September 13, 2011

Strong isolation between Neandertals and modern humans

In my comments on the recent Hammer et al. paper on Africa, I noted that the naive Out of Africa edifice has been shattered by recent discoveries, and attempts to patch it up by postulating a little admixture here, a little admixture there are largely unconvincing. We need a radically new model, so it was a nice surprise for me to see a new paper that attempts to do just that in a quantitative manner.

Currat & Excoffier address two of the holes of the current working model:
  1. Why Chinese don't have less Neandertal admixture than Europeans, even though Neandertals were a West Eurasian distributed species
  2. Why no Neandertal mtDNA is detected in modern samples
I will begin with #2, which is tangential to the paper's overall simulation framework. The authors write:
We thus simulated the genealogy of 20 samples of 100-mtDNA sequences scattered over all Eurasia, and we estimated the fraction of these current lineages to be of Neanderthal ancestry by using a conservative interbreeding success of 2% for the hybrids. Among 10,000 simulations of this process, we could never observe any mitochondrial sequence of Neanderthal origin in our samples. We thus conclude that an interbreeding success smaller than 2% for Neanderthal-human hybrids is fully compatible with limited Neanderthal nuclear introgression and with no introgression of mtDNA.
This is contradicted by another recent study by Ghirotto et al. focused entirely on the problem of Neandertal mtDNA (non)-intogression. I tend to side with G. et al. on this issue, as we currently have not only 100-strong samples from around the world, but, literally, tens of thousands of mtDNA samples, and perhaps even more if we account for commercial testing: no Neandertal sequences have turned up. So, I don't think it's the case that Neandertal mtDNA admixture is so low that we're not finding it in small (100-strong) samples.

In any case, the current paper's major contribution is tackling the problem of the equidistribution of Neandertal admixture across Eurasia, despite the fact that Neandertal was a West Eurasian distributed species.
The solution is simple: extend the Neandertal range eastwards, all the way to the Altai. That way, there is no mystery why French and Chinese have the same degree of Neandertal admixture: the former picked it up en route to France, the latter in their eastward journey to China. The extension of the Neandertal range to the east is supported by the close relationship of the Denisovan genome to Neandertals as well as the Okladnikov sample from Uzbekistan, so it is not simply a kludge, but a reasonably well-supported position.

I remain unconvinced for a variety of reasons:
  1. The extension of the Neandertal range works only if we assume that the Chinese ended up in China largely via an inner Asian route, rather than a coastal migration that has garnered increasing support from analyses of mtDNA and Y-chromosomes
  2. Moreover, even if the ancestors of the Chinese did follow a route via the heartland of Eurasia, that would not explain why south Eurasians would also possess the same levels of Neandertal admixture; that would require that not only East Eurasians, but also Australo-Melanesians and "Ancestral South Indians" to follow an inner Asian route that would bring them in protracted contact with Neandertals
  3. The expansion model also sidesteps the big elephant in the room: the fact that modern peoples are not necessarily descended primarily from the early modern humans who lived in the same localities as themselves. Neolithic and post-Neolithic events are increasingly thought to have shaped gene pools, so it is not clear, for example, to what extent the modern French are descended from Upper Paleolithic Europeans who ended up in France at the time when Neandertals were still in existence, and thus had the opportunity to mate with them along their route.
With respect to point #1, the authors write:
Although we have modeled the Asian range to extend up to the Altai region north of the Himalayas, we cannot be certain that the ancestors of East Asians migrated through this region. However, the facts that Papua New Guineans show signals of hybridization with another hominin (Denisovan) (2) and that their ancestors are likely to have followed a coastal southern route to the Pacific (19, 20) suggest that the Denisovan range must have extended more to the south and that the ancestors of East Asians may have indeed traveled north of the Himalayas, above the Denisovan range.
A northward migration route for East Asians and a southward one for Australo-Melanesians would solve the problem of "why Chinese are as Neandertal-admixed as the French" but would create a new problem of "why Papuans are also as Neandertal-admixed as the French", unless we derive everybody from the Altai.

The authors write:
Under our model of hybridization during range expansions, similar amounts of Neanderthal ancestry in France and China (Fig. S2) are more often observed if the geographical range of Neanderthals extended up to the Altai Mountains north of the Himalayas. Indeed, a hybridization range restricted to Europe, the Middle East, and the Caucasus region (including the brown and green areas in Fig. 1) would always lead to a much larger Neanderthal introgression level in Europe than in China (Figs. S2 and S3, light bars).
It is important to note the differences between this model and the one in the Reich et al. and Green et al. papers from last year; the latter postulated a brief, but relatively intense episode of admixture in West Asia, and the subsequent colonization of the world by (slightly Neandertal-admixed) modern humans with minimal subsequent admixture.

The Currat & Excoffier favored model, on the other hand, posits exceptionally low admixture rates (or mostly infertile hybrids) but with repeating admixture events across a large geographical range that is extended far to the east.

I am not very supportive of either model:
  • R&G can't convincingly explain why modern West Eurasians have the same amount of Neandertal "admixture" as every other Eurasian. They inhabited the same space as Neandertal for tens of thousands of years and had maximal opportunities to admix with them
  • C&E save the phenomena by postulating an eastward Neandertal population, but it is not clear to me how relevant the inner Asian route was to the peopling of Asia as a whole
In any case, the C&E paper is an important contribution to the emerging new debate about human origins. The central prediction of their model, that modern human/Neandertal admixture was no easy thing is certainly consistent with everything we know about human behavior. Homogamy seems to be a tendency of our species, and heterogamy between populations separated by hundreds of thousands of years of separate evolution cannot have been frequent.

Hopefully, next year, we will see even more realistic simulation studies that include the second elephant in the room, archaic admixture in Africans. The discovery that Africans possess a degree of archaic African admixture from Homo populations that branched off before the split of modern humans and Neandertals leads to the inevitable conclusion that Eurasians (who lack this archaic African admixture) would appear closer to Neandertals based on the D-statistic used by R&G.

Happy were the days when human evolution could be viewed as a simple tree branching process, but now we have at least four players in the field (modern humans, Neandertals, Denisovans, archaic Africans) and we are bound to have new full ancient H. sapiens genomes. Things are bound to get more interesting.

UPDATE:

Laurent Excoffier wrote to me in an e-mail:
Hi Dienekes,

I've read your blog and liked your quite balanced discussion. However I have
two comments on your main concerns:

1) You seem to conclude that our model assumes that Chinese only followed a
northern route, but what we portray in our Figure 3 is a map representing
introgression event densities, which does not necessarily represent the
migrations followed by the ancestors of Chinese people. In our simulations,
migrants could indeed go South of the Himalaya, but since we assumed there
were no Neanderthals there, no admixture events occurred in this region. So,
our model does not strictly assume a main northern migration route to Asia,
but rather a concentric migration both North and South of the Himalayas,
with some later secondary contacts in Eastern Asia.
An important message of our paper is to say that whatever the migration
routes to Asia, Asian populations having interacted with Neanderthals for
about the same amount of time (or in space) would have about the same final
level of introgression. So we would not (and never did) say that East Asian
have followed a migration route North of the Himalayas.

2) Concerning you big elephant, we indeed believe that current Europeans are
mainly descending from Paleolithic people, and we have a specific paper on
this (Currat M, Excoffier L (2005) The effect of the Neolithic expansion on
European molecular diversity. Proceedings of the Royal Society London B 272,
679-688), and a later paper examining and explaining why local genes are likely
to massively introgress the genome of populations invading the territory of
a local species (Currat M, Ruedi M, Petit RJ, Excoffier L (2008) The
hidden side of invasions: Massive introgression by local genes. Evolution
62, 1908-1920). It thus appears very likely to us that the Paleolithic gene
pool of Europeans has largely persisted in modern humans, despite posterior
Neolithic migrations, if we assume that Neolithic populations mixed to some
appreciable extent with local paleolithics during their expansion over
Europe.

Anyway, it is good that our paper promotes discussions on the dynamics of
the settlement, and that spatial aspects of human evolution are now given
some importance.

best

laurent
With respect to point #1, I acknowledge that the C&E model does not explicitly choose either a northern or southern migration route for the ancestors of the Chinese. However, I do believe that it effectively chooses a mainly northern route. This is a simple consequence of geometry: concentric radiation of migrants results in more migrants flowing through the larger volume of the interior of the Eurasian continent (where they would have the opportunity to admix with eastern Neandertals) rather than through the narrower coastal band (where they would not).

It is also an interesting question whether the concentric model would lead to non-distinguishable admixture levels in East Asians and Australo-Melanesians, since -under the concentric migration model- the latter would mostly travel through regions of low-to-non-existent Neandertal occupation, whereas the former would not.

With respect to point #2, it is of course a hotly debated issue whether modern Europeans are largely descended from Paleolithic, Neolithic, or even post-Neolithic migrants. My reasons for preferring a more recent ancestry are:
  • First, the published ancient DNA work, which showed a chasm between Mesolithic and modern mtDNA in central Europe, and seemingly non-existence of the main current European Y-haplogroup R1 in both Central Europe and France Neolithic sites (Treilles).
  • Second, the observation of Fst values between European and West Asian populations are approximately 1/3 of those between West and East Eurasians. If the separation between Europeans and West Asians was effected close to the time of the Upper Paleolithic colonization of Europe (40ky), and would thus have diverged genetically only slightly less than West Eurasians and East Asians have.
The issue can only be settled, of course, if we are ever lucky enough to obtain ancient DNA from Upper Paleolithic Europeans.

PNAS doi: 10.1073/pnas.1107450108

Strong reproductive isolation between humans and Neanderthals inferred from observed patterns of introgression

Mathias Currat, and Laurent Excoffier

Recent studies have revealed that 2–3% of the genome of non-Africans might come from Neanderthals, suggesting a more complex scenario of modern human evolution than previously anticipated. In this paper, we use a model of admixture during a spatial expansion to study the hybridization of Neanderthals with modern humans during their spread out of Africa. We find that observed low levels of Neanderthal ancestry in Eurasians are compatible with a very low rate of interbreeding (<2%), potentially attributable to a very strong avoidance of interspecific matings, a low fitness of hybrids, or both. These results suggesting the presence of very effective barriers to gene flow between the two species are robust to uncertainties about the exact demography of the Paleolithic populations, and they are also found to be compatible with the observed lack of mtDNA introgression. Our model additionally suggests that similarly low levels of introgression in Europe and Asia may result from distinct admixture events having occurred beyond the Middle East, after the split of Europeans and Asians. This hypothesis could be tested because it predicts that different components of Neanderthal ancestry should be present in Europeans and in Asians.

15 comments:

German Dziebel said...

"a coastal migration that has garnered increasing support from analyses of mtDNA and Y-chromosomes"

There's no support for a coastal route in either mtDNA and Y-DNA. The reason the whole idea of a coastal route exists is because scientists originally believed that humans must have bypassed Neandertals by skirting their range at a tangent.

"So, I don't think it's the case that Neandertal mtDNA admixture is so low that we're not finding it in small (100-strong) samples."

It has nothing to do with the sample size. mtDNA is just too variable and has lots of hot spots with back mutations - traces of Neandertal lineages are poorly detectable by genetic methods but they must be there, judging by all other systems evidence. Read this: "Recombination or mutation rate heterogeneity? Implications for Mitochondrial Eve," by Erika Hagelberg // TRENDS in Genetics Vol.19 No.2 February 2003.

"extend the Neandertal range eastwards, all the way to the Altai. That way, there is no mystery why French and Chinese have the same degree of Neandertal admixture."

Nonsense. With no Neandertals around, Amerindians have higher frequencies of some Neandertal-identical haplogroups such as X chromosome B006 and HLA A*02, C*07:02 (comp. Vindija) than any Old World groups.

The Neandertal admixture theory - in any form - misinterprets signs of common descent of humans from non-African hominids as archaic interbreeding events.

᧞eandertalerin said...

"I am not very supportive of either model"

So, what's your model, if I can ask you?

"The discovery that Africans possess a degree of archaic African admixture from Homo populations that branched off before the split of modern humans and Neandertals..."

Is that finally confirmed? As far as we know, there's no hominid genome to compare with. There's even no known hominid in the region, and the current dates for the interbreeding event (35.000 years bp) are quite odd, considering this hominid lived there for 700.000 years. Maybe we should be more cautious before assuming this archaic admixture really took place?

"...leads to the inevitable conclusion that Eurasians (who lack this archaic African admixture) would appear closer to Neandertals based on the D-statistic used by R&G."

Maybe. But what about Melanesians? And Yorubas? The last ones don't appear to have any neandertal nor archaic african admixture. And B006? It is not found in sub-saharan Africa, and it's shared with neandertals.

Dienekes said...

So, what's your model, if I can ask you?

My model is that Eurasians lack archaic African admixture, and that the latter is partly responsible for inflated diversity values observed in modern Africans.

I don't discount the possibility of real Neandertal admixture, however the D-statistic model that was used to prove that admixture in the original papers needs to be updated to account for the possibility that the observed pattern is due to archaic admixture in Africans.

Maybe. But what about Melanesians? And Yorubas? The last ones don't appear to have any neandertal nor archaic african admixture. And B006? It is not found in sub-saharan Africa, and it's shared with neandertals.

Melanesians do appear to have an archaic element not shared by Eurasians, and the simplest explanation is that they indeed have some type of archaic admixture.

As for Yoruba, I don't believe they have no archaic African admixture. Please refer to my comments on the Hammer et al. paper; the 2% that they managed to find is an underestimate of archaic admixture (as they themselves suggest), because drift and recombination has obliterated the evidence for non-recent admixture events; non-recent admixture events persist IMO as inflated diversity values in all modern Sub-Saharan populations.

Annie Mouse said...

By far the simplest explanation for a consistent Neanderthal admixture is that it occurred before the European/Asian split.

If the the so called "Neanderthal genes" truly are Neanderthal, a fact that I have yet to be convinced.

Andrew Oh-Willeke said...

"Indeed, a hybridization range restricted to Europe, the Middle East, and the Caucasus region (including the brown and green areas in Fig. 1) would always lead to a much larger Neanderthal introgression level in Europe than in China (Figs. S2 and S3, light bars)."

To expand a bit on Annie Mouse - the Neanderthal admixture that we see today occurred before the European/Asian split in the Middle East (and ended fairly early on after perhaps 1,000-2,000 years when the two species were in contact there).

Admixture probably did keep happening in Europe. But, modern Europeans derived very little of the ancestry from Upper Paleolithic Cro-Mags of 50,000kya to 30,000kya who would have had greater levels of Neanderthal admixture. Much of Europe was depopulated in the LGM. There were massive infusions of new unadmixed populations post-LGM, in the Neolithic and in any subsequent demographic migrations into Europe. The higher levels of Neanderthal admixture in Upper Paleolithic Cro-Mags was profoundly diluted by later demographic waves to the point of being smaller than the random variation seen in modern populations.

Denisovians are too distinct from Neanderthal and have a separate source population so they can be Neanderthal sources in Asia. The genetic legacy of Neanderthals is largely non-overlapping between East and West, so it could be mostly separate admixture events in parallel and separate proto-European and proto-Asian communities just post-separation rather than in a single community. If the process and duration were similar it would produce the same result.

Re the mtDNA issue: Human moms pregnant from Neanderthal dads have kids in human tribes and their descendants don't go extinct. Neanderthal moms pregnant from human dads have kids in Neanderthal tribes and their descendants go extinct. As long as human-Neanderthal relationships don't last more than a few months (98% of the time; a couple flukes wouldn't screw up the pattern badly enough to allow the mtDNA lines to go), it just works out that way.

Re Lack of Neanderthal Y-DNA: Haldane's Law.

Re very low fertility: This is a problem, because we need enough fertility for hybrids to have six successive generations of fertile offspring so 2% fixation can be reached. Lower admixture rates accomplish almost the same thing as lower ferility of mixed couples and/or hybrids.

Onur said...

My model is that Eurasians lack archaic African admixture, and that the latter is partly responsible for inflated diversity values observed in modern Africans.

I don't discount the possibility of real Neandertal admixture, however the D-statistic model that was used to prove that admixture in the original papers needs to be updated to account for the possibility that the observed pattern is due to archaic admixture in Africans.


That is one of my favorite models.

My other favorite model is that modern humans admixed with Neanderthals, however limited, only prior to the emergence of a cognitive evolution in modern humans that triggered the Upper Paleolthic Revolution (thus before the splits of Eurasians) and after that completely shunned admixing with Neanderthals (even if they sexed with them, they rejected children born from such sexual intercourses) seeing them much inferior.

AP said...

Laurent Excoffier's points make sense. A single Southern Route can also explain the admixture that possibly occurred in the near/mid east. Plus we do have Upper Paleolithic mtDNA that indicates continuity for U and H through the present.

terryt said...

"Happy were the days when human evolution could be viewed as a simple tree branching process, but now we have at least four players in the field (modern humans, Neandertals, Denisovans, archaic Africans)"

I'm sure we will also find South Asian and SE Asian contributions too.

"So,
our model does not strictly assume a main northern migration route to Asia,
but rather a concentric migration both North and South of the Himalayas,
with some later secondary contacts in Eastern Asia".

Which is the way I see it.

"There's no support for a coastal route in either mtDNA and Y-DNA. The reason the whole idea of a coastal route exists is because scientists originally believed that humans must have bypassed Neandertals by skirting their range at a tangent".

German. I agree with you on that. And the hypothesis was also invented to explain an apparently short time frame between the arrival in Australia and the development of the Upper Paleolithic. I think we can now see that those two events are not necessarily closely related. 'Modern Human' does not always equal 'Upper Paleolithic Human'.

As for 'no support for a coastal route in either mtDNA and Y-DNA' perhaps we should drop the word 'coastal'. I see y-hap F and mtDNA M as having moved south of the Himalayas through South Asia, but not along the coast. They appear to have entered SE Asia via Northeast India. But Y-hap C and mtDNA N do not easily fit a route through South Asia. On the other hand their modern distribution
quite easily fits a route north of the Himalayas and then down the east coast of Eurasia.

Onur said...

thus before the splits of Eurasians

into the various groups that would form the bases of the non-African races.

eurologist said...

Onur,

One of my favorite models likewise is what you described as "your other favorite model." People too easily forget that there could have been more than one ooA, that there is evidence for a very early ooA (which is also favored by climate), but no to scant evidence of AMHs before the UP in Europe, most of West Asia, or the East even in places where we would expect to find them (were it not for the presence of Neanderthals and perhaps Denisovans and erectus).

Where did AMHs live all this time? The place that had vast regions with a similar climate and landscape as the Africa AMHs were used to is the Indian subcontinent (during that time!). If admixture with Neanderthals took place on the route to there and at the climatic fringes (remember that Neanderthals expanded eastward just at exactly this time), admixture had several 10,000 years to accumulate and mix to a fairly homogenous level, and no more admixture is required once UP people left to the West and East.

The fact that Chinese Neanderthal admixture is different from European is not that difficult to explain, after ~40,000 years, if you also realize that most admixture is "private" anyway, if I understand Hawks correctly. I.e., if you pick two Europeans, most of their admixtures are single bits that do not match - except for pieces that are highly selected, of course, which would be expected to be homogenous but different in populations with vastly different climate and fauna.

Onur said...

Eurologist,

So, if I understood you correctly, this is your model for all non-Africans:

First Out of Africa, Neanderthal admixture on the way to India, refuge in the Indian subcontinent, and then some ten thousands years later Out of India.

Your model sounds very similar to Luis Aldamiz' model. In contrast, I didn't define any specific refuge location in my model. May I ask you your evidence for your Indian refuge scenario?

BTW, my two favorite models don't exclude each other, as both Africans and non-Africans may have various levels of admixture from various sorts of archaic humans.

᧞eandertalerin said...

"however the D-statistic model that was used to prove that admixture in the original papers needs to be updated to account for the possibility that the observed pattern is due to archaic admixture in Africans."

How can we update it, if we have no DNA from these hypothetical archaic Africans?

"Melanesians do appear to have an archaic element not shared by Eurasians, and the simplest explanation is that they indeed have some type of archaic admixture."

Yet if we "update" the data having in mind this archaic-African admixture, wouldn't the 5% of Melanesians be affected as well?

"As for Yoruba, I don't believe they have no archaic African admixture. "

As far as we know, Hammer only studied the genomes of some African populations. If these archaic hominids lived in central Africa, isn't that odd to find that so many different African groups interbred with them only 35.000 ybp, considering that they're not closely related to each other?

Another paper from Wall & Plagnol mentioned that West Africans perhaps have 5% of archaic admixture, but I wouldn't trust it, since the same authors in another paper also claimed that Europeans were as much as 14% archaic.

Dienekes said...

How can we update it, if we have no DNA from these hypothetical archaic Africans?

We'll never have DNA from archaic Africans from the jungles of Central Africa. I'm sure, however, that over the next years FGS and better computational models will allow us to mine even more archaic African DNA from the
genomes of modern Sub-Saharans, beyond what Hammer et al. were able to achieve.

Yet if we "update" the data having in mind this archaic-African admixture, wouldn't the 5% of Melanesians be affected as well?

No, because Melanesians have an excess of this relative to other Eurasians, hence Denisovan admixture estimates should be fine.

isn't that odd to find that so many different African groups interbred with them only 35.000 ybp, considering that they're not closely related to each other?

I am not sure what you find odd. What I find odd is that modern humans admixed with archaics only 35ky BP, if anatomically modern humans existed in Africa 195ky BP. The picture is still very fluid and oddities like this will keep cropping up until the dust settles.

n/a said...

"The issue can only be settled, of course, if we are ever lucky enough to obtain ancient DNA from Upper Paleolithic Europeans."

This is evidently in the works.

terryt said...

"Human moms pregnant from Neanderthal dads have kids in human tribes and their descendants don't go extinct. Neanderthal moms pregnant from human dads have kids in Neanderthal tribes and their descendants go extinct".

That does explain the evidence.

"Where did AMHs live all this time? The place that had vast regions with a similar climate and landscape as the Africa AMHs were used to is the Indian subcontinent"

Yes, but ... both mtDNA M and N appear to have undergone a period of drift before they expanded. This period of drift is most easily explained as a period spent in some region of limited extent. India hardly fits that scenario. In fact the sudden expansion of M fits a sudden arrival in India very well.

"May I ask you your evidence for your Indian refuge scenario?"

Yes.

"If these archaic hominids lived in central Africa, isn't that odd to find that so many different African groups interbred with them only 35.000 ybp, considering that they're not closely related to each other?"

Perhaps not. Pygmy mtDNA is not exclusive until relatively 'recently', even amoung Western groups. For example most Western Pygmies are L1c1a, a haplogroup shared with many non-Pygmy inhabitants of Gabon. So admixture with Central African rainforest inhabitants may have been just 35,000 years ago, long after 'modern' humans had left Africa.