Stumbled on a provocative claim in Nick Lane’s Power, Sex, Suicide: Mitochondria and the Meaning of Life, which was originally published in 2005 but was then updated in 2018:

“Much of our internal heat is generated by dissipating the proton gradient across the mitochondrial membranes (see page 183). Since the proton gradient can either power ATP production or heat production, we are faced with alternatives: any protons dissipated to produce heat cannot be used to make ATP. (As we saw in Part 2, the proton gradient has other critical functions too, but if we assume that these remain constant, they don’t affect our argument.) If 30 per cent of the proton gradient is used to produce heat, then no more than 70 per cent can be used to produce ATP. Wallace [Douglas C.] and colleagues realized that this balance could plausibly shift according to the climate. People living in tropical Africa would gain from a tight coupling of protons to ATP production, so generating less internal heat in a hot climate, whereas the Inuit, say, would gain by generating more internal heat in their frigid environment, and so would necessarily generate relatively little ATP. To compensate for their lower ATP production, they would need to eat more. Wallace set out to find any mitochondrial genes that might influence the balance between heat production and ATP generation, and found several variants that plausibly affected heat production (by uncoupling electron flow from proton pumping). The variants that produced the most heat were favoured in the Arctic, as expected, while those that produced the least were found in Africa.”

These claims look a bit too good to be true and clearly come from the candidate gene era, but thermogenesis due to uncoupling is not necessarily a complex trait but possibly a narrowly defined cellular phenotype. Plus, I imagine the amount of room for biologically meaningful variation is likely quite low given the importance of bioenergetics. There is also other evidence that bioenergetic traits have been subject to selection effects over time. So these claims may be plausible.

I’ve tried to briefly checked them out:

Are there robust differences in the expression of uncoupling protein across populations that experience climates with different temperatures consistent with a thermal-adaptation model?

The unsurprising answer is no. If we go to a population database that samples different continental ancestries like gnomAD, we can see that the most promising lead from the candidate gene area studies (rs1800592) actually shows a rough reversal of the frequencies expected for the thermal adaption hypothesis. Compare the left panel of the figure below from a 2010 study of 1000 people (PMID:20802238) to the gnomAD continental frequencies.

This doesn’t rule out bioenergetics-related adaptation based on local climates and caloric availability. I believe the recent Reich pre-print found some evidence of this in ancient West Eurasian populations, but it underscores the importance of not being seduced by simple biological stories when even seemingly simple physiological stories are plausible.