David Reich’s book has stirred up a lot of the usual suspects. Most are non-geneticists, but there are a few geneticists among them. Sorta kinda. Here’s an interesting example – something by Dan Graur, a geneticist currently at the University of Houston.
Graur says: “For selection to operate and counteract the effects of random genetic drift, the effective population size should be large. Unfortunately, the long-term effective population size for all the humans in the world is barely 10,000—lower than that of chimpanzee. By necessity, the effective population size of each “race” separately is much smaller. So, the chances that 74** loci will experience significant changes in allele frequencies simultaneously in each of the four populations is zero. With one locus, a change in human allele frequency may occur, albeit very rarely, as evidenced by the case of lactase persistence in North European and the case of the EDAR allele in East Asians. To imagine that 74 alleles change frequency in concert and that the 74 alleles have successfully battled genetic drift and recombination in merely 2,000 generations requires an extremely naïve and unsophisticated view of the evolutionary process.
Moreover, according to Augustine Kong, whom David Reich quotes, educational attainment is a deleterious trait that is selected against. The selection against educational attainment, according to Kung, is extremely powerful, so much so that differences in allele frequencies are observable after merely 3-5 generations (100 years). If one believes these claims, one should explain how come there are so many university graduates in Iceland. With such huge purifying selection against education, it’s a miracle that any one in Iceland knows how to read and write.
Kong A. et al. 2017. Selection against variants in the genome associated with educational attainment. PNAS 114: E727-E732
Moore DS, Shenk D. 2017. The heritability fallacy. WIREs Cogn Sci, 8:e1400. doi: 10.1002/wcs.1400
Okbay A et al. 2016. Genome-wide association study identifies 74 loci associated with educational attainment. Nature 533:539–542 ”
The interesting thing about this is that Graur has had prestigious jobs in areas relating to evolution, while apparently understanding nothing about it. Graur served as associate editor of the journal of Molecular Biology and Evolution from 1995 to 2011. Since 2009, he serves as associate editor of the journal Genome Biology and Evolution. From 2009 to 2011 he has held the position of Councillor for the Society for Molecular Biology and Evolution, won the Humboldt prize in 2011, and was elected a fellow of AAAS in 2015. In spite of this, every single statement in this short note is incorrect.
Graur says “For selection to operate and counteract the effects of random genetic drift, the effective population size should be large.” Well, it would be nice if the effective population size were large, but that is by no means necessary. German Shepherds were created in the 1890s, by interbreeding several breeds of dogs ( and wolves !), followed by selection for desired traits. That didn’t require tens or hundreds of thousands of dogs. Thoroughbreds are the fastest horses in existence: their effective population size is around 100. Dachshunds are smaller than they were in the 1970s – but then everybody knows they’re contrary enough to violate the laws of genetics. Born that way, probably.
I could go on – and on, and on. Every example we have of selection on domesticates is a counterexample to what Grauer is saying here. No farmer he. Does he think that some ancient geneticist corralled a million aurochsen in order to breed the domesticated cow?
A large population size is nice, because it generates more favorable mutations, but that is not necessary. Selection, at least in the short run, does not require any new mutations at all. Nice, because a large population size reduces the strength of chance, so the probability of the change in the trait under selection stalling or reversing in a given generation is lower – the results of selection become more predictable, especially in the short run. Next, Grauer talks about the long-term effective population size (EPS) in humans being around 10,000 – which he seems to think is too low for selection to work. He’s wrong, but that number is also the wrong number. He’s talking about the neutral-theory effective population size, which is a function [ harmonic mean] of long-term population size over the last million years or so – and it’s the wrong ‘effective population size’ for selection questions. There are different values of EPS for different questions. The right one is the “Variance EPS, which tells you the theoretical population size that yields the same noise in allele frequency change, right now. It basically measures how little a population is susceptible to drift.” The correct value of EPS, for different continental branches of humanity, has been hundreds of thousands to millions for a long time, well before agriculture was developed.
Graur is making a very basic mistake: he is perfectly happy making an argument to which there are many counterexamples. They don’t bother him. Even one counterexample should bother you. As I said, every domesticate is a counterexample to what he’s saying. We know of many genetically caused differences between human populations, such as height, skin color, disease resistance, ability to efficiently utilize certain foods, tolerate high altitude, low temperatures, high levels of arsenic, etc etc. By ‘we’ I mean anyone who can read. Some of those genetic differences are caused by new forms of one or a few genes ( like sickle-cell), others by small shifts in the frequency of many alleles that influence the trait.
Graur notes, incredulously, Augustine Kong’s work showing that the frequencies of variants influencing educational attainment are changing rapidly – in a way that decreases potential for educational attainment. We’re getting dumber. He asks how there could be so many university graduates in Iceland. But this kind of strong selection against potential for educational attainment – selection against intelligence – has been predicted for many years by people that understood that A. intelligence is highly heritable and B. On the whole smarter people, especially smarter women, have significantly fewer kids than average in contemporary society. This is almost certainly a new phenomenon, less than a hundred years old, just as mass higher education and artificial birth control are new things. Although Kong showed this directly, through genetic analysis, the idea is hardly new. R.A. Fisher talked about it, in The Genetical Theory of Natural Selection [published in 1930]. Of course Fisher understood polygenic selection pretty well too, having been the main developer of the theory of quantitative selection. Demographers in the 1950s talked about it. Cyril Kornbluth understood it ( ‘The Marching Morons’ ): but to be fair, Kornbluth was no ordinary man.
West Hunter 
Also: 74 genes evolving in concert is exactly what we’d expect if those genes all contribute to a trait that is under selection.
A person who understands quantitative genetics would know this. If not B, then not A.
WTF is he talking about??? There are a lot more than 74 alleles that are known to differ between continental populations! Nothing to “imagine” – it’s a sold fact.
Or, is he saying that the 74 changes must occur simultaneously (“in concert”)? If so, it’s beyond stupid. It’s exact same idiotic argument that ID proponents use to argue that basic evolutionary theory cannot be true. Flagellum must have come with all parts at once!
He isn’t denying that there can’t be allele differences among populations, he’s saying any differences are almost always due to drift and not selection.
Still stupid.
*frequency differences
re the discussion in the previous post: I haven’t gone through the math of this paper, but it says that the variance EPS also has a harmonic mean in it somewhere, at least under some simplifying conditions:
https://www.su.se/polopoly_fs/1.246487.1461150280!/menu/standard/file/Olssonetal_CharVarEffSize_TheorPopBiol_2013.pdf (page 94)
What’s the deal?
It’s also on the wiki page on Effective Population Size.
I responded in that last post. Basically, if you’re trying to find a single Variance EPS over a multiple generation period wherein the population was actually changing, then you’d use the harmonic mean. That would tell you how drifty the population was, on average over that time period. I.e. over the time period, the fluctuating population would have drifted as much as an ideal population fixed at the Variance EPS.
Now, if you’re interested in the strength of drift during the time period A -> B, anything prior to A is irrelevant (hence irrelevant to Variance EPS during that period). So any population history prior to the split of human continental groups is irrelevant to the strength of drift after the split. That’s not true for Inbreeding EPS; a tiny bottle neck earlier in our species history would still imply a big reduction in heterozygozity today. SInce Inbreeding EPS is calculated from heterozygosity (not drift), it remains very low today.
If A = B, then you get the result that the driftiness of the population right now depends only on its characteristics right now. Humanity has an enormous population with equal sex ratios and pretty low reproductive skew right now. So, it drifts very little; selection can easily dominate it. This has been true for millennia, no doubt.
Graur evidently thinks that since the Inbreeding EPS is like 10,000 today, it follows that drift is strong and you’d need strong selection to overcome it. That is just plain wrong.
Of course this is all academic. If it turned out that racial differences in IQ were due to drift, that wouldn’t make them any less real, would it?
You could easily have populations drift apart near the founding. But if one drifted in a direction that lowered fitness, natural selection tend to push it right back, especially as larger population size reduced the power of chance.
Thanks, I get it now!
I have little background in evolution besides reading ‘The Genetical Theory of Natural Selection’ [after reading your many recommendations for it] and even / I / knew that that tweet exchange was ridiculous.
Anyways, I read Reich’s book, and while it’s very good at catching you up to date with ancient genetics, he is, like you pointed out, supremely dismissive of phenotypic data of differences even pointing or being a clue to genetic differences.
I also wish he’d spent more time on predictions by evolutionary biologists or anthropologists or others like yourself that were either falsified or confirmed. He touches on that a bit when discussing what anthropologists used to think, but doesn’t spend too much time on it.
“On the whole smarter people, especially smarter women, have significantly fewer kids than average in contemporary society.”
If this continues, one wonders if it will lead to greater sexual dimorphism in human intelligence (leading more to the BOSSS (Boss – secretary) model of human mating, as coined by Heartiste).
That would take a long, long, long time.
That comment about selection against education was outrageous. I remember the first time I went through a CS chamber. When I took off my mask, I inhaled and my first thought was: wow, that didn’t do anything. Of course, that lasted for maybe half a second.
I bet he wouldn’t use this sort of logic when to argue we should go ahead and reintroduce lead housepaint.
Graur has a new post up where he claims that heritability is the correlation between phenotype and genotype (rather than the square thereof). Is he incapable of getting anything right?
The Soviet domesticated fox breeding experiment apparently achieved fully domesticated foxes in less than 20 years, starting with a breeding population (apparently) in the low hundreds or less.
Thanks. I heard of the 10,000 EPS for humans too and was wondering what to make of it.
Th decreasing intelligence trend was the subject of some of Murray’s work. Low IQ individuals are having more children than the high IQ assortive maters. Paul Gigot has referred to some of these effects as “The Roe Effect.”
“So, the chances that 74** loci will experience significant changes in allele frequencies simultaneously in each of the four populations is zero. With one locus, a change in human allele frequency may occur, albeit very rarely, as evidenced by the case of lactase persistence in North European and the case of the EDAR allele in East Asians. To imagine that 74 alleles change frequency in concert and that the 74 alleles have successfully battled genetic drift and recombination in merely 2,000 generations requires an extremely naïve and unsophisticated view of the evolutionary process.”
As soon as I read this, I knew Graur wasn’t ‘misinformed’, he’s just a particular breed of sociopath whose realised the formulae for everlasting fame and success in our meme-ified culture. His statement almost exactly fits some of those lewontinesque favorites that lefties parrot so confidently to this day: “sweeping statement x rules out any changes at all except the very rare case of Y that you have probably heard right wingers quote”
He’s constructed an argument deliberately to counteract the most common objection to the ‘humans haven’t changed at all’ argument, therefore placing himself at the head of the ‘most quotable smart lefty’ queue on this topic. He’s preening to be the next Lewontin.
The last bit is the real tell: “…requires an extremely naïve and unsophisticated view of the evolutionary process.” This line is designed to literally be catnip to the average mid-wit lefty desperate to feel smarter than these fact-quoting-racists.
I was spit-balling an article in the vein of Reich’s facesaving Overton overturn on genetic differences in race, only this time on “why we need to rethink multiculturalism”. When you get inside these people’s heads its surprising how easily the ‘prop-up in-group ego/malign out-group’ rhetoric flows.
I posted a comment on Graur’s post about Jensen, which cites Lewontin’s argument about seeds in different environments.
Nevan Sesardic has pointed out that Lewontin was attacking a straw man.* There are three possible arguments from within-group heritability (WGH) to between-group-heritability (BGH).
High WGH entails a non-zero BGH.
High WGH, by itself, inductively establishes a non-zero BGH.
High WGH, together with some collateral empirical information, inductively establishes a non-zero BGH.
Jensen’s position was (3), not (1) or (2).
Jensen: “I have explained in greater detail that heritability coefficients by themselves cannot answer the question of genetic differences between groups, but when used along with additional information concerning the amount of relevant environmental variations within groups and overlap between groups, can enter into the formulation of testable hypotheses that could reduce the heredity-environment uncertainty concerning group differences”
(1969) Reducing the Heredity-Environment Uncertainty’, in Environment, Heredity & Intelligence – Harvard Educational Review Series No.2) 220.
James Flynn, who proposes an environmental cause of group differences, also pointed out that Jensen never argued that heritability estimates alone could decide the issue.
“Far too many of Jensen’s critics have not taken up the challenge to refute him in any serious way, rather they have elected for various forms of escape, the most popular of which has been to seize on an argument put forward by the distinguished Harvard geneticist Richard C. Lewontin…
the real message of Lewontin’s example is that we can ignore high [heritability] only if there exists a highly specific and unusual set of circumstances. Therefore, it is absurd to say that high [heritability] estimates within black and white respectively are irrelevant. Their relevance consists precisely of this: they force us to look for a plausible candidate for the role of [X-factor)”
(1980) Flynn – Race, IQ and Jensen.
The late behavioural geneticist David C. Rowe did some work in the late 90’s looking for such environmental factors.
(2000) Sesardic – Philosophy of Science that Ignores Science: Race, IQ & Heritability http://www.ln.edu.hk/philoso/staff/sesardic/POS-2000.pdf
Waiving away the technical details of the many blundered thing which is Graur’s post, is he really saying that Reich is an idiot when it comes to genetics — that none of these basic points occurred to Reich, or the large number of people who reviewed Reich’s book? If what Graur claims is true, what on earth could Reich be going on about when he says we should be prepared to discover average genetic differences between races on social traits?
I suspect that Graur really believes in every one of those errors. But I can check further.
Not many of the people reviewing Reich’s book are competent to evaluate what he’s talking about. I can think of three. And most of those face space limitations.
Actually, I meant something different by “people who reviewed Reich’s book” (though I can see now how my phrase wasn’t clear), namely, those who reviewed his book before it was published. A good number of them are quite competent.
That is, read the book before it was published for errors.
Reich mentions a number.
What Graur claims is not true. There are interesting upcoming GWAS results.
Do smarter men father fewer children? If the other way around, mightn’t that offset the dysgenic trend among the smartest women at the top?
Smarter men father about the same, smart women less. The net trend is negative: that’s been known for maybe 60 years. Suspected for at least 90.
Do smarter men father fewer children? If the other way around, mightn’t that offset the dysgenic trend among the best educated women?
I recall Jayman on this, or maybe it was Audacious Epigone, but know nothing about the statistics myself. Can eugenisis at the bottom of society offset dysgenesis at the top?
I’m surprised this guy is a geneticist. Under extreme selection, it looks like the effective population size can be as low as 2 (or possibly even 1 nowadays, with cloning).
Helen Dean King, whose rats are ancestral to many of today’s lab rat lineages, was able to prove Darwin wrong 100 years ago: even most extreme inbreeding works, if one ramps up selection and progeny output.
https://books.google.com/books/about/Studies_on_Inbreeding.html?id=5WDXAAAAMAAJ&printsec=frontcover&source=kp_read_button
Side note: couldn’t that be the recipe for spell-checking genomes/breeding superhumans? Provided that the “experiment” goes on for a couple centuries? Of course, the (lineage of) experimenter(s) has to be rather ruthless.
Reblogged this on Quaerere Propter Vērum.
A new article says EDAR causes differences in survival among nursing infants:
https://www.pnas.org/content/115/19/E4426