Author here. Not sure why this turned up on HN today, but feel free to ask questions.
To what extent can we tell this apart from the fact almost every university student leaves their hometown, to attend university?
https://www.biorxiv.org/content/biorxiv/early/2025/05/18/202...
Regional religiousness shows higher genetic correlations with personality (openness and conscientiousness) and less with SES and health traits than political preferences do, which implies additional dimensions of geographic clustering beyond high versus low SES.
Like say everyone tended to marry withing their own class at different points in history, they might tend to be more related to other people with ancestors in similar classes, than in societies that abandoned it.
Not endorsing it but would be interesting if past social systems affected DNA.
There's a famous paper where they map the first two principal components of a bunch of humans and get a map of Europe out.
Interesting!
Can you provide a link to this paper?
But yes the key message is, there is geographic clustering at genetic level.
I know the discussions are politically fraught. But if I understand the summary, your findings lean toward the determinism side. Is that fair? How do you think of the dichotomy? Thanks!
Absolutely not. I don't think any serious geneticist is a genetic determinist, in fact it's hard to even know what that means... DNA without an appropriate environment is nothing but a long stringy molecule!
In fact, the main impact of this paper was to help make geneticists aware that genes are confounded with geographic environments. That (plus much other research!) is one reason why researchers are now putting a lot of emphasis on family-based designs. In those, you can get truly causal estimates of the effect of a genetic variant or of a whole polygenic score, due to the "lottery of meiosis" that randomly give you genes from either your mum or dad.
Now you could equally argue that the paper shows geographic environments are confounded with genes. That's true too, though sadly a lot of social science still proceeds as if it wasn't the case.
(wasn't suggesting that)
However, the randomness is great enough that first cousins do share approximately 1/8 of their genes, second cousins 1/32, third cousins 1/128 and so on. The rapid decrease of shared genes among relatives in successive generations also means that if mating in a population is really random, the population rapidly becomes homogeneous.
It is only assortative mating which generates and maintains distinctive subpopulations.
Whatever genetic signal and influence is found between groups A and groups B, A and B are so highly variable within themselves that none of this is applicable to comparing individuals beyond a perhaps slightly weighed dice.