Gene Expression

Update: Update at the bottom….

In reference to the sequencing of the Neandertal genome, Kambiz at states:

I have one little gripe with the New York Times article. Wade quotes a geneticist, Dr. Bruce Lahn saying there is, “evidence from the human genome suggests some interbreeding with an archaic species.” Has he not read Paabo’s paper 2004 PLoS paper, “No Evidence of Neandertal mtDNA Contribution to Early Modern Humans?” There is significant mtDNA evidence to stongly conclude that humans and Neandertals did not interbread directly.

I left a comment on that website, but I will elaborate a bit here. In reference to Bruce Lahn, he almost certainly has read that paper, but,

1) mtDNA can take you only so far.

2) Lahn is implicitly pushing a new paradigm in how we view human origins, a “genocentric” perspective, to borrow a phrase.

To understand in detail about why mtDNA must be viewed with caution, read RPM’s take down of its use in molecular ecology, or John Hawks’ post which addresses the issue from a paleoanthropological angle. If you want it short and sweet: one locus isn’t going to tell you the whole story, and that is more and more true as you push back into time. Reading the mtDNA lineage is reading the mtDNA lineage, not the sum totality of our evolutionary genomic history.

But let’s move past that issue. I am convinced by other data from other loci (autosomal & Y surveys) that the preponderance of our ancestry as modern human beings derives from Africa within the last 200-50 K years. What Lahn was alluding too, and was not particularly well articulated in the article, is that phylogeny and phenotype do not always track. Even if the vast majority of alleles are identical by descent from African populations, that does not mean that some of our alleles do not derive from “archaic” populations long resident in other regions of the world. What Lahn is pointing to is the possible persistence of locally favorable genetically coded traits as one population replaces another via enough interbreeding for the favored alleles to “jump” between demes. In a word: introgression.

To comprehend this better a little baby-formalism is in order.

The probability of fixation, the likelihood that an allele will reach ~100% in the population, is approximately 1/(2N) in the case of a neutral (neither favored nor unfavored) mutant, and 2s for a positively selected mutant. The relationship between these two values allows one to intuit the relative power of selection vs. drift within populations. Let’s ignore drift, 1/(2N), and focus on selection, 2s. The variable s indicates the selection coefficient, roughly the proportional deviation above population mean fitness when an individual carries that allele against an average genetic background. In plain English, if you have allele x, you’ll have more kids by increment y. A selection coefficient of 0.02 indicates a 2% increase in ftness. If a mutation occurred de novo within a large population (remember, let us ignore drift), then its probability of fixation would be 2 times its selection coefficient, so in the case specified earlier, 4%, or a 1 out of 25 chance that the mutant will sweep through the population from one mutational event and become “wild type.”

But what does this have to do with Neandertals and interbreeding? Everything. The baby-formalism above isn’t tied to the functional-molecular genetic cases of de novo mutations within populations, but can be used to imagine circumstances where new alleles, operationally mutations, are introduced into the population from without. Concretely, a breeding event between indivduals from discrete populations where there is an exchange of disjoint alleles is equivalent to a mutational event. When the individuals go back to their populations of origin (imagine a woman returning impregnated from a sojourn), they will bring back the new genes. If they are favorable they may sweep through the population and fix, but this does not mean that the rest of the exogenous genome will be dragged along (though some might “hitchhike” and generate linkage disequilibrium, as you see in the HapMap), segregation and recombination will break apart associations so that the selectively favored locus will rise in frequency while other alien alleles will be effected by other neutral or selective dynamics independently. To extend the baby-formalism further, consider this case:

1) An allele that is present to fixation in Neandertals but not present in modern humans has a selection coefficient of 0.02 when introduced into the modern human genetic background

2) Neandertals and modern humans share a deme-deme border for tens of thousands of years

3) There were widely spaced intrebreeding events

The probability of fixation of the new allele when it jumps from Neandertal to modern human is 4%. In other words, there is a 96% chance of extinction. Since the rare breeding events are spaced apart in time let’s assume that for each event the extinction period is shorter than the next event. You can see where I’m going with this, the probability of a sequence of extinctions can be modeled as 0.96n, where n represents the number of introductions. At 18 introductions of the mutation the expectation is less than 50% that fixation will not have occurred at least once (and of course biologically once it is fixed a new introduction is a moot point, and the genetic background will have shifted so that the allele is no longer favored against it. Independence is a little lie I introduced just to make the baby-formalism more infantile). In other words, genomes are dynamic, and they can “pick up” all sorts of local goodies through just a few breeding events.

A clear and resounding objection to this is that humans and Neandertals were not interfertile. The reality is that the separation of the two populations was only 500,000 years BP, so it is almost certainly not true, tigers and lions are interfertile and they have been distinct for 2 million years. Another objection is that Neandertals looked too different. Again, the Watusi have had Twa (pygmy) queens, resulting in the matrimony of the tallest and shortest of human populations. Neandertals were distinctive looking, but I am skeptical that they were alien enough that one could never conceive of copulation with them. Additionally, we know humans have sex with animals all the time, or at least frequently enough that the dynamic I allude to above could result in allleles jumping populations.

Finally, what’s the payoff here, are there candidates for genes that could have jumped? Yes, there are, and one is mentioned in the article, at least implicitly, MC1R, which plays some role in melanin production. In Europeans this locus is extremely polymorphic, that is, it has a lot of diversity. This requires explaining, and some authors have posited negative frequency dependent selection, where low frequency traits hang around in the population because they are beneficial, but if they become too frequent they will be selected against. I won’t hit on this, but address the issue for why this is introduced in the first place: the coalescence, the common ancestors, for many of the MC1R alleles in Europe are very deep, and some have estimated that the level of extant
diversity would require 800,000 years to accumulate via neutral non-selective processes. What does this have to do with bumping Neandertal uglies? Neandertals separated from our own lineage 500,000 years ago, and emerged from archaic populations that have lived in Europe for as long as a million years. In other words, Neandertals had enough “time” to build up neutral alternative allelic variants which confered a selective advantange vis-a-vis new arrivals. What might have happened was simply that dark skinned Africans arrived on the scene, replaced the native Neandertals, but managed to pick up extremely beneficial locally adapted alleles for coloration from the predecessor population, who obviously had a long time to be optimized toward the local climatic regime.

But this isn’t what Lahn is pointing to in any case, the alleles are a lot more exciting than that. If you want to know, ask Evolgen, he’s keeping the secret too :)

Update: Just to be clear, my use of independent probabilities above was simply to suggest that the expectation of extinction decreases to triviality with only a few introductions of the allele.


  1. #1 Hamsterbaffle
    July 22, 2006

    Something I’ve always wondered, and that maybe someone here can address, is this:

    Wouldn’t any mtDNA evidence only indicate that human males did not produce fertile offspring with neandertal females? Since mitochondria are only passed through the mother’s line, this seems to follow. It wouldn’t rule out the possibility of neandertal males producing fertile offspring with human females, would it?

    Where am I wrong?

  2. #2 razib
    July 22, 2006

    It wouldn’t rule out the possibility of neandertal males producing fertile offspring with human females, would it?

    in which case you’d see the evidence in the Y. it does seem that Y & mtDNA tend to support out-of-africa as the preponderant source of our ancestry, but not necessarily the sum totality of our genetic heritage as most seem to assume.

  3. #3 Hamsterbaffle
    July 24, 2006

    Thanks for the response. So do we have evidence that Neanderthals did not contribute to (European or other) human genetic heritage through the Y, as we do for mtDNA?

    I remember hearing about the mtDNA evidence for the first time a few years ago, and every story I read about it basically stated that this proves humans and neanderthals didn’t interbreed. It was obvious to me that the data didn’t say that at all.

    So is there some finding I’ve missed which indicates no neanderthal lineage in the Y?

  4. #4 razib
    July 24, 2006

    . So do we have evidence that Neanderthals did not contribute to (European or other) human genetic heritage through the Y, as we do for mtDNA?

    some prelim. analysis of the Y has occurred to my knowledge, though not as much as the mtDNA (which is rather copious in quantity so extraction is easier). same general result, they are different. but the same arguments apply, if you push far back in time, by their nature neutral uniparental lineages will probably not leave descendents.

  5. #5 Luis
    July 26, 2006


    I can’t make sense of some of your reasonings. First of all the separation between the lineages leading to H. neanderthalensis and to H. sapiens did not diverge just 500,000 years ago but are rather separated by at least 900,000 years (and probably even more). 900,000 years BP is the approximate date of humans (I guess the generic term H. erectus applies still then for all) arriving to Europe, where (in adition to West Asia) H. neanderthalensis evolved (via H. heidelbergensis and H. antecessor). The lineage leading to H. sapiens instead remained in Africa (most likely East Africa) and the two branches therefore were separeted for about one million years or maybe even more (hard to say without full genetic data from the Neanderthal branch).

    That makes sapiens and neanderthals at least half as distant as tigers and lions. And while tigers and lions can be interbred in captivity, such interbreeding doesn’t seem to happen in the wilderness. I guess that you still have a point in hybrids being potentially viable (maybe) but that doesn’t necessarily mean that interbreeding actually happened or that hybrids were actually succesful and not stoned to death or drowned after birth out of racist (“specist”?) prejudice (just a guess). Hybrids of tigers and lions are possible but have some problems, following with your example. They are sometimes fertile but their offspring are very fragile and would surely die in wild conditions without leaving any genetic legacy.

    Anyhow, the main barrier to hybridation between lions and tigers is not just genetic but specially behavioural: male lions, for instance, seem to find the mating play of female tigers unappealing or confusing. They are very different beyond their gentics: they are also different in their behaviour and that’s most important, specially regarding such a cultural animal as humans.

    Also, why would sapiens interbreed with neanders only in Europe in the Upper Paleolithic and not in SW Asia in the Middle Paleolithic, just after our ancestors left Africa? Neanderthals were in West and Central Asia also then and our ancestors surely met (or maybe avoided) them in their early travels.

    But the main objection I have to the surprising insistance of some on a suppossed hybridation is that modern West Eurasians don’t look at all like Neanderthals. If hybridation happened and specially if it was driven by need of adaptation we should see Neanderthal traits as prominent occipitals, lower foreheads or extremely strong stocky (but short) bodies in their modern descendants, at least to a point. None of those traits are present among Western Eurasians.

    And then you have the MtDNA…

    All evidence points in the direction of lack of succesful hybridation (even if there could be some isolated cases without legacy). What do you have in favor of succesful hybridation? Nothing but a “maybe”…

    Maybe Nessie exists too… but where’s the hardcore evidence? Nowhere.

  6. #6 razib
    July 26, 2006

    What do you have in favor of succesful hybridation? Nothing but a “maybe”…

    well…you don’t understand genetics very well. most of what you are saying is irrelevant.

    1) the 500 K mark is derived from genetic data. mtDNA data, which you use to support your later arguments.

    2) just because there are morphological gradations across the range of homo does not necessitate saying that group A and group B entered into cladogenesis 900 K BP.

    3) evidence like this points strongly to archaic introgression (e.g., neandertals).

    4) if you don’t understand the link at #3 (which i suspect you don’t), then you don’t understand what i’m trying to say. the point with MC1R explicitly addresses your “nessie” crack.

    5) i’m not writing a textbook here, if you don’t want to do the leg work and understand the genetics here, don’t take on a patronizing tone, it is fucking bullshit, and impolite to behave in such a fashion in my house.

  7. #7 Steve C
    July 26, 2006

    Interesting point about a specific adaptive genes being ‘plucked’ from one population into another without many other genes tagging along.

    Could it have worked the other way around. Could Neanderthals have plucked certain traits from modern humans arriving in Europe – such as a fuller capacity for language, abstract thought and cultural development – and under strong selective pressure rather rapidly begin to look and act like modern humans?

    I raise this because of Greg Cochoran’s contention back on April Fools Day that there was still mostly Neanderthal populations around. Though I’m very skeptical, if this is the case, then this is the sort of process that would make a mostly Neanderthal population look indistinguishable from modern populations.

    In fact, one could imagine a form of convergent evolution where strong selective pressures push a mostly Neanderthal population and a mostly modern population in the same direction. It’s perfectly possible that a mostly Neanderthal population could still have modern human mtDNA, so would not be detected in that way.

  8. #8 razib
    July 26, 2006

    It’s perfectly possible that a mostly Neanderthal population could still have modern human mtDNA, so would not be detected in that way.
    :) well, i don’t know, greg hasn’t told me, though europeans also have recently coalesced Y lineages. but your idea is intriguing….

  9. #9 Luis
    July 26, 2006

    LOL! I never meant to be impolite. I just meant to question your reasoning (and that of all the people still oddly enthusiastic about multiregionalism and/or hybridation hypothesis against all evidence).

    My apologies if I have offended you. It wasn’t my intention. I was just trying to make my point clear and truly I don’t see what is in my post to be offended about.

    Anyhow, let’s go point by point:

    1/You probably know that chronologies derived from genetic data have huge margins of error. You find much more recent chronologies of that sort (with much better samples) that yield dates ranging from 11,000 BCE to 400 CE (this example is taken from a study in South Asian R1a – if I don’t recall badly – and obviously leaves a huge margin for speculation). If the Neanderthal MtDNA chronolgy has proportional error margins, well… Genetics is still in its infancy.

    2/ I still expect that date to be the minimal for the gentic divergence between the two species/subspecies.

    3/I’m not sure if I understand well enough that but the introgression can also be digging in our own (Sapiens) archaisms, can’t it?

    4/ Sadly since I had a computer brakdown a few days ago, I can’t open PDFs online. I know I must fixt it fast but sadly can’t do it right now. Skipped

    5/ Ahem… I will skip that part too if you don’t mind.

  10. #10 razib
    July 26, 2006

    3/I’m not sure if I understand well enough that but the introgression can also be digging in our own (Sapiens) archaisms, can’t it?

    yes. there is evidence for introgression in east asians too (from their own local archaics).

  11. #11 Rory
    July 28, 2006

    This business of MC1R interests me, so I’m glad you brought it up. There was some press a few years ago about red hair in Europe possibly coming from neanderthals, which came out of a paper by J. Rees on the age of this mutation in MC1R being around ~25,000bp. This age was later disputed using a much older date for the mutation. For the European red hair-associated Arg151cys and Arg160Trp variants, Harding et al. (2000) estimated an age of ~80,000 years. For Asp294His, and Ser316Ser, they estimated an age of <= 30,000 years. Are these some of the alleles that you’re referring to? Is it time to bring back the red hair argument? Curious as to your thoughts.
    There was also a paper by Johnson and Coffin in 1999 about the thousands of nonfunctional human endogenous retrovirus’s we have coursing through our blood, which would seem relevant to your argument here. And Harding et al. did a study of allelic ß-globin sequences that would also seem to agree.

  12. #12 razib
    July 28, 2006

    And Harding et al. did a study of allelic ß-globin sequences that would also seem to agree.

    see rogers & wooding 2004 for a differing perspective.

  13. #13 NuSapiens
    August 9, 2006

    Razib, Gene Thugs Represent!

    It’s been shown that some mtDNA lines are associated with metabolic adaptations related to climate. This is a serious violation of the neutrality requirement for tracking things, especially far back in time. Adam and Eve are cute stories, but I wonder whether they will be around after another generation.

    Neanderthal disappeared during a climate shift in Europe. If Neanderthal mtDNA was adapted to Euro-tundra, I wouldn’t expect that to stick around very long during a thaw period. If at all, it might be in the far north.