One point that Charles Murray makes effectively in Human Diversity is that human migration necessarily creates different genetic patterns.
If we start with a tribe of 200 people, and 100 of them break off and move to a new location, those 100 cannot possibly take with them a representative sample of the gene pool of the whole tribe. There are many possible genetic combinations, but by arithmetic the new tribe can only take 100 combinations with them.
Here are some thoughts I have about that;
1. An assumption that Murray makes, which I believe is accurate, is that there is not much re-mixing. The old tribe and the new tribe cross-breed very little, if at all. In prehistoric days, the physical separation made cross-breeding unlikely. Also mutual tribal suspicion.
2. My guess is that some human colonies survived, and some didn’t. The ones that survived preserved their genetic tendencies and cultural traits.
3. Surviving colonies tend to stay put. If the colony really thrives and its population increases, then it will send out more migrant colonies.
4. My guess is that when long-distance communication and transportation was primitive, failing colonies tended to just fail. As technology improved, failing colonies would be more inclined to migrate, because they have better knowledge of where life might be better.
5. Of all of the colonies that humans ever created, only a relatively few were successful. When your genome is used to speculate about your ancestry, it is linking you back to one or more of the successful colonies.
6. If only a small proportion of colonies were successful, then of all of the viable combinations of genes, only a few will be present. Evolution will not have selected with extreme rigor. Yes, some of the colonies failed because of weak genes. But others failed because of bad luck or bad culture. And not very many combinations of genes were tried.
7. I think that this picture reinforces my skepticism about polygenic scores ever being able to explain much of the observed variation in heritable traits. We will observe some combinations of genes with great frequency, making additional sampling from those populations redundant from a statistical perspective. My intuition continues to be that we are now or soon will be at the point of greatly diminishing returns to increased sample size.
8. It is not just prehistoric migration that follows the colony model. Consider David Hackett-Fisher’s Albion’s Seed. Consider the Bosnian community in St. Louis, the Hmong community in Minneapolis-St. Paul, etc.
9. The more that a migrating colony marries endogamously and brings strong cultural beliefs when it migrates to a larger society, the longer it can persist without without dissolving into that society. Consider Orthodox Jews.
10. What will emerge from the migration process is populations with differences in both genetic makeup and cultural practices. Most of these differences are random, as opposed to selective. This will make it difficult to pin down the extent to which differences in outcomes across populations have genetic causes.
I paused my reading of “Human Diversity” after part 1 so I may be talking out of certain anterior aspects of my anatomy here. I think it’s important to think beyond the primate troop level when thinking about the evolution of human civilization; the big man and chiefdom societies described in the first half of Robert Wright’s “Nonzero”.
It might be useful to think about the nature (e.g. founder effects) vs environment (leaf diet) of bonobos vs common chimps. If the Westermarck Effect is real in humans, the assumption about limited inter-tribe marriage is false, though it might also explain the 5+ year difference in male/female marriage age.
It might be useful to trace back Kling’s St. Louis Bosnians as the southern slavs have a well documented recent history with several massive migration/conquest/integration turning points independent of changes in the gene pool. We are all familiar with Germanic barbarian tribes flowing South after the collapse of the Roman Empire but it is not widely known that Slavic barbarians migrated out of an area in the Ukraine around the same time. The Italian maritime city-state of Ragusa slowly transformed into the Slavic city-state we now know as Dubrovnik. Roman script and Catholicism vs Cyrillic script and Orthodox Christianity vs Arabic script and Islam all after the cultural influence of neighboring Greece had waned. A fantastic natural experiment in the evolution of civilization and recent enough to be well documented by outsiders.
It might be useful thinking about reef fish when thinking about human traits and migration. I had an AHa moment many years ago while reading Paul Humann’s “Reef Fish Behavior”. Fresh water fish like pike and muskey follow the terrestrial animal pattern of non-overlapping ecosystems while you get this weird phenomena in reef fish of multiple near identical species (e.g. French Angelfish and Grey Angelfish) and species competing in the same niche (e.g. various parrotfish species) living side-by-side.
The reason is that reef fish are dispersed randomly after being hatched. The adults live on a different reef than their parents and the juvenile phase is purely pelagic, independent of reeds altogether.
Shouldn’t you consider the importance of migrants out of highly endogamous groups with strong traits as a way of testing for genes importance? If the original group stays relatively isolated while the migrants cross breed or assimilate into a local culture elsewhere, the persistence of inherited traits correlated to the outgroups performance in the new area that move in parallel with the originals would seem to be a good way of screening for relative importance. The same is true for example when considering children adopted by a different race. In the US, the performance of non-white kids adopted by white parents on SAT and similar tests tend to mirror the pattern of the overall kids in the country rank-ordered by ethnic group. Averages of mixed race children tend to be similar to the averages of their parent ethnicities. Etc. There are surely ways of teasing out info, with appropriate caveats of course. Much better than just assuming — as is now common — that many traits are “socially constructed.”
On 6, it’s not clear to me what is meant by “selected with extreme rigor”. Works like Cochran and Harpending’s, The 10,000 Year Explosion: How Civilization Accelerated Human Evolution show that there has been a lot of relatively rapid and significant (>1SD shift in group mean) selection causing major genetic shifts underlying expression of traits which were more adaptive under different local circumstances and pressures. It’s only recently that even a good portion of humanity escaped from Malthusian conditions, and Malthusian conditions impose lots of selective pressure.
Ok, I suppose there is the theoretically most high-altitude-conditions-tolerant human possible, and I doubt that Tibetans (once part of the greater mainland Chinese population), native Andeans, or Ethiopians are close to that ideal, nevertheless, their groups diverged genetically in significant ways from other humans to adapt to their unique circumstances and challenges.
From Wikipedia, “The adaptation account of the Tibetans has become the fastest case of human evolution in the scientific record, as it is estimate to have occurred in less than 3,000 years … over 98% of humans from other parts of the world normally suffer symptoms of altitude sickness in these regions, often resulting in life-threatening trauma and death.” That’s pressure (heh, or lack thereof I guess.)
Most of the adaptation is thought to be from variants in just three genes, one which may have been, ah, “reproductively appropriated” from the Archaic Human population of Denisovans which had lived in the area for much longer periods. As soon as the variant became available, in short order, evolution created a 78% frequency difference between Tibetans and mainland Chinese.
If one was doing the ordinary work of doing the special linear algebra statistical analysis of a large set of vectors of (DNA, High Altitude Adaptation), these genes pop out like a sore thumb, and would be very accurately predictive of the individual having the trait.
I think similar things could be said for lactose tolerance in the last 10K years (~100% in Irish and Danes, ~0% in Thailand and Vietnam) and cognitive ability in certain groups (Cochran says +1SD for Ashkenazi Jews in the 800 years from 800-1600 AD.)
If rigorous means “perfection”, then sure. But even small groups have enough variation that new pressures can drive major changes in just dozens of generations. It’s when those pressures are relaxed that the rigor of selection falls.
On 7. In a comment on the previous post on the subject, I’ve expressed the reason for my optimism for increased n to yield substantially better predictive models, even from the point of view of decreasing errors in models arrived at by sophisticated statistical analysis.
But fortunately we have more than statistics. We understand the basic principles of biochemistry involved and we are learning about all the variants in proteins and their shapes, roles, and interactions. “This gene regulates sensitivity of cartilage producing cells and growth plate closure to sex hormones, and variant A is more sensitive, causes earlier fusion, and shorter stature on average.”
My belief is that there will be a positive feedback between GWAS-derived insights and “Physical Biochemistry” insights. GWAS gives us a hint to look at a particular protein and gene. We start examining the variants of that gene, and the differences in chemistry of its interactions. We discover important differences in interactions between one variant and a variant of a different protein from a different gene. We “seed” the next analysis of the GWAS samples with this insight, and we infer a better, more predictive model, which in turn points us in the direction of other candidates and functional mechanisms.
This feedback, for example, would be quick helpful in identifying and correcting for the kind of “collateral confounder” you proposed earlier, though I’m not sure how important that theoretical possibility is in practice due to genetic recombination and chromosome crossover.
At a more fundamental level, I don’t think that the constraint of sets of available codes is really such a limitation, as we aren’t seeking to explain the maximum potential variance that could emerge from all possible viable codes, but only the (still significant) variance we observe within those constrained sets.
If we want to know why lactose tolerance or altitude tolerance varies in a particular population group – say one on the margin of areas where other populations went to fixation, and maybe with 50% of people having the trait – we don’t need to observe every possible level of trait for every possible human to figure out the genes likely responsible.
Indeed, on the contrary, it seems to me that the fact that the population group is all closely related with lots of similarity of variants for most genes means that it would be easier, not harder, to pull out the gene differences that are responsible for traits that still have high levels of variance despite that close-relatedness and redundancy in sampling.
That is a good example; I hope your insight proves true. I think the other key advantage is being able to use GWAS as an alternative to large scale twin and adoption studies for questions/tests that were not previously performed.
MIT has an online course in computational evolutionary biology that might help you understand these issues more clearly. I’m pretty sure you aren’t afraid of mathematical models, and it might give you a bridge to the extensive literature that addresses many of your questions in excruciating detail.
Evolution does not select. If you survive and thrive, you leave many copies of your genes. If you survive and do OK, you leave fewer copies. If you die, you don’t leave any. Differential reproductive success rates is the “selection” in natural selection – not the selection of fit individuals. That is, Selection is not on the individual, it is on the population.
Suppose Alan dies without offspring. Suppose Bill has a dozen kids and over a hundred grandkids. It’s fair to say that evolution has favored Bill over Adam; Bill’s genes are greatly increased in the next generations and Adam’s have decreased. That’s the micro view of evolution.
If you take a macro view, individuals can be regarded as mere temporary containers holding a random selection of a population’s genes. In this view, the fitness of each gene is its average effect on the reproductive efficiency of the containers that hold it (which average is strongly influenced by the other genes in the containers). Evolution is the process of some genes becoming more common in the population and others less common over many generations.
But remember that the micro view and the macro view are different ways to look at the same process. A “God’s eye” view would combine the scope of the macro view with the detail of the micro view. Unfortunately it would also be too complex to be useful by humans.
(And yes, I’ve ignored a bunch of stuff about inclusive fitness and invasive migration which complicates, but doesn’t change, my point).
As a general rule, culture mutates greatly several times in a millennium (aside from the bits that are necessary to sustain life) while a millennium of genetic evolution (for humans) has very modest results. If you’re thinking on evolutionary timescales, you can often ignore cultural effects in much the same way that you can ignore today’s weather when contemplating the shape of a hill that was formed by countless years of wind and rain.