Lewontin started it, but responsibility is infinitely divisible, so that’s no excuse. Later practitioners are just as guilty.
He tried to argue away the possibility of phenotypic differences between different human populations using the magic of genetic statistics. Looking at overall genetic variation in humans (why?) most is within-group, rather than between groups. That’s true of dog breeds, as well, which is why they’re all really the same, even though they may seem to vary wildly in size, appearance, lifespan, and behavior.
Which is why Pygmies aren’t really short: they’re faking it.
This whole approach is just nonsense. First, you can sometimes see big between-population phenotypic differences driven by a single SNP: the east-Asian EDAR variant causes many of the morphological characteristics you see in East Asians, and it’s almost fixed. Naturally, that one SNP accounts for an infinitesimal fraction of overall genetic variation.
Selection can also create big phenotypic differences between populations through moderate frequency changes in many alleles of small effect, and that doesn’t result in much between-group genetic variation either. Those tamed foxes aren’t going to be wildly genetically different from their source population, in terms of overall variation. They differ in the genes that influence their tameness, that plus drift from the domestication bottleneck. Mostly this means that the frequencies of alleles that favor tameness has increased: I wouldn’t bet on many hard sweeps, although there would eventually be some, given time. But fox domestication didn’t take much time: true human-symp foxes, ones that jump into your lap, showed up by generation 10. This is a behavior that you never, ever see in wild foxes. Except, sometimes, when they have rabies.
Height is highly polygenic, almost as much so as IQ. Pygmies are about six standard deviations shorter than Europeans or Bantus: what magic principle says that you couldn’t see comparably large differences in IQ? There is none. Although the evidence we suggests that average population IQ doesn’t vary quite that much – probably ranges over something like 3 std, from low to high.
If we knew the alleles that influence height – a fair number, not necessarily all of them – we could probably determine the extent to which genetic differences causes the difference in height between two populations, say between southern and northern Europeans. Height-favoring alleles are definitely more common in northern Europeans, so at least some of the difference in average height is genetic. At any rate, this approach works if the populations are not too divergent.
Now this is perhaps not quite so simple if the local population has a significant number of alleles affecting height that are private, not shared with the populations we have previously studied.
For example, a quite recent study indicates that SNPs associated with height in Europeans contribute to adult height in Pygmies, but a number of other local alleles (not found in Europeans) do as well. I would guess the Pygmy variants affecting growth hormone metabolism are not found in Europeans.
Pygmies are diminutive, obviously, but not in exactly the same way that Europeans are. If you tried to estimate Pygmy height using only the alleles that influence height in Europeans, you’d predict that Pygmies were short, all right, but in real life they’d be a good deal shorter than your estimate – because you’re not counting the private height alleles. You would need to supplement your list of height variants with a huge GWAS study of Pygmies, in order to find those local variants. If you want to find small-effect variants, you’ll probably need to include all living Pygmies.
Assuming that you do your original GWAS study in Europeans, in what other populations would your predictions work decently without local supplementation? Hard to be sure, since there might be local variants of strong effect – but I’d guess that it would probably work on everybody outside of Africa (OK, maybe not Melanesians), might work with the more common kinds of sub-Saharan Africans (Bantus and Nilotics, say) and would be least likely to work with Pygmies and Bushmen, who are the most divergent.
I started out this piece with the aim of kicking Lewontin’s idiotic heirs, but then got interested in the details, how to actually figure something out. When will I learn?