Counting populations from millennia away: how big were ANI and ASI?

When two populations meet and mix, the demographic shape of each affects what the resulting population looks like. A small population mixing with a large one disappears genetically faster than one mixing with a comparable partner. The ancestry-proportion structure of a mixed population depends, in part, on the relative sizes of its parents.

So a question naturally follows from Story 1's framing: how big were ANI and ASI before they mixed? Were they comparable populations meeting on roughly equal demographic terms, or was one substantially larger than the other?

Population genetics has tools to estimate this — indirectly, but well — from the DNA of living people. Apply them to ANI and ASI and you find the populations weren't equal partners.

What "population size" actually means here

Population genetics doesn't directly measure census size — the number of people walking around. It measures effective population size, abbreviated Ne, a statistical quantity that captures how much genetic diversity a population maintains.

Roughly: Ne is the size of an idealized, randomly-mating, non-selected population that would produce the same level of genetic variation as the real population. In practice, Ne is usually 2 to 10 times smaller than census size — partly because real populations have unequal sex ratios, family-based reproductive variance, generational overlap, and population fluctuations that all reduce effective diversity below what naive census-size models would predict.

This sounds like an academic distinction, but it matters. A population with a census of 100,000 spread across many small isolated villages might have a much smaller Ne than a smaller but well-connected population, because the isolation amplifies genetic drift. So when geneticists report Ne estimates, they're capturing something about both raw demography and population structure together.

How Ne is estimated

Several methods, all of which converge on broadly similar numbers when applied to the same data:

Coalescent modeling. Take a sample of modern genomes, identify shared ancestry tracts, and back-infer the population size required to produce the observed pattern of shared ancestry. Tools like PSMC and MSMC do this with different statistical assumptions; both have been applied to South Asian samples ^[S2].

Linkage-disequilibrium decay (the same physics as Story 1's mixing dates). The rate at which linked variants drift apart over generations depends on Ne — smaller populations show faster drift. Apply this principle to specific time slices and you can reconstruct an Ne trajectory.

Identity-by-descent (IBD) segments. When two people share long stretches of identical DNA, those stretches reveal recent common ancestry. The rate of long IBD sharing in a population is a sensitive measure of recent Ne.

These methods don't always agree on absolute numbers — Ne estimates can differ by a factor of two depending on assumptions — but they agree on relative magnitudes and trajectories.

What the methods say about ANI and ASI

The picture that emerges across multiple studies [S1, S2, S3]:

ANI appears to have maintained a substantial effective population size — tens of thousands — through the relevant period. As Article 2 in this series showed, ANI itself isn't one population but a mixture of an Iranian-Neolithic-derived component and a Steppe component, both of which were themselves part of much larger Eurasian metapopulations. The combined ANI lineage was, demographically, well-connected to the Eurasian Bronze and Iron Age world.

ASI appears to have had a substantially smaller effective size — closer to ten thousand or less. The deepest indigenous component of South Asian ancestry, sometimes called Ancestral Ancestral South Indian (AASI) in the literature, shows particularly strong bottleneck signatures: small effective population size sustained over long periods, consistent with relatively isolated hunter-gatherer demographics.

This isn't a small difference. By the time ANI and ASI met, ANI's combined effective population was probably several times larger than ASI's. That has implications.

What this means

Two implications follow from this asymmetry, each of which matters for how we read the demographic history.

First: the fact that ASI ancestry is well-preserved across modern Indian populations — typically 30% to 70% depending on group — is interesting precisely because ASI was the smaller partner. In a simple admixture scenario where a small population is absorbed into a large one, the smaller population's ancestry gets diluted out over generations. The fact that ASI ancestry is robustly present across the subcontinent suggests that whatever mixing happened wasn't a simple absorption. It was reciprocal in some structurally important way — perhaps because the meeting happened at a regional scale where ASI populations were locally numerous, or because subsequent isolation preserved their ancestry contribution.

Second: the demographic asymmetry has implications for how we interpret the cultural and linguistic history. Smaller populations don't easily impose their language or institutions on larger ones unless they hold disproportionate power. The puzzle of whether ANI groups (especially Steppe-derived) imposed Indo-European languages on a much larger ASI substrate is partly a demographic puzzle: it requires either a fairly violent or politically unequal meeting, or a much slower process of cultural absorption. The genetics constrains the demographic shape; archaeology and linguistics fill in the social one.

What's still being refined

Ne estimates depend heavily on assumptions about generation length, mutation rate, recombination patterns, and population structure. Different methods on the same data can give Ne estimates that differ by a factor of two. The Ne trajectories — how population size changed over time — are robust, but the absolute numbers carry uncertainty.

What's solid: ANI and ASI weren't comparable populations. They were demographically asymmetric, with ANI carrying a substantially larger effective population size deep into prehistory. What that meant for the cultural and political dynamics of their meeting is a different question — and one Article 4 in this series gets into.