search this blog

Friday, January 6, 2017

The Metal Age invader that never was #2


I just figured out how the Metal Age invader that never was came about. Take a look at this slide from an FTDNA presentation by Dr. Michael Hammer.


Hammer's taken the ADMIXTURE analysis from Lazaridis et al. 2013 and labeled the clusters according to what he thinks they represent. FTDNA then basically copied the analysis and the cluster labels for their "Ancient Origins" test.

But those clusters do not represent what Hammer thinks they do, in other words ancient populations like Western Hunter-Gatherers (WHG) or Ancient North Eurasians (ANE). Ancient ancestry proportions are listed in other parts of the Lazaridis et al. paper, like table S.14.10.

Moreover, Hammer appears to be under the impression that the Metal Age invader (that never was) came from Asia, and today it is the Kalasha people of the Hindu Kush who carry the highest ancestry proportion from this ancient population (that, of course, never actually was).

First of all, the Bronze Age Yamnaya pastoralists and related groups (i.e. the real Metal Age invaders) expanded both into Europe and Asia from the Pontic-Caspian Steppe. Obviously, the Pontic-Caspian Steppe is located in what is now known as Eastern Europe, not Asia.

Secondly, the Kalasha people carry about the same amount of Yamnaya-related ancestry as present-day Northern and Eastern Europeans, not three or four times as much, which, as per the figure above, is what Hammer seems to think.

Formal stats-based Yamnaya-related (or Steppe_EMBA) ancestry proportions for Europeans are listed in Haak et al. 2015 and for South Asians, including the Kalasha, in Lazaridis et al. 2016. See below.



See also...

Who's your (proto) daddy Western Europeans?

Late PIE ground zero now obvious; location of PIE homeland still uncertain, but...

21 comments:

Onur Dincer said...

Thanks for the insightful input, Davidski. As an FTDNA project admin, I am going to mention this blog post to the FTDNA staff.

Onur Dincer said...

If you have any more suggestions, feel free to tell us.

Davidski said...

Well, I think FTDNA needs to pull that test and Michael Hammer should edit his slide show presentation in a few places in time for the next FTDNA conference.

Onur Dincer said...

If you were to design a similar test, how would your modeling be (if you want, you can write your answer to me by email or directly to FTDNA in form of suggestion)?

Davidski said...

Something like this...

http://eurogenes.blogspot.com.au/2015/03/yamnaya-related-ancestry-proportions-in.html

Except with Anatolian and Iranian farmer clusters, as opposed to "Pre-Yamnaya" and "Middle Eastern".

Onur Dincer said...

So you propose a replacement of the EEF cluster with an Anatolian farmer cluster and addition of an Iranian farmer cluster.

Davidski said...

Yes, I've already experimented with that type of model, and it does seem to work, but I haven't had time to perfect it yet.

Onur Dincer said...

Thanks for your input again. I will share this blog entry along with your suggestions in the comments section with the FTDNA staff.

truth said...

The Haak et al. graphic bars seems bullshit to me, I am Spaniard and when trying to emulate with nMonte Stats with the same populations used (LBK_EN, Loschbour, Yamnaya) I get about 20% Loschsbour, not 0% as the graph suggests, and I'm not basque at all.

Davidski said...

Models based on f4 stats like in that Haak graph show more "southern" ancestry, like EEF for modern Europeans and Armenian-related for Yamnaya. PCA and Admixture based models give more Euro hunter-gatherer admixture.

I don't think there's a perfect method to do this, but it's important to at least be in the ballpark.

Gioiello said...

Another fake study from Levantinists (Daniel E. Platt et al., Mapping Post-Glacial expansions: The Peopling of Southwest Asia )
They believed to demonstrate that there were three refugia in Southwest Asia (one, in the north, centered around the eastern coast of the Black Sea, the second, with a more Levantine focus, and the third in the southern Arabian Peninsula), but they, after their analysis, had to admit, above all for the believed Middle Eastern haplogroups (J1 and J2), what we know from so long, that the expansion happened from Caucasus, and hope that they will understand what I am saying from so long, that they were before the hgs of the European hunter-gatherers. Unfortunately for them aDNA didn't please their desires (and prejudices). If actually Indo-Europeans and hg. R1b came from Yamnaya (what I don't think) will see next.
I'd be glad that they demonstrated that only 1 R1b1-M18 is in one of these places as in their supplement:
Marker Zalloua Lab Ferri 2010 Arredi, 2004 Robino Cinnioglu Zalloua Lab Zalloua Lab Bosch Battaglia Battaglia Balanovsky Battaglia Zalloua Battaglia Zalloua lab Battaglia ElSibai2009, thi study Chiaroni, Luis 2004 de Filippo Tofanelli Battaglia Zalloua Lab Zalloua Lab Tofanelli Chiaroni Ferri 2008 Tofanelli Battaglia ElSibai2009; this study Chiaroni de Filippo Mohamad Zalloua Lab Chiaroni Zalloua Lab Abu Amero Zalloua2008a,b; this study our Lab ZallouaNP Zalloua2008a,b; this study Tofanelli Tofanelli Chiaroni, Luis 2004 ElSibai2009, this study Chiaroni Adams Tofanelli Tofanelli Cadenas Battaglia Battaglia Adams Ambrozio,Hernandez Ambrozio,Dugoujon Tofanelli ElSibai2009,Haber2011, this study Chiaroni Fadhloui 2011 Zalloua2008a,b; this study Tofanelli
YHRD ISOGG 2011 Afghanistan Albania Algeria Algeria Armenia Armenia Bahrain Balkan Balkan Bosnia_Herzegovina Caucasus Caucasus Chad Croatia Cyprus Czech Republic Egypt Egypt Ethiopia Ethiopia Hungaria Iran Iraq Iraq Iraq Italy Italy Italy Jordan Jordan Kenya Kuwait Kuwait KSA KSA KSA Lebanon Libya Malta Morocco Morocco Oman Oman Palestine Palestine Portugal Portugal Qatar Qatar Sardinia Slovenia Spain Spain Spain Sudan Syria Syria Tunisia Tunisia Tunisia
M18 R1b1a1 R1b1c1 x x x x x x x x x x x x x x x x x x x x x x x x x

AWood said...

Far too much weight is put into a 24,000 year old boy who died in Siberia. Although there is a mild similarity between he and modern, or even Yamnayan Europeans, he is not by any measure the same thing...not even close. If we're going to allege some "Metal Age Invader", let's use someone from that period to make that comparison shall we?

Samuel Andrews said...

Hammer: "Asian Steppe Invader"
Davidski: "Eastern Europe"

Eastern Europe works best but some people then might forget that most of Eastern Europe in 3300 BC was MN or WHG or SHG or EHG not Steppe. Then again Steppe ancestry in Europe at some point traveled through most of Eastern Europe.

I think Hammer may have called them Asian Steppe invaders because of the Huns and Mongols and Sycthians and Conan The Barbarian. The idea of a Steppe invader is pretty common.

Plus I think the title Asian invader exemplifies the idea their genetic input is small, that they invaded Europe for a short time in 2600 BC, didn't make settlements, and disappeared. In reality they made Europe their home and an important ancestor for many Europeans as anyone else is.

Romulus the I2a L233+ Proto Balto-Slav, layer of Corded Ware Women said...

Check this out

http://www.nature.com/articles/srep40338

looks new??

Onur Dincer said...

As I promised, I mentioned Davidski's suggestions in this blog entry to the FTDNA staff. I will let you know their answer as soon as I receive it.

Gioiello said...

@ Romulus
"Check this out
http://www.nature.com/articles/srep40338
looks new??"

Read my harsh criticism above!

Olympus Mons said...

@Gioello,
What criticism? a rant against a, in my opinion, very good scientific paper. operative word being scientific.

Just tries to assert point of origin of some haplogroups based on variance in specific places. Is this something that shouldn't be done? is it something that is not a correct scientific postulate? - Well to me it is.

As was the paper "Different waves and directions of Neolithic migrations in the Armenian Highland" a 2014 paper, that asserted that the Highest variance for several haplogroups.
this new Daniel platt paper just confirms that the 2014 paper found regarding J2 as well. However the 2014 paper also had something remarkable by asserting beyond any doubt that.... in comparison with all analyzed populations from the Near East, Europe, and Anatolia, the haplogroup R1b1a2-M269 occurs with the highest genetic variances in the western parts of the Armenian plateau, in Sasun and Salmast.

Do you know where Salmast is? - Its exactly where Shulaveri Shomu (Aratashen and Arkanashen) was! - Hummmm, who is the guy that keeps saying that R1b-M269 dispersal was the Shulaveri dispersal in 4900bc?

Anonymous said...

http://www.nature.com/articles/srep40338

The key thing about this study so far as I can see is that it is about the genetic history of modern SW Asians. Not the world. I am also not a big fan of STR studies, the mutations are too labile for deep ancestry studies in my opinion.

I found the paper interesting but I could not see how they came to the conclusion that there was any kind of southern Arabian refugium or Levantine refugium. It did not fit their own data. Specifically the trees in Figure 2. The autosomal data is not displayed or properly discussed IMO.

From Figure 2.
J1 was in the Caucasus in the near East neolithic according to their data. Arabia and the Levant are very late off shoots.`Ancient DNA puts J1 in Hungary in 5500 BCE but it this is not apparent from their data, which admittedly is focused on modern SW Asia family lines. So a forgivable omission. It just means no J1 went back to SW Asia.

J2 (of more interest in European genetics) was also shown in the Caucasus on the early Near Eastern neolithic and was in North Africa 6kya. No mention of Arabia and the Levantine line is amongst the youngest. J2 was also in France by 4500BCE (aDNA) so clearly it had already hit the Mediterranean genetic spin cycle 5k years before the supposed 1k European age in SW Asian populations. I am going to guess that these 1k year old European J2 lines in SW Asia came back into SW Asia with the first crusades.

All SW Asian E1b1b1 is shown as originating in North Africa prior to the Near Eastern neolithic and expanding (Natufians etc) into pretty much everywhere from there. SW Asian "Ëuropean" lines are listed as 10k years. It is not clear to me if this means that European E1b1b1 has been diverging in Asia for 10 thousand years or if more lines migrated into SW Asia reflecting 10k years of mutation elsewhere. My guess is both scenarios are true for this haplogroup.
Arabian and Levant E1b lines in are amongst the youngest in this group also.

Figure 3
The autosomal k3 analysis seems to be that basis of the refugium ideas but it is based on modern populations and why did they choose K3? It seems completely arbitrary. Plus it is all based on a limited list of population choices. The absence of north africa/Egypt as an option is glaring especially in view of the E1b results. Clearly the haplogroup and autosomal data points to A LOT of Caucasian ancestry in the modern middle east (about half for everyone but some arabians). I found that surprising.

In summary
So if you focus on SW Asian ancestry ONLY, their haplogroup tree results are believable for these three haplogroups. But so far as I can tell the 3 refugium idea is unfounded and unsupported.

Chewie said...

This might be a silly question but how is it explained that the Basque samples have little (in the first graph) to low ( as in the second graph) yamnaya ancestry but a high incidence of Y haplogroups R1b? I have previously read that R1 haplogroups are thought to have spread with indo European languages and from the steppe, however this doesn't seem to match the ancestry tests?

Is it some type of founder effect?

I am eager to hear an answer!

Thanks

Davidski said...

Yes, the R1b in Basques generally shows low diversity and appears to be from a strong founder effect.

Unknown said...

> expanded both into Europe and Asia from the Pontic-Caspian Steppe. Obviously, the Pontic-Caspian Steppe is located in what is now known as Eastern Europe, not Asia.

Didn't they expand from Mongolia to Pontic-Caspian Steppe first?