This post is such a good summary of the "wallasea problem".
Yet I am not convinced by the Mondal et al work.
First, even by stretching their model parameters to the max, they obtained only a small difference in coalescence times between PNG-Afr and Asi-Afr compared to real data, (compare fig 1 and 4 from their paper).
Second, surely there must be other Eurasian populations that could experience strong bottlenecks and long isolation. Yet the coalescence time between all Eurasians and Afr Yoruba is very consistent, and doesn't exceeds 75 my, that is likely happened after Toba.
I thing the right explanation is the most obvious one -that there was an early HS population in SEA that was mostly but not fully replaced by newcomers after 50 ky but left some genetic trace.
This is a great comment, my only thing to add is that the number of sampled populations of living people is still not really very large, and most of the Eurasian groups come from shared histories in the time frame we're looking at. I do wonder whether the "population Y" signature in South America may reflect a similar effect of demography, certainly it's another case of a long and extended series of founder effects.
Also, the dates in Mondal et al model don't look quite correct. The HS-Neanderthal split is now estimated closer to 800 ky, not less than 600 ky, and there is a deep history of interbreeding between them from around 300ky, not reflected in their model. The alleged "exit from Africa" at 62 ky is unlikely, since by that time there were already sister Y haplo groups F, C, D, E and only E is found in Africa, it is untenable that hugely successful F emerged in Africa and then went extinct there.
The big rise of Eurasian populations began around 55 ky, not in the Holocene as shown in the model, and by 45 ky the former likely already exceeded African population.
It is extremely likely that early HS and Denisovans were already present in SEA. The alleged early HS are always referred to as "extinct ex-OoA", but I prefer to call them middle-paleolithic people that were later mostly replaced by upper paleolithic people coming from West Eurasia.
These are great questions. I don't agree with every point but I do want to encourage more close examination of this kind.
Some small differences of opinion: There is not very clear "Middle Paleolithic" in SE Asia and in Indonesia in particular, at least not in the sense the term is usually defined for India and areas further to the west. The restructuring of technology for all ancient people who inhabited this region is a fascinating piece of the dynamics that I think current ideas are not capturing. If archaeology is relevant to understanding these dispersals at all (which I don't necessarily assume) then the lack of transfer of western technokits into this region is an important element to explain.
Regarding Y haplogroups, I probably have less trouble than many people in accepting a large and rapid divergence of Y haplogroups with near-extinction of haplogroups from closely-related populations. In my expectation, this is likely mediated by natural selection on the fully-linked NRCY. I think the same of most Late Pleistocene and Holocene connections based on Y haplogroups. This makes me different from many people who think about Y haplogroups a lot, I am aware. (But then, I am supervising a PhD on Y haplogroups and have published on them, so there's that!)
You are entirely correct that the timeline of Neanderthal-African divergence is a key element, although this is less critical for methods that bring in linkage like MSMC because the mutation rate difference is not as influential on the estimation of recent events as the recombination rate. Incorporating the deep time element into these models, such as the recurrent gene flow into Neanderthals from Africa, is a challenge considering the number of parameters that must come in for this. That may result in equifinality of models even for reasonable situations, and may not be solvable without ancient DNA more densely sampled than today.
I think you're well-reasoned in considering some African-origin people to have been in SE Asia and island SE Asia in the early Late Pleistocene, I don't think this is unlikely but I do encourage exploring alternatives including a more modern-looking Denisovans. We don't have great ways of testing this.
"In my expectation, this is likely mediated by natural selection on the fully-linked NRCY. I think the same of most Late Pleistocene and Holocene connections based on Y haplogroups. This makes me different from many people who think about Y haplogroups a lot, I am aware."
I've long had a suspicion that the ultimate breakthrough of OOA/proto-Eurasian population must have had some sort fertility-specific physiological component, not just social/cultural or even neurological. Some mutations which accumulated in the Y (or the X) that somehow influenced the fertility patterns of the OOA group - the rate of conception, improved spermatogenesis, sex ratio at birth, etc - that directly contributed to the massive population break-out after the bottleneck.
I would really like to see some researchers attempt to actually test this out by looking at all the defining mutations of our major Y clades - C, D, E, F, and also some of their associated major down-stream sub-clades (R1b, J1, E1b1, O1, etc) - and see if they can identify specific mutations which would indicate these clades were directly positively selected. It would be interesting to see the same thing on a broader scope looking at the CT macro-level too, and also A and B.
I agree that archaeology there is puzzling. For example, UP technology spread like fire very quickly around vast regions of Eurasia up to south Siberia around 45 ky ago, but in India it was gradual and in SEA not visible until recently. But genetic data suggests modern people were already there at least 40 my, and likely longer. Why it is not showing in stone tools is interesting.
Yes the dental data too, but there were also genetic studies I recall that put HS-N split close to 800 ky, significantly earlier than was thought previously.
John, is my interpretation of the RCCR method correct here?
Say you have two populations, popA and popB, with an RCCR value of 0.5 (on a scale from 0 to 1) at 50k ybp. Does that essentially mean that 50% of the unique drift that defines popA as popA distinct from popB (and vice versa) had accumulated by 50k ybp?
Looking at "Table 1 Best fitted parameter values of model A," they have the separation of African and OOA populations at 62k ybp, with a 10k year long bottleneck. The estimated effective population size of OOA during that time is only 726 - that seems like a very, very low population size to be able to ride out a 10k year bottleneck without just going extinct.
This extremely small bottleneck size extended over a long time is a common feature of all out-of-Africa models. It emerges from the mathematical genetics in a pretty straightforward way. Genetic variation will tend to persist through a bottleneck that lasts only a short time. It takes a bottleneck of around of the same order in generations as the effective size to really bring genetic variation down to where we see it in populations outside Africa. For an effective size of ~700, that means hundreds of generations, which is here estimated as ~350 or so, or 10,000 years.
You're entirely right in thinking that this is extreme. If you designed a biodome with this effective size and expected it to stay viable for 10,000 years, you'd be taking a long shot.
On the other hand, we are subject to a backward-looking bias. There may have been dozens or even hundreds of equivalently small groups or local populations. Many did become extinct. We've even found DNA from some of those. In the long run, one of them survived and was later remarkably successful. Its very growth was probably a big part of the failure of many of the others. Looking backward, we're tempted to see the 10,000 years of small population size as the story, but the real story was the interruption of gene flow among this network for a long time. That we don't have a great explanation for at the moment.
I think this is a very important point that deserves a separate post. Most people misunderstand the concept of a strong bottleneck. It feels "the whole humanity at some point was just 1000 individuals, how could they survive?". In fact at any given time moment the number of living individuals was much larger than bottleneck estimates. They were just scattered among many small groups that lost connection to each other. They went extinct, expanded, split, expanded again and eventually very few of them won out and expanded greatly, while others left zero or very small genetic footprint.
This post is such a good summary of the "wallasea problem".
Yet I am not convinced by the Mondal et al work.
First, even by stretching their model parameters to the max, they obtained only a small difference in coalescence times between PNG-Afr and Asi-Afr compared to real data, (compare fig 1 and 4 from their paper).
Second, surely there must be other Eurasian populations that could experience strong bottlenecks and long isolation. Yet the coalescence time between all Eurasians and Afr Yoruba is very consistent, and doesn't exceeds 75 my, that is likely happened after Toba.
I thing the right explanation is the most obvious one -that there was an early HS population in SEA that was mostly but not fully replaced by newcomers after 50 ky but left some genetic trace.
This is a great comment, my only thing to add is that the number of sampled populations of living people is still not really very large, and most of the Eurasian groups come from shared histories in the time frame we're looking at. I do wonder whether the "population Y" signature in South America may reflect a similar effect of demography, certainly it's another case of a long and extended series of founder effects.
Also, the dates in Mondal et al model don't look quite correct. The HS-Neanderthal split is now estimated closer to 800 ky, not less than 600 ky, and there is a deep history of interbreeding between them from around 300ky, not reflected in their model. The alleged "exit from Africa" at 62 ky is unlikely, since by that time there were already sister Y haplo groups F, C, D, E and only E is found in Africa, it is untenable that hugely successful F emerged in Africa and then went extinct there.
The big rise of Eurasian populations began around 55 ky, not in the Holocene as shown in the model, and by 45 ky the former likely already exceeded African population.
It is extremely likely that early HS and Denisovans were already present in SEA. The alleged early HS are always referred to as "extinct ex-OoA", but I prefer to call them middle-paleolithic people that were later mostly replaced by upper paleolithic people coming from West Eurasia.
These are great questions. I don't agree with every point but I do want to encourage more close examination of this kind.
Some small differences of opinion: There is not very clear "Middle Paleolithic" in SE Asia and in Indonesia in particular, at least not in the sense the term is usually defined for India and areas further to the west. The restructuring of technology for all ancient people who inhabited this region is a fascinating piece of the dynamics that I think current ideas are not capturing. If archaeology is relevant to understanding these dispersals at all (which I don't necessarily assume) then the lack of transfer of western technokits into this region is an important element to explain.
Regarding Y haplogroups, I probably have less trouble than many people in accepting a large and rapid divergence of Y haplogroups with near-extinction of haplogroups from closely-related populations. In my expectation, this is likely mediated by natural selection on the fully-linked NRCY. I think the same of most Late Pleistocene and Holocene connections based on Y haplogroups. This makes me different from many people who think about Y haplogroups a lot, I am aware. (But then, I am supervising a PhD on Y haplogroups and have published on them, so there's that!)
You are entirely correct that the timeline of Neanderthal-African divergence is a key element, although this is less critical for methods that bring in linkage like MSMC because the mutation rate difference is not as influential on the estimation of recent events as the recombination rate. Incorporating the deep time element into these models, such as the recurrent gene flow into Neanderthals from Africa, is a challenge considering the number of parameters that must come in for this. That may result in equifinality of models even for reasonable situations, and may not be solvable without ancient DNA more densely sampled than today.
I think you're well-reasoned in considering some African-origin people to have been in SE Asia and island SE Asia in the early Late Pleistocene, I don't think this is unlikely but I do encourage exploring alternatives including a more modern-looking Denisovans. We don't have great ways of testing this.
"In my expectation, this is likely mediated by natural selection on the fully-linked NRCY. I think the same of most Late Pleistocene and Holocene connections based on Y haplogroups. This makes me different from many people who think about Y haplogroups a lot, I am aware."
I've long had a suspicion that the ultimate breakthrough of OOA/proto-Eurasian population must have had some sort fertility-specific physiological component, not just social/cultural or even neurological. Some mutations which accumulated in the Y (or the X) that somehow influenced the fertility patterns of the OOA group - the rate of conception, improved spermatogenesis, sex ratio at birth, etc - that directly contributed to the massive population break-out after the bottleneck.
I would really like to see some researchers attempt to actually test this out by looking at all the defining mutations of our major Y clades - C, D, E, F, and also some of their associated major down-stream sub-clades (R1b, J1, E1b1, O1, etc) - and see if they can identify specific mutations which would indicate these clades were directly positively selected. It would be interesting to see the same thing on a broader scope looking at the CT macro-level too, and also A and B.
I agree that archaeology there is puzzling. For example, UP technology spread like fire very quickly around vast regions of Eurasia up to south Siberia around 45 ky ago, but in India it was gradual and in SEA not visible until recently. But genetic data suggests modern people were already there at least 40 my, and likely longer. Why it is not showing in stone tools is interesting.
When you say 800k, are you referencing this analysis from several years ago based on dental trait comparison? https://www.ucl.ac.uk/news/2019/may/neanderthals-and-modern-humans-diverged-least-800000-years-ago
Most analyses I've seen over the past few years do tend place the human/Neandersovan divergence in the 500-600k range https://www.sciencedirect.com/science/article/pii/S0092867423014034
Yes the dental data too, but there were also genetic studies I recall that put HS-N split close to 800 ky, significantly earlier than was thought previously.
John, is my interpretation of the RCCR method correct here?
Say you have two populations, popA and popB, with an RCCR value of 0.5 (on a scale from 0 to 1) at 50k ybp. Does that essentially mean that 50% of the unique drift that defines popA as popA distinct from popB (and vice versa) had accumulated by 50k ybp?
Looking at "Table 1 Best fitted parameter values of model A," they have the separation of African and OOA populations at 62k ybp, with a 10k year long bottleneck. The estimated effective population size of OOA during that time is only 726 - that seems like a very, very low population size to be able to ride out a 10k year bottleneck without just going extinct.
This extremely small bottleneck size extended over a long time is a common feature of all out-of-Africa models. It emerges from the mathematical genetics in a pretty straightforward way. Genetic variation will tend to persist through a bottleneck that lasts only a short time. It takes a bottleneck of around of the same order in generations as the effective size to really bring genetic variation down to where we see it in populations outside Africa. For an effective size of ~700, that means hundreds of generations, which is here estimated as ~350 or so, or 10,000 years.
You're entirely right in thinking that this is extreme. If you designed a biodome with this effective size and expected it to stay viable for 10,000 years, you'd be taking a long shot.
On the other hand, we are subject to a backward-looking bias. There may have been dozens or even hundreds of equivalently small groups or local populations. Many did become extinct. We've even found DNA from some of those. In the long run, one of them survived and was later remarkably successful. Its very growth was probably a big part of the failure of many of the others. Looking backward, we're tempted to see the 10,000 years of small population size as the story, but the real story was the interruption of gene flow among this network for a long time. That we don't have a great explanation for at the moment.
I think this is a very important point that deserves a separate post. Most people misunderstand the concept of a strong bottleneck. It feels "the whole humanity at some point was just 1000 individuals, how could they survive?". In fact at any given time moment the number of living individuals was much larger than bottleneck estimates. They were just scattered among many small groups that lost connection to each other. They went extinct, expanded, split, expanded again and eventually very few of them won out and expanded greatly, while others left zero or very small genetic footprint.
Noted. I agree this is a good topic for further writing as it is a common misunderstanding.