Hard sweeps and natural selection obscured by Bronze Age admixture

The above is the map from the Online Ancient Genome Repository. You can see the variation by region. There’s a lot of ancient DNA in Europe. Very little in Asia. And only moderate amounts elsewhere.

The map is from a new preprint, Ancient human genomes reveal a hidden history of strong selection in Eurasia:

The role of selection in shaping genetic diversity in natural populations is an area of intense interest in modern biology, especially the characterization of adaptive loci. Within humans, the rapid increase in genomic information has produced surprisingly few well-defined adaptive loci, promoting the view that recent human adaptation involved numerous loci with small fitness benefits. To examine this we searched for signatures of hard sweeps – the selective fixation of a new or initially rare beneficial variant – in 1,162 ancient western Eurasian genomes and identified 57 sweeps with high confidence. This unexpectedly extensive signal was concentrated on proteins acting at the cell surface, and potential selection pressures include cold adaptation in early Eurasian populations, and oxidative stress from carbohydrate-rich diets in farming populations. Critically, these sweep signals have been obscured in modern European genomes by subsequent population admixture, especially during the Bronze Age (5-3kya) and empires of classical antiquity.

So the “big thing” that they found here is that admixture obscures signals of selection. More precisely, it obscures signals of hard selective sweeps, the classical variant where a single position in a single haplotype rises up in frequency rapidly due to positive selection.

If you read further into the paper you note that they believe admixture, due to the mixing of backgrounds, attenuates the signal of hard sweeps, and may even imply that these hard sweeps are soft sweeps through the mixing of distinct genetic backgrounds. I honestly didn’t follow that too closely, but I guess it depends on the selection coefficient and rate of mixing. They are reporting lots of selection events of >1%, and I wonder about how credible this is (Haldane’s dilemma?).

That being said, the functional significance of these selection events is important. Basically, they look like adaptations to climate and changes in diet. What authors seem to be suggesting here is that the shift in lifestyle and expansion of farmers in the early Holocene was a pretty big deal, and the mixing between various divergent streams during the Bronze Age muddled the signals.

If the authors are right, that means that ancient DNA is going to be very big for understanding the trajectory of selection, because it’s not just going to be subtle polygenic changes.