January 26, 2014

Brown-skinned, blue-eyed, Y-haplogroup C-bearing European hunter-gatherer from Spain (Olalde et al. 2014)

There is nothing like a little ancient DNA weirdness to start off 2014, which promises to be as exciting as 2013 was.

The new study La Brana 1 identifies it as ancestral in the SLC24A5 locus in which virtually all Europeans are derived. This comes in the heels of the Loschbour preprint which identified that sample from Luxembourg as also being ancestral. Taken together, it's now clear that hunter-gatherers from Mesolithic Western Europe were brown.

Curiously, it now seems that both Europe and India were (in part) inhabited by brown people and became lighter by a process of admixture + selection. The process went "all the way" in Europe, but a cline of pigmentation was sustained in India.

The other finding (not mentioned in the abstract) is that La Brana 1 belonged to Y-haplogroup C6! This is a low-frequency European clade of haplogroup C. So now, we have evidence that haplogroup C is not eastern Eurasian (as the presence of its subclades in Australia, India, East Asia, and the Americas might suggest), but a pan-Eurasian entity. It remains to be seen whether this C-in-Europe can be pushed further back in time, but finding it in Mesolithic Iberia reduces the chance that it's some random eastern Eurasian who made it to the outskirts of Europe recently.

Finally, La Brana 1 has derived alleles at loci associated with pathogen resistance. This might be important, because a common hypothesis is that Europeans developed this type of resistance during the Neolithic as they started interacting with the pathogens of domesticated species and started living in less-hygienic higher-density settlements.


Nature (2014) doi:10.1038/nature12960

Derived immune and ancestral pigmentation alleles in a 7,000-year-old Mesolithic European

Iñigo Olalde et al.

Ancient genomic sequences have started to reveal the origin and the demographic impact of farmers from the Neolithic period spreading into Europe1, 2, 3. The adoption of farming, stock breeding and sedentary societies during the Neolithic may have resulted in adaptive changes in genes associated with immunity and diet4. However, the limited data available from earlier hunter-gatherers preclude an understanding of the selective processes associated with this crucial transition to agriculture in recent human evolution. Here we sequence an approximately 7,000-year-old Mesolithic skeleton discovered at the La Braña-Arintero site in León, Spain, to retrieve a complete pre-agricultural European human genome. Analysis of this genome in the context of other ancient samples suggests the existence of a common ancient genomic signature across western and central Eurasia from the Upper Paleolithic to the Mesolithic. The La Braña individual carries ancestral alleles in several skin pigmentation genes, suggesting that the light skin of modern Europeans was not yet ubiquitous in Mesolithic times. Moreover, we provide evidence that a significant number of derived, putatively adaptive variants associated with pathogen resistance in modern Europeans were already present in this hunter-gatherer.

Link

338 comments:

«Oldest   ‹Older   201 – 338 of 338
Hamar Fox said...

Annie Mouse,

East Asian admixture. Which we know happened, and is visible in the physical features of some Western Europeans.

Huh? The only Western Europeans with any degree of East Asian admixture are Scandinavians, and, within Western Europe, only Scandinavian populations can produce individuals with unambiguous East Asian traits with any frequency. East Asian admixture is of course the reason why Finns and Russians are closer to East Asians than Mal'ta is in this paper's f3 statistics, but neither Finns nor Russians are Western European.

German Dziebel,

If Amerindians were derived from MA-1, we wouldn't have seen any pull toward Amerindians in Western Europe.

You never gave me your source, so I can only assume by Western European you mean LB1, and are referring to the PCA included in this paper. First of all, LB1 isn't pulled toward Amerindians but toward East Eurasians in general. Second, modern Europeans are 7000 years of drift further from East Eurasians than is LB1. f3 statistics employ an outgroup, which controls for drift, hence f3 statistics show that LB1 is within the European range, being only slightly more shifted toward Amerindians than unadmixed modern Europeans, and less shifted toward Amerindians and East Eurasians in general than recently admixed Europeans (Finnish and Russian). This makes perfect sense, since modern Europeans have Neolithic admixture that LB1 largely or wholly lacked, and Neolithic migrants were genetically more distant from Eastern Eurasians than were European and Siberian hunter gatherers (Lazaridis et al.)

If this isn't what you were referring to, then I'd definitely need to see your source. Since all life on Earth is related, and me, you, my cat, your goldfish, that greenfly you swatted the other day are all varying degrees of distant cousin, differences and 'pulls' between them can only be relative, so I ask relative to which population(s) are Western Europeans pulled toward Amerindians but not toward East Asians?

I'll elaborate on what I said here too:

So even if you think this is because Amerindians were the initial centre of it all and their progeny split into W. and E. Eurasians, which developed their own way, essentially erasing detectable bonds with the other Eurasian half of the picture, any subsequent admixture from Amerindians into Mal'ta should not be part of that process and hence the global relationships of the most recent admixture from America shouldn't be disguised in MA-1.

What this means is that, if W. and E. Eurasians stem from a greater genetic pool, i.e. that of Americans, then the unique modern characteristics of both of these groups are determined by inheritance of a subset of Amerindian diversity (selected down or lost through bottle-necking) and unique evolution. This would create W. and E. Eurasian signatures unique from Amerindians. But this must be a one-time process. The chances of later populations from America dovetailing unnoticeably into W. or E. Eurasians would be impossible, because the same essentially random selection of Amerindian traits is unlikely, and unique evolution is impossible to replicate. So, in short, additional migrations of Amerindian-rooted populations into MA-1, for instance, would introduce a different signal into that population than the signal that resulted from the first out-of-America wave/waves. So the only way geography would be relevant to MA-1's levels of Amerindian admixture relative to other W. Eurasians is if he did indeed receive additional waves of admixture. But he has no East Asian 'admixture', even though these later waves would be statistically very likely to carry the genes, ultimately present in Amerindians, that East Asians inherited from the American genetic pool but W. Eurasians did not.




terryt said...

"if W. and E. Eurasians stem from a greater genetic pool, i.e. that of Americans, then the unique modern characteristics of both of these groups are determined by inheritance of a subset of Amerindian diversity (selected down or lost through bottle-necking) and unique evolution ... because the same essentially random selection of Amerindian traits is unlikely, and unique evolution is impossible to replicate".

That logic is not going to work with German. He appears to believe that every time a part of a population moves away it carries an exact sample of all the genes present in the source population. He does not accept bottlenecks ever happen.

"Amerindian admixture in Europe has been ascertained in a number of studies".

Not so. All that has been confirmed is that Europeans and Amerindians share a genetic element from a common source. One is not 'descended from' the other.

"We do know from Rasmussen et al. that Anzick is closer to MA-1 than modern Amerindians".

Almost certainly because the 'modern Amerindians' include northern populations that have been later admixed with further levels of East Asian genetic elements.

"It proves that LB and Amerindians both have a similar amount of Mal'ta affinity."

But that doesn't at all mean that Amerindians and LB have exactly the same genes from that ancestral population.

"So effectively MA-1 and LB both have an Amerindian component or, let's be honest with ourselves, Amerindian ancestry".

The 'Amerindian ancestry' bit doesn't automatically follow at all from the first part. All we can say is that members of the three groups share genetic elements but none is 'descended' (and certainly not totally) from any other.

Hamar Fox said...

Oops, two corrections:

East Asian admixture is of course the reason why Finns and Russians are closer to East Asians than Mal'ta is in this paper's f3 statistics

Here I obviously meant La Brana and not Mal'ta.

And here, a correctly place 'are' should clear up the ambiguity:

being only slightly more shifted toward Amerindians than unadmixed modern Europeans are

Tobus said...

@German:
No, it's not. EDF 5a,b are consistent with the inference of shared ancestry between MA-1, LB and Amerindians.
Amerindian admixture in Europe has been ascertained in a number of studies....LB is very clearly admixed population.

You seem to be shifting the goalposts a little bit - your original position was that LB has increased Amerindian affinity relative to Europeans, but you now seem to be saying that LB's "Amerindian affinity" is the same as the rest of Europe. This I can agree with - LB has the same affinity to Amerindians as modern Europeans do.

it's hard to say whether they add additional weight to the data in EDF 5d but they don't contradict it.

Of course they don't contradict it - they are both raw data. What graph a) contradicts is your interpretation from this data that LB has increased Amerindian affinity relative to modern Europeans.

I think you've managed to come up with another pseudoscientific antic

f3(A, X, O) only measures A vs X. Are you honestly disputing this?

It's common descent or admixture. There's no other reason.
Don't play hide and seek with me - what are those "alternative possibilities"? Convergence?

Non-shared affinity is the obvious one. X shares 10% of A's genome, Y shares a different 10%. Both will show up as having the same amount of affinity to A even though there is no affinity between X and Y.

Your mistake is assuming that an A-X affinity and a A-Y affinity necessitates an X-Y affinity. X and Y can each have individual common ancestry and/or admixture with A without necessarily having it with each other.

So how do we confirm whether the A-X affinity and the A-Y affinity is common or whether X and Y have different parts of A's genome? I put it to you that a direct measurement of X-Y affinity will reflect this - if X-Y have high affinity then it's reasonable to say that most of this affinity is the same affinity shown X, but if A-B shows little or no affinity then we can only conclude that the affinity each shows to X is unrelated.

In the case in point LB's affinity to MA-1 over Europeans can be explained as due to him being 7,000 years closer to their common ancestor. Amerindian affinity to MA-1 over Europeans can be explained as due to ancient admixture. In order to determine how much of MA-1 related DNA in LB is the same as the MA-1 related DNA in modern Amerindians we need a graph with LB or Amerindians on an axis. Luckily we have such a graph, EDF 5a which shows LB with no increased affinity to Amerindians relative to Europeans. Graphs 5c and 5e show Amerindians have roughly the same LB affinity as Central/South Asians, not directly related to the point but again not showing any special relationship between LB and Amerindians.

I note that if you compare 5c and 5d you can see that Amerindians have less affinity to Sardinians than C/SA but the same affinity to LB as C/SA - this could be due to C/SA having more Sardinian affinity than LB affinity (Middle Easterners are likewise shifted), but it could also suggest that Amerindians have more LB affinity than Sardinian affinity (you might argue for the latter based on the raw numbers, but you can't compare raw scores with a different X population - it's potentially a totally different sized set of SNPs). In either case Sardinians are only one population representing one extreme of European variation, and 5a still shows LB to be within the range of all modern Europeans relative to Amerindians.

German Dziebel said...

@Hamar Fox

"You never gave me your source, so I can only assume by Western European you mean LB1, and are referring to the PCA included in this paper."

I was referring to West Eurasians in general coming off of the Lazaridis ancient DNA paper and the earlier Patterson 2012 paper. But La Brana is a good case in point. You should peruse my recent exchanges with Tobus in this string where I suggest that EDF5 d in Olalde unambiguously places LB within the Amerindian cluster on X axis just like in EDF 5a, b LB falls into the West Eurasian cluster. This proves that LB is an admixed population and that it shares with MA-1 common ancestry in Amerindians. There are more details than that upstream but this is the gist.

"What this means is that, if W. and E. Eurasians stem from a greater genetic pool, i.e. that of Americans, then the unique modern characteristics of both of these groups are determined by inheritance of a subset of Amerindian diversity (selected down or lost through bottle-necking) and unique evolution. This would create W. and E. Eurasian signatures unique from Amerindians. But this must be a one-time process."

I'm fine with this. West Eurasians and East Eurasians split from the parental New World population in pre-Malta times. I think Tobus is favoring this scenario over a recent admixture from Amerindians into East Asians. In this case, such traits as EDAR, shovel-shaped incisors, etc. shared between East Asians and Amerindians must have been lost in West Eurasian populations. This very well may be the case as we see elevated (compared to modern Europeans) frequencies of shoveling in the dental profile of some Gravettian sites and in Catalhouyuk.

Tobus said...

@German:
You should peruse my recent exchanges with Tobus in this string where I suggest that EDF5 d in Olalde unambiguously places LB within the Amerindian cluster on X axis just like in EDF 5a, b LB falls into the West Eurasian cluster. This proves that LB is an admixed population and that it shares with MA-1 common ancestry in Amerindians.

I have informed, explained and exemplified to you multiple times this logic has a major flaw, but for some reason you just can't (or won't?) see it.

I think Tobus is favoring this scenario over a recent admixture from Amerindians into East Asians.

You think wrong, but I'm kind of getting used to that... listening to what other people are saying was obviously not a high priority in your vastly superior education.

German Dziebel said...

@Tobus

"LB has increased Amerindian affinity relative to Europeans, but you now seem to be saying that LB's "Amerindian affinity" is the same as the rest of Europe. This I can agree with - LB has the same affinity to Amerindians as modern Europeans do."

No, it's not the same. In EDF 5d LB is clearly pulled out of the West Eurasian cluster and falls into the Amerindian cluster. It's not as close to Amerindians as MA-1 but it's closer to Amerindians than modern West Eurasians.

"What graph a) contradicts is your interpretation from this data that LB has increased Amerindian affinity relative to modern Europeans."

The graph a) shows (to a certain degree because we - due to your laziness - don't have the actual values) that LB is closer to Amerindians than to East Asians. EDF 5d ascertains that LB is closer to Amerindians than modern West Eurasians.

"f3(A, X, O) only measures A vs X. Are you honestly disputing this?"

This is exactly how creationists debate. They often require scientists to chew it up for them to the last morsel because they live to the very end with a hope that they will be able to escape from the prison of facts and logic. We are not talking about a single populations. We have a sample of many. They are plotted together on the basis of their respective f3 values. f3 measures the same thing for every population. Hence, when all the populations are plotted together their f3 values display their relative proximity to each other. Naturally, these populations form clusters based on common descent/admixture. More recently diverged populations form tight clusters. More remotely related populations show pulls toward each other on the scale of the whole plot.

"X shares 10% of A's genome, Y shares a different 10%. Both will show up as having the same amount of affinity to A even though there is no affinity between X and Y."

Nonsense. French and Orcadians are both closer to Sardinians than Yoruba. They are close to each other, too. And lo and behold they share common descent. Why would it be different for LB and Amerindians. You're forgetting that LB falls INTO the Amerindian cluster against the X axis in EDF 5d.

"In order to determine how much of MA-1 related DNA in LB is the same as the MA-1 related DNA in modern Amerindians we need a graph with LB or Amerindians on an axis. Luckily we have such a graph, EDF 5a which shows LB with no increased affinity to Amerindians relative to Europeans. Graphs 5c and 5e show Amerindians have roughly the same LB affinity as Central/South Asians, not directly related to the point but again not showing any special relationship between LB and Amerindians."

EDF 5d is perfectly enough to establish the affinity between Amerindians, MA-1 and LB. In your previous comment, you agreed with this. You also agree that LB and modern Western Eurasians have Amerindian affinities. So i'm going to ignore your above attempt at filibustering the debate. What EDF 5 a,b is telling us is that modern DNA is not always enough to establish genetic affinity between populations that diverged long before the formation of the modern population clusters. Ancient DNA is necessary. Now that we have MA-1 and Anzick it becomes possible to see that ancient extreme west Europeans (LB) are related to modern extreme easterners (Amerindians). It's the same phenomenon as only the availability of ancient Armenian allows us to see that modern Armenian is an Indo-European language. Without this ancient attestation, it wouldn't have been possible to classify modern Armenian as related to modern Romance languages.

"Luckily we have such a graph, EDF 5a which shows LB with no increased affinity to Amerindians relative to Europeans."

We would need to have Anzick in the Amerindian sample. I predict that Anzick would create for LB a stronger pull toward Amerindians than the one that Karitiana shows.

German Dziebel said...

@TerryT

"That logic is not going to work with German. He appears to believe that every time a part of a population moves away it carries an exact sample of all the genes present in the source population. He does not accept bottlenecks ever happen."

A bottleneck shouldn't be a null hypothesis. It''s more of a special case that needs to be documented rather than assumed. Even in the case of such large and genetically diverse populations as Basques, historically recent Basque migrants in North America don't differ much (are no less diverse) genetically from European Basques. Same for a number of Afro-American populations. The fact that the most divergent Y-DNA lineage A00 is found in African Americans suggests that bottlenecks can't be posited as a general model, only as a special case.

In the case of demographically small and subdivided Pleistocene populations the notion of a bottleneck is even less viable as a general model. That's why we can easily have "Amerindian admixture" in Europe. Notably, no "African admixture" has so far been reported.

"Not so. All that has been confirmed is that Europeans and Amerindians share a genetic element from a common source. One is not 'descended from' the other."

No modern population is descended from another modern population. The presence of the Amerindian genetic component in Eurasia suggests that the common source population for modern Amerindians and modern Eurasians resided in the New World. Otherwise, we would've been talking about a an "Antarctic" or an "Australian" component.

Grey said...

@Tobus
"But despite at least 40-50kya of modern humans in low UV environments we only have evidence for light skin evolving just twice, and both of these in the last 10 or at most 20kya"

I don't think the EEF had the time or numbers to whiten all of Europe before record history so the three other possibilities are:

1. Albinism happens. It doesn't usually spread but maybe once it did.

2. Neanderthal admixture.

or simplest option

3. ANE or ANE/WHG people were already white or white-ish when they expanded into Europe so it wasn't just the EEF.

The Neanderthal option would be the most fun though.

terryt said...

"The Neanderthal option would be the most fun though".

I agree completely. And not impossible either.

"A bottleneck shouldn't be a null hypothesis. It''s more of a special case that needs to be documented rather than assumed. Even in the case of such large and genetically diverse populations as Basques, historically recent Basque migrants in North America don't differ much (are no less diverse) genetically from European Basques. Same for a number of Afro-American populations".

Comparing the size of the Basque reion to the whole af Africa is ridiculaous. And African-Americans do not contain a representative sample of African populations. They are primarily West African and mostly from a comparatively small region of West Africa.

"The presence of the Amerindian genetic component in Eurasia suggests that the common source population for modern Amerindians and modern Eurasians resided in the New World".

That is a huge lep of faith. Your 'out of America' hypothesis is absolutely full of inconsistencies, but I will point them out on Dienekes' Clovis blog.

German Dziebel said...

@Tobus

"You think wrong, but I'm kind of getting used to that... listening to what other people are saying was obviously not a high priority in your vastly superior education."

Yes, I'm totally confused about what you're saying.

"I have informed, explained and exemplified to you multiple times this logic has a major flaw, but for some reason you just can't (or won't?) see it.

I can't see what doesn't exist. But I have precisely identified a dozen of flaws on your part. The data we are having it hand is far from perfect, but I don't think I've admitted any flaws of interpretation or data reporting.

Tobus said...

@German:
I hear what you are saying and understand what you are getting at 100%, but you are making a fundamental mistake in your interpretation of the data. I've tried to point it out to you multiple times but you don't see to hear or understand me.

We have already resolved a couple of misunderstandings on your part, such as you believing if A is closer to B and to C, then B must be closer to A than to C, and you thinking D-statistics work differently to how they do. Now we seem to be stuck at another problem - you don't understand how these outgroup f3 plots work.

For f3(X, A, O), the result is calculated by (O-X)(O-A) where O, X, and A are the allele frequencies of each allele in the panel. Traditionally O is a population being tested as an admixture of X and A, but in these "outgroup" f3 stats, O is deliberately chosen as an outgroup to both X and A and the result is a measurement of X and A's shared genetic drift since divergence from O. The charts in EFD 5 plot the results of a number of separate f3 calculations so they can be visually compared.

As you can see from the maths involved there is absolutely no measurement of X-Y affinity when you plot f3(X, A, O) and f3(Y, A, O). When two populations get similar scores this means (and only means) that they share a similar amount of drift relative to O. The f3 stats are pure numbers and don't supply any reason for this correlation, we must infer the relationship based on other data.

Your mistake with graph d) is one of confusing correlation with causation - if I say "all people wearing glasses have poor eyesight", I hope you can see it's a logical fallacy to equate this with "all people with poor eyesight wear glasses" (some wear contact lenses, some just squint). In the case in point the maths of the f3 stat means that "all populations with shared MA-1 affinity will plot together on graph d)", but it's a logical fallacy to equate this with "all populations that plot together on graph d) share MA-1 affinity". The correlation between LB and Amerindians relative to MA-1 is there, but you are assuming the cause of this correlation must be the same as the cause for the Russian/French correlation against Sardinians when there is no logical reason to assume this must be the case.

You are entirely correct and justified to posit the possibility that LB's affinity to MA-1 on graph d) is due to common Amerindian affinity and hence LB must have more Amerindian affinity than modern Europeans, but you are not correct or justified to use graph d) as proof that this is the case when other data, such as graph a), explicitly rejects this conclusion.

Tobus said...

@Grey:
I don't think the EEF had the time or numbers to whiten all of Europe before record history

Stuttgart at 7000 years ago already had half on the known lightening alleles, so we can assume an origin some time earlier than this (at least 9-10 kya?). Recorded history of "white" people starts at around 4000 ago (2000 BC). That gives 5-6,000 years. Research says that the SLC45A2 sweep is one of the strongest ever seen and Jablonski estimated that under selection populations can change colour in as little at 100 generations (~2500 years)... so it's not impossible in terms of timing.

Neanderthal admixture is pretty much ruled out - no modern humans (white or otherwise) have the known Neanderthal lightening gene, and no Neanderthal has any of the known modern human lightening genes.

Hamar Fox said...

German,

I was referring to West Eurasians in general coming off of the Lazaridis ancient DNA paper and the earlier Patterson 2012 paper.

The significance of one population pulling toward another depends on the type of analysis. For example, f3 is designed specifically to tease out relationships by trying to force them to 'pull' toward or away from other populations. A single f3 analysis tells us only whether a population is more related to one or another population, or sits right in between, but says nothing about which of the populations are admixed, if any. Only over several f3 comparisons does this picture emerge, based on an 'if X is related to Y, and Y is related to Z, but X isn't related to Z' type logic.

On the other hand, in PCA analyses, one population lurching toward another and away from other populations that share the majority of said population's heritage, usually is evidence of admixture, especially when the PCA is comprehensive (Eurogenes' latest is a good place to look), and when there are plenty of unadmixed/relatively unadmixed reference populations included in the analysis. Naturally in PCAs, unadmixed populations make no gesture toward the admixed population, so the fact that East Africans plot between West Africans and West Eurasians does not in any way affect the relationship between West Africans and West Eurasians, provided the analysis doesn't get confused by a lack of samples from either or both of the parental populations.

Lazardis largely used formal admixture analyses (f3 and/or f4, IIRC), and the conclusion from a series of statistical tests was clearly that Amerindians 'pulled' toward W. Eurasians, while the relationship between W. Eurasians and E. Eurasians (represented by Onge and, I think, Han) remained unaffected. Again, this was based on formal statistics, but the logic is the same as what I said about interpreting PCAs: the relationship of the possible parental populations to each other is key.

But La Brana is a good case in point. You should peruse my recent exchanges with Tobus in this string where I suggest that EDF5 d in Olalde unambiguously places LB within the Amerindian cluster on X axis just like in EDF 5a, b LB falls into the West Eurasian cluster. This proves that LB is an admixed population and that it shares with MA-1 common ancestry in Amerindians. There are more details than that upstream but this is the gist.

EDF 5d shows that LB is similar to Amerindians in their relationship to MA-1 along the x axis. I can only really repeat what Tobus has already said about this. Both Amerindians and mesolithic Europeans are related to Mal'ta.

On a side note, LB's full genome analysis shows him to have the same Amerindian component as all other mesolithic Europeans so far assessed by the same calculator (Dodecad Globe4):

http://genetiker.wordpress.com/2014/01/29/admixture-analyses-of-the-complete-la-brana-1-genome/

Considering LB's relationship to Karitiana is much weaker than his relationship to MA-1, I'd say this supports my earlier contentions that the k=4 'Amerindian' component in Europeans is actually more of a MA-1-like component -- not necessarily an ancestral connection (MA-1 was not ancestral to LB), but a measure of similarity.

This very well may be the case as we see elevated (compared to modern Europeans) frequencies of shoveling in the dental profile of some Gravettian sites and in Catalhouyuk.

Then the process must have begun early:

"Yet, the Upper Paleolithic artifacts from the 23,000-year-old Mal’ta site near Lake Baikal in south-central Siberia (Medvedev et al. 1996) have been found in association with skeletal remains that bear similar morphology with contemporary anatomically modern humans teeth from Europe thus providing the evidence for links between Siberia and the West during the Upper Paleolithic."

http://dienekes.blogspot.co.uk/2005/11/y-chromosomes-of-south-siberians.html

Rokus said...

'Both Amerindians and mesolithic Europeans are related to Mal'ta.'

Since Mal'ta is older than both Amerindians and mesolithic Europeans, per definition this can't mean that MA-1 is a mixture of such modern components, something apparently suggested by some commentors. Instead, Mal'ta should be considered ancestral to its constituent components. Those ancestry components of Mal'ta Boy (MA-1) thus rather 'reappear' after millennia of spatial isolation and drift in western Eurasia (dark blue, 34%), the Americas (orange and pink, 17%), central and south Asia (green, 37%) and even Oceania (purple, 4%). As I explained elswhere, this strongly indicates an early worldwide human expansion having MA-1 near the epicenter. Apparently, East Asian, Siberian, Near Eastern and African components were not directly involved in the initial peopling of Northern Eurasia and the Americas, while the higher levels of 'dark blue' and especially 'green' suggest a longer period of shared drift within the wider expanse of Northern Eurasia against an earlier Amerindian split-off.

German Dziebel said...

@Tobus

"We have already resolved a couple of misunderstandings on your part, such as you believing if A is closer to B and to C, then B must be closer to A than to C, and you thinking D-statistics work differently to how they do."

Let's not talk about my typos, let's continue to work through your logical flaws and unscientific biases.

"if I say "all people wearing glasses have poor eyesight", "

Your Halle Berry and eye-glasses analogies are distracting. We have plenty of real evidence to work with.

"The correlation between LB and Amerindians relative to MA-1 is there, but you are assuming the cause of this correlation must be the same as the cause for the Russian/French correlation against Sardinians when there is no logical reason to assume this must be the case."

No, it is how scientific logic works. One should interpret a specific case in the light of a general rule. On all plots in Olalde (and elsewhere) Middle East gravitates toward the Yoruba pole because Middle Easterners share genetic affinity with Sub-Saharan Africans. On EDF 5b East Asians are shifted all the way rightwards on the Han axis. All modern West Eurasians and East Asians cluster with their own "kind." MA-1 and LB share genetic affinity because they are pulled toward each other. EDF 5d suggests that Mesolithic LB is a subset of Amerindian variation when a Paleolithic sample with Amerindian affinities is used.

What you are trying to say is that we need to have more data to test this inference. I would agree with that. But you should stop confusing data and logic. My logic provides the best theory to explain the existing data.

"You are entirely correct and justified to posit the possibility that LB's affinity to MA-1 on graph d) is due to common Amerindian affinity and hence LB must have more Amerindian affinity than modern Europeans, but you are not correct or justified to use graph d) as proof that this is the case when other data, such as graph a), explicitly rejects this conclusion."

It doesn't explicitly reject "this conclusion." EDF 5 d is not a "conclusion", it's evidence. So we need to harmonize the two pieces of evidence. The most natural way to do it is to hypothesize that a terminal Pleistocene Amerindian sample (such as Anzick) or an even more ancient Amerindian-derived sample (MA-1) creates a stronger pull for an ancient extreme West Eurasian sample than a modern Amerindian sample. Modern Amerindians are closer to modern East Asians than to modern West Eurasians (following geography) and to ancient West Eurasians than to modern East Asians. It's clear that in EDF 5c, d (with MA-1 as an anchor) West Eurasians swap places with East Asians vis-a-vis Amerindians strongly suggesting that West Eurasians carry Amerindian admixture or that they descended from Amerindians. And in EDF 5d, LB falls with other Amerindians compared with modern West Eurasians because LB is 7,000 years closer to the Amerindian source. What creates this pull on the MA-1 axis is the fact that MA-1 is part Amerindian. If MA-1 were West Eurasian, the West Eurasian cluster would've showed up on the extreme right on the MA-1 anchored axis. MA-1 and LB would've been within the range of West European variation, while Amerindians would've been pulled into their orbit from somewhere in the middle. But the data explicitly rejects this as Amerindians remain rightmost regardless of whether Karitiana or MA-1 are used as an anchor.

German Dziebel said...

@Hamar Fox

"EDF 5d shows that LB is similar to Amerindians in their relationship to MA-1 along the x axis. I can only really repeat what Tobus has already said about this. Both Amerindians and mesolithic Europeans are related to Mal'ta."

I just have to refer you back to my latest response to Tobus. The relationship between LB and Amerindians to MA-1 is due to the Amerindian component in MA-1. This is very easy to see on EDF 5: LB and MA-1 are pulled out of the West Eurasian cluster, while Amerindians remain rightmost on the x axis regardless of whether MA-1 or Karitiana are used as an anchor. If MA-1 was related to LB as a West Eurasian population and West Eurasians admixed into Amerindians, we would've seen West Eurasians positioned rightmost on the MA-1 anchored x axis, plus LB and MA-1 would've been everywhere part of the West Eurasian cluster. Amerindians would've been somewhere in the middle and pulled toward West Eurasians on the MA-1 anchored x axis.

"LB's full genome analysis shows him to have the same Amerindian component as all other mesolithic Europeans "

Exactly. Olalde data confirms that Mesolithic Europeans such as LB are more Amerindian than modern Europeans consistent with the progressive weakening of an ancient genetic signal.

"Considering LB's relationship to Karitiana is much weaker than his relationship to MA-1, I'd say this supports my earlier contentions that the k=4 'Amerindian' component in Europeans is actually more of a MA-1-like component -- not necessarily an ancestral connection (MA-1 was not ancestral to LB), but a measure of similarity."

Again, regardless of the relative proximity between LB and MA-1 compared to LB and Karitiana, this component is Amerindian both in MA-1 and in LB because MA-1 is closer to Amerindians than to the rest of West Eurasians (Raghavan).

@Rokus

"Since Mal'ta is older than both Amerindians and mesolithic Europeans, per definition this can't mean that MA-1 is a mixture of such modern components, "

Whether MA-1 as a sample is older than modern Amerindians seems to be irrelevant. It's a matter of attestation, not genetic derivation. Anzick is even closer to MA-1 than modern Amerindians suggesting that, at least in the case of America, modern components are good proxies for ancient components.

Rokus said...

Correction:

Apparently, East Asian, Siberian, Near Eastern and African components were not available during the initial Northern Eurasian expansion: MA-1 does not say anything about their possible participation during later events that involve the Americas.

Rokus said...

"Since Mal'ta is older than both Amerindians and mesolithic Europeans, per definition this can't mean that MA-1 is a mixture of such modern components, "

Whether MA-1 as a sample is older than modern Amerindians seems to be irrelevant. It's a matter of attestation, not genetic derivation. Anzick is even closer to MA-1 than modern Amerindians suggesting that, at least in the case of America, modern components are good proxies for ancient components.

Of course the relative age of the components to the sample is relevant. Modern components came into being by drift and admixture. Raghavan's model was found to have the best predictive accuracy using nine modern reference populations, what still doesn't exclude admixture for each component though it considerably enhances the contribution of drift for each component. It may be easily reconstructed that MA-1's components drifted apart in various derived populations, whether or not those derived populations also experienced additional admixture. Completely predictable that Anzick should have been closer to MA-1, since having experienced less drift it still wouldn't have been as much closer to MA-1 if the latter had become admixed with something else.

Hamar Fox said...

German,

I just have to refer you back to my latest response to Tobus. The relationship between LB and Amerindians to MA-1 is due to the Amerindian component in MA-1. This is very easy to see on EDF 5: LB and MA-1 are pulled out of the West Eurasian cluster

LB is only definitively pulled out of the W. Eurasian cluster in EDF 5d, i.e. in his relation to MA-1. In his relation to both Han and Karitiana vis-a-vis other W. Eurasians, he's only just outside the main blue cluster and in an ambiguous relationship with the other two European populations (obviously Russians and Finns). But it can be seen in the same figure that all of the Europeans less related to Han and Karitiana are also more related to Sardinians than is LB. Since we know farmers were less Eastern-shifted than hunter gatherers, and Sardinians are the most neolithic-descended European population, LB's position is consistent with Lazaridis et al.

If MA-1 was related to LB as a West Eurasian population and West Eurasians admixed into Amerindians, we would've seen West Eurasians positioned rightmost on the MA-1 anchored x axis

Not if we take into account two things: 'W. Eurasian' isn't a monolith. As a genetic concept, it is internally differentiated, and at present we know that it involved, minimally, two major components, at least as concerns Europeans; i.e. European and Siberian hunter-gatherers, and the Near Eastern-like farmers. Since MA-1 is overwhelimingly more related to one of these groups (HG) than to the other (farmers), then to consider the major effects of admixture of the latter with the former in the profile of modern Europeans is essential in understanding why their relationship to MA-1 is slightly diminished. think of it as a milder case of what happens to Middle Eastern populations in the same figure as a result of SSA admixture.

Second, we need to account for the effects of direct and indirect relationships. If MA-1 admixed directly or in greater proportion into the Americas, while his relationship with populations further west was simply one of relatively recent common ancestry (or, if there was admixture into Europeans, it was less), then what we see in EDF 5 is consistent. To go back to my East African analogy from before: They may well be more strongly related to SW Asians than SW Asians are to NE Europeans (in f3, not PCA), even though SW Asians and NE Europeans are both W. Eurasian, because a direct exchange of genes with one population (in sufficient amounts), however divergent overall, may prove to be stronger than a mere ancestral relationship with a less divergent population.

Exactly. Olalde data confirms that Mesolithic Europeans such as LB are more Amerindian than modern Europeans consistent with the progressive weakening of an ancient genetic signal.

But the signal is stronger in some modern groups than in others, even within Europe, and the dampening of the signal in modern Europeans is consistently proportionate to their farmer vs. mesolithic ancestry. Why so? If Sardinians and, say, Orcadians are equally descended from Amerindians, then why is the signal stronger in the one than the other? Unless for social or environmental reasons, the ancestors of Sardinians reproduced on average in shorter generations, so that they'd be a greater number of generations removed from Amerindians than their contemporaries, or Orcadians received additional waves of admixture in addition to the natural waning of the signal, I don't see how the signal declining with time fits the data.

The signal's almost complete absence in populations that aren't European, S. Asian or E. Asian also doesn't fit with a single wave out of the Americas. But it does fit with the fact that Europeans and S. Asians are related to the MA-1 portion of Amerindians (since MA-1 has a strong Europeans and S. Asian affinity) and East Asians are related to, well, the East Asian portion of Amerindians' ancestry.



terryt said...

"let's continue to work through your logical flaws and unscientific biases".

Let's consider German's as well. Take this for example:

"The most natural way to do it is to hypothesize that a terminal Pleistocene Amerindian sample (such as Anzick) or an even more ancient Amerindian-derived sample (MA-1) creates a stronger pull for an ancient extreme West Eurasian sample than a modern Amerindian sample".

This 'logic' is totally flawed. In trying to make the case that MA-1 is derived from Amerindians rather than being ancestral to them you first of all assume MA-1 is derived (as shown by your calling it an 'ancient Amerindian-derived sample'). You then proceed, based on that assumption, to 'prove' it is indeed an 'ancient Amerindian-derived sample'. That's 'logic'?

Tobus said...

@German:
Let's not talk about my typos

Typos?!?! What typos?

Modern Amerindians are closer ... to ancient West Eurasians than to modern East Asians

Nonsense. Look at graphs 5a and 5b - on both graphs East Asians and Amerindians show higher affinities to the subject poplations (Karitiana and Han respectively) than LB and MA-1 do.

LB falls with other Amerindians compared with modern West Eurasians because LB is 7,000 years closer to the Amerindian source. What creates this pull on the MA-1 axis is the fact that MA-1 is part Amerindian. If MA-1 were West Eurasian, the West Eurasian cluster would've showed up on the extreme right on the MA-1 anchored axis.

This is where your false assumption is most apparent - West Eurasian *are* on the far right of the MA-1 axis, only slightly further away than Amerindians (0.18 vs, 0.19), yet you assume only the Amerindian aspect is responsible for explaining the LB affinity. In terms of percentage this 0.01 difference in the f3 score equates to MA-1's drift being shared proportionally 52.5% by Amerindians and 47.5% by Europeans making either population roughly equally likely as the source. LB doesn't plot with Amerindians in any other chart and on the LB axis (EDF 5e) MA-1 plots right in the middle of Europeans, nowhere near the Amerindians. All this points very strongly to LB being related to MA-1 purely on the West European side, and graph 5a confirms this as LB has no increased Amerindian affinity relative to Europeans. Considering that LB is 7,000 years (about 30%) closer to MA-1 than Europeans, is it not worth considering that it's this greater affinity with MA-1's West Eurasian component that pulls him 0.01 closer to MA-1 in the f3 stats?

German Dziebel said...

@Tobus

"Typos?!?! What typos?"

When I wrote West Eurasians instead of Amerindians this guiding you and Hamar Fox down the wrong path of believing that I can't see how a population A can be closer to population B and more removed from population C or something like that.

"This is where your false assumption is most apparent - West Eurasian *are* on the far right of the MA-1 axis, only slightly further away than Amerindians (0.18 vs, 0.19), yet you assume only the Amerindian aspect is responsible for explaining the LB affinity."

What other aspect? MA-1 is a Paleolithic offshoot of Amerindians, modern Amerindians are the closest to it, followed by Mesolithic Europeans (LB) who fall within the modern Amerindian variation, followed in turn by modern West Eurasians. Modern West Eurasians, on the y axis, have an additional vertical pull toward Sardinians compared to LB which reflects a Neolithic admixture from the Middle East. My explanation of this pattern is the most reasonable one. The Paleo-Mesolithic substrate has Amerindian roots, the Neolithic superstrate has Middle Eastern roots.

You are not happy with the numbers? That's the data we have. And the pattern is clear: the most divergent Amerindians are at 0.20, the least divergent West Eurasians are 0.175. The most divergent East Asians are at 0.172, the least divergent ones are at 0.161.

"Look at graphs 5a and 5b - on both graphs East Asians and Amerindians show higher affinities to the subject poplations (Karitiana and Han respectively) than LB and MA-1 do."

Look at graphs 5c and 5d: Amerindians are closer to West Eurasians (in fact ancient and modern) than they are to modern East Asians. Or, modern and ancient Eurasians are closer to Amerindians than modern East Asians. And 5c and 5d are anchored in an ancient sample located in East Asia. So clearly there was no split between East Asians and Amerindians followed by a West Eurasian injection into Amerindians. There was either a split between Amerindians and West Eurasians followed by a split between West Eurasians and East Asians, or, as graphs 5a and 5b tells us, there was a split within Amerindians followed by a migration to West Eurasia and East Asia.

"LB doesn't plot with Amerindians in any other chart"

On 5 c and e when the x axis is anchored in LB, Amerindians get a pull upwards to 0.17 whereas their usual spot is at 0.16 when the x axis is anchored in Sardinians. This confirms a special connection between Amerindians and LB.

"And on the LB axis (EDF 5e) MA-1 plots right in the middle of Europeans, nowhere near the Amerindians."

Duh, this confirms that MA-1 is a hybrid Amerindian/West Eurasian population. And on all the plots where MA-1 is shown, it's pulled out of the West Eurasian cluster toward Amerindians (rightwards). And even on 5e it's the least "La Brana" than most other West Eurasians. How does the fact that MA-1 has West Eurasian affinities change the fact that MA-1 has Amerindian affinities? That's what is called an admixed population.

German Dziebel said...

@Tobus (contd.)

""All this points very strongly to LB being related to MA-1 purely on the West European side."

Impossible! LB is Amerindian when the analysis if anchored in MA-1. If LB was related to MA-1 on the West Eurasian side, it would've fallen with the rest of West Eurasians on the MA-1 anchored axis.

"graph 5a confirms this as LB has no increased Amerindian affinity relative to Europeans. "

Already explained why this is so.

"Considering that LB is 7,000 years (about 30%) closer to MA-1 than Europeans,"

As a side note, although this is a fact, EDF 5a,b don't reflect this at all. LB falls with the rest of West Eurasians.

"is it not worth considering that it's this greater affinity with MA-1's West Eurasian component that pulls him 0.01 closer to MA-1 in the f3 stats?"

No, it's not. The pull would've been much stronger if it were on the West Eurasian side. Way more rightwards than Amerindians. Between West Eurasians and Amerindians, Amerindians are most MA-1-like and Anzick confirms this.

German Dziebel said...

@Hamar Fox

There's not much I disagree with in what you wrote, but I also don't see much of a pushback against the out-of-America hypothesis.

"LB is only definitively pulled out of the W. Eurasian cluster in EDF 5d, i.e. in his relation to MA-1. In his relation to both Han and Karitiana vis-a-vis other W. Eurasians, he's only just outside the main blue cluster and in an ambiguous relationship with the other two European populations (obviously Russians and Finns)."

I discussed it with Tobus - most likely, considering the time that has elapsed since the Amerindian-West Eurasian split and the small size of Amerindian populations, it takes an Amerindian-derived ancient sample such as MA-1 to make an Amerindian pull in such a westernmost sample as LB visible.

"But it can be seen in the same figure that all of the Europeans less related to Han and Karitiana are also more related to Sardinians than is LB. Since we know farmers were less Eastern-shifted than hunter gatherers, and Sardinians are the most neolithic-descended European population, LB's position is consistent with Lazaridis et al."

Yes, I agree. Modern West Eurasians are pulled away from Eastern non-Africans because of Neolithic admixture. Still, all of them are pulled toward the Amerindian pole (roughly at 0.16 in EDF 5a).

". Since MA-1 is overwhelimingly more related to one of these groups (HG) than to the other (farmers), then to consider the major effects of admixture of the latter with the former in the profile of modern Europeans is essential in understanding why their relationship to MA-1 is slightly diminished. "

Again, I have no problem seeing how the ancient Amerindian HG signal got diluted by Neolithic farmers. A combination of factors such as time, geography and external admixture have created the observable divergence between modern West Eurasians and modern Amerindians.

"But the signal is stronger in some modern groups than in others, even within Europe, and the dampening of the signal in modern Europeans is consistently proportionate to their farmer vs. mesolithic ancestry. Why so? If Sardinians and, say, Orcadians are equally descended from Amerindians, then why is the signal stronger in the one than the other? "

Some got more affected by the farmer gene flow from the Middle East in the past 10,000 years than the others. Or even later the coming of nomadic Indo-Europeans from the East has replenished the supply of Amerindian-derived genes in Western Europe creating a somewhat choppy landscape.

"The signal's almost complete absence in populations that aren't European, S. Asian or E. Asian also doesn't fit with a single wave out of the Americas."

It's just weaker but it's there - Middle East, South Asia, central Asia or what have you - they are all pulled rightwards toward the Karitiana pole in EDF 5a and e and even more so toward the MA-1 pole in 5 c, d. We see this Amerindian signal everywhere on those ADMIXTURE graphs. It's muted some places more than others but it's there.

"But it does fit with the fact that Europeans and S. Asians are related to the MA-1 portion of Amerindians (since MA-1 has a strong Europeans and S. Asian affinity) and East Asians are related to, well, the East Asian portion of Amerindians' ancestry."

There's no such thing as the "MA-1 portion of Amerindians." It's just a sample giving us an idea of the ancient distribution of some modern components. There's the Amerindian portion of MA-1 and the fact that South Asian GREEN in Raghavan is not seen in the New World at all, while it's present in MA-1 suggests that its emergence postdates the migration of ancient Amerindians to South Siberia.

Tobus said...

@German:
When I wrote West Eurasians instead of Amerindians this guiding you and Hamar Fox down the wrong path of believing that I can't see how a population A can be closer to population B and more removed from population C or something like that.

You mean when you said "I can't wrap my mind around the logic whereby MA-1 is closer to LB than to other Europeans but LB is closer to other Europeans than to MA-1?

What other aspect?

The West Eurasian one.

Look at graphs 5c and 5d: Amerindians are closer to West Eurasians (in fact ancient and modern) than they are to modern East Asians.

Why do you insist on making assumptions of affinity between two populations from their f3 stats against a third? It's not a valid interpretation. We have direct measurements of affinity against both Han and Karitiana so there's no need to make something up based on how they plot against MA-1. The direct measurements show Amerindians are clearly closer to East Asians than they are to West Eurasians.

On 5 c and e when the x axis is anchored in LB, Amerindians get a pull upwards to 0.17 whereas their usual spot is at 0.16 when the x axis is anchored in Sardinians. This confirms a special connection between Amerindians and LB.

I already discussed this - the raw f3 numbers are based on a calculation of allele frequency differences in the A and Yoruba populations, you can't assume the numbers on the axes of different f3s are comparable because the allele frequencies being multiplied will be different in each case. The relative positions of the the other populations in graphs c/d *could* suggest an LB-Amerindian affinity, but note that the same shift is present in East Asian, and Oceanian populations as well and that the Middle Easterners and C/S Asians are shifted the other way. This is consistent with a generalised EEF vs WHG difference, not an specific MA-1 or Amerindian affinity. Any Amerindian-specific affinity this includes is still within the bounds of modern European affinity as shown by graph a.

Duh, this confirms that MA-1 is a hybrid Amerindian/West Eurasian population.

... or that Amerindians contain MA-1 admixture. The point holds that there a string West Eurasion/LB/MA-1 lineage and the shift of graph d) is possibly caused by it (definitely when you consider graph a).

Tobus said...

@German (cont.):
Impossible! LB is Amerindian when the analysis if anchored in MA-1. If LB was related to MA-1 on the West Eurasian side, it would've fallen with the rest of West Eurasians on the MA-1 anchored axis.

You don't seem to get it, "LB gets the same score as Amerindians" does not equate to "LB is Amerindian". LB falls 0.01 (about 5%) closer to MA-1 than Europeans, consistent with him being 7,000 years closer. The correlation with Amerindians is an artefact due to Amerindians having their own slightly increased (~5%) affinity to MA-1 over Europeans. You are confused because you expect LB to have Amerindian affinity so you assume it must be the cause, but all the other graphs (a, b, c and e) show that he doesn't.

Already explained why this is so.

Your "explanation" for this is to simply ignore the data as incorrect and posit a model that's contradictory to it. How can you honestly justify saying "this finding confirms the 'easternmost' (Amerindian-like) affinities of European hunters-gatherers" when a direct measure of Amerindian affinity places the LB hunter-gatherer in the modern European cluster? You made a generalised statement that is contradicted by the data - just admit it and let's put this to rest.

As a side note, although this is a fact, EDF 5a,b don't reflect this at all. LB falls with the rest of West Eurasians.

EDF 5a,b aren't measuring LB against MA-1, they're measuring both LB and MA-1 against Sardinians. This difference is shown by MA-1 scoring lower than LB does. Your consistent misunderstanding of what is being presented on these graphs is dumbfounding... I dread to think what you make of 5e where MA-1 plots inside the modern European cluster.

No, it's not. The pull would've been much stronger if it were on the West Eurasian side.

Yet there'd be no pull at all if it were from the Amerindian side? LB plots at the least MA-1 end of the Amerindian cluster, yet following the exact same reasoning you applied to the European side, LB should be further right than Amerindians if the pull was from the Amerindian side. It seems you are being very inconsistent in your expectations here... is there a good reason why LB would plot further away from MA-1 than all but one Amerindian sample despite being 7,000 years closer?

German Dziebel said...

@Tobus

"You mean when you said "I can't wrap my mind around the logic whereby MA-1 is closer to LB than to other Europeans but LB is closer to other Europeans than to MA-1? "

Yes, this may be the one. Can you get over this already?

"Why do you insist on making assumptions of affinity between two populations from their f3 stats against a third? It's not a valid interpretation. "

Cease and desist your creationist argumentation! It's a perfectly valid assumption. In f3 charts all Europeans form a cluster, all East Asians form a cluster, all Amerindians form a cluster and all those clusters are based on shared genetic history.

"We have direct measurements of affinity against both Han and Karitiana so there's no need to make something up based on how they plot against MA-1. The direct measurements show Amerindians are clearly closer to East Asians than they are to West Eurasians."

The direct measurements (EDF 5a, 5e) show that all Eurasian populations are shifted toward modern Amerindians to a varying degree. MA-1 is an ancient Amerindian-derived population n Siberia, which allows us to see how this "shift" occurred in history. It was likely a split between two Amerindian groups that later gave West Eurasians vs. East Asians.

"This is consistent with a generalised EEF vs WHG difference, not an specific MA-1 or Amerindian affinity."

No, all Eurasian populations are shifted toward Amerindians on the x axis. MA-1 makes the pull even stronger, again supporting the Amerindian nature of MA-1. And in Lazaridis WHG is closer to Amerindians than to other eastern non-Africans.

"Amerindians contain MA-1 admixture. The point holds that there a string West Eurasion/LB/MA-1 lineage and the shift of graph d) is possibly caused by it (definitely when you consider graph a)."

This is counterfactual. MA-1 is mtDNA U and Y-DNA R - two lineages not present in Amerindians. ADMIXTURE, too, doesn't show West Eurasian BLUE in America but it does show up in MA-1. Plus we have Amerindian ADMIXTURE colors all over Eurasia.

"The correlation with Amerindians is an artefact due to Amerindians having their own slightly increased (~5%) affinity to MA-1 over Europeans. You are confused because you expect LB to have Amerindian affinity so you assume it must be the cause, but all the other graphs (a, b, c and e) show that he doesn't."

Like I said, other graphs show that all Eurasians are shifted toward Amerindians, so LB just makes this pull a tad stronger due to it being 7,000 years closer to the split between two ancient Amerindian populations, one of which gave rise to West Eurasians.

"How can you honestly justify saying "this finding confirms the 'easternmost' (Amerindian-like) affinities of European hunters-gatherers" when a direct measure of Amerindian affinity places the LB hunter-gatherer in the modern European cluster?"

And this modern European cluster as a whole is shifted toward Amerindians in those direct tests. Once an ancient Amerindian-derived MA-1 sample is included, LB leaves it's modern European cluster and falls into the Amerindian cluster.

German Dziebel said...

@Tobus (contd.)

"You made a generalised statement that is contradicted by the data - just admit it and let's put this to rest."

What is your problem, Tobus? Your stubbornness, lack of education and pseudoscientific biases have taken you down the road of intellectual bankruptcy if not ethical delinquency. You are not using any facts in this debate. You're just trying to appeal to judges.

"EDF 5a,b aren't measuring LB against MA-1, they're measuring both LB and MA-1 against Sardinians."

My point was that not every chart shows the 7,000 yo difference between LB and modern Europeans. They all show different sides of the story. SO your expectation that LB should always show an increased affinity to modern Amerindians is unjustified.

"LB plots at the least MA-1 end of the Amerindian cluster, yet following the exact same reasoning you applied to the European side, LB should be further right than Amerindians if the pull was from the Amerindian side."

No, if MA-1 were a West Eurasian population related through the West Eurasian line to LB, on the one hand, and to Amerindians, on the other, then West Eurasians would've been further to the right from Amerindians on the MA-1 anchored axis and LB-1 would have been closer to MA-1 than to Amerindians. But in reality, LB is part of an Amerindian cluster on the MA-1 anchored axis.

terryt said...

@ German:

"MA-1 is mtDNA U and Y-DNA R - two lineages not present in Amerindians".

And yet you claim MA-1 is American in origin. How do you explain its absolute lack of American haploid DNA? I look forward to you appropriately convoluted reply.

Hamar Fox said...

German,

I discussed it with Tobus - most likely, considering the time that has elapsed since the Amerindian-West Eurasian split and the small size of Amerindian populations, it takes an Amerindian-derived ancient sample such as MA-1 to make an Amerindian pull in such a westernmost sample as LB visible.

To test whether this is correct, we can make the following prediction: If we replace MA-1 with Anzick-1 vs. Sardinians, LB will pull toward Anzick-1 in a similar fashion to what we see in EDF 5d; i.e. LB will be closer to Anzick-1 than even the modern European populations that are closer to modern Amerindians than is LB (in EDF 5a).

All the data needed to check this is available, and someone may even have done it already.

Yes, I agree. Modern West Eurasians are pulled away from Eastern non-Africans because of Neolithic admixture. Still, all of them are pulled toward the Amerindian pole (roughly at 0.16 in EDF 5a).

Correct, but it's also expected if all modern W. Eurasians are related to the MA-1-like element in Amerindians. It's clear that the HG element in modern W. Eurasians is overwhelmingly more related to MA-1 than the Neolithic-derived element in them, but this doesn't preclude at least a mild relationship between the Neolithic element and MA-1.

One way to check the general Amerindianness of non-ENAs without a MA-1 connection would be to analyse the relationship of sub-Saharan Africans and Amerindians vs. SSAs and East Asians. If SSAs are closer to Amerindians relative to East Asians than can be explained by W. Eurasian admixture into Amerindians (since W. Eurasians are more African-related than E. Asians), then we would have a population unconnected to MA-1 that shows evidence of an Amerindian connection.

We see this Amerindian signal everywhere on those ADMIXTURE graphs. It's muted some places more than others but it's there.

Oceanic and paleo African components also typically generate noise in almost all populations. An example of Australasian noise:

https://docs.google.com/spreadsheet/ccc?key=0ArAJcY18g2GadGlpc3JQaVdQbS1QTWF3SzNjTVdfZEE#gid=0

There's the Amerindian portion of MA-1 and the fact that South Asian GREEN in Raghavan is not seen in the New World at all, while it's present in MA-1 suggests that its emergence postdates the migration of ancient Amerindians to South Siberia.

Anzick-1 shows 12.11% Finnish and 6.61% Burusho in 'old world' calculators (i.e. when an 'Amerindian' component is absent). This is the same ratio of N. European to South-Central Asian in MA-1 and suggests that a population with MA-1's genetic make-up may comprise ~20% of Anzick-1's ancestry.

Source: http://genetiker.wordpress.com/2014/02/16/analyses-of-the-anzick-1-genome/

Tobus said...

@German:
No, all Eurasian populations are shifted toward Amerindians on the x axis. MA-1 makes the pull even stronger

This is the crux of the problem - you are assuming that a shift towards MA-1 is a shift towards modern Amerindians. MA-1 is *not* a modern Amerindian population, he's a paleo-Siberian with affinity not just to Amerindians, but to Europeans and Central/South Asians as well. Depending on which data you use, the amount of DNA MA-1 shares with Amerindians ranges from 16% (ADMIXTURE) to ~50% (my quick and dirty extrapolation from f3 stats not considering C/S Asians), with Raghavan estimating "14 to 38%". It's plain logic to me that any population that shows some affinity to MA-1 can either a) share all of this with Amerindians, b) share some of this with Amerindians or c) share none of this with Amerindians. You seem to be implying that a) must always be the case. Can you explain the reasoning behind the assumption that any affinity to MA-1 has to overlap with MA-1's affinity to Amerindians.

German Dziebel said...

@Tobus

"You seem to be implying that a) must always be the case. Can you explain the reasoning behind the assumption that any affinity to MA-1 has to overlap with MA-1's affinity to Amerindians."

Sometimes it's good to look at the data when you're struggling to make a choice between 3 seemingly equally plausible options. Direct measurement with Karitiana always creates a rightward pull for nearly all Eurasian populations. MA-1 does the same thing and the pull is even stronger. When compared to MA-1, LB is Amerindian.

@Hamar Fox

"To test whether this is correct, we can make the following prediction: If we replace MA-1 with Anzick-1 vs. Sardinians, LB will pull toward Anzick-1 in a similar fashion to what we see in EDF 5d; i.e. LB will be closer to Anzick-1 than even the modern European populations that are closer to modern Amerindians than is LB (in EDF 5a)."

Yes, agree, and one indication that this indeed may be the case is Rasmussen et al.'s plot showing that Anzick is shifted further to the right than modern Amerindians on the x axis anchored in MA-1.

"If SSAs are closer to Amerindians relative to East Asians than can be explained by W. Eurasian admixture into Amerindians (since W. Eurasians are more African-related than E. Asians), then we would have a population unconnected to MA-1 that shows evidence of an Amerindian connection."

SSAs and Amerindians tend to be most divergent from each other. East Asians are always more African shifted compared to Amerindians. This suggests that the Amerindian-West Eurasian connection implies gene flow from Amerindians to West Eurasians and not the other way around.

"It's clear that the HG element in modern W. Eurasians is overwhelmingly more related to MA-1 than the Neolithic-derived element in them, but this doesn't preclude at least a mild relationship between the Neolithic element and MA-1."

Lazaridis: All European ancient samples including Stuttgart (Neolithic!) have a stronger affinity to Amerindians than to other ENAs.

"it's also expected if all modern W. Eurasians are related to the MA-1-like element in Amerindians. "

But since this "MA-1 element" doesn't continue in modern East Asians, Siberians and extreme NE Asians (Chukchees) as much as it does in modern Amerindians, we have to treat it as an "Amerindian" element in an LGM-dated South Siberian population. We obviously need more data to test it but considering that New World harbors 2/3 of world linguistic diversity (leaving out PNG), Amerindian antiquity should be taken seriously.

German Dziebel said...

@TerryT

"And yet you claim MA-1 is American in origin. How do you explain its absolute lack of American haploid DNA. I look forward to you appropriately convoluted reply.?"

I will disappoint you, Terry, with a rather simple answer. Genomewide, MA-1 is majority West Eurasian and minority Amerindian. There are well known examples when original haploid lineages of a population are completely replaced with admixed lineages. E.g., Munda mtDNA is completely South Asian, while they likely came to India from SE Asia. So this is what happened with MA-1 - it's original Amerindian lineages were replaced with West Eurasian lineages. A very natural explanation, it seems.

Alternatively, the whole mtDNA tree is full of problems, so it may turn out that mtDNA U lineages are in fact a subset of B lineages. Remember that the whole R11'B6 clade is defined by a supposedly "reverse" mutation T16189C! If, instead, it's deemed an ancestral retention (Denisova and Neandertals are C16189), then B 11'B6 branched off before U.

terryt said...

"Genomewide, MA-1 is majority West Eurasian and minority Amerindian".

I thought you had consistently claimed MA-1 was mainly Amerindian.

"There are well known examples when original haploid lineages of a population are completely replaced with admixed lineages".

But here you seem to be claiming the admixture into MA-1 was from a non-existent population. Doesn't make sense.

"Alternatively, the whole mtDNA tree is full of problems"

'full of problems'? Why have you not yet attempted to provide at least the basis of an alternative phylogeny if you are so sure of that?

"since this 'MA-1 element' doesn't continue in modern East Asians, Siberians and extreme NE Asians (Chukchees) as much as it does in modern Amerindians, we have to treat it as an 'Amerindian' element in an LGM-dated South Siberian population".

Isn't it far more likely that the East Asian element in Siberia and northwestern NOrth America the result of later expansion of a 'purer' East Asian population?

Tobus said...

@German:
Sometimes it's good to look at the data when you're struggling to make a choice between 3 seemingly equally plausible options.

Absolutely... now can you show me LB having increased affinity with Amerindians on any other graph other than the one we're trying to interpret?

Direct measurement with Karitiana always creates a rightward pull for nearly all Eurasian populations.

Direct measurement with Karitiana confirms that LB has no more "pull" to Amerindians than modern Europeans do.

MA-1 does the same thing and the pull is even stronger.

You are just restating your "pull to MA-1 is pull to Amerindian" assumption. You yourself acknowledge that "genomewide, MA-1 is majority West Eurasian and minority Amerindian." so I ask again, by what reasoning are you making the assumption that LB's affinity with MA-1 *must be* due to MA-1 having a small amount of Amerindian affinity and cannot be due to MA-1's predominately non-Amerindian affinity?

When compared to MA-1, LB is Amerindian.

When compared to Karitiana, LB is European, C/S Asian and Papuan. When compared to Han, LB is European and C/S Asian... when compared to LB, Mal'ta is European, when compared to Karitiana, MA-1 is C/S Asian. All this "when compared to A, X is Y" nonsense doesn't get us anywhere, because it's an invalid interpretation of the data. A similar f3 score on a single isolated axis that isn't replicated elsewhere doesn't prove that the two populations are related, it's just a chance correlation.

If we want to know if LB has more affinity to Amerindians than modern Europeans do, all we need to do is look at the Karitiana axis - the direct measurement of what we want to know.

German Dziebel said...

@Tobus

"now can you show me LB having increased affinity with Amerindians on any other graph other than the one we're trying to interpret?"

I can show you only what the data shows me. And the data is unambiguous that all Eurasians are shifted toward modern Amerindians, West Eurasians are closer to modern Amerindians than East Asians if an ancient Amerindian-derived sample is used and that LB falls within the modern Amerindian variation when an ancient Amerindian-derived sample is used. It's a perfectly meaningful picture. You can keep racking your brains around why LB doesn't show a special pull toward modern Amerindians but this is just your obsessive pet peeve.

"Direct measurement with Karitiana confirms that LB has no more "pull" to Amerindians than modern Europeans do."

Yes, they all show a pull toward Amerindians confirming that LB and modern West Eurasians have Amerindian descent. You are stubbornly denying it in an effort to preserve your pseudoscientific bias. MA-1 has more Amerindian descent than LB as MA-1 is closer to Amerindians on all the charts. LB falls into the modern Amerindian cluster when we look at Eurasia through the MA-1 lens. What's your problem?

"When compared to Karitiana, LB is European, C/S Asian and Papuan. When compared to Han, LB is European and C/S Asian... when compared to LB, Mal'ta is European, when compared to Karitiana, MA-1 is C/S Asian. All this "when compared to A, X is Y" nonsense doesn't get us anywhere..."

You are a plain creationist bigot. Scientific data doesn't take you anywhere? All the patterns you've identified are perfectly natural. Compared to Karitiana, LB, CS Asian and Papuan come closer to each other. Compared to Han, Europeans and C/S Asians come closer to each other. MA-1 is everywhere pulled toward Karitiana. West Eurasians are pulled toward Amerindians compared to East Asians. These are just the various iterations of the "eastern" pull observed in West Eurasia. The eastern pull peaks in Amerindians.

@terryT

"I thought you had consistently claimed MA-1 was mainly Amerindian. "

Yes, F3 statistic in Raghavan shows that MA-1 is closer to Amerindians than to any other population. What's going on here is that the Amerindian component in MA-1 is likely older than the West Eurasian one, although the latter has become more dominant over the years. This means that MA-1 is a smaller Amerinndian population that absorbed a lot of West Eurasian genes. There are plenty of those in the contemporary U.S.

"Isn't it far more likely that the East Asian element in Siberia and northwestern NOrth America the result of later expansion of a 'purer' East Asian population?"

"Unlikely" is not even the right word here. It's plain nonsense. I went over all the evidence and logic with you already multiple times. Your brain is just not capable of retaining information. The real question is whether there was a sizable migration of Amerindians to East Asia 12,000 years ago or the split between Amerindians and East Asians happened once some 50,000 years. A reverse migration of East Asians to the New World 10,000 years ago or later is a myth.

terryt said...

"Alternatively, the whole mtDNA tree is full of problems, so it may turn out that mtDNA U lineages are in fact a subset of B lineages. Remember that the whole R11'B6 clade is defined by a supposedly 'reverse' mutation T16189C!"

I though you claimed I knew nothing about mitochodrial DNA yet you you are showing your ignorance. It is not just the R11'B6 clade that is defined by 16189C. The mutation has the effect of combing three haplogroups, R11'B6, B4'5 and R24. I am more than happy to accept we are dealing with three separate R-Derived haplogroups here. As a result you need to explain which of these three R24'B groups you are suggesting U is a subset of. As the phylogeny stands U can be though of as simply being a fourth R-derived haplogroup.

And even if it can be shown U is a subset of one of the three B'24 haplogroups we are still left a long way from America. All three B subset look almost certain to have first appeared somewhere near South China/Southeast Asia, nowhere near America. U can in no way be shown to descend from any America haplogroup.

"So this is what happened with MA-1 - it's original Amerindian lineages were replaced with West Eurasian lineages. A very natural explanation, it seems".

Except that you have not proposed any possible original Amerindian lineage that could have been replaced. It seem s that as we investigate your claim further and further we have more and more difficulty discerning exactly what you believe. What is it you are actually claiming?

terryt said...

"F3 statistic in Raghavan shows that MA-1 is closer to Amerindians than to any other population".

How do you reconcile that with: 'Genomewide, MA-1 is majority West Eurasian and minority Amerindian'. In fact:

"the Amerindian component in MA-1 is likely older than the West Eurasian one"

Surely that can only be so if the West Eurasian component had a completely separate origin to the Amerindian componenet. In other words it does not descend from any out of America population. So I'm left still pondering what it is exactly you are proposing originally emerged from America: an Upper Paleolithic population or an earlier population. You have certainly eliminated any population that forms the basis of the human species. In fact I remember you saying in effect that the Australian Aborigines and Papuans do not have Amerindian ancestry. And you have claimed many times that Africans definitely do not have Amerindian ancestry. So we have multiple populations spread around the world. I'm very much inclined to agree with that idea but we're left with the problem of what exactly are you proposing emerged from America?

"The real question is whether there was a sizable migration of Amerindians to East Asia 12,000 years ago or the split between Amerindians and East Asians happened once some 50,000 years".

We have no evidence at all for the latter proposition, and very litttle for the first.

"they all show a pull toward Amerindians confirming that LB and modern West Eurasians have Amerindian descent"

They all have a common origin, not necessarily in America.

"West Eurasians are pulled toward Amerindians compared to East Asians"

Yes, West Eurasians have very little East Eurasian ancestry, unlike Amerindians.

Tobus said...

I can show you only what the data shows me. And the data is unambiguous that all Eurasians are shifted toward modern Amerindians
You can keep racking your brains around why LB doesn't show a special pull toward modern Amerindians

So, you finally accept that LB doesn't show a special pull to Amerindians in this data? The only thing I'm racking my brain about is why it took you 6 weeks to see what I pointed out to you on day 1.

Scientific data doesn't take you anywhere?

It's not the scientific data that's the problem - it's your unscientific misinterpretation of the scientific data. Take some time to understand the maths behind the f3 stat and thus what these plots actually represent - the relationships you are concluding from isolated correlations are well beyond the scope of what the f3 stat actually measures. The data is fine - it's invalid and unconfirmed conclusions that take us nowhere.

German Dziebel said...

@Tobus

"So, you finally accept that LB doesn't show a special pull to Amerindians in this data? The only thing I'm racking my brain about is why it took you 6 weeks to see what I pointed out to you on day 1."

Your comprehension of written stimuli is that of a pre-schooler. All the EDF graphs in Ollade show that all Eurasians are pulled toward Amerindians, and West Eurasians, including LB, are pulled toward Amerindians more than a number of other Eurasians, e.g., West Asians. One of the graphs shows that LB is a subset of Amerindians if an ancient Amerindian-derived ancient sample is used as an anchor. Why does it take you so long to accept these irrefutable facts? The answer is simple - you are a pseudoscientific bigot and a troll.

"Take some time to understand the maths behind the f3 stat and thus what these plots actually represent - the relationships you are concluding from isolated correlations are well beyond the scope of what the f3 stat actually measures."

I understand them perfectly. It's you who somehow imagines that in each trilateral comparison f3 measures a different thing - sometimes it's actual common genetic history, sometimes it's something else. If West Eurasians, East Asians and Amerindians form clusters, it's common genetic history, but if LB falls into an Amerindian cluster it's noise in the data. This is patently flawed and betrays your creationist strategy of questioning the integrity of scientific evidence for the sake of the triumph of your abstract and bookish "knowledge."

Tobus said...

@German:
All the EDF graphs in Ollade show that all Eurasians are pulled toward Amerindians, and West Eurasians, including LB, are pulled toward Amerindians more than a number of other Eurasians, e.g., West Asians.

So do you, or do you not, agree that this data shows La Brana has an affinity to Amerindians that is within the range of modern Europeans?

I understand them perfectly. It's you who somehow imagines that in each trilateral comparison f3 measures a different thing - sometimes it's actual common genetic history, sometimes it's something else.

No you don't understand at all - outgroup f3 stats always measure the exact same thing, it's just a different measurement to what you think it is and so you are getting confused. The f3 stat is a mathematical formula that calculates affinity between populations A and X relative to an outgroup. It NEVER measures common genetic history between the various X populations - in fact it CANNOT do such a thing because the mathematics simply don't allow for it. Common genetic history is a subjective interpretation of the data that can be made IF and ONLY IF other data supports such a conclusion. In the case of populations in similar geographical and temporal regions that cluster on every axis, such a conclusion is warranted. In instances where there is no geographical or temporal correlation and there is only a common "cluster" on a single axis, we need corroboration from other data. In the case of LB the conclusion you are drawing is explicitly rejected by direct measurement on the Karitiana axis, so it's an invalid conclusion.

Your problem stems from confusing what the data is actually measuring (pop X is this far from pop A) with a possible interpretation when scores are similar (pop X and Y both share the same genes with pop A) - one is fact and one is theory, but you seem to think they are the same thing. The facts presented by the f3 stats are the same in every case, but the interpretation of these facts is subject to consistency with other facts - so a conclusion that Russians and French share common ancestry with Sardinians is OK (because it's consistent with all other data, here and elsewhere), but a conclusion that LB and Amerindians share common ancestry with MA-1 isn't (because it's inconsistent with the Karitiana axis). Do you see the difference?

German Dziebel said...

@Tobus

I thought you can't get any more obtuse. But it turns out you can!

On the one hand, you arguing that "this data shows La Brana has an affinity to Amerindians that is within the range of modern Europeans." On the other hand, you're doubting my conclusion that "LB and Amerindians share common ancestry with MA-1."

So, now your reading of mathematics is telling you that West Eurasians and LB share genetic history with Amerindians, MA-1 shares genetic history with LB and modern West Eurasians, Amerindians share genetic history with MA-1 but then lo and behold Amerindians don't share genetic history with LB?

A chimp would do a better job with math!

So here's the correct interpretation of the data: 1. Direct measurement with modern Karitiana shows that all Eurasians, ancient and modern, are shifted toward Amerindians. LB as an ancient, 7,000 year old West Eurasian sample is no different. (It could've been shifted more toward Yoruba, which we would expect under out-of-Africa but it's not.) 2. MA-1 shows the greatest proximity to Amerindians out of all West Eurasian populations followed by LB and other ancient West European samples. 3. MA-1, which is located in East Asia in close proximity to the New World, is genetically closer to LB than to modern West Eurasians. 4. But MA-1's strongest relationship is with modern and ancient Amerindians due to it's possessing a genome-wide "component" most strongly expressed in modern South Amerindians. 5. Mesolithic Europeans such as LB fall within the modern Amerindian spectrum of variation compared with MA-1 demonstrating that they are closer to Amerindians than modern West Eurasians.

Tobus said...

@German:
On the one hand, you arguing that "this data shows La Brana has an affinity to Amerindians that is within the range of modern Europeans." On the other hand, you're doubting my conclusion that "LB and Amerindians share common ancestry with MA-1."

My argument has always been that LB has no increased affinity to Amerindians realtive to modern Europeans - it seems you are now agreeing with this point: LB has the same level of affinity to Amerindians that modern Europeans do, yes?

So here's the correct interpretation of the data:

Let's see..:
1. Agreed (although we'd only expect a "shift to Yoruba" if LB diverged from Karitiana before other Europeans)
2. Agreed (assuming if "followed by LB and other ancient West European samples" means LB and other West European samples as a group - if you mean followed by LB then followed by other West Europeans then you are wrong)
3. Agreed
4. Agreed (although who has the "strongest relationship" depends on which data you use - the ADMIXTURE/PCA show a different picture)
5. I agree with the first part restating the data ("Mesolithic Europeans such as LB fall within the modern Amerindian spectrum of variation compared with MA-1"), but the conclusion you draw from that data is not correct ("demonstrating that they are closer to Amerindians than modern West Eurasians.") - affinity to MA-1 is not the same as affinity to Amerindians.

All in all, pretty good - you just fall at the last hurdle. What is your reasoning behind assuming that affinity to MA-1 must be due to the small part of Amerindian-like DNA he has and not due to his much larger non-Amerindian components?

Remember your own point 1. - when measured directly against Karitiana LB shows no difference to modern European samples. That's contradictory to your conclusion in point 5.

German Dziebel said...

@Tobus

"My argument has always been that LB has no increased affinity to Amerindians realtive to modern Europeans - it seems you are now agreeing with this point: LB has the same level of affinity to Amerindians that modern Europeans do, yes?"

No, Olalde EDF 5d clearly shows that LB is closer to Amerindians than other West Eurasians. See also below.

"(although we'd only expect a "shift to Yoruba" if LB diverged from Karitiana before other Europeans)"

So your twisted logic leaves open the possibility that a 7,000 year old sample diverged from Karitiana after modern Brits.

"(assuming if "followed by LB and other ancient West European samples" means LB and other West European samples as a group - if you mean followed by LB then followed by other West Europeans then you are wrong) ."

We don't have evidence how all ancient West Eurasian samples stack up against Amerindians. So, yes, as a group.

"I agree with the first part restating the data ("Mesolithic Europeans such as LB fall within the modern Amerindian spectrum of variation compared with MA-1")."

Halleluja! Mind you, though, that TerryT will excommunicate you.

"the conclusion you draw from that data is not correct ("demonstrating that they are closer to Amerindians than modern West Eurasians.") - affinity to MA-1 is not the same as affinity to Amerindians."

Raghavan et al. have a nice graph showing that Amerindians are the closest to MA-1 compared to other populations, so the conclusion MA-1, LB and Amerindian share the same ancestry is correct. All other f3 charts support this conclusion. If you want to prove that it's not, you need to marshall some evidence.

"What is your reasoning behind assuming that affinity to MA-1 must be due to the small part of Amerindian-like DNA he has and not due to his much larger non-Amerindian components?"

Apart from what I wrote above and throughout, the fact that a) LB falls into the Amerindian cluster on the MA-1 axis and b) direct measurement with Karitiana shows that all Eurasians are shifted toward Amerindians proves that the affinity between MA-1 and LB is due to Amerindian and not West Eurasian ancestry. Especially so, since MA-1 is in close geographic proximity to Amerindians.

"Remember your own point 1. - when measured directly against Karitiana LB shows no difference to modern European samples. That's contradictory to your conclusion in point 5."

No contradiction here. 5 is a logical subset of 1. All Eurasians are shifted toward Amerindians in direct measurements. An ancient West Eurasian sample, which falls with modern West Eurasian variation (and not with Yoruba or Yoruba-shifted West Asians), shows special proximity to Amerindians when an even older Amerindian-derived sample from East Asia is used. Compare the relationship between MA-1 and modern East Asians vs. LB vs. modern West Eurasians. MA-1 doesn't fall with modern East Asians. Instead it falls with modern Amerindians. LB falls with modern West Eurasians and doesn't fall with Yoruba. The LB pull toward Amerindians, as extreme non-Yorubas, is indirect in EDF 5a and direct in EDF 5d, when ancient Amerindian-derived sample is used. EDF 5a captures more of a macro Eurasian-Amerindian perspective, while EDF 5d drills down to the precise affinity of a West Eurasian ancient sample. It's a perfect picture illustrating out-of-America.

Tobus said...

@German:
No, Olalde EDF 5d clearly shows that LB is closer to Amerindians than other West Eurasians.

*sigh* EDF 5d measures affinity to MA-1, not Amerindians. It's graph 5a that measures LB's affinity to Amerindians.

Raghavan et al. have a nice graph showing that Amerindians are the closest to MA-1 compared to other populations, so the conclusion MA-1, LB and Amerindian share the same ancestry is correct

Whatever ancestry they may all share, LB shows the same affinity to Amerindians as modern Europeans do.

a) LB falls into the Amerindian cluster on the MA-1 axis

A measurement of affinity to MA-1 is not a measurement of affinity to Amerindians.

b) direct measurement with Karitiana shows that all Eurasians are shifted toward Amerindians proves that the affinity between MA-1 and LB is due to Amerindian and not West Eurasian ancestry.

No, it proves the exact opposite. It proves that whatever draws LB closer to MA-1 in graph 5d does not also draw him closer to Amerindians in graph 5a. It can't be shared Amerindian drift that causes the shift in 5d, if it were we'd see it in 5a as well.

An ancient West Eurasian sample, which falls with modern West Eurasian variation ... shows special proximity to Amerindians when an even older Amerindian-derived sample from East Asia is used.

A measurement of "special proximity" to MA-1 is not a measurement of "special proximity" to Amerindians.


The LB pull toward Amerindians, as extreme non-Yorubas, is indirect in EDF 5a and direct in EDF 5d,

The Karitiana axis is a direct measurement of Amerindian affinity, you said so yourself above. Your spurious logic has now twisted the truth to the point that you have just reversed fact and fiction. I suggest you go back to basics, read graph EDF 5a, and confirm for yourself that LB's Amerindian affinity is within the range of modern Europeans.


German Dziebel said...

@Tobus

" EDF 5d measures affinity to MA-1, not Amerindians. It's graph 5a that measures LB's affinity to Amerindians."

It measures both. MA-1 is an Amerindian-like population. Keep denying the evidence!

"No, it proves the exact opposite. It proves that whatever draws LB closer to MA-1 in graph 5d does not also draw him closer to Amerindians in graph 5a. It can't be shared Amerindian drift that causes the shift in 5d, if it were we'd see it in 5a as well."

It does - all of modern Eurasians are shifted toward modern Amerindians in EDF 5a. It means all of them have Amerindian ancestry. If it weren't for Amerindians, an ancient sample like LB would have been in the range of West Asians, or somewhere in the 0.14 range on the X axis in 5a. So 5a shows that LB has more Amerindian ancestry than modern West Asians and in 5d we see that LB has more of that ancestry than modern West Europeans. All consistent with time, geography and a founding Amerindian migration.

"I suggest you go back to basics, read graph EDF 5a, and confirm for yourself that LB's Amerindian affinity is within the range of modern Europeans. "

And EDF 5d shows that LB falls within the range of modern Amerindians. This is exactly what we'd expect from a hybrid population - it's part of one cluster along one axis and part of another cluster along the other axis.

Tobus said...

@German:
It measures both. MA-1 is an Amerindian-like population

No, it only measures affinity to MA-1 - that's hardcoded in the maths. You yourself agree that MA-1 is much more non-"Amerindian-like" than he is "Amerindian-like" so an automatic assumption of Amerindian affinity from the MA-1 axis is unfounded.

It does - all of modern Eurasians are shifted toward modern Amerindians in EDF 5a. It means all of them have Amerindian ancestry.

Outgroup f3s don't measure ancestry from the subject population. They measure shared drift, in this case with Amerindians since divergence from the Yoruba branch.

I'm also not sure what "shift" you are talking about - in 5a Europeans (and LB) show the same affinity to Amerindians as Central/South Asians, and less than Papuans and East Asians. The only population with consistently lower scores are the West Asians, who are known to have recent African-admixture. If anything, Europeans here represent the baseline that other shifts are relative to.

So 5a shows that LB has more Amerindian ancestry than modern West Asians and in 5d we see that LB has more of that ancestry than modern West Europeans.

5a and 5d don't use the same subject populations - the "ancestry" (really shared drift) measured in 5a is not the same as the one measured in 5d.

And EDF 5d shows that LB falls within the range of modern Amerindians

Only in terms of affinity with MA-1. The point we are arguing is affinity with Amerindians, in which case graph 5a is the measurement we want.

It's plain to see that LB's affinity with Amerindians is within the range that modern Europeans show - hence he does not have any increased or special affinity with Amerindians.



German Dziebel said...

@Tobus

"Outgroup f3s don't measure ancestry from the subject population. They measure shared drift, in this case with Amerindians since divergence from the Yoruba branch."

In the other string, you are arguing that f3 "estimates ancestry." Should we keep just one string alive to help you avoid schizophrenic comments?

"No, it only measures affinity to MA-1 - that's hardcoded in the maths."

Where do you get your bizarre ideas? Where does it say that plots showing a whole bunch of human populations actually should be ignored as estimates of the degree of mutual affinity between those plotted populations? You should recommend labs to actually publish only a plot with one population in it and no values against the axes because the laxity of showing multiple populations severely violates Tobus's "maths."

"I'm also not sure what "shift" you are talking about."

You should've said so from the very beginning! Considering that you yourself acknowledged that all West Eurasians are equally Amerindian-shifted, I thought you could read the charts for yourself. Don't hesitate to ask questions if you're lost. I don't bite! The shift can be seen when you observe that all the Eurasian populations, to a varying degree, are shifted away from 0 on the very left of the x axis toward Karitiana on the very right. If there were no Amerindian shift, all the Eurasian populations would've been nicely lined up in the area of 0 on the x axis (this means more African shifted) and Amerindians wouldn't have been so far to the right. This is consistent with Rosenberg et al. 2002 that identified not one ancestral component, African, but two African and Amerindian that are found in variable degrees across the world populations. It's just ancient DNA from Eurasia shows more Amerindian affinity than African.

"Only in terms of affinity with MA-1. The point we are arguing is affinity with Amerindians, in which case graph 5a is the measurement we want."

You asked for it, Tobus! So, let's review what anchoring it in Karitiana gives us. EDF 5a shows an all-Eurasian shift toward modern Amerindians. Although LB is an ancient sample, it falls with modern West Europeans and not with modern West Asians. This means it has a stronger Karitiana pull than West Asians. Modern West Europeans are derived from a population that's most closely related to LB and they stay in the general "area" of LB. MA-1 is further shifted toward Karitiana than LB and modern West Europeans. MA-1 nearly falls into the East Asian cluster on Karitiana-anchored EDF 5a and 5e. At any rate, it's closer to East Asians than to West Europeans. It's also East Asian geographically. So, modern East Asians are derived from a population most closely related to MA-1. This disproves Raghavan's conclusion that MA-1 and East Asians are not proximate. When directly compared to Karitiana, they are!

If EDF 5a gives us a macro-Eurasian perspective, EDF 5d drills down to the specific affinities of LB. LB can't be derivative of modern European populations simply because ancients are not derived from moderns. Agree? EDF 5d shows that LB is pulled out of the modern West European cluster and falls with modern Amerindians when directly compared with MA-1. This confirms the special proximity between LB, MA-1 and Amerindians and the origin of both modern East Asians and modern West Europeans from an Amerindian-like population.

Tobus said...

@German:
In the other string, you are arguing that f3 "estimates ancestry."

No, in the other thread I said that ADMIXTURE "estimates ancestry", not f3 stats - try to pay attention.

Where do you get your bizarre ideas?

From reading the sources that created the f3 stats and a solid understanding of high-level maths. It's explained fairly well in the Mal'ta SI - f3(A, X; O) measures shared drift between A and X since divergence from O. If two populations get similar scores, there's no guarantee (or even indication) in the f3 stats that it's from the same cause.

This seems to be the main sticking point, so I'll present a simple analogy: If Alice lives 10km from the station and Bob lives 10km from the station, how far does Alice live from Bob?

If there were no Amerindian shift, all the Eurasian populations would've been nicely lined up in the area of 0 on the x axis (this means more African shifted) and Amerindians wouldn't have been so far to the right.

I'm glad I asked, you clearly have some misconceptions about the f3 stats. A score of zero (which isn't where the graph starts BTW) wouldn't indicate any affinity with Yoruba - these stats measure shared drift since divergence with Yoruba, so a score of zero would mean no affinity to Yoruba OR Karitiana (only a negative score would indicate affinity with Yoruba). What you call an "Amerindian shift" really just means a period of time when the two populations were the same after their ancestors diverged from Yoruba - if you look at the Sardinian axis you can see the same effect, but I wager you're not willing to call it a "European shift".

MA-1 nearly falls into the East Asian cluster on Karitiana-anchored EDF 5a and 5e. At any rate, it's closer to East Asians than to West Europeans

You're doing it again - assuming a genetic relationship between populations based solely on their f3 stats with a third population. MA-1 is not closer to East Asians than to West Europeans. You know this, but you're flawed understanding of f3 is leading you further and further into lala land.

So, modern East Asians are derived from a population most closely related to MA-1

.. and there it is, a patently ridiculous statment contradicted by all other available data including the other f3 plots that directly measure the relationship. How do you reconcile your "MA-1 is closest to East Asians" conclusion against the MA-1 axis in EDF 5c/d and the Han axis in EDF 5b? Ignore the concrete direct measurements in favour of your unfounded speculations?

This confirms the special proximity between LB, MA-1 and Amerindians..

No, just between LB and MA-1... the "Amerindian" bit is speculation on your part, and explicitly ruled out by 5a.

Tobus said...

@German:
Yes, it's a general belief. A mistaken belief, I would add.

Yes, you would, but as we've ssen you wouldn't be able to provide an alternative model that works with the known genetic data.

What is it that you want to discuss in the main body of the paper that suggests that f3 statistics works only for one population at a time?

I don't know - the fact that each f3 stat only measures one population at a time against the axis maybe? According to the paper they "computed an ‘outgroup f3-statistic’, where the deviation from 0 will be a function of the shared genetic history of two populations A and B in their unrooted history with the outgroup O".. where do you think it says anything about f3 measuring distance to a population not even included in the formula?

I've noticed that you like to skip facts because they seem like nonsense to you.

Unfortunately you obscure any relevant facts in your posts with liberal amounts of personal attacks, spurious logic, subjective opinions and the occasional conspiracy theory. If there is a genuine fact with actual relevance to the discussion that I've skipped over, please feel free to restate it - minus the superfluous posturing.

Go to Rasmussen et al. EDF 6. Take any chart whatsoever...Report back to me about what you found.

I found that none of Y-axes actually measure distance between Europans and East Asians, they only measure the distance of various populations to Anzick and modern Amerindian samples. In each case East Asians are closer to the Amerindian subject than Europeans are.

The first two set of X-axes however were much more informative, directly measuring distances to East Asians and Europeans respectively. In every case Amerindians were closer to the subject population than the East Asian or European sample being compared, confirming the relationship diagram you agreed to above (WEu-------SAm-NAm---EAs). East Asians are indeed closer to Amerindians than they are to Europeans... but we both knew that already, didn't we?

I honestly don't know how what you're saying here is relevant.

It's not, you just said something stupid so I corrected you... it appears however that you really do believe there is some kind of GPS-sensor in DNA that allows it to mutate directly relative to distance... it's not relevant though, so let's drop it.

German Dziebel said...

@Tobus

" If two populations get similar scores, there's no guarantee (or even indication) in the f3 stats that it's from the same cause."

If you got it from the " sources that created the f3 stats", please provide an actual quote. Also, what other causes besides ancestry/admixture do you have in mind? Divine intervention?

"a solid understanding of high-level maths... If Alice lives 10km from the station and Bob lives 10km from the station, how far does Alice live from Bob."

You should find a golden middle between "high maths" and "kitchen logic." F3 stats correctly places all modern West Eurasians into a cluster, all Amerindians into a cluster, all East Asians into a cluster (they all live the same no. of km from the "station" and they all live in their respective "same place, or Alice is 11 km away from the station and Bob is 10 km from the station and Alice and Bob are close neighbors"). Why do you need your fantasy examples if you already have a real geographic landscape baked into the plots in question?

"A score of zero (which isn't where the graph starts BTW) wouldn't indicate any affinity with Yoruba - these stats measure shared drift since divergence with Yoruba, so a score of zero would mean no affinity to Yoruba OR Karitiana (only a negative score would indicate affinity with Yoruba)."

Again, look at any chart and you'll see that both axes, which "don't even start with zero," are anchored in Yoruba. It's a continuum - more Karitiana means less Yoruba and the other way around. Look again at Rosenberg et al. 2002 and you'll see that East Asians are more Amerindian and less African than West Eurasians that are more African and less Amerindian.

"What you call an "Amerindian shift" really just means a period of time when the two populations were the same after their ancestors diverged from Yoruba."

Tobus is again en route from high maths back to ideology. An Amerindian shift means Amerindian ancestry or admixture. This does mean that two populations - Amerindian and another one - were one New World population at some point. Yoruba likely picked up an African hominin substrate, which made them divergent from a Eurasian base, not the other way around.

"if you look at the Sardinian axis you can see the same effect, but I wager you're not willing to call it a "European shift"."

Why not? It is a "European shift." Notably, It's less dramatic than the Amerindian shift. Also of note is the fact that, when compared with MA-1, modern Amerindians are not "European shifted"; it's Europeans who are Amerindian-shifted.

"You're doing it again - assuming a genetic relationship between populations based solely on their f3 stats with a third population. MA-1 is not closer to East Asians than to West Europeans."

Next to the real data, your assertions mean nothing. Look at EDF 5a and EDF 5e: MA-1 is closer to East Asians than to West Eurasians. You yourself have pushed us to focus only on the Karitiana-anchored plots. Be careful what you wish for: Raghavan made you believe that MA-1 doesn't have anything to do with East Asians. But lo and behold F3 data actually refutes it when Karitiana is used as an anchor.

"How do you reconcile your "MA-1 is closest to East Asians" conclusion against the MA-1 axis in EDF 5c/d and the Han axis in EDF 5b"

Easily! MA-1 experienced gene flow from a West Eurasian source, so it got pulled away from Amerindians and East Asians. The pull is pretty strong, which is consistent with ADMIXTURE results in which a minority of MA-1 ancestry is Amerindian. MA-1 is a hybrid population, remember?

terryt said...

" MA-1 experienced gene flow from a West Eurasian source"

Doesn't that immediately blow your 'out of America' model out of the water? You appear now to concede there were people outside America who have made a significant contribution to the modern human genetic makeup. Or is it that we are all still unsure what you and Alvah actually believe?

German Dziebel said...

@TerryT

"Doesn't that immediately blow your 'out of America' model out of the water? You appear now to concede there were people outside America who have made a significant contribution to the modern human genetic makeup."

What the f3 data is telling us in the very minimum is that MA-1 is a hybrid between Amerindians and West Eurasians. Just like LB. I would be fine if we stop here and table further conversations until more data comes in. This still means that there was an out-of-America migration to Eurasia prior to 24,000 years. This refutes Raghavan's nonsensical conclusion that Amerindians are a mix of West Eurasians and East Asians.

West Eurasian admixture into an older, in-situ Amerindian-derived population in South Siberia can easily explain why MA-1 is shifted away from East Asians in MA-anchored plots, while appearing closer to East Asians than to West Eurasians in Karitiana-anchored plots. East Asians evolved from an Amerindian-like population that hadn't experienced West Eurasian admixture. If we interpret f3 plots as good proxies for a history of population divergence, then West Eurasian "admixture" in MA-1 becomes incipient West Eurasian divergence from an Amerindian-like population in South Siberia (MA-1) (imagine a refugium in which an originally small Amerindian population developed into a proto-West Eurasian population and then expanded and spread across Europe), with modern East Asians diverging from another Amerindian-like population in post-Mal'ta times. Again, this would be consistent the fact that MA-1 is closer to East Asians than to West Eurasians on Karitiana-anchored plots, while being closer to modern West Eurasians in MA-anchored plots.

Tobus said...

@German:
Patterson (2012) explains the original f3 stat as an indicator of whether O is an admixture of the X and A populations ("The observation of a significantly negative value of f3(C; A,
B) is thus evidence of complex phylogeny in C."..."F3 is guaranteed to be positive if C is unadmixed"), and Rhaghavan (2013) explains the use of "outgroup f3" as an estimation of common ancestry between X and A since divergence from O (the result of f3(A, B; O) "will depend on how much genetic history is shared between populations A and B". Neither of these source say that a comparison of f(X, A; O) and f3(Y, A; O) can be used as a measurement of the distance or shared ancestry between X and Y, nor do they make any such interpreation in the discussion of their results. It's something you seem to have just made up yourself.

The maths also provide no method for measuring X/Y affinity from f3(X, A; O) and f(Y, A; O). The original f3 formulae and central determinant of the outgroup f3 (the other factors are solely based on O and affect the magnitude, but not the sign) is Sum((O-X)(O-A)) where each letter represents the allele frequency of each SNP in each population. Note that it will always be positive if X and A either both have less or both have more of the allele than O, and can only be negative if one has more and one has less than O. Note also that there is no direct comparison between X and A - the result is the product of their individual differences to O, a positive result can be obtained even if A and X have very different allele frequencies, so long as they are both more or both less frequent than in O.

So if two different X populations get the same f3 result against A/O this simply means that the sum of their "(O-X)" parts ends up at the same figure - but there is no requirement that they have to have the same O-X value (or even sign) at each individual allele, and thus can be quite unrelated to each other. Consider 2 alleles, both very rare in O and both at 50% in A, but one fixed in X and absent in Y and the other fixed in Y and absent in X (such as we might see if X and Y share different parts of A's DNA):

f3(X, A; O) will be (0.01-1)(0.01-0.5) + (0.01-0)(0.01-0.5) = 0.485
f3(Y, A; O) will be (0.01-0)(0.01-0.5) + (0.01-1)(0.01-0.5) = 0.485

This proves that two X populations can get the same f3 score without sharing any genetic affinity with each other. Your assumption of X/Y affinity based on similar f3 scores is just that - an assumption - with no inherent scientific, mathematical or logical validity. It's true that populations with shared ancestry will plot together, but it's just as true that not all populations that plot together *have to* have shared ancestry. The only reliable measurement is one directly against the relevant axis.

Tobus said...

@German (cont):

It's a continuum - more Karitiana means less Yoruba and the other way around.

You are completely wrong - f3(X, A; O) isn't measuring X on an "O to A" scale. It's measuring whether X and A have a signature of common post-O genetic history. All populations in these plots have significantly positive results indicating zero affinity to Yoruba, regardless of relative magnitude. Yoruba is an outgroup, it's purely a more/less Karitiana range.

An Amerindian shift means Amerindian ancestry or admixture
Why not? It is a "European shift."

.. and so means European ancestry or admixture?

Look at EDF 5a and EDF 5e: MA-1 is closer to East Asians than to West Eurasians.

Are we looking at the same graphs? MA-1 is higher than East Asians on both the Sardinian and La Brana axes, proving the exact opposite of what you are saying.

Easily! MA-1 experienced gene flow from a West Eurasian source, so it got pulled away from Amerindians and East Asians.

So is MA-1 closer to, or further from East Asians than Europeans? He can't be both, so make up your mind.

What the f3 data is telling us in the very minimum is that MA-1 is a hybrid between Amerindians and West Eurasians.

The f3 data tells us no such thing. Stop making things up.

MA-1 is closer to East Asians than to West Eurasians on Karitiana-anchored plots

"X is closer to Y when anchored in A" means absolutely nothing - the actual genetic distance between East Asians and Europeans doesn't change depending on what population is chosen for A. If you think some populations "come closer" when using an indirect f3 measurement, that's a sure sign that your interpretation of affinity from that measurement is false. Stick to the direct measurements and you won't be led astray.

German Dziebel said...

@Tobus

"Rhaghavan (2013) explains the use of "outgroup f3" as an estimation of common ancestry..."

You were supposed to find a direct quote supporting what's otherwise remains to be an absurd claim that f3 measures different things for different populations every time it's run and that all the plots that show all the populations in relation to each other shouldn't be used as representative of relative shared genetic history between the plotted populations. Instead, you are trying to prove something else? What is it?

Here's a very clear quote describing the meaning of the f3 stat with a practical conclusion for a number of populations: Native Americans vs. Siberians compared to MA-1:

"Since genetic drift in the lineage specific to the outgroup is expected to be constant regardless of which populations A and B are used (in the absence of gene flow), the remaining variation between statistics will depend on how much genetic history is shared between populations A and B. This statistic f3(outgroup;MA-1,population B) was quantified for a set of 147 worldwide populations (including the novel genotype data from this study) using ~300k SNPs where data was available for MA-1. The highest f3 -statistics were observed for Native Americans, followed by Siberian and northern European populations."

Take also Fig SI 26 in Raghavan. It says: "All western Eurasian populations are closer to Karitiana than to Han." Take any other chart and read its caption. It says the same thing. Repeat. Recalibrate your meds.

Raghavan's conclusion that gene flow went from MA-1 to Amerindians and not the other way around was justified by an appeal to the fact that East Asians and West Eurasians are not close enough to each other. They thought that this means that West Eurasians don't have the East Asian component. Here's for you a comparison across two f3 runs. Their oversight is that both West Eurasians and East Asians have an Amerindian component, which is why on a Karitiana-anchored axis they are close to each other and MA-1 is closer to East Asians than to West Eurasians.

"Your assumption of X/Y affinity based on similar f3 scores is just that - an assumption - with no inherent scientific, mathematical or logical validity."

Can you submit your complaint to Raghavan and other researchers because they don't have a problem thinking across f3 runs? I don't have any patience for your pseudomathematical rants.

"You are completely wrong - f3(X, A; O) isn't measuring X on an "O to A" scale. It's measuring whether X and A have a signature of common post-O genetic history. All populations in these plots have significantly positive results indicating zero affinity to Yoruba, regardless of relative magnitude. Yoruba is an outgroup, it's purely a more/less Karitiana range."

Do you have a quote to support your claim? The plots clearly refer to Yoruba vs. Sardinian, Yoruba vs. Karitiana, etc. scale. So, quite naturally Middle Easterners are the least Amerindian and the most Yoruba-like. This is consistent with all other non-f3 plots.

"Are we looking at the same graphs? MA-1 is higher than East Asians on both the Sardinian and La Brana axes, proving the exact opposite of what you are saying."

Same charts, different mental abilities. MA-1 is closer to East Asians than to West Eurasians on EDF 5a on both axes and it's closer to East Asians than to West Eurasians on the X axis on EDF 5e. In both cases, the X axis is Karitiana-anchored.

German Dziebel said...

"So is MA-1 closer to, or further from East Asians than Europeans? He can't be both, so make up your mind."

When compared with Karitiana, MA-1 is closer to East Asians than to West Eurasians. This proves that MA-1 and East Asians are Amerindian derived. Unlike MA-1, East Asians don't have a West Eurasian component, hence East Asians and West Eurasians are far apart when the plot is anchored in Han.

What you are failing to understand is that both "deep ancestry" and "admixture" are blended together in f3 plots. MA-1, LB, East Asians and Amerindians share "deep Amerindian ancestry." MA-1 and LB has a large amount of subsequent West Eurasian admixture that neither Amerindians, nor East Asians have. East Asians also have subsequent "Mongoloid" admixture that neither southern Amerindians, nor West Eurasians have. Northern Amerindians seem to have it, but since they are close to southern Amerindians, it's unlikely that northern Amerindians got it from East Asia. It's likely it was the other way around. But this late out-of-America admixture happened much later than MA-1.

"The f3 data tells us no such thing. Stop making things up."

It is. And your heckling won't change this fact. In EDF 5a, MA-1 is closer to Amerindians than to West Eurasians on the Y axis. It's shifted away from West Eurasians toward Amerindians on the Karitiana-anchored X axis. On EDF 5b, it's part of the West Eurasian cluster on the Han-anchored axis, while Amerindians are the closer to MA-1 than West Eurasians on MA-anchored axis in EDF c, d.

"If you think some populations "come closer" when using an indirect f3 measurement, that's a sure sign that your interpretation of affinity from that measurement is false. Stick to the direct measurements and you won't be led astray."

Nonsense.

terryt said...

"This refutes Raghavan's nonsensical conclusion that Amerindians are a mix of West Eurasians and East Asians".

No. It means the results would fit with either direction. Apart from you and Alvah most scientists involved in the subject (and I hesistate to call either of you 'scientists' or involved in the subject') accept the direction that fits all other data.

"West Eurasian admixture into an older, in-situ Amerindian-derived population in South Siberia can easily explain why MA-1 is shifted away from East Asians in MA-anchored plots"

MA-1 is not 'shifted away from East Asians'. It has no East Asian.

"imagine a refugium in which an originally small Amerindian population developed into a proto-West Eurasian population and then expanded and spread across Europe"

So where was the West Eurasian population that mixed into the 'Amerindian-derived population in South Siberia' hiding before it mixed? If such a popualtion existed that means not Eurasians are by no means descended from Amerindians.

terryt said...

Another inconsistency from German:

"East Asians also have subsequent 'Mongoloid' admixture that neither southern Amerindians, nor West Eurasians have. Northern Amerindians seem to have it, but since they are close to southern Amerindians, it's unlikely that northern Amerindians got it from East Asia".

If 'Northern Amerindians ... are close to southern Amerindians' it is extremely unlikely the Mongoloid element arose in America. So now we have at least three different populations that provided admixture to this hypothetical Amerindian population that moved into the wider world: African, Papuan/Ausralian, Mongoloid and West Eurasian. The amount of Amerindian element in modern populations keeps getting smaller and smaller. Where will it end? with no contribution?

Tobus said...

@German:
Instead, you are trying to prove something else? What is it?

What I proved was that two X populations can get the same f3 score in f3(X, A; O) yet not have any shared ancestry with each other. Hence a similar f3 score does not *always* have to indicate shared ancestry between the two X populations.

Here's a very clear quote describing the meaning of the f3 stat.. "the highest f3-statistics were observed for Native Americans, followed by Siberian and northern European populations"

The quoted comments are about how the 3 populations (scored on the MA-1 axis. Note that there is nothing said here about the affinity between the populations themselves.

Take also Fig SI 26 in Raghavan. It says: "All western Eurasian populations are closer to Karitiana than to Han."

Katitiana is the X axis, Han is the Y axis... again this quote is about direct measurements against the A populations, not about affinity between the various X populations.

Can you submit your complaint to Raghavan and other researchers because they don't have a problem thinking across f3 runs

Can you show me where you think they are doing this - the two examples provided already are using direct measurements against the A population, which is the same method I used to arrive at the "LB has no increased Amerindian affinity" conclusion from EDF 5a. What you are doing with EDF 5d is implying a direct causal correlation between two X populations with the same score, I can't find any instance of Rhagavan (or any other scientist) doing this... no surprise since it's not a valid way to interpret the data.

Do you have a quote to support your claim?

You just quoted Rhagavan saying "the remaining variation between statistics will depend on how much genetic history is shared between populations A and B" (note: no mention of distance from O, and distinctly different to "whether B has more affinity to A or to O" or similar phrasing)... plus I already quoted you the f3 formula that shows exactly what f3 measures as well as quotes detailing what positive and negative scores mean. You do realise they chose Yoruba on purpose because it's an outgroup with equal affinity to all populations: "Because the Yoruban individual is genetically an outgroup to all the individuals that we will choose as populations 1 and 2, then these outgroup f3 statistics are proportional to the amount of genetic drift shared between populations 1 and 2 since their divergence with the outgroup".

Tobus said...

@German (cont):
MA-1 is closer to East Asians than to West Eurasians on EDF 5a on both axes...

Neither axis in EDF 5a measures closeness to East Asians. You want the X-axis on 5b for that.

..and it's closer to East Asians than to West Eurasians on the X axis on EDF 5e

The X axis on EDF 5e doesn't measure closeness to East Asians. You want the X-axis on 5b for that.

When compared with Karitiana, MA-1 is closer to East Asians than to West Eurasians.

So you believe MA-1 is closer to East Asians to West Eurasians? Perhaps you want to read Rhagavan again.

What you are failing to understand is that both "deep ancestry" and "admixture" are blended together in f3 plots

I totally understand that - that's why I think LB's MA-1 affinity (from "deep ancestry") doesn't have any connection to Amerindians' MA-1 affinity (from admixture)... this is supported by EDF 5a which doesn't show any increased Amerindian affinity in LB.

What *you* fail to understand is the the calculation used to generate f3(X, A; O) and f3(Y, A; O) doesn't measure the distance between X and Y.

In EDF 5a, MA-1 is closer to Amerindians than to West Eurasians on the Y axis.

The Y axis in EDF 5a doesn't measure closeness to Amerindians. You've got this one totally back to front since the Y axis measures affinity to Sardinians (West Eurasian proxy) and MA-1 scores higher than Amerindians.

Nonsense.

Then prove it - show me how the absolute genetic distance between two samples can change when somebody somewhere does some maths and plots them on a graph. The variation you are seeing is not real, it's just a side effect of using an unsigned one-dimensional measurement to various points on the genetic spectrum. LB and Amerindians are the same distance away from MA-1, but in different directions.

German Dziebel said...

@Tobus

"What I proved was that two X populations can get the same f3 score in f3(X, A; O) yet not have any shared ancestry with each other. Hence a similar f3 score does not *always* have to indicate shared ancestry between the two X populations."

You don't understand what "proving" means. You need to show that, say, Russians may end up in the same cluster as Papuans in f3 plots. Since this never happens, f3 stats works well as a way to compare ancestry across populations.

"Note that there is nothing said here about the affinity between the populations themselves."

Well it does, Siberians, Amerindians and northern Europeans are mentioned as having varying degrees of the ancestral component. They are being compared with each other.

"Katitiana is the X axis, Han is the Y axis... again this quote is about direct measurements against the A populations, not about affinity between the various X populations."

The source I quoted says: "ALL western Eurasian populations are closer to Karitiana than to Han."

How many entities do you think the English word "all" covers? One? What's your native language, Tobus? Piraha?

"Can you show me where you think they are doing this?"

I think it is Piraha. I just gave you an example when Raghavan et al. use cross-f3 comparisons to justify their conclusion of the directionality of gene flow between MA-1 and Amerindians. You delete my answer and repeat your question.

" I can't find any instance of Rhagavan (or any other scientist) doing this... no surprise since it's not a valid way to interpret the data."

See above. You need to provide cautionary statements from Raghavan et al and other authors that say "don't interpret our synthetic plots to infer world population history because individual f3 runs don't compare with each other. They are all measuring different thing and the continental clusters that we arrived at are purely coincidental."

"You do realise they chose Yoruba on purpose because it's an outgroup with equal affinity to all populations: "Because the Yoruban individual is genetically an outgroup to all the individuals that we will choose as populations 1 and 2, then these outgroup f3 statistics are proportional to the amount of genetic drift shared between populations 1 and 2 since their divergence with the outgroup".

But when you compare all the f3 scores with each other and plot all the populations together, you can see that some populations are closer to the "outgroup" than the others. This is because they branched off first after the separation of the outgroup and stayed in the same general area or because the "outgroup" population later admixed with the "in-group" population. Geography becomes a predictive factor in both cases. Plus ancient samples represent populations that are closer to the outgroup because they are closer to them in time. MA-1 is a 24,000 year old sample but it is shifted away from West Eurasians but not in the direction of Yoruba but in the direction of Karitiana. This again speaks to the fact that Yoruba is not an unadmixed human population but a composite of Eurasians with an archaic African substrate. Hence, we don't find the "African component" in some of the most ancient Eurasian remains such as MA-1. Because it was absorbed into their descendants who entered Africa. Contrariwise, it's Amerindians, who are firmly stuck in the rightmost extreme of f3 plots and are the closest to MA-1 that behave like a true outgroup.

"Neither axis in EDF 5a measures closeness to East Asians. You want the X-axis on 5b for that....The X axis on EDF 5e doesn't measure closeness to East Asians. You want the X-axis on 5b for that."

Nonsense.

German Dziebel said...

"So you believe MA-1 is closer to East Asians to West Eurasians? Perhaps you want to read Rhagavan again"

When looked through the Karitiana lens, yes. Only when we assume that MA-1 is an unadmixed West Eurasian population that we conclude that it's unrelated to East Asians. But once MA-1 is seen as a mix of ancestral Amerindians and subsequent West Eurasians, it becomes naturally close to East Asians because East Asians have the same Amerindian component. The actual data refutes Raghavan's conclusion. This is exactly where a cultural mythology-driven, anti-Amerindian bias came in and derailed the normal scientific process.

"this is supported by EDF 5a which doesn't show any increased Amerindian affinity in LB."

What's in reality supported by EDF 5a is the hypothesis that all Eurasians are Amerindian-shifted including all the available ancient samples, LB and MA-1 specifically. LB is Amerindian-shifted and not Yoruba-shifted compared to West Eurasians, although they are 7,000 years older than modern West Eurasians.

"The Y axis in EDF 5a doesn't measure closeness to Amerindians. You've got this one totally back to front since the Y axis measures affinity to Sardinians (West Eurasian proxy) and MA-1 scores higher than Amerindians"

It's closer to Amerindians on the Y axis than to modern West Eurasians (Sardinians are modern West Eurasians). The fact that it's higher than Amerindians on the Sardinian anchored axis is additional evidence that Amerindians contributed genes to MA-1 and only after that MA-1 became "West Eurasian" in its genetic orientation.

"LB and Amerindians are the same distance away from MA-1, but in different directions."

What different directions? What are you talking about? MA-1 is more Amerindian, LB is less Amerindian. MA-1 is closer to the New World geographically, LB is farther away.

terryt said...

"The source I quoted says: 'ALL western Eurasian populations are closer to Karitiana than to Han.'"

Yes, because all West Eurasians share an element carried into Karitiana, and Amerindians in general, that is not present in East Asians. Such a scenario is impossible if both West Eurasians and East Asians descend from Amerindians. I would have thought that was the obvious conclusion and showed with certainty that Karitiana is the admixed population, not either West Eurasians or East Asians.

Tobus said...

@German:
You need to show that, say, Russians may end up in the same cluster as Papuans in f3 plots

EDF 5a X-axis. EDF 5e X-axis.

Despite me being able to show what you asked for, your insistence that I need to show any particular case scenario that you cherry-pick is not a logical refutation of my point. All I need is a single case between any two populations to prove my point is valid, and there are plenty of these.

Siberians, Amerindians and northern Europeans are mentioned as having varying degrees of the ancestral component.

Exactly! The ancestral component is the axis, and all the discussion is how much of it they have. There is no discussion of the plot showing close they are to each other, just how close they are to the A population.

How many entities do you think the English word "all" covers?

In this case it means all the samples they have data for (close to 20 from a quick manual count). But what's your point - they are talking about these populations' distances to the axis, not the distances between these populations. See the black diagonal line at Y = X? This represents an equal score on both axes - they are saying all their samples score higher on the X axis than on the Y axis, nothing about genetic distance between their samples.

I just gave you an example when Raghavan et al. use cross-f3 comparisons to justify their conclusion of the directionality of gene flow between MA-1 and Amerindians. You delete my answer and repeat your question.
.. I assume referring to "Raghavan's conclusion that gene flow went from MA-1 to Amerindians and not the other way around was justified by an appeal to the fact that East Asians and West Eurasians are not close enough to each other.

Rahgavan's conclusion was based on D-stats, not f3 runs (see SI 14.6 section). It was an obvious attempt by you to avoid the question, so yes I ignored it and asked the question again (which again you've failed to answer).

You need to provide cautionary statements from Raghavan et al and other authors that say "You need to provide cautionary statements from Raghavan et al and other authors that say "don't interpret our synthetic plots to infer world population history"

I've provided quotes saying what f3 *can* do, what it *does* measure and how it *is* used, but you want a quote saying how it *can't* be used, what it *doesn't* measure and how it *isn't* used. You are shifting the burden of proof - insisting that I prove your claim is false instead of you proving that it's true. I've proved that f3 measures shared drift in populations X and A since divergence from O. If you want to claim that f3's on different X populations can be used measure the distance between the X populations then you're going to have to prove how this is possible. The creators don't say it can do it, the mathematics doesn't allow for it, and there's no examples of it in any published papers - it sees you just made it up.

But when you compare all the f3 scores with each other and plot all the populations together, you can see that some populations are closer to the "outgroup" than the others.

Please read the source material again - an affinity to the outgroup is shown by a statistically significant negative score, and which of the two populations has this affinity is not measured. You're making the same correlation/causation mistake again - yes, populations with Yoruba affinity will get a lower score (it would mask part of the shared drift), but not every population that is shifted left *has to* have Yoruba affinity (a population which diverged from Yoruba at the same time as the A population will have no shared drift with A and thus score 0).

Tobus said...

@German (cont):
Nonsense.

Please provide an explanation of how (O-A)(O-X) on either axis in graph EDF 5a measures affinity to East Asians.

When looked through the Karitiana lens, yes.

Can you explain in concrete (ie mathematical and allele frequency terms) exactly what you mean by "the Karitiana lens"? I suspect if you actually understood how this "lens" is formulated you wouldn't be placing so much weight on how it makes the non-Karitiana populations look in relation to each other.

The actual data refutes Raghavan's conclusion. This is exactly where a cultural mythology-driven, anti-Amerindian bias came in and derailed the normal scientific process.

Unless of course you can't prove your "f3 measures distance between X populations" theory has any basis - then the opposite bias could be said to be responsible.


What's in reality supported by EDF 5a is the hypothesis that all Eurasians are Amerindian-shifted including all the available ancient samples, LB and MA-1 specifically

And as this whole discussion is supposed to be about, that the LB "shift" is within the range as the modern European "shift"... agreed?

LB is Amerindian-shifted and not Yoruba-shifted compared to West Eurasians

Huh? Where is this in EDF 5a? LB be shows no shift relative to Europeans.


What different directions? What are you talking about?

Look at the PCA in Rhagavan SI 10 - see how MA-1 is not one of the extremes, that he's somewhere in the middle? If we get a ruler and measure distance to MA-1, can you see that a score of "10" can be "10 to the left" or "10 to the right"? This is the direction I'm talking about.

Now f3 stats measure something different to PCA so the relative positions are going to be different, but in terms of this left/right direction f3 simply ignores it and makes the subject population the extreme right in all cases. If you imagine drawing a line vertically through the entire PCA graph at MA-1's position on the X-axis, and then folding at this line so the right-hand side comes over and superimposes on the left hand side, you get a rough visual approximation of what the f3 plot is doing to get MA-1 as the right-most extreme - it's only presenting the raw distances and not the direction. I think you can see how populations that are in reality quite far apart on the PCA have now "become closer" due to the right-hand side being superimposed on the left-hand side. Populations on opposite sides may now even be on top of each other!

So now go back to EDF 5d and "unfold" it in your mind, taking the populations on the right of MA-1 in the PCA over to the right of MA-1 in the f3. You can now see how Amerindians and LB only cluster due to the lack of direction in f3 - in reality they are the same distance from MA-1 but in different directions... this is what I am talking about.

terryt said...

From the other post German has been contributing to:

"surely you didn't just dress up like an American Indian and traipse around Russia for a few years?"

What a brilliant discovery. It explains everything.

German Dziebel said...

@Tobus

"Despite me being able to show what you asked for, your insistence that I need to show any particular case scenario that you cherry-pick is not a logical refutation of my point. All I need is a single case between any two populations to prove my point is valid, and there are plenty of these."

Not a single one. All continental populations form their own clusters in f3.

"The ancestral component is the axis, and all the discussion is how much of it they have. There is no discussion of the plot showing close they are to each other, just how close they are to the A population."

Baloney. Their proximity of the 147 populations to each other is measured in terms of the degree of the expression of the ancestral component. Amerindians came out the closest to MA-1 followed by Siberians and Northern Europeans. It means that Northern Europeans are closer to Siberians who in turn are closer to Native Americans in reference to MA-1.

"In this case it means all the samples they have data for (close to 20 from a quick manual count). But what's your point - they are talking about these populations' distances to the axis, not the distances between these populations. "

You must be kidding me...

"Rahgavan's conclusion was based on D-stats, not f3 runs (see SI 14.6 section)."

Now you're going to argue that D tests cannot be compared with each other either or that 4-population test measure a unique reality shared by 4 populations but it's an altogether different reality from what's measured in another D test? Both f3 and D tests are measures of admixture and they mutually consistent. Patterson 2012 used f3 to determine the directionality of gene flow from northern Europe to southern Europe. On another occasion Raghavan directly compares D and f3 results and finds them consistent with each other:

"...To further investigate the evidence of gene flow between MA-1 and different Native American populations, f3(Yoruba; MA-1, Native American) was obtained using 52 Native American populations. Eskimo-Aleut speakers were found to have a lower degree of shared drift with MA-1, but no statistically significant differences existed between populations designated as having 100% 'First American' ancestry by Reich et al. (2012). This is consistent with admixture between the MA-1 lineage and Native Americans occurring prior to the diversification of Native American populations. To test this explicitly, Dfreq(Han, MA-1; X, Karitiana) tests were performed where X was one of 51 Native American populations other than Karitiana. In agreement with the outgroup f3 statistics, all populations of entirely ‘First American’ ancestry were consistent with forming a clade with Karitiana to the exclusion of MA-1 and Han (Figure SI 24). However, in tests involving Eskimo-Aleut populations a significant skew was observed where Karitiana was closer to MA-1 (Figure SI 24). "

"You are shifting the burden of proof - insisting that I prove your claim is false instead of you proving that it's true. "

All the continental groups are clearly defined in f3 tests. The burden of proof is on you to prove that it's coincidental. And you've failed the test. You can't provide a single example of two closely related populations placed in two vastly different clusters by f3 statistics. And all the quotes from the creators of the methods in question confirm the obvious truth: f3 scores are scalable across populations.

German Dziebel said...

@Tobus (contd.)

" but not every population that is shifted left *has to* have Yoruba affinity (a population which diverged from Yoruba at the same time as the A population will have no shared drift with A and thus score 0)."

That's a special case that requires special proof. For all practical purposes, we can ignore it as a general principle.

"Unless of course you can't prove your "f3 measures distance between X populations" theory has any basis - then the opposite bias could be said to be responsible."

The burden of proof is on you. f3 measures deep ancestry and admixture between populations using different populations as best representatives of ancestral components. Your resistance to this simple fact tested against dozens of samples is creationist in nature.

" exactly what you mean by "the Karitiana lens"

It's when Karitiana is used as a source population or the source of an ancestral component.

"And as this whole discussion is supposed to be about, that the LB "shift" is within the range as the modern European "shift"... agreed?"

LB has more ancient Amerindian ancestry than modern West Eurasians. We went over this already multiple times. EDF 5a shows an Amerindian shift for all Eurasian populations, EDF 5d pulls LB out of the West Eurasian cluster and places it within the Amerindian cluster.

"Huh? Where is this in EDF 5a? LB be shows no shift relative to Europeans."

Under an out-of-Africa scenario, since it's 7,000 years older than modern Europeans, we would expect it to be more Yoruba shifted than modern Europeans. But it's clearly Amerindian shifted and not Yoruba shifted. All of West Eurasians are Amerindian shifted, but LB is additionally Amerindian shifted as comparison with the ancient Amerindian-derived MA-1 sample indicates.

"Look at the PCA in Rhagavan SI 10 - see how MA-1 is not one of the extremes, that he's somewhere in the middle?"

It's the same in f3 when the axes are anchored in Karitiana and Sardinian.

"You can now see how Amerindians and LB only cluster due to the lack of direction in f3 - in reality they are the same distance from MA-1 but in different directions... this is what I am talking about."

What direction again? Amerindians and LB share common genetic ancestry. That's what f3 measures. PCA measures distances without peeling the onion on how those distances came about. f3 is more versatile because it allows you to compare distances measured against a population assumed to be ancestral, unadmixed or underived, and then change those populations around and measure again.

Tobus said...

@German:
Not a single one.

Europeans and Papuans get the same scores on the EDF 5a and 5e X-axes. Amerindians and C/SAsians get similar scores on 5c's Y-axis. Other axes cluster Papuans and Middle Easterners, C/S Asians and East Asians, Papuans and C/S Asians, etc. etc.

Baloney
You must be kidding me

Are we reading the same quotes? Show me where Rahagavan estimates the distance between X populations based on their distance to A.

Now you're going to argue that D tests ...

We're not talking about D-stats, please stay focused.

You can't provide a single example of two closely related populations placed in two vastly different clusters by f3 statistics

That's a strawman fallacy - as I've said all along closely related populations will *always* get similar f3 stats, but not *all* populations that get similar f3 scores *have to* be closely related.

What *you* need to prove is that a close affinity is the *only* way two populations can get a similar f3 score. The creators don't say that, the maths doesn't say that and there's no examples of experts using that... perhaps it's not true?

That's a special case that requires special proof. For all practical purposes, we can ignore it as a general principle.

I totally disagree, it's a fundamental aspect of the f3 stat that you should take the time to understand - this isn't a simple "A or O" comparison. Please read the sources and try to understand what is really being measured here, I'm happy to help explain the maths if it's beyond you.

The burden of proof is on you

I've proven my claim with statements from the creators, a mathematical proof and examples of it being used in practice. If you want to posit a different usage it's up to you to prove it's valid.

It's when Karitiana is used as a source population or the source of an ancestral component.

... in mathematical and allele frequency terms, how does the "source" or "ancestral" component affect the resulting score?

LB has more ancient Amerindian ancestry than modern West Eurasians

Circular argument - you are assuming this ancestry is the cause of 5d and then using 5d to prove it exists. Can you show me this supposed ancestry in any other graph?

Under an out-of-Africa scenario, since it's 7,000 years older than modern Europeans, we would expect it to be more Yoruba shifted than modern Europeans.

Again, please take some time to read how f3 stats work. These charts are not a simple Yoruba->PopA continuum.

It's the same in f3 when the axes are anchored in Karitiana and Sardinian.

... and LB shows no increased affinity to Amerindians in either of those two axes.

What direction again? Amerindians and LB share common genetic ancestry. That's what f3 measures.

The f3 plots clearly show that LB has the same degree of common ancestry with Amerindians as modern Europeans do. The only axis that shows a difference between LB and Europeans is the MA-1 axis. This axis forces East Asians, Amerindians, Europeans and Central/South Asians to all be on the same side of MA-1, when we know they are in fact on opposite sides, creating an illusion of correlation when really there are two (or more!) genetic ancestries being overlaid. It's not a valid conclusion that populations "clustering" solely on the MA-1 axis have to have shared common ancestry.

Tobus said...

@terry:What a brilliant discovery. It explains everything.

Yes it does - my guess is he smoked some weed and then went to watch Dances With Wolves... there he saw the Hollywood-inspired light and has been on a personal Spirit Quest to spread the one true gospel ever since.

German Dziebel said...

@Tobus

"Europeans and Papuans get the same scores on the EDF 5a and 5e X-axes. Amerindians and C/SAsians get similar scores on 5c's Y-axis. Other axes cluster Papuans and Middle Easterners, C/S Asians and East Asians, Papuans and C/S Asians, etc. etc."

Looks like you believe that continental populations are unrelated to each other, hence you brush it off any mathematical evidence for their relationship. Europeans and Papuans are similarly related to Amerindians on one axis, while obviously drawn apart from each other on the other. That's exactly what one would expect.

You need to provide evidence for noise in the plots by showing an individual population radically shifting its continental cluster depending on the anchor population used.

"The creators don't say that, the maths doesn't say that and there's no examples of experts using that... perhaps it's not true?"

It's true. The creators created these methods to measure ancestry. You are the only one that's denying it. This is why you're a science denier.

"in mathematical and allele frequency terms, how does the "source" or "ancestral" component affect the resulting score? "

Just read the methodology section in Patterson et al. 2012 and others to find out.

"Again, please take some time to read how f3 stats work. These charts are not a simple Yoruba->PopA continuum. "

They are. It's explicitly says so on all the plots. Find a quote proving that they are not. You go from 1 to 0 and then from 0 to -1. The scale is continuous.

"Circular argument - you are assuming this ancestry is the cause of 5d and then using 5d to prove it exists."

There's no circular argument here. The method is designed to measure ancestry.

"That's a strawman fallacy - as I've said all along closely related populations will *always* get similar f3 stats, but not *all* populations that get similar f3 scores *have to* be closely related."

Now your argument is truly circular. What you're doing is importing your assumptions about who is related to whom and then sorting the data to fit your assumption. And this is not my strawman fallacy. It's your recurrent creationist fallacy: the animal origin of man is impossible because it contradicts the Bible, hence all the evidence that scientists furnish in its favor must be noise. The origin of the Old World man from the New World man is impossible, hence all the evidence in its favor must be noise.

"I totally disagree, it's a fundamental aspect of the f3 stat that you should take the time to understand - this isn't a simple "A or O" comparison. Please read the sources and try to understand what is really being measured here, I'm happy to help explain the maths if it's beyond you."

You can disagree but you can't prove your point. You need to show, using actual populations, that the east-west cline between Yoruba and Amerindians (or between Middle Easterners and Amerindians in truncated plots) that's present in all plots (f3, PCA or else) is noise when it comes to f3 plots.

"The f3 plots clearly show that LB has the same degree of common ancestry with Amerindians as modern Europeans do."

This plot shows that all Eurasians have Amerindian ancestry. LB is an ancient sample. When an older ancient sample is used, it shows that LB has more Amerindian ancestry than other West Eurasians.

German Dziebel said...

@Tobus (contd.)

"It's not a valid conclusion that populations "clustering" solely on the MA-1 axis have to have shared common ancestry."

It is. LB and other West Eurasians already have Amerindian ancestry as EDF 5a and 5e amply prove. 5d shows that all of them get pulled toward MA-1 and Amerindians but LB is pulled the most to the effect that it becomes more Amerindian than the rest of West Eurasians, which is consistent with its being 7,000 years older than modern West Eurasians.

"creating an illusion of correlation when really there are two (or more!) genetic ancestries being overlaid."

I love what you are smoking!

"he only axis that shows a difference between LB and Europeans is the MA-1 axis. "

Tobus, as a non-scientist you have failed to set a valid procedure for testing the amount of Amerindian ancestry in LB. Instead of pointing to EDF 5a, you should've proposed to wait for another relevant ancient sample to show up. Now we only have MA-1 (plus a number of other ancient European samples that show increased Amerindian affinities) and I 100% agree that we need more but your constant milking of EDF 5a for a "proof" that LB doesn't have increased Amerindian ancestry is counter-productive. I understand you are an obsessive compulsive individual who hates losing but you need to grow up.

German Dziebel said...

@Terry and Tobus

"What a brilliant discovery. It explains everything."

Congratulations! I'm happy to see my two little creationists making their own little discoveries. If that out-of-Africa thing doesn't work out, you can always find solace working for Scotland Yard.

Tobus said...

@German;
You need to provide evidence for noise
Just read the methodology section in Patterson et al. 2012 and others to find out.
Find a quote proving that they are not.
The method is designed to measure ancestry.
You need to show, using actual populations,

I've gave my proof 3 posts ago... now where's *your* proof? Put up or shut up German.


German Dziebel said...

@Tobus

"I've gave my proof 3 posts ago... now where's *your* proof?"

The method measures ancestry between populations. This is what all the papers describing the method explicitly say. But you're saying that it's not. You're talking to yourself. A anonymous commenter with no credentials suddenly provided a "proof" for his own wishful thinking.

Tobus said...

@German:The method measures ancestry between populations. This is what all the papers describing the method explicitly say.

As all the papers describing the method explicitly state, the method measures shared drift between the X and the A population: "these outgroup f3 statistics are proportional to the amount of genetic drift shared between populations 1 and 2 since their divergence with the outgroup" (Olalde 2014)

You are claiming that the results of two such f3 measurements can be used to measure the "ancestry" between the two X populations used. Provide proof that this is true or stop saying it.

German Dziebel said...

@Tobus

"You are claiming that the results of two such f3 measurements can be used to measure the "ancestry" between the two X populations used. Provide proof that this is true or stop saying it."

They are all plotted together and their mutual relationships are explicit. With the same outgroup and the same anchor they are perfectly compatible. Populations used to have been analyzed using pairwise comparisons. Are you questioning this method, too, because only two populatiosn are being compared at a time?

The scholars forgot that for the readers such as Tobus they need to chew every little bit up.

Tobus said...

@German:They are all plotted together and their mutual relationships are explicit.

Yes German, I know you are assuming X population relationships from their relative positions in plots.... but please provide some evidence that this is actually a valid thing to do.

German Dziebel said...

@Tobus

"Yes German, I know you are assuming X population relationships from their relative positions in plots.... but please provide some evidence that this is actually a valid thing to do."

The data and the methods to analyze it are very explicit. They make absolute sense (continental population clusters are clearly identified, differences between X populations relative to Amerindians are proportionate to geographic distance from America and proximity to Africa, only one variable in a formula changes with every application). I encourage you to write to Nick Patterson who sometimes visits blogs (or anyone else) and ask him your specific question. And ask for an extended reply, so we don't debate ambiguities. Be neutral and objective in the way you formulate your question and report back even if it contradicts your beliefs.

German Dziebel said...

@Tobus

This should resolve the dispute.

Lazaridis published un update on his study of Ancient Eurasian samples:

http://biorxiv.org/content/biorxiv/suppl/2014/04/05/001552.DC4/001552-3.pdf

Table S14.7, p. 96, confirms that LB, just like all other ancient European samples, has Amerindian affinity. And it's f4 stats.

Tobus said...

@German:
The data and the methods to analyze it are very explicit.

And they reject your interpretation - see the quotes and mathematical proof above.

They make absolute sense (continental population clusters are clearly identified,

Continental clusters share similar genetic histories and so cluster as expected.

differences between X populations relative to Amerindians are proportionate to geographic distance from America and proximity to Africa

Differences between non-contentinal X populations are random and contradictory depending on the A population used. (eg Papuans identical with Europeans on the Karitiana axis, yet furthest from them on the Saridinian axis.)

This should resolve the dispute...Table S14.7, p. 96, confirms that LB, just like all other ancient European samples, has Amerindian affinity.

The statement I'm contesting is that LB has *more* Amerindian affinity than *modern* Europeans. A table showing that he has the *same* Amerindian affinity as *ancient* Europeans (which BTW is significantly *less* than MA-1) doesn't confirm that statement, and it is and always has been, directly contradicted by EDF 5a.


terryt said...

"LB, just like all other ancient European samples, has Amerindian affinity".

Didn't we already know that to be the case? But note: 'has Amerindian affinity' does not mean 'is descended from Amerindians'. A distinction you are not prepared to accept. It means simply that Amerindians and Europeans both descend partly from a common source population. I would have thought that has been obvious for some time now.

German Dziebel said...

@Tobus

"And they reject your interpretation - see the quotes and mathematical proof above."

Your quote selection and mathematical proof are substandard and can't be used to question the actual data and the my superior interpretive logic.

"Continental clusters share similar genetic histories and so cluster as expected."

Hmm, this makes me think you are a polygenecist.

"Differences between non-contentinal X populations are random and contradictory depending on the A population used. (eg Papuans identical with Europeans on the Karitiana axis, yet furthest from them on the Saridinian axis.)"

There's no contradiction here: they are equally removed from Amerindians but in two different directions. As you notice, the Y axis orders populations according to their continental provenance. The X axis measures their genetic proximity to Amerindians. So Europeans and Papuans are equally removed from Amerindians genetically but they are radically apart geographically. This is what two migrations in different directions would cause.

"The statement I'm contesting is that LB has *more* Amerindian affinity than *modern* Europeans. A table showing that he has the *same* Amerindian affinity as *ancient* Europeans (which BTW is significantly *less* than MA-1) doesn't confirm that statement, and it is and always has been, directly contradicted by EDF 5a."

You were contesting every statement that came from me. You were arguing that LB is not related to MA-1 through the Amerindian line. The new Lazaridis data proves that LB has Amerindian affinity, which is predictably less than MA-1 (because MA-1 is older and closer to Amerindians geographically) but both populations are linked through their Amerindian affinity. Modern Europeans are derived from a population related to LB, so it's irrelevant if they are equally removed from Amerindians or not. They should be by virtue of the fact that LB is 7,000 years older than modern Europeans and it doesn't skew African, it skews Amerindian. Olalde EDF 5 d perfectly reflects the Amerindian roots of LB and the fact that ancient Eurasians are more Amerindian than modern Eurasians.

Hamar Fox said...

I don't see any contradictions or surprises at all. Reading through the comments at Eurogenes, it's simply mind blowing how so few people get what should be so obvious. For example, they're talking about the affinity of LB to NE Siberians relative to other E. Asians as being related to his C Y-DNA, without considering that NE Siberians have MA-1-like admixture that renders them more similar to European HGs than populations that lack this admixture.

It's not as bad here, but I agree with Terry: the explanations are right there, laid out for us, and there's not really any excuse for not getting it at this point.

The f3 statistics seem to take drift (or at least some of it) into account. This is seen from the fact that PCAs place LB closer to Eastern populations (relative to other Europeans) than the f3 statistics do. LB does also seem to have some farmer ancestry. The small difference that remains between LB and modern Europeans is caused by higher 'Sardinian' ancestry in modern Europeans.

LB doesn't have ANE ancestry, so his relation to Amerindians is purely by way of his membership in the Eurasian HG clade, which his non-'Sardinian' portion of ancestry belongs to. The Motala HGs do seem to have ANE, so if they were included in the f3 statistics, they'd be closer to MA-1 (and hence Amerindians) than LB, especially considering that they lack 'Sardinian' ancestry.

I really don't see anything that doesn't fit or make sense. The newest ADMIXTURE analysis from Lazaridis et al. shows that the 'East Eurasian' portion of the hunter-gathers' ancestry at low K-levels resolves into Amerindian and NE Siberian at higher K-levels, where before it was only Amerindian. This is a perfect demonstration of how ADMIXTURE can get things the wrong way round, since we know NE Siberians are the ones mixed with Eurasian HGs, not the reverse. If HGs were descended from Amerindians, why would ADMIXTURE choose to identify portions of their ancestry (however small) as NE Siberian over Amerindian? It would have to favour a false signal over a real one, which should and would be much stronger.

Tobus said...

@German:
they are equally removed from Amerindians but in two different directions.

Yes! And the same principle could also apply to LB and Amerindians on the MA-1 axis, no? Same distance, different directions?

German Dziebel said...

@Hamar Fox

"LB doesn't have ANE ancestry,"

Just read Olalde: LB and MA-1 share common ancestry. So your whole argument falls apart.

@Tobus

"Yes! And the same principle could also apply to LB and Amerindians on the MA-1 axis, no? Same distance, different directions?"

Nonsense. In Olalde EDF 5a Papuans and Europeans share the same amount of Amerindian affinity but they form different continental clusters. In Olalde EDF 5d, LB has an increased share of Amerindian/Amerindian-derived (MA-1) ancestry, while still exhibiting proximity to the West Eurasian geographic cluster because they are a European population. Everything is simple and logical.

terryt said...

"my superior interpretive logic".

Which we are yet to see any evidence of.

"it's simply mind blowing how so few people get what should be so obvious... I agree with Terry: the explanations are right there, laid out for us, and there's not really any excuse for not getting it at this point".

Yes, it seems so obvious if you approach the problem with an open mind rather than trying to fit the data to some preconceived notion.

"And the same principle could also apply to LB and Amerindians on the MA-1 axis, no? Same distance, different directions?"

You're not going to get German to agree with that. He is the individual here most hobbled by some preconceived notion.

Hamar Fox said...

German,

Just read Olalde: LB and MA-1 share common ancestry. So your whole argument falls apart.

Having shared ancestry =/= one group being descended from the other. In fact, this reply should be unnecessary, since I already said this in my last comment:

LB doesn't have ANE ancestry, so his relation to Amerindians [and MA-1] is purely by way of his membership in the Eurasian HG clade, which his non-'Sardinian' portion of ancestry belongs to.

I added the bracketed part, but my point was hardly cryptic.

Tobus said...

@German:Nonsense. In Olalde EDF 5a Papuans and Europeans share the same amount of Amerindian affinity but they form different continental clusters

.. and LB and Amerindians aren't from different continents?

If "same distance different direction" can apply to two X populations in 5a, then you have to at least admit the *possibility* that it might also apply to two X populations in 5d... no? Consistency not something they teach you in humanities school?


@terry:You're not going to get German to agree with that.

Probably not, but it's so obvious I can't help point it out.

German Dziebel said...

@Tobus

"If "same distance different direction" can apply to two X populations in 5a, then you have to at least admit the *possibility* that it might also apply to two X populations in 5d... no? Consistency not something they teach you in humanities school?"

I know that my elite education in the natural sciences, social sciences and humanities is hard on your ego, but try to bring any education you may have to bear on understanding one thing: you are inventing "possibilities" that are not reflected on any of the generated plots or tables. Before talking about consistency, at least try to abstain from reveries. MA-1 is not more Amerindian than modern Amerindians, as EDF 5a shows. On EDF 5a MA-1 is more Amerindian-shifted than the West Eurasian cluster, hence when the axis is anchored in MA-1 and LB falls into the Amerindian cluster, it means it has Amerindian ancestry. Lazaridis 's f4 stats confirm this. LB and MA-1 are of course geographically distinct as their respective positions on the Y axis illustrate, but genetically both MA-1 and LB are Amerindian-derived.

@Hamar Fox

"LB doesn't have ANE ancestry, so his relation to Amerindians [and MA-1] is purely by way of his membership in the Eurasian HG clade, which his non-'Sardinian' portion of ancestry belongs to."

If MA-1 is not ANE, then I'm confused which populations are. LB has confirmed Amerindian ancestry (Olalde, Lazaridis), it shares ancestry with MA-1 (Olalde), which has Amerindian ancestry (Raghavan's data), that's all that matters, it seems.

Simon_W said...

@ Hamar Fox

LB does also seem to have some farmer ancestry

La Brana has farmer ancestry? Where did you get this from? I don't see it in the ADMIXTURE analysis.

Tobus said...

@German:
you are inventing "possibilities" that are not reflected on any of the generated plots or tables

Hang on a minute - you just posited this possibility yourself two posts ago to explain the contradictory positions of the various X populations. To wit: "There's no contradiction here: they are equally removed from Amerindians but in two different directions."... are you saying this is not the case anymore?

On EDF 5a MA-1 is more Amerindian-shifted than the West Eurasian cluster, when the axis is anchored in MA-1 and LB falls into the Amerindian cluster, it means it has Amerindian ancestry.

This is the assumption I've been asking you to provide some hard evidence for over the last few weeks... what is your proof that LB's shared drift with MA-1 *has to be* at the exact same loci as MA-1's shared drift with Amerindians. There's no requirement for this to be the case in the maths and such an interpretation contradicts EDF 5a.... can you please provide some actual solid evidence (not just your opinion!) that your assumption here is valid?

Hamar Fox said...

Simon W,

La Brana has farmer ancestry? Where did you get this from? I don't see it in the ADMIXTURE analysis.

My bad. I've seen it mentioned a lot in comments, and I've seen LB-1 get double figure scores for 'Mediterranean' and 'Basque' ancestry, so that's what I was thinking of. But you're right, it's not supported in the Lazaridis et al. data. His scores in S. European components, in that case, probably reflect the WHG influence in those components, and not the reverse.

But now that leaves me again with the problem of why LB is so much more related to European Neolithics and modern Sardinians than MA-1 is (as seen, for example, in EDF 5 a and b in the Olalde et al. paper). True, the Neolithics and modern Sardinians have WHG ancestry and little to no ANE ancestry, but is there some other connection between EEFs and WHGs that EEFs and ANEs lack? It would be interesting to see how MA-1 and LB-1 compare in relation to a 'southern' population with very little WHG or ANE ancestry, such as the Maltese.

German Dziebel said...

@Tobus

"This is the assumption I've been asking you to provide some hard evidence for over the last few weeks... what is your proof that LB's shared drift with MA-1 *has to be* at the exact same loci as MA-1's shared drift with Amerindians. There's no requirement for this to be the case in the maths and such an interpretation contradicts EDF 5a.... can you please provide some actual solid evidence (not just your opinion!) that your assumption here is valid?"

It doesn't contradict EDF 5a. In 5a, all Eurasians are shifted toward Amerindians. MA-1 is significantly shifted toward Amerindians and falls outside of the West Eurasian cluster. In EDF 5d, when we use MA-1 as an anchor, LB does the same thing: it falls out of the West Eurasian cluster and falls with Amerindians.

"can you please provide some actual solid evidence (not just your opinion!) that your assumption here is valid?"

f4 stats from Lazaridis confirm it.

"here's no requirement for this to be the case in the maths."

Then your maths is no more than myths.

Tobus said...

@German:
It doesn't contradict EDF 5a.

Yes it does. If LB's shift out of the European cluster on the MA-1 axis is due to higher shared drift with Karitiana (another X population), then LB should also be shifted out of that cluster on the Karitiana axis (the direct measurement). He's not, contradicting your yet-to-be-substantianted-by-any-solid-evidence theory.

f4 stats from Lazaridis confirm it.

You're just regurgitating previous arguments now - those f4 show that LB has the *same* affinity to Amerindians as other *ancient* Europeans... you're interpretation of 5d is the he has *more* affinity than *modern* Europeans.

Then your maths is no more than myths.

So prove it!!

German Dziebel said...

@Tobus

"If LB's shift out of the European cluster on the MA-1 axis is due to higher shared drift with Karitiana (another X population), then LB should also be shifted out of that cluster on the Karitiana axis (the direct measurement). He's not, contradicting your yet-to-be-substantianted-by-any-solid-evidence theory."

You're pitting one kind of data against another. In reality, they work very well together.

"You're just regurgitating previous arguments now - those f4 show that LB has the *same* affinity to Amerindians as other *ancient* Europeans... you're interpretation of 5d is the he has *more* affinity than *modern* Europeans."

You've been denying that LB has any Amerindian ancestry and that it's related to MA-1 on the "west Eurasian" side. All the data refutes this. It's irrelevant if modern Europeans have the same amount of Amerindian ancestry as LB because ultimately they are derived from a LB-like population and LB is derived from an MA-1-like population and MA-1 is derived from an Amerindian population.

"So prove it!!"

No, you have to show with better maths that f3 and f4 stats create a wrong system of relationshios between the populations on a global plot. Considering you haven't done it, it's clear that you are just bluffing and shifting the burden of responsibility to cover it up.

German Dziebel said...

@Tobus (contd.)

"If LB's shift out of the European cluster on the MA-1 axis is due to higher shared drift with Karitiana (another X population), then LB should also be shifted out of that cluster on the Karitiana axis (the direct measurement)."

Another way to see the greater proximity between LB and Amerindians as compared to modern West Eurasians is Olalde EDF 5a and EDF 5c. When the Y axis is anchored in LB, Amerindian and West Eurasians are closer to each other than when the Y axis is anchored in Sardinians. The proximity is 0.01, which roughly the same value by which LB is pulled out of the West Eurasian cluster in EDF 5d.

One needs to remember that Amerindians are much more divergent from the rest of Eurasians, as EDF 5a compared to EDF 5b and the rest of the plots attest, so when Amerindians and West Eurasians are pulled closer together on the Y axis, this can only mean Amerindian admixture in West Eurasians via a LB-like population, not West Eurasian admixture in Amerindians.

Tobus said...

@German:
You've been denying that LB has any Amerindian ancestry

As I've said from day 1, LB has the same amount of Amerindian affinity as modern Europeans do (EDF 5a).

and that it's related to MA-1 on the "west Eurasian" side.

LB's increased affinity to MA-1 over Europeans in EDF 5d is affinity to MA-1's "Eurasian" side yes. If it were to MA-1's "Amerindian" side then LB would be shifted out of the European cluster on EDF 5a as well. It's plain to see from these 2 charts that LB's increased affinity to MA-1 does not correspond with an increased affinity to Amerindians.

you have to show with better maths that f3 and f4 stats create a wrong system of relationshios between the populations on a global plot

You are the one claiming a relationship that a) the creators don't claim, b) no experts use, c) isn't measured by the maths and d) produces contradictory results. Either show me the scientific reasoning and mathematical working behind this proposed X-to-X relationship or shut up about it.

Patterson (2012) is an excellent resource to explain how the f3 stats are calculated. Please read this and let me know if you find anything to support your claims... if not then you just made it up and are unable to prove how it works.

When the Y axis is anchored in LB, Amerindian and West Eurasians are closer to each other than when the Y axis is anchored in Sardinians. The proximity is 0.01, which roughly the same value by which LB is pulled out of the West Eurasian cluster in EDF 5d.

Something you probably didn't realise about the f3 calculation is that swapping the X and A populations has no effect: (O-X)(O-A) is mathematically identical to (O-A)(O-X). This means if we take the Amerindian values from both these axes (LB and Sardinian) we get the same values that LB and Sardinians would get on the Karitiana axis. In this case 0.16 (Sard.) and 0.17 (LB).

If you look at EDF 5a you can confirm this - the leftmost blue dot is at 0.16 and LB is at 0.17. You can also see that Sardinians aren't the only modern European population tested, and that one population (Russians? Finns?) scores higher than LB (~.175), another gets the same as LB and a few are only marginally below LB. I'm not sure which dictionary you use, but if LB's score on the Karitiana axis is higher than some Europeans and lower than others, this means LB's affinity to Amerindians is "in the range of modern Europeans".




German Dziebel said...

@Tobus

"As I've said from day 1, LB has the same amount of Amerindian affinity as modern Europeans do (EDF 5a)."

We're going in circles. LB shows increased affinity to Amerindians over modern Europeans in EDF 5a. EDF 5a shows that all Eurasians are Amerindian shifted. EDF 5c shows that LB makes Europeans closer to Amerindians, while Sardinians don't. f4 stats confirms that LB has Amerindian affinity. Another way to look at it: LB is shifted toward MA-1 and Amerindians in EDF 5d compared to other West Eurasians , MA-1 is shifted toward Amerindians in EDF 5a compared to other West Eurasians (including LB), hence LB has Amerindian affinity to a greater degree than modern Europeans.

"LB's increased affinity to MA-1 over Europeans in EDF 5d is affinity to MA-1's "Eurasian" side yes."

That's not what you were claiming, but fine. But it's clear that LB falls with Amerindians and not with any Eurasians on EDF 5d, so it's not Eurasian affinity but Amerindian.

"You are the one claiming a relationship that a) the creators don't claim, b) no experts use, c) isn't measured by the maths and d) produces contradictory results. Either show me the scientific reasoning and mathematical working behind this proposed X-to-X relationship or shut up about it."

All the creators interpret the plots (f4 and f3) just like I do - various Eurasian groups have a varying degree of "Amerindian admixture," or that "among all modern populations MA-1 is closest to Amerindians," which is exactly the same inference. It is measured by the available maths to the extent maths can measure genetic affinity. And the results are very consistent, although nuanced,a s they should be. Your objections are frivolous and ill-advised. There's nothing more to talk about.

"Something you probably didn't realise about the f3 calculation is that swapping the X and A populations has no effect: (O-X)(O-A) is mathematically identical to (O-A)(O-X). This means if we take the Amerindian values from both these axes (LB and Sardinian) we get the same values that LB and Sardinians would get on the Karitiana axis."

But you probably don't realize that Amerindians have much higher values on the Karitiana axis than Europeans have on the Sardinian axis.

"In this case 0.16 (Sard.) and 0.17 (LB)."

Dah, that was my point. LB is closer to Karitiana than Sardinians.

"one population (Russians? Finns?) scores higher than LB (~.175), another gets the same as LB and a few are only marginally below LB."

Yes, that's true. And it transpired in Lazaridis, too. This is likely the effect of later "East Asian" gene flow into Eastern Europe (EDF 5b has a slight Han-shift in that Finn or Russian population). If you take it out, you'll be left with LB being Amerindian shifted compared to modern Europeans (unaffected by recent East Asian gene flow) in EDF 5a as well.

"this means LB's affinity to Amerindians is "in the range of modern Europeans".

The above proves you wrong.

Tobus said...

@German:
LB shows increased affinity to Amerindians over modern Europeans in EDF 5a.

Huh?? LB falls inside the Europeans cluster on EDF 5a

That's not what you were claiming, but fine.

Yes, it's exactly what I'm claiming - I just claimed it!

But it's clear that LB falls with Amerindians and not with any Eurasians on EDF 5d,

EDF 5d measures affinity to MA-1, LB only "falls with" Amerindians on that plot in the same way that Papuans "fall with" Europeas in 5a.

All the creators interpret the plots (f4 and f3) just like I do

None of them say LB is closer to Amerindians than modern Europeans are - that's the difference.

Dah, that was my point. LB is closer to Karitiana than Sardinians.


The above proves you wrong.

Sorry what proves me wrong? LB is in the range of modern Europeans, it's a plainly observable fact... you suggesting that he wouldn't be in the range if we take out the populations that score higher than him out only proves that point.

So can we please agree - this data shows LB having an affinity with Karitiana that is within the range of modern European affinity with Karitiana.






German Dziebel said...

@Tobus

"EDF 5d measures affinity to MA-1, LB only "falls with" Amerindians on that plot in the same way that Papuans "fall with" Europeas in 5a."

Yes, in both cases because of shared genetic affinity via a third population. LB falls outside of the West Eurasian cluster and falls into the Amerindian cluster. Amerindians are the closest to MA-1 among all modern human populations (Raghavan Fig. 1c, f3 stats), so for all practical purposes this third population MA-1 equals Amerindians.

"None of them say LB is closer to Amerindians than modern Europeans are - that's the difference."

They can't chew everything up for you. Think on your feet. Or better listen to me and learn.

"Sorry what proves me wrong? LB is in the range of modern Europeans, it's a plainly observable fact... you suggesting that he wouldn't be in the range if we take out the populations that score higher than him out only proves that point.

So can we please agree - this data shows LB having an affinity with Karitiana that is within the range of modern European affinity with Karitiana."

You keep trying to sell me your beliefs. Do you work as a full-time preacher, Tobus? I don't buy your beliefs because they are not supported by facts. EDF 5d clearly establishes Amerindian affinity of LB in higher proportion than modern Europeans. It's likely that EDF 5a shows LATER gene flow into northern Europe from Asia, which made some European populations as close to Amerindians and East Asians as LB, so EDF 5d represents the original situation.

Tobus said...

@German:
Yes, in both cases because of shared genetic affinity via a third population

...and most importantly, a relative lack of genetic affinity when measured directly against each other.

Amerindians are the closest to MA-1 among all modern human populations (Raghavan Fig. 1c, f3 stats), so for all practical purposes this third population MA-1 equals Amerindians.

This logic doesn't hold - "closest" doesn't mean "equals".

We have examples of real-life practical purposes here, so let's compare the X-axes of EDF 5c and 5e in Olalde for example - you can see that we get extremely different results when we use MA-1 in the place of Karitiana. So contrary your unfounded assumption, MA-1 is clearly *not* equal to Americans for practical purposes.

You seem to have taken the fact that MA-1 shares some DNA with Amerindians (less than 1/3 by all reckonings) and exagerrated it all out of proportion to the point where you now believe that MA-1 actually *was* an Amerindian. You seem to think that any affinity to MA-1 affinity MUST also be affinity to Amerindians, but this is not the case... it's plain to see from the data that affinity to MA-1 is not the same as affinity to Amerindians - that's why their f3 axes look so different.

You keep trying to sell me your beliefs

And here's that pattern again! You are saying my point is a "belief" when it's a clearly verifiable statement of fact (black star is inside the blue dots on the Karitiana axis), and in fact it's *you* that is basing your argument of a false belief - that MA-1 is an Amerindian. You are falsely accusing me of something that you yourself are in fact doing.

EDF 5d clearly establishes Amerindian affinity of LB in higher proportion than modern Europeans.

Incorrect! As I've pointed out multiple times now and as is clearly written under the X-axis - EDF 5d establishes affinity to MA-1. Affinity to MA-1 is not the same as affinity to Amerindians, you can see this for yourself by comparing how differently the populations score on the MA-1 axis to the Karitiana axis.

It's likely that EDF 5a shows LATER gene flow into northern Europe from Asia...

Whatever reason you want to explain it away with, this data shows LB's affinity to Amerindians falls into the same range shown by modern Europeans. That's just a basic description of EDF 5a, no?


German Dziebel said...

@Tobus

"And here's that pattern again! You are saying my point is a "belief" when it's a clearly verifiable statement of fact (black star is inside the blue dots on the Karitiana axis), and in fact it's *you* that is basing your argument of a false belief - that MA-1 is an Amerindian. You are falsely accusing me of something that you yourself are in fact doing."

My criticism of you is fully in line with facts about your approach to manipulating scientific data. You follow your belief and not evidence or logic. Here's what's uncontroversial. EDF 5a shows that all Eurasians have Amerindian affinity. EDF 5d takes LB out of the West Eurasian cluster and places it into the Amerindian cluster. f4 confirms that LB has Amerindian (and not only MA-1 affinity). f3 supports the notion that MA-1 is the closest to Amerindians out of all modern human populations, which means that they can form a broad cluster, with gaps of attestation. Nowhere in Olalde can you see MA-1 as divergent from Amerindians. The fact that LB falls into the West Eurasian cluster in EDF 5a can easily be explained as subsequent gene flow from Asia into Russians and Finns. Gently remove a couple of rightmost dots in the West Eurasian cluster and you'll see more similarity between EDF 5a and EDF 5d. LB will be an outlier from the West Eurasian cluster toward Amerindians.

"We have examples of real-life practical purposes here, so let's compare the X-axes of EDF 5c and 5e in Olalde for example - you can see that we get extremely different results when we use MA-1 in the place of Karitiana."

Of course, they are different. MA-1 is 24,000 years old. Any sample of that antiquity will reduce the distance between different living continental populations because it's closer to the time of their separation (BTW, you can hopefully see for yourself that f3 measures genetic distance between X populations just fine), but it's still most proximate to Amerindians, hence forms a cluster with them, not with East Asians, West Eurasians or Papuans. And since it's physically found outside of America, an out-of-America migration is the most natural explanation.

"Whatever reason you want to explain it away with, this data shows LB's affinity to Amerindians falls into the same range shown by modern Europeans. That's just a basic description of EDF 5a, no?"

You are not capable of basic descriptions. You are only capable of biased descriptions. The basic description of EDF 5a is that all Eurasian populations have Amerindian ancestry. When you want to drill down to such an ancient sample as LB, you need to use other plots. Once you see the difference between EDF 5a and 5d, you shouldn't dismiss EDF 5d as an inconvenient fact but find the best way to reconcile the two. And the best way is to take EDF 5d as a starting point because it's actually anchored in an ancient sample and then explaining EDF 5a as obscured by LATER gene flow. Once you remove the easternmost populations from the West Eurasian cluster because they were likely affected by LATE, local gene flow from Asia, you'll arrive at the same basic pattern in EDF 5a as in EDF 5d in which a westernmost, ancient European sample, LB, shows an Amerindian ancestry at a higher proportion than modern Europeans, with the exception of those that were exposed to LATER gene flow from Asia.

Tobus said...

@German:
Here's what's uncontroversial. ...

You are going to extraordinary lengths to explain WHY LB falls into the same range as modern Europeans in graph 5a, so why is it so hard for you to admit THAT he does... if he didn't you wouldn't need to be explaining it away. So please confirm what everyone can see and what you are avoiding but making up excuses for: LB's affinity to modern Americans (for whatever reason!) is within the range shown by modern Europeans.

Gently remove a couple of rightmost dots in the West Eurasian cluster and you'll see more similarity between EDF 5a and EDF 5d.

So you fudge the data until it says what you want? Nice.

Of course, they are different. MA-1 is 24,000 years old.

So let's not pretend that MA-1 is the same as Amerindians "for all practical purposes" then shall we? Affinity to MA-1 is clearly not the same as affinity to Amerindians, the Karitiana axis is the only reliable measure of Amerindian affinity we have in this data.

Any sample of that antiquity will reduce the distance between different living continental populations because it's closer to the time of their separation

It may well do your head in, but this is a mistaken assumption (which I also made earlier) - the age of the samples doesn't affect the f3 stats at all. f3 measures shared drift from the time of separation from the outgroup to the time of separation of the X1 and A populations. Drift in either the X or A population after their divergence won't affect the results, since the f3 only counts alleles that X and A both have (mathematically, either O-X or O-A will be zero if only X or A has the allele). Effectively the f3 stops counting at the time of X/A divergence - as long as the sample is after that point it makes no difference how close to it it is.

(BTW, you can hopefully see for yourself that f3 measures genetic distance between X populations just fine)

No, I can't see that at all... where are you seeing this? What I see is that we'd get totally different distances depending on whether MA-1 or Karitiana was used (East Asian/Amerindians distance changes from 0.1 to 0.025, European/Papuan from 0 to 0.02, South Asians and East Asians would swap who's closest to Amerindians... etc. etc. etc.). I can't see any consistent way of measuring the distance between the X populations using this data.

The basic description of EDF 5a is that all Eurasian populations have Amerindian ancestry. When you want to drill down to such an ancient sample as LB, you need to use other plots.

LB is clearly plotted on EDF 5a so there's no need to go a different plot, and there's no other plot that measures LB against Karitiana anyway. Can you please expand your "basic description" to include where LB sits relative to these Europeans?

You are not capable of basic descriptions. You are only capable of biased descriptions.

And yet it's you saying "Remove some dots from the graph!" and "Ignore the only graph with the data!"... it seems like *you* have a preconceived outcome you want to show and *you* are manipulating the data to get it, but surprise, surprise it's you accusing me of do what you are in fact doing yourself.... again.

I'll also point you to the latest <a href="http://www.sciencemag.org/content/early/2014/04/23/science.1253448>Skoglund paper</a> which shows LB having significantly less Anzick (and MA-1) affinity than the Swedish hunter-gatherers (Table S13 in the SI). While it doesn't directly compare LB with modern Europeans, it does confirm that LB's "Amerindian" aspect was not particularly remarkable in his ancient context, and might make his appearance in the modern European cluster in 5a more palatable for you.

terryt said...

"So you fudge the data until it says what you want?"

He has always done that. Or simply ignores inconvenient data. Calls himself a 'scientist'? Hilarious. At least we can laugh at him.

German Dziebel said...

@Tobus

"You are going to extraordinary lengths to explain WHY LB falls into the same range as modern Europeans in graph 5a, so why is it so hard for you to admit THAT he does... if he didn't you wouldn't need to be explaining it away. So please confirm what everyone can see and what you are avoiding but making up excuses for: LB's affinity to modern Americans (for whatever reason!) is within the range shown by modern Europeans."

No, as EDF 5d shows. But you can keep ignoring the data to stay with your belief.

"So you fudge the data until it says what you want? Nice.

This is not fudging, it's just teaching you how to analyze it. But for a creationist bigot such as yourself, any analysis equals fudging with divinely revealed truth.

"So let's not pretend that MA-1 is the same as Amerindians "for all practical purposes" then shall we? Affinity to MA-1 is clearly not the same as affinity to Amerindians, the Karitiana axis is the only reliable measure of Amerindian affinity we have in this data."

Nonsense. MA-1 is the closest to Amerindians out of all modern human populations. Not to Papuans, not to East Asians, not to West Eurasians. This means it can form an ancestral clade only with Amerindians. EDF 5d shows it nicely.

"It may well do your head in, but this is a mistaken assumption (which I also made earlier) - the age of the samples doesn't affect the f3 stats at all."

Of course it does as it elicits ancestral relationships obfuscated by later gene flow. Historical thinking is just another competence that you're missing.

"LB is clearly plotted on EDF 5a so there's no need to go a different plot, and there's no other plot that measures LB against Karitiana anyway. Can you please expand your "basic description" to include where LB sits relative to these Europeans?"

LB is an ancient sample and it requires another ancient sample to figure out its placement. If LB had SSA admixture, it would have belonged left of the West Eurasian cluster (with Middle East) and Amerindian admixture would move it on the right and lumped it with West Eurasians who could easily had less SSA admixture. But EDF 5d clearly shows that LB has an exclusive eastern ancestral affinity to MA-1 and Amerindians and modern West Eurasians are shifted away from it either through divergence or through greater SSA admixture. It's most likely the former.

" I can't see any consistent way of measuring the distance between the X populations using this data."

Why do you need it in the first place? To measure an outcome of the divine plan, not an evolutionary process? You are in the wrong business then.

"I'll also point you to the latest <a href="http://www.sciencemag.org/content/early/2014/04/23/science.1253448>Skoglund paper</a> which shows LB having significantly less Anzick (and MA-1) affinity than the Swedish hunter-gatherers (Table S13 in the SI). While it doesn't directly compare LB with modern Europeans, it does confirm that LB's "Amerindian" aspect was not particularly remarkable in his ancient context, and might make his appearance in the modern European cluster in 5a more palatable for you."

We have perfect data from the latest Lazaridis stating in plain English that LB has Amerindian affinity. Keep tryin

German Dziebel said...

@Tobus (contd.)

g to weasel your way out of the data trap!

"East Asian/Amerindians distance changes from 0.1 to 0.025, European/Papuan from 0 to 0.02, South Asians and East Asians would swap who's closest to Amerindians."

All of this is very consistent. West Eurasians and South Asians stay together in both EDF 5a and EDF 5d. East Asians maintain their special proximity to Papuans but are also drawn close to West Eurasians in both MA-1 anchored plot and in the Karitiana-anchored plot. (And in Tianyuan anchored plot, for that matter.) The notable difference between EDF 5a and EDF 5d is that West Eurasian and West Eurasian-like populations (such as South Asians) and East Asians and East Asian-like populations (such as Papuans) swap their positions vis-a-vis Amerindians but this is precisely what we would expect since MA-1 is an admixed West Eurasian-Amerindian population, while East Asians are not.

Tobus said...

@German:
No, as EDF 5d shows

EDF 5d shows affinity to MA-1, not Amerindians!! How many times do I need to remind you of that?

This is not fudging, it's just teaching you how to analyze it.

No, it's definitely fudging - you are saying we should ignore some of the results on the chart so it fits your position. If we take the chart as it is, without fudging, LB's affinity to Amerindians is clearing in the range of modern Europeans... agreed?

Nonsense.

Nonsense what? Are you saying that MA-1 and Amerindians *ARE* identical to each other for all practical purposes? Perhaps you can tell me the scores of all the continental groups on both the MA-1 and the Karitiana axes.

Of course it does as it elicits ancestral relationships obfuscated by later gene flow.

We're simply trying to agree on what Amerindian affinity LB shows relative to modern Europeans - whether some earlier affinity has been "obfuscate by later gene flow" or not, is not an issue: this data shows that LB's Amerindian affinity is within the range of modern Europeans, plain and simple.

LB is an ancient sample and it requires another ancient sample to figure out its placement.

We're not talking about some ethereal ancient placement, we're talking about his placement relative to a specific modern population... why are you trying to avoid the obvious fact that LB scores within the range of modern Europeans?

Why do you need it in the first place?

I don't need it - you presented it as proof that LB and modern Europeans relative scores on the MA-1 axis means that LB is closer to Amerindians than Europeans... contrary to what the direct measurement against Karitiana shows.

We have perfect data from the latest Lazaridis stating in plain English that LB has Amerindian affinity.

Not contested - LB has Amerindian affinity, and it's within the same range shown by modern Europeans.

All of this is very consistent.

Not if you are still proposing that the difference between two f3s is a measure of genetic affinity between the two X populations - using that logic Papuans are both identical to, and the most divergent from, Europeans... a clearly contradictory interpretation.

German Dziebel said...

@Tobus

"EDF 5d shows affinity to MA-1, not Amerindians!! How many times do I need to remind you of that?"

It's like saying that the Y axis shows proximity to Sardinians but not to (other) West Eurasians. Sardinians are West Eurasians, just like MA-1 are Amerindians.. admixed with West Eurasians. This is why Amerindians are still rightmost in EDF 5d but MA-1 is midway between Amerindians and West Eurasians in EDF 5a.

I guess your hangup is that you secretly hope MA-1 will turn out to be a UFO-derived population that admixed into West Eurasians and Amerindians and then returned to its launchpad on Mars.

"LB's affinity to Amerindians is clearing in the range of modern Europeans... agreed?"

No. Unless you believe Amerindian affinity in Europe came from Vikings and are trying to propose that Vikings started to bring Amerindian mates to Europe 7,000 years ago. LB is an ancient sample - it needs to be compared to another ancient sample and then assess modern variation in light of this ancient variation. If we have a robust enough ancient sample but still see that our modern sample has an expected amount of "unlost" ancient affinity such as Amerindian affinity, then some of the modern samples must have replenished it through recent gene flow from a different area, which is geographically more proximate to the New World.

"Are you saying that MA-1 and Amerindians *ARE* identical to each other for all practical purposes? Perhaps you can tell me the scores of all the continental groups on both the MA-1 and the Karitiana axes."

MA-1 is a mix between an Amerindian and a West Eurasian populations(s). I've told you this a gazillion times. Of course, they are not the same on all charts but they are the same for all practical purposes in their Amerindianness. Anzick scores even higher than modern Amerindians on an MA-1 axis, so it's likely it's even more Amerindian than MA-1.

"Not if you are still proposing that the difference between two f3s is a measure of genetic affinity between the two X populations - using that logic Papuans are both identical to, and the most divergent from, Europeans... a clearly contradictory interpretation."

Of course, for a creationist like you "descent with modification" sounds like a contradictory interpretation. For a scientist like me there's no contradiction between Papuans and Europeans sharing the same amount of Amerindian descent and an exorbitant amount of their specific continental "modifications."

terryt said...

" For a scientist like me"

Ha, ha, ha.

Tobus said...

@German:
It's like saying that the Y axis shows proximity to Sardinians but not to (other) West Eurasians.

All West Eurasians plot together on all axes and in all other genetic measurements and they all live in the same geographic area and temporal period. MA-1 doesn't plot with Amerindians on any axis, nor in any other genetic measurements and he lived on a different continent 24,000 years ago... by what logic are you proposing these are the same?

I guess your hangup is that you secretly hope MA-1 will turn out to be a UFO-derived population

Then you guess wrong.

No. Unless you believe Amerindian affinity in Europe came from Vikings and are trying to propose that Vikings started to bring Amerindian mates to Europe 7,000 years ago

What are you talking about?! All I'm doing is describing the data - how does LB plot relative to modern European on the Karitiana axis?

MA-1 is a mix between an Amerindian and a West Eurasian populations(s). I've told you this a gazillion times.

Actually you repeatedly state that MA-1 *IS* Amerindian - it's me who has to repeatedly remind you that in fact only a small part of their DNA is shared.

Of course, they are not the same on all charts but they are the same for all practical purposes in their Amerindianness.

Perhaps in their "Amerindianness", but it's the majority non-"Amerindianness" that explains LB's (and everybody else's) differing positions on EDF 5a and 5d. It seem to me like your obsession with MA-1's minority Amerindian component has blinded you to more obvious and parsimonious explanations.

there's no contradiction between Papuans and Europeans sharing the same amount of Amerindian descent and an exorbitant amount of their specific continental "modifications."

So you accept that two populations can share the same f3 score, yet not show any elevated genetic affinity with each other in a direct measurement?

German Dziebel said...

@Tobus

"All West Eurasians plot together on all axes and in all other genetic measurements and they all live in the same geographic area and temporal period. MA-1 doesn't plot with Amerindians on any axis, nor in any other genetic measurements and he lived on a different continent 24,000 years ago... by what logic are you proposing these are the same?"

By the logic called "evolutionary." Amerindians are the rightmost on the MA-1 axis. They are shifted toward Amerindians with respect to West Eurasians on all other plots. They are closer to the New World geographically than European populations. Populations are related to each other in evolutionary terms. But in your "polygenist" world all populations are locked into their respective geographic corners and are defined by fixed "genetic distance" from each other. A typical creationist position applied not to species but to human populations.

"Then you guess wrong."

Of course, you're going to deny it. But if you deny the admixed nature of MA-1, then you need to come up with a real modern geographic cluster with which it shares an ancestral relationship. If it's not Amerindian, or West Eurasian, or Papuan, or East Asian, then it must be UFO-derived.

"All I'm doing is describing the data - how does LB plot relative to modern European on the Karitiana axis?"

Well, you are describing it wrongly. And I give you the logical outcome of your description. It does look like a caricature to you and this is exactly how your "logic" looks to me. You define an ancient sample as "belonging" to a modern cluster. The fact that LB is not plotted with the rest of ancient European samples shouldn't allow you to interpret it as part of a modern cluster. An MA-1-anchored axis clearly shows a pull away from that cluster toward Amerindians and MA-1.

"Actually you repeatedly state that MA-1 *IS* Amerindian - it's me who has to repeatedly remind you that in fact only a small part of their DNA is shared."

MA-1 is ancestrally Amerindian with a heavy admixture from a West Eurasian population. Just like many modern Native American populations. What's not clear about it?

" It seem to me like your obsession with MA-1's minority Amerindian component has blinded you to more obvious and parsimonious explanations."

Nonsense.

"So you accept that two populations can share the same f3 score, yet not show any elevated genetic affinity with each other in a direct measurement?"

No, I can't accept your creationist argument. What I accept is that populations that share the same f3 score on an Amerindian axis share deep Amerindian descent, while exhibiting later continental modifications (divergence, admixture).

Tobus said...

@German:
By the logic called "evolutionary."... Populations are related to each other in evolutionary terms.

The logic that lets everyone agree that the French, Sardinians, Russians etc. all form a "cluster" on the Sardinian axis just isn't comparable to this "evolutionary" logic you are using to suggest a similar relationship between Amerindians and a 24,000 Eurasian sample that they share less than 30% of their genome with. I get your reasoning, it's just so far-fetched an inconsistent - East Asians would be a much better proxy for Amerindians than MA-1, not that we need one anyway since we have Karitiana in the data.

But if you deny the admixed nature of MA-1, then you need to come up with a real modern geographic cluster with which it shares an ancestral relationship

MA-1 has an ancestral relationship with modern (and ancient!) Europeans, Central/South Asians and Amerindians... no need for UFOs!

Well, you are describing it wrongly.

Then can you please describe to me how LB plots in relation to modern Europeans in EDF 5a?

A-1 is ancestrally Amerindian with a heavy admixture from a West Eurasian population.

No, read Raghavan again - MA-1 is ancestrally Eurasian and gave genes to both modern Europeans and Amerindians.

No, I can't accept your creationist argument.

What's creationist about it?

What I accept is that populations that share the same f3 score on an Amerindian axis share deep Amerindian descent

Do you therefore accept that ALL populations than get the same f3 score on ANY axis share deep descent from that axis? Or is this a special case you made up to support your prior assumptions?



German Dziebel said...

@Tobus

"The logic that lets everyone agree that the French, Sardinians, Russians etc. all form a "cluster" on the Sardinian axis just isn't comparable to this "evolutionary" logic you are using to suggest a similar relationship between Amerindians and a 24,000 Eurasian sample that they share less than 30% of their genome with."

That's precisely creationist logic applied to modern populations. See https://www.youtube.com/watch?v=z6kgvhG3AkI at about 40:30: dogs were always dogs, finches always finches. Creationists accept observational science but reject evolutionary science.

Meanwhile, it's perfectly natural for two populations that diverged 24,000+ years ago to share less than 30% of genes. 30% is the old 100%.

"MA-1 has an ancestral relationship with modern (and ancient!) Europeans, Central/South Asians and Amerindians... no need for UFOs!"

if MA-1 didn't come from either Europe, S/C Asia or America but spawned all of these populations, then it must be the UFO...

"Then can you please describe to me how LB plots in relation to modern Europeans in EDF 5a?"

Modern Europeans are close to a 7,000 year old Western European sample in the degree of Amerindian affinity. Or, modern Europeans are closer to a 7,000 year old Western European sample in the degree of Amerindian affinity than to 24,000 year old South Siberian sample.

"Do you therefore accept that ALL populations than get the same f3 score on ANY axis share deep descent from that axis? Or is this a special case you made up to support your prior assumptions?"

Sure, but then you compare the maximum f3 scores from one axis to the maximum f3 scores on another axis. Amerindians are more divergent on their axis than West Eurasians are on theirs or East Asians on theirs. So, if we were to collapse the two axes into one, we would end up with Amerindians diverging first, followed by everyone else. The only caveat is that East Asians may get their higher Amerindian score from Holocene admixture from North America.

terryt said...

"Creationists accept observational science but reject evolutionary science".

It seems you have a very poor understanding of evolutionary science because you have a unique version of how evolution proceeds. As shown by:

"Meanwhile, it's perfectly natural for two populations that diverged 24,000+ years ago to share less than 30% of genes. 30% is the old 100%".

It would not be 'perfectly natural' at all. Unless we have either a prodigious level of mutation or an extreme level of drift in both populations. You are ignoring the fact that most genetic mutations are either neutral or actually harmful and so mutation as an evolutionary driver is not particularly rapid. On the other hand drift leads to a reduction of genetic diversity and so we would see such a reduction in one or other of the populations under consideration. Where is the evidence for such drift in East Asians, West Eurasians or Amerindians?

A change that gave rise to a residual genetic similarity of just 30% between any two populations over such a time period as you claim here could only be the product of admixture. And another example of you lack of understanding when it come to evolutionary science:

"if MA-1 didn't come from either Europe, S/C Asia or America but spawned all of these populations, then it must be the UFO..."

MA-1 was found in South/Central Eurasia and its presence there is very unlikely to have been the product of a UFO. Its history before the period of its existence does not include either America or Europe yet the population it was part of has made a genetic contribution to both regions. I would have thought that was perfectly obvious and straightforward for anyone with even the slightest understanding of evolutionary biology to see.

"The only caveat is that East Asians may get their higher Amerindian score from Holocene admixture from North America".

For which there is not the slightest bit of evidence.

Tobus said...

@German:
Meanwhile, it's perfectly natural for two populations that diverged 24,000+ years ago to share less than 30% of genes. 30% is the old 100%.

Yes, but it's not perfectly natural to then assume that these two populations are the same for all practical purposes. MA-1 is not an Amerindian, pure and simple. If you want to measure affinity to Amerindians, use the Karitiana axis, not the MA-1 axis.

if MA-1 didn't come from either Europe, S/C Asia or America but spawned all of these populations, then it must be the UFO...

MA-1 didn't "spawn" all these populations, he shares varying degrees of common ancestry with all West Eurasians and admixed into an extant proto-Amerindian population, and possibly into some of the Eurasian lineages as well. It seems likely that both LB's and Stuttgart's MA-1 affinity is all from a common ancestor, with no gene flow from MA-1's lineage after their divergence (see the f3 tree in the latest Skoglund for instance).

Modern Europeans are close to a 7,000 year old Western European sample in the degree of Amerindian affinity.

Can you give a touch more detail - does the 7,000yo have more or less Amerindian affinity than modern Europeans do? Or is he somewhere within the range of modern Europeans (less then some, more than others)?

Sure,

So you are saying that:
Amerindians and East Asians share deep European ancestry? (5a Y)
C/S Asians and Middle Easteners share deep European ancestry? (5a Y)
Europeans and Central/South Asians share deep Han affinity? (5b X)
Amerindians and C/S Asians share deep La Brana ancestry? (5c Y)
Papuans and Middle Easteners share deep MA-1 ancestry? (5c X)
East Asians and C/S Asians share deep MA-1 ancestry? (5c X)

Amerindians are more divergent on their axis than West Eurasians are on theirs or East Asians on theirs.

Outgroup f3 doesn't measure divergence, it measures shared drift. Amerindians have more shared drift with each other than West Eurasians/East Asians do, not divergence from a common ancestor (in fact the opposite make more sense, the *closer* two populations are to their common ancestor the higher shared drift they will have - these f3 indicate Amerindians have been a discrete population for *less* time than any of the other samples).

So, if we were to collapse the two axes into one

We can't do this, the axes aren't correlated on any common baseline.

German Dziebel said...

@Tobus

"Yes, but it's not perfectly natural to then assume that these two populations are the same for all practical purposes. MA-1 is not an Amerindian, pure and simple."

MA-1 is an Amerindian population shifted toward West Eurasians due to added West Eurasian ancestry.

"MA-1 didn't "spawn" all these populations, he shares varying degrees of common ancestry with all West Eurasians."

MA-1 is closer to Amerindians than to West Eurasians, as Raghavan's data shows. It gotta belong to some kind of cluster (or be closer to one modern continental cluster than to others), otherwise you're postulating a ghost population brought about by a UFO landing.

"Can you give a touch more detail - does the 7,000yo have more or less Amerindian affinity than modern Europeans do? "

It has more Amerindian affinity than modern Europeans, although we only have LB and MA-1 displayed on those plots, hence we don't have a full view of how ancient Eurasian samples square off against modern Europeans in terms of Amerindian affinity.

"So you are saying that:
Amerindians and East Asians share deep European ancestry? (5a Y)
C/S Asians and Middle Easteners share deep European ancestry? (5a Y)
Europeans and Central/South Asians share deep Han affinity? (5b X)
Amerindians and C/S Asians share deep La Brana ancestry? (5c Y)
Papuans and Middle Easteners share deep MA-1 ancestry? (5c X)
East Asians and C/S Asians share deep MA-1 ancestry? (5c X)."

And you're saying that MA "MA-1 didn't "spawn "shares varying degrees of common ancestry with all West Eurasians [plus Papuans and South Asians, no?) and admixed into an extant proto-Amerindian population."

"Amerindians have more shared drift with each other than West Eurasians/East Asians do."

No, Amerindians have more shared Amerindian drift (ancestry) with each other than West Eurasians and East Asians. In other words, West Eurasians and East Asians are less Amerindian than Amerindians.

"these f3 indicate Amerindians have been a discrete population for *less* time than any of the other samples)."

Nonsense. It's like saying that Amerindians have been Amerindians for less time than any other populations have been Amerindian. You are still not getting the absurdity of your argument - a discrete Amerindian component has begun to differentiate at a time when Europeans, Middle Easterners and Papuans have first diverged from Africans and the populations have carried it around for tens of thousands of years in a test-tube before spilling it over in Beringia.

terryt said...

"MA-1 is an Amerindian population shifted toward West Eurasians due to added West Eurasian ancestry".

How can it possibly be 'shifted toward West Eurasians due to added West Eurasian ancestry' if modern humans all came from America? What population was present that enabled it to be 'shifted' at all? Consistent inconsistency once more?

"In other words, West Eurasians and East Asians are less Amerindian than Amerindians".

Of course they are. West Eurasians are less Amerindian than Amerindians because Amerindians are admixed with East Asians, and East Asians are less Amerindian than Amerindians because Amerindians are admixed with West Eurasians. Simple. Or at least I would have thought anyone with multiple degrees, even if not in genetics or even biology, would have found it so. But it appears your creationist belief system prevents you from seeing the obvious.

Tobus said...

@German:
MA-1 is an Amerindian population shifted toward West Eurasians due to added West Eurasian ancestry.

Read Raghavan again - the evidence shows that MA-1 is a Eurasian population that gave DNA to Amerindians, not the other way around. I thought your theory was that MA-1 was the first European anyway - the "Amerindian ancestor" of all Eurasians, no?

It has more Amerindian affinity than modern Europeans

In EDF 5a? Are we looking at the same plot? Doesn't the 7,000yo Western European sample fall *inside* the range shown by modern Europeans on that graph?

And you're saying that MA shares varying degrees of common ancestry with all West Eurasians [plus Papuans and South Asians, no?) and admixed into an extant proto-Amerindian population

The point I was making was that you are applying rationale to one set of data that produces nonsensical results when applied to rest of the data (same f3 score = shared deep ancestry). If you are going to draw a conclusion from one graph, please take the time to ensure it holds true in all the other graphs as well.

To answer your question, all populations share varying degrees of shared ancestry with all the other populations. In terms of MA-1 he has a fair bit of shared ancestry with South Asians (the same level as many Europeans have), but much less with Papuans. This is consistent with Papuans diverging from MA-1's lineage earlier than South Asians and Europeans.

In other words, West Eurasians and East Asians are less Amerindian than Amerindians.

That goes without saying, my point is that we'd expect all non-African groups to have roughly similar drift since divergence from Yoruba, but Amerindians appear to have significantly more than everybody else. We can rule out the idea of Amerindians diverging earlier than Yoruba because then *they'd* be the outgroup and these f3 would be negative. One possible explanation might be that Amerindians contain both East Asian and European ancestry/admixture and so have the combined drift from both lineages, giving them a shared drift component larger than any single branch has.

German Dziebel said...

@Tobus

"Read Raghavan again - the evidence shows that MA-1 is a Eurasian population that gave DNA to Amerindians, not the other way around."

Raghavan made a wrong conclusion from the data. The good thing is that he actually published what the data actually shows. MA-1 shares genetic ancestry with Amerindians before Eurasians.

"In EDF 5a? Are we looking at the same plot? Doesn't the 7,000yo Western European sample fall *inside* the range shown by modern Europeans on that graph?"

Since the past happens before the present, we're looking at EDF 5d first to determine what falls into what. LB falls into a modern Amerindian range, while modern Europeans don't. This means LB has more Amerindian ancestry than modern Europeans. Then we look at EDF 5a and observe that a couple of modern European populations entered the "orbit" defined by the Amerindian-shifted LB. From this we conclude that they must carry later (post 7,000 years) Amerindian-shifted admixture, which may have come from Siberia/East Asia. And those populations are Finnish and Russian, so it all makes sense.

"We can rule out the idea of Amerindians diverging earlier than Yoruba because then *they'd* be the outgroup and these f3 would be negative."

No, you shouldn't rule it out. Amerindians did indeed diverge before Yoruba but Yoruba admixed with extinct "archaic" Africans. It's those that had diverged before Amerindians.

"East Asian and European ancestry/admixture and so have the combined drift from both lineages, giving them a shared drift component larger than any single branch has."

Impossible, as this would have driven Amerindian heterozygosity up, not down.

terryt said...

"Raghavan made a wrong conclusion from the data".

My humble friend. You have just absolutely proved we have no need to anything you say seriously. Your creationist mind-set is absolute. As demonstrated by:

"Impossible, as this would have driven Amerindian heterozygosity up, not down".

That would certainly not be the case if the Amerindian ancestors had undergone a population bottleneck before entering America. Of course you refuse to even consider such a possibility because it goes against your creationist belief. From the interpretations of the data you have consistently offered we can only conclude that you obviously believe humans were first created from nothing in America. And when they moved into Eurasia they then mixed with what can only be described as non-existent populations in Western Eurasia and in China before reaching Africa. Where they mixed with chimpanzees. I am absolutely certain you are the only individual who sees our prehistory in that manner. You are only able to do so because you are determined to ignore the evidence haplogroup diversity and distribution provide as well as the pattern of the EDAR370A mutation's geographic spread. Even the data you quote as 'proof' can just as easily be interpreted in alternative ways to what you insist is the only way.

Tobus said...

@German:
Raghavan made a wrong conclusion from the data

The evidence is fairly conclusive German, no matter how you spin it.

LB falls into a modern Amerindian range, while modern Europeans don't. This means LB has more Amerindian ancestry than modern Europeans

As I've pointed out I don't know how many times now, EDF 5d measures shared drift with MA-1 *not* Amerindians. EDF 5d cannot tell us if the X populations have more or less Amerindian than each other, only if they have more or less MA-1 with each other. This plots shows LB has more MA-1 than modern Europeans.

Then we look at EDF 5a and observe that a couple of modern European populations entered the "orbit" defined by the Amerindian-shifted LB.

This plot shows not only that LB has the same amount of shared drift with Amerindians that modern Europeans have, but also that LB has *less* shared drift with Amerindians than MA-1 does. This proves that LB's higher score in EDF 5d cannot be completely accounted for by Amerindian affinity alone.

No, you shouldn't rule it out. Amerindians did indeed diverge before Yoruba but Yoruba admixed with extinct "archaic" Africans. It's those that had diverged before Amerindians.

That wouldn't change the f3 results, they only measure alleles in common, not private alleles (and if Yoruba had lost most of their common "Asian" alleles in the admixture they'd have reduced, not increased, heterozygosity so we know this can't be the dase). Even with archaic admixture, Africans would still have a higher shared drift with non-Amerindians if your suggestion were correct.

Impossible, as this would have driven Amerindian heterozygosity up, not down

Amerindian heterozygosity *is* up from ancient levels (such as what we see in Loschbour.).. plus we know Amerindians underwent a bottleneck after the admixture event which has made their heterozygosity lower than their modern relatives.

German Dziebel said...

@Tobus

"The evidence is fairly conclusive German, no matter how you spin it."

Yes it's conclusive that MA-1 is an Amerindian-derived population and Amerindian admixture is found all over Eurasia.

"As I've pointed out I don't know how many times now, EDF 5d measures shared drift with MA-1 *not* Amerindians. EDF 5d cannot tell us if the X populations have more or less Amerindian than each other, only if they have more or less MA-1 with each other. This plots shows LB has more MA-1 than modern Europeans."

Your objections are invalid. EDF 5d shows affinities to both Amerindians and MA-1 because MA-1 is an Amerindian derived population.

"This plot shows not only that LB has the same amount of shared drift with Amerindians that modern Europeans have, but also that LB has *less* shared drift with Amerindians than MA-1 does. This proves that LB's higher score in EDF 5d cannot be completely accounted for by Amerindian affinity alone."

OK, you admit that LB has Amerindian affinity. Good. Science can truly do magic with people! You can also see that LB has more MA-1 and Amerindian affinity than modern Europeans. It's pulled out of the West Eurasian cluster and placed into the Amerindian cluster. If MA-1 were a pure West Eurasian population, you would have seen all modern West Eurasians forming a cluster with LB and MA-1 in the rightmost corner of the axis. Amerindians would've been in the middle on the chart. This is not the case. In reality Amerindians claim unique affinity to MA-1 that continues in LB, to a lesser degree in modern Europeans and to still lesser degree in modern West Asians. Of course, shared drift with Amerindians decreases with distance from the New World precisely because other shared drift clusters have emerged since the founding out-of-America event. MA-1 participated in at least one other major shared drift cluster, which is West Eurasian, hence it's closer to LB than Amerindians are. Amerindians are pure, MA-1 is admixed.

"Amerindian heterozygosity *is* up from ancient levels (such as what we see in Loschbour.).. plus we know Amerindians underwent a bottleneck after the admixture event which has made their heterozygosity lower than their modern relatives."

Loschbour further proves that Amerindian level of homozygosity is an ancient trait. Heterozygosity goes up everywhere but in America it has gone up at a slower rate than elsewhere. Amerindians didn't undergo any bottleneck. You just invented a bottleneck after an admixture event from two heterozygous populations just make your idea work. I call it pseudoscience. And for a good reason.

terryt said...

"EDF 5d shows affinities to both Amerindians and MA-1 because MA-1 is an Amerindian derived population".

That has got to be the most circular logic I have ever encountered. You start with an assumption, then use that assumption to 'prove' the original assumption. I thought you claimed to be intelligent.

" If MA-1 were a pure West Eurasian population"

No-one has ever claimed MA-1 is 'a pure West Eurasian population' except in your imagination. Central Eurasian, that contributed genetically to the West Eurasian population: yes. And, incidentally, contributed genetically to the Amerindian population. That is obvious for even the most cursory examination of haplogroups.

German Dziebel said...

@TerryT

"That has got to be the most circular logic I have ever encountered."

You are the most circular reader my logic has ever encountered. Look at the data: Amerindians are the most MA-1 population on EDF 5d. The only was this can be explained is that MA-1 is an Amerindian-derived population. Otherwise, West Eurasians would have been the most MA-1 like population.

terryt said...

"Amerindians are the most MA-1 population on EDF 5d. The only was this can be explained is that MA-1 is an Amerindian-derived population".

I agree that is 'the only' explanation if you have already made up your mind MA-1 is Amerindian. If you are prepared to consider all the possible options it become immediately obvious that your interpretation is not the 'only' one.

"Otherwise, West Eurasians would have been the most MA-1 like population".

Complete rubbish, as usual. Seems you are more than happy to accept non-existent non-Amerindian populations in Eurasia that MA-1 has bred with, and other non-existent populations that East Asians bred with yet are unable to accept actual documented other populations that involved extremely likely inter-breeding events. On what grounds do you insist on taking that position?

Tobus said...

@German:
Yes it's conclusive that MA-1 is an Amerindian-derived population and Amerindian admixture is found all over Eurasia.

Funny, every expert in the field comes to the exact opposite conclusion from that very same data.

EDF 5d shows affinities to both Amerindians and MA-1 because MA-1 is an Amerindian derived population.

That's funny, because it shows very different affinities than does the Karitiana axis.

You can also see that LB has more MA-1 and Amerindian affinity than modern Europeans.

He only has more affinity on the MA-1 axis, not on the Karitiana axis.

Loschbour further proves that Amerindian level of homozygosity is an ancient trait.

Quite the contrary, he proves that Amerindian heterozogisity is *higher* than an ancient level.

Amerindians are the most MA-1 population on EDF 5d. The only was this can be explained is that MA-1 is an Amerindian-derived population.

Funny, I've already provided an alternative explanation to that - basal affinity as opposed to direct ancestry.






German Dziebel said...

@Tobus

"Funny, every expert in the field comes to the exact opposite conclusion from that very same data."

Experts are mortals. If people have believed in something for a very long time, they can't walk away from it in a split second.

"That's funny, because it shows very different affinities than does the Karitiana axis."

They are not different as in "contradictory," they are different as in "mutually consistent." MA-1 is as removed from Karitiana on the Karitiana axis as Karitiana is plotted on the MA-1 axis. Kariatiana is rightmost on both. All Eurasians are Karitiana and MA-1 shifted. MA-1 pulls all of West Eurasians, especially LB to the Karitiana pole. That's all.

"He only has more affinity on the MA-1 axis, not on the Karitiana axis."

But since MA-1 is Amerindian derived, this is an effect of a couple of modern West Eurasians moving into the LB orbit in post-MA-1 times due to secondary East Asian gene flow.

"Quite the contrary, he proves that Amerindian heterozogisity is *higher* than an ancient level."

They are close enough. There may be another modern Amerindian population that's more homozygous than Loschbour. These are details. What's important is that heterozygosity grows with time, Amerindians are much slower in picking up heterozygosity than Europeans or especially Africans and Amerindians is the best example from modern human populations of the levels of heterozygosity found among archaic hominins such as Neandertals and Denisovans.

"Funny, I've already provided an alternative explanation to that - basal affinity as opposed to direct ancestry."

So, MA-1 is "basally" related to Amerindians? I thought you argued for direct West Eurasian ancestry in modern Amerindians.

terryt said...

"If people have believed in something for a very long time, they can't walk away from it in a split second".

So I've noticed.

Tobus said...

@German:
If people have believed in something for a very long time, they can't walk away from it in a split second.

You are living proof of that.

They are not different as in "contradictory," they are different as in "mutually consistent."

East Asians, C/S Asians, Europeans, LB etc. all get different MA-1 scores (EDF 5d) to their Amerindian scores (EDF 5a). The two are clearly not the same.

this is an effect of a couple of modern West Eurasians moving into the LB orbit in post-MA-1 times due to secondary East Asian gene flow.

Whatever the reason, LB plots inside the European cluster against Karitiana. We know his MA-1 affinity can't be due to shared Karitiana alleles because he doesn't have as many Karitiana alleles as MA-1 does yet he gets the same score as them - it has to be due to some other aspect of MA-1... and with Amerindian out of the picture there's only West Eurasian left.

They are close enough. There may be another modern Amerindian population that's more homozygous than Loschbour.

And there also may be other Eurasians more homozygous than Loschbour - particularly those around 24kya in Sibieria. An admixture of two of these plus a bottleneck (more recently confirmed in the Skiffles paper) could easily lead to modern Amerindian heterozyosity levels.

So, MA-1 is "basally" related to Amerindians?

No, it's West Eurasians and MA-1 who share a basal relationship, Amerindians have direct ancestry from the admixture event. Although Amerindians have less overall affinity to MA-1, their shared drift score is higher because the affinity they do have is from a branch much closer to MA-1 than the common ancestor he shares with West Eurasians.


German Dziebel said...

@Tobus

"You are living proof of that."

A sample of one is your method. A sample of millions is mine.

"East Asians, C/S Asians, Europeans, LB etc. all get different MA-1 scores (EDF 5d) to their Amerindian scores (EDF 5a). The two are clearly not the same."

Sure, MA-1 is West Eurasian-admixed. While Amerindians (and East Asians) are not.

"LB plots inside the European cluster against Karitiana. We know his MA-1 affinity can't be due to shared Karitiana alleles because he doesn't have as many Karitiana alleles as MA-1 does yet he gets the same score as them - it has to be due to some other aspect of MA-1... and with Amerindian out of the picture there's only West Eurasian left."

LB does not get the same score as Karitiana on the MA-1 axis. Karitiana is further to the right and Anzick would be even further. It can't be the West Eurasian component as this would make all of modern West Eurasians more MA-1 than Amerindians. This is clearly not the case. Amerindians are more MA-1 like than West Eurasians. But LB is closer to MA-1 than other modern Europeans, so it's clear that it's ancestors of modern Amerindians that pull MA-1 and to a lesser degree LB away from modern Europeans. From your UFO theory it follows that ancestors of modern Amerindians landed in East Asia. But I prefer to place Amerindian ancestors where their descendants are historically found - in the New World and derive them from earlier hominins such as Denisovans and Neandertals. And to derive East Asians and West Eurasians from Amerindians.

LB does not plot inside the European cluster. It's the modern European cluster that's close to ancient west Eurasians of which only LB is depicted in Olalde EDF. From now on I will ignore your pseudoscientific interpretations of Olalde EDF 5 data.

terryt said...

"A sample of one is your method. A sample of millions is mine".

Completely avoiding considering evidence that contradicts your belief is your speciality. For example:

"MA-1 is West Eurasian-admixed".

What can it possibly be 'admixed' with if everyone originated in America? I keep reading your comments in the forlorn hope you might actually produce some evidence to support your belief but so far: nothing. I should stop wasting my time.

Tobus said...

@German:
It can't be the West Eurasian component as this would make all of modern West Eurasians more MA-1 than Amerindians.

Why would it? LB himself is not more MA-1 than Amerindians on that graph, so why would you expect Europeans to be if LB has *more* of the West Eurasian part of MA-1 than Europeans do.. you make no sense.

so it's clear that it's ancestors of modern Amerindians that pull MA-1 and to a lesser degree LB away from modern Europeans.

Well MA-1 was never European to begin with, so needs no "pull", and LB is related to MA-1 much more than he is to Amerindians, so the "pull" to MA-1 may have nothing at all to do with Amerindians ancestors, beyond that fact that MA-1 himself *is* a partial ancestor to modern Amerindians.

LB does not plot inside the European cluster.

LB = black star. Europeans = blue dots. Black star is inside the blue dots. Go figure.

German Dziebel said...

@Tobus

"Why would it? LB himself is not more MA-1 than Amerindians on that graph, so why would you expect Europeans to be if LB has *more* of the West Eurasian part of MA-1 than Europeans do.. you make no sense."

In another string you claim that MA-1 is a West Eurasian population genetically. This means that West Eurasians should be closer to MA-1 than Amerindians. But they are not.

"Well MA-1 was never European to begin with, so needs no "pull", and LB is related to MA-1 much more than he is to Amerindians, so the "pull" to MA-1 may have nothing at all to do with Amerindians ancestors, beyond that fact that MA-1 himself *is* a partial ancestor to modern Amerindians."

What kind of construct is "partial ancestor"? Is it Tobusian for "admixed population"? But apart from that LB is indeed related more closely to MA-1 than to Amerindians, which is perfectly consistent with MA-1 being older, hence closer to Amerindians, and in geographical proximity to America and LB being all the way in Europe and younger, hence more removed from Amerindians.

"LB = black star. Europeans = blue dots. Black star is inside the blue dots. Go figure."

LB = black star. MA-1 = green star. Blue dots are a subset of the Amerindian-shifted ancient Eurasian "cluster" staked out by the two samples available at the time of study. Figured out!

Tobus said...

@German:
This means that West Eurasians should be closer to MA-1 than Amerindians. But they are not.

They are as a group, see the PCA and ADMIXTURE results. It's only when you subdivide "West Eurasians" into separate population groups that Amerindians appear closer. MA-1 is still over 70% non-Amerindian. I've already explained one scenario that could account for this, I'm sure your extensive imagine could come up with other scenarios that fit the wider data as well.

What kind of construct is "partial ancestor"? Is it Tobusian for "admixed population"?

No, it's the opposite - it means one of the ancestral lineages of an admixed population... a bit like how your mother's side is part of your ancestry and your father's side is another. MA-1 is part of Amerindian ancestry, a Paleo-"East Asian" population is another.

Blue dots are a subset of the Amerindian-shifted ancient Eurasian "cluster" staked out by the two samples available at the time of study.

Whatever your definition, the black star is *inside* the cluster blue dots, not outside it.

terryt said...

German, why don't you just start a discussion with Dr. Clyde Winter? The two of you deserve each other.

«Oldest ‹Older   201 – 338 of 338   Newer› Newest»