March 14, 2012

Junk scholastic paper promotes Neolithic replacement doctrine

This argument on whether Europeans descend mostly from Paleolithic aborigines or Neolithic immigrants is becoming ideological and going beyond the scientific issue of fact-checking and having different opinions. Actually facts are being selected and manipulated and interpretations are becoming truly incompatible.

What for me means that modern Central Europeans are clearly NOT direct descendants of Neolithic settlers, who had much greater abundance of rare lineages like N1a, means for Qiaomei Fu that they were indeed the ancestors.


Qiaomei Fu has recycled the Neolithic, Epipaleolithic and Forest Neolithic (said to be foragers but not really true) aDNA from Haak 2005 and 2010, Sampietro 2007, Bramanti 2009, Krause 2010, Deguilloux 2011, Lacan 2011 and Gamba 2011. See Jean Manco's Ancient Eurasian DNA page for more details.

From the text, it appears that they are using here mostly Bramanti's 2009 sample, with an extra sample from Krause (Kostenki's U2). This may be justified up to a point because Bramanti has been one of the few to test for coding region markers, so most of the other samples haplogroup assignations are questionable (but they are often not U, which is clearly defined in the HVS-I mutational set). However the choice of discarding Italian data from Caramelli 2005, which clearly demonstrated R0(xH) in one sample and N* in another, is not justifiable on any grounds.

It is not justifiable either that no space is dedicated to discuss the sample, why did they chose these or those and why did they ignore the others. In such non-existent space they could have discussed the 19 known true Paleolithic European samples, not considered by them, that are definitely not U, even if we can't know for sure (with some exceptions) what exactly they were.

In that space they could have also discussed why they chose to consider Forest Neolithic peoples from the Baltic as "hunter-gatherers" or how, excluding them their sample is reduced to n=11, of which all are U. Adding to these two more from Chandler 2005, the overall European true-Paleolithic mtDNA count is, if my maths are correct, n=32, of which 13 are U variants (six U5, four U4, two U* and one U2). As mentioned above 19 are something else than U and in some cases I am quite certain that it is H.

The 'forager' sample: Red: true Paleolithic, Green: Forest Neolithic, Lemon: Forest Chalcolithic

Another thing they have not considered is excluding repeated sequences located in the same necropolis, often the same tomb, surely meaning close relatives, which therefore should count as just one. They could also ponder that saying U as such is almost meaningless and that its subclades should be considered instead.

Not really as important as the huge blanks left by the survey and the significance of 19 true Paleolithic sequences which are not U (60%), most of which exist in the huge and most important areas left blank in this sample: West Europe, Italy, Balcans...

That alone should be enough to disqualify this paper as ideological junk. 

But there is more: the Neolithic European samples are not considered in their local context, the Catalan samples are not compared with modern Catalans (bad fit) nor the German samples with modern Germans (another bad fit) but they are magically and conveniently pooled to make a pan-European set which more or less seems to fit with modern Europeans after some suspicious filtering they claim makes an unbiased dataset.

I have already wasted more than enough time in this piece of crap and really gets me angry: an academically qualified scholar like Fu, much more his mentor and co-author Svante Päbo should know better than pulling this ideological piece of propaganda.

Enough said unless the comments section demands something more.

See also: European ancient mtDNA maps page at this blog.

20 comments:

  1. The "hunter-gatherers" seem to carry some neolithic lineages, like T, K, and J: were they already mixed?

    Also: how can we be sure that these hunter-gatherers weren't descended from another ancient expansion from the middle east, or alternatively, that the neolithics were the ancient hunter-gatherers?

    And I also agree they forgot to sample most of Europe. So, the paper it's not conclusive as it pretends to be, I understand: if we suspect there was some H in southern Europe and North Africa in the Mesolithic, these scientits should've paid more attention to these regions.

    ReplyDelete
  2. "The "hunter-gatherers" seem to carry some neolithic lineages, like T, K, and J: were they already mixed?"

    Which sample are you talking about? Ostorf?

    "Also: how can we be sure that these hunter-gatherers weren't descended from another ancient expansion from the middle east, or alternatively, that the neolithics were the ancient hunter-gatherers?"

    Don't ask me! I'm demolishing their paper not defending it. Fu's email is somewhere on the paper: ask him (her?)

    "And I also agree they forgot to sample most of Europe. So, the paper it's not conclusive as it pretends to be, I understand: if we suspect there was some H in southern Europe and North Africa in the Mesolithic, these scientits should've paid more attention to these regions".

    In absolutely full agreement. It's a repetition of the argument that we already saw for example at Dienekes' years ago, when the Bramanti and Haaks papers were published.

    ReplyDelete
  3. Which sample are you talking about? Ostorf?

    Figure 1 of the article. The cake that has N=23 hunter-gatherers. They're 83% U and the rest are these three lineages, mostly considered to be of Neolithic origin.
    Also, not all U's are the same: we know this clade is about 60.000 years old, if there are different clades of U, not just U5, this means they probably were already mixed from various population of West-Central Eurasian origin.

    ReplyDelete
  4. They are all form Ostorf (Funnelbeaker 'Forest Neolithic' from Mecklemburg).

    "Also, not all U's are the same: we know this clade is about 60.000 years old, if there are different clades of U, not just U5"...

    I also mentioned that: U5, U4, U* and U2 are not closely related: they must have diverged at least 30 Ka ago (Kostenki 1) or even earlier.

    ReplyDelete
  5. Sorry to say that this is another "garbage in, garbage out" paper.

    Could have been so much better, since there is so much information on H subgroups and diversity that could have been compared with outside of the region. What a waste of effort.

    ReplyDelete
  6. I was just suggested this other paper: Deguilloux 2012 (free access). It reaches to the same conclusions but it is infinitely more elegant, detailed, cautious and amiable to read. It still ignores the data from Chandler 2005 but admits one Paleolithic ("Mesolithic") H from Villabruna (which I cannot confirm but as R*, as it was only sequenced the HVS-I).

    Another paper (this one tentatively favoring adaptive selection in mtDNA) I have been suggested today is Pickard 2008 (part of a book apparently, available at academia.edu, you need an account to read but you can join as "independent researcher"). The most interesting aspect here is that it does indeed cite Chandler 2005 in regards to the transition in Portugal.

    ReplyDelete
  7. I don't agree that this is a "junk" paper. The main point of the paper is to show that you must augment mtDNA from living people with ancient DNA if you are going to get a sense of the rate of growth of populations.

    It's true that further study needs to be done on other mtDNA types such as T and HV, but the methods used in the paper were limited by the availabilty of ancient DNA samples.

    It's true that further sampling of other regions and mtDNA types is certainly needed to understand the growth of hunter-gatherers, pasturalists and farmers. Still that doesn't detract from the ultimate finding of the paper regarding the relative growth rate of hunter-gatherers and farmers.

    You can see my further comments at my blog:

    http://www.linearpopulationmodel.blogspot.com/2012/03/finding-echo-of-hunter-gatherer-refuges.html

    ReplyDelete
  8. I do not understand why you say that: my whole argument is that the lack of data and the extrapolation of a handful of samples localized in time and space is not meaningful, specially if you disregard the details (the devil is in the details and you can't see the full picture if you do not dedicate enough time to the details).

    I have just been suggested as complementary reading (together with the two papers in the previous comment) yet another very similar recent study by Marie Lacan (direct download PDF). This one is somewhat more detailed, including a sample of her own (Chalcolithic France) and suggests that maybe there was a stabilization in the Chalcolithic.

    Still a "cursory read" by myself finds that the variation of H (from c. 20% to more than 40%) is not explained this way and that some of the detected variation may be related to the differences in the two samples, with the Neolithic sample being 77% Central European and the Chalcolithic sample 76% SW European (mostly French).

    Still, repeating from my email, I understand that it looks this way (Neolithic-Chalcolithic-Modern, I will ignore Paleolithic/Epipaleolithic until a better sample exists, which at least covers Europe better and does not include Forest Neolithic outliers from the Baltic):

    Stable clades: U (U4, U5 and K), HV(xH), J (mostly J1) and I. (Maybe also X).

    Decreasing clades: N (N1a), which collapses to near zero, and T (mostly T2).

    Expanding clade: H.

    I think that this does not make much sense: I can accept that the collapse of N1a and decrease T may correspond to localized flows and certain bias of the dataset but the doubling of mtDNA H after Neolithic and, it seems now, Chalcolithic is most anomalous.

    Of all the ancient samples ever reported only a few have more than 40% H:

    1. Sunghir Gravettian: almost certain H17'27 (n=2 but same sequence)
    2. Taforalt Oranian: majority reported H (based on HVS-I) but hard to confirm even one
    3. Vilabruna Late UP 1/1 (HVS-I - doubtful)
    4. Portugal Epipaleolithic: H reported in 5/9 samples based on HVS-I, two of which are quite clearly H1b.
    5. Portugal Neolithic: H reported in the vast majority of 23 samples (half of which is unmistakably so)
    6. Austrian LBK (one of Bramanti's samples, tested for AluI)
    7. Tyrol Neolithic 1/2 (HVS-I - doubtful).
    8. Many Chalcolithic Iberian and Basque samples (not too clear)

    Notice that doubtful is not negative, just not sufficiently proven.

    Is it possible that there was an expansion from Iberia in the Chalcolithic through the Atlantic primarily? H1 has a very Atlantic distribution as does H3 in a more reduced SW zone and certain subclades of R1b-South, specifically the "Irish" clade.

    Notice that this would not mean that H or R1b are that recent but that they experienced re-expansions as part of the expansion of a population in which they were dominant. Based on Chandler 2005 and what we know of Iberian and Basque ancient DNA in general, that re-expansion would have been from Iberia probably, and would be closely related to Megalithism.

    I would suggest that the civilization of South Portugal, specially the culture of Vila Nova de Sao Pedro, centered around the still poorly studied Castro do Zambujal, was central: first with Megalithism and then with both Megalithism and Bell Beaker, of which it was a major center, specially around 2000 BCE.

    This is becoming a serious possibility.

    ReplyDelete
  9. PS- Marnie: the surname is Fu, not Qiaomei. How do I know? Because it says cite as: Fu, Q... blah-blah.

    It also bugged me because I know that East Asian surnames typically go first, however accommodation to West European standards in publications in English has become common and similarly Spanish authors, who officially style two surnames, now often publish with both surnames tied with a "-".

    ReplyDelete
  10. OK. Thanks, for spotting this. I changed it.

    ReplyDelete
  11. Just read carefully through both Ricaut et al and Deguilloux et al. I don't see anything that refutes the Fu hypothesis. Fu focuses on the Megalithic to Neolithic transition. The paper deliberately does not use mtDNA haplogroups for which there are not statistically significant ancient DNA sample sizes. Thus it focuses on U and H.

    Ricaut et al. attempts to use small sample sizes, but admits that there are difficulties with this approach: "The small size of the Mesolithic sample (n = 15) greatly reduces the statistical power of this analysis."

    Both Ricaut et al and Deguilloux et al mention the importance of the Neolithic to Chalcolithic transition. It is true that the Fu paper does not specifically address this transition. However, Ricaut does mention that most of mtDNA were established in Europe before the Chalcolithic.

    I don't see any fundamental disagreements between these three papers. They are complementary in their conclusions.

    ReplyDelete
  12. I did not say that any of those papers refute Fu. Actually they are all quite similar but less crappy (they admit limitations here and there and produce some more detail and even some alternative speculations).

    The main exception is the Prickard paper which is indeed more critical overall of the whole model but suggests that the changes are due to biological selection, what is most unlikely.

    "The small size of the Mesolithic sample (n = 15) greatly reduces the statistical power of this analysis."

    Exactly. And never mind that the sample is extremely biased towards one single region: Central Europe.

    "It is true that the Fu paper does not specifically address this transition".

    Another sign of the weak understanding that this paper exudes.

    ReplyDelete
  13. Actually another big problem I see with Fu is that he claims that the Neolithic mtDNA pool is modern, even if it has less than half than modern amounts of H.

    The same may be said of Ricaut (wrongly mentioned above as Lacan, who is just a co-author) but in regard to Chalcolithic: while Ricaut sees that Neolithic mtDNA is NOT modern, he fails to see the same problem in his France-biased Chalcolithic mtDNA sample, when nothing fundamental really changes in his samples this way.

    You can't claim that an mtDNA pool with c. 20% H is modern European, when this one has >40% H. Either something has changed A LOT since then or the sample is accidentally biased in favor of exotic clades.

    This last may be another explanation: IF different ethnic groups lived then side by side (but effectively segregated) with different funerary practices (burial vs. leaving the corpse to the vultures, for example, or burial in monument/village vs. burial with yew sapling, which is perishable and would be almost impossible to locate) we would only see the genetics of some, and not the others.

    The weak point of this hypothesis is that the lack of a modern mtDNA pool persists through millennia and it's difficult to imagine that this segregation would be so persistent in time.

    So the segregation must have been through more defined barriers, as long geographic distances. I.e. the source of the modernization of the mtDNA pool are a distinct population of some sort.

    There's where I would place the Portuguese... or something of the like.

    ReplyDelete
  14. I'll focus on one point.

    You say "Actually another big problem I see with Fu is that he claims that the Neolithic mtDNA pool is modern, even if it has less than half than modern amounts of H."

    Fu et al paper:
    The unbiased European percentage of mtDNA H is 41%. For Croatia, the percentage of H is 49%.

    Ricaut et al:
    0% Mesolithic, 23% Neolithic, 18% Chalcolithic, 42% Modern

    Eupedia mtDNA haplogroup frequencies:
    http://www.eupedia.com/europe/european_mtdna_haplogroups_frequency.shtml

    I don't see any discrepancy here between Fu et al., Ricaut et al., and other data for the percentage of H in modern European populations.

    ReplyDelete
  15. The Prickard paper simply makes an argument for natural selection, noting the disappearance of N1a. It also points out that the study of J and T haplogroups are important in the study of the Neolithic expansion. This paper is quite conjectural.

    Again, I don't see that their points refute the work of Fu.

    There are statistical limitations of working with small sample sizes. Researchers use varying approaches to draw conclusions when confronted with small sample sizes. That's the difference between Fu and Ricaut.

    Only more ancient mtDNA samples will shield further light on this difficult area of research. I'm glad that the Max Planck Institute is on it.

    ReplyDelete
  16. The discrepancy is between Neolithic (or Chalcolithic in Ricaut's case) and modern. 18-23% is the half of 42%.

    This means that if there was admixture from some other source at 50-50 apportions, then the other source had to have at least 60% mtDNA H, if the admixture was at 25-75 apportions, then the immigrant population (25%) would have to be at c. 100% mtDNA H. So we must accept a major immigration in the surveyed areas after Neolithic, although the source is not clear at all.

    It is the lack of enough H what really bugs me.

    However in the Chandler 2005 'non-paper', Portuguese Neolithic peoples had approx. 18/23 H lineages (at least 9 unmistakable H variants and then other 9 R0-CRS, what is almost for sure H1 or H2a). That means 78% H, what allows for them to be a source. And Portugal is one of the few areas of Europe where you can imagine people out-migrating in the Chalcolithic, through the Megalithic and later also the Bell Beaker cultural phenomenons.

    ReplyDelete
  17. These questions will have to be addressed with further ancient samples, as Ricaut et al. so elegantly state.

    The Prickard paper mentions Portugal. You also. I wouldn't expect Portugal to be representative of the larger process of LBK Neolithicization that took place. I haven't read closely enough to understand the specifics of the Neolithic in Portugal, but it is very likely that the process there is different from that along rivers like the Danube and its tributaries.

    Thanks for the links to the Ricaut, Prickard and Deguilloux papers.

    ReplyDelete
  18. Maybe but there's precisely where Fu et al. fuck it up, because they want to answer without sufficient data - and instead of admitting that we have many more questions than answers, they happily claim Neolithic continuity. That's why Fu's paper is junk.

    "I haven't read closely enough to understand the specifics of the Neolithic in Portugal, but it is very likely that the process there is different from that along rivers like the Danube and its tributaries".

    Indeed. Portugal (or more precisely SW Iberia, as it may exclude North Portugal and include some border areas from Spain) was essentially the farthest west participant of the Cardium Pottery Neolithic. Soon after that it was the first place on Earth to begin burying people in dolmens (trilithons), which are the architectural and conceptual landmark of the Megalithic phenomenon (to be more precise I usually speak of "Dolmenic Megalithism", so we can ignore other constructions like stone rings, etc., which may have some peculiarities).

    Only a millennium later this phenomenon arrived to Brittany and soon after spread through all Atlantic Europe and some inland or Mediterranean areas. These include: South, West and North Iberia, most of France, Ireland, Wales, SW England, Scotland, Switzerland, South and NW Germany (two different cultures), Denmark and Scania, parts of the Low Countries... That is almost all Western Europe.

    Later, after the advance of Indoeuropeans/Kurgan Peoples in Central and Northern Europe (which totally displaced Megalithism for their single burial customs, usually in kurgan), Megalithism expanded instead to the Western Mediterranean (South Italy and North Africa specially).

    Within this process, not far from Lisbon, it arose what is now known as Castro do Zambujal, a massive city (?) at the center of many other towns of the same culture, which is the Western counterpart of Los Millares and later El Argar. This now almost forgotten civilization was later also central in at least some stages of the Bell Beaker phenomenon, which again covers all the area of our interest.

    At that time Iberia was importing from the Baltic (amber) to North Africa (ivory) and Atlantic sea routes must have been most important.

    It has little to do with LBK but it has a lot to do with Chalcolithic Europe as a whole, first with the Dolmenic Megalithism (phenomenon and then the Bell Beaker, in their cultural, religious and trading dimensions but probably also political aspects as well (hard to say: they do not seem to have written anything).

    ReplyDelete
  19. Maju, it's too harsh to use f*d up as a descriptor. Let's just say that the paper focuses on H and U and unfortunately uses only one randomly sampled modern site in Croatia to check their European wide sampling.

    You're the expert on Megalithism and clearly have read a lot about it. Are there sites in these Megalithic areas where large samples of ancient DNA are available? Similar to the LBK sites sampled in Haak?

    ReplyDelete
  20. There should be: a characteristic of Dolmenic Megalthism is "collective" (serial) burial in dolmens, usually in extended position, what IMO is a trait of Paleolithic continuity, the same we have in Dniepr-Don and its Baltic 'forest neolithic' offshoots (Pitted Ware and such), which also retained the use of ochre on corpses. Instead Balcanic, LBK and many Cardium Pottery burials used flexed (fetal) burial position, which began in West Asia. While this is only one trait it seems to me a major one, because burial traditions are hard to change, specially in traditional societies, while pottery or even the economy are more easily changed.

    So there should be plenty of remains, however I know that in Portugal, critically acidity of the soils is a major problem for conservation of the skeletons. Not that this was a problem for Chandler et al. 2005, who obviously could access some 32 individuals from the Epipaleolithic (9) and Neolithic (23), as I have mentioned once and again.

    While the experts proceed to test for coding region mutations, the results already say enough via HVS-I: of the 23 Neolithic samples, 12 are clear like water and they are: 9 H and 3 U5 (75% H), and the rest is 9 R0-CRS (most probably H1 or H2a), 1 R* and 1 HV0 (surely V). If we use common sense extrapolation here, 78% of the Neolithic Portuguese mtDNA pool was H. But even discarding all even slightly unclear cases, 75%, what is a a lot (and a lot more than they have today, what may have been caused by immigration since the Bronze Age).

    "Similar to the LBK sites sampled in Haak?"

    Haak only tested for HVS-I: his samples are at least as useless as most others. It was Bramanti 2009 who tested for coding region sites, clearifying the picture and surprising all: the supporters of Paleolithic Continuity because in Central Europe at least there was, it seems, a discontinuity upon arrival of Neolithic and the supporters of Neolithic replacement because the Neolithic replacement was almost as different from modern mtDNA pools as the Paleolithic one. So they must have been replaced as well, it seems.

    It is this replacement what bugs me. I'm absolutely unsatisfied by the claims of Fu of continuity since Neolithic or Ricaut of continuity since Chalcolithic either: both are actually showing that in those times, in the sampled areas at least, there was a clear low level of H. This one must have come from somewhere else and that's why I'm suggesting Portugal and Megalithism as origin and vehicle.

    ReplyDelete

Please, be reasonably respectful when making comments. I do not tolerate in particular sexism, racism nor homophobia. Personal attacks, manipulation and trolling are also very much unwelcome here.The author reserves the right to delete any abusive comment.

Preliminary comment moderation is... ON (your comment may take some time, maybe days or weeks to appear).