Ancient DNA connects large-scale migration with the spread of Slavs
PMCID: PMC12507669
PMID: 40903570
Abstract
The second half of the first millennium ce in Central and Eastern Europe was accompanied by fundamental cultural and political transformations. This period of change is commonly associated with the appearance of the Slavs, which is supported by textual evidence 1 , 2 and coincides with the emergence of similar archaeological horizons 3 – 6 . However, so far there has been no consensus on whether this archaeological horizon spread by migration, Slavicisation or a combination of both. Genetic data remain sparse, especially owing to the widespread practice of cremation in the early phase of the Slavic settlement. Here we present genome-wide data from 555 ancient individuals, including 359 samples from Slavic contexts from as early as the seventh century ce . Our data demonstrate large-scale population movement from Eastern Europe during the sixth to eighth centuries, replacing more than 80% of the local gene pool in Eastern Germany, Poland and Croatia. Yet, we also show substantial regional heterogeneity as well as a lack of sex-biased admixture, indicating varying degrees of cultural assimilation of the autochthonous populations. Comparing archaeological and genetic evidence, we find that the change in ancestry in Eastern Germany coincided with a change in social organization, characterized by an intensification of inter- and intra-site genetic relatedness and patrilocality. On the European scale, it appears plausible that the changes in material culture and language between the sixth and eighth centuries were connected to these large-scale population movements. Analyses of ancient human DNA show that cultural and political transformations in Central Europe during the second half of the first millennium ce were associated with movements of Slavic populations into Germany, Poland and Croatia.
Full Text
The term Slavs first appears as an ethnonym in the course of the sixth century in Constantinople and later in the west (Box 1 and Supplementary Note 1.1). Written sources locate them initially north of the Lower Danube, and later in the Carpathian Basin, the Balkans and the Eastern Alps (Extended Data Fig. 1). Many came under the rule of the Avar steppe empire along the Middle Danube (567 ce to around 800 ce). In the seventh century, there is evidence for the presence of Slavs in much of East-Central and Southeastern Europe. Where Slavs lived, Roman, Germanic and other pre-Slavic infrastructures were usually replaced by rather simple ways of life, archaeologically characterized by small settlements of pit houses, cremation burials, handmade, undecorated pottery and modest, low-metal material culture, known as the Prague-Korchak group. (Supplementary Note 1.2). More complex social systems and regional rulership developed later in the contact zones with Byzantium and the Christian west.
The similarity of early Slavic cultures was often attributed to a swift spread of Slavs from Northeast of the Carpathians, although debates continue, not only about their geographical origin (Supplementary Note 1,1). In Poland, the non-native (allochthonist) view assumes Slavic origin from Ukraine–Belarus, whereas the native (autochthonist) concept asserts that their ancestors inhabited Polish territory since the Bronze Age. Some scholars doubt Slavic expansion by migrations and assume that there was ‘Slavicisation’ of existing populations (Supplementary Note 1,2). Previous modern and ancient DNA studies have supported gene flow into the Northern Balkans and the Russian Volga-Oka region, but also argued for population continuity in Poland, so that the scale and sequence of these movements and their association with ‘Slavic’ material culture has remained unclear. Eventually, this cultural transformation led to the replacement of Germanic and other languages in East-Central and Southeastern Europe and the introduction of Slavic languages, which today represent the largest linguistic group in Europe. Yet, this presumed joint spread of language and material culture is difficult to trace, given that the first longer texts in Slavic were written in the late ninth century.
We selected skeletal remains from 591 ancient individuals from 26 different sites from Central and Eastern Europe (Supplementary Tables 1 and 2), creating, in combination with previously published data, a dense sampling transect for three regions: (1) Elbe-Saale Region in Eastern Germany as the main study area; (2) the Northwestern Balkans; and (3) Poland–Northwestern Ukraine (Extended Data Fig. 2 and Supplementary Tables 7–10). Complementary to these three transects, we generated new data and collected published data from the Baltics and Northwestern Russia to form a reference transect in the east. After hybridization DNA capture and quality filtering (Methods), genome-wide data for 555 unique individuals with a median coverage of 538k single nucleotide polymorphisms (SNPs) (on 1240k data) were available for analysis, including 359 ancient individuals from the SP, as well as 205 individuals predating the cultural transformations connected to the emergence of the Slavs (Fig. 1 and Supplementary Table 1). We analyse the ancient genome-wide data (Supplementary Table 11) together with an extended dataset of more than 11,500 present-day Europeans (Supplementary Table 3), covering all major Slavic-speaking groups, including data from more than 600 individuals belonging to the Sorbian minority in Eastern Germany.
To visualize genome-wide ancestry diversity before and after the spread of Slavic groups, we performed principal component analysis (PCA) on 10,528 present-day Europeans and projected our newly reported and other relevant ancient genome-wide data onto their genetic variation (Fig. 2). When comparing the SP samples to earlier and present-day data from our three study regions, we observe that the genetic composition within the transects changed markedly between about 600 and 800 ce. In general, the Roman and MP samples that predate the arrival of Slavic groups show high genetic heterogeneity in PCA space, with most samples from Germany and Poland clustering with present-day continental Northern German, Dutch and Scandinavian populations (Extended Data Fig. 3 and Supplementary Figs. 6, 7, 10 and 12), whereas the Roman and MP individuals from Croatia cluster with present-day Italian and Eastern Mediterranean populations (Fig. 2c and Supplementary Note 3).
In Eastern Germany and the Northwestern Balkans, most of the genetic diversity within the Roman and MP clusters follows a north–south cline along PC1. For the Northwestern Balkans, this heterogeneity has been attributed to increasing Eastern Mediterranean ancestry that arrived subsequently to the incorporation of the region into the Roman Empire. More unexpectedly, we detect a high number of MP individuals with non-local, Southern European ancestry in the Elbe-Saale region of Eastern Germany, although this area was never part of the Roman Empire. Using qpAdm, we measure on average between approximately 15% and 25% of Southern European ancestry in all 4 MP sites of the region (Extended Data Fig. 3).
Both PCA-based MOBEST analysis and F4 statistics indicate that this non-local ancestry was most probably derived from contemporaneous source populations in Italy and/or the Northern Balkan Peninsula (or other areas of the Roman Empire where people of this ancestry were located) (Extended Data Fig. 3 and Supplementary Figs. 3–5). Previous studies already identified mixed communities of northern and southern ancestry in Hungary and Northern Italy that were interpreted as amalgamation between Northern European newcomers and the local romanized population. In contrast to these earlier results, we do not find evidence that the two different ancestries were correlated to differences in material culture (Supplementary Table 4). Applying a generalized linear model, we demonstrate that neither the presence of grave goods overall, nor certain types of artefacts (such as weapons or brooches) are significantly correlated with either PCA position or ADMIXTURE profiles (Supplementary Fig. 60). Instead, we find the only significant (P < 0.05) correlation between ancestry and material culture among the burial constructions, where we show that individuals buried in pits feature on average higher Northern European ancestry (Supplementary Fig. 60c). The spatial organization of the burials was also not determined by similarity in ancestry. Instead, we observe that individuals were buried close to their biological relatives, within small kin groups composed of individuals with Northern European, Southern European or mixed ancestry, reflecting a high degree of admixture between individuals with different ancestry backgrounds during the MP. Consequently, our data from Eastern Germany demonstrate that the cosmopolitan character of the Roman Empire not only affected the incorporated territories but also facilitated exchange and mobility along its borders and beyond into barbarian lands (Barbaricum), resulting in an unprecedented genetic diversity in Central Europe during and, in the case of Eastern Germany, even after its existence. Although the causes and circumstances of their movement to the Elbe-Saale region remain open for speculation, these newcomers apparently adapted the fashions and traditions of the local populations, resulting in a rather homogenous material culture within a group of individuals with diverse genetic backgrounds.
However, this diversity had collapsed in the subsequent SP (Supplementary Note 6). In contrast to the preceding MP, the genetic profile of Eastern Germany during the SP has shifted considerably and clusters nearly exclusively with present-day Slavic-speaking populations (for example, Poles and Belarussians), indicative of a fundamental replacement of genetic ancestry (Fig. 2b,c). A similar pattern is seen in the Northwestern Balkans, Poland–Northwestern Ukraine as well as the Volga-Oka region in Russia, illustrating that this influx of new genetic material was not limited to certain regions but affected wide areas of Central and Eastern Europe, consistent with the rather simple, very similar archaeological horizons observed during the SP (Supplementary Figs. 10–12). To formally test whether these patterns observed from PCA are consistent with gene-flow events from the east into our study regions, we used F-statistics to quantify genetic affinities of SP individuals to preceding MP and succeeding present-day groups (Fig. 3a–c and Supplementary Tables 17–19). The divergence between pre-Slavic and Slavic-associated groups is verified both in the distribution of genetic distances (FST) (Supplementary Fig. 19) as well as shared alleles (F4) (Supplementary Figs. 17–20) (Supplementary Note 4.2). Both on the population and the individual scale, SP individuals from all three study regions uniformly show less genetic affinity to the preceding local populations than to ancient and present-day groups from Eastern Europe and Baltics (Supplementary Figs. 34–38 and Supplementary Notes 4.2 and 4.4.1).
Our results reveal that SP individuals display Baltic or Northeastern European-related ancestry that was previously absent in the three study regions. To quantitatively estimate this influx, we decomposed ancestral sources using a supervised clustering approach implemented in the software ADMIXTURE. Specifically, we assembled modern populations into 12 metapopulations that serve as proxies for the source ancestries in Central Europe (Methods and Supplementary Note 4.1). Applying our ancestry decomposition to the ancient genome-wide data, we find that (despite differences in the local trajectories) Northeastern European ancestry (BAL, represented by present-day individuals from Belarus, Lithuania and Latvia) was either completely absent or only a minor ancestry component throughout most of prehistory in our study transects (Fig. 3d and Extended Data Fig. 3), accounting for 6 ± 2%, 5 ± 1% and 7 ± 2% of the total MP ancestry in the Northwestern Balkans, Eastern Germany and Poland–Northwestern Ukraine, respectively. However, consistent with PCA (Supplementary Fig. 12a–c) and F4 statistics (Fig. 3a–c), BAL ancestry increased after 600 ce and became the largest ancestry component in all three study regions, reaching 47 ± 2%, 65 ± 1% and 63 ± 2%, respectively, during the SP. Outside our three study transects, we furthermore identify a major surge of BAL ancestry (from 0% in the MP to approximately 27%) in the Avar-associated population of Mödling, Austria, confirming an early arrival in the Pannonian Basin in the seventh century ce as reported by written sources, followed by substantial admixture with local groups (Extended Data Fig. 4). Only in Northwestern Russia do we detect a different trajectory: in the Volga-Oka area, the Slavic transition coincides with a significant decrease of BAL ancestry (from 65 ± 2% to 55 ± 7%), suggesting that the SP newcomers originated from a region further to the west of the Volga-Oka area where they incorporated additional ancestry not local to Eastern Europe.
The source for the incoming Northeastern European ancestry appears to be the same in all four regions. To showcase this shared descent, we applied ancIBD to identify segments that are identical by descent (IBD) that are shared between the MP and SP populations. We highlight that SP groups in Croatia, Eastern Germany and Poland–Ukraine share comparably large amounts of IBD with each other, despite the vast geographic distance between the three study regions, but share nearly no segments with the preceding populations (Fig. 3e and Supplementary Table 37). This IBD-sharing signal, including a large fraction of segments longer than 16 cM, clearly indicates that ancient individuals from Slavic-associated contexts descend from a common source population that migrated westwards and southwards at most a few generations earlier across Central Europe (Extended Data Figs. 5 and 10). Such evidence for large-scale population movement also explains the previously detected pattern of high levels of sharing of IBD between present-day pairs of individuals across Eastern Europe (Supplementary Fig. 56) and rejects the idea that this signal was caused predominantly by consistently low population densities.
To obtain a finer-scale characterization of genetic ancestries across space and time, we applied a hierarchical cluster detection approach to a network of around 2,500 individuals constructed from these pairwise IBD-sharing similarities (Supplementary Table 38 and Supplementary Note 4.3.2). We identify a large IBD-sharing community that contains most of our new and published SP individuals as well as multiple other contemporary samples from Central and Southeastern Europe. Within this larger cluster, we identify two distinct sub-communities: one primarily includes SP individuals from north of the Carpathian Mountains, whereas the other comprises individuals buried further south. This separation may reflect two geographically diverging waves of expansion or different patterns of incorporation of the local populations (Extended Data Fig. 5). Yet, at least sporadic gene flow from Eastern Europe into Pannonia and the Balkans must have already occurred during the Iron Age and Roman Period, as we identify a substantial number of individuals within the SP cluster buried in Austria, Hungary, Serbia and Montenegro during the time period from 500 bce to 300 ce, predating the large-scale population movements of the sixth and seventh centuries (Supplementary Figs. 29a, 32a,b and 33).
Using newly generated early medieval data from the Polish site Gródek, Hrubieszów County, near the Ukrainian border, which represents some of the oldest Slavic inhumation burials from Poland (dating between the seventh and ninth centuries ce), as a proximal source (both in time and space) for the incoming BAL-enriched ancestry, we calculate using qpAdm that approximately 82 ± 1%, 83 ± 6%, 93 ± 3% and 65 ± 4% of the local gene pool in the Northwestern Balkans, Eastern Germany, Poland–Northwestern Ukraine and the Volga-Oka valley, respectively, were replaced during the SP by migrants from Eastern Europe (referred here to as ‘SP ancestry’; Methods) (Fig. 4c and Supplementary Tables 47 and 51). These results contradict a model of substantial population continuation from the Iron Age or MP to the Middle Ages in present-day Western and Central Poland, where previous research claimed an autochthonous origin of the SP gene pool (Extended Data Fig. 4). Yet more samples are needed to assess the overall degree of genetic replacement over the larger area. Applying qpAdm to model present-day groups using ancient source populations, we show that Eastern European ancestry is the dominant genetic component in all Slavic-speaking populations today and is also found in neighbouring non-Slavic-speaking groups in Central Europe and regions bordering to the south (Extended Data Fig. 7 and Supplementary Note 5). We measure the highest proportions of Eastern European ancestry in present-day Ukraine, Belarus and Poland, from where it gradually decreases to the east and south (Extended Data Fig. 7 and Supplementary Table 41). Notably, we observe a profound duality to the west, in Eastern Germany, with the present-day German-speaking population from Saxony exhibiting around 40% SP ancestry and the Slavic-speaking Sorbs of Upper Lusatia (Saxony) exhibiting 88% SP ancestry (comparable to modern Poles) (Extended Data Fig. 7). This agrees with previous studies on the genetic isolation of the Sorbs and is consistent with them representing the descendants of these Slavic groups that were minimally (or at least less) integrated into the reproductive networks of the expanding German-speaking settlement east of Elbe and Saale from the twelfth century onwards. Conversely, we suggest that the German eastward expansion and earlier Frankish conquest is probably associated with the reduction in SP ancestry observed in the German-speaking population.
a, Contours indicate the averaged MOBEST maximum probability at search time 1,950 years before present for 20 individuals from Niederwünsch (denoting the mean prediction of the geographic regions where the ancestors of these individuals originated). This is supplemented by five lines of evidence: (A) ancient and present-day groups from the Baltics show the highest genetic similarity to SP individuals; (B) Bronze Age and Iron Age individuals from Estonia, Ingria and Karelia are less related to SP individuals than groups from Lithuania and Latvia; (C) populations in Western Russia feature too high proportions of Steppe and/or Siberian ancestry; (D) SP individuals are enriched in EEF and depleted in WHG ancestry compared with Bronze Age and Iron Age populations from the Baltics; (E) Putative migration directions inferred using pairwise mean sIBD sharing values between SP sites (n > 2). Made with Natural Earth. b, Comparison of linguistic split times (left) and genetic admixture dates in SP groups (right). Divergence date distributions for the Balto-Slavic and Slavic subgroups were extracted from a sample of 37,004 trees. Genetic admixture dates were obtained using DATES. Error bars indicate 2 × s.d. H, historical. c, Sex-biased admixture in four MP and four SP populations. Shown are non-local ancestry proportions on the autosomes, X chromosome and the Y chromosome (Y-chromosome haplogroups R1a, N and I2 for SP populations; E, G, J and T for MP populations). Points denote qpAdm (autosomes and X chromosome) or maximum likelihood (Y chromosome) estimates. Estimates were obtained as described in Supplementary Notes 7.2 using ancient source groups. The corresponding data can be found in Supplementary Table 47. Error bars indicate 2 × s.d. ♂ indicates an excess of non-local males in the admixture process; ♀ indicates a non-local female bias. The size of the symbols denotes the strength of the sex bias (with |z| > 2 being considered significant).
Both F4 and FST statistics identify the highest genetic similarity between SP individuals and present-day populations from the Baltics, Poland and Belarus (Supplementary Figs. 17–22). These are also the regions where BAL and SP ancestry (here approximated by medieval samples from Gródek) are maximized today and where the highest proportions of R1a haplotypes (specifically R1a-M458 and R1a-M558) are found among the male population. In patterns of haplotype sharing between the ancient and modern individuals, this similarity was mirrored in a distinctive IBD signal (Extended Data Fig. 6 and Supplementary Figs. 23–25): SP individuals from all three study regions share more and longer IBD fragments with Eastern Europeans than with any other Eurasian group, establishing direct genetic relatedness between present-day Balto-Slavic speakers and SP individuals in Central and Southeastern Europe (Extended Data Fig. 6 and Supplementary Note 4.3.1). This pattern of excess affinity to Northern and Northeastern Europeans is not only evident in the comparison with present-day data but also in the archaeogenetic record: Comparing the SP individuals to other ancient samples, we show that they, independent of their geographic origin, share the highest drift and largest sum of IBD with Bronze and Iron Age groups from Lithuania, Latvia and Estonia, and are (as shown by F-statistics) more closely related to these individuals than to any other population in post-Neolithic Europe (for IBD see Supplementary Fig. 32c; for F3 and F4 see Supplementary Figs. 34–38; Supplementary Note 4.4.1).
Yet, in contrast to the Bronze Age Baltic samples, we note that SP individuals from all study regions exhibit substantially less Western hunter-gatherer (WHG) and more early European farmer (EEF) ancestry (Supplementary Figs. 41 and 42a and Supplementary Note 4.4.2). This suggests that the SP groups in Central Europe were already admixed, most probably between a WHG and Steppe ancestry-enriched Baltic Bronze Age-related source from the sub-Neolithic forest zone and at least one EEF-enriched source from the south. Using qpAdm, we identify various groups in Southeastern and East-Central Europe that constitute working proxies for such an EEF-enriched donor, yet we are not able to precisely identify the most likely representative (Supplementary Fig. 43). Across all fitting two-way models (P > 0.01) (and most non-fitting), the admixture proportions are highly similar, with the Eastern German and Polish-Northwestern Ukrainian SP samples receiving around 71% Baltic (95% confidence interval: 66.5%–76%) and around 29% (95% confidence interval: 24%–33.5%) EEF-enriched ancestry (Supplementary Fig. 43b and Supplementary Table 34). However, we highlight that the demographic trajectories that led to the formation of the SP gene pool were potentially more complex than a simple two-way admixture event. Although we calculate similar estimates of Baltic Bronze Age-derived ancestry applying a non-negative least squares approach based on PCA- and ADMIXTURE results, mirroring previous results from genome-wide genealogies, all models profit from the inclusion of an additional Western European source (Supplementary Figs. 40, 45 and 46). Thus, which vector population(s) ultimately transmitted EEF-enriched ancestry to the Northeast cannot be resolved fully for now (Supplementary Note 4.4.2-3).
Assuming a two-way admixture process, using DATES (distribution of ancestry tracts of evolutionary signals) we obtained an average date of approximately 1000 bce for this admixture event that formed the SP gene pool (972 bce ± 250 for Niederwünsch and 906 bce ± 362 for Poland_EMA, respectively) (Fig. 4b and Supplementary Fig. 35). Of note, these DATES estimates overlap with the more recent part of the distribution of divergence estimates between Baltic and Slavic languages (Fig. 4b). Both phylogenetic analysis of cognate-coded basic vocabulary data (Fig. 4b and Extended Data Fig. 8) and most Indo-European linguists date the disintegration of Proto-Balto-Slavic to the second millennium bce. However, since the Bayesian linguistic estimates are on average shifted a few centuries older than the admixture estimates, we highlight the possibility that the admixing Baltic-related groups spoke a language that had already begun to diverge from the language or dialect continuum of the populations further north, the former eventually becoming the Slavic languages and the latter the (present-day) Baltic languages. To identify the most plausible geographic location for this initial formation of the SP gene pool, we applied MOBEST to perform spatiotemporal interpolation of the genetic affinities of SP individuals from the study regions to approximately 5,660 previously published ancient samples from Western Eurasia, obtaining similarity probabilities across Europe that can be interpreted as proxies for geographical origin at a specific time. We set the prediction time to 1,950 years before the present, providing us the most likely origin of an individual at this time point (and thus before the demographic transition in Central Europe). Averaging the probability surfaces, we infer a region spanning the south of Belarus and north of Ukraine as the best spatial proxy for the origin of the SP individuals in our three study transects (Fig. 4a and Supplementary Figs. 33 and 34). Such a range would agree well with the area where many linguists propose the earliest development of Slavic languages and archaeologists locate the origin of Slavic-associated material culture (Supplementary Fig. 44); however, more ancient DNA (aDNA) data are needed to conclusively assess the genetic landscape of this region.
From there, Northeastern European ancestry is likely to have spread east, west and south, admixing with or even replacing the local gene pools (Fig. 4a). Although we cannot precisely measure the onset of this expansion or its duration, we highlight that DATES estimates for admixture between local and immigrant ancestries in SP individuals are generally recent and similar across the study transects, consistent with admixture processes starting in the sixth and early seventh century and agreeing with historically recorded arrival dates of Slavic groups in these regions (Extended Data Fig. 1). The detection of substantial genetic introgression from the northeast into regions in which Slavic came to be spoken indicates that the diffusion of Slavic language and Eastern European-derived ancestry were related, although the degree of their overlap cannot be ascertained. This provides a plausible explanation for the high genetic relatedness across present-day Slavic-speaking groups, which was previously linked to the spread of the Slavic languages. However, we highlight that such a simplified model does not capture the more complex regional dynamics that emerge from historical and archaeological evidence, and are still evident in language boundaries that do not correspond to genetic differences across the Balkans and Central Europe. To investigate possible sex biases in these expansion and admixture processes, we compared estimates of SP-related ancestry on the X chromosomes and the autosomes to identify proportion differences indicative of male-biased admixture (Fig. 4c). Notably, we find no evidence for sex bias in any of the SP populations in Germany, Croatia, Poland or Russia (|z | <2; Fig. 4c and Supplementary Table 47). However, we observe that the previously undetected gene flow of Southern European-related ancestry into the MP population of Eastern Germany was significantly female-biased in most studied sites (Fig. 4c and Supplementary Fig. 59d).
The Slavic groups that we studied also showed fundamentally different social organization compared with the preceding MP population (Supplementary Note 8). Most notably, we highlight more intense inter-site and intra-site genetic relatedness in the Elbe-Saale region (Extended Data Fig. 2c), reflected by patrilineally organized pedigrees that comprise large numbers of individuals (Extended Data Figs. 9 and 10a,b). The cemeteries of the preceding MP in Eastern Germany were characterized by small units of biological relatedness, mostly consisting of fewer than four first- and second-degree relatives. At the site level, we identified for each individual an average of 1.16 ± 0.18 close relatives (here defined as all relationships up to third degree) (in Brücken specifically: 0.64 ± 0.14; Supplementary Fig. 62). This pattern is also mirrored in IBD sharing within sites (which also captures distant genetic relatedness greater than third degree), with the proportion of pairs of individuals that share any IBD larger than 12 cM ranging between 1 ± 0.4%, 6.8 ± 1.8% and 5.2 ± 1.8% in Brücken, Deersheim and Obermöllern, respectively.
By contrast, during the SP, we show that the number of close relatives at the sites increased nearly sixfold to 6.41 ± 0.4%. We even observe one case of seven offspring from the same couple (Extended Data Figs. 9 and 10b). Notably, four of the seven siblings had reached reproductive age, with three of them having offspring. As most of them were male, we can assume that several grown-up daughters might had gone elsewhere to marry. Moreover, the majority of offspring being male points towards additional unsampled female siblings (to statistically account for an equivalent number of females born). Notably, for all unions (in which at least one parent was identified on site), we find 52 sons (62% of the offspring; 95% confidence interval: 51–72%) but only 32 daughters (38% of the offspring; 95% confidence interval: 28–49%) (exact binomial test; P = 0.03753).
Although these extensive kinship networks evidenced a high degree of relatedness among all individuals within sites, we do not find a single case of close consanguinity (defined here as offspring of first cousin unions or closer) (Supplementary Fig. 54d). This shows profound knowledge of the lineages and deliberate avoidance of consanguinity. We also identify at least 11 cases of individuals reproducing with multiple partners, pointing to polygamy or serial monogamy. Despite a 2.7:1 ratio of half-siblings sharing the same father (95% confidence interval: 0.43–0.91) versus those sharing the same mother (95% confidence interval: 0.09–0.57), we do not find a single instance of levirate marriages as practiced in late Avar-period communities in the Carpathian Basin.
In parallel with the increase of genetic interconnectedness within the sites, we also observe that the organization of the cemeteries changed, reflecting in the spatial layout the extended pedigrees. Although close relatives were buried significantly closer together than non-related individuals in both the MP and the SP, only during the SP did cemeteries feature a significant correlation between genetic and spatial distances, suggesting that the cemeteries were planned and structured around these larger kin groups (Mantel statistic based on Spearman’s rank correlation; P = 0.0001 for both Niederwünsch and Steuden) (Extended Data Fig. 9 and Supplementary Figs. 63 and 65). This signal is most prominent in the site of Steuden, where at least 27% of all variance in spatial distances between graves is explained by genetic relatedness.
Although the sex ratio of adults across the SP sites is balanced (Exact binomial test; P = 1 for Steuden, P = 0.08794 for Niederwünsch), females have on average significantly fewer close relatives than males (Fisher’s exact test; P = 0.008) and feature overall an increased pairwise mismatch rate compared with males (Wilcoxon rank sum test; test statistic (W) = 13,976,553, P = 0.04218; Supplementary Fig. 67) and a lower sum of IBD segments shared within sites (Welch two-sample t-test; t = −3.707, d.f. = 355.82, P = 0.0002431) (Extended Data Fig. 10c). However, females share more IBD segments between different sites than males, thereby demonstrating higher inter-site relatedness among females in contrast to higher intra-site relatedness among males (Welch two-sample t-test; t = 2.9513, d.f. = 64.952, P = 0.004398) (Extended Data Fig. 10c).
This demonstrates that exogenous origin was more common for females than for males, suggesting a patrilocal inheritance organization, and agrees well with the nearly exclusively patrilineally organized pedigrees (for example, 88% patrilineal lineages (95% confidence interval: 68–97%) versus 12% matrilineal lineages (95% confidence interval: 4–32%)) and the significant underrepresentation of female offspring at the sites (Extended Data Figs. 9 and 10b and Aupplementary Fig. 54). Across all SP sites in Eastern Germany, we find only one instance of mitochondrial haplogroups being transmitted further than one daughter generation. This contrasts with the preceding MP population, for which we detect no difference in the number of close relatives between males and females (Fisher’s exact test; P = 0.74) and no difference in pairwise mismatch rate in general (Wilcoxon rank sum test; W = 3,215,752, P = 0.06), potentially suggesting a less strictly patrilineal social system before the arrival of Slavic groups (Supplementary Fig. 67). However, owing to the overall smaller number of identified relatives, these tests might be less statistically conclusive and underestimate signals of MP female exogamy and patrilineal practices.
Patterns of patrilocal organization into kin groups are broadly similar across regions and might have contributed to the previously described turnover and homogenization of the paternal gene pool (Extended Data Fig. 10b). In particular, we find an identical pattern of correlation of spatial and genetic distances in SP Velim, Croatia (Mantel statistic based on Spearman’s rank correlation; P = 0.0001; Supplementary Figs. 64 and 65d) as well as evidence for patrilocality and female exogamy (for example, more close biological relatives among males than among females (Fisher’s exact test; P = 0.027) and higher pairwise mismatch rates among females compared to males (Wilcoxon rank sum; W = 156,550, P = 0.02377; Supplementary Fig. 67g–i)), mirroring the social stratification observed in the Elbe-Saale region. By contrast, at Velim the mean number of close relatives in the site is significantly lower than in Niederwünsch or Steuden (1.01 ± 0.17). Within a shared pattern of patrilocality, SP fine-scale organization differed substantially across Central Europe owing to the complex, regionally contingent nature of this expansion. Rather than simple replacement, partial integration of the local population was probably dependent on the fortunes of specific groups or families. However, the substantial number of individuals from sites across Eastern Germany, Croatia and Poland–Northwestern Ukraine who share comparably large amounts of IBD with each other confirms that these Slavic-associated groups were closely linked as the result of a shared biological origin and recent geographical expansion.
In modern ethnic and national terminology, ‘Slavs’ denotes all speakers of Slavic languages and/or citizens of the Slavic nation states. This concept of an ethnic collective spanning several nations is much more marked than among the Germanic or Romance speakers in the rest of Europe (Supplementary Note 1). The extent to which Slavic identity mattered diverged; it was important for nineteenth and twentieth century Slavic nationalisms, Pan-Slavism and Russian imperialism, but regional or national allegiances often carried more weight. The prejudices of their western neighbours who tended to regard Slavs as culturally inferior reinforced sentiments of Slavic commonality. The question of Slavic origins, addressed in this Article, had a crucial role in ideological debates about the unity and the significance of the Slavs. It is therefore important to be precise in the scholarly use of the term. In research about the early Slavs, the meanings of the term diverge. In written sources since the sixth and seventh century in Byzantium and the west, groups of Slavs or Wends increasingly appear in a wide range of lands beyond and along the Danube and the Elbe rivers. We can make use of different sources to understand how large parts of Europe became Slavic: outside perceptions about Slavs in texts; archaeological traces of shared cultural practices among early Slavs (particularly the Prague-Korchak culture); linguistic reconstructions of a common Slavic language prior to the particular Slavic idioms; and shifts in ancestry of the medieval gene pool, which point to migrations. We should not take these disciplinary results as proxies for each other as attributes of a coherent people called Slavs; yet, they provide different perspectives on the Slavicisation of Europe during the Early Middle Ages. Combining them allows us to overcome simplistic theories of an expansion of the Slavs and instead understand the common dynamic and the different ways in which Slavic peoples began to form in many parts of Europe. We therefore use Slavs for populations named in this way in contemporary texts, without implying that they self-identified as such. These Slavic groups can be localized, but hardly circumscribed. We do not use genetic or archaeological features in regions where Slavs spread to distinguish between Slavs and non-Slavs, or between speakers and non-speakers of Slavic languages, although we assume that these phenomena overlapped to a considerable degree.
Within tables and figures, we refer to groups of published individuals by the names given in the Allen Ancient DNA resource v.54.1. Sample sizes, context information and publication names can be found in Supplementary Tables 7–11. In the main text and Supplementary Notes, we used the following abbreviations for archaeological time periods: N, Neolithic; C, Chalcolithic; EBA, Early Bronze Age; MBA, Middle Bronze Age; LBA, Late Bronze Age; IA, Iron Age; RA, Roman Age; EMA, Early Middle Ages; MA, Middle Ages; H, historical.
Provenance information for samples from all archaeological sites is given in Supplementary Information Note 2, together with short descriptions of each site, the institution owning the samples (or custodians of the samples), the responsible co-author who obtained permission to analyse and the year of the permission granted.
We merged our ancient DNA data with previously published datasets of ancient individuals reported by the Reich Lab in the Allen Ancient DNA Resource v.54.1 (https://reich.hms.harvard.edu/allen-ancient-dna-resource-aadr-downloadable-genotypes-present-day-and-ancient-dna-data) (1240k SNP panel). For comparisons with present-day groups, we compiled and curated a high-resolution, quality-filtered reference dataset containing genotypes for 426,135 SNPs (the intersection of several different Affymetrix and Illumina chip types) from 12,176 contemporary individuals sampled from 49 (mostly European and West Asian) populations from previously published datasets as described in Gretzinger et al.. Sample sizes are given in Supplementary Table 3. We produced four different datasets:
We applied ADMIXTURE in supervised mode using modern reference populations at K = 12. This analysis was run on haploid data with the parameter “haploid” set to all (=“*”). To obtain point estimates for populations, we averaged individual point estimates and calculated the s.e.m. As modern references we used the groupings listed in the Supplementary Notes 4.1. The Q matrix of this ADMIXTURE analysis was also used as input for FSTruct as described by the authors.
We use Bayesian phylogenetic inference to estimate the ages of language divergence as described. Analyses were performed on the IE-CoR database that stores data on cognate relationships (shared word origin) between 161 Indo-European languages, in a reference set of 170 basic meanings (https://iecor.clld.org). Divergence date distributions for the Balto-Slavic and Slavic subgroups were extracted from the sample of 37,004 trees resulting from the main analysis of Heggarty et al.. A critical evaluation of this approach is discussed in Supplementary Note 9. However, we highlight that these computational estimates for the splits of the Baltic and Slavic languages compare well with estimates produced by methods of traditional Indo-European linguistics. These estimates are based on the reconstructed rapidity of diversification relative to that observed in the other branches of Indo-European (some of which is recorded in writing thousands of years before Baltic and Slavic), the nature of contacts and convergence between Balto-Slavic and other languages (IE and non-IE), and the character of the reconstructible Proto-Balto–Slavic vocabulary—but also attempted links to the archaeological picture. The dating of the split of Proto-Balto–Slavic is generally in agreement with the results of traditional historical linguistics as described above; and earlier glottochronological approaches have yielded similar dating—for example, an estimation at the fifteenth and fourteenth century bce.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Sections
"[{\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Sec8\", \"MOESM1\", \"Fig5\", \"MOESM1\"], \"section\": \"Main\", \"text\": \"The term Slavs first appears as an ethnonym in the course of the sixth century in Constantinople and later in the west (Box 1 and Supplementary Note\\u00a01.1). Written sources locate them initially north of the Lower Danube, and later in the Carpathian Basin, the Balkans and the Eastern Alps (Extended Data Fig. 1). Many came under the rule of the Avar steppe empire along the Middle Danube (567 ce to around 800 ce). In the seventh century, there is evidence for the presence of Slavs in much of East-Central and Southeastern Europe. Where Slavs lived, Roman, Germanic and other pre-Slavic infrastructures were usually replaced by rather simple ways of life, archaeologically characterized by small settlements of pit houses, cremation burials, handmade, undecorated pottery and modest, low-metal material culture, known as the Prague-Korchak group. (Supplementary Note\\u00a01.2). More complex social systems and regional rulership developed later in the contact zones with Byzantium and the Christian west.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\", \"MOESM1\"], \"section\": \"Main\", \"text\": \"The similarity of early Slavic cultures was often attributed to a swift spread of Slavs from Northeast of the Carpathians, although debates continue, not only about their geographical origin (Supplementary Note\\u00a01,1). In Poland, the non-native (allochthonist) view assumes Slavic origin from Ukraine\\u2013Belarus, whereas the native (autochthonist) concept asserts that their ancestors inhabited Polish territory since the Bronze Age. Some scholars doubt Slavic expansion by migrations and assume that there was \\u2018Slavicisation\\u2019 of existing populations (Supplementary Note\\u00a01,2). Previous modern and ancient DNA studies have supported gene flow into the Northern Balkans and the Russian Volga-Oka region, but also argued for population continuity in Poland, so that the scale and sequence of these movements and their association with \\u2018Slavic\\u2019 material culture has remained unclear. Eventually, this cultural transformation led to the replacement of Germanic and other languages in East-Central and Southeastern Europe and the introduction of Slavic languages, which today represent the largest linguistic group in Europe. Yet, this presumed joint spread of language and material culture is difficult to trace, given that the first longer texts in Slavic were written in the late ninth century.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM3\", \"MOESM3\", \"Fig6\", \"MOESM3\", \"MOESM3\", \"Fig1\", \"MOESM3\", \"MOESM3\", \"MOESM3\"], \"section\": \"New ancient DNA data\", \"text\": \"We selected skeletal remains from 591 ancient individuals from 26 different sites from Central and Eastern Europe (Supplementary Tables 1 and 2), creating, in combination with previously published data, a dense sampling transect for three regions: (1) Elbe-Saale Region in Eastern Germany as the main study area; (2) the Northwestern Balkans; and (3) Poland\\u2013Northwestern Ukraine (Extended Data Fig. 2 and Supplementary Tables 7\\u201310). Complementary to these three transects, we generated new data and collected published data from the Baltics and Northwestern Russia to form a reference transect in the east. After hybridization DNA capture and quality filtering (Methods), genome-wide data for 555 unique individuals with a median coverage of 538k single nucleotide polymorphisms (SNPs) (on 1240k data) were available for analysis, including 359 ancient individuals from the SP, as well as 205 individuals predating the cultural transformations connected to the emergence of the Slavs (Fig. 1 and Supplementary Table 1). We analyse the ancient genome-wide data (Supplementary Table 11) together with an extended dataset of more than 11,500 present-day Europeans (Supplementary Table 3), covering all major Slavic-speaking groups, including data from more than 600 individuals belonging to the Sorbian minority in Eastern Germany.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig2\", \"Fig7\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"Fig2\", \"MOESM1\"], \"section\": \"Genetic shifts in Central Europe\", \"text\": \"To visualize genome-wide ancestry diversity before and after the spread of Slavic groups, we performed principal component analysis (PCA) on 10,528 present-day Europeans and projected our newly reported and other relevant ancient genome-wide data onto their genetic variation (Fig. 2). When comparing the SP samples to earlier and present-day data from our three study regions, we observe that the genetic composition within the transects changed markedly between about 600 and 800 ce. In general, the Roman and MP samples that predate the arrival of Slavic groups show high genetic heterogeneity in PCA space, with most samples from Germany and Poland clustering with present-day continental Northern German, Dutch and Scandinavian populations (Extended Data Fig. 3 and Supplementary Figs. 6, 7, 10 and 12), whereas the Roman and MP individuals from Croatia cluster with present-day Italian and Eastern Mediterranean populations (Fig. 2c and Supplementary Note\\u00a03).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig7\"], \"section\": \"Genetic shifts in Central Europe\", \"text\": \"In Eastern Germany and the Northwestern Balkans, most of the genetic diversity within the Roman and MP clusters follows a north\\u2013south cline along PC1. For the Northwestern Balkans, this heterogeneity has been attributed to increasing Eastern Mediterranean ancestry that arrived subsequently to the incorporation of the region into the Roman Empire. More unexpectedly, we detect a high number of MP individuals with non-local, Southern European ancestry in the Elbe-Saale region of Eastern Germany, although this area was never part of the Roman Empire. Using qpAdm, we measure on average between approximately 15% and 25% of Southern European ancestry in all 4 MP sites of the region (Extended Data Fig. 3).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig7\", \"MOESM1\", \"MOESM1\", \"MOESM3\", \"MOESM1\", \"MOESM1\"], \"section\": \"Genetic shifts in Central Europe\", \"text\": \"Both PCA-based MOBEST analysis and F4 statistics indicate that this non-local ancestry was most probably derived from contemporaneous source populations in Italy and/or the Northern Balkan Peninsula (or other areas of the Roman Empire where people of this ancestry were located) (Extended Data Fig. 3 and Supplementary Figs. 3\\u20135). Previous studies already identified mixed communities of northern and southern ancestry in Hungary and Northern Italy that were interpreted as amalgamation between Northern European newcomers and the local romanized population. In contrast to these earlier results, we do not find evidence that the two different ancestries were correlated to differences in material culture (Supplementary Table 4). Applying a generalized linear model, we demonstrate that neither the presence of grave goods overall, nor certain types of artefacts (such as weapons or brooches) are significantly correlated with either PCA position or ADMIXTURE profiles (Supplementary Fig. 60). Instead, we find the only significant (P\\u2009<\\u20090.05) correlation between ancestry and material culture among the burial constructions, where we show that individuals buried in pits feature on average higher Northern European ancestry (Supplementary Fig. 60c). The spatial organization of the burials was also not determined by similarity in ancestry. Instead, we observe that individuals were buried close to their biological relatives, within small kin groups composed of individuals with Northern European, Southern European or mixed ancestry, reflecting a high degree of admixture between individuals with different ancestry backgrounds during the MP. Consequently, our data from Eastern Germany demonstrate that the cosmopolitan character of the Roman Empire not only affected the incorporated territories but also facilitated exchange and mobility along its borders and beyond into barbarian lands (Barbaricum), resulting in an unprecedented genetic diversity in Central Europe during and, in the case of Eastern Germany, even after its existence. Although the causes and circumstances of their movement to the Elbe-Saale region remain open for speculation, these newcomers apparently adapted the fashions and traditions of the local populations, resulting in a rather homogenous material culture within a group of individuals with diverse genetic backgrounds.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\", \"Fig2\", \"MOESM1\", \"MOESM1\", \"Fig3\", \"MOESM3\", \"MOESM3\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\"], \"section\": \"Genetic shifts in Central Europe\", \"text\": \"However, this diversity had collapsed in the subsequent SP (Supplementary Note\\u00a06). In contrast to the preceding MP, the genetic profile of Eastern Germany during the SP has shifted considerably and clusters nearly exclusively with present-day Slavic-speaking populations (for example, Poles and Belarussians), indicative of a fundamental replacement of genetic ancestry (Fig. 2b,c). A similar pattern is seen in the Northwestern Balkans, Poland\\u2013Northwestern Ukraine as well as the Volga-Oka region in Russia, illustrating that this influx of new genetic material was not limited to certain regions but affected wide areas of Central and Eastern Europe, consistent with the rather simple, very similar archaeological horizons observed during the SP (Supplementary Figs. 10\\u201312). To formally test whether these patterns observed from PCA are consistent with gene-flow events from the east into our study regions, we used F-statistics to quantify genetic affinities of SP individuals to preceding MP and succeeding present-day groups (Fig. 3a\\u2013c and Supplementary Tables 17\\u201319). The divergence between pre-Slavic and Slavic-associated groups is verified both in the distribution of genetic distances (FST) (Supplementary Fig. 19) as well as shared alleles (F4) (Supplementary Figs. 17\\u201320) (Supplementary Note\\u00a04.2). Both on the population and the individual scale, SP individuals from all three study regions uniformly show less genetic affinity to the preceding local populations than to ancient and present-day groups from Eastern Europe and Baltics (Supplementary Figs. 34\\u201338 and Supplementary Notes\\u00a04.2 and 4.4.1).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\", \"Fig3\", \"Fig7\", \"MOESM1\", \"Fig3\", \"Fig8\"], \"section\": \"Spread of SP ancestry across Europe\", \"text\": \"Our results reveal that SP individuals display Baltic or Northeastern European-related ancestry that was previously absent in the three study regions. To quantitatively estimate this influx, we decomposed ancestral sources using a supervised clustering approach implemented in the software ADMIXTURE. Specifically, we assembled modern populations into 12 metapopulations that serve as proxies for the source ancestries in Central Europe (Methods and Supplementary Note\\u00a04.1). Applying our ancestry decomposition to the ancient genome-wide data, we find that (despite differences in the local trajectories) Northeastern European ancestry (BAL, represented by present-day individuals from Belarus, Lithuania and Latvia) was either completely absent or only a minor ancestry component throughout most of prehistory in our study transects (Fig. 3d and Extended Data Fig. 3), accounting for 6\\u2009\\u00b1\\u20092%, 5\\u2009\\u00b1\\u20091% and 7\\u2009\\u00b1\\u20092% of the total MP ancestry in the Northwestern Balkans, Eastern Germany and Poland\\u2013Northwestern Ukraine, respectively. However, consistent with PCA (Supplementary Fig. 12a\\u2013c) and F4 statistics (Fig. 3a\\u2013c), BAL ancestry increased after 600 ce and became the largest ancestry component in all three study regions, reaching 47\\u2009\\u00b1\\u20092%, 65\\u2009\\u00b1\\u20091% and 63\\u2009\\u00b1\\u20092%, respectively, during the SP. Outside our three study transects, we furthermore identify a major surge of BAL ancestry (from 0% in the MP to approximately 27%) in the Avar-associated population of M\\u00f6dling, Austria, confirming an early arrival in the Pannonian Basin in the seventh century ce as reported by written sources, followed by substantial admixture with local groups (Extended Data Fig. 4). Only in Northwestern Russia do we detect a different trajectory: in the Volga-Oka area, the Slavic transition coincides with a significant decrease of BAL ancestry (from 65\\u2009\\u00b1\\u20092% to 55\\u2009\\u00b1\\u20097%), suggesting that the SP newcomers originated from a region further to the west of the Volga-Oka area where they incorporated additional ancestry not local to Eastern Europe.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig3\", \"MOESM3\", \"Fig9\", \"Fig14\", \"MOESM1\"], \"section\": \"Spread of SP ancestry across Europe\", \"text\": \"The source for the incoming Northeastern European ancestry appears to be the same in all four regions. To showcase this shared descent, we applied ancIBD to identify segments that are identical by descent (IBD) that are shared between the MP and SP populations. We highlight that SP groups in Croatia, Eastern Germany and Poland\\u2013Ukraine share comparably large amounts of IBD with each other, despite the vast geographic distance between the three study regions, but share nearly no segments with the preceding populations (Fig. 3e and Supplementary Table 37). This IBD-sharing signal, including a large fraction of segments longer than 16\\u2009cM, clearly indicates that ancient individuals from Slavic-associated contexts descend from a common source population that migrated westwards and southwards at most a few generations earlier across Central Europe (Extended Data Figs. 5 and 10). Such evidence for large-scale population movement also explains the previously detected pattern of high levels of sharing of IBD between present-day pairs of individuals across Eastern Europe (Supplementary Fig. 56) and rejects the idea that this signal was caused predominantly by consistently low population densities.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM3\", \"MOESM1\", \"Fig9\", \"MOESM1\", \"MOESM1\", \"MOESM1\"], \"section\": \"Spread of SP ancestry across Europe\", \"text\": \"To obtain a finer-scale characterization of genetic ancestries across space and time, we applied a hierarchical cluster detection approach to a network of around 2,500 individuals constructed from these pairwise IBD-sharing similarities (Supplementary Table 38 and Supplementary Note\\u00a04.3.2). We identify a large IBD-sharing community that contains most of our new and published SP individuals as well as multiple other contemporary samples from Central and Southeastern Europe. Within this larger cluster, we identify two distinct sub-communities: one primarily includes SP individuals from north of the Carpathian Mountains, whereas the other comprises individuals buried further south. This separation may reflect two geographically diverging waves of expansion or different patterns of incorporation of the local populations (Extended Data Fig. 5). Yet, at least sporadic gene flow from Eastern Europe into Pannonia and the Balkans must have already occurred during the Iron Age and Roman Period, as we identify a substantial number of individuals within the SP cluster buried in Austria, Hungary, Serbia and Montenegro during the time period from 500 bce to 300 ce, predating the large-scale population movements of the sixth and seventh centuries (Supplementary Figs. 29a, 32a,b and 33).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Sec9\", \"Fig4\", \"MOESM3\", \"MOESM3\", \"Fig8\", \"Fig11\", \"MOESM1\", \"Fig11\", \"MOESM3\", \"Fig11\"], \"section\": \"Spread of SP ancestry across Europe\", \"text\": \"Using newly generated early medieval data from the Polish site Gr\\u00f3dek, Hrubiesz\\u00f3w County, near the Ukrainian border, which represents some of the oldest Slavic inhumation burials from Poland (dating between the seventh and ninth centuries ce), as a proximal source (both in time and space) for the incoming BAL-enriched ancestry, we calculate using qpAdm that approximately 82\\u2009\\u00b1\\u20091%, 83\\u2009\\u00b1\\u20096%, 93\\u2009\\u00b1\\u20093% and 65\\u2009\\u00b1\\u20094% of the local gene pool in the Northwestern Balkans, Eastern Germany, Poland\\u2013Northwestern Ukraine and the Volga-Oka valley, respectively, were replaced during the SP by migrants from Eastern Europe (referred here to as \\u2018SP ancestry\\u2019;\\u00a0Methods) (Fig. 4c and Supplementary Tables 47 and 51). These results contradict a model of substantial population continuation from the Iron Age or MP to the Middle Ages in present-day Western and Central Poland, where previous research claimed an autochthonous origin of the SP gene pool (Extended Data Fig. 4). Yet more samples are needed to assess the overall degree of genetic replacement over the larger area. Applying qpAdm to model present-day groups using ancient source populations, we show that Eastern European ancestry is the dominant genetic component in all Slavic-speaking populations today and is also found in neighbouring non-Slavic-speaking groups in Central Europe and regions bordering to the south (Extended Data Fig. 7 and Supplementary Note\\u00a05). We measure the highest proportions of Eastern European ancestry in present-day Ukraine, Belarus and Poland, from where it gradually decreases to the east and south (Extended Data Fig. 7 and Supplementary Table 41). Notably, we observe a profound duality to the west, in Eastern Germany, with the present-day German-speaking population from Saxony exhibiting around 40% SP ancestry and the Slavic-speaking Sorbs of Upper Lusatia (Saxony) exhibiting 88% SP ancestry (comparable to modern Poles) (Extended Data Fig. 7). This agrees with previous studies on the genetic isolation of the Sorbs and is consistent with them representing the descendants of these Slavic groups that were minimally (or at least less) integrated into the reproductive networks of the expanding German-speaking settlement east of Elbe and Saale from the twelfth century onwards. Conversely, we suggest that the German eastward expansion and earlier Frankish conquest is probably associated with the reduction in SP ancestry observed in the German-speaking population.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\", \"MOESM3\"], \"section\": \"Formation of the SP gene pool.\", \"text\": \"a, Contours indicate the averaged MOBEST maximum probability at search time 1,950 years before present for 20 individuals from Niederw\\u00fcnsch (denoting the mean prediction of the geographic regions where the ancestors of these individuals originated). This is supplemented by five lines of evidence: (A) ancient and present-day groups from the Baltics show the highest genetic similarity to SP individuals; (B) Bronze Age and Iron Age individuals from Estonia, Ingria and Karelia are less related to SP individuals than groups from Lithuania and Latvia; (C) populations in Western Russia feature too high proportions of Steppe and/or Siberian ancestry; (D) SP individuals are enriched in EEF and depleted in WHG ancestry compared with Bronze Age and Iron Age populations from the Baltics; (E) Putative migration directions inferred using pairwise mean sIBD sharing values between SP sites (n\\u2009>\\u20092). Made with Natural Earth. b, Comparison of linguistic split times (left) and genetic admixture dates in SP groups (right). Divergence date distributions for the Balto-Slavic and Slavic subgroups were extracted from a sample of 37,004 trees. Genetic admixture dates were obtained using DATES. Error bars indicate 2\\u2009\\u00d7\\u2009s.d.\\u00a0H, historical. c, Sex-biased admixture in four MP and four SP populations. Shown are non-local ancestry proportions on the autosomes, X chromosome and the Y chromosome (Y-chromosome haplogroups R1a, N and I2 for SP populations; E, G, J and T for MP populations). Points denote qpAdm (autosomes and X chromosome) or maximum likelihood (Y chromosome) estimates. Estimates were obtained as described in Supplementary Notes\\u00a07.2 using ancient source groups. The corresponding data can be found in Supplementary Table 47. Error bars indicate 2\\u2009\\u00d7\\u2009s.d. \\u2642 indicates an excess of non-local males in the admixture process; \\u2640 indicates a non-local female bias. The size of the symbols denotes the strength of the sex bias (with |z|\\u2009>\\u20092 being considered significant).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\", \"MOESM1\", \"Fig10\", \"MOESM1\", \"MOESM1\", \"Fig10\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\"], \"section\": \"Formation and origin of SP ancestry\", \"text\": \"Both F4 and FST statistics identify the highest genetic similarity between SP individuals and present-day populations from the Baltics, Poland and Belarus (Supplementary Figs. 17\\u201322). These are also the regions where BAL and SP ancestry (here approximated by medieval samples from Gr\\u00f3dek) are maximized today and where the highest proportions of R1a haplotypes (specifically R1a-M458 and R1a-M558) are found among the male population. In patterns of haplotype sharing between the ancient and modern individuals, this similarity was mirrored in a distinctive IBD signal (Extended Data Fig. 6 and Supplementary Figs. 23\\u201325): SP individuals from all three study regions share more and longer IBD fragments with Eastern Europeans than with any other Eurasian group, establishing direct genetic relatedness between present-day Balto-Slavic speakers and SP individuals in Central and Southeastern Europe (Extended Data Fig. 6 and Supplementary Note\\u00a04.3.1). This pattern of excess affinity to Northern and Northeastern Europeans is not only evident in the comparison with present-day data but also in the archaeogenetic record: Comparing the SP individuals to other ancient samples, we show that they, independent of their geographic origin, share the highest drift and largest sum of IBD with Bronze and Iron Age groups from Lithuania, Latvia and Estonia, and are (as shown by F-statistics) more closely related to these individuals than to any other population in post-Neolithic Europe (for IBD see Supplementary Fig. 32c; for F3 and F4 see Supplementary Figs. 34\\u201338; Supplementary Note\\u00a04.4.1).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM3\", \"MOESM1\", \"MOESM1\", \"MOESM1\", \"MOESM1\"], \"section\": \"Formation and origin of SP ancestry\", \"text\": \"Yet, in contrast to the Bronze Age Baltic samples, we note that SP individuals from all study regions exhibit substantially less Western hunter-gatherer (WHG) and more early European farmer (EEF) ancestry (Supplementary Figs. 41 and 42a and Supplementary Note\\u00a04.4.2). This suggests that the SP groups in Central Europe were already admixed, most probably between a WHG and Steppe ancestry-enriched Baltic Bronze Age-related source from the sub-Neolithic forest zone and at least one EEF-enriched source from the south. Using qpAdm, we identify various groups in Southeastern and East-Central Europe that constitute working proxies for such an EEF-enriched donor, yet we are not able to precisely identify the most likely representative (Supplementary Fig. 43). Across all fitting two-way models (P\\u2009>\\u20090.01) (and most non-fitting), the admixture proportions are highly similar, with the Eastern German and Polish-Northwestern Ukrainian SP samples receiving around 71% Baltic (95% confidence interval: 66.5%\\u201376%) and around 29% (95% confidence interval: 24%\\u201333.5%) EEF-enriched ancestry (Supplementary Fig. 43b and Supplementary Table 34). However, we highlight that the demographic trajectories that led to the formation of the SP gene pool were potentially more complex than a simple two-way admixture event. Although we calculate similar estimates of Baltic Bronze Age-derived ancestry applying a non-negative least squares approach based on PCA- and ADMIXTURE results, mirroring previous results from genome-wide genealogies, all models profit from the inclusion of an additional Western European source (Supplementary Figs. 40, 45 and 46). Thus, which vector population(s) ultimately transmitted EEF-enriched ancestry to the Northeast cannot be resolved fully for now (Supplementary Note\\u00a04.4.2-3).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig4\", \"MOESM1\", \"Fig4\", \"Fig4\", \"Fig12\", \"Fig4\", \"MOESM1\", \"MOESM1\", \"MOESM1\"], \"section\": \"Formation and origin of SP ancestry\", \"text\": \"Assuming a two-way admixture process, using DATES (distribution of ancestry tracts of evolutionary signals) we obtained an average date of approximately 1000 bce for this admixture event that formed the SP gene pool (972 bce\\u2009\\u00b1\\u2009250 for Niederw\\u00fcnsch and 906 bce\\u2009\\u00b1\\u2009362 for Poland_EMA, respectively) (Fig. 4b and Supplementary Fig. 35). Of note, these DATES estimates overlap with the more recent part of the distribution of divergence estimates between Baltic and Slavic languages (Fig. 4b). Both phylogenetic analysis of cognate-coded basic vocabulary data (Fig. 4b and Extended Data Fig. 8) and most Indo-European linguists date the disintegration of Proto-Balto-Slavic to the second millennium bce. However, since the Bayesian linguistic estimates are on average shifted a few centuries older than the admixture estimates, we highlight the possibility that the admixing Baltic-related groups spoke a language that had already begun to diverge from the language or dialect continuum of the populations further north, the former eventually becoming the Slavic languages and the latter the (present-day) Baltic languages. To identify the most plausible geographic location for this initial formation of the SP gene pool, we applied MOBEST to perform spatiotemporal interpolation of the genetic affinities of SP individuals from the study regions to approximately 5,660 previously published ancient samples from Western Eurasia, obtaining similarity probabilities across Europe that can be interpreted as proxies for geographical origin at a specific time. We set the prediction time to 1,950 years before the present, providing us the most likely origin of an individual at this time point (and thus before the demographic transition in Central Europe). Averaging the probability surfaces, we infer a region spanning the south of Belarus and north of Ukraine as the best spatial proxy for the origin of the SP individuals in our three study transects (Fig. 4a and Supplementary Figs. 33 and 34). Such a range would agree well with the area where many linguists propose the earliest development of Slavic languages and archaeologists locate the origin of Slavic-associated material culture (Supplementary Fig. 44); however, more ancient DNA (aDNA) data are needed to conclusively assess the genetic landscape of this region.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig4\", \"Fig5\", \"Fig4\", \"Fig4\", \"MOESM3\", \"Fig4\", \"MOESM1\"], \"section\": \"Formation and origin of SP ancestry\", \"text\": \"From there, Northeastern European ancestry is likely to have spread east, west and south, admixing with or even replacing the local gene pools (Fig. 4a). Although we cannot precisely measure the onset of this expansion or its duration, we highlight that DATES estimates for admixture between local and immigrant ancestries in SP individuals are generally recent and similar across the study transects, consistent with admixture processes starting in the sixth and early seventh century and agreeing with historically recorded arrival dates of Slavic groups in these regions (Extended Data Fig. 1). The detection of substantial genetic introgression from the northeast into regions in which Slavic came to be spoken indicates that the diffusion of Slavic language and Eastern European-derived ancestry were related, although the degree of their overlap cannot be ascertained. This provides a plausible explanation for the high genetic relatedness across present-day Slavic-speaking groups, which was previously linked to the spread of the Slavic languages. However, we highlight that such a simplified model does not capture the more complex regional dynamics that emerge from historical and archaeological evidence, and are still evident in language boundaries that do not correspond to genetic differences across the Balkans and Central Europe. To investigate possible sex biases in these expansion and admixture processes, we compared estimates of SP-related ancestry on the X chromosomes and the autosomes to identify proportion differences indicative of male-biased admixture (Fig. 4c). Notably, we find no evidence for sex bias in any of the SP populations in Germany, Croatia, Poland or Russia (|z\\u2009|\\u2009<2; Fig. 4c and Supplementary Table 47). However, we observe that the previously undetected gene flow of Southern European-related ancestry into the MP population of Eastern Germany was significantly female-biased in most studied sites (Fig. 4c and Supplementary Fig. 59d).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\", \"Fig6\", \"Fig13\", \"Fig14\", \"MOESM1\"], \"section\": \"Social changes in Eastern Germany\", \"text\": \"The Slavic groups that we studied also showed fundamentally different social organization compared with the preceding MP population (Supplementary Note\\u00a08). Most notably, we highlight more intense inter-site and intra-site genetic relatedness in the Elbe-Saale region (Extended Data Fig. 2c), reflected by patrilineally organized pedigrees that comprise large numbers of individuals (Extended Data Figs. 9 and 10a,b). The cemeteries of the preceding MP in Eastern Germany were characterized by small units of biological relatedness, mostly consisting of fewer than four first- and second-degree relatives. At the site level, we identified for each individual an average of 1.16\\u2009\\u00b1\\u20090.18 close relatives (here defined as all relationships up to third degree) (in Br\\u00fccken specifically: 0.64\\u2009\\u00b1\\u20090.14; Supplementary Fig. 62). This pattern is also mirrored in IBD sharing within sites (which also captures distant genetic relatedness greater than third degree), with the proportion of pairs of individuals that share any IBD larger than 12\\u2009cM ranging between 1\\u2009\\u00b1\\u20090.4%, 6.8\\u2009\\u00b1\\u20091.8% and 5.2\\u2009\\u00b1\\u20091.8% in Br\\u00fccken, Deersheim and Oberm\\u00f6llern, respectively.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig13\", \"Fig14\"], \"section\": \"Social changes in Eastern Germany\", \"text\": \"By contrast, during the SP, we show that the number of close relatives at the sites increased nearly sixfold to 6.41\\u2009\\u00b1\\u20090.4%. We even observe one case of seven offspring from the same couple (Extended Data Figs. 9 and 10b). Notably, four of the seven siblings had reached reproductive age, with three of them having offspring. As most of them were male, we can assume that several grown-up daughters might had gone elsewhere to marry. Moreover, the majority of offspring being male points towards additional unsampled female siblings (to statistically account for an equivalent number of females born). Notably, for all unions (in which at least one parent was identified on site), we find 52 sons (62% of the offspring; 95% confidence interval: 51\\u201372%) but only 32 daughters (38% of the offspring; 95% confidence interval: 28\\u201349%) (exact binomial test; P\\u2009=\\u20090.03753).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\"], \"section\": \"Social changes in Eastern Germany\", \"text\": \"Although these extensive kinship networks evidenced a high degree of relatedness among all individuals within sites, we do not find a single case of close consanguinity (defined here as offspring of first cousin unions or closer) (Supplementary Fig. 54d). This shows profound knowledge of the lineages and deliberate avoidance of consanguinity. We also identify at least 11 cases of individuals reproducing with multiple partners, pointing to polygamy or serial monogamy. Despite a 2.7:1 ratio of half-siblings sharing the same father (95% confidence interval: 0.43\\u20130.91) versus those sharing the same mother (95% confidence interval: 0.09\\u20130.57), we do not find a single instance of levirate marriages as practiced in late Avar-period communities in the Carpathian Basin.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig13\", \"MOESM1\", \"MOESM1\"], \"section\": \"Social changes in Eastern Germany\", \"text\": \"In parallel with the increase of genetic interconnectedness within the sites, we also observe that the organization of the cemeteries changed, reflecting in the spatial layout the extended pedigrees. Although close relatives were buried significantly closer together than non-related individuals in both the MP and the SP, only during the SP did cemeteries feature a significant correlation between genetic and spatial distances, suggesting that the cemeteries were planned and structured around these larger kin groups (Mantel statistic based on Spearman\\u2019s rank correlation; P\\u2009=\\u20090.0001 for both Niederw\\u00fcnsch and Steuden) (Extended Data Fig. 9 and Supplementary Figs. 63 and 65). This signal is most prominent in the site of Steuden, where at least 27% of all variance in spatial distances between graves is explained by genetic relatedness.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\", \"Fig14\", \"Fig14\"], \"section\": \"Social changes in Eastern Germany\", \"text\": \"Although the sex ratio of adults across the SP sites is balanced (Exact binomial test; P\\u2009=\\u20091 for Steuden, P\\u2009=\\u20090.08794 for Niederw\\u00fcnsch), females have on average significantly fewer close relatives than males (Fisher\\u2019s exact test; P\\u2009=\\u20090.008) and feature overall an increased pairwise mismatch rate compared with males (Wilcoxon rank sum test; test statistic (W)\\u2009=\\u200913,976,553, P\\u2009=\\u20090.04218; Supplementary Fig. 67) and a lower sum of IBD segments shared within sites (Welch two-sample t-test; t\\u2009=\\u2009\\u22123.707, d.f.\\u2009=\\u2009355.82, P\\u2009=\\u20090.0002431) (Extended Data Fig. 10c). However, females share more IBD segments between different sites than males, thereby demonstrating higher inter-site relatedness among females in contrast to higher intra-site relatedness among males (Welch two-sample t-test; t\\u2009=\\u20092.9513, d.f.\\u2009=\\u200964.952, P\\u2009=\\u20090.004398) (Extended Data Fig. 10c).\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig13\", \"Fig14\", \"MOESM1\", \"MOESM1\"], \"section\": \"Social changes in Eastern Germany\", \"text\": \"This demonstrates that exogenous origin was more common for females than for males, suggesting a patrilocal inheritance organization, and agrees well with the nearly exclusively patrilineally organized pedigrees (for example, 88% patrilineal lineages (95% confidence interval: 68\\u201397%) versus 12% matrilineal lineages (95% confidence interval: 4\\u201332%)) and the significant underrepresentation of female offspring at the sites (Extended Data Figs. 9 and 10b and Aupplementary Fig. 54). Across all SP sites in Eastern Germany, we find only one instance of mitochondrial haplogroups being transmitted further than one daughter generation. This contrasts with the preceding MP population, for which we detect no difference in the number of close relatives between males and females (Fisher\\u2019s exact test; P\\u2009=\\u20090.74) and no difference in pairwise mismatch rate in general (Wilcoxon rank sum test; W\\u2009=\\u20093,215,752, P\\u2009=\\u20090.06), potentially suggesting a less strictly patrilineal social system before the arrival of Slavic groups (Supplementary Fig. 67). However, owing to the overall smaller number of identified relatives, these tests might be less statistically conclusive and underestimate signals of MP female exogamy and patrilineal practices.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"Fig14\", \"MOESM1\", \"MOESM1\", \"MOESM1\"], \"section\": \"Social changes in Eastern Germany\", \"text\": \"Patterns of patrilocal organization into kin groups are broadly similar across regions and might have contributed to the previously described turnover and homogenization of the paternal gene pool (Extended Data Fig. 10b). In particular, we find an identical pattern of correlation of spatial and genetic distances in SP Velim, Croatia (Mantel statistic based on Spearman\\u2019s rank correlation; P\\u2009=\\u20090.0001; Supplementary Figs. 64 and 65d) as well as evidence for patrilocality and female exogamy (for example, more close biological relatives among males than among females (Fisher\\u2019s exact test; P\\u2009=\\u20090.027) and higher pairwise mismatch rates among females compared to males (Wilcoxon rank sum; W\\u2009=\\u2009156,550, P\\u2009=\\u20090.02377; Supplementary Fig. 67g\\u2013i)), mirroring the social stratification observed in the Elbe-Saale region. By contrast, at Velim the mean number of close relatives in the site is significantly lower than in Niederw\\u00fcnsch or Steuden (1.01\\u2009\\u00b1\\u20090.17). Within a shared pattern of patrilocality, SP fine-scale organization differed substantially across Central Europe owing to the complex, regionally contingent nature of this expansion. Rather than simple replacement, partial integration of the local population was probably dependent on the fortunes of specific groups or families. However, the substantial number of individuals from sites across Eastern Germany, Croatia and Poland\\u2013Northwestern Ukraine who share comparably large amounts of IBD with each other confirms that these Slavic-associated groups were closely linked as the result of a shared biological origin and recent geographical expansion.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\"], \"section\": \"\", \"text\": \"In modern ethnic and national terminology, \\u2018Slavs\\u2019 denotes all speakers of Slavic languages and/or citizens of the Slavic nation states. This concept of an ethnic collective spanning several nations is much more marked than among the Germanic or Romance speakers in the rest of Europe (Supplementary Note\\u00a01). The extent to which Slavic identity mattered diverged; it was important for nineteenth and twentieth century Slavic nationalisms, Pan-Slavism and Russian imperialism, but regional or national allegiances often carried more weight. The prejudices of their western neighbours who tended to regard Slavs as culturally inferior reinforced sentiments of Slavic commonality. The question of Slavic origins, addressed in this Article, had a crucial role in ideological debates about the unity and the significance of the Slavs. It is therefore important to be precise in the scholarly use of the term. In research about the early Slavs, the meanings of the term diverge. In written sources since the sixth and seventh century in Byzantium and the west, groups of Slavs or Wends increasingly appear in a wide range of lands beyond and along the Danube and the Elbe rivers. We can make use of different sources to understand how large parts of Europe became Slavic: outside perceptions about Slavs in texts; archaeological traces of shared cultural practices among early Slavs (particularly the Prague-Korchak culture); linguistic reconstructions of a common Slavic language prior to the particular Slavic idioms; and shifts in ancestry of the medieval gene pool, which point to migrations. We should not take these disciplinary results as proxies for each other as attributes of a coherent people called Slavs; yet, they provide different perspectives on the Slavicisation of Europe during the Early Middle Ages. Combining them allows us to overcome simplistic theories of an expansion of the Slavs and instead understand the common dynamic and the different ways in which Slavic peoples began to form in many parts of Europe. We therefore use Slavs for populations named in this way in contemporary texts, without implying that they self-identified as such. These Slavic groups can be localized, but hardly circumscribed. We do not use genetic or archaeological features in regions where Slavs spread to distinguish between Slavs and non-Slavs, or between speakers and non-speakers of Slavic languages, although we assume that these phenomena overlapped to a considerable degree.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM3\", \"MOESM3\", \"MOESM1\"], \"section\": \"Naming\", \"text\": \"Within tables and figures, we refer to groups of published individuals by the names given in the Allen Ancient DNA resource v.54.1. Sample sizes, context information and publication names can be found in Supplementary Tables 7\\u201311. In the main text and Supplementary Notes, we used the following abbreviations for archaeological time periods: N, Neolithic; C, Chalcolithic; EBA, Early Bronze Age; MBA, Middle Bronze Age; LBA, Late Bronze Age; IA,\\u00a0Iron Age; RA, Roman Age; EMA, Early Middle Ages; MA, Middle Ages; H, historical.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\"], \"section\": \"Permissions for archaeological research\", \"text\": \"Provenance information for samples from all archaeological sites is given in Supplementary Information Note\\u00a02, together with short descriptions of each site, the institution owning the samples (or custodians of the samples), the responsible co-author who obtained permission to analyse and the year of the permission granted.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM3\"], \"section\": \"Dataset\", \"text\": \"We merged our ancient DNA data with previously published datasets of ancient individuals reported by the Reich Lab in the Allen Ancient DNA Resource v.54.1 (https://reich.hms.harvard.edu/allen-ancient-dna-resource-aadr-downloadable-genotypes-present-day-and-ancient-dna-data) (1240k SNP panel). For comparisons with present-day groups, we compiled and curated a high-resolution, quality-filtered reference dataset containing genotypes for 426,135 SNPs (the intersection of several different Affymetrix and Illumina chip types) from 12,176 contemporary individuals sampled from 49 (mostly European and West Asian) populations from previously published datasets as described in Gretzinger et al.. Sample sizes are given in Supplementary Table 3. We produced four different datasets:\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\"], \"section\": \"\", \"text\": \"We applied ADMIXTURE in supervised mode using modern reference populations at K\\u2009=\\u200912. This analysis was run on haploid data with the parameter \\u201chaploid\\u201d set to all (=\\u201c*\\u201d). To obtain point estimates for populations, we averaged individual point estimates and calculated the s.e.m. As modern references we used the groupings listed in the Supplementary Notes\\u00a04.1. The Q matrix of this ADMIXTURE analysis was also used as input for FSTruct as described by the authors.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM1\"], \"section\": \"Language dating\", \"text\": \"We use Bayesian phylogenetic inference to estimate the ages of language divergence as described. Analyses were performed on the IE-CoR database that stores data on cognate relationships (shared word origin) between 161 Indo-European languages, in a reference set of 170 basic meanings (https://iecor.clld.org). Divergence date distributions for the Balto-Slavic and Slavic subgroups were extracted from the sample of 37,004 trees resulting from the main analysis of Heggarty et al.. A critical evaluation of this approach is discussed in Supplementary Note\\u00a09. However, we highlight that these computational estimates for the splits of the Baltic and Slavic languages compare well with estimates produced by methods of traditional Indo-European linguistics. These estimates are based on the reconstructed rapidity of diversification relative to that observed in the other branches of Indo-European (some of which is recorded in writing thousands of years before Baltic and Slavic), the nature of contacts and convergence between Balto-Slavic and other languages (IE and non-IE), and the character of the reconstructible Proto-Balto\\u2013Slavic vocabulary\\u2014but also attempted links to the archaeological picture. The dating of the split of Proto-Balto\\u2013Slavic is generally in agreement with the results of traditional historical linguistics as described above; and earlier glottochronological approaches have yielded similar dating\\u2014for example, an estimation at the fifteenth and fourteenth century bce.\"}, {\"pmc\": \"PMC12507669\", \"pmid\": \"40903570\", \"reference_ids\": [\"MOESM2\"], \"section\": \"Reporting summary\", \"text\": \"Further information on research design is available in the\\u00a0Nature Portfolio Reporting Summary linked to this article.\"}]"
Metadata
"{}"