Tuesday, September 10, 2013

More Analysis of Neanderthal Introgression

"The timing and history of Neandertal gene flow into modern humans." S. Sankararaman et al.
   Previous analyses of modern human variation in conjunction with the Neandertal genome have revealed that Neandertals contributed 1-4% of the genes of non-Africans with the time of last gene flow dated to 37,000-86,000 years before present. Nevertheless, many aspects of the joint demographic history of modern humans and Neandertals are unclear. We present multiple analyses that reveal details of the early history of modern humans since their dispersal out of Africa. 
   1.We analyze the difference between two allele frequency spectra in non-Africans: the spectrum conditioned on Neandertals carrying a derived allele while Denisovans carry the ancestral allele and the spectrum conditioned on Denisovans carrying a derived allele while Neandertals carry the ancestral allele. This difference spectrum allows us to study the drift since Neandertal gene flow under a simple model of neutral evolution in a panmictic population even when other details of the history before gene flow are unknown. Applying this procedure to the genotypes called in the 1000 Genomes Project data, we estimate the drift since admixture in Europeans of about 0.065 and about 0.105 in East Asians. These estimates are quite close to those in the European and East Asian populations since they diverged, implying that the Neandertal gene flow occurred close to the time of split of the ancestral populations.  
[Ed. This is probably in the time frame of ca. 50,000 to 86,000 years ago within the range of estimated admixture times.]
   2.Assuming only one Neandertal gene flow event in the common ancestry of Europeans and East Asians, we estimate the drift since gene flow in the common ancestral population. We show that an upper bound on this shared drift is 0.018. Because this is far less than the drift associated with the out-of-Africa bottleneck of all non-African populations, this shows that the Neandertal gene flow occurred after the out-of-Africa bottleneck.  
[Ed. Note that if the effective population size of Out of Africa modern humans fell before it recovered, the bottleneck could have happened well after the initial Out of Africa migration.]
   3.We use the genetic drift shared between Europeans and East Asians, in conjunction with the observation of large regions deficient in Neandertal ancestry obtained from a map of Neandertal ancestry in Eurasians, to estimate the number of generations and effective population size in the period immediately after gene flow. These analyses suggest that only a few dozen Neandertals may have contributed to the majority of Neandertal ancestry in non-Africans today.
[Ed. I'd love to get a look at this data.]

Via a Dienekes Anthropology Blog post on the 2013 ASHG conference abstracts (emphasis his; bracketed comments mine).

* Notably, this study does not discriminate between a single admixture event in a single genetic population followed promptly by a population split into Western and Eastern components, and two similar parallel admixture events, one with proto-West Eurasians and one with proto-East Eurasians, that takes place after the populations split.

Point one does, however, confirm inferences I have made previously from the small amount of overlap found in the particular Neanderthal genes found in West Eurasians and East Eurasians respectively, which you would not expect if Neanderthal admixed genes had reached fixation in a single unstructured population very long before the West Eurasian-East Eurasian split of "Out of Africa" modern humans into two separate populations with very little gene exchange.

* Point two is expected from the absence of Neanderthal genes in Africans except to the extent of gene flow from back migration from the Out of Africa population, and from the presence of Neanderthal genes in all non-Africans.

While understated in the abstract, the bigger revelation of point two is that Neanderthal gene flow probably took place quite a while after the Out of Africa event, rather than at the outset when one might naively have expected the Out of Africa population to be at its smallest prior to its expansion into "virgin territory."  Other studies have suggested based on statistical analysis of modern population genetic data that the Out of Africa population contracted before it expanded.

Neanderthal admixture may have happened not long after the effective Out of Africa population size hit bottom, or perhaps more accurately, Neanderthal admixture prior to then while the effective population size was falling was likely to be lost to the gene pool through drift, while Neanderthal admixture during the immediately following population expansion was likely to be preserved in the expanding population.

* Point three, suggesting that there were only a few dozen instances of Neanderthal admixture that account for almost all Neanderthal genes in modern humans is the most fascinating when it comes to building a narrative and understanding how this happened.

Assuming an effective population size around that time for Out of Africa modern humans of 3,000 to 20,000 and a time span of admixture that may have been somewhere in the range of 900 to 24,000 years (about 30 to 800 generations), we are talking about an entire modern human Out of Africa population the size of a small city in which there were one or two half-Neanderthal children born each generation or so (of course, more concentrated and more sparse scenarios are possible to some extent).

Also, based on reasoning from the absence of Neanderthal Y-DNA and mtDNA, I have previously concluded that Neanderthal hybrids who were born into modern human communities probably almost always had modern human mothers and Neanderthal fathers, and that live births were overwhelmingly of half-Neanderthal girls rather than boys due to issues of hybrid compatibility often expressed as Haldane's law.  An abstract of a recent paper by Reich in the same post corroborates the notion that hybrid incompatibility was an issue from direct genetic evidence:
We built a map of Neandertal ancestry in modern humans, using data from all non-Africans in the 1000 Genomes Project. We show that the average Neandertal ancestry on chromosome X of present-day non-Africans is about a fifth of the genome average. It is known that hybrid incompatibility loci concentrate on chromosome X. Thus, this observation is consistent with a model of hybrid incompatibility in which Neandertal variants that introgressed into modern humans were rapidly selected away due to epistatic interactions with the modern human genetic background.
Source Paper: "Insights into population history from a high coverage Neandertal genome. 
D. Reich for the Neandertal Genome Consortium.

In my view, this evidence, taken as a whole, supports a scenario in which modern human-Neanderthal couplings were episodic events, perhaps one night stands, perhaps seasons affairs, perhaps rapes, rather than sustained, marriage-like relationship in which a Neanderthal individual was integrated permanently into a modern human community, or visa versa.  There may have been an undetectable small number of exceptions that proved the rule, of course.

* I also continue to think that the Neanderthal introgression legacy in modern humans probably shows only half the picture.  Hybrid offspring of modern human men and Neanderthal women were probably also born, but if hybrid children were matrilocal, they would have disappeared when their Neanderthal tribes, in general, went extinct.

Are Archaic Ancestry Percentages A Function Of Phenotypic Invisibility?

One point it would be interesting to work out would be how long it would take before Neanderthal introgression reached fixation in the community and hybrid Neanderthals ceased to be a distinct and recognizable sub-community within Out of Africa modern humans.  In similar models that I have run (aims at understanding the future of race relations), it takes surprisingly few generations (five to ten or so) for a population to become almost completely admixed in the absence of endogamy norms, but in a community where some people are strongly endogamous and a minority are not, unadmixed people are very swiftly comprised only of people who have a strong endogamy norm.

Ultimately, no unadmixed populations persisted in either West Eurasia or East Eurasia, presumably because after a few generations of admixture people with a small percentage of Neanderthal admixture became phenotypically indistinguishable from people with none.  But, at first, it is hard to imagine that discernibly hybrid Neanderthal girls would have been on a completely level playing field in finding mates as girls without any Neanderthal admixture.  

One would expect that a Neanderthal ancestry proportion that was phenotypically indistinguishable from a non-admixed individual would have a proportion of Neanderthal admixture similar to the amount which is indistinguishable for instance, in a black-white mixed race individual, since Neanderthals and modern humans would have been more phenotypically distinct from each other than any two modern humans.  

Experience tends to show that a black-white mixed race individual can "pass for white" in the vicinity of 1/8th to 1/16th African ancestry, about 6.25%-12.5%.  The current Neanderthal admixture level in modern humans is just a little bit below that threshold (which would have been a bit lower for Neanderthal-modern human mixed species individuals due to the greater differences between the two to start with).  Indeed, one might wonder if this "pass for modern human" threshold played an important part in determining the ultimate level of admixture that took place on the theory that only people who could pass for modern human would have been fully integrated into the modern human community and contribute genetically to the population in the long run.  

Otzi the Iceman and the estimated peak level of admixture in Denisovan introgressed populations both seem to be close to 8% archaic admixed, again suggesting the phenotypic indistinguishability fraction as a cultural threshold with important long term effects.


terryt said...

"In similar models that I have run (aims at understanding the future of race relations), it takes surprisingly few generations (five to ten or so)"

In animal breeds that allow 'breeding up' 1/16 is usually sufficient to ignore that minority ancestry. By that time the hybrid is indistinguishable from the 'purebred' animal. As you say, that equates to 6.25 %.

Anonymous said...

I wonder about skin colour. As I understand it, these cro magnon or modern humans out of Africa were black, while the neandertals were white. i have read somewhere that modern humans in Europe continued to be black or at least dark skinned until they started doing more agriculture (less vitamin D). Or did both the start of agriculture and the whtening of the skin have to do with immigrants from the Middle East?
I know lots of mixed people here in Norway, some also who are born in tropical countries. From just looking around it seems to me that a child born here will be more white than half way between his or her mother and father, while the mixed children born under a tropical sun are just as dark as other, non-mixed people from the same place. Could this be true? Of course my own observations could well be just an anomaly. I know children who are one quarter black, but you would never know from looking at them.

andrew said...

The skin color issue isn't easy to resolve. Neanderthal genes at loci that relate to skin color in modern humans are different from those found in any modern humans, so we can only guess at how these genes expressed themselves. The best guesses are that Neanderthals were lightly pigmented but that light pigmentation in modern humans in Northern Europe was a case a convergent evolution rather than introgression of archaic hominin genes.

The oldest autosomal genome of a modern human (an individual about 24,000 years old in Siberia a bit to the east of Lake Baikal) suggests a person who was at least as darkly pigmented as a modern Southern European.

As far as I know, there is almost no ancient DNA from modern humans for the next 18,000 years or so (until the Copper Age and Otzi). And, at those ages skin and hair are generally not preserved. So, we have no direct evidence at all about pigmentation in Cro-Magnons (the first modern humans in Europe) or any other pre-Neolithic revolution modern humans, except the conclusions we choose to draw from cave paintings and petroglyphs (which were not necessarily even intended to be photo-realistic to the best of the artists abilities).

We base our best estimates of when diversification in pigmentation happened in Eurasians on mutation rate dating of pigmentation genotypes in modern populations relative to ancestral genotypes in Africa. Broadly speaking, scholarly estimates range from ca. 20kya to 60kya, with the Ma'lta genome released this month favoring an earlier date.

The most widely accepted view of the evolution of light pigmentation in West Eurasians sees it as a selective adaptation to living at higher latitudes, not to agriculture. Light pigmentation in Northern Europeans likely evolved in hunter-gatherers before agriculture was adopted there and later migrants from Southern Europe and West Asia who brought agriculture to the North were probably darker skinned than the new immigrant populations of the early Neolithic and metal ages, at least initially. But, since lighter skin confers selective advantage at high latitudes due to issues like Vitamin D (which has an immune function rivaling Vitamin C and hence is important in disease resistance) this phenotype would have been positively selected for within admixed populations migrating from lower latitudes even if the admixture percentage was quite low. Light pigmentation genes would have increased in frequency in high latitude populations much more rapidly than one would expect in a selectively neutral gene allele. If anything, modern Middle Easterners have gotten somewhat lighter in the last four or five thousand years due to migration into the region from Northern Europe and Central Asia.

The place where you are born has no impact on your genotype which is determined at conception with roughly equal contributions from each parent. But, it is not impossible that childhood exposure to sunlight has epigenetic effects that impact the phenotype that is expressed given a person's genome.

But, I am quite skeptical of your observations. There is considerably phenotypic variation in pigmentation even among mixed race children of the same parents.

Also, I suspect that if there is variation it is more likely to be due to cryptic (i.e. unknown) European ancestry being more common in people who appear black and migrate to Norway before having children than in people who do so only after having children.

Another factor that could be related to skin color in mixed race children born in different places could be behavioral differences. People who grow up around light skinned people are more likely to be more careful about exposing themselves to the sun and thus having tanning opportunities than people who grow up around dark skinned people.

andrew said...

Also, keep in mind that modern humans left Africa ca. 100,000 to 125,000 years ago by best archaeological estimates and certainly by 65,000-75,000 years ago.

The first modern humans entered Europe about 40,000 years ago.

There was a 25,000-85,000 year ago (with the older dates favored by the more recent research) between the departure of modern humans from African and their first appearance in Europe. And, it is quite possible that there was more population structure (i.e. geographic diversity in genotype and phenotype) at the time of Out of Africa than there is today when Bantu expansion driven by new agricultural technologies and iron implements within Africa has homogenized the population genetics of a large part of Africa. African were probably more diverse in genes and in appearance at the time that the first modern humans left Africa than they are today. So, some of the differences between modern Eurasians and modern Africans may predate the first migration of modern humans out of Africa.

Since modern humans migrated to Europe from well to the North of the majority of sub-Saharan Africans, the proto-Eurasians were probably more lightly pigmented than the average African even on day one in Arabia.

terryt said...

I don't know if you've seen this:



"We date the coalescence of the light skin associated allele at 22–28 KYA."

Andrew's comment:

"The most widely accepted view of the evolution of light pigmentation in West Eurasians sees it as a selective adaptation to living at higher latitudes"

That's consistent with the paper. However you will note in the comments section that Eurologist is adamant that the mutation occurred in 'NW India, Pakistan, and Afghanistan'.