Linear Population Model is a no comments blog for the understandable reason that it takes too much time to moderate blog comments, so rather than posting comments there, I'll engage with some of the concerns raised by Marnie here at this blog.
Marine opens up the issue with this premise:
You may know that the Eurogenes blog (along with some very prominent researchers) has been heavily and stubbornly promoting the theory of a very sudden invasion of Europe during the Copper Age from either the Ponto-Caspian Steppe or the Central Asian Steppe.After setting up the position expressed by Davidiski at the Eurogenes blog fairly accurately and discussing some other relevant data, Marnie gets to the bottom line issue he is raising:
So what's with thinking that R1 lineages were simply confined to the Central Eurasian steppe for 20,000 years (and not to Europe until the Copper age)?The most important data supporting the Eurogenes position come from several data sets of Central and Eastern European ancient DNA (e.g. Germany and Hungary).
There definitely is a shift of modern Europeans between the Mesolithic and today. I agree with that.
But I don't agree that the shift happened due only to replacement from the Ponto-Caspian or Central Eurasian Steppe and only during the Copper Age.
It's just as likely that the genetic shift of Europeans is due to Western Europeans fusing with populations of Finland, Scandinavia, the Baltic, Balkans (including Greece), Central Europe, the Ukraine, Russia, Anatolia, and the Levant, starting in the Mesolithic and continuing to the present day. It would have been a complex process, with waves of people possibly moving both in and out of the Steppe and in and out of Europe, since the R/Q/P split.
Ancient DNA samples from this region cluster into three groups.
1. Samples that precede the Neolithic revolution (the Neolithic revolution begins when the first farmers appears in this part of Europe; this initial Neolithic archaeological culture is often called the "Linear Pottery Culture" or LBK for short) including Upper Paleolithic and Mesolithic samples and also samples from European hunter-gatherer populations in the region (as determined based upon the archaeological context of the ancient DNA) who were contemporaneous with the first farmers of the region.
These samples are predominantly mtDNA U4 and U5, and the handful of autosomal samples that are available are very distinct from those of subsequent farming populations which derive only part of their ancestry from European hunter-gather like sources. There are very few pre-Neolithic European samples of ancient Y-DNA available, but all that are available are Y-DNA I2.
While there is a decent case that Southern Europe saw some enrichment in its mtDNA diversity in the Mesolithic era, I don't think that the case is nearly as strong for what became LBK cultural zone in the early Neolithic era.
2. Samples from farmers of the early Neolithic era (i.e. from the beginning of the LBK culture until roughly the start of the Corded Ware culture in the Copper Age).
These samples have a minor component of mtDNA haplogroups similar to those found in the Mesolithic era and among hunter-gatherers contemporaneous with them, but with many new mtDNA haplogroups that aren't found in older samples. The dominant Y-DNA haplogroup is G2.
3. Samples from the Copper Age (roughly coincident with the arrival of the Corded Ware culture) to the present in this region.
From the Copper Age onward, the ancient DNA strongly resembles the modern population genetics of the region. It is highly enriched in mtDNA haplogroup H relative to the early Neolithic where it was uncommon, or the Mesolithic era where mtDNA H was absent.
The frequency of Y-DNA G2 is dramatically reduced and Y-DNA I2 is also quite uncommon, while, Y-DNA R1a (and really a very specific subset of Y-DNA R1a which a most recent common ancestor estimated to date from the Copper Age by admittedly imperfect mutation rate dating methods) is predominant. In some regions, there is a blend of Y-DNA R1b (and again actually a specific European subset of R1b with similar estimate time depth), the predominant Y-DNA type of much of Western Europe and areas near the Southern Baltic Coast of Europe, near the biogeographic boundary between R1a predominance and R1b predominance that roughly corresponds to the historical ranges of the Bell Beaker people (or cultures in continuity with them) in the West, and with the Corded Ware people (or cultures in continuity with them) in the East, during the Copper and Bronze Ages in Continental Europe.
The limited ancient autosomal DNA fits this clustering with a shift from the LBK DNA profile towards the modern profile represented by a new ancestry component that is detected, appearing in ancient DNA starting around the Copper Age.
There is also Bronze Age ancient DNA and physical anthropology data showing genetic continuity and continuity in physical appearance and bone structure of people from the Corded Ware region of Europe all of the way to the Tarim Basin at the fringe of the East Asian highlands of greater China, where mummies reveal individuals who were Eastern European in their genes, coloration and bone structure, who we know spoke a long lost Indo-European language called Tocharian until around the time of an expansion of linguistically Altaic peoples across the Eurasian steppe all of the way to Europe itself.
Evidence from the time depth of language families in Europe also tends to support this analysis.
In the interests of not letting this post get too stale, I am posting it now. But, I hope to enrich it with additional links and references over the next few days.
UPDATE: A new ancient DNA find reported at Eurogenes appears to further support this hypothesis.
25 comments:
For what it's worth: I agree completely with your analysis.
You may want to correct this: "Mesolithic samples (i.e. the first farmers in this part of Europe whose initial archaeological culture is often called the "Linear Pottery Culture" or LBK for short)". I'm aware that you know it's wrong, because it contradicts what you write immediately afterwards (and also because I know you know better).
"From the Copper Age onward, the ancient DNA strongly resembles the modern population genetics of the region.".
Isn't it from the Late Bronze Age onwards? We know nothing (AFAIK) from the Middle Bronze Age and Early Bronze Age samples like those from Unetice (some would place them still in the Late Chalcolithic but whatever) are still way off modern genetic pools, as are those from the Late Chalcolithic (or "Copper Age").
My impression is that in the Chalcolithic-Bronze transition the modern genetic pools were not fully formed in Germany (actually in the North-Central region of Germany, where more of the samples come from). This suggests a further process of blending among the various pools, which show some notable differences (some are high in mtDNA H, others lacking H, etc.)
Another issue is how much of what we know for those regions is extrapolable to the rest of Europe. For example I'd expect the genetic pool of Portugal to have changed only with the Celtic invasion of c. 700 BCE and not before (but lack data to confirm). Of many other regions we can say nearly nothing, including France, Britain, all Southern Spain, Italy, the Balcans...
Anyhow, I think that Marnie's criticism is well placed, at least in general terms. It's well known (at least to me) that Davidski has a bias on this matter and that he claims things that are at the very least hard to prove and that often clash with reputable studies on the matter as the recent one by Underhill et al., by which most of the R1a expansion does not seem at all related to Indoeuropeans (Central-East Europe may be the exception but not the rule).
I addressed your concern. The parenthetical was intended to modify the word Neolithic earlier in the sentence but I can see how it is unclear and substantially rewrote it.
I would expect the gene pool of Portugal to change a lot ca. 2900-2500 BCE (a big double digit percentage shift, although not as big as elsewhere due to incorporation of indigeneous populations in the expanding culture) and then change further only modestly ca. 700 BCE with the Celts (ca. 8% contribution).
There would certainly be some regional variation in amount and timing, but the broad outlines would be pretty similar up to a shift in regional timing roughly proportionate to effective travel distance suitably defined.
I think that the claim that significant R1 expansion (R1a or R1b) is taking place in Europe in the Mesolithic (aka Epipaleolithic) is not well supported at all by the available ancient DNA and reasonable inference from it, even though it might be plausible in the absence of ancient DNA data to contradict it. The Davidiski narrative fits with multiple lines of evidence (linguistic, modern population genetic, ancient DNA, archaeological) while there really isn't a comparable narrative to explain an earlier expansion that is consistent with the evidence.
"I would expect the gene pool of Portugal to change a lot ca. 2900-2500 BCE"
On what grounds? Do you spouse the "colonial" hypothesis for the rise of civilization in Iberia maybe?
I beg you to segregate your reasoning of R1a (recent Holocene expansion) and R1b (with a clearly older timeline and different centers and patterns). Making a parallel between R1a and R1b really don't makes any sense whatsoever: they are different animals. Anyhow the direct evidence we have for pre-Metal Y-DNA is yet very small and mostly restricted to Northern Europe (and if something we know is that Northern or Central European results can't be extrapolated to other regions).
Certainly, R1a and R1b expansions are different processes by different populations, although they are not that greatly separated in time and seem to be driven by quite similar technological advantages for the expanding populations that may even have a common source.
There was urban civilization in Iberia pre-Bell Beaker, but the technologies that appeared around then converted this civilization from a localized Southwestern Iberian phenomena to a more dynamic one that profoundly influenced a much larger region. I believe that these technologies arose because people from somewhere well to the east o Iberia brought R1b and the knowledge with them, mostly likely from somewhere between Georgia and Crete.
There may have been significant integration by this migrants into established Iberian civilization, both culturally (e.g. adopting some preexisting megalithic religious concepts) and in terms of migrants marrying local women. The extent of this is harder to measure, because mtDNA H that characterizes and is parallel to the R1b expansion was more widespread albeit at lower frequencies in earlier periods and because I don't have enough of a firm command of sub-phylogenies of mtDNA H (and many studies don't report sufficiently detailed subtyping) to get a clear enough picture. But, I have little doubt that the outsiders were the superstrate culture.
The pre-metal Y-DNA data points aren't really numerous, but there are enough Western European early Neolithic results that are uniform enough and similar enough to samples elsewhere (all Y-DNA G2a dominated), and enough Mesolithic/early Neolithic boundary hunter-gatherer Y-DNA samples (all I2 dominated), to make this scenario a pretty strongly favored hypothesis so far.
The strong correlation between early Neolithic autosomal DNA from all sources, and from Sardinia (which is not rich in R1b compared to Western Europe) is also quite powerful, and boosts the power of low sample sizes in a way that one can't for mtDNA to the same degree, although mtDNA X2 frequencies in Western Europe also support my hypothesis.
"R1a and R1b expansions are different processes by different populations, although they are not that greatly separated in time and seem to be driven by quite similar technological advantages for the expanding populations that may even have a common source".
What's your evidence for all this? I see absolutely nothing to back these conjectures.
Very especially the best scholarly age estimate to date for R1 is Underhill's and it implies that the expansion of R1a and R1b (surely several branches and several processes) are very much different and, above all, that R1b is much much older than R1a.
"There was urban civilization in Iberia pre-Bell Beaker, but the technologies that appeared around then converted this civilization from a localized Southwestern Iberian phenomena to a more dynamic one that profoundly influenced a much larger region".
Again I see nothing to back this. Megalithism is 2500 years older in parts of Iberia than Bell Beaker and in most of Europe it is some 1500 years older as well. The influence that can be tracked to Iberia (or wherever else in Atlantic Europe) is much older than Bell Beaker. Not just that, the key "technologies" that we find in BB such as copper metallurgy, architecture and archery are attested in SW Europe (and other places) long before BB. BB is a dynamic and colorful trade network or something like that but it's not that innovative in terms technological AFAIK but before BB there was Artenac and before Artenac there was already Zambujal and Los Millares, all of which continue in existence with BB. BB is so much hyped!
And you know why is that? Because earlier, when it was still possible that BB originated in Central Europe, it could arguably back an older IE expansion and that appealed (even beyond all reason) to certain people, who put feelings and prejudices before logic and evidence. Now the aura persists, even if BB can't anymore justify such ideas.
"The pre-metal Y-DNA data points aren't really numerous, but there are enough"...
They are clearly not enough at all. There's not a single sample from a Gokhem-like site: all are from Stuttgart-like sites (or in some cases older but only one West of the Rhine, which is not even I2 but something very odd: C1).
"Marnie at the Linear Population Model blog takes issue with the widely held view that Y-DNA R1a arrived in Europe as a result of a copper age expansion of Indo-Europeans into the regions of Europe where it is found today."
1. What I said, based on evaluation of correlation in admixture components, is that the ANE component is not widely correlated with low level components associated with an archaic admixture process.
2. It is this archaic admixture process, and the low level admixture components associated with it, that primarily are the cause of the strong "eastward shift" on PCA plots of Europeans (with the exception of Basques and Sardinians.)
3. I then said that the archaic admixture components (that cause the eastward shift) are not widely correlated with the R1* ydna haplogroup.
4. There does appear to be a weak correlation between the eastward shifting archaic admixture components and the ydna R1a haplogroup.
5. There does appear to be a weak correlation between the eastward shifting archaic admixture components and the ydna N haplogroup.
5. There does *not* appear to be a correlation between the eastward shifting archaic admixture and the ydna R1b haplogroup.
That is the point of my article, which, for someone who claims to have a background in physics, shouldn't be lost on you.
@Marnie: by "archaic eastward shift" you mean ANE and Eastern Asian genetics, i.e. an "archaic" (ancient pre-UP for the largest part) within Eurasian H. sapiens, right? Otherwise (if you mean for example Neanderthal or Denisovan admixture) I fail to make sense of your words.
@Maju
First of all, ANE is a component that comes up in the software package Admixture.
Depending on which populations you choose for your Admixture run, components associated with Northern Europeans will vary.
So the definition of ANE is rather tenuous.
In the Lazarides paper, "Ancient North Eurasian" is labeled on the PCA plot (Figure 1B).
http://biorxiv.org/content/biorxiv/early/2013/12/23/001552.full.pdf
Closest cousins to the labeled ANE point on the plot are the MA1 and AG2 ancient DNA samples, and then modern groups like Finns, Mordovians, Russians and Ukrainians.
On an Admixture run, (for instance Rasmussen et al, 2014)
http://www.linearpopulationmodel.blogspot.com/2015/01/ancient-north-eurasians-north-south.html
Mordvinians, Russians, and Ukrainians show two major components, for K=10, 11. There are also minor components.
On the Rasmussen run in this paper, the major components are shown in dark blue and green.
Note that Orcadians and French (far left) also have these two components in almost the same combination as Ukrainians, Russians and Mordovinians.
The main difference is that Ukrainians, Russians and Mordovinians have low level components associated with groups in Northeast Asia. (for example Chukchi's, Yakuts).
The "ANE" component that Davidski is referring to appears to be the green component on the Rasmussen et al. admixture run.
Note that the Mordovinians, Russians and Ukrainians pull toward MA1 and the Ancient North Eurasians on the Lazarides PCA.
The French much less so.
So it cannot be argued that the "ANE" component is uniformly associated with the shift toward the MA1 ancient DNA on PCA plots.
It has to be something else that is causing the "shift" and that something else is low level admixture of these Northeast Asian components.
Again, at some point, this low level components became incorporated in Mordovinians, Russians and Ukrainians (and also Maris, Finns and the Chuvash).
However, this low level is not correlated with R1b dominant populations.
Like I said, it does appear to be loosely associated with R1a and N ydna haplogroup populations.
In short, Maju, regarding your comment "archaic eastward shift" you mean ANE and Eastern Asian genetics":
The "shift" is from Northeast Asian gene flow, which is only loosely associated with some, but not all, populations that have the green component shown on the Rasmussen et al. admixture run.
ANE es Mal'ta 1, no un componente zombie. The definition is not at all "tenuous" but just the generalizing extrapolation of affinity with Ma1 (call it "paleo-Siberian", I like it better in fact). And that is particularly clear in Lazaridis' paper.
I believe that Davidski sticks to this definition but prove me wrong. It is not the Indo-Afgan "green" component in the Rasmunssen graph you mention, not at all. The modern population closest to ANE are Kets, although these are seldom used in such analyses (not sure the reason), then come Native Americans and only then the tenuously admixed populations of Eastern Europe (and maybe other parts of Eurasia).
"... it cannot be argued that the "ANE" component is uniformly associated with the shift toward the MA1 ancient DNA on PCA plots".
There seems to be some sort of misunderstanding here (in your interpretation or in Davidski's usage, unsure) but in any case I would indeed appreciate if ANE/Ma1 is replaced in the near future by a more realistic proxy of the origin of that admixture: Yamna.
This is why we are all eagerly awaiting the Reich/Patterson new paper.
"The "shift" is from Northeast Asian gene flow"...
Not so simple: analysis of Siberian genetics show that ANE (Ma1, AG) is "intermediate" between Europe/West Asia and Native Americans, that it has no relation with core East Asia and that Siberians themselves form clines between these three polarities.
→ http://forwhattheywereweare.blogspot.com/2013/12/the-malta-adna-findings.html
You're missing this issue: there's nothing that we can clearly call "NE Asian": it's very blurry and clinally diverse.
From the European point of view, there are two Siberian polarities: one tending to Ma1 and therefore to Kets and Native Americans and the other (more restricted to some specific populations like Uralics) tending to core East Asians (Chinese, Koreans, etc.) These are two different influences that only very slightly overlap.
Erratum: "ANE es Mal'ta 1, no un componente zombie" should be in English: "ANE is Mal'ta 1, not a zombie "component"".
I need some coffee, heh!
@Maju
"ANE es Mal'ta 1, no un componente zombie. The definition is not at all "tenuous" but just the generalizing extrapolation of affinity with Ma1 (call it "paleo-Siberian", I like it better in fact). And that is particularly clear in Lazaridis' paper."
It is not clearly defined in the Lazarides paper (2013). You really can't tell from the PCA what "ANE" is specifically associated with.
As you know, MA1 had R1* type ydna.
However, this does not mean that all R1* downstream shares this "Ancient North Eurasian" affinity.
Regarding which component "ANE" is connected to, and Davidski's and Lazarides' analysis, you're right. It is not clear what they mean by "ANE".
Instead of posting PCA plots, I'd like to see an Admixture run like the one in the Rasmussen et al paper.
That would clear up a lot of the confusion.
ANE is described clearly in Lazaridis et al. as Mal'ta-1, exactly the same as EEF is Stuttgart and WHG is Lochsbour. This may be a pro or a con but there is no doubt: all analysis are done with these three ancient samples as references and very particularly the so-called "fateful triangle".
"As you know, MA1 had R1* type ydna".
Yes but 2/3 of West and South Eurasia also have some sort of R1 or related lineages (R2, Q, P1*) so I don't think it's very important. Some people are emphasizing, even truly hyping, this detail but I tend to totally ignore it, more so when the most ANE populations worldwide today are not R1 carriers, but Q1 carriers instead, and when I postulate that these lineages spread around all West Eurasia (including Central Asia and beyond into NE Siberia, etc.) within the Early Upper Paleolithic. It's just normal, fully within my expectations, that Mal'ta carried P-something and the only thing that informs us of is that my model of P expansion is very plausible, as R1 was already formed c. 24 Ka ago (and probably some 10 Ka earlier as well, or even more).
@Maju
"ANE is described clearly in Lazaridis et al. as Mal'ta-1"
OK. If that's the definition you want to use, here is a link to a Figure from the Kostenki 14 paper, which shows an Admixture run with K14, MA1, Loschbour, Gok2, Iceman and Stuttgart.
http://dienekes.blogspot.com/2014/11/genome-of-kostenki-14-upper-paleolithic.html
MA1 has five major components:
Hunter-Gatherer (about 25%)
America (about 15%)
South Asia (about 15%)
Siberia (about 5%)
Central Asia (about 20%)
K14:
Hunter-Gatherer (about 30%)
America (< 2%)
East Asia (< 2%)
South Asia (about 20%)
Siberia (< 3%)
Central Asia (about 10%)
Middle East (about 10%)
Africa (about 5%)
Locshbour:
Hunter-Gatherer (about 95%)
Gok2:
Middle East (about 35%)
Hunter-Gatherer (about 55%)
In Europeans today, there are no Europeans look anything like MA1 on an Admixture run. The "South Asian", "American" and "Siberian" components appear at the less than about 2% in all Europeans, with the exception of a few populations like Finns, Mordovians, some Russians and the Chuvash. In those cases, the "South Asian", "American" and "Siberian" components are less than about 5%.
So, at most, an MA1 like population could not have contributed to Western Europeans at more than about the 4% level.
"all analysis are done with these three ancient samples as references and very particularly the so-called "fateful triangle". "
So, Davidski should put up an Admixture run with the "Fateful triangle" populations. That will clarify how much an ANE type population (as defined as being MA1 like) could have contributed to the populations of Europe.
I'm quite confident that it will be a lot less than he is implying.
"Yes but 2/3 of West and South Eurasia also have some sort of R1 or related lineages (R2, Q, P1*) so I don't think it's very important."
Well, it's important if you are trying to argue for a very specific timeframe for the contribution of this "ANE" population to Europeans.
R1 and R2 split a long time ago. Before MA1.
Davidski and David Reich are trying to use MA1 to infer a very specific timeframe for the contribution of an MA1 like population to Europeans.
First, the contribution of an MA1 like population is much less than they suggest.
Second, you're right to suggest that there were probably R* and Q carrying population in Central Asia, West Asia, India, Siberia and even Eastern Europe at the time of MA1. These population probably carried different variations of the admixture components we see in K14 and MA1.
So, a lot of autosomal variation for R* and Q carriers.
That's my central point.
We can't infer that all R* carriers had autosomal components that all looked exactly like MA1.
So too much is being inferred from this assumed "Ancient North Eurasian" population.
Marnie: Ma1 and K14 are (because of their age) around half way the divergence between West and East Eurasians. Hence I'd expect them to be about just 2/3 like West Eurasians, the rest being reflection of their not-so-distant relation with the common Eurasian roots.
And if you look at supp. fig. 20 in that paper you see something like that at K=3-5, which are the depths showing the continental scale differences. At K=5 for example K14 appears as 2/3 West Eurasian (bingo!) and 1/3 other affiliations, mostly Australasian (Papuan). Ma1 is rather like 1/2 West Eurasian, 1/4 Native American and 1/4 Australasian, showing almost no other affiliations at all.
The Papuan affiliation can be explained maybe by the origin of Y-DNA K and mtDNA N in SE Asia (to be researched) and the NA affiliation makes total sense because their Altai UP ancestors were at the origin of the Q1 migration eastwards some millennia earlier.
You can butcher the components further but doesn't give us any more information in fact. Maybe you can speculate about that African extra in K14 being "Basal Eurasian" or whatever but little more.
You can't also reverse-engineer the ADMIXTURE analysis as you try to. It won't work because the results are sample-dependent and ancient samples are not causing them but rather "suffering" them.
You can indeed run a "supervised" Admixture run with some of those ancient samples as the fixed references and then you'd get the results you want. Something like that but even more refined was done by Lazaridis et al. It's all over the paper (and supp. materials, which you can't afford not to read if you want to be critical) but a simple example can be seen in this graph (notice that PC2 is not at all as large as PC1 but I don't know the exact correction to be made so left it as found).
"So, Davidski should put up an Admixture run with the "Fateful triangle" populations."
That's already done with more refined means by Lazaridis et al. (again in the supp. materials). Feel free to ask him anyhow. ANE ranges specifically from 5% in Sardinians and 10% in Basques to 18% in Estonians and 17% in Scots.
"Well, it's important if you are trying to argue for a very specific timeframe for the contribution of this "ANE" population to Europeans."
ONLY if you happen to believe like Davidski that somehow Ma1 and R1 are casually related, not if you believe as I do that it's mere coincidence and even maybe a reflection of R1 or R1b expansion to Europe (let's not forget that Ma1 is Gravettian, just like Crô-Magnon 1 and other of the earliest European remains).
...
...
"R1 and R2 split a long time ago. Before MA1".
Before Ma1 necessarily, yes. So what? It's not like it is ancestral to R1a or R1b, it's just another random R1-nothing of the kind you find in Pakistan and nearby areas. It's a symptom, not a cause of anything. A symptom of the overall westward expansion of P1 from ~ Bengal, a symptom of the human wave from Sundaland that eventually triggered the Upper Paleolithic in West or Central Asia and so dramatically replaced Neanderthals in their territory.
"Davidski and David Reich are trying to use MA1 to infer a very specific timeframe for the contribution of an MA1 like population to Europeans."
Well, that can be inferred from the level of affinity (%ANE) in ancient samples, so it's a legitimate goal and one that should clarify the Indoeuropean intrusion in most of Europe, where it's more and more obviously that is exotic.
"First, the contribution of an MA1 like population is much less than they suggest."
Possibly but WHY? I find extremely difficult to properly estimate the levels of ANE ex-novo penetration without enough ancient samples. Which is the baseline reference? Basques, Sardinians, Stuttgart, Bologna or Motala? Very likely it's different in each region, so I request more more ancient samples from before and after IE intrusion in each region.
"We can't infer that all R* carriers had autosomal components that all looked exactly like MA1."
Not at all: that's like sampling some random guy from Dublin and concluding that Chadic peoples and Uyghurs all look like the Irish just because they also have R1b. That R1 from Altai is no doubt just one among many that already prowled the western half of Eurasia at that same time, surely including some R1b even (although not yet R1a - but certainly its remote ancestor).
"So too much is being inferred from this assumed "Ancient North Eurasian" population."
Davidski is no doubt and some of his followers maybe but that's far from being a generalization.
@Maju
Returning to the central point, I look forward to seeing an admixture run from Davidski and David Reich showing K14, MA1, and any ancient DNA they are using to make inferences, along with modern populations.
Without that, we can't even get to square one with inferences about movements of MA1 like populations into Europe. Not in the Upper Paleolithic. Not in the Mesolithic and not in the Neolithic.
MA-1 did not have R1*, he had R*, on another branch from R1'2 even. Probably his patrilineage split from that of surviving R more than 5 thousand years earlier.
Our really old aDNA necessarily comes from cold places like Siberia, so we are also in the dark about where else these components were present - Siberia could be quite peripheral.
Given its present distribution ANE presumably *was* widespread through Siberia, but R1 need not have been until very recently.
@capra internetensis
"MA-1 did not have R1*, he had R*, on another branch from R1'2 even. Probably his patrilineage split from that of surviving R more than 5 thousand years earlier. "
from this paper:
http://mbe.oxfordjournals.org/content/early/2014/12/13/molbev.msu327.full.pdf+html
"Structure within the West Eurasian hg R: The TMRCA of hg R is 19ka, and within both hgs R1a and R1b comprise young star like expansions"
see the tree in the supplementary data.
From the TMRCA and the tree, the R1a/b split can be framed, timewise, between 19ka and about 15ka.
Later that 15ka is unlikely for the R1a/b split (if the tree in the supplementary data is correct.)
"Given its present distribution ANE presumably *was* widespread through Siberia, but R1 need not have been until very recently."
I doubt that the R1/R2 split occurred very recently.
Anyway, it's pointless to debate it. I'm sure better dates for the R1/R2 and R1a/R1b splits will be out soon.
My main point, made in previous comments in this thread, is that it is very sloppy to be inferring the R1* hg autosomal signature from the Mal'ta autosomal signature.
You can't infer what R1 hg rich, R1b hg rich, or R1a rich populations looked like autosomally from the Mal'ta sample.
Also, these populations probably evolved over time.
Currently, there are no R hg ancient DNA samples after Mal'ta (until the Iron Age).
Very sloppy to be making grand inferences from such a sparse and uncertain set of data.
The fact that this kind of garbage continues to be published in journals such as Nature speaks to fact that Nature is turning into a pseudo-scientific rag.
@Capra: Good observation. I have (for the first time) carefully read SI5 in Raghavan 2013 and they do seem to even place it at a pre-R now extinct branch. There's no such thing as R1-2 today: that's R; so if Ma1's lineage does not reach at the split point the modernly defined R node (even if it is close to it), it should be described as P1* or, if you wish, pre-R (as it is clearly in the branch leading to R and not to Q or other P1* branches).
However its own private drift (see fig SI 5a) suggests that R and R1 were already developed.
Re. R1b I'm comparing that figure with the partly similar one of Underhill 2014 (fig. 5) and I'm not sure what exactly is called R1b in Raghavan's figure. It's certainly not R1b-M343 but either R1b-141 or (most likely) R1b-M412, which are further derived subsets. I'd need to check the references used (all NA-something but one labeled "reference") but I will here assume that it's R1b-412, which is the main European branch.
So my impression with this comparison is that Ma1's chronology (24 Ka BP) is 2/3 towards present between the R and R1b-M412 nodes. So R1b-M343 was already formed back then (this is true even if I replace M412 by S141, the difference is not game-changing). R1b (as such: M343) therefore MUST be some 30-40 Ka old (very roughly). However the observable European expansion is more recent (maybe 15 Ka BP).
Take your time and refine these hurried estimates but what is apparent from those graphs is that, chronologically speaking, Ma1 lived in a time when R1b-M343 had already coalesced. The formation of its main sublineages R1b1c-V88 and R1b1a-L320 is probable also but less clear (for lack of data). He also lived in a time prior to the R1b-M412 expansion in Europe - which seems to be of Magdalenian or (less likely) later chronology (but always quite older than the R1a one).
"I doubt that the R1/R2 split occurred very recently."
Certainly not: in Raghavan's fig. SI 5a is totally clear that it is much older than the lifetime of Ma1. However we don't know if R2 as we know it today (M479) had coalesced yet (it's a poorly researched haplogroup).
R1 at the very least had coalesced already as well, meaning that the R1a/R1b split had also happened too.
I have just argued that R1b must also have coalesced by that time, however this was not yet the derived branches we see today in Europe apparently. These lineages formed later and I'm chewing on the implications, because if you think that Neolithic is too recent a date, then one must consider Magdalenian being original from West Asia, and that is difficult to argue for (not totally impossible however).
"Currently, there are no R hg ancient DNA samples after Mal'ta (until the Iron Age)."
There is one but doesn't help much either: recently it was announced informally (at a commentary in another publication, unpublished data) that some Afanasevo burials have R1a(xZ93), what the commenter argued it shows a European origin of this culture. The study will be published soon, I understand.
It does give a minimum age for the R1a node of at least 6000 years (surely more) but still the irregularity of full sequence drift-parameter at modern branches implies that the "molecular clock" is hard to apply at such nearby chronologies, even with full chromosome data, still leaving a very narrow window for the possibility of European R1b being of Neolithic origin. The window is becoming extremely narrow, so, together with other evidence (geostructure of European R1b fits better with Late UP expansion, greater WHG parameter in Gokhem), I lean towards the Magdalenian/Epipaleolithic chronology.
But we need more data to be safe, really.
PS- What is clear is that European R1b has nothing to do with IE (Kurgan) expansion. It's necessarily older than that (and shows a negative cline towards the IE homeland, where it basically does not exist at all nor has any diversity worth mentioning). The lack of R1b findings among early Neolithic sequences make it look rather pre-Neolithic as well (re-expanding surely within Megalithism from Atlantic "refugia").
@Maju
I can't agree strong enough with the statement that "European R1b has nothing to do with IE (Kurgan) expansion" and also
"The window is becoming extremely narrow, so, together with other evidence (geostructure of European R1b fits better with Late UP expansion".
Thanks for sharing your ideas.
Post a Comment