Iranian farmers and Steppe horsemen: why ANI is two ancestries, not one

In Story 1 we said that almost every Indian alive today is a mixture of two ancestral populations, ANI and ASI ^[S5]. That framing is correct and useful, but it leaves an obvious puzzle the dating itself surfaced: why did Dravidian-speaking populations in the south finish their mixing roughly a thousand years before Indo-European-speaking populations in the north?

The cleanest answer comes from realizing that ANI isn't really one population. It's itself a mixture — of two ancestral components that arrived in South Asia at very different times.

The first wave: Iranian Neolithic farmers

Around 8,000 to 9,000 years ago, farming spread eastward from the Fertile Crescent across the Iranian plateau. By roughly 7,000 years ago, populations carrying Iranian Neolithic farmer ancestry had reached Mehrgarh, in present-day Balochistan — the earliest known farming community in South Asia ^[S2].

Over the next several millennia, these populations mixed with the indigenous South Asian populations already present in the subcontinent — populations whose ancestry traces back tens of thousands of years before farming. The result: a "core" South Asian gene pool with Iranian Neolithic ancestry mixed with much older indigenous ancestry ^[S1]. The Indus Valley Civilization, when it arose around 4,500 years ago, was built by populations descending from this earlier admixture.

The cleanest single-individual data point is the 2019 Rakhigarhi genome — a Mature Harappan individual from a site in present-day Haryana, dated to about 4,500 years ago. The profile: Iranian-Neolithic-related ancestry mixed with deeply rooted indigenous South Asian ancestry, with no detectable Steppe ancestry ^[S4]. Whatever the Indus Valley Civilization was, the people who built it carried only the first wave.

The second wave: Bronze Age Steppe pastoralists

A second wave arrived later. Starting around 4,500 years ago and intensifying through 4,000 to 3,500 years ago, populations from the Eurasian Steppe — the open grasslands stretching across present-day Kazakhstan and southern Russia — began moving south through Central Asia ^[S3]. These were mobile pastoralist groups, archaeologically associated with the Yamnaya cultural horizon and its descendants. They carried a distinct genetic signature — the same Steppe Middle/Late Bronze Age signal that, around the same time, also spread west into Europe.

By approximately 3,500 to 3,000 years ago, populations descending from this Steppe wave had reached Northwest South Asia — the region that today is Punjab, Haryana, and parts of Pakistan.

Steppe ancestry today is not evenly distributed across India. It is highest — typically 15 to 25 percent of total genome — in northwestern populations, especially Indo-European speakers. It is lower in the Hindi belt, Bengal, and Maharashtra; lower still in the Dravidian-speaking south; and very low in tribal communities and isolates ^[S1].

Resolving the puzzle

Now Story 1's thousand-year gap resolves cleanly.

The ANI ancestry that ended up in South Indian populations came mostly from the first wave — Iranian Neolithic farmer ancestry, which had been mixing with deep indigenous populations for thousands of years before the Steppe wave arrived. By the time Steppe ancestry reached the subcontinent, the South's mixing was largely complete. The Moorjani LD-decay analysis, applied to Dravidian-speaking populations, picks up that earlier event.

The North got both waves. The first wave mixed with indigenous populations to form the Indus-type gene pool. Then, around 4,000 to 3,500 years ago, Steppe ancestry arrived and admixed on top — and that admixture continued for centuries. The Moorjani LD-decay analysis on Indo-European populations picks up the more recent, Steppe-driven event, which by their estimate finished around 2,000 years ago.

In short: the South finished mixing earlier because most of its mixing was effectively complete before the Steppe wave arrived. The North finished mixing later because there was a second wave to mix in.

How we know

The disentangling of ANI's two parental sources is one of the firmer findings in the field, anchored in multiple independent lines of evidence.

The most direct is ancient DNA itself. Sample individuals from before the Steppe wave — like Rakhigarhi and the broader Indus Periphery cluster from Narasimhan 2019 — and you find Iranian Neolithic plus indigenous ancestry, with no Steppe component. Sample individuals from after — Bronze Age and Iron Age individuals from sites in the Swat Valley and elsewhere in northwestern South Asia — and you start seeing Steppe ancestry layered on top. The temporal sequence is direct, not inferred.

The statistical methods used to model these populations — qpAdm, qpGraph, ADMIXTURE — let researchers test specific demographic hypotheses against the data. The Narasimhan 2019 paper presents extensive supplementary analyses using all of these. Different methods, different sample subsets, the same broad picture.

What remains under active research is finer detail. How fast did the Steppe wave actually move? Was it a single migration or sustained pastoralist contact? How much of the linguistic transformation often associated with Indo-European arrival maps to the genetic signal versus happening through cultural diffusion without migration?

But the basic two-source picture for ANI — and the temporal separation between Iranian Neolithic and Steppe components — is one of the things the field has converged on. The thousand-year gap that Story 1 surfaced as a puzzle is, at this point, a puzzle the data answers.