Widespread horse-based mobility arose around 2200 BCE in Eurasia

The study addresses when and how domestic horses became integrated into widespread human mobility and whether horses drove major human migrations in Eurasia, notably the Yamnaya expansion. DOM2 (modern domestic horse lineage) emerged in the western Eurasian steppes during the third millennium BCE, with archaeological indicators such as Sintashta chariots suggesting horse-based mobility developed no earlier than late third millennium BCE. Contradictory claims include proposed DOM2 ancestry in horses from Corded Ware contexts, osteological signs of horsemanship in steppe-related individuals, and shared Indo-European horse terminology implying earlier equestrianism. Evidence from Botai (central Asia, ~3500 BCE) suggests horse-focused subsistence, but interpretations of milk consumption, harnessing, and management practices are debated. The study aims to test for synchronous shifts in horse and human ancestry, reconstruct horse demographics for domestication bottlenecks, and detect markers of controlled reproduction (close-kin mating and shortened generation times) to clarify the timing and scale of horse husbandry and mobility.

Prior work indicates DOM2 originated in the western Eurasian steppes and spread alongside technological innovations like Sintashta chariots (c. 2200–1800 BCE), suggesting late third millennium BCE onset of long-distance horse mobility. Other studies have proposed earlier DOM2 genetic influence in Europe (e.g., in Corded Ware horses) and osteological evidence for horsemanship among Yamnaya and even pre-Yamnaya individuals. Linguistic evidence of shared horse-related terms across Indo-European languages has been interpreted as signs of established steppe equestrianism. At Botai (~5500 years ago), archaeological evidence for harnessing and horse milk use has been reported, though challenged by subsequent analyses; dental wear interpreted as bit wear and stable sex ratios of assemblages have been questioned. DOM2 and Botai represent genetically distinct lineages, complicating inferences about early domestication and use. The paper revisits these debates with expanded genomic time series and integrated population genetic modeling to test prior claims (e.g., lack/presence of DOM2 in CWC horses) and to contextualize archaeological/linguistic interpretations.

• Assembled and analyzed 475 ancient horse genomes spanning >50,000 years across Eurasia, plus 77 modern horse genomes (40 domestic breeds and 6 Przewalski’s horses). Generated 124 new genomes (median 1.40× coverage) and obtained 401 radiocarbon dates (140 new), creating a dense time-stamped series across the domestication process.
• Population structure and ancestry: Used Struct-f4 (K=9) to infer ancestry components and profiles; performed Locator analyses to predict geographic origins of specimens; built population graphs including additional groups to test gene flow scenarios; applied multidimensional scaling and qpAdm mixture modeling to model ancestry proportions and test steppe contributions to European horses.
• Temporal and spatial analyses: Compared ancestry profiles before vs after 2150 BCE in central Europe, Carpathian and Transylvanian Basins, and Anatolia; mapped steppe-related ancestry presence over time and space to distinguish natural dispersal vs husbandry-driven movements.
• Demographic inference: Reconstructed effective population size trajectories (N) using linkage disequilibrium-based methods (GONE) over 200 generations prior to ~1864 BCE, focusing on early DOM2 and Sintashta-associated horses. Quantified demographic bursts and bottlenecks.
• Inbreeding: Assessed runs of homozygosity (ROH) across size categories (≥1 cM bins up to ≥15 cM) to detect close-kin mating signals over time and across regions.
• Generation time estimation: Developed two single-genome proxies:
  1. Mutation clock: estimated generations since divergence from outgroup(s) using accumulated mutations.
  2. Recombination clock: inferred generations since the MRCA from recombination patterns. Validated via coalescent simulations under varying inbreeding and demography. Applied to all radiocarbon-dated genomes to estimate historical generation times, and used GAM regressions to model temporal changes while controlling for depth and population structure.
• Comparative analyses: Evaluated Botai (~3500 BCE) and descendant group (Borly4) for demography, ROH, and generation times; compared DOM2 vs non-DOM2 relatives (Yamnaya/Turganik/Steppe Maykop contexts, older steppe and Carpathian/Transylvanian horses).

• No evidence for DOM2 ancestry in Corded Ware (CWC) horses: • Locator and improved population graphs reject DOM2 contribution to CWC horses (P < 1e−5 for better-fitting graphs without DOM2). CWC horses cluster with pre-Yamnaya central European lineages (ENEOCZE, NEOPOL). qpAdm two-way models: CWC ≈ 32.4% ENEOCZE + 67.6% northern European (FBPWC); adding a third source yields ≤1.7% steppe contribution. • A distinct ancestry cline peaks in CWC horses and declines westwards and eastwards; this CWC-related ancestry persists in European tarpans (~45.1%) until ~1868 CE but is largely absent in modern domestic horses.
• Timing of DOM2 spread and replacement: • First typical DOM2 in Carpathian Basin ~1822 BCE (1895–1749 BCE); last local HUNG profile ~2033 BCE (2120–1945 BCE). Regional turnover narrowed to ~2033–1945 BCE, consistent with earliest DOM2 outside steppe: Moldavia ~2063 BCE, Anatolia ~2125 BCE, Czechia ~2037 BCE—post-dating human steppe ancestry arrivals by ≥600 years. Thus, Yamnaya migrations (~3000 BCE) and DOM2 spread are chronologically incompatible.
• Steppe-related horse ancestry in Europe is ancient and natural: • Steppe-like ancestry occurs in Europe long before domestication claims, found in Austria (~3300 BCE), Transylvanian Basin (~4200 BCE), Poland (~5210–5006 BCE), Anatolia (~6396–4456 BCE), Iberia (~5299–1900 BCE), and Upper Palaeolithic France (~21909–12090 BCE), implying natural post-LGM dispersals rather than husbandry.
• DOM2 demographic burst and bottleneck: • LD-based reconstructions show ~13.7-fold N increase within 30 generations before ~1864 BCE. Using average generation time 8 (7–12) years, widespread horse-based mobility rose ~2190 BCE (2310–2160 BCE). Sintashta-only subset yields ~2100 BCE (2200–2075 BCE). • A strong domestication bottleneck with minimal N ≈ 500 diploids starts ~2664 BCE (3064–2564 BCE), indicating plummeting availability of breeding horses during steppe migrations.
• Reproductive control indicators: • Early DOM2 show first long ROH (≥15 cM), consistent with close-kin mating in Central Asia and Anatolia—behavior avoided in wild but common in managed breeding. • Generation time contraction: abrupt ~2.1-fold acceleration to ~3.5 years around 2200–2100 BCE (from long-term average ~7.4 years). This shift affects DOM2 but not contemporaneous non-DOM2 relatives (Yamnaya/Turganik/Steppe Maykop) or older regional horses.
• Botai husbandry signals: • In 36 Botai horses: no close-kin mating evidence (no long ROH), but markedly shortened generation times, comparable in magnitude to DOM2 breeding, specific to Botai and descendant Borly4 (~3000 BCE). • Botai population experienced ~2.4-fold expansion starting ~80 generations before settlement (~4140 BCE; 4460–4060 BCE with 8-year gen time), matching humid, open steppe conditions; followed by decline over last 20 generations (~140–240 years), consistent with a prey-pathway domestication response (corralling and reproductive control via earlier breeding) to resource depletion.
• Broader context: • The rise of widespread horse-based mobility (~2200 BCE) aligns with earliest DOM2 outside steppe, earliest Akkadian horse imagery, and evidence of conflicts/crises from Balkans to Egypt and the Indus, and the 4.2 ka aridification event, suggesting increased demand for mobility, grazing, and territorial control. • Generation times accelerated again in the last ~200 years with modern breeding; race breeds (e.g., Thoroughbreds) show less acceleration due to extended reproductive lifespans.

The data refute the hypothesis that domestic horses (DOM2) facilitated Yamnaya or earlier steppe-related human migrations into Europe. Genomic continuity of local horse lineages in central and southeastern Europe persists until the end of the third millennium BCE, and DOM2 expansion post-dates the Yamnaya horizon by centuries. Steppe-like horse ancestry in Europe predates domestication by millennia, indicating natural dispersals rather than managed movement. The precise timing of a severe domestication bottleneck (~2664 BCE) followed by a rapid demographic surge and reproductive control practices (close-kin mating and accelerated generation times around 2200–2100 BCE) marks the onset of widespread horse-based mobility. These practices likely underpinned the rapid Eurasian spread and replacement of local horse lineages. At Botai (~3500 BCE), the combination of shortened generation times without close-kin mating, a prior population expansion under favorable climate, and a subsequent decline is consistent with a regional husbandry focused on meat procurement via the prey domestication pathway rather than long-distance mobility. Overall, the results disentangle horse domestication, management strategies, and human mobility narratives, placing large-scale equestrian mobility in the late third millennium BCE and highlighting the role of deliberate reproductive control in enabling rapid expansion. Implications extend to archaeological interpretations of technology (chariots), socio-political changes (conflicts and territorial expansion), and linguistic dispersals, suggesting that while equestrian practices could have existed earlier, they were not demographically or geographically extensive enough to drive major human genetic turnovers prior to ~2200 BCE.

This study compiles an unprecedented time series of ancient horse genomes to date the emergence of widespread horse-based mobility to around 2200 BCE, contemporaneous with the rapid expansion of DOM2 and the adoption of strong reproductive control (close-kin mating and shortened generation times). It demonstrates no substantive DOM2 involvement in Corded Ware or earlier European horses, and shows that steppe-related horse ancestry in Europe reflects ancient natural dispersals. At Botai, evidence supports localized husbandry with accelerated reproduction but without inbreeding, consistent with the prey domestication pathway. Methodologically, the work introduces and validates genome-based mutation and recombination clocks to infer historical generation times, expanding the toolkit for detecting reproductive control across species and time. Future research directions include applying these methods to other domesticates and hominins to assess how environmental or cultural shifts affected generation times, and integrating broader archaeological datasets to refine the interplay between animal management, climate events, and human mobility.