Post by Admin on Jun 20, 2023 22:30:05 GMT
To quantify the impact of north-eastern European migrations into the Balkans, we used f4 rotating-source qpAdm tests with proximate sources dating to the Iron Age, Bronze Age and the Roman era, where Russia Ingria IA served as a proxy for Balto-Slavic-related ancestry. Our qpAdm models recover remarkably high Balto-Slavic-related ancestry in the late Roman and Post-Roman populations of Croatia (50-65%), Montenegro (45-65%), North Macedonia (30-50%), and Serbia (50-55%) (Table S11). These admixture proportions are almost identical to those proposed by studies on modern Slavic-speaking populations using different methods, which recover the ancestry of South Slavs as 55-70% Balto-Slavic-related, with the remainder originating from the pre-Slavic inhabitants of the Balkans (48), further affirming the accuracy of our models.
Figure 6.
Mobest analysis of Late Bronze Age-Iron Age, Medieval, and post-Medieval samples from the region of Albania, plotting the probability surface that identifies the highest genetic-geographical match at the mean date of the respective individual. The higher the probability surface (light yellow-green), the closer the genetic match. Only a single sample from the post-Medieval population of Bardhoc is included here, as all individuals from this region display the same probability surface (Fig. S7A-C). The latitude and longitude coordinates with the best fit for the examined individuals are provided in EPSG:3035 projection in Table S19.
Considering the abovementioned Anatolian and Balto-Slavic ancestry shifts in the Roman and Medieval Balkans, we model the ancestry of samples from Medieval Albania. In contrast to neighbouring populations, the samples from Medieval South-Eastern (Shtikë, 889-989 calCE) and North-Eastern (Kënetë, 773-885 calCE) Albania (hereafter Albania Mdv) experience only a minor shift on their PC position from the BA-IA to the Migration Period (Fig. 5A, B), hinting at large-scale genetic continuity for over 2500-3000 years. This is reflected in the ancestry makeup of Albania Mdv, as in ultimate f4 qpAdm models they derive 17% of their ancestry from Iran N-like sources (2-5% in the BA-IA), while their IGHG ancestry increases only marginally (0-4% in BA-IA, 6% in Medieval times) (Fig. 3C; Table S5). Proximate qpAdm models comprising Balkan BA-IA sources and proxies for Anatolian-Levantine (East Anatolia BA IA; Syria Ebla EMBA) ancestry also replicate this ancestry shift (Table S12). The most strongly supported result was a two-way model where Albania Mdv derive 85% of their ancestry from BA-IA-Hellenistic Albania, and 15% from either the Anatolian or Levantine proxy, with equal support (Table S12).
To further resolve the observed ancestry patterns, we ran a second proximate model using Roman era West Balkan sources, together with a Slavic proxy (Russia Ingria IA), where we recovered Albania Mdv as either 100% Roman West Balkan, or 85% Roman West Balkan and 15% Slavic-related (Table S13). Accordingly, Albania Mdv cluster together with Roman era West Balkan samples that derive only a small proportion of their ancestry from Anatolian populations [Croatia (Beli Manastir, Gardun, Omišalj, Sisak, Trogir, Velić, Zadar); Montenegro (Doclea); Serbia (West-Balkan-shifted samples)] (Fig. 5B) and show little to no increase in IGHG ancestry (Fig. 3C).
Populations of largely unadmixed palaeo-Balkan ancestry persisted in pockets in other regions of the collapsing Roman Empire as well. Two samples from Early Avar Pannonia (550-650 CE) cluster with populations from Iron Age Bulgaria and Greece on the PCA (Fig. 5C), a relationship corroborated by f3-statistics (Fig. S6), qpAdm (Table S14), and IBD-sharing (Table S21) (50). Such outliers may affirm historical reports of the Avars undertaking mass resettlements of Roman subjects from the area of Thrace towards their Khaganate in Pannonia (62, 63).
Given that some linguistic hypotheses suggest a mixed West Balkan and Thracian origin for the Medieval population of Albania (2, 64), we undertake an f4 qpAdm test combining Roman era West Balkan, East Balkan (the two outliers from Early Avar Hungary), and Slavic-related (Russia IA Ingria) sources. Albania Mdv is once again effectively modelled as being 100% of Roman era West Balkan origin (Croatia Roman Gardun, Montenegro Roman Doclea), or as a two-way mix of a Roman period West Balkan source (85%) and a Slavic-related source (15%) with high support (p = 0.16; SE = 0.04) (Table S15). Two-way or three-way models with West Balkan + East Balkan or West Balkan + East Balkan + Slavic-related sources also passed with low statistical support (SE = 0.12 and SE = 0.09-0.14, respectively), recovering Albania Mdv as deriving 45-60% of their ancestry from the two Thracian-shifted Avar era outliers (Table S15), a finding that may also receive support from f3-statistics (Fig. S2F). Furthermore, models using the Avar era outliers as the sole palaeo-Balkan source for Albania Mdv receive high support (p = 0.72; SE = 0.03; Table S15). However, it is unlikely that the Avar era outliers are realistic local sources due the fact that such significant ancestry shifts would have pulled Albania Mdv samples towards the direction of southeastern Balkan populations on the PCA, which is not the case (Fig. 5A, B). Overall, the PCA and f4 qpAdm statistical models suggest that the Medieval population of Albania was minimally affected by demographic changes during the Roman era, in stark contrast to adjacent regions such as Croatia and Serbia (Fig. 5A). However, currently unsampled urbanised areas in Albania such as Durrës and Shkodër, may have comprised populations of more complex ancestry.
Although we show that some West and East Balkan populations persisted largely unadmixed in late Roman and early post-Roman times (Fig. 5A), such ancestry profiles cannot be found post-900 CE, as Mobest analysis of Albania Mdv shows affinities only to populations from Italy, which maintained a larger proportion of Eastern Mediterranean ancestry compared to contemporary sampled locations in southeastern Europe (Fig. 6B, C). This suggests that the region of modern Albania served as a refugium of Iron Age West Balkan ancestry throughout the demographic and social upheaval that took place during the Migration Period (9, 17, 61). However, our qpAdm models cannot discriminate whether West palaeo-Balkan ancestry in Medieval Albania originated from indigenous populations or other incoming palaeo-Balkan groups from adjacent regions such as the northern Adriatic coast or the Balkan interior. Indeed, IBD-sharing suggests that the sample from Kënetë shares small (10.5 cM) segments with BA and Roman individuals from neighbouring Velika Gruda in Montenegro and Zadar in Croatia, respectively (Table S20) (50). Conversely, the individual from Shtikë shares short segments (11-14.5 cM) with geographically distant Hun and Avar-era commoners from Hungary (Table S20), suggesting that possible trade networks between Medieval Albania and the Avar Khaganate (based on archaeological artefacts) (8, 15, 40) may have also involved genetic exchange.
Figure 6.
Mobest analysis of Late Bronze Age-Iron Age, Medieval, and post-Medieval samples from the region of Albania, plotting the probability surface that identifies the highest genetic-geographical match at the mean date of the respective individual. The higher the probability surface (light yellow-green), the closer the genetic match. Only a single sample from the post-Medieval population of Bardhoc is included here, as all individuals from this region display the same probability surface (Fig. S7A-C). The latitude and longitude coordinates with the best fit for the examined individuals are provided in EPSG:3035 projection in Table S19.
Considering the abovementioned Anatolian and Balto-Slavic ancestry shifts in the Roman and Medieval Balkans, we model the ancestry of samples from Medieval Albania. In contrast to neighbouring populations, the samples from Medieval South-Eastern (Shtikë, 889-989 calCE) and North-Eastern (Kënetë, 773-885 calCE) Albania (hereafter Albania Mdv) experience only a minor shift on their PC position from the BA-IA to the Migration Period (Fig. 5A, B), hinting at large-scale genetic continuity for over 2500-3000 years. This is reflected in the ancestry makeup of Albania Mdv, as in ultimate f4 qpAdm models they derive 17% of their ancestry from Iran N-like sources (2-5% in the BA-IA), while their IGHG ancestry increases only marginally (0-4% in BA-IA, 6% in Medieval times) (Fig. 3C; Table S5). Proximate qpAdm models comprising Balkan BA-IA sources and proxies for Anatolian-Levantine (East Anatolia BA IA; Syria Ebla EMBA) ancestry also replicate this ancestry shift (Table S12). The most strongly supported result was a two-way model where Albania Mdv derive 85% of their ancestry from BA-IA-Hellenistic Albania, and 15% from either the Anatolian or Levantine proxy, with equal support (Table S12).
To further resolve the observed ancestry patterns, we ran a second proximate model using Roman era West Balkan sources, together with a Slavic proxy (Russia Ingria IA), where we recovered Albania Mdv as either 100% Roman West Balkan, or 85% Roman West Balkan and 15% Slavic-related (Table S13). Accordingly, Albania Mdv cluster together with Roman era West Balkan samples that derive only a small proportion of their ancestry from Anatolian populations [Croatia (Beli Manastir, Gardun, Omišalj, Sisak, Trogir, Velić, Zadar); Montenegro (Doclea); Serbia (West-Balkan-shifted samples)] (Fig. 5B) and show little to no increase in IGHG ancestry (Fig. 3C).
Populations of largely unadmixed palaeo-Balkan ancestry persisted in pockets in other regions of the collapsing Roman Empire as well. Two samples from Early Avar Pannonia (550-650 CE) cluster with populations from Iron Age Bulgaria and Greece on the PCA (Fig. 5C), a relationship corroborated by f3-statistics (Fig. S6), qpAdm (Table S14), and IBD-sharing (Table S21) (50). Such outliers may affirm historical reports of the Avars undertaking mass resettlements of Roman subjects from the area of Thrace towards their Khaganate in Pannonia (62, 63).
Given that some linguistic hypotheses suggest a mixed West Balkan and Thracian origin for the Medieval population of Albania (2, 64), we undertake an f4 qpAdm test combining Roman era West Balkan, East Balkan (the two outliers from Early Avar Hungary), and Slavic-related (Russia IA Ingria) sources. Albania Mdv is once again effectively modelled as being 100% of Roman era West Balkan origin (Croatia Roman Gardun, Montenegro Roman Doclea), or as a two-way mix of a Roman period West Balkan source (85%) and a Slavic-related source (15%) with high support (p = 0.16; SE = 0.04) (Table S15). Two-way or three-way models with West Balkan + East Balkan or West Balkan + East Balkan + Slavic-related sources also passed with low statistical support (SE = 0.12 and SE = 0.09-0.14, respectively), recovering Albania Mdv as deriving 45-60% of their ancestry from the two Thracian-shifted Avar era outliers (Table S15), a finding that may also receive support from f3-statistics (Fig. S2F). Furthermore, models using the Avar era outliers as the sole palaeo-Balkan source for Albania Mdv receive high support (p = 0.72; SE = 0.03; Table S15). However, it is unlikely that the Avar era outliers are realistic local sources due the fact that such significant ancestry shifts would have pulled Albania Mdv samples towards the direction of southeastern Balkan populations on the PCA, which is not the case (Fig. 5A, B). Overall, the PCA and f4 qpAdm statistical models suggest that the Medieval population of Albania was minimally affected by demographic changes during the Roman era, in stark contrast to adjacent regions such as Croatia and Serbia (Fig. 5A). However, currently unsampled urbanised areas in Albania such as Durrës and Shkodër, may have comprised populations of more complex ancestry.
Although we show that some West and East Balkan populations persisted largely unadmixed in late Roman and early post-Roman times (Fig. 5A), such ancestry profiles cannot be found post-900 CE, as Mobest analysis of Albania Mdv shows affinities only to populations from Italy, which maintained a larger proportion of Eastern Mediterranean ancestry compared to contemporary sampled locations in southeastern Europe (Fig. 6B, C). This suggests that the region of modern Albania served as a refugium of Iron Age West Balkan ancestry throughout the demographic and social upheaval that took place during the Migration Period (9, 17, 61). However, our qpAdm models cannot discriminate whether West palaeo-Balkan ancestry in Medieval Albania originated from indigenous populations or other incoming palaeo-Balkan groups from adjacent regions such as the northern Adriatic coast or the Balkan interior. Indeed, IBD-sharing suggests that the sample from Kënetë shares small (10.5 cM) segments with BA and Roman individuals from neighbouring Velika Gruda in Montenegro and Zadar in Croatia, respectively (Table S20) (50). Conversely, the individual from Shtikë shares short segments (11-14.5 cM) with geographically distant Hun and Avar-era commoners from Hungary (Table S20), suggesting that possible trade networks between Medieval Albania and the Avar Khaganate (based on archaeological artefacts) (8, 15, 40) may have also involved genetic exchange.