"The Roman Empire expanded through the Mediterranean shores and brought human mobility and cosmopolitanism across this inland sea to an unprecedented scale. However, if this was also common at the Empire frontiers remains undetermined. The Balkans and Danube River were of strategic importance for the Romans acting as an East-West connection and as a defense line against “barbarian” tribes. We generated genome-wide data from 70 ancient individuals from present-day Serbia dated to the first millennium CE; including Viminacium, capital of Moesia Superior province. Our analyses reveal large scale-movements from Anatolia during Imperial rule, similar to the pattern observed in Rome, and cases of individual mobility from as far as East Africa. Between ~250-500 CE, we detect gene-flow from Central/Northern Europe harboring admixtures of Iron Age steppe groups. Tenth-century CE individuals harbored NorthEastern European-related ancestry likely associated to Slavic-speakers, which contributed >20% of the ancestry of today's Balkan people.
"A key feature of the data is two parallel genetic clines running along PC1 (Fig. 1). We call the first the “Balkan Iron Age cline”, with southern Balkan populations such as Bronze Age and Iron Age Aegean groups on the right extreme closer to Near Eastern populations (larger values in PC1), northern populations such as Slovenian Iron Age groups on the left extreme closer to Central European populations (smaller values in PC1), and a Bulgarian Iron Age individual and Bronze Age and Iron Age Croatian groups taking intermediate positions but closer to the southern and northern extremes, respectively. This Iron Age cline is mirrored by the “presentday Balkan cline”, which is shifted towards the upper-left of the plot (lower values in PC1 and higher values in PC2) with respect to the Iron Age cline but maintains 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. This suggests that present-day populations are not direct descendants without admixture of Iron Age groups from the same region, and that similar demographic shaped Balkan populations from North to South over the past 2,000 years."
"Consistent with this, we model the ancestry of this Balkans Iron Age Clusteras predominantly deriving from Iron Age (IA) groups from nearby areas in the Balkans, with 67% Aegean Bronze Age-related ancestry and the remainder Slovenia Iron Age-related ancestry (Figure 2; Supplementary section 12.1). A local origin is supported by a high frequency of Ychromosome lineage E-V13, which has been hypothesized to have experienced a Bronze-to-Iron Age expansion in the Balkans and is found in its highest frequencies in the present-day Balkans 17. We interpret this cluster as the descendants of local Balkan Iron Age populations living at Viminacium, where they represented an abundant ancestry group during the Early Imperial and later periods (~47% of sampled individuals from the 1-550 CE).
"The other major cluster (44% of the samples from Viminacium between 1-250 CE) is represented by individuals who projected towards ancient and present-day Eastern Mediterranean groups in PCA (Figure 1A), close to ancient individuals from Rome during Imperial times 3. Their ancestry can be modelled as deriving deeply from Chalcolithic Western Anatolian groups (Figure 2; Supplementary section 12.2), and we refer to this cluster as the Near Eastern-related cluster. The same signal of arrivals individuals with Anatolian/Near Eastern ancestral origins is also evident in Rome during the same period 3, consistent with largescale gene-flow originating from the major eastern urban centers of the Empire (such as Constantinople, Antioch, Smyrna and Alexandria). These results suggest that immigration from the east was a common feature across urban centers in the Roman Empire, including in border areas and large cities/military outposts such as Viminacium. Individuals with Eastern Mediterranean ancestry could have high social status: 3 out of the 4 individuals buried in two sarcophagi (each containing a male-female pair) with exceptionally rich grave goods at the Rit necropolis in Viminacium belonged to the Near Eastern-related cluster, while the remaining one belonged to the Balkans Iron Age-related cluster. This kind of burial was common in the Eastern Roman settlements for aristocratic members of society 20. Individuals from this cluster were also more likely to be inhumated in a wooden coffin rather than freely buried, which could also be an indication of higher social prestige."
Three individuals from ~1-250 CE did not fit into the two major clusters. Two males from Viminacium could be modelled using Iron Age individuals from Northwest Europe as their only source (Figure 2; Supplementary section 12.5), pointing to a Northwestern European origin also supported by the R1b-U106 paternal lineage, which was not been detected in the Balkans in earlier periods but was found at high frequencies in Germanic-speaking areas, both in ancient and present-day individuals. The most remarkable outlier is male I15499, excavated at Pirivoj necropolis in Viminacium, who projects outside West Eurasian genetic diversity (Figure S7). When we incorporated African populations onto the PCA (Figure S8), he projected within the variation of present-day East African populations and close to early Christians from Northern Sudan from 500-800 CE 21 who provide a good fit for his ancestry in qpAdm (Figure 2; Supplementary section 12.4). An Eastern African ancestral origin agrees with his uniparental markers mtDNA L2a1j and Y-chromosome E1b-V32, both common in East Africa today 17,22. Archeological examination of I15499’s grave found an oil lamp depicting an eagle, the symbol of Roman legion (Figure S2C). Although lamps are a common finding in Viminacium graves 23, not many depict military iconography. We hypothesize that this male was a Roman legionary or auxiliary stationed at Viminacium. We cannot determine if he was a Roman citizen, although auxiliary military service for a prolonged period of time resulted in citizenship. Historical evidence also points to African recruits being tapped to reinforce the Roman Danubian limes 24."
"At Slog, we found one directly radiocarbon dated individual with a clear Near Eastern ancestral origin, likely from the Northern Levant (Figure 1B Figure 2; Supplementary section 12.3), as well as directly radiocarbon dated individuals belonging to the Balkans Iron Age-related cluster. This confirms that the two major ancestry clusters from 1-250 CE period co-existed at least three centuries in the Danubian limes. The legacy of Balkans Iron Age groups persists in admixed form in later groups including present-day Balkan populations (see below), whereas the Near Eastern-related ancestral legacy eventually ebbed in favor of Northern/Eastern European-related ancestry, similar to the patterns observed in the city of Rome itself 3."
"These findings support the hypothesis that such individuals were part of a cosmopolitan group comprising a large proportion of individuals in Imperial towns and cities who over time were demographically overwhelmed by populations in the countryside or by faster reproductive rates of rural or populations without as much Near Eastern influence."
"The legacy of Balkans Iron Age groups persists in admixed form in later groups including present-day Balkan populations (see below), whereas the Near Eastern-related ancestral legacy eventually ebbed in favor of Northern/Eastern European-related ancestry, similar to the patterns observed in the city of Rome itself 3. These findings support the hypothesis that such individuals were part of a cosmopolitan group comprising a large proportion of individuals in Imperial towns and cities who over time were demographically overwhelmed by populations in the countryside or by faster reproductive rates of rural or populations without as much Near Eastern influence.
"We found highly similar ancestry trajectories across time in Rome and Viminacium, with a strong Anatolian/Near Eastern influence during the Imperial period that resulted in a large portion of the analyzed individuals in both cities having Near Eastern ancestry, followed by a resurgence of local ancestry after the Empire’s decline 3. These results highlight how mobility from the Easternmost areas of the Empire was a common feature of large cities and towns from the capital city of Rome to the Danubian limes, but that demographically these populations were a veneer without long-lasting influences, suggesting either that they were greatly outnumbered by local rural populations, or that their reproductive rates were much lower than that of local rural populations, consistent with evidence that cities and towns in the Roman empire did not successfully reproduce themselves demographically and instead constantly had to be repopulated through immigration 29. In the Imperial period, genetic data suggest that a large proportion of this immigration derived from the Eastern Mediterranean highlighting the centrality of this region in the period of intense human connectivity during Imperial Rome. Conversely, the decline in the geographic scale and number of people involved in transMediterranean movements following the Empire’s decline is reflected in the fact that in later periods, Eastern Mediterranean influence largely disappeared in both the city of Rome and in the large towns of the Balkans."
(Finally, someone got it!)
We also observe new ancestry during this period at Mediana, Slog necropolis at Timacum Minus and Viminacium (mostly at Pecine and Vise Grobalja necropoli), as early as the 4thcentury CE. A cluster of 10 individuals from these necropoli is shifted in PCA from the Balkans Iron Age-related cluster toward Central/Northern European ancient and present-day populations (Figure 1B). This group which we refer to as Central/Northern European cluster, could be modeled as deriving from two main sources: ~38% related to the local Balkans Iron Age substratum (we use the Balkans Iron Age-related cluster as a proxy for this type of ancestry) and 50% Central/Northern European ancestry (we use as a proxy individuals from a roughly contemporaneous Langobard-associated cemetery in Hungary 25). To obtain a fitting model, a significant proportion of ancestry (~14%) related to contemporaneous nomadic steppe groups (proxied in our analysis by Late Sarmatians from the Eastern Pontic-Caspian steppe 26) is also needed (Figure 2; Supplementary section 12.6). This is even more evident in two individuals from the Pecine necropolis in Viminacium (referred to as Steppe cluster), who could be modelled as deriving ~43% of ancestry from the Balkans Iron Age-related cluster and 57% ancestry from Late Sarmatian-related Steppe groups (Figure 2; Supplementary section 12.7). Ychromosome lineages also provide evidence for gene-flow, as 5 of 7 males in the Central/Northern European and Steppe cluster belonged to two lineages not found in the Balkans earlier: haplogroup I1 with a strong Northern European distribution and haplogroup R1a-Z645, common in the Steppe during the Iron Age and early 1st millennium CE 26–28. The Roman Empire had a prolonged history of contact with Germanic tribes, whose homelands were in Northern Europe between the Rhine and Vistula rivers. During the Great Migration period groups that coalesced as the Goths moved southwards, and settled at the Black Sea north coast prior to their entry in the Roman Empire 6. Our observations are consistent with the hypothesis that such tribes interacted with Steppe-related nomadic populations reaching the Eastern European plateau, and incorporated their ancestry into their gene pool before moving into the Balkans. However, the occurrence and manner of this interaction needs to be clarified with a more thorough sampling of this region and time period.
"The remaining five individuals clustered in the West-Eurasian PCA (Figure 1C) on top of the “present-day Balkan genetic cline”, close to present-day Serbianspeaking individuals that we newly genotyped for this study, but this apparent similarity is a projection artifact as their ancestry could not be fitted using the same qpAdm models (Supplementary section 12.8). To understand this, we performed a PCA using present-day Germanic- and Slavic-speaking populations (Supplementary section 9; Figure S9) that we expected would be sensitive to more recent drift separating Central, Northern and Eastern European populations. The Kuline individuals are more shifted towards present-day Slavicspeaking populations as compared to individuals in the Central/Northern European cluster, agreeing with the presence of Y-chromosome lineage I2-L621 in Kuline, which is common in present-day Slavic-speaking groups and absent in earlier periods. In light of these results, we modeled the ancestry of the Kuline individuals as a mixture of 56% deriving from the local Balkan Iron Age substratum and 44% deriving from Northeastern European Iron Age groups, and obtained a good statistical fit (Figure 2; Supplementary section 12.8). Our results point to a strong demographic impact of Eastern European groups in the Balkans during the Medieval period, likely associated to the arrival of Slavic-speaking populations. Yet, our results rule out a complete demographic replacement, as we observe a significant portion of local Iron Age Balkan ancestry in Kuline individuals. Interestingly, we found sex bias when modeling the X chromosome of the individuals of this necropolis (Supplementary section 12.8). Perhaps the immigrant groups were constituted by a higher number of women, who therefore impacted more greatly in the demographics of the post-Roman Balkans. However, these findings have only been observed in the Kuline individuals with North-European related ancestry (n=5), we suggest more data will be needed to obtain more evidence Slavic sex bias in the Balkans. To explore whether this Northeastern European ancestry signal persisted in present-day Balkan and Aegean populations, we attempted to model present day groups by using the same qpAdmmodel used for the Kuline individuals (Supplementary section 13). Present-day Serbs, Croats and the rest of central/northern Balkan populations yielded a similar ancestral composition as the Kuline individuals, with approximately 50% Northeastern European-related ancestry admixed with ancestry related to Iron Age native Balkan population (Figure 3), implying substantial population continuity in the region over the last 1,000 years. This ancestry signal significantly decreases in more southern groups, but it is still presents in populations from mainland Greece (~30%) and even the Aegean islands (7-20%)."