What population geneticists mean by 'recent selection'—and what it tells us about post-Harappan India

When a population geneticist says "recent selection," they're not talking about your grandparents' generation. They mean evolutionary changes detectable in the genome over the last five to ten thousand years—a timeframe that, in South Asia, overlaps the Harappan urbanization, the arrival of Indo-Aryan speakers, and the spread of agriculture across the subcontinent.

The distinction matters because selection leaves traces in DNA that decay over time. Strong, recent selection creates long stretches of identical chromosome segments around the selected gene—what geneticists call extended haplotype homozygosity. Older selection events blur those signals through recombination, the shuffling that happens each generation. So when researchers scan Indian genomes for selection signatures, they're looking at a specific temporal window, one that happens to align with major cultural and demographic transitions that archaeologists and linguists have argued about for decades.

How the scans work

Selection scans rely on several statistical tests. The extended haplotype homozygosity (EHH) test looks for unusually long stretches of shared DNA around a variant, suggesting it rose to high frequency quickly. The integrated haplotype score (iHS) compares haplotype length between ancestral and derived alleles at the same site. A third approach, the population branch statistic (PBS), compares allele frequency differences across three populations to identify variants that changed rapidly in one branch after it split from the others.

Each test has a different time resolution. EHH and iHS are most sensitive to selection in the last ten to twenty thousand years. PBS can detect older events but works best when the selection happened after population divergence. For South Asia, that means most scans pick up changes since the Holocene—the period when farming spread, cities rose along the Indus, and steppe pastoralists migrated into the northwest.

What shows up in Indian genomes

Lactase persistence—the ability to digest milk sugar into adulthood—is one of the clearest signals. The European variant, tagged by the SNP rs4988235, appears in northern India but at lower frequencies than in Europe. Studies have found this allele in northwest Indian populations, where it likely arrived with steppe-related ancestry sometime after 2000 BCE. The selection signal is strong enough to show up in iHS scans, meaning it rose in frequency recently and rapidly.

Skin pigmentation genes also carry selection signatures. The derived, lighter-pigmentation alleles at SLC24A5 and SLC45A2 are common in northern India and show extended haplotypes consistent with recent positive selection. These alleles are nearly fixed in European populations and present at intermediate frequencies in South Asia, with a north-to-south cline. The pattern suggests they entered the subcontinent with steppe ancestry and were favored, though the selective pressure—whether UV protection, vitamin D synthesis, or something else—remains debated.

Immune loci show more complex patterns. HLA genes, which help the immune system recognize pathogens, are under balancing selection—evolution favors multiple variants rather than one winner. But other immune genes show directional selection. Variants associated with response to Mycobacterium tuberculosis and other pathogens common in dense agricultural settlements appear in PBS scans comparing South Asian populations to African and East Asian ones. The inference is that as people shifted from hunter-gatherer lifestyles to farming and urban living, exposure to crowd diseases changed the selective landscape.

Height-related loci present a puzzle. Genome-wide association studies identify hundreds of variants that collectively influence stature, and some show selection signatures in European populations. In India, the signals are weaker and harder to interpret. Part of the difficulty is that height is polygenic—shaped by many genes of small effect—and selection on such traits leaves subtler traces than selection on single genes like lactase.

The post-Harappan timeframe

The "recent" window captured by selection scans overlaps almost precisely with the period archaeologists call the post-Harappan, beginning around 1900 BCE when Indus cities declined. That's also when genetic studies place the arrival of steppe-related ancestry in northern India, based on ancient DNA from sites like Rakhigarhi and Roopkund.

The coincidence is useful. If lactase persistence or skin pigmentation alleles entered with steppe ancestry and then rose in frequency under selection, the genetic signal provides an independent line of evidence about timing and process—independent, that is, from pottery styles, burial practices, or linguistic reconstructions.

But the overlap also creates ambiguity. Selection takes time. An allele that entered the subcontinent at 2000 BCE and reached, say, thirty percent frequency by 500 BCE might show a selection signal, but pinning down exactly when the selection happened—or how strong it was—requires modeling that depends on assumptions about generation time, population size, and migration rates. Those parameters are uncertain for Bronze Age South Asia.

What doesn't show up

Some adaptations leave no trace in selection scans because they happened too long ago. Skin color variation in Africa, shaped by hundreds of thousands of years of evolution, doesn't show extended haplotypes—the signal has been erased by recombination. Similarly, if an adaptation reached fixation (one hundred percent frequency) long ago, there's no variation left to detect.

Other adaptations may be purely cultural. The spread of rice agriculture across South Asia involved changes in diet, settlement, and social organization, but whether it also drove genetic adaptation—beyond immune responses to new pathogens—is unclear. Selection scans haven't identified strong signals around genes for starch digestion in Indian populations the way they have for amylase copy number variation in some East Asian groups.

The absence of evidence isn't evidence of absence, but it does constrain the narrative. Population genetics can tell us about lactase and pigmentation because the selection was recent and strong. It's quieter about adaptations that were older, weaker, or didn't involve discrete genetic variants.

The stratigraphic view

From a field archaeologist's perspective, selection scans are like dating a buried layer. You get a range, not a point. The signal tells you something happened in the last several thousand years, but pinning it to a century—or linking it cleanly to a specific migration or cultural shift—requires triangulation with other data. Ancient DNA from stratified contexts, isotopic evidence of diet change, and dated skeletal series all help narrow the window.

The geneticists are clear about the limits. Selection scans identify candidates—loci worth investigating. Confirming that selection actually occurred, and why, requires functional studies, ecological context, and historical evidence. The exciting part is that for post-Harappan India, all those lines of evidence are starting to converge.