Genetic orthodoxy?

Share on FacebookShare on Google+Email this to someoneTweet about this on Twitter

John Hawks, in a post on scientists who dispute the acceleration hypothesis (acceleration deniers?), makes reference to “the Stanford school of genetic orthodoxy”. So what is this?

Essentially, he’s referring to the current paradigm (I’m as much of a fan of hyperbole as anyone else, but paradigm is clearly the more appropriate word here) in the field of population genetics about the peopling of the world. The story goes like this: a small set of individuals from an ancestral population in Africa moved somewhere in the Middle East, and grew. Then from there, a small set of individuals moved nearby in each direction and settled. Ditto for those populations, and so on. These “serial bottlenecks” kept occurring until the entire world was populated, replacing the individuals that were there before them.

The observation that solidified this paradigm comes from this paper, which showed an impressive negative correlation between distance from East Africa and genetic diversity, consistent with each population containing a subset of the diversity of the populations it came from. Since then, that sort of approach has been used in a number of similar applications, including this nice one on the peopling of the Americas.

Further support for this paradigm comes from more recent work modeling human demography–it’s simply not true that this out-of-Africa hypothesis is enforced like an orthodoxy. See, for example this paper entitled “Statistical evaluation of alternative models of human evolution” (lest you think that alternative models of human evolution aren’t being evaluated), which concludes for a single origin of humans in Africa. This doesn’t test the “serial bottleneck” model, but does address the multiregional hypothesis, which I think is the major point for Hawks. Or consider a more recent paper, which attempts (with moderate success) to infer the colonization history of the world. The results favor out-of-Africa, as well as serial bottlenecks (though theses bottleneck, it must be noted, were essentially built into their model).

Now, new data may alter some of these models somewhat–David Reich and other claim here (in a News and Views article) that they see evidence for multiple waves of migration from Africa in PCA analysis, though it remains to be seen how those results hold up.

I’m not sure what Hawks thinks of these papers–for all I know, they’re making the multiregional hypothesis into a statistical straw man that is easily demolished, but the point remains that the consolidation of these observations into a paradigm is not entirely without reason. The statistical methods and genetic data are available to challenge it, and skeptics (I know many) are more than welcome to try their hand.

Labels:

29 Comments

  1. well, i’m not weighing in on either “side” here, but i would also add that my minimal knowledge of palaeonthropology using morphological cladistic analysis of fossils also paints the same picture. perhaps john can point out the problem with that, i don’t know, but that certainly changes the picture when you have two disciplines offering the same results.

  2. “does address the multiregional hypothesis, which I think is the major point for Hawks.” 
     
    You’re wrong. He disagrees with the models of past effective population size being used. For example, that Lohmueller paper posited a European bottleneck that lasted for 7703 generations: at 25 years a generation, that’s 192,000 years – a neat trick, considering that the expansion of modern humans out of Africa only happened about 50k years ago.  
     
    The idea that the parsimonious explanation of a mild excess of nonsynonymous SNPS in Europeans compared to Africans (55.4% versus 47%) is a bottleneck among people outside Africa that is four times longer than there _were_ any modern humans outside of Africa is questionable. 
    I think it is possible that humans were somehow less well-adapted to ecological situations that they had never faced in Africa: that would selected for more change than among people who stayed at home. If you were talking about any other species that had expanded out of Africa into temperate climates (worse than that, during the Ice Age) that’s what you would expect to see.  
     
    It’s called natural selection.

  3. He Helletal-Auton-Falush paper looks intrinsically interesting, except that it finds an unexpected relationship between the Orcadians (Scotland), the Han Chinese, and the Yakut (NE Siberian Turks). There’s also a second unexpected relationship between the Orcadians and the Sardinians. This looks like an artifact to me. 
     
    On the other hand, it seemingly reports the Hazara-East Asian relationship as unexpected, whereas this is a well-documented case of Mongol backmigration.  
     
    They deliberately avoided contamination by historical and contemporary knowledge, which is good I think, but I have trouble imagining an Orkneys — Yakutsk — Sardinia connection.

  4. No, “paradigm” clearly isn’t the right word. A paradigm is a philosophical/methodological framework within which theories are proposed and tested, governing what sorts of things we’re looking at, how we’re looking at them, and what sorts of questions we’re asking. What you’re describing is a specific *theory* (or family of closely related theories) about human population history that’s supposed to explain the patterns of variation we see — which Hawks contends has not actually been adequately tested against other, more plausible models. He has pretty good reasons for thinking that, which I’m pretty sure he’d be happy to share. 
     
    Merely saying “the results favor out-of-Africa” doesn’t make it so — they genuinely might, but it’s a meaningless assertion unless you explore the assumptions underlying it explicitly. If the point of this post was just to say “hey, there’s a reason why these smart people believe this” then fine, but I feel no wiser about what that reasoning actually is. It’s nice that a paper “concludes for a single origin of humans in Africa”, but shouldn’t we be more interested in premises and deductions than conclusions?

  5. I don’t think there is any argument with some form of the general Out-of-Africa model. What Hawks is addressing, I think, is the widespread fear-of-selection and consequent invocation of bottlenecks to explain everything. There is no support for these supposed bottlenecks outside the particular data sets that are being explained away. 
     
    This bottlenecks-everywhere mentality is part of a larger funny kind of  bias in human genetics. The paradigm recently in genomic surveys has been to do careful elegant analysis, beautiful computations on huge datasets, then wave the arms back and forth about bottlenecks with no thought of testing any hypotheses. All the attention is to the technology, none to the scientific issues supposedly being addressed.  
     
    Another wonderful example is the Olshen et al. paper discussed here a while ago. Beautiful data showing with absolutely no ambiguity that there never was any Ashkenazi bottleneck. But the authors paid no attention, prattled on about the Ashkenazi bottlenecks (3 of them!) anyway without even looking at their data. 
     
    We have had the same same tradition a long time with the mtDNA and Y chromosome tree-makers. Careful genetcs, careful computation, then stare at the output and make something up to explain it. No testing of prior hypotheses ever-the concept is foreign. 
     
    Henry

  6. For example, that Lohmueller paper posited a European bottleneck that lasted for 7703 generations: at 25 years a generation, that’s 192,000 years 
     
    i’m not making any argument about the lohmueller paper, in fact I didn’t read it all that closely. I know the authors think that positive selection is a small subset of their SNPs, and so shouldn’t alter their results, but I think you and Hawks are right in that it should be considered more.  
     
    in any case, that paper has nothing to do with the “Stanford orthodoxy”

  7. If the point of this post was just to say “hey, there’s a reason why these smart people believe this” then fine, but I feel no wiser about what that reasoning actually is 
     
    well to the first point, yes, that was essentially my point. read the excoffier paper–if there was something other than complete or near-complete replacement, you’d be able to see it in the data.

  8. re:bottlenecks. any model for recent human history that doesn’t include some form of strong genetic drift (via bottlenecks or small Ne or something) in non-African populations simply will not fit the data well. a very nice paper that serves as an example: 
     
    http://www.nature.com/ng/journal/v39/n10/abs/ng2116.html 
     
    Our analysis shows that East Asian and northern European ancestors shared the same population bottleneck expanding out of Africa but that both also experienced more recent genetic drift, which was greater in East Asians. 
    some models of demography are more reasonable than others, and the data should allow this to be tested.

  9. “re:bottlenecks. any model for recent human history that doesn’t include some form of strong genetic drift (via bottlenecks or small Ne or something) in non-African populations simply will not fit the data well. a very nice paper that serves as an example:” 
     
    Or genetic draft. 
     
    Henry

  10. I’m curious why the results of Mungo Man’s mtDNA analysis are so often left out of these discussions of the out-of-Africa model. Mungo Man is considered to have fully modern anatomy, and has been dated to around 40kya; yet his MtDNA derives from an even deeper clade than the deepest divisions between the Khoisan and the rest of the world– meaning Mungo was not descended from our mitochondrial “Eve”.  
     
    This seems to be in direct conflict with the premise that all beyond-Africa modern humans are a subset of African genetic diversity. While the mitochondrial Eve of modern people shouldn’t necessarily be the same mitochondrial Eve of anatomically modern people 40kya, under the current out-of-Africa model, one still shouldn’t expect to find evidence of the deepest clade division at the far end of our species’ advancing edge.  
     
    Most modern Australian Aborigines have the N mtDNA haplotype, with a smattering of M thrown in. Not only is Mungo Man not an N or M, he?s not even an L3. In fact, he?s not an L2, L1, or L0. He?s pre- ?L?. Yet he’s anatomically modern. 
     
    If one were going to imagine a piece of evidence that supports mixture between modern and pre-modern people, it would be hard to think of something better than this? the presence of an ancient, deep-clade mt haplotype in an anatomically modern skeleton at the extreme edge of our species range.

  11. I don’t like terms like “out-of-Africa”. Unless Mungo Man grew up in his mother’s pouch, or evolved from an orang utan, or bore a suspiciously elongated quartzite skull then his ancestry is ultimately African as well. 
     
    The question is when (not if) his matrilineal ancestors left Africa. Are there features in his DNA which correlate to bottlenecks in the development of other 40000-year-old control samples? If this hasn’t even been tested then Mr Mungo can tell us nothing pro or con about any form of out-of-Africa.

  12. There is lots of worry out there that the Mungo mtDNA sequence is an artifact. 
     
    Henry

  13. David, I also have a conflicted relationship with the term “out-of-Africa” for the same reasons you mention, but at this point, I don’t think the term is going anywhere. For my purposes, I use it only to describe the migration of anatomically modern Homo sapiens out of Africa. Homo erectus left previously from the same place, true, but the term, for some reason, is usually only applied to modern man. Or at least that’s my understanding of it. 
     
     
    You wrote: “Are there features in his DNA which correlate to bottlenecks in the development of other 40000-year-old control samples? If this hasn’t even been tested then Mr Mungo can tell us nothing pro or con about any form of out-of-Africa.” 
     
    I respectfully disagree. Let me tell you why. (These are my own thoughts.)  
     
    In the current model, if Mungo Man’s mt haplotype arrived in Australia via that first wave of modern man to cross the Red Sea, then that would mean his mt haplotype was present, at some frequency, in not only that first population to leave Africa, but also in the African population of which that population was a subset. This is why genetic diversity decreases with distance from Africa, and why geneticists agree Africa is the homeland of our species without having to once look at a bone. Each step along the path of a migration route produces new populations which are genetic subsets of the populations from which they derive.  
     
    So lets think about this. This would logically require that, in addition to being present in Africa, Mungo man’s halplotype would also have to be present at both the very root of modern Man’s out-of-Africa expansion AND, at the very least, every single step along the migration route to Australia. Even given a skipping-stone view of this migration, one is left to wonder why haplotypes M and N left genetic signals in populations along the way, but Mungo’s mt did not. Unless one is willing to accept that Mungo’s MtDNA was independently lost from every far flung population of modern man in the world, then the best explanation is regional admixture of non-moderns.

  14. Dave, here’s more of my reasoning.  
     
    I have no doubt that the proto-San population of 40kya had deeper clade diversity than in humans living today (with diversity loss occuring through chance over time within this population) but all evidence in living people today point to only a single clade of this original ancient diversity ever having left Africa– descendents of the L3 clade, M and N. 
     
    Even if the proto-San of 40kya had clade diversity going back to the first Homo, and even if much of that diversity has been lost between then and now, it wouldn’t change the fact that Mungo’s mtDNA must have been from a deeper clade than any carried across the Red Sea by the first moderns out of Africa. We know that because if it was a part of that population, then its abscence in living populations today would require it to have been independently lost from the proto-San, and from every other population refugia created en route between Africa and Ausralia. The math doesn’t work.  
     
    While mtDNA diversity is lost over time, the vast majority of that loss must necessarily occur in the original ancestral population for the simple reason that that is where the diversity is. Descendant populations aren’t that diverse to begin with, (they are, by definition, a sub-set) and by the time you get to the number of descendant populations that were produced on the migration route to Australia, it becomes mathematically untennable that each of those populations happened to lose, by chance, Mungo’s mt, while retaining, by chance, N and M. Mungo’s mt could not have been a part of that migration.  
     
    Henry, I beame less concerned about Mungo’s DNA being an artifact of contimination or error when I saw that it was compared to chimps, Neanderthals, as well as to modern humans, and it was placed on the Homo branch intermediate to Neanderthals and modern man. It could still be an error, of course, but a very lucky error, because Mungo’s exact position on that tree would– again — be just what you’d imagine mitochondrial evidence of regional pre-modern mixture to look like.

  15. Bottlenecks make selection more plausible. If newcomers were constantly showing up from environments with other selection pressures, then selection would be slower.  
     
    Say you are trying to create a new breed of dog in your backyard through selection for the traits you want. The fence around your yard acts as a bottleneck. If the fence is only three feet high, your selective breeding operation is going to go very slowly because stray dogs keep jumping into your backyard and impregnating your bitches. If your fence is ten feet high, you’ll get results faster. Your fence is your bottleneck and the bigger the better for selection to work.

  16. Part of the problem of this discussion is that people focus on the partial bottlenecks like, say, the Red Sea, but ignore the really big bottlenecks, like say the Atlantic Ocean between Africa and South America, which was a near absolute bottleneck until 1492 or the Pacific between South America and Australia. Back in 2005, I drew a map of the “Roads Not Taken:” 
     
    http://www.vdare.com/Sailer/050320_leroi.htm

  17. Also, another issue is the old junk gene vs. functional gene dichotomy. The Stanford School focuses on the most neutral genes it can find because they don’t want to know about things like convergent evolution. They are just interested in the genealogy of races.  
     
    From their point of view, the fact that sub-Saharan Africans and some Melanesians look (and perhaps act) fairly similar is no interest. The important thing is that their neutral genes show that they diverged tens of thousands of years ago. In contrast, the Utah School (to coin a term) is interested in the tropical selection pressures that either made them grow more alike or kept them alike over huge periods of time.

  18. steve, re: bottlenecks 
     
    1) if you are doing selective breeding then homogenizing the genetic background can make it easier for selection by freeing up more additive genetic variance. 
     
    2) but the extent to which “fence jumping” matters is subject to the parameter of migration between demes per generation. if the migration is like 1% per generation, and you’re selecting 20% of the population in a given generation through truncating above the 20th percentile on a quantitative trait then i don’t think a 1% dilution per generation would matter. 
     
    3) finally, bottlenecks tend to reduce variation. if #1 is not operative, then the power of selection is going to be weaker per generation all things controlled since you have less variance to operate over.

  19. Henry, I beame less concerned about Mungo’s DNA being an artifact of contimination or error when I saw that it was compared to chimps, Neanderthals, as well as to modern humans, and it was placed on the Homo branch intermediate to Neanderthals and modern man. It could still be an error, of course, but a very lucky error, because Mungo’s exact position on that tree would– again — be just what you’d imagine mitochondrial evidence of regional pre-modern mixture to look like. 
     
    It’s been suggested that postmortem damage at mutational hotspots may produce such weird results. 
     
    Apropos John Hawks, his latest post today discusses a fictional story in Nature with a “surprise ending” that may delight GNXP readers. But isn’t there a second surprise ending that Hawks has missed?

  20. Unlike Drosophila, human patterns of diversity (i.e pi) show very little dependence on the recombination rate (beyond that predicted from the divergence, i.e. the mutation rate). Thus large scale genetic draft is unlikely to be operating in humans. This is one of a number reasons why selection has been somewhat dismissed as the primary explanation of the large scale reduction of diversity outside Africa.

  21. G: 
     
    Very nice–a real hypothesis test. Can you give us some good sources? 
     
    Henry

  22. The lake Mungo sequence is a numt, or nuclear insertion of mitochondrial DNA. These are extremely common and can make obtaining genuine mtDNA sequences difficult with PCR-based methods. 
     
    The particular numt that the Mungo study sequenced is fixed in modern humans as far as anyone knows. The mitochondrial lineage that was the source of the insertion predates modern mtDNA diversity but postdates the chimp/human split. Thus, when considered as an mtDNA homologue it makes Mungo appear outside modern diversity. However, since the sequence is paralogous the comparison is not valid.

  23. I think the claim that the multiregional/OOA hypotheses are rarely tested is a bit disingenuous. Most or all of the genetic predictions of the OOA can be accommodated by multiregionalism, however the reverse is not true. I don’t think there is any data incompatible with at least the weak form of the OOA yet found. This, of course, means that there isn’t data incompatible with multiregionalism. Multiregionalism has not been falsified.  
     
    This doesn’t, however, imply that the two hypotheses are equally likely given what we know. 
     
    We have spent a lot of time looking for data that could contradict OOA and exclusively support multiregionalism. We just don’t seem to find any.

  24. G emailed this to me; we’ve been having issues with the commenting system. 
     
    A couple of refs are: 
    Hellmann et al 2005 
    hellmann et al 2003 
    http://www.ncbi.nlm.nih.gov/pubmed/16140990  
    http://www.ncbi.nlm.nih.gov/pubmed/12740762 
     
    a couple of recent papers have claimed to find more of a correlation beyond that seen in divergence: 
     
    Hellmann et al 07 
    http://www.ncbi.nlm.nih.gov/pubmed/18411405 
     
    at least some of this is explained by changing recombination rates and potentially biased gene conversion: 
     
    http://www.ncbi.nlm.nih.gov/pubmed/17044736 
    Spencer et al 
     
    Also the large scale genetic draft you are suggesting seems to run counter to Hawks et al, which suggested that there were few full sweeps and many partial sweeps (which do not reduce diversity dramatically).

  25. Jason, thanks for your insight into Mungo Man’s MtDNA. I’d read about the sequence in question being an insertion, but I’m unclear about your comment, “Thus, when considered as an mtDNA homologue it makes Mungo appear outside modern diversity. However, since the sequence is paralogous the comparison is not valid.” 
     
    I can’t see how Mungo’s mt sequence having been inserted in our nuclear DNA could render the comparison not valid. To me, Mungo’s mtDNA answers the question of where that particular stretch of nuclear DNA came from in our species, but it doesn’t provide a way to prove Mungo’s placement outside modern diversity could be in error. What am I missing? Are you suggesting that what the researchers thought was Mungo’s MtDNA wasn’t MtDNA at all, but instead a stretch of nuclear DNA that we all have? Because if not, it seems that if Mungo’s mtDNA IS that particular sequence, then there’s still no way around it being outside modern diversity.

  26. Kosmo, 
     
    Yes, what the researchers thought was Mungo’s mtDNA is actually nuclear DNA that we all have. 
     
    It is not Mungo’s mtDNA that was inserted into the nucleus, it was the Mungo’s pretty distant ancestor, an ancestor that we all share.

  27. ‘No, “paradigm” clearly isn’t the right word’. All right, paranickel.

  28. “Yes, what the researchers thought was Mungo’s mtDNA is actually nuclear DNA that we all have” 
     
    And with that little piece of information, the entire issue shifts beneath my feet. Regional mixture between moderns and premoderns now lacks its smoking gun. 
     
    Thanks again, Jason.

  29. “Yes, what the researchers thought was Mungo’s mtDNA is actually nuclear DNA that we all have” 
     
    Do you have a reference for that? The original authors of the Mungo Man DNA paper were aware of the insert and claim that their DNA is not an insert, but a mitochondrial variant related to it.

a