Caste Endogamy's Genetic Signature: How Far Back Do the Founder Effects Go?

Ask a population geneticist when caste became a genetic boundary, and the honest answer is: the DNA can tell you when people stopped marrying outside a group. It cannot tell you why they stopped, or what anyone called the rule at the time.

That distinction sits at the center of two studies that have shaped how geneticists talk about endogamy (marriage within a defined group) in South Asia: Moorjani et al. 2013 in the American Journal of Human Genetics, and Narasimhan et al. 2019 in Science. A third paper, Nakatsuka et al. 2021 in Cell, pushed the analysis further into jati-level (endogamous sub-caste) founder events. Together they give a rough, and genuinely surprising, timeline — while stopping well short of what "caste hardened around year X" headlines have sometimes claimed.

What a founder event actually looks like in the genome

When a population shrinks to a small breeding group and then stays closed — no new members marrying in from outside — a statistical signature builds up. Everyone in that group starts sharing unusually long stretches of DNA inherited from the same handful of ancestors. Geneticists call this a founder event, and it's the same phenomenon that produces elevated rates of certain recessive diseases in populations like Ashkenazi Jews or Finns.

Moorjani et al. (2013) used a method based on the decay of "linkage disequilibrium" — the tendency of nearby genetic variants to be inherited together, which breaks down at a predictable rate each generation once mixture stops — to estimate when major ancestral groups in India stopped mixing with each other. Their estimate placed the end of large-scale admixture between what the paper's ancestry framework calls Ancestral North Indian (ANI) and Ancestral South Indian (ASI) populations at roughly 1,900 to 4,200 years ago, with most groups clustering toward the more recent end of that window. After that point, the paper found, groups across India became "strongly endogamous" — that is, marriage stopped crossing those ancestry lines in most of the populations they sampled.

Narasimhan et al. (2019), a much larger ancient-DNA study spanning Central and South Asia, extended this picture. It documented the demographic events that produced the ANI/ASI mixture in the first place — the arrival and blending of Steppe pastoralist ancestry with earlier South Asian populations — and reinforced that widespread admixture across social groups gave way to endogamy in roughly the first millennium BCE, a shift the authors link to the period described in later Indian textual tradition as the era of formalized varna and jati rules, without claiming the genetics prove that connection.

Jati by jati, the picture gets sharper — and stranger

Nakatsuka et al. (2021) is the paper most directly relevant to the pitch's question. Working with a larger set of present-day Indian populations organized by jati and language group, the authors found that founder events are not evenly distributed. Some jatis show founder events with effective population sizes small enough to produce measurable elevated risk for specific recessive diseases — a pattern the paper compares explicitly to Finland and Ashkenazi Jewish populations, both well-documented founder populations in medical genetics.

Crucially, the paper found that the strength of a jati's founder signature does not track cleanly with caste rank, geography, or language alone. Some groups traditionally placed low in caste hierarchies show very strong founder effects; some high-ranked groups show weaker ones. The determining factor, as the authors frame it, is simply how small and how closed the marrying population was and for how long — which is a demographic fact, not a moral one.

What the data doesn't say

None of these papers can date when the word "jati" or the rules of endogamy were first written down, discussed, or enforced — that is a question for textual history, not genomics. Sanskrit textual tradition places prescriptive statements on marriage and varna in texts like the Manusmriti, but dating those texts precisely, and knowing how widely their prescriptions were actually followed on the ground, is a separate scholarly problem that DNA cannot settle.

There's also a statistical limit worth naming plainly: linkage-disequilibrium dating methods estimate when large-scale mixture stopped, averaged across many generations. They are not a precise calendar date. A founder event dated to "around 2,000 years ago" carries a wide confidence interval, and different populations within that same broad estimate could have closed off decades or centuries apart from each other.

So the responsible summary is narrower than it sounds in headlines: the genomic data show that most sampled Indian populations transitioned from widespread admixture to strong endogamy sometime in the last few thousand years, most heavily clustered in the first millennium BCE, and that by the time these jatis appear in the genetic record today, many have been endogamous long enough to produce founder effects comparable to other well-studied isolated populations worldwide. Whether that closure was sudden or gradual, uniform or patchy, driven by ritual prescription or by ordinary social inertia — those questions remain open, and the DNA alone won't close them.