Drift, selection, and history

Now that we're in an age where natural selection on functional genes is recognized to have occurred in historic time, stories like last week's paper on the Manchu dynasty Y chromosome patriline have taken on an additional element of complexity.

That's because of the possibility that these newly common variants might carry some selective advantage.

The difference between selection and drift is a question of function: does any gene on the Y chromosome do anything that would have caused an allele to increase markedly in frequency during the past few hundred years?

With most genes, we wouldn't have to ask the question. First, it would be much less plausible for a biparentally inherited allele to increase so rapidly in frequency. With the Y chromosome, there is the chance that a father with a huge number of mates (and offspring) would transmit the opportunity for similar high reproduction to his many sons. Inheritance of cultural power is a cultural mechanism, not a genetic one. Hence, there is the chance of great concentration of reproduction into a single patriline, and the consequent increase in frequency of the patriline's haplotype. An autosomal gene would have only a fraction of the reproductive boost, because the father would have two copies, each would be transferred to only half the offspring, and each son (and daughter) would receive another copy from his (or her) mother. After a few generations, there is even a fair chance that neither of the founder's alleles would be left, replaced by the alleles coming in from the wives, son's wives, and so on. Inbreeding would prop up the frequency of the original founder's alleles within the dynastic lineage, but it would limit the spread of those alleles across the region. Hence, if we observed a huge recent increase in the frequency of an autosomal variant, the most likely hypothesis would be selection.

Second, autosomal genes give us the possibility to localize selection to a single locus. Recombination breaks up chromosomes every generation, so that after a relatively few generations, different parts of chromosomes have different fates. A functional gene under selection will carry a relatively small part of the chromosome with it, a small part that can be identified through linkage disequilibrium. Thus, functional genes under selection can be identified to a relatively small area, often to the gene itself. In contrast, the nonrecombining portion of the Y chromosome is completely linked, so that there is no chance to find out which portion of the Y might have been under recent selection. In principle, one could find whether the successful haplotype had any new functional variants, but in practice genetic studies do not assay the complete sequence but only a small number of markers. Without tying the reproductive excess of a haplotype to a particular functional variant, it is much harder to make the argument for selection.

Personally, I think that the extreme founder effect is the most reasonable alternative, at least for the time being. We know that some rulers in Asia had very high offspring number, and that many of their offspring were made rulers of regions or nations and had similar opportunities. Over a few generations, this extraordinary situation would greatly increase the frequency of the Y chromosome haplotypes carried by such patrilines. Even if geneticists have not yet documented the relationship between the haplotypes and the patrilines, history makes the argument very plausible.

But what's interesting to me is that the original work on these Y chromosome haplotypes does not draw this distinction between founder effect and selection in any clear way. For example:

The historically documented events accompanying the establishment of the Mongol empire would have contributed directly to the spread of this lineage by Genghis Khan and his relatives, but perhaps as important was the establishment of a long-lasting male dynasty. This scenario shows selection acting on a group of related men; group selection has been much discussed (Wilson and Sober 1994) and is distinguished by the property that the increased fitness of the group is not reducible to the increased fitness of the individuals. It is unclear whether this is the case here. Our findings nevertheless demonstrate a novel form of selection in human populations on the basis of social prestige (Zerjal et al. 2003:720).

This is a peculiar use of the word "selection". It is sort of analogous to "kin selection", which is not really natural selection at all, but instead is a structured behavioral pattern that advantages kin compared to nonkin. It's a confusing part of evolutionary biology that "kin selection" has the word "selection" in it, but it is the sort of thing you learn not to confuse. Likewise, it's not "group selection" unless there is some function on the Y chromosome that advantages the group (possibly at a cost to the individual).

Is this a "novel form of selection in human populations on the basis of social prestige"? I don't think so -- I think this is just normal cultural inheritance of power.

But consider what such a hypothesis would actually mean. It would have to imply that there actually has been selection on some Y chromosome variant because of a function leading to social prestige. This hypothesis would be along the lines of the idea that Genghis Khan's Y chromosome caused the Mongol conquests (and the subsequent assignment of strong reproductive advantages to their rulers). It is functionally similar to the idea that reproductive dominance and aggression may be linked to some Y chromosome gene (or genes).

We don't have any reason to think such things are true. The hypothesis would certainly imply a very different genetic history for the Y chromosome than usually assumed.

But here the hypothesis is, stuck in a paper on the "genetic legacy of the Mongols", without any exploration of its meaning. It's one of those things that a careful reader might take as a sign -- a sign that the genetic universe has some strange stuff waiting for us. It's a time to be precise about what we know and don't know.

I wonder how many of these patrilines the Genographic Project will turn up in areas with different histories?