Selection on height in Sardinians

Evidence of polygenic adaptation at height-associated loci in mainland Europeans and Sardinians:

Adult height was one of the earliest putative examples of polygenic adaptation in human. By constructing polygenic height scores using effect sizes and frequencies from hundreds of genomic loci robustly associated with height, it was reported that Northern Europeans were genetically taller than Southern Europeans beyond neutral expectation. However, this inference was recently challenged. Sohail et al. and Berg et al. showed that the polygenic signature disappeared if summary statistics from UK Biobank (UKB) were used in the analysis, suggesting that residual uncorrected stratification from large-scale consortium studies was responsible for the previously noted genetic difference. It thus remains an open question whether height loci exhibit signals of polygenic adaptation in any human population. In the present study, we re-examined this question, focusing on one of the shortest European populations, the Sardinians, as well as on the mainland European populations in general. We found that summary statistics from UKB significantly correlate with population structure in Europe. To further alleviate concerns of biased ascertainment of GWAS loci, we examined height-associated loci from the Biobank of Japan (BBJ). Applying frequency-based inference over these height-associated loci, we showed that the Sardinians remain significantly shorter than expected (~ 0.35 standard deviation shorter than CEU based on polygenic height scores, P = 1.95e-6). We also found the trajectory of polygenic height scores decreased over at least the last 10,000 years when compared to the British population (P = 0.0123), consistent with a signature of polygenic adaptation at height-associated loci. Although the same approach showed a much subtler signature in mainland European populations, we found a clear and robust adaptive signature in UK population using a haplotype-based statistic, tSDS, driven by the height-increasing alleles (P = 4.8e-4). In summary, by examining frequencies at height loci ascertained in a distant East Asian population, we further supported the evidence of polygenic adaptation at height-associated loci among the Sardinians. In mainland Europeans, we also found an adaptive signature, although becoming more pronounced only in haplotype-based analysis.

The whole literature on selection and height is confused. This is definitely an unformed and new area of exploration, so I wouldn’t put my money on any particular result. But, it is important to note I think that the association of particular genetic variants with differences in height is stronger than the signature of selection on those variants. Second, the preprint is hard to follow because there are all sorts of factors like ascertainment in the huge datasets necessary to do analysis on polygenic traits that date from the way the data were generated in the late 2000s (as well as new datasets coming online).

I think looking at variants in East Asians, and how they impact Europeans, is pretty neat. Obviously, some of the variants that impact polygenic traits are going to be rare, and so not shared between populations, but a lot of it is probably “standing variation” that dates back to before the Out of Africa event. In other words, the key thing is to look at differences in frequencies of alleles which are present in most populations, not different alleles which are not present in all populations.

One element that jumps out at me is the trajectory of selection, and how much is due to events that date deep into the past, to such an extent that it might not make sense to talk about populations as we understand them today. So, for example, they talk about selection events going back to beyond 10,000 years…but all the populations that we survey today did not really exist that deeply in time. This doesn’t mean that selection didn’t happen. “Populations” is a human construct, alleles are alleles. They may have been subject to selection in a variety of populations which admixed themselves out of existence in turn (there was selection for larger brains on and off for millions of years up until about ~200,000 years ago in various hominin populations).

The strongest selection result in this preprint seems to be that something is going on with Sardinians, the most direct descendants of Neolithic farmers. As noted on Twitter I think this has more to do with the nature of calorie restriction, or lack thereof, than selection on height per se. A lot more has to be done on understanding how the “secondary products revolution” (going from simple cereal farming to agro-pastoralism) impacted on human nutrition to understand selection on height, which does seem to be a reoccurring signal across human groups.