Post by Admin on Jan 23, 2021 1:01:13 GMT
EF ancestry in Fatyanovo Culture individuals
Then, we turned to the Bronze Age Fatyanovo Culture individuals and determined that they carry maternal (subclades of mtDNA hg U5, U4, U2e, H, T, W, J, K, I, and N1a) and paternal (chrY hg R1a-M417) lineages (Table 1, fig. S1, and tables S2 to S4) that have also been found in CWC individuals elsewhere in Europe (14–16, 18, 27). In all individuals for which the chrY hg could be determined with sufficient depth (n = 6), it is R1a2-Z93 (Table 1 and tables S1 and S2), a lineage now spread in Central and South Asia, rather than the R1a1-Z283 lineage that is common in Europe (36). R1a2-Z93 is also not rejected for the individuals that were determined with less depth due to missing data on more apical markers (table S2).
On the PCA, the Fatyanovo individuals (and the Estonian CWC individual) group together with many European Late Neolithic/Bronze Age (LNBA) and Steppe Middle/Late Bronze Age individuals on top of modern Northern and Eastern Europeans (Fig. 2A). This ancient cluster is shifted toward Anatolian and European EF compared to Steppe Early/Middle Bronze Age populations, including the Yamnaya. The same could be seen in ADMIXTURE analysis where the Fatyanovo individuals are most similar to LNBA Steppe ancestry populations from Central Europe, Scandinavia, and the Eastern Baltic (Fig. 2B and fig. S2). These populations are composed of the “WHG” (blue) and “ancient Caucasus/Iran” (olive) component and small amounts of the “Russian Far East” (orange) component, similarly to Yamnaya populations. However, the European LNBA populations (including Fatyanovo) also display a component most frequent in Anatolian and European EF populations (light green), which is not present in the Yamnaya from Russia.
We studied the affinities of the Fatyanovo individuals by comparing FST, outgroup f3, and D statistics’ results of different populations and found that Fatyanovo shares more with European EF populations and modern Near Easterners than Yamnaya Samara does (Fig. 3C; figs. S3, S4C, and S5C; and tables S5 to S8). This signal can also be seen when using either autosomal or X chromosome (chrX) positions from the 1240K dataset (https://reich.hms.harvard.edu/downloadable-genotypes-present-day-and-ancient-dna-data-compiled-published-papers) instead of the autosomal positions of the HO dataset with less single-nucleotide polymorphisms (SNPs) (fig. S5, A and B, and tables S11 and S12). We also compared the affinities of Yamnaya Samara and Fatyanovo directly with D statistics and saw that Fatyanovo is significantly more similar (Z > 3) to most EF populations than to Yamnaya Samara, and the latter, in turn, is significantly more similar to most Steppe populations than to Fatyanovo (table S13). We studied the apparent EF input in Fatyanovo further using admixture f3 statistics and got significant results (Z < −3) for admixture between different Yamnaya groups and a wide variety of EF populations (table S14). Furthermore, when comparing outgroup f3 and D statistics’ results for Fatyanovo and Central CWC, there are no clear differences in their affinities to different ancient or modern population groups (Fig. 3D, fig. S3D, and tables S5 to S8).
Because the previous analyses suggested that the genetic makeup of the Fatyanovo Culture individuals was a result of admixture between migrating Yamnaya individuals and contemporary European populations, we used two complementary methods (qpAdm and ChromoPainter/NNLS) to determine suitable proxies for the admixing populations and the mixing proportions (Fig. 4 and tables S15 to S17). We tested qpAdm models including Yamnaya from Samara or Kalmykia and a variety of EF populations one at a time and found that the two EF populations, with the highest P values with both Yamnaya populations, are Globular Amphora and Trypillia (P values of 0.02/0.16 and 0.06/0.26, respectively) (table S15). The admixture proportions are 65.5%/66.9% Yamnaya Samara/Kalmykia + 34.5%/33.1% Globular Amphora and 65.5%/69.6% Yamnaya Samara/Kalmykia + 34.5%/30.4% Trypillia, respectively (Fig. 4 and table S15). The proportions are similar (69 to 75% Yamnaya + 25 to 31% EF) for Central and Baltic CWC (Fig. 4 and table S15). We also tested these four models with the preceding Volosovo Culture HG BER001 added to “right” populations to see whether there is shared drift between Volosovo and Fatyanovo that would cause the models with admixing populations without this drift to be rejected. All four models are still not rejected (P values of 0.97/0.57 and 0.98/0.59, respectively) (table S15), suggesting that no Volosovo contribution is needed to model Fatyanovo. Ancestry proportions were also calculated using the ChromoPainter/NNLS pipeline with the results 37%/38% Yamnaya Samara + 63%/62% Globular Amphora/Trypillia for Fatyanovo and 51 to 60% Yamnaya + 40 to 49% EF for Central/Baltic CWC (table S16). Although the estimated proportion of EF ancestry is higher for Fatyanovo compared to the other groups in both cases, the difference is significant (P value of 0.005 to 0.03) only with Trypillia in the model. Note that qpAdm calculates admixture between populations, while ChromoPainter/NNLS uses single individuals as sources, which might influence the results. Although two-way admixture between Yamnaya and an EF population is enough to explain the genetic variation in Fatyanovo, qpAdm models with an HG population added are also not rejected with EHG, WeRuHG, and Volosovo (P values of 0.14 to 0.98) (table S17). Fatyanovo can be modeled as 60 to 63% Yamnaya Samara + 33 to 34% Globular Amphora + 3 to 6% HG. The results are similar for Central and Baltic CWC with 2 to 11% of HG ancestry (except with Volosovo as a source for Central CWC, which gets a negative mixture coefficient).
Fig. 4 qpAdm admixture modeling results.
(A) Models with Yamnaya Samara or Kalmykia and Globular Amphora as sources for Fatyanovo, Central, and Baltic CWC populations. (B) Models with Yamnaya Kalmykia and Globular Amphora as sources for Fatyanovo individuals with the oldest radiocarbon date on top and the youngest on the bottom. (C) Models with Yamnaya Samara or Kalmykia and Trypillia as sources for Fatyanovo, Central, and Baltic CWC populations.
We estimated the time of admixture for Yamnaya and EF populations to form the Fatyanovo Culture population using DATES (37) as 13 ± 2 generations for Yamnaya Samara + Globular Amphora and 19 ± 5 generations for Yamnaya Samara + Trypillia. If a generation time of 25 years and the average calibrated date of the Fatyanovo individuals (~2600 cal BCE) are used, this equates to the admixture happening ~3100 to 2900 BCE.
Next, we investigated the possible difference in affinities between Fatyanovo and other CWC groups (Central and Baltic CWC) and found that the populations are similar to each other because a one-way qpAdm model cannot be rejected for Fatyanovo = Central CWC (P value of 0.26) or Central CWC = Baltic CWC (P value of 0.48) (table S18). On the other hand, there is considerable variation in admixture proportions within populations, visible on PCA (Fig. 2) and ADMIXTURE (fig. S2A) and confirmed by per-individual qpAdm models showing 4 to 47% Globular Amphora ancestry in Fatyanovo (Fig. 4B and table S15) and 7 to 55% in the other two groups (table S15). We tested whether the variation in ancestry correlates with time using the second PC component (PC2) or the qpAdm ancestry proportions and the calibrated radiocarbon dates of the individuals. We found that there is no correlation between time and ancestry proportions in Fatyanovo (P value of 0.92/0.23, respectively), but there is a significant change toward more EF ancestry in Baltic CWC using PC2 (P value of 0.0003) and in both Central and Baltic CWC using qpAdm proportions (P value of 0.02 for both).
Furthermore, we confirmed the presence of sex-biased admixture previously seen in CWC individuals from Estonia, Poland, and Germany (27, 38, 39) in the Fatyanovo. To that end, we first compared autosomal and chrX outgroup f3 results (fig. S5D and tables S11 and S12). Two-sample two-tailed t tests assuming unequal variances showed that the mean f3 value for EF populations is significantly lower than for HG/Steppe ancestry populations based on autosomal positions (P value of 0.000001) but not based on chrX positions (P value of 0.14). Next, we calculated admixture proportions on chrX data using qpAdm and the same models as with autosomal data (table S15). Only two of the four models presented for autosomal data (Yamnaya Samara + Globular Amphora/Trypillia) yielded a significant P value (0.13/0.36, respectively) due to the low number of chrX SNPs available. The confidence intervals (CIs) were extremely wide with Trypillia, but the chrX data showed 40 to 53% Globular Amphora ancestry in Fatyanovo, in contrast with the 32 to 36% estimated using autosomal data. The sex-biased admixture is also supported by the presence of mtDNA hg N1a in two Fatyanovo individuals—an hg frequent in Linear Pottery Culture (LBK) EFs but not found in Yamnaya individuals so far (16, 18, 40)—and by all males carrying chrY hg R1a-M417, which appeared in Europe after the Steppe migration (15, 18).
Last, we looked for closely related individuals in the Fatyanovo Culture sample set using READ (41). There were no confirmed cases of second-degree or closer relatives (fig. S6), although a second-degree relationship cannot be ruled out for some pairs as the 95% CIs of their point estimates overlap with those of the second-degree relatedness threshold.
Then, we turned to the Bronze Age Fatyanovo Culture individuals and determined that they carry maternal (subclades of mtDNA hg U5, U4, U2e, H, T, W, J, K, I, and N1a) and paternal (chrY hg R1a-M417) lineages (Table 1, fig. S1, and tables S2 to S4) that have also been found in CWC individuals elsewhere in Europe (14–16, 18, 27). In all individuals for which the chrY hg could be determined with sufficient depth (n = 6), it is R1a2-Z93 (Table 1 and tables S1 and S2), a lineage now spread in Central and South Asia, rather than the R1a1-Z283 lineage that is common in Europe (36). R1a2-Z93 is also not rejected for the individuals that were determined with less depth due to missing data on more apical markers (table S2).
On the PCA, the Fatyanovo individuals (and the Estonian CWC individual) group together with many European Late Neolithic/Bronze Age (LNBA) and Steppe Middle/Late Bronze Age individuals on top of modern Northern and Eastern Europeans (Fig. 2A). This ancient cluster is shifted toward Anatolian and European EF compared to Steppe Early/Middle Bronze Age populations, including the Yamnaya. The same could be seen in ADMIXTURE analysis where the Fatyanovo individuals are most similar to LNBA Steppe ancestry populations from Central Europe, Scandinavia, and the Eastern Baltic (Fig. 2B and fig. S2). These populations are composed of the “WHG” (blue) and “ancient Caucasus/Iran” (olive) component and small amounts of the “Russian Far East” (orange) component, similarly to Yamnaya populations. However, the European LNBA populations (including Fatyanovo) also display a component most frequent in Anatolian and European EF populations (light green), which is not present in the Yamnaya from Russia.
We studied the affinities of the Fatyanovo individuals by comparing FST, outgroup f3, and D statistics’ results of different populations and found that Fatyanovo shares more with European EF populations and modern Near Easterners than Yamnaya Samara does (Fig. 3C; figs. S3, S4C, and S5C; and tables S5 to S8). This signal can also be seen when using either autosomal or X chromosome (chrX) positions from the 1240K dataset (https://reich.hms.harvard.edu/downloadable-genotypes-present-day-and-ancient-dna-data-compiled-published-papers) instead of the autosomal positions of the HO dataset with less single-nucleotide polymorphisms (SNPs) (fig. S5, A and B, and tables S11 and S12). We also compared the affinities of Yamnaya Samara and Fatyanovo directly with D statistics and saw that Fatyanovo is significantly more similar (Z > 3) to most EF populations than to Yamnaya Samara, and the latter, in turn, is significantly more similar to most Steppe populations than to Fatyanovo (table S13). We studied the apparent EF input in Fatyanovo further using admixture f3 statistics and got significant results (Z < −3) for admixture between different Yamnaya groups and a wide variety of EF populations (table S14). Furthermore, when comparing outgroup f3 and D statistics’ results for Fatyanovo and Central CWC, there are no clear differences in their affinities to different ancient or modern population groups (Fig. 3D, fig. S3D, and tables S5 to S8).
Because the previous analyses suggested that the genetic makeup of the Fatyanovo Culture individuals was a result of admixture between migrating Yamnaya individuals and contemporary European populations, we used two complementary methods (qpAdm and ChromoPainter/NNLS) to determine suitable proxies for the admixing populations and the mixing proportions (Fig. 4 and tables S15 to S17). We tested qpAdm models including Yamnaya from Samara or Kalmykia and a variety of EF populations one at a time and found that the two EF populations, with the highest P values with both Yamnaya populations, are Globular Amphora and Trypillia (P values of 0.02/0.16 and 0.06/0.26, respectively) (table S15). The admixture proportions are 65.5%/66.9% Yamnaya Samara/Kalmykia + 34.5%/33.1% Globular Amphora and 65.5%/69.6% Yamnaya Samara/Kalmykia + 34.5%/30.4% Trypillia, respectively (Fig. 4 and table S15). The proportions are similar (69 to 75% Yamnaya + 25 to 31% EF) for Central and Baltic CWC (Fig. 4 and table S15). We also tested these four models with the preceding Volosovo Culture HG BER001 added to “right” populations to see whether there is shared drift between Volosovo and Fatyanovo that would cause the models with admixing populations without this drift to be rejected. All four models are still not rejected (P values of 0.97/0.57 and 0.98/0.59, respectively) (table S15), suggesting that no Volosovo contribution is needed to model Fatyanovo. Ancestry proportions were also calculated using the ChromoPainter/NNLS pipeline with the results 37%/38% Yamnaya Samara + 63%/62% Globular Amphora/Trypillia for Fatyanovo and 51 to 60% Yamnaya + 40 to 49% EF for Central/Baltic CWC (table S16). Although the estimated proportion of EF ancestry is higher for Fatyanovo compared to the other groups in both cases, the difference is significant (P value of 0.005 to 0.03) only with Trypillia in the model. Note that qpAdm calculates admixture between populations, while ChromoPainter/NNLS uses single individuals as sources, which might influence the results. Although two-way admixture between Yamnaya and an EF population is enough to explain the genetic variation in Fatyanovo, qpAdm models with an HG population added are also not rejected with EHG, WeRuHG, and Volosovo (P values of 0.14 to 0.98) (table S17). Fatyanovo can be modeled as 60 to 63% Yamnaya Samara + 33 to 34% Globular Amphora + 3 to 6% HG. The results are similar for Central and Baltic CWC with 2 to 11% of HG ancestry (except with Volosovo as a source for Central CWC, which gets a negative mixture coefficient).
Fig. 4 qpAdm admixture modeling results.
(A) Models with Yamnaya Samara or Kalmykia and Globular Amphora as sources for Fatyanovo, Central, and Baltic CWC populations. (B) Models with Yamnaya Kalmykia and Globular Amphora as sources for Fatyanovo individuals with the oldest radiocarbon date on top and the youngest on the bottom. (C) Models with Yamnaya Samara or Kalmykia and Trypillia as sources for Fatyanovo, Central, and Baltic CWC populations.
We estimated the time of admixture for Yamnaya and EF populations to form the Fatyanovo Culture population using DATES (37) as 13 ± 2 generations for Yamnaya Samara + Globular Amphora and 19 ± 5 generations for Yamnaya Samara + Trypillia. If a generation time of 25 years and the average calibrated date of the Fatyanovo individuals (~2600 cal BCE) are used, this equates to the admixture happening ~3100 to 2900 BCE.
Next, we investigated the possible difference in affinities between Fatyanovo and other CWC groups (Central and Baltic CWC) and found that the populations are similar to each other because a one-way qpAdm model cannot be rejected for Fatyanovo = Central CWC (P value of 0.26) or Central CWC = Baltic CWC (P value of 0.48) (table S18). On the other hand, there is considerable variation in admixture proportions within populations, visible on PCA (Fig. 2) and ADMIXTURE (fig. S2A) and confirmed by per-individual qpAdm models showing 4 to 47% Globular Amphora ancestry in Fatyanovo (Fig. 4B and table S15) and 7 to 55% in the other two groups (table S15). We tested whether the variation in ancestry correlates with time using the second PC component (PC2) or the qpAdm ancestry proportions and the calibrated radiocarbon dates of the individuals. We found that there is no correlation between time and ancestry proportions in Fatyanovo (P value of 0.92/0.23, respectively), but there is a significant change toward more EF ancestry in Baltic CWC using PC2 (P value of 0.0003) and in both Central and Baltic CWC using qpAdm proportions (P value of 0.02 for both).
Furthermore, we confirmed the presence of sex-biased admixture previously seen in CWC individuals from Estonia, Poland, and Germany (27, 38, 39) in the Fatyanovo. To that end, we first compared autosomal and chrX outgroup f3 results (fig. S5D and tables S11 and S12). Two-sample two-tailed t tests assuming unequal variances showed that the mean f3 value for EF populations is significantly lower than for HG/Steppe ancestry populations based on autosomal positions (P value of 0.000001) but not based on chrX positions (P value of 0.14). Next, we calculated admixture proportions on chrX data using qpAdm and the same models as with autosomal data (table S15). Only two of the four models presented for autosomal data (Yamnaya Samara + Globular Amphora/Trypillia) yielded a significant P value (0.13/0.36, respectively) due to the low number of chrX SNPs available. The confidence intervals (CIs) were extremely wide with Trypillia, but the chrX data showed 40 to 53% Globular Amphora ancestry in Fatyanovo, in contrast with the 32 to 36% estimated using autosomal data. The sex-biased admixture is also supported by the presence of mtDNA hg N1a in two Fatyanovo individuals—an hg frequent in Linear Pottery Culture (LBK) EFs but not found in Yamnaya individuals so far (16, 18, 40)—and by all males carrying chrY hg R1a-M417, which appeared in Europe after the Steppe migration (15, 18).
Last, we looked for closely related individuals in the Fatyanovo Culture sample set using READ (41). There were no confirmed cases of second-degree or closer relatives (fig. S6), although a second-degree relationship cannot be ruled out for some pairs as the 95% CIs of their point estimates overlap with those of the second-degree relatedness threshold.
