|
Post by Admin on Dec 8, 2023 19:17:22 GMT
A genetic history of the Balkans from Roman frontier to Slavic migrations Summary The rise and fall of the Roman Empire was a socio-political process with enormous ramifications for human history. The Middle Danube was a crucial frontier and a crossroads for population and cultural movement. Here, we present genome-wide data from 136 Balkan individuals dated to the 1st millennium CE. Despite extensive militarization and cultural influence, we find little ancestry contribution from peoples of Italic descent. However, we trace a large-scale influx of people of Anatolian ancestry during the Imperial period. Between ∼250 and 550 CE, we detect migrants with ancestry from Central/Northern Europe and the Steppe, confirming that “barbarian” migrations were propelled by ethnically diverse confederations. Following the end of Roman control, we detect the large-scale arrival of individuals who were genetically similar to modern Eastern European Slavic-speaking populations, who contributed 30%–60% of the ancestry of Balkan people, representing one of the largest permanent demographic changes anywhere in Europe during the Migration Period. Introduction At its peak in the 2nd century CE, the Roman Empire stretched from Mesopotamia and Arabia in the east to Britain in the west and from the Rhine and Danube rivers in the north to the Sahara Desert in the south.1 The massive extraction and mobilization of resources from western Britain to the eastern desert of Egypt by the imperial polity stimulated the movement of humans via both coercive and consensual processes, effectively restructuring populations across this vast zone. The Balkan Peninsula has been a historic crossroad of eastern and western Mediterranean cultures, as well as continental European influences from the north and Mediterranean from the south. From the 1st to the 6th century CE, the Roman Empire’s Middle Danube frontier in present-day Croatia and Serbia was a zone of defense, confrontation, and exchange with populations living north of the frontier. This region was also a source of significant mineral wealth and a crucial hinge in a ∼2,000-km-long corridor of military and communications infrastructure linking the Black Sea to the Black Forest.2 Following the establishment of Roman control in the early 1st century CE, the region became increasingly urbanized and culturally “Romanized.” Between ca. 268 and 610 CE, more than half of all Roman emperors belonged to families originating in the Middle Danube.3 In late antiquity, the region experienced numerous invading groups labeled by historical sources as Goths, Huns, Gepids, Avars, Heruls, Lombards, or Slavs.4 Non-Romans were also increasingly recruited into the Roman army from peoples across the northern frontier. Various Germanic groups settled in the Danubian region, and some late antique cultural artifacts (and associated human remains) have been attributed to “Germanic”-related influence.4 Nevertheless, the Roman Empire retained some control over this frontier zone into the second half of the 6th century. However, over the later 6th and 7th centuries, as the Roman Empire (ruled from Constantinople, ∼1,000 km away) was confronted by pandemic plague and environmental, political, and military challenges, Roman control over this frontier was lost.5,6 The end of imperial hegemony in the Balkans coincided with further population movements patchily attested in the historical record, including the arrival of the Slavs, whose migration to the region was, much like the arrival of Germanic groups in post-Roman Britain, significant enough to have a particularly lasting impact, reflected in the south Slavic languages widely spoken in the Balkans today.7 Slavic-associated ancestry in present-day populations has been identified as far as the Peloponnese8 (the southern tip of the Balkan Peninsula in present-day Greece), but the degree, timing, and character of permanent demographic impacts across the region have been poorly understood. Although historians have explored Roman imperialism through the lenses of geopolitics, institutions, cultures, and economics, the scale of the Roman Empire’s impact on the population history of its constituent territories is only now becoming understood through the recovery and analysis of ancient DNA. Ancient DNA can complement or challenge conventional archaeological and textual evidence, offering direct insights into individual histories and processes of population change, including social groups whose movements have hitherto been mostly invisible in elite-dominated sources. In fact, archaeogenetic studies are starting to confirm the hints preserved in the documentary record of the empire’s remarkable capacity to foster mobility and mixture.9,10,11 For instance, a man from Roman York in northern England (ancient Eboracum) showed affinities to modern Middle Eastern populations,12 and individuals with a high proportion of North African ancestry were found in southern Iberia.13 A study of 48 skeletons from Rome’s hinterland in the Imperial period showed that at the peak of the Empire, genetic ancestry became much more heterogeneous than in previous periods and shifted toward Near Eastern populations,14,15 and a similarly dramatic shift was shown to extend deep into central Italy.16 Archaeological DNA is also being used to trace the timing, nature, and extent of migrations and population change in post-Roman Europe, from Anglo-Saxon England17 to Lombard Italy.18 The Middle Danube frontier, a crucial axis for the Roman Empire, has not been systematically characterized using archaeogenetic data. To explore the population history of the Balkans (bounded by the Adriatic, the Central Mediterranean, the Aegean Seas, and, to the north, by the Middle and Lower Danube and Sava rivers) in the high Imperial (ca. 1–250 CE), late Imperial (ca. 250–550 CE), and post-Roman (ca. 550–1000 CE) periods, we present new genomic data from 136 ancient individuals from present-day Croatia and Serbia and 6 from Austria, the Czech Republic, and Slovakia, along with information on the archaeological context of their burial (Data S1, section 1). This dataset furnishes insights into the population dynamics of a vital frontier zone, including changes likely associated with the introduction of Slavic languages and the making of modern Balkan populations.
|
|
|
Post by Admin on Dec 8, 2023 19:18:50 GMT
Results Data generation We extracted DNA from 146 ancient Balkan samples (Data S2, Table 1), of which 136 yielded genome-wide data after in-solution hybridization enrichment with either the “1240k19,20” panel of about 1.23 million single-nucleotide polymorphisms (SNPs) or the “Twist” panel that targets an enlarged set of 1.43 million SNPs (the same set of core SNPs and supplementary content)21 (STAR Methods). The individuals were excavated from 20 different sites (Figures 1A and 1B) representing a variety of regions and archaeological contexts including, among others, Viminacium (Kostolac, Serbia), the capital of the Roman Upper Moesia province located at the confluence of the Mlava River and the Danube, where we report data from 57 individuals from 6 different necropolises22 (Data S1, section 1); Roman colonies such as Iader (Zadar, Croatia) on the Adriatic coast and Siscia (Sisak, Croatia) and Mursa (Osijek, Croatia) on the Pannonian road from the Adriatic to the Danube; military fortresses such as Tilurium (Gardun, Croatia) and Timacum Minus (Ravna, Serbia)23; and Early Medieval necropolises such as Jagodnjak (Croatia) and Nuštar-Dvorac (Croatia). To place the results in a geographic and temporal context, we also generated genome-wide data from six Early Medieval Central European individuals from present-day Austria, the Czech Republic, and Slovakia, Affymetrix Human Origins SNP array24 data from modern Serbs (n = 37) (Data S2, Table 2), and 38 new radiocarbon dates (Data S2, Table 9). Figure 1 Overview of ancient Balkan individuals analyzed in this study For genome-wide analyses, we filter out 13 newly reported individuals with fewer than 20,000 SNPs and/or with evidence of contamination (STAR Methods) and include 15 individuals with previously reported genomic data25 from present-day Croatia, Albania, North Macedonia, Greece, Romania, and Bulgaria, for a total of 138 Balkan individuals mostly dated to ∼1–1000 CE (Figures 1A and 1B; Data S2, Table 1). High ancestry heterogeneity To study the 138 Balkan individuals, we performed principal-component analysis (PCA) by projecting them and other ancient individuals from relevant periods and regions onto the axes computed on 1,036 present-day West-Eurasian (WE) individuals (Figures 1C and S1) genotyped on the Affymetrix Human Origins array. A key feature of the data is the presence of two parallel genetic clines running along PC1 (Figure 1C). The first, which we call the “Bronze-Iron Age Balkan cline,” includes southern (Aegean) Bronze and Iron Age groups on the right extreme closer to Near Easterners (larger values in PC1) and northern Bronze and Iron Age groups from modern Croatia and Serbia on the left extreme closer to Central/Northern/Eastern European populations (lower values in PC1); Bronze-Iron Age groups from Bulgaria and Albania take intermediate positions. This Bronze-Iron Age cline is paralleled by the “present-day Balkan cline,” which is shifted upward (higher values in PC2) with respect to the Iron Age cline but displays in PC1 the same geographical pattern of southern Balkan populations such as the Greeks on the right and northern Balkan populations such as Croatians on the left (Figure 1D). The maintenance of the same geographical pattern along PC1 in both clines points to some degree of local continuity from the Iron Age across the entire region, along with the strong impact of migration from outside the Balkans, affecting all groups from north to south over the past 2,000 years. Irrespective of modern nation-state boundaries, populations in our study area have been shaped by similar processes of migration and change. Balkan individuals in our 1st millennium CE transect showed higher ancestry heterogeneity in PCA compared with previous Iron Age Balkan populations (variances in PC1 and PC2 values are significantly different with p = 0.045 and 0.0046, respectively), with most spreading along either the present-day or the Bronze-Iron Age Balkan clines. This suggests that key demographic events involved in the formation of present-day groups had already taken place by ∼1000 CE. The remaining individuals plot far beyond the two Balkan genetic clines and likely represent cases of sporadic long-distance mobility that provide evidence concerning the regions acting as demographic sources for the Balkans during this period. Given the high ancestry heterogeneity observed in our dataset, even within the same sites and necropolises, we estimate ancestry proportions separately for each individual. We used qpAdm26,27 with Balkan Iron Age populations as the local ancestry source and earlier and contemporaneous populations from neighboring regions as proximate sources for newly arriving ancestries (Data S1, section 4). Large-scale demographic input from western Anatolia Around half of the 45 individuals between ∼1 and 250 CE can be fitted with qpAdm models featuring only Balkan Iron Age groups (Figure 2A) and are characterized by a high frequency (5 out of 10) of Y chromosome lineage E-V13 (Data S1, section 2), which has been hypothesized to have experienced a Bronze-to-Iron Age expansion in the Balkans.28 These individuals, sampled from Roman towns such as Viminacium, Tragurium (Trogir), and Mursa (Osijek), are consistent with being direct descendants of local Balkan Iron Age populations (Figure 2A), pointing to a high degree of integration of the local population into Roman society. Despite the exceptional number of Roman colonies in the region and the large military presence along this frontier, there is little ancestry contribution from populations long established in the Italian Peninsula, a pattern exemplified by the almost complete absence in our Balkan transect of Y chromosome lineage R1b-U152, the most common paternal lineage in Bronze Age and Iron Age populations in the Italian Peninsula.15,16,29 The prevalence of cremation burials in the earliest centuries could bias the sample, but even after the transition to inhumation burial around the 2nd century, ancestry contributions from populations of Italian descent are not detectable. Rome’s cultural impact on the Middle Danube was deep, but our findings suggest that it was not accompanied by large-scale population movement from the metropole, at least by the descendants of central Italian Iron Age populations.
|
|
|
Post by Admin on Dec 10, 2023 0:18:09 GMT
Figure 2 A diversity of ancestral origins The Roman Empire did, however, stimulate demographic change in the Balkans. In this early period, ∼1/3 of the individuals (15 of the 45) fall beyond the Balkan clines in PCA (Figures 1C and S4) but close to Near Easterners and can be modeled as deriving their ancestry predominantly from Roman/Byzantine populations from western Anatolia and, in one case, from Northern Levantine groups (Figure 2A; Data S2, Table 6). Most of these individuals were excavated at four different Viminacium necropolises, but we also found them at other urban centers such as Tragurium (Trogir) and Iader (Zadar). A very strong demographic shift toward Anatolia is also evident in Rome and central Italy during the same period15,25 and demonstrates long-distance mobility plausibly originating from the major eastern urban centers of the Empire such as Ephesus, Corinth, or Byzantium/Constantinople, and our results show that these migrants had a major demographic impact not only on the Imperial capital but also on other large towns on the Empire’s northern periphery. Our data also provide insights concerning the social dynamics of this demographic process. Unlike the Balkan Iron Age ancestry group whose sex ratio was evenly balanced (11 females of the 22 individuals), the 12 adult individuals with full Anatolian/Levantine ancestry included only 2 women (p = 0.019 for a one-side binomial test). This points to a larger contribution of Near Eastern men but could also result from different burial customs between the sexes. People with Anatolian ancestry and people with local Balkan ancestry were not spatially segregated in burial nor, for the most part, culturally distinct in burial customs or grave goods. They admixed and were buried at the same necropolises, even side by side as in tomb G-148 at Rit necropolis (Figure 2D). However, the evidence may also point to some degree of social stratification, because all 3 individuals at the Rit necropolis of Anatolian origin were buried in stone sarcophagi with exceptionally rich grave goods (Figure 2D; Data S1, section 1).35 The main source of migrants to the region shifted away from Anatolia after ∼300 CE (Figure 2A), but together with the ancestral legacy of local Balkan Iron Age groups, Anatolian-related ancestry persisted in admixed form into the later Medieval individuals (Figure 2A) with a mean of 23% (95% CI = 17%–29%), indicating that this was a deep and lasting demographic impact. Migrants from sub-Saharan Africa and North Africa Our newly reported data also revealed sporadic long-distance mobility. Three men who likely lived in the 2nd or 3rd centuries CE fell outside European and Near Eastern variability (Figure S1), close to present-day and ancient Africans (Figure S2A). Proximal qpAdm modeling confirmed these observations (Figure 2A; Data S2, Table 6) with 33% and 100% North African ancestry for individuals I26775 (Iader) and I32304 (Viminacium Pećine), respectively, whereas I15499 (Viminacium Pirivoj) could be modeled using only ancient East African populations, supporting an East African ancestral origin and agreeing with his uniparental markers mtDNA L2a1j and Y chromosome E1b-V32, both common in East Africa today.28,36 The individual of East African ancestry was buried with an oil lamp depicting Jupiter-related eagle iconography (Figure 2C; Data S1, section 1), not a common finding in Viminacium graves.37 Isotopic analysis of tooth roots showed that he was also an outlier with respect to dietary habits during childhood (Figure 2B), with elevated δ15N and δ13C values indicating the likely consumption of marine protein sources,38 unlike individuals from Pirivoj and other necropolises whose values (Figure 2B) were similar to the Roman-Period population from Sirmium30 and consistent with a largely C3-based diet with a significant portion of animal protein consumption.38 Thus, he likely spent his early years elsewhere, possibly in East Africa, the land of his ancestors, and although we will never know his whole life story, whether as a soldier, slave, merchant, or migrant, it encompassed a long journey that ended with his death in adolescence on the northern frontiers of the Roman Empire.
|
|
|
Post by Admin on Dec 11, 2023 5:58:54 GMT
Figure S1 PCA with ancient samples projected onto the PCs computed on present-day West-Eurasian individuals, related to Figure 1 Figure S2 PCAs with ancient samples projected onto the PCs computed on present-day West-Eurasian populations and additional individuals, related to Figure 1 From internal to external migration during late antiquity Beginning in the 3rd or 4th century CE, we observe individuals who are admixed with ancestry related to Central/Northern European and Pontic-Kazakh Steppe populations (Figure 4A; Data S2, Table 6). These two ancestry types tend to colocalize in the same individuals, suggesting that the stream of migrants into the Balkans included people who were admixtures of these two sources, although there are some exceptions, such as two contemporaries from Viminacium, Pećine necropolis, who can be modeled as having 36%–50% Pontic-Kazakh Steppe-related ancestry without any contribution of Central/Northern European ancestry (Figure 1C; Data S2, Table 6). Individuals bearing these ancestries were buried at the same necropolises (such as Pećine and Više Grobalja at Viminacium) as individuals with predominantly local and Anatolian ancestries, often with overlapping radiocarbon dates, and displayed 42%–55% of Balkan Iron Age-related ancestry (Data S2, Table 6). By contrast, only 2 of the 9 males belonged to Y chromosome lineages found among individuals with a fully local ancestry profile (Data S1, section 2), with the rest belonging to three haplogroups: I1 and R1b-U106 with a strong Northern European distribution and haplogroup R1a-Z93, which was common in the Steppe during the Iron Age and early 1st millennium CE.39,40,41 Such discrepancy between the autosomal and Y chromosome signals could be explained by a patrilineal social organization such as has been deduced for early Germanic societies42 that resulted in the persistence of the incoming male lineages due to social selection for reproductive success among male offspring from these lineages and its observation in the admixed individuals in our transect. People with these ancestry profiles present evidence of different dietary patterns, too, as shown by significantly elevated δ13C values (p = 0.001) for individuals bearing ancestry from Pontic-Kazakh Steppe groups (Figure 2B), likely pointing to a C4-rich diet.43 The appearance of individuals with admixed Central/Northern European and Pontic-Kazakh Steppe ancestry inside the Roman Empire in late antiquity reflects the Roman encounter with various trans-frontier populations in this period. Notably, many individuals reflect a prior process of population admixture between these two sources that likely occurred beyond the Roman frontier, perhaps indicative of, e.g., the formation of diverse confederations under Gothic leadership.44 Furthermore, although the Roman Empire intermittently lost military control of this frontier from the middle of the third century on, it is noteworthy that many individuals with these ancestries appear integrated into Roman society well before the final breakdown of Roman control of the region. This pattern confirms the importance of processes such as migration, recruitment, and settlement (whether sanctioned by the imperial government or not) in the demographic history of the region in late antiquity, a period of intense interaction and exchange across the Danube border.45 It is also noteworthy that only 3 individuals show >80% ancestry related to Central/Northern European and Pontic-Kazakh Steppe groups. Perhaps the fewer samples whose date range falls in the 6th century CE (n = 10) obscure the importance of the direct migration of large, predominantly Germanic groups into the region. However, it is just as important to observe that many individuals belonging to archaeological contexts identified by cultural criteria (Data S1, section 1) as belonging to various Germanic groups reflect a process of admixture with local populations. At Kormadin, for instance, in what has conventionally been identified as a “Gepid” cemetery, of the four individuals tested, we identified two who model as completely local Iron Age Balkan ancestry and two, including one child aged 5–7 years, who display local Iron Age Balkan, Central/Northern European, and Pontic-Kazakh Steppe ancestry. It is also unexpected to find that Central/North European and Pontic-Kazakh Steppe ancestries vanished after 700 CE (95% CI for the sum of these two ancestry proportions = 0%–3%) (Figure 4A; Data S2, Table 6). Although the relatively small differentiation between Central/North European and Eastern European ancestries could have resulted in the misassignment of small proportions of Central/North European ancestry as Eastern European ancestry, this result is supported by the complete absence (Data S1, section 2) of Y chromosome lineages clearly associated with Central/North European and Pontic-Kazakh Steppe ancestry (I1, R1b-U106, and R1a-Z93) in the 24 men in our transect who lived after 700 CE (95% CI for the frequency of those haplogroups = 0%–12%). Although this absence could reflect unknown sampling bias, it suggests that the population size of incoming Central/North European groups may have been limited as compared with the local Iron Age population and/or that selective demographic processes—out-migration, differential mortality due to urbanism or military service—acted to prevent a long-lasting demographic impact of these groups. Slavic migrations and the formation of the present-day Balkan gene pool By 700 CE, a new type of ancestry appeared across all the Balkan regions covered by our sampling. In a PCA projection onto diverse WE populations (Figure 1C), these individuals fall at similar positions as the earlier group with Central/Northern European and Pontic-Kazakh Steppe-related ancestry. However, we can distinguish their ancestry with a PCA setup more sensitive to recent drift separating Central/Northern and Eastern European populations (Figure 3A). Several Balkan individuals before 700 CE plot close to present-day Central and Northern European Germanic-speaking populations, overlapping individuals from Langobard-associated cemeteries in Hungary and Northern Italy18 displaying Central/Northern European-related ancestry (CNE_EarlyMedieval). After 700 CE, we observe a clear shift toward present-day Eastern European Slavic-speaking populations in the ancient Balkan transect, a shift mirrored by present-day Balkan populations (Figure 3A). Accordingly, Eastern European-related populations share more alleles (Z = 9.85) with Balkan individuals after 700 CE compared with before 700 CE (Figure 3B). The differential affinities of Balkan individuals with the strongest Central/Northern European shift in PCA (Z = 1.99) and Balkan individuals with the strongest Eastern European shift in PCA (Z = −3.41) is evident using f4-statistics of the form f4 (OldAfrica, Test; Eastern European-related, Central/Northern European-related) (Figure 3B). Corroborating these results, qpAdm models (Data S1, section 4) with Central/Northern European and Pontic-Kazakh Steppe groups yield a very poor fit (p = 2.70×10−15; Data S2, Table 7) for the group of Balkan individuals with the strongest Eastern European shift, and we were able to obtain a better fit with variable proportions of Balkan Iron Age-related, Anatolian-related, and Eastern European-related ancestry (p = 0.049; Data S2, Table 7). As an Eastern European-related proxy, we used a group of Early Medieval individuals excavated in western Hungary, the Czech Republic, eastern Austria, and western Slovakia (CEE_EarlyMedieval). This group fell within the variation of present-day Eastern European Slavic-speaking populations, very close to the Balkan individuals in our dataset with the strongest Eastern European-related shift (Figures 3A and S3).
|
|
|
Post by Admin on Dec 11, 2023 21:25:46 GMT
Figure 3 Arrival of ancestry related to Eastern European populations after 700 CE We present evidence that Eastern European ancestry was sporadically present in the Balkans long before the Slavic migrations of late antiquity. Indeed, a woman who probably died in the 2nd or 3rd centuries CE and was buried at Više Grobalja presents unmixed Eastern European ancestry (Figure 4A), offering a remarkable illustration of how small-scale individual percolation into the dynamic economy of the Roman Empire may have preceded larger-scale migration. The vast majority of the individuals with Eastern European ancestry in our dataset appear in the 7th–10th centuries and are of admixed ancestry (Figure 4A). The Slavic migrations started as early as the 6th century,46 and our dataset may not reflect the early phases, although it provides insights into its dynamics. Of the seven Balkan individuals with more than 90% East European-related ancestry who were more likely to be migrants, three were females. This finding, together with a ∼50/50 ratio of local versus non-local (R1a-Z282, I2a-L621, and Q1a-L715) Y chromosome lineages (Data S1, section 2), hints at different social dynamics operating during this event compared with previous periods: here, females and males make major contributions. We have evidence of the interaction between the two groups at the individual level. At the fortified settlement of Brekinjova Kosa (Bojna, Croatia), two adult men dated to 770–890 cal CE were buried together in the same pit, the younger one (I26748) with healed skull trauma and with 97% Eastern European-related ancestry and the older one (I26749) with ancestry entirely deriving from Balkan Iron Age populations. Additionally, at the site of Dvorac (Nuštar, Croatia), a woman (I28390; Grave 52) with ∼90% Eastern European ancestry had a son (I34800; Grave 50-A; 64% of this ancestry) (Figure 4A; Data S2, Table 6) with an unsampled man who, like the main group of individuals from the site, must have had ∼30% Eastern European ancestry, demonstrating a direct case of incorporation of non-local women that could exemplify some of the social processes at play. The finding of a pair of 10th-century twins with southwestern European ancestry at Timacum Minus again demonstrates sporadic mobility from far-away regions in the Middle Danube.
|
|