. 4. Temporal Y chromosome and autosomal PC2 variation in Bohemia.
(A) Temporal distribution of Y chromosome haplogroups by culture. Schematic of phylogenetic relationships between Y chromosome lineages is shown along y axis. Dashed vertical lines demarcate respective (colored) cultural group into early and late phases. (B) Temporal variation in PC2 showing the genetic variation of males and females within each cultural group.

The greatest genetic differentiation within early CW individuals can be found at Vliněves. The fst value between the three highest and three lowest early CW individuals on PC2 from Vliněves is greater than pairwise comparisons of all modern-day European populations (Fig. 5 and table S27).



Fig. 5. Genetic distances of Early CW individuals from Vliněves.
Pairwise Fst between the three highest (VLI076, VLI088, and VLI090; VLI_High) and three lowest (VLI008, VLI079, and VLI009; VLI_Low) early CW on PC2 from Vliněves in the context of modern European pairwise Fst (2 SE plotted; table S27). Inset: PCA of 1141 modern West Eurasian individuals (gray points) on which early CW individuals from Vliněves (green symbols with black outline) were projected (see also Fig. 2B). The highly differentiated pairs of modern European populations labeled in the main figure are colored in the inset PCA.