August 25, 2011

No evidence for Neandertal admixture from mtDNA

I've been quite skeptical of the interpretation of the published Neandertal genome as evidence for Neandertal admixture in Eurasians. Of course, such admixture is a possibility, but I remain unconvinced that the Neandertal admixture thesis has managed to shoulder out other interpretations.

A new paper shows how the mtDNA evidence is inconsistent with Neandertal admixture, by looking at ancient Homo sapiens, Neandertal, and modern human mtDNA. The basic idea is simple, that if admixture with Neandertals did take place at the levels predicted (~4%), then we would expected to see some Neandertal-like mtDNA in either ancient or modern humans, but we don't.

There are some alternative explanations for the lack of Neandertal-like mtDNA in modern humans that I have considered:
  1. Modern human-Neandertal crossings were female-male exclusively.
  2. Neandertal mtDNA sequences were weeded out by natural selection, perhaps because Neandertal mtDNA was well-adapted to extreme cold and became maladaptive in modern human bodies, as the latter had better cold-protection technology (tight clothes, for example), and the glaciers retreated anyway.
I can entertain #1, although it's often the case the the group that has the upper hand often mates with the females of the disadvantaged group, and not vice versa. In any case, I don't see any evidence for it, and the lack of archaic-like Y-chromosomes in Eurasians is probably inconsistent with it.

I had a certain degree of appreciation for #2 until recently. If Europeans had Neandertal admixture, and Neandertal mtDNA was maladaptive to a postglacial technological environment, then selection is a reasonable hypothesis.

However, if Neandertal admixture affected all Eurasians equally, as the current theory holds, then there are plenty of people in North Asia that continued to live in glacial-like temperatures and with predominantly meat diets until the present. Why would Neandertal mtDNA be maladaptive in them?

At present I retain my agnosticism on the subject of Neandertal admxture: the Neandertal genome studies have certainly proven one thing: that archaic humans are not irrelevant to the human story. How they are relevant remains to be seen.


AJPA DOI: 10.1002/ajpa.21569

No evidence of Neandertal admixture in the mitochondrial genomes of early European modern humans and contemporary Europeans

Silvia Ghirotto et al.

Neandertals, the archaic human form documented in Eurasia until 29,000 years ago, share no mitochondrial haplotype with modern Europeans. Whether this means that the two groups were reproductively isolated is controversial, and indeed nuclear data have been interpreted as suggesting that they admixed. We explored the range of demographic parameters that may have generated the observed mitochondrial diversity, simulating 3.0 million genealogies under six models differing as for the relationships among contemporary Europeans, Neandertals, and Upper Palaeolithic European early modern humans (EEMH), who coexisted with Neandertals for millennia. We compared by Approximate Bayesian Computations the simulation results with mitochondrial diversity in 7 Neandertals, 3 EEMH, and 150 opportunely chosen modern Europeans. A model of genealogical continuity between EEMH and contemporary Europeans, with no Neandertal contribution, received overwhelming support from the analyses. The maximum degree of Neandertal admixture, under the model of gene flow supported by nuclear data, was estimated at 1.5%, but this model proved 20–32 times less likely than a model without any gene flow. Nuclear and mitochondrial evidence might be reconciled if smaller population sizes led to faster lineage sorting for mitochondrial DNA, and Neandertals shared a longer period of common ancestry with the non-African's than with the African's ancestors.

Link

31 comments:

Aaron said...

Maybe Neanderthal women weren't very attractive.

No seriously, I appreciate the agnosticism on this topic especially since the Neanderthal genome was sequenced to only 1.3x coverage using degraded DNA.

What I don't understand is why we think they were so much different than modern humans. I mean, they diverged from humans 250,000 years ago while Sub-Saharan Africans and Eurasians diverged 150,000 years ago. If they existed still today they would just be considered another ethnic group, not another species. There would be no surprise if there was interbreeding between two ethnic groups.

᧞eandertalerin said...

@Aaron: Neandertals diverged from modern humans between 500.000 and 1.000.000 years ago, according to their genomes and mtDNA. Anatomically they're clearly different from modern humans (including africans).

The neandertal genome has a coverage of a little more than 60%, and contamination levels have been proved to be really low. It was extremely degradated, but precisely, this was the hard work scientists had while reconstructing it, using the DNA from 3 females.

It's no suprise they share no mtDNA with modern humans. Denisovans don't either; the reasons for this are still unknown, but maybe they were lost by drift.

Dienekes said...

@Aaron: Neandertals diverged from modern humans between 500.000 and 1.000.000 years ago, according to their genomes and mtDNA. Anatomically they're clearly different from modern humans (including africans).

I believe that is the date of the divergence in modern-Neandertal genes, which precedes the actual divergence of the populations.

Blackbird said...

That there is 4% of the Neanderthal genome in current non-africans does not imply the same level of past hybridization. It could have been a much more reduced level of actual hybridization, several orders of magnitud less, but the observed 4% is comprised by genes that were selected for afterwards as they did well in the modern human genomes. I guess we will learn more when a better coverage neanderthal genome/s is available

aeolius said...

Just a stab, but a horse-donkey cross is much more likely Jack x Mare =mule
then Jenny x Stallion=hinny
Would that possible also happen in Sapien x Neanderthal?

Andrew Oh-Willeke said...

The mtDNA is the easier issue.

All one has to suppose is that hybrid children ended up in a community that included their mother, and hybrid mothers were either single parents and/or that cross-cultural couples were matrilocal (the former seems more likely to me, with the hybrids in both directions being either children of rape or children of short term liasons that did not produce long term pair bonding on the patrilocal basis that appears to have been the Neanderthal pattern from the one case where ancient DNA has been able to discern an answer to this question). Thus, there may have been incidental sexual intercourse between Neanderthals and modern humans in tribes located near each other, but not Neanderthal-modern human "marriage" that brought Neanderthal women into modern human tribes. (Even, if modern human women married into Neanderthal tribes patrilocally, or modern human men married into Neanderthal tribes matrilocally, their descendants would have disappeared when Neanderthal tribes as a whole went extinct.)

This is a pretty weak assumption given the absence in the fossil record of mixed Neanderthal-modern human tribes, and the fact that "marriage" takes far more sophisticated interaction than a mere one night stand or instance of rape.

Thus, hybrids with human mothers had human mtDNA and went on to leave descendants in their human tribes that eventually peopled the Earth. In contrast, hybrids with Neanderthal mothers would have had Neanderthal mtDNA and would leave descendants in their Neanderthal tribes which eventually went extinct.

Haldane's law, in turn, explains the absence of patriline Neanderthal DNA in modern humans. (Hybrids strongly tend not to be heterozygotes.)

terryt said...

"Just a stab, but a horse-donkey cross is much more likely Jack x Mare =mule
then Jenny x Stallion=hinny"

The reason for that has nothing to do with compatability. Hybrids born to a horse mother are larger than those to a donkey mother, and so are more useful.

"It's no suprise they share no mtDNA with modern humans. Denisovans don't either; the reasons for this are still unknown, but maybe they were lost by drift".

Drift is the most likely expalnation as far as I'm concerned. No ancient 'modern' human haplgroups survive in Europe either, especially Y-haps. European haplogroups may not be much older in Europe than the Neolithic.

Amanda S said...

I think that there might be some interesting mathematics in this because, only a proportion of individuals that have reproduced and still have surviving descendents, would have descendents which exclusively belong to either the male or female line. There must be some sort of numbers which help to predict the likelihood of being a maternal or paternal line ancestor, such as the number of living male or female descendents one has or the number of generations that have passed in which such descendents or still present.

Onur said...

If Neanderthal-modern human admixture really happened, I think the most plausible explanation for the complete disappearance of Neanderthal mtDNA in post-hybridization modern humans is that Neanderthal mtDNA in modern humans were lost very early through drift when the modern human population level was very low in Eurasia (this isn't so difficult in uniparental DNA).

However, if Neandertal admixture affected all Eurasians equally, as the current theory holds, then there are plenty of people in North Asia that continued to live in glacial-like temperatures and with predominantly meat diets until the present. Why would Neandertal mtDNA be maladaptive in them?

Maybe Neanderthal mtDNA in modern humans were wiped out by natural selection already before the colonization of the cold latitudes by modern humans.

German Dziebel said...

"There are some alternative explanations for the lack of Neandertal-like mtDNA in modern humans that I have considered:

Modern human-Neandertal crossings were female-male exclusively.
Neandertal mtDNA sequences were weeded out by natural selection, perhaps because Neandertal mtDNA was well-adapted to extreme cold and became maladaptive in modern human bodies, as the latter had better cold-protection technology (tight clothes, for example), and the glaciers retreated anyway."

There's a third one. All traces of "Neanderthal admixture" are in fact evidence of common descent with Neanderthals that non-Africans preserved better than Africans and Asians, Amerindians and Papuans better than Europeans and Africans. European mtDNA's show a striking lack of lineages from mhg M, which does suggest that some of the original Eurasian diversity was lost in Europeans.

apostateimpressions said...

Perhaps it is anachronistic to expect Neanthertals and Sapiens to have mixed in some sort of muliticulti 'melting pot', where a Sapiens man took a Neanderthal wife (Neanda lol) to his family. Presumably Neanderthals and Sapiens were aggressive competitors with deep-seated hostility and whose respective numbers were limited by the same environment.

Conceivably Neanderthal males raped Sapiens women and some of the children were raised by Sapiens. That being so, the genetic mixture wouldnt show up in mtDNA.

Even today we can observe situations where rapes go predominantly in one direction. Consider the situation in Oslo for instance where every single one of the hundreds of rapes perpetrated in the city over the last decade have been committed by recent immigrants. That too wouldnt show up in mtDNA.

Or else any mixed offspring born of Neanderthal women were raised among Neanderthals or killed by them. No Neanderthal mtDNA would show up in Sapiens unless a hybrid were born to a Neanderthal woman yet raised and assimilated into Sapiens in the first generation.

eurologist said...

Andrew,

I was thinking about something similar to your thoughts. I agree that hybrid offspring would have very likely been raised in their respective mother's clans.

There may have been a long, 80,000-year window for hybridization. However, there is little record of AMHs as generalists, populating extremely different biotopes before ~50,000 years ago. Thus, it is likely that for a long time, AMHs outside Africa did what they knew best and what their body had evolved for: hunting in dryish, hot, savanna-like places. Only at the fringe would they be in contact with Neanderthals or Denisovans.

Neanderthals were clearly cold-adapted and liked to hunt in cool, forested areas. We don't know much about the latter, except that their type locality also conforms to that. Hybrid offspring would have had a hard time following the mobile lifestyle of AMHs in their hot environments, unless they happened to inherent the necessary attributes from the AMH side, and were male (motivated and capable of joining male AMH groups in the hunt).

Sure, there may have been a small fraction of hybrid women leaving for AMH camps (because they looked and acted more similar/ more familiar, or because Neanderthals expelled them) - but they would have been social outcasts and likely not preferred partners for those males that tend to have the most children. Same goes for their female children and grandchildren, since their non-AMH admixture was likely visible for at least three generations.

saphorr said...

Isn't there a third possibility, that Neanderthal mtDNA was just lost through genetic drift?

Population sizes at the time would have been low, and that's supported by the frequency of the Neanderthal genes.

Why couldn't some Neanderthal autosomal DNA have survived, while all the pure Neanderthal matrilines and patrilines went extinct?

Dienekes said...

Isn't there a third possibility, that Neanderthal mtDNA was just lost through genetic drift?

Genetic drift is what this paper considers, and it's not enough. If there was admixture at the levels predicted by the autosomal evidence, then drift alone could not have made Neandertal mtDNA extinct.

It's either negative selection, or much lower levels of Neandertal admixture.

terryt said...

"All one has to suppose is that hybrid children ended up in a community that included their mother ... and/or that cross-cultural couples were matrilocal ... long term pair bonding on the patrilocal basis that appears to have been the Neanderthal pattern from the one case where ancient DNA has been able to discern an answer to this question"

I think you could be on to something there. I suspect that matrilocality is much more ancient than patrilocality as it is reasonably common in many herd animals. And, because Y-chromosome lines tend to be replaced more easily than mtDNA lines any Neanderthal male haplogroups in hybrids with 'modern' human women would be unlikely to survive for too many generations.

"European mtDNA's show a striking lack of lineages from mhg M, which does suggest that some of the original Eurasian diversity was lost in Europeans".

I think it is more likely that M never arrived in Europe.

"there is little record of AMHs as generalists, populating extremely different biotopes before ~50,000 years ago".

Yes, and that is interesting in itself.

Fanty said...

"Genetic drift is what this paper considers, and it's not enough. If there was admixture at the levels predicted by the autosomal evidence, then drift alone could not have made Neandertal mtDNA extinct."

I dont know.

There are those compter simulations with drift, that show that a group of 50X and 50Y can very easily lose one of the 2 groups in 5 generations only, by totaly random drift. While this clan of 100 members would still have 50/50 mixed aDNA.

And if one imagines, that it seems that half of the population of modern countries stem from single males that lived less than 10K years ago suggests that in the past 10K years 99.999% of all existing lineages have gone extinct by whatever reason (all the other males that this single man hunted with, whos lineages havent made it....) .

German Dziebel said...

"I think it is more likely that M never arrived in Europe."

Since mtDNA hg C has been found in Europe at low frequencies, it's likely that some of "M" lineages made it to Europe but then were lost through drift.

terryt said...

"Since mtDNA hg C has been found in Europe at low frequencies, it's likely that some of 'M' lineages made it to Europe but then were lost through drift".

It's even more likely that just line C made it to Europe, and that reasonably recently.

terryt said...

"There are those compter simulations with drift, that show that a group of 50X and 50Y can very easily lose one of the 2 groups in 5 generations only, by totaly random drift".

The figure I've read is one lineage/generation. So 100 couples, each with individual haplogroups, would reduce to one Y and one mtDNA in 100 generations, about 2000-2500 years.

" it seems that half of the population of modern countries stem from single males that lived less than 10K years ago suggests that in the past 10K years 99.999% of all existing lineages have gone extinct by whatever reason"

And that certainly explains the lack of any Neanderthal haplogroups in the modern population.

terryt said...

Sorry. That loss of one per generation depends on stable population numbers.

eurologist said...

Yes, and that is interesting in itself.

I just read the entry about HLA loci in Hawk's blog, who writes:
"The HLA pattern may suggest a more widespread pattern of mixture across Asia, which was later overwritten by population movements of people who didn't have Denisovan ancestry."

I know it is just speculation at this point, but I have argued before that y-DNA haplogroup DE may be associated with such an early, almost unsuccessful ooA event perhaps 130,000 - 100,000 years ago.

The thing is, CT(xDE, or CF) left a very nice and very old trail along its way. DE did not.

German Dziebel said...

"It's even more likely that just line C made it to Europe, and that reasonably recently."

It's unlikely that a separate C lineage, not identical to Amerindian ones, found in Iceland is a recent arrival. Hg C fits a broader pattern of Europe being dominated by later N-derived lineages (U, etc.), with deeper M (C, D in Saami) and N (X) clades surviving on the margins and/or at low frequencies.

Andrew Oh-Willeke said...

The trouble with genetic drift is that genetic drift does little to cull lineages during periods of population expansion.

Also, you need to be able to have a drift scenario that entirely wipes out Neanderthal mtDNA and Neanderthal Y-DNA while preserving 2.5% to 4% autosomal DNA, which means that the drift that removes the matrline and patriline ancestry needs to take place a very small number of generations after the initial admixture event, or with a lower effective population size than is indicated by other data.

The more admixture events you have, the more improbable a genetic drift scenario becomes. With a minimum effective population size for AMH during the admixture era of say 4,000, you need more than 100 admixture events (in a gradualist scenario, stretched out over thousands of years, one every generation or two on average) to get 2.5% autosomal reaching fixation absent strong selective effects (which are seen in HLA but not in most of the other Neanderthal genome which is spread over a wide variety of loci at low frequencies in modern humans).

Wiping out a lineage that is introduced once with genetic drift in a population that is rapidly expanding (as frontier AMHs interacting with Neanderthals would have been) can happen. Wiping out a lineage that is introduced a hundred separate times with genetic drift in a population that is rapidly expanding is dramatically less likely.

(Yes, you can have intermediate scenarios, for example, like one of ten modern human clans reaching fixation at say 25% Neanderthal and then admixing in turn with other early human populations, that don't require so many independent admixture events, but losing an mtDNA lineage that accounts for 25% of autosomal DNA in the hybrid clan is also highly improbable unless you have an organizing prinicple.)

In short, the math of making random genetic drift remove all Neanderthal mtDNA and Y-DNA from the AMH population in a long term mixed marriage scenario while leaving 2.5% to 4% autosomal, if one assumes that most of the Neanderthal DNA was selectively neutral and have reasonable assumptions about effective AMH population size and a generally expanding AMH population just doesn't work out.

terryt said...

"Hg C fits a broader pattern of Europe being dominated by later N-derived lineages (U, etc.)"

Most likely the oldest survivng mtDNA haplogroups in Europe. And most are R-derived rather than basal N, which implies a slightly later arrival than N's original expansion, or even that of M.

"with deeper M (C, D in Saami)"

Surely one would have to think that the Saami are even later arrivals in Europe. They tend to be confined to the very north, a regionthat was under ice until around 10,000 years ago. Any C and D in Europe almost certainly derives from already-diversified eastern C and D haplogroups.

"and N (X) clades surviving on the margins and/or at low frequencies".

And those N claders on the margins are most likely survivals from before the R-derived haplogroups' expansion.

"The trouble with genetic drift is that genetic drift does little to cull lineages during periods of population expansion".

But I seriously doubt that we can assume continuous population expansion since the Early Upper Paleolithic, or even since modern humans first entered Europe.

"The more admixture events you have, the more improbable a genetic drift scenario becomes".

Admixture alone would also lower the diversity of pre-existing haplogroups.

"In short, the math of making random genetic drift remove all Neanderthal mtDNA and Y-DNA from the AMH population in a long term mixed marriage scenario while leaving 2.5% to 4% autosomal, if one assumes that most of the Neanderthal DNA was selectively neutral and have reasonable assumptions about effective AMH population size and a generally expanding AMH population just doesn't work out".

But the modern European haplogroups are perhaps no older than the Neolithic yet I doubt anyone would consider that all autosomal DNA in modern Europeans is that recent.

German Dziebel said...

"which implies a slightly later arrival than N's original expansion, or even that of M."

Exactly. Europeans must have lost most of their ancient "eastern" M and N components as they colonized Europe from Asia.

"Surely one would have to think that the Saami are even later arrivals in Europe. They tend to be confined to the very north, a region that was under ice until around 10,000 years ago."

You're surely wrong. The Saami are remnants of the original population of Europe that migrated up north with the retreat of the ice. Their "eastern" component, which diminished in frequencies since the evolution of the "western" U, etc. component, gives us an idea of what Europe looked like in the Upper Paleolithic.

terryt said...

"The Saami are remnants of the original population of Europe that migrated up north with the retreat of the ice".

Yes, 'with the retreat of the ice'. They almost certainly entered the ice free regions of Europe from where the majority of C and D haplogroups are found: from the east. So they give us no idea of 'what Europe looked like in the Upper Paleolithic'.

German Dziebel said...

"They almost certainly entered the ice free regions of Europe from where the majority of C and D haplogroups are found: from the east. So they give us no idea of 'what Europe looked like in the Upper Paleolithic'."

And how did they absorb and differentiate within mtDNA U5 shared with Berbers, my friend? See here http://www.ncbi.nlm.nih.gov/pubmed/15791543. The Franco-Cantabrian refuge is a good explanation. Is it in Europe in your opinion? Terry, you have to study the evidence before you make your statements.

terryt said...

"And how did they absorb and differentiate within mtDNA U5 shared with Berbers, my friend?"

Surely you're not claiming every mtDNA line entered Europe together? I could find the link you gave by the way. The Franco-Cantabrian refuge is a good explanation is sufficient for European U but I doubt that C or D are that old in Europe. They are both basically eastern haplogroups.

Onur said...

My understanding is that Saami are a mixture of the oldest extant Europeans with late (a few thousand years before present) Mongoloid immigrants from the east. I think the Mongoloid components of Saami are entirely from the people who brought the Uralic tongue to the present-day Saami area while the Caucasoid components of Saami are primarily (but not entirely) pre-Uralic and local to the present-day Saami area.

terryt said...

"I think the Mongoloid components of Saami are entirely from the people who brought the Uralic tongue to the present-day Saami area"

And probably includes mtDNA haplogroups C and D.

Onur said...

And probably includes mtDNA haplogroups C and D.

yes