Network of large pedigrees reveals social practices of Avar communities

The study addresses how to reconstruct kinship practices and social organization in past societies where archaeological and historical records are fragmentary. Although biological relatedness does not perfectly mirror social kinship, ancient DNA can be a powerful tool to infer elements of past kinship systems, particularly when entire cemeteries are densely sampled to capture the extent of relationships across ancient populations. Focusing on the Avars—steppe-origin groups who dominated eastern central Europe from the late sixth to early ninth centuries AD—the authors note that despite written sources preserving Turkic titles and indicating a Central Asian political structure, direct evidence of Avar social practices has been lacking. By exhaustively sampling four Avar-period cemeteries on the Great Hungarian Plain and integrating genomic, isotopic, archaeological, anthropological and historical data, the authors aim to reconstruct pedigrees, evaluate population structure, and infer kinship practices, mobility, and chronological refinements. This approach seeks to reveal whether Avar communities exhibited patrilineal organization, patrilocality, female exogamy, and other social strategies typical of Eurasian pastoralist steppe peoples and to detect potential community replacements over time.

The paper situates its research within prior work that has used ancient DNA to reconstruct pedigrees and assess social organization in prehistoric and historic contexts, and within historical and anthropological studies of Eurasian steppe societies. It references historical sources documenting Turkic political titles among Avars and comparative anthropological research identifying patrilineal, patrilocal, and exogamous marriage practices among steppe pastoralists. It also notes previous genomic studies of Avar elites and other steppe-descended groups, acknowledging that sparse sampling strategies in earlier work limited the detection of extended relatedness networks across sites. The authors leverage and expand upon these literatures by applying comprehensive, cemetery-wide sampling to recover multigenerational pedigrees and network connections.

• Sites and sampling: Four fully excavated Avar-period cemeteries from the Great Hungarian Plain were selected to represent two main regions divided by the Tisza River: Danube–Tisza interfluve (DTI: Kunpeszér, KUP; Kunszállás, KFJ) and Transtisza (TT: Rákóczifalva, RK; Hajdúnánás, HNJ). At RK, 279 of 308 Avar graves were sampled, plus 56 pre-Avar individuals (2nd–6th centuries) to capture local transitions.
• Genomic data generation: After quality control, genome-wide data targeting approximately 1,240,000 SNPs were obtained for 424 individuals at an average coverage of 2.6×.
• Isotopes and radiocarbon: New 87Sr/86Sr, δ13C, and δ15N isotope measurements were produced for 154 individuals (RK, KUP, KFJ). Fifty-seven new radiocarbon dates were generated for RK. Bayesian modelling of radiocarbon dates incorporated pedigree-derived relative chronological constraints to refine phases and reduce date uncertainties by up to ~60%.
• Kinship inference: Close biological relatedness (first-, second-, third-degree) was identified using KIN software optimized for low-coverage ancient DNA. Pedigrees were reconstructed from dense kinship networks within sites.
• Haplotype IBD networks: ancIBD was used to detect pairwise identity-by-descent segments (>12 cM threshold), constructing networks to assess within- and between-site connections, sex-specific connectivity, and degree centrality. Adults-only subnetworks were analyzed and compared by sex via permutation and statistical tests (Kolmogorov–Smirnov, Welch’s t-test).
• Population genetics: PCA, qpWave/qpAdm admixture modelling, and DATES were applied to estimate ancestry components, admixture proportions, and timing, with a focus on northeastern Asian (eastern Eurasian Steppe) vs. western Eurasian ancestries. Comparisons were made across DTI vs. TT regions and across Avar phases (early, middle, late).
• Y-chromosome and mtDNA lineages: Male-line haplogroups were assigned and contrasted with mtDNA haplogroups to assess sex-biased patterns.
• Archaeological/anthropological contextualization: Spatial mapping of graves, grave goods (e.g., horse harnesses, belt sets), burial constructions, and age-at-death distributions were integrated with pedigree clusters to infer social organization, marriage patterns, and site use over time.

• Dense kinship within sites: No close genetic relatives were detected across sites, but extensive within-site relatedness was observed. A total of 373 first-degree pairs (235 parent–child; 138 siblings) and >500 second-degree pairs were identified, enabling reconstruction of 31 pedigrees (2–146 individuals), some spanning up to nine generations.
• Strict patrilineal descent and patrilocality: Pedigrees overwhelmingly trace male-line descent. Fathers are consistently descendants of founding male(s), while mothers largely lack parents at the site, indicating female exogamy. Spatial clustering of graves aligns with pedigree structure.
• Sex bias and exogamy: In RK, 202 of 276 individuals with genetic data belonged to pedigrees; 64 were unrelated, of whom 51 were female (mostly young adults), consistent with exogamous partners integrated into patrilocal communities.
• Y vs. mtDNA diversity: Marked contrast between limited Y-chromosome lineages and diverse mtDNA lineages. RK pedigrees 1–8 and 12 show Y haplogroups J1a-Z2317 and J2b-CTS11760 only (versus ~50 mtDNA haplogroups). KFJ and KUP predominantly show N1a-Y16220 among males (versus ~20 mtDNA haplogroups). HNJ pedigree 1 shows Q1a-L715.
• Multiple partnerships and levirate unions: Numerous cases of individuals with multiple reproductive partners (RK: 15 male, 7 female cases; also present in HNJ and KFJ). Evidence for levirate unions in RK (5 cases) and KFJ (2 cases), and indirect evidence at KUP, involving father–son pairs, full brothers, paternal half-brothers, and paternal uncle–nephew.
• Avoidance of consanguinity: No long runs of homozygosity consistent with consanguinity were detected, even for distant levels (e.g., second-cousin). Only one inferred related-partner case at sixth-degree, aligning with reported steppe marital rules.
• Female-mediated intercommunity links: IBD networks show adult females have fewer total within-site IBD connections but higher ratios of between-site connections than males (P < 0.05), and seven females (no males) unrelated within site share IBD connections with other sites, indicating female exogamy sustained intercommunity cohesion.
• Community replacement at RK without ancestry shift: Mid–7th-century shift in RK marked by replacement of dominant patriline: early J1a lineages give way to J2b in later pedigrees (3,4,5), mirrored by changes in burial customs and diet (δ13C, δ15N), spatial reorganization of graves, and reduction of horse-harness graves. Despite this, overall ancestry profiles and descent system remained consistent.
• Ancestry structure and admixture: Most individuals (88%) carry northeastern Asian ancestry traceable to the eastern Eurasian Steppe, with varying western Eurasian admixture. Median NE Asian ancestry ranges from ~100% at KUP (DTI) to ~32% at RK (TT). Post-arrival admixture with local Carpathian Basin populations is estimated at ~20%, with much of admixture predating the Avar period.
• Regional structuring: Genetic and network analyses show more intra- than inter-regional connections (DTI vs. TT), with DTI sites (KUP, KFJ) sharing N1a lineages and high connectivity, and TT sites (e.g., HNJ and nearby Hortobágy-Árkus) also closely connected. The Kunbábony elite burial shows the highest number of between-site IBD links among analyzed individuals.
• Isotopic homogeneity: Strontium isotopes are consistent with local/regional values, suggesting high regional continuity and limited evidence for first-generation long-distance migrants being buried at these cemeteries.

The findings reconstruct detailed, multigenerational pedigrees that uncover a social system organized around strict patrilineal descent, patrilocal residence, and female exogamy—features characteristic of Eurasian pastoralist steppe societies. Biological relatedness corresponds closely to archaeological spatial organization and grave-good distributions, indicating substantial overlap between social and biological kinship structures. The observed absence of consanguinity, even with prevalent levirate and multipartner unions, implies adherence to marital rules prohibiting close-agnatic marriages for several generations and suggests robust intergenerational memory of ancestry. Female exogamy emerges as a key mechanism in maintaining cohesion between local communities, as supported by IBD network analyses showing female-biased inter-site connections. The mid–seventh-century RK community replacement—detected through relatedness networks and Y-lineage turnover (J1a to J2b)—is accompanied by shifts in burial customs and diet but not by major changes in genome-wide ancestry, pointing to local political realignment rather than large-scale immigration. Together with isotopic homogeneity, these results challenge hypotheses of repeated large-scale steppe migrations during the Avar period, instead supporting regional continuity with local mobility and evolving social dynamics within the Carpathian Basin.

The study demonstrates that exhaustive, cemetery-wide ancient DNA sampling enables reconstruction of extensive pedigrees that reveal Avar social organization built on patrilineal descent, patrilocality, female exogamy, avoidance of consanguinity, and frequent levirate and multipartner unions. Biological kinship patterns align with burial spatial structure and grave goods, indicating strong overlap of social and biological relatedness. Networks of communities were tightly centered on male descent lines and interconnected primarily via exogamous women. A community replacement at RK in the mid–seventh century—evident from Y-lineage turnover, burial practice changes, and dietary shifts—did not entail an ancestry shift, highlighting how relatedness networks can detect social/political transformations masked by genetic continuity at the ancestry level. The paper does not explicitly outline future research directions.

• Geographic and sampling scope: Despite dense sampling, only four cemeteries were exhaustively analyzed. Cross-site close kinship was not observed, and inclusion of previously published, sparsely sampled sites introduced sampling strategy biases that limited the ability to detect the full extent of inter-site connections.
• Isotopic interpretability: Strontium isotope homogeneity across sites and periods suggests regional continuity but also limits detection of long-distance migrants if their isotopic signatures overlap local baselines; thus, first-generation long-distance migrants may go undetected.
• Inference boundaries: While relatedness patterns indicate a community replacement at RK, absence of skeletal trauma precludes attributing the shift to violent causes. The social interpretation of levirate vs. potential polyandry in some cases remains partly inferential.
• Sex and age structure: The strong sex bias among unrelated individuals (predominantly young adult females) informs exogamy inferences but could also reflect preservation, mortuary selection, or sampling constraints beyond genetic kinship detection.