Wednesday, December 07, 2005

Out-of-Africa again & again...again   posted by Razib @ 12/07/2005 08:47:00 PM

Carl Zimmer has a post put, Tree or a Trellis, which summarizes Alan Templeton's Out-of-Africa Again & Hypothesis. No biggie, Templeton's point is that a one locus genealogy does not necessarily represent the whole of a species' evolutionary population history. Allan Wilson's work in the 1980s with mitochondrial Eve was a case in conflation of gene history with population history when the data was translated into the public domain. We still haven't gotten past this, Seven Daughters of Eve, Journey of Man and The Real Eve are three recent books that use this genes-as-prehistory methodology. The History and Geography of Human Genes by L. Cavalli-Sforza was an explicit attempt to do this, and a forerunner, but at least he focused on more than one locus. Recent work has tended to skew toward mtDNA and non-recombinant Y (NRY) lineages, which do not exhibit the nasty tendency toward recombination characteristic of autosomal loci over long periods of time (nasty because recombination destroys easy inference via derived character states, which can be used to create phylogenetic treees). As I have noted before, our own personal genealogies are certainly reticulated loops of spaghetti all knotted up. The utilization of mtDNA and NRY offer us clean and tidy narratives which result in a genius of storytelling, and Wilson's work in the 1980s certainly (right I think) highlighted the important role that Africa has played in our species' demographic expansion over the past few hundred thousand years. But the facts on the ground are rather messier, and John Hawks' post on selection and mtDNA are required reading if you want to scan through the depths a bit. This older post, which was a cut & paste of an email from Henry Harpending, highlights the possible importance of selection across various loci working in ways we might not anticipate. The social effect of biases in the transmission of the uniparental loci (mtDNA ~ female and NRY ~ male) can not be discounted. Consider the pre-modern practice of the Namboothiri Brahmins of Kerala of only allowing the first born male to marry a Brahmin woman. All other sons entered into consortships with women of the Nair caste. This resulted in two tendencies, a) most Namboothiri women did not marry and remained cloistered virigins b) many Nairs had Namboothiri fathers. Over the generations the mtDNA lineages diagnostic for Namboothiris (imagine that they were migrants to the area) would be greatly outnumbered by those for Namboothiri NRY. Cross-cultural anthropology tells us that in societies with stratification the reproductive potential of an elite son is far greater than an elite daughter (both because of the biology of constraints upon the number of female gestations vs. male inseminations, and because of the frequency of polygyny vs. polyandry). You don't need to be a deep thinker to consider that assaying the NRY gene genealogy might give you a different impression about population history than mtDNA would.

Henry's paper Genomics refutes an exclusively African origin for modern humans is a good introduction to the nuances of this area if you can't wait for Templeton's paper. The reality is that we need to move beyond seeing genes as simply phylogenetic tokens, genes that do things (functional loci) are a dynamic currency which floats in the world markets of selection. MHC alleles are not shared by chimpanzees and humans because the two species have only recently moved past a speciation event, or because introgression via hybridization is common (or at least we don't think it is), rather, long term selection (either overdominance or frequency dependent) is preserving ancient variants on the order of tens of millions of years.1 More recently, and perhaps more significantly for our purposes (I believe that genetic loci like MHC where diversity is positively selected for are rare), the allele which confers lactose tolerance has experienced widespread selection across the Western half of Eurasia over the past 10,000 years. It is likely that the allele in question first began to increase in frequency in northern Europe, and was transmitted via deme-to-deme genetic exchanges subsequently. Yet, just because the gene arose in northern Europeans and now has attained a relatively high frequency through western Eurasia, it does not follow that a relatively high frequency of the genes of populations outside of northern Europe are northern European. Independent assortment and recombination break genetic associations, so that over hundreds of generations the allele that confers lactose tolerance has become decoupled from its predominant northern European genetic background.2 The point in challenging the Out-of-Africa only hypothesis is not to revive classical anangenetic Multiregionalism, that story is just as simplistic as Out-of-Africa only.

To be honest, I am skeptical that this sort of detail is every going to make it into the popular press books, and I believe that humans have an bias to conflate their intuitions related to recent ancestral genealogy with that of the evolutionary genetic scale.3 Nevertheless, it can't hurt to correct the misimpressions of those who don't understand, and one can always fight the good fight against the likes of Bryan Sykes, who stand to profit from confusing the public as to the relevance of one genetic locus.4 As I noted before, Richard Dawkins has promoted Out-of-African Again & Again in his book The Ancestor's Tale, so I am mildly hopeful that the tide might be turning. And remember, this isn't all theoretical, we already have candidates for loci which have crossed over between archaic H. sapiens populations...and more are in the pipeline!

Addendum: For the visually oriented, make sure to check out Carl's copy of Templeton's diagram which illustrates his model.

1 - Obviously neutral alleles go extinct due to random walk processes. But even alleles with fitness impacts can eventually go extinct if a new mutant with the same fitness begins to increase in frequency, as relative to each other the two variants are neutral. The preservation of a lot of diversity on the MHC loci is strongly suggestive of selection for the specific alleles.

2 - As someone pointed out to me, just because that the gene increased began to rise in frequency first among northern Europeans, it does not imply that it did not exist within the standing genetic variation of populations across the world. One could hypothesize that the spread of dairy-culture throughout the world resulted in situ selection upon the allele within the local population, but my understanding is that the character of genetic diversity does imply that non-European variants are derived. Additionally, a priori it seems more plausible that migration could introduce far more genetic variation over the small period of time that cultural transitions often occur over. Consider that the probability of fixation of a positively selected allele as 2s, wife-trading between tribes seems likely to introduce many copies of 2s (that is [number of women carrying the gene] X 2s), in comparison to the number copies of the gene in question within the population.

3 - Recombination is not important over 2-3 generations, ergo, it isn't something we encounter much in our daily experience.

4 - Spencer Wells work with the Genographic Project seems a bit more innocuous, instead of being out to make a big buck, he seems more interested in pushing forward his Y chromosomal research and collecting data by overselling its implications and relevance. That's science.