Made visible · the genome

Three ancestries made South Asia. Then endogamy froze them in place.

Almost every person in the subcontinent today is a blend of three ancient sources — Iranian-plateau farmers, Steppe pastoralists, and the oldest indigenous hunter-gatherers. The mixing happened, then it stopped: a hard turn to marrying-within that locked the blend into thousands of walled groups. The genome records both the mixture and the wall.

The three sources.

what every Indian genome is built from

Ancient DNA has resolved the deep ancestry of South Asia into three components that combined in different proportions to make everyone on the "Indian cline" — the gradient running from the northwest to the south:

Schematic, not to scale. The three ancient sources combine into two mixed groups — Ancestral North Indians (ANI, more Iranian+Steppe) and Ancestral South Indians (ASI, more AASI) — whose blend forms the modern cline. Direction of arrows is established by ancient DNA; proportions vary by region, language and caste. Sources: Reich et al. 2009; Narasimhan et al. 2019; Kerdoncuff et al. 2024.

"Three" is the cline — not everyone.

who the model leaves out

Before going deeper, a correction the headline number hides. The three-ancestry model describes the Indian cline — the north-south gradient that most caste populations and Indo-European- and Dravidian-speakers sit on. It does not describe the whole subcontinent. The Austroasiatic (Munda) and Tibeto-Burman populations are bracketed out — they carry their own East-Asian-related ancestry from separate movements, and they simply are not on the cline the three-source model was built to explain.

Kerdoncuff et al. (2024) itself notes Austroasiatic- and East-Asian-related groups received the Iranian source but keeps them off the headline three; Munda input per Chaubey 2010, Tätte 2018, Rau 2019 (this platform's Munda findings). verified · scite

The oldest layer: one migration, 50,000 years ago.

AASI · the indigenous substrate

Beneath everything is the AASI — Ancient Ancestral South Indians — descended from the people of a single major migration out of Africa around 50,000 years ago, with minimal contribution from any earlier wave. They are distantly related to the Andamanese (the Onge serve as the genetic proxy), and crucially they have never been directly sampled — their existence is reconstructed from the fragments they left in everyone else. This is the deepest indigenous layer, and it is strongest in the south and in tribal groups.

Kerdoncuff, Skov, Patterson et al. (2024), ~2,700 whole genomes — single OoA migration ~50 kya; Yang (2022) review; AASI/Onge proxy per Narasimhan et al. 2019. doi:10.1126/science.aat7487 verified · scite

The Iranian layer, and a correction that matters.

"Iranian-related" ≠ "from Iran"

The second component is Iranian-plateau farmer-related ancestry — and the Kerdoncuff study adds a specific finding: it traces to a common source in the early Neolithic cultures of Central Asia, feeding into the ancestors of ASI, ANI, and the Austroasiatic- and East-Asian-related groups alike. But the standing caution from the earlier work holds and must be stated every time: "Iranian-related" is a statistical label, not a passport. The Iranian-plateau and Indus genomes descend from different hunter-gatherer groups who began farming without large movements between them. It marks a shared deep ancestry, not a caravan from Iran.

Kerdoncuff et al. (2024) — common Central Asian Neolithic source; Narasimhan et al. (2019), Science — the "Iranian-related ≠ migration" qualification. doi:10.1126/science.aat7487 verified · scite

The Steppe layer, and the date the whole argument turns on.

after 4,000 years ago · the Indo-Aryan-linked signal

The third component is Steppe pastoralist-related ancestry — the Yamnaya-derived signal that reached South Asia via Central Asia. Two facts, both cited, both load-bearing:

• The Steppe ancestry in South Asia has the same profile as Bronze Age Eastern Europe — one movement of people affected both regions, and it likely spread the shared Indo-Iranian and Balto-Slavic linguistic features.
• The formation of ANI and ASI as we know them post-dates 4,000 years ago: northern groups on the Indus Periphery cline mixed with Steppe-ancestry populations; southern groups mixed primarily with the AASI. The north-south genetic split is younger than the Indus cities.

Narasimhan et al. (2019), Science — Steppe profile = Bronze Age E. Europe; Yang (2022) — ANI/ASI formation post-4,000 ya. doi:10.1126/science.aat7487 verified · scite

One more thing the honest reader should hold. This research is produced under genuine political pressure — scholars working on it have faced harassment, and the field knows it. But note which way the pressure pushes: nationalism wants the Steppe migration gone, so fear biases toward understating it, never inflating it. That a paper published under that pressure still reports a real Steppe intrusion is a sign the evidence is strong enough to survive the chill — not weak enough to be politics. When you see hedging here, suspect it softens the finding; it does not manufacture one.

Then the mixing stopped.

the endogamy shift · caste written in the genome

The Kerdoncuff study's sharpest contribution is what came after the admixture. Following the mixing, India underwent a major demographic shift toward endogamy — marrying strictly within the group — producing extensive homozygosity and identity-by-descent sharing: long stretches of genome shared because populations descend from small numbers of common ancestors and stopped marrying out. Thousands of groups, each a genetic island.

This is caste, visible in the DNA. The founder events and the walls the caste order built are not an interpretation here — they are measurable in the length and frequency of shared genomic segments. The genome remembers when a community stopped marrying its neighbours.

The sequence: deep indigenous layer (~50 kya) → Iranian-related + Indus admixture → Steppe admixture (post-4,000 ya) → the endogamy shift that froze the blends into caste-bounded groups (intensifying in the last ~2,000 years). Dates per Narasimhan 2019, Kerdoncuff 2024, Yang 2022. The endogamy wall is measurable in genome-wide homozygosity.

The archaic ghost in the Indian genome.

Neanderthal and Denisovan

One striking Kerdoncuff finding, worth its own line: Indians carry ~1–2% archaic ancestry (Neanderthal + Denisovan), and by assembling the fragments across ~2,700 genomes the study recovered ~50% of the introgressing Neanderthal genome and ~20% of the Denisovan — more than any previous archaic-ancestry study — and found Indians have the largest variation in Neanderthal ancestry of any world region. The subcontinent is not only deep in human history; it is a reservoir of the archaic humans our species met and mixed with.

Kerdoncuff, Skov, Patterson et al. (2024). verified · scite

What the genome does NOT tell you.

the ethical guardrail

This is the most important section on the page. Genetic ancestry is not identity, and it does not confirm or deny anyone's oral history, community memory, or sense of who they are.

Biddanda, Bandyopadhyay, de la Fuente Castro et al. (2022) — genetic vs oral histories, a community-led study. verified · scite

What this establishes.

the ledger