Evidence page · the third arrival
Eleven million people speak a language their fathers brought and their mothers never spoke.
The Munda languages — Santali, Ho, Mundari, Korku, Bhumij — are the westernmost branch of Austroasiatic, a family whose centre of gravity is Southeast Asia. Their speakers live in Jharkhand, Odisha and West Bengal, surrounded by Indo-European and Dravidian. Close to 100% of their maternal lineages match other Indian populations. Around 65% of their paternal heritage is closer to Southeast Asian than to South Asian variation. This is the third male-biased arrival in the South Asian record — and unlike the other two, its descendants ended up at the bottom of the order rather than the top.
The split record.
two lines, two continents
When, and how many.
the founders
what the fathers brought against what the mothers kept
| Finding | Detail |
|---|---|
| Admixture dates | 2,000–3,800 years ago, varying by Munda population — from genome-wide data on 102 Munda speakers, not from uniparental markers alone. |
| Lineage coalescence | Indian O2a-M95 lineages sit within a clade nested inside East and Southeast Asian variation, coalescing in the last 5,000–7,000 years — the upper bound for the main gene-flow episode. |
| Four founders | Four downstream branches — B418, B419, B426, M1284 — account for around 77% of all O2a-M95 carriers in South Asia, coalescing about 5,000 years ago. 59% of Mundari and 62% of Khasi men assign to them. |
| And they sort by group | B418 is most frequent in North Munda communities and absent among Manipuri populations. B419 is virtually absent among Mundari speakers and found almost exclusively among Tibeto-Burman speakers in Manipur and Bangladesh. Different founding men, different modern communities. |
The whole page is an asymmetry between two records. Reading them side by side, sortable, is the argument.
Seventy-seven per cent of a lineage carried by millions descends from four men. That is the founder-effect signature this platform documents elsewhere — the same shape as the Vysya bottleneck, the same shape as the steppe expansion, produced by the same mechanism of patrilineal descent groups concentrating reproduction.
The direction is contested, and honestly so.
a live scientific dispute
| Position | Evidence for | Difficulty |
|---|---|---|
| Into India from Southeast Asia | Indian O2a-M95 nests within East/Southeast Asian variation; young coalescence in India; genome-wide admixture dates of 2,000–3,800 years; the Austroasiatic family's centre of diversity is Southeast Asia. | Must explain the high mtDNA diversity in Munda populations — resolved by the sex-biased model. |
| Out of India to Southeast Asia | An independent estimate places O-M95's origin among Indian Austroasiatics at ~65,000 years ago; high maternal diversity in Munda groups; consistent with the coastal route out of Africa. | The nested phylogeny is difficult to reconcile with an Indian origin, and STR-based dates have been superseded by whole-Y sequencing. |
The into-India reading is better supported by the current evidence: high-resolution Y sequencing places the Indian lineages inside a Southeast Asian clade, and genome-wide admixture dating agrees with it. The 65,000-year figure rests on short-tandem-repeat variation, a method the whole-chromosome phylogenies have largely displaced.
What would change the assessment: ancient DNA from eastern India and mainland Southeast Asia across the relevant window. Almost none exists. The disagreement is methodological — which molecular clock to trust — not ideological, and it is worth saying so, because arguments about direction of movement in South Asia usually are not.
The comparison that matters.
three arrivals, three outcomes
| Arrival | From | When | Became | Rank today |
|---|---|---|---|---|
| L1-M22, J2 | Zagros / Iranian plateau | 8,000–6,000 BP | The agricultural population; with AASI, the Indus | Middle — farming and landholding castes |
| R1a-Z93 | Steppe via Sintashta | 4,000–3,000 BP | The priestly stratum | Top |
| O2a-M95 | Southeast Asia | 3,800–2,000 BP | Munda-speaking communities | Scheduled Tribes |
Every one of these is a male-biased arrival that brought its language and married local women. If arrival itself conferred status, the Munda would not be where they are. The Santal, Ho, Munda and Korku are Scheduled Tribes — dispossessed of forest and land, among the poorest communities in India, and the population at the centre of the Jharkhand and Odisha mining displacements.
So rank was not set by who arrived. It was set by what the arrivals arrived into, and what they subsequently did. The steppe-derived lineage entered a world where ritual authority was becoming heritable property, and captured it. The Munda lineage entered forest and upland country and stayed there — and when the state reached that country, it classified them the way it classified everyone on marginal land.
What the language actually is.
not a dialect, not a survival
Munda is a full branch of Austroasiatic, a family of over 150 languages stretching from India to Vietnam. Santali alone has around seven million speakers — more than Welsh, Icelandic and Estonian combined — and is one of the twenty-two languages scheduled in the Indian Constitution, written in Ol Chiki, a script created for it in the 1920s by Raghunath Murmu.
The Austroasiatic family's other South Asian branch is Khasi, in the Meghalaya hills — matrilineal, and still intact. Two branches of one family: one in the eastern uplands practising descent through women, one in the Chota Nagpur plateau. The Khasi carry O2a-M95 at 62% assignable to the same four founders. The paternal line is shared; the inheritance systems are not.
The honest limits.
- Sources: the ~65% paternal and ~100% maternal figures, O2a frequencies of 19–95%, and the father-tongue characterisation from Chaubey et al. and Tätte et al. 2019 (Scientific Reports, 102 Munda genomes, doi); the four founder branches and their group distribution from the O2a-M95 phylogeny study; the ~65,000-year Indian-origin estimate from Kumar et al. (BMC Evolutionary Biology, 1,222 individuals, 25 populations).
- The direction of movement is unresolved. This page reports both positions and states which the current evidence favours; it does not treat the question as closed.
- Y-haplogroup frequencies are not genome-wide ancestry. The 65% figure describes paternal lineages; the genome-wide Southeast Asian component in Munda populations is a separate and smaller measurement.
- "Arrival" describes gene flow, not an event. The dates span 1,800 years across different Munda populations and describe admixture, not a migration with a start and end.
- Munda communities are not treated here as a proxy for any ancient population.
The claims, and the direction that is contested
The page says its own direction of inference is disputed. The inspector marks which of its statements are inference rather than observation.