« Hix from the Stix & Boyz from the Hood @ Yale | Gene Expression Front Page | Fat chicks »
January 24, 2003

"Racial Diversity"

Anthropologist Henry Harpending e-mails me the following:

Mitochondrial DNA paints an extreme picture of greater African diversity. It is just one locus and hence expected to be unreliable. Large collections of repeat polymorphisms show that there is a diversity cline away from Africa so that Japanese are about 15% less diverse than Africans, Amerindians maybe 25% less diverse. These are all probably neutral and hence passive indicators of population history.

Khoisan people are, according to their neutral genes, just generic Africans, but they sure don't look anything like [that] to their neighbors. They also have no 7R alleles at the dopamine D4 receptor locus while the frequency in their neighbors is 20%. There are a lot of indications in the genetics literature that a lot of selection is going on in our species, and I expect that the differences in appearance and in DRD4 and in other behaviorally significant loci are maintained by selection in the face of gene flow that has homogenized the neutral genome.

If something like this is going on then we have to rethink our ideas about race or whatever the euphemism of choice is. There may be more to it than Steve's notion [of] race as extended family: what we see with our eyes as races may reflect shared selective regimes, especially sexually selected regimes, rather than only shared common ancestry. That is why I think that the argument about who are the real Europeans is pointless until we know whether we are talking about passive
reflections of shared ancestry or some kind of shared regime of selection.

These considerations also make me less enthusiastic than [name deleted] about what large numbers of SNPs [single nucleotide polymorphism] are going to do for us. It may be that what we really want to know about is shared selective regimes while SNPs will just give us a passive picture of extent of shared ancestry. For example I don't expect that SNPs would help us very much to understand the differences between bulldogs and boxers.


Clarification Let me elaborate with one of Henry's examples of a divergence between the neutral loci that reflect ancestry and those that point to shared selection: the Basques. Henry points that that though the Basque share over 90% of their ancestry with their neighbors according to neutral markers (y chromosome & mtDNA), they are over 50% Rh-, far higher than the surrounding populations. How did this happen?

While neutral markers would tend to "blend," causing gentle clines (what you see in Cavalli-Sforza's colored-coded maps), selectively influenced loci, whether adaptive, sexual or otherwise, would have sharper gradients. So, each generation there might have been 1% intermarriage between non-Basques and Basques, but those with Rh incompatibilities suffered far higher miscarriage rates, because the Basque were mostly Rh- to begin with. The non-Basque who were Rh compatible were more likely to pass on their genes, which matched the Basque norm. On the other hand, they would also pass on their neutral markers, which would tend to shift the Basque haplotype frequencies more toward their neighbors. It seems likely that sexual selection also plays a role-for instance, skin color differences of great range between groups that are putatively close.

This is a complicated topic, and I'm sure I'm garbling Henry's presentation, but it could explain anomalies like the fact that white looking Brazilians have so many non-white neutral genes, there is sexual selection going on to preserve a certain phenotype. It also is an explanation for the fact that though the Khoisan look far different from their Bantu-speaking neighbors, their neutral genes indicate them to be close relatives-adaptive and sexual pressures differentiated the two populations despite convergence and blending of their neutral genes.

So I suppose one can ask, let us say you have individual A, B and C. A & B are cousins, but diverge sharply in a host of phenotypes, while A & C are close matches-perhaps close enough that many would think them siblings though they lack much shared ancestry (relatively speaking). An ancestral relationship would give a false impression of congruency between individuals A & B, while ignoring the similarities between A & C. The key is probably to simply not use any one standard universally, in populations effected by the same selective pressures ancestry is still important. But where a whole host of phenotypic traits tend to separate two groups with a recent ancestral history, we might be inclined to be more careful about using neutral markers to guide us.

Remember, if we are using “race” as a tool to aid in evaluating social policy or medical decisions, it makes little difference what method we use as long as we achieve some utility. Imagine as one last example, three populations, X, Y and Z. X & Y have recent shared ancestry, but Y & Z have adapted to similar local conditions. The pathogenic environments of Y & Z have pushed them toward similar phenotypes with selective pressures having resulted in the latter two groups being exclusively of one blood group. If the X population that is ancestrally close to Y still has a diverse mixture of blood groupings, we should ignore shared ancestry if looking to make life or death decisions on transfusions where we had blood shortages and could not test for immune matches-we might assume a priori that Y & Z would be matches, excluding those from group X.

Those who assert that "race" is absurd indicate that there is a "tall race" and a "short race," a "straight hair race" and a "curly hair race." Those of us who do not totally discard the concept rebut that race has utility, that phenotype tends to match genotype, and certain populations cluster together. I am not denying that. But human beings are not a unitary whole-we are as Dawkins et. al. would say a mob of myriad genes, our bodies are simply vehicles for their perpetuation. The mob dictates to us the facts on the ground, and often our own concepts, reflecting our emphasis on the individual as the atomic unit of organization and classification, do not map so well and encapsulate the nuances that nature has imparted to us. All we can do is muddle onward....

Posted by razib at 09:19 PM

I don't see how shared selective regimes in any way weakens Steve's notion of race as "extended family." Are porpoises closer to fish because shared selective regimes made them look like fish? Are Pandas not really raccoons because shared selective regimes with bear ancestors made them look like bears? Are bats really birds then?

"That is why I think that the argument about who are the real Europeans is pointless until we know whether we are talking about passive
reflections of shared ancestry or some kind of shared regime of selection." -- Henry Harpending

This statement is a complete non-sequitur. Steve's notion stands totally unscathed and intact.

Posted by: Unadorned at January 25, 2003 10:59 AM

henry's argument is a little hard to explain cog-but remember that the comparisons you make do not apply because the examples you give do not interbreed. i will elaborate as best i can later....

Posted by: razib at January 25, 2003 01:56 PM

I would not defend my "a racial group is a partially inbred extended family" definition not on the grounds that it explains all phenoytpic similarities in this world. Instead, it provides a highly useful organizing tool for some knowledge, but it's utility is precisely linked to its modesty of aims.

For example, Lithuanians, Sudanese Dinkas, and Canadian Blackfoot Indians all tend to be tall, but calling them "the tall race" is not terribly useful.

Similarly, black Africans and some Melanesians from the New Guinea area are very similar in looks in a variety of ways, but they are relatively unrelated genealogically. Apparently, natural and sexual selection pressures in these two tropical environments have been quite similar. Thus, it's less useful to call them a single race.

It would be good to have a term for peoples who share certain phenotypic traits yet doesn't imply close common ancestry. Any suggestions?

Another useful term would be for peoples who share a single phenotypic trait due to common ancestry, but in general are not closely related geneaologically overall. Consider the economically important mutation for lactose tolerance. We don't know whether it appeared only once or several times, but it's quite possible that it appeared only once and that both Lithuanians and Dinkas inherited it from the same single ancestor, even though, relatively speaking, they are not particularly closely related genealogically. For example, Lithuanians share more overall ancestry with certain lactose intolerant Mediterranean peoples, while Dinkas' family trees are more like their non-herding and lactose intolerant neighbors in a general sense. Still, they may well both have a single common ancestor with this wonderful mutation. I would suggest calling these kind of attenuated but real racial groupings a "trace."

Posted by: Steve Sailer at January 27, 2003 12:00 PM

It would be good to have a term for peoples who share certain phenotypic traits yet doesn't imply close common ancestry. Any suggestions?





(in honor of Mr. 'Race Without Color') Diamonds

...or, (oh god, I should stop)

heirsatz (at least I didn't say 'shamcestors')


Akin (hard "A". Its a double-meaning. The word usually means 'related', b/c etymologically the schwa-sounding "a" in 'akin' means "for of", but the prefix "a" more commonly means 'without', as in asexual, which is pretty much what someone would have to be to actually write this stuff down.)

Now does anyone have any good ideas?

Posted by: Jason Malloy at January 30, 2003 12:09 AM