Thursday, August 30, 2012

High Resolution Denisovan Genetics

Several commentators have reviewed the last high resolution sequencing of the Denisovian genome, but John Hawks does has done a lot of research on the ancient Neanderthal genome has some of the most insightful comments.

There is Denisovan admixture in Sahul populations but not other Asians

First, he notes that the detailed findings confirm prior results on Densisovan admixture:

The mixture with a whole-genome sample from Papua New Guinea is estimated at 6% Denisovan ancestry. Confirming the later paper by Reich and colleagues, the new analysis finds no significant evidence of Denisovan ancestry in a mainland south Chinese (Han Dai) individual, and can exclude it down to a very small fraction:

However, in contrast to a recent study proposing more allele sharing between Denisova and populations from southern China, such as the Dai, than with populations from northern China, such as the Han, we find less Denisovan allele sharing with the Dai than with the Han (although non-significantly so, Z = –0.9). Further analysis shows that if Denisovans contributed any DNA to the Dai, it represents less than 0.1% of their genomes today.
That is a mystery to be explained. How did Asians end up lacking any evidence of Denisovan ancestry, when the peoples of Sahul (Australia and New Guinea) have six percent? . . . The early modern humans who were the ancestors of present Sahulian peoples surely came from Asia, and they surely mixed with Denisovans there somewhere, right? But today there's no sign that present Asian peoples descended from those early Asian peoples.

A Narrative The Could Explain The Geography Of Denisovan Admixture

My most recent analysis of this mystery was in a post from a week and a half ago. First, I argue that Denisovans are either Homo Erectus or an admixed derivative of them, and that Homo Flores, in turn, is a Homo Erectus subspecies that has experienced island dwarfism. I'll restate and expand upon one plausible narrative below.


To be clear, I'm not claiming that the evidence proves that this narrative is true. Instead, I want to suggest that this is a narrative that would consistently explain everything we know so far without making any assumptions more outlandish than necessary to fit the facts. It don't claim that there aren't only plausible narratives that could also be true, only that I can't come up with any more plausible narratives and neither has anyone else. This is a hypothesis, not a proof. I have also taken the indulgence of relying on memory rather than looking up the published articles that back some of my assertions on Southeast Asian population genetics that I read a couple of years ago or so, and hence not linking to them, in order to save time in the process of drafting an already long post.

An Admixture In Flores Only Scenario

In my previous post, I concluded my analysis with the following summary:

[I]t isn't inconceivable to me that almost all "Denisovian" admixture in modern humans is actually traceable to a dozen or two instance of admixture with members of the species Homo Flores on the island of Flores, and that a strong flight instinct of Homo Erectus population members who did not experience dwarfism and had someplace to which they could flee, together with a modern human perception of those who did not experience dwarfism as more primate than fellow hominin, could have meant that there was no meaningful amount of archaic hominin admixture anywhere else in Southeast Asia or East Asia (with a possible exception probably involving even fewer instances of admixture relative to the founding population in the Philippines).

Why Not A Dilution Hypothesis?

One of the key reasons to favor a Flores admixture location over a mainland Asian one, in the 20/20 hindsight of the known facts that I didn't mention in my previous post is that too much of the population genetic makeup of island Southest Asia is too old to be explained by a land bridge crossing along the then geographically united Sunda penninsula during the Last Glacial Maximum. Likewise it is not a good fit to a predominant population source during the Austronesian expansion that is the source for the languages of the region just a few thousand years ago.

Genetically, some island Southeast Asia looks a bit like some of the populations of the far Northeastern corner of Europe. They show some of the highest levels of pre-Neolithic ancestry on their respective continents, only island Southeast Asia didn't have to be entirely repopulated after the Last Glacial Maximum from scratch, so its Paleolithic ancestry component is much older. (See also this paper on indigenous Malaysian genetics, this paper on Eastern Indonesian genetics, this paper, Island Southeast Asian genetics, here, this paper on Indonesian Y-DNA, Indonesian autosomal genetics, here, here, here, here, here, here, here, here, here, and here.)

This, together with the very low percentages of Denisovan admixture in island Southeast Asians relative to the Sahul populations, makes the most sensible scenario to explain the low levels of Neanderthal admixture seen in Europeans, i.e. archaic ancestry dilution by post-admixture waves of migrants look much less convicincing in Asia than it does in Europe.

The numbers for a dilution hypothesis work in Europe and are supported by the recent evidence of the Otzi genome. But, to explain the results via dilution, you need the Sunda and South Chinese populations to have experienced sixty times as much dilution as the Sahul populations, yet to retain in island Southeast Asia a very large percentage of early Paleolithic population genetics that is quite distinct from that of East Asia. After controlling for inferred levels of Austronesian gene flow, there is still a huge disparity in Denisovan admixture and even less of an explanation for how one Paleolithic modern human population could have so totally replaced a prior one. The genes of island Southeast Asian populations don't show that there was near total population replacement when these islands were joined to the mainland as part of the Sunda penninsula around 20,000 years ago. In short, the numbers just don't work to explain the geography of a Denisovan admixture with dilution from subsequent migration waves alone.

Suggestive Evidence To Favor Denisovans Flight And/Or Slaughter

The new high resolution study's conclusion that the Denisovans may have had a quite low effective population size is congruent, however, with a "flee don't f--k" scenario outside Flores, all of the way to Siberia. I've hypothesized for some time, based on the inferior tool kit found in Asia prior to the arrival of modern humans, that the population density of archaic Asians may have been quite a bit lower than the population density of Neanderthals or modern humans hunter-gatherers in comparably favorable environments. This inference is also suggested on the ground that Homo Erectus was also probably less intelligent than either Neanderthals or modern humans were given their brain case sizes relative to body size.

A greatly outnumbered, technologically inferior, less intelligent population of Asian Denisovans (i.e. late Asian Homo Erectus) would probably be more inclined to flee in the face of newcomers when they were so clearly outmatched, rather than trying to stand their ground and co-exist as the Neanderthals, who were probably a more equal match for modern human newcomers, apparently did. Since the Denisovan's had a smaller population, they might have a greater capacity to flee as well, at least on the mainland, where they weren't cornered.

Of course, seeing some of their peers who didn't flee get slaughtered by incoming modern human tribes could easily have helped motivate the Denisovan decision to flee rather than fight or co-exist with modern humans. We can probably never know if how violent the first contact events were for sure. Given the record of megafauna extinctions likely caused by modern humans, the income modern human tribes would probably have been capable of carrying out a Denisovan genocide leaving only a tiny diasporan population in places like Siberia and Flores, whether or not they actually did so.

How Did Denisovans Hold On So Long Before Collapsing?

My best guess is that the main reason that intellectually and technologically outmatched Denisovans were able to hold out for so long despite potential competition from rival archaic and modern hominin populations is that the jungles of Southeast Asia and the mountains of Zomia were an effective barrier to the migration of populations of archaic hominins large enough to be threatening to small tribes of Denisovan hunter-gatherers.

The Toba Eruption Opens The Gates To Southeast Asia

This may have changed when ash fall from the Toba explosion ca. 75,000 temporarily killed of the jungle. The eruption would also have reduced the population size of the Denisovans in the area between South Asia and Southeast Asia who used to live there. Even if the Toba explosion by itself wasn't sufficient to wipe out a Denisovan population, if it had had the luxury of recooperating without competition from other slightly more fit hominins, it may have so depleted Denisovan populations in this region that they promptly went extinct in this range under the additional indignity of an inmigrating population of modern humans who outnumbered them and were even slightly better hunters in the suddenly unfamiliar volcano ash ravaged landscape than they were, particularly in light of how static their lithic culture was, which suggests that they may have not been very innovative or adaptable.

The ash fall area from the Toba eruption is much more consistent with an event that opens the door for migration from South Asia to Southeast Asia than it is with an event that drives the large Out of Africa migration in a single wave all the way to Asia (see also here). This region would have been much harder hit by the Toba eruption than the more distant territory in South Asia where we know that modern humans encountered Toba ash fall. So any barrier effect that the presence of existing Denisovan tribes presence may have posed would have also been eliminated at least temporarily.

Toba ash would have fallen upon and reduced the habitability of all of what is now Burma, Thailand, Malaysia, Cambodia, Laos and Vietnam, the island of Hainan in South China, much of the island of Borneo, all of the island of Sumatra in Indonesia, the island of Palawan in the Phillipines, and just a little bit of western Java. But, most of the island of Java, the Philippines (except the island of Palawan), and everything east of the Wallace line, would have had no direct effects from the Toba eruption.

The parts of island Southeast Asia and the Phillipines other than the island of Java and a few islands in the Phillipines that weren't mostly within the ashfall area of the Toba eruption is a good match for the areas where there appear to be traces of Denisovan admixture, while none of the areas within the ash fall region of the Toba eruption show any trace of Denisovan admixture.

Denisovan Cave itself, in Siberia, would have been connected to the part of East Asia where Toba's ash did not reach, and would have been within the northern reach of the archaeologically estimated range of Homo Erectus. So, modern humans populating the vacuum left by the Toba eruption may have encounted very few Denisovan's relative to their own numbers if this is when they made their entry into the Asian continent (a date consistent with reasonably calculated mutation rate ages in Asia) until they had made their way all the way from India to South China.

Furthermore, it appears that the Northern part of the Denisovan range in Siberia and maybe even Northern China, may have been entirely depopulated during the last glacial maximum, and much of the intermediate territory consists of highlands in Zomia, Tibet, and Mongolia (see the paleoclimate map here) where I am not aware of any evidence of Homo erectus archaeology. If the Denisovans like the Neanderthals didn't have the ability to make weather proof shelters of their own and craft warm clothing, these places may have been inaccessible to them. The only lowlands between the Denisovan Cave (which can be reached via mountain valley passes for the most part that are warmer than the Central Siberian steppe to the north, or through the Central Siberian Plateau during the summer), and the eastern fringe of the Toba ash fall area in mainland Asia, are in Eastern China, Korea and Manchuria. The modern human edge over the Denisovans in these areas which were not as tropical, may have been greater than it was in Southeast Asia and South China.

The archaeological evidence shows that modern humans had already left Africa at least 25,000 to 30,000 years before this eruption. And, we know that there were modern humans in South Asia who survived the Toba eruption who could have taken advantage of the new migration options the temporary destruction of Southeast Asian jungles by ash fall could have opened up. We also know from African population genetics, that a large tropical jungle such as the Congo could be just as formidable a barrier to the migration of sustainable paleolithic hunter-gatherers as an ocean or a high mountain range.

When the Toba eruption opened, it the gates of the region to modern human populations already resident in South Asia, and they surged into Southeast Asia. The Denisovans retreated before there was any meaningful opportunity for admixture all the way to Siberia. Island Southeast Asians had less of an opportunity to flee, but these quite small populations of non-dwarf late Homo Erectus may have seemed too threatening to the early modern human populations to be tolerated and may have instead have been simply wiped out (or at least were not integrated socially enough to give rise to admixture).

Denisovan Matings May Have Been Less Fruitful Than Neanderthal Matings Due To Greater Genetic Distance

Given that Denisovans were more genetically distant from modern humans than Neanderthals (probably more like 1,800,000 years of separation instead of 500,000 years of separation), it may also have been the case that the yield of fertile offspring from mating between Denisovans and modern humans may have been lower, and hence a longer period of co-existence may have been necessary for a comparable level of introgression into the modern human population to accumulate.

Prolonged Co-Existence On Flores Would Have Allowed Plenty Of Time For Introgression Even With Very Few Fertile Hybrids

But, this could have happened during the archaeologically long period of co-existence (tens of thousands of years) between modern humans and Homo Flores on the island of Flores which is geographically cheek by jowl. In this context, introgression could accumulate over a considerable amount of time (perhaps 25,000-30,000 years) even if the interspecies couplings weren't very fertile on average. If the modern human effective population size was small, it wouldn't take many successful interspecies hybrid children per generation on average in the 1,000 generations of co-existence between a hypothesized Toba explosion driven entry of modern humans in the region and the colonization of Sahul by modern humans.

Alternatively, perhaps an initial modern human population in Flores had an effective population size of only a dozen or two individuals (or less), which is quite possible if the gene pool of the Sahul migrants was subsequently expanded by a small number of Upper Paleolithic people as I suggest below, given current levels of genetic diversity of these populations. This would make sense in a scenario in which the original modern humans in Sahul were shipwrecked there and didn't have the maritime skills to intentionally navigate across the Wallace line or to Sahul until Upper Paleolithic people arrived. If this was true, and admixture took place only in the initial moment of the most dire population bottle neck, one could obtain all of the observed level of Denisovan admixture in the Sahul populations from just one or two pregnancies in the first generation. It should probably be possible to estimate the effective population size of the Denisovan population that contributed to the Sahul gene pool with enough high resolution genomes from that population to describe the diversity in the parts of their genome not found in other modern humans, and an analysis that showed that this effective population was just one or two individuals wouldn't surprise me, although it certainly isn't compelled by what we know so far.

In a slight variation of the shipwreck scenario that has the virtue of relying less on random chance, perhaps there was just one instance of Denisovan admixture (either just one fertile hybrid child, or one Denisovan-modern human mated couple with multiple children) on the island of Java. Perhaps due to hunting and gathering competition from modern humans, and perhaps due to warfare with modern humans as well in which the Denisovan's were outnumbered and outfought (and at hunter-gatherer population densities, the populations of both groups on the island of Javan would have been quite small - there were perhaps 45,000 Neanderthals by a census measure in all of Europe and Java was much smaller than Europe), the Densivoan of Java were pushed over into extinction soon thereafter. Perhaps the family of the hybrid child was persecuted by the other modern humans on the island. Perhaps the hybrid child and a small group of supporters made up of extended family and close friends were willing, in order to flee this persecution in a way that would leave it forever behind them attempt a risky one way trip across the straight and hence across Wallace line with inferior early Paleolithic rafts. In this scenario, the migrants would have had no interest in even trying to return to Java and hence no incentive to refine their successful boat design. This scenario would also produce the admixture ratios we see today in the Sahul population.

Expansion Into Sahul May Have Required An Upper Paleolithic Boost

The expansion from Flores or some other island on the east side of the Wallace line to the Sahul may have been faciliated by advantages associated with the arrival of a second wave of modern humans, who were integrated with first wave modern humans before expanding further to the Sahul. This second wave may have been expanding themselves due to the advantages, whatever they were, that drove the larger Upper Paloelithic revolution.

For example, in a scenario that also keeps the initial population size of modern humans in Flores low enough to make it sensible for them to treat Homo Flores on a more equal footing that their kin in Southeast Asia had previously, perhaps the first wave modern human residents of Flores arrived only by maritime misadventure and didn't have the boating technology to return back across the Wallace line to their fellow modern humans. This group of shipwrecked men and women wouldn't have been nearly as impressive and intimidating to the indigeneous archaic hominins as the intentionally migrating populations who crossed the less daunting waters on the other side of the Wallace line, and as dwarves, wouldn't have been seemed like as much of a threat to the shipwrecked men and women.

These first wave modern humans may not have had the capacity to intentionally navigate across the Wallace line or beyond to Sahul in large enough numbers to establish a sustainable population. This capacity may have arrived only much later when a small population of much more competent mariners at the most distant fringe of Upper Paleolithic revolution population expansion, informed by the technological advances of the Upper Paleolithic revolution, arrived and were integrated into the existing population in Flores. By the time the wave of newcomers appeared on the scene, the first wave modern human population may have already received considerable Homo Flores introgression into the small modern human population of Flores over a thousand generations of co-existence on that island that was carried on by the migrants to the Sahul.

This two wave migration theory would explain another of the long standing mysteries of Paleolithic migration in Asia. There is pretty good archaeological evidence pointing to the time that Sahul was first colonized by modern humans about 45,000 years ago. The genetic evidence indicates that the founding population of Sahul was quite small but was phylogenetically quite diverse and also that the Papuan and Australian proto-populations underwent a schism almost immediately upon arrival, with both branches having almost all parts of this phylogenetic diversity. Yet, there were modern humans in South Asia at the time of the Toba eruption and there were modern humans in mainland Southeast Asia at least 63,000 years ago.

Why did it take so much longer for them to reach the Sahul than the rest of Asia?

Modern humans made it from Beringia to the far reaches of South America in just a few thousand years and the distance from mainland Southeast Asia to Papua New Guinea and Australia is much shorter. A scenario with a first wave of modern humans who managed to cross the Wallace line only by accident and a second wave of Upper Paleolithic modern humans who could do it on purpose a thousand generations later, would explain this gap. One of the things that may have made the Upper Paleolithic era revolutionary may have been a revolution in boating technology.

The measured levels Neanderthal admixture in Asians relative to Europeans contradicts prior research.

Second, Hawks notes that their finding on the relative levels of Neanderthal admixture differ from prior research:

To me at the moment, this is the most interesting paragraph of the new paper:
Interestingly, we find that Denisovans share more alleles with the three populations from eastern Asia and South America (Dai, Han, and Karitiana) than with the two European populations (French and Sardinian) (Z = 5.3). However, this does not appear to be due to Denisovan gene flow into the ancestors of present-day Asians, since the excess archaic material is more closely related to Neandertals than to Denisovans. We estimate that the proportion of Neandertal ancestry in Europe is 24% lower than in eastern Asia and South America (95% C.I. 12–36%). One possible explanation is that there were at least two independent Neandertal gene flow events into modern humans. An alternative explanation is a single Neandertal gene flow event followed by dilution of the Neandertal proportion in the ancestors of Europeans due to later migration out of Africa. However, this would require about 24% of the present-day European gene pool to be derived from African migrations subsequent to the Neandertal admixture.

This is a very interesting result, partially because it is the opposite of what we are finding. As I explained earlier this year, we are finding Europeans to share more Neandertal alleles than Asians do. The difference in our results has been much smaller than 24%; really only an increase of less than 0.5% on the whole genome, or maybe 10% relative to the overall amount in Europe (which is on the order of 3%).

Hawks considers some possible sources of the discrepency. Given the flat contradiction between the studies using methods whose differences are subtle, it is hard to know the cause of the discrepency. On balance, my intuition is to give greater credit to the findings of Hawks group, because his group studied much larger sample sizes and their is some meaningful variation in Neanderthal admixture levels in Eurasians in the 1000 Genomes sample. The large discrepency may simply be a matter of random chance whose statistical significance (Z=5.3) is overestimated because the data are more structured than the statistical assumptions from the new study, using samples from just three European and two Asian populations assumes.

The question of whether there was one round of Neanderthal admixture, or two or more with some regional differentiation that just happened to produce roughly similar levels of Neanderthal admixture since the processes were similar and played out in similar ways, is beyond the scope of this post and complicated by the discrepencies in the reported empirical results.

Denisovan Effective Population Size Was Small And Fell When Modern Humans Appeared

Third, he notes that there are indications in the genomic evidence of population decline in Denisovans and Neanderthals to a low effective population size coinciding temporally more or less with the arrival of modern humans:

It has become possible to make some good estimates of demographic history using only a single diploid genome, using a technique developed by Li and Durbin. Meyer and colleagues applied this technique to the Denisova genome, finding that its genetic history contrasts with that of living human populations:

To estimate how Denisovan and modern human population sizes have changed over time we applied a Markovian coalescent model to all genomes analyzed. This shows that present-day human genomes share similar population size changes, in particular a more than two-fold increase in size before 125,000–250,000 years ago (depending on the mutation rates assumed). Denisovans, in contrast, show a drastic decline in size at the time when the modern human population began to expand.

There is not yet enough data from Neandertal genomes to apply the same method, but to the extent that we understand their diversity, they show a similar picture. These archaic humans in Eurasia had much, much smaller effective population sizes than the ancient population of Africa.

Of course, we know for a fact that the Denisovan are no more, that they co-existed at least briefly with modern humans, and that modern human population size has grown over time. Any model that indicated something different when applied to the data would have to have been wrong, and the confidence intervals for the dates provided by the model are wide enough to fit all of the plausible scenarios. There is also considerable mischief that goes into interpreting what effective population size means in the context of a very late relict population of a species on the verge of extinction that Hawks explores a little.

But, as I've noted above, there are plausible reasons to think that Asian Homo Erectus might have had a fairly small effective population size that fell when modern humans arrived and that the Denisovan reflects Asian Homo Erectus ancestry.


terryt said...

"The new high resolution study's conclusion that the Denisovans may have had a quite low effective population size is congruent, however, with a 'flee don't f--k' scenario outside Flores, all of the way to Siberia"

But the paper suggests a high level of inbreeding in the Altai region. This is inconsistent with any real connection with SE Asian H. erectus.

"That is a mystery to be explained. How did Asians end up lacking any evidence of Denisovan ancestry, when the peoples of Sahul (Australia and New Guinea) have six percent? . . . The early modern humans who were the ancestors of present Sahulian peoples surely came from Asia, and they surely mixed with Denisovans there somewhere, right?"

I have always maintained that Y-DNA C and mt-DNA N entered East Asia via a northern route. Such a route would easily explain the 'mystery'.

"But today there's no sign that present Asian peoples descended from those early Asian peoples".

Again I have an explanation. I have also long suggested that somewhere in East Asia was a third ancient Homo species that has contributed genes to modern humans. These people were perhaps not exposed to the Denisova people before they began their own expansion, basically once the Upper Paleolithic had been introduced to northern China".

"But, to explain the results via dilution, you need the Sunda and South Chinese populations to have experienced sixty times as much dilution as the Sahul populations"

That is quite possibly the case though. All it requires is to drop the idea that Y-DNA O is an ancient member of the SE Asian population. Personally I am much more inclined to the idea that O's expansion was responsible for carrying the Upper Paleolithic technology south through China from its origins in the Far East (northwest China).

"The genes of island Southeast Asian populations don't show that there was near total population replacement when these islands were joined to the mainland as part of the Sunda penninsula around 20,000 years ago".

Again, the genetic evidence does so very much if you are prepared to accept that O is post Upper Paleolithic, and perhaps as recent as Early Neolithic, in SE Asia.

"We estimate that the proportion of Neandertal ancestry in Europe is 24% lower than in eastern Asia and South America (95% C.I. 12–36%)".

The fact that there is ANY Neanderthal ancestry in eastern Asia argues strongly for an early human movement east through Central Asia. Therefore such movement must have been possible at times despite what Maju, for example, likes to believe.

terryt said...

I've just had a look at John Hawks' blog.

"Maybe Denisovans didn't live in South or Southeast Asia at all. If not, that demands that we explain how Australians got their genes".

I think he is correct here. I have just explained 'how Australians got their genes'.

"We must, I think, conclude that there was at least one, and possibly several episodes of massive population movement across South and Southeast Asia"

Again I think I have established that is so. Especially if we concede that Y-DNA O and associated mt-DNA haplogroups did not originate in SE Asia but further north.

andrew said...

"at hunter-gatherer population densities, the populations of both groups on the island of Javan would have been quite small - there were perhaps 45,000 Neanderthals by a census measure in all of Europe and Java was much smaller than Europe"

Javan has an area of about 50,000 square miles. Europe has an area of about 4,000,000 square miles, but perhaps half of that territory would have been unavailable to Neaderthals because it was too far North.

So, back of napkin, Javan could support about 1,125 Neanderthals at a European Neanderthal population density, and by hypothesis, quite a bit few Homo Erectus, but perhaps somewhat more modern humans.

andrew said...

I totally agree that the last place that East and SE Asians could have picked up their "Neanderthal element" in in South Asia and probably something like Pakistan and perhaps North India based on archaeological evidence of the extent of Mousterian technology.

I also know that the island is called Java, even though my fingers apparently don't.

The point is using Neanderthal population density figures (which honestly may be a bit on the high side) is simply to estimate archaic hominin population density in Java. Now, one doesn't actually need particularly large Denisovan population density to get the observed level of Denisovan admixture, so long as the admixture takes place with a very small modern human population and that population is the only one that makes it across the Wallace line. It is also useful simply to have a sense of the hard numbers we are talking about to conceptualize the situation better.