search this blog

Saturday, August 6, 2016

Yamnaya dogs (?)

Just in at bioRxiv:

Abstract: Europe has played a major role in dog evolution, harbouring the oldest uncontested Palaeolithic remains and having been the centre of modern dog breed creation. We sequenced the whole genomes of an Early and End Neolithic dog from Germany, including a sample associated with one of Europe’s earliest farming communities. Both dogs demonstrate continuity with each other and predominantly share ancestry with modern European dogs, contradicting a Late Neolithic population replacement previously suggested by analysis of mitochondrial DNA and a Late Neolithic Irish genome. However, our End Neolithic sample possesses additional ancestry found in modern Indian dogs, which we speculate may be derived from dogs that accompanied humans from the Eastern European steppe migrating into Central Europe. By calibrating the mutation rate using our oldest dog, we narrow the timing of dog domestication to 20,000-40,000 years ago. Interestingly, the extreme copy number expansion of the AMY2B gene found in modern dogs was not observed in the ancient samples, indicating that the AMY2B copy number increase arose as an adaptation to starch-rich diets after the advent of agriculture in the Neolithic period.

And on page 17:
The age of the samples provide a time frame, between ~7,000 and 5,000 years ago, for CTC to obtain its additional Indian­like ancestry component. Considering that CTC shows similar admixture patterns to Central Asian and Middle Eastern modern dog populations, as seen in the PCA (Figure 2) and ADMIXTURE (Supplementary Figure S8.3.2.) analysis, and that the cranium was found next to two individuals associated with the Neolithic Corded Ware Culture, we speculate that the Indian­-like gene flow may have been acquired by admixture with incoming populations of dogs that accompanied steppe people migrating from the East. Moreover, ADMIXTUREGRAPH and ​ f4 statistics support the possibility that the Indian and the wolf ancestry are the consequence of the same admixture event, involving a dog population that carried the two ancestries. This scenario is further supported by the model estimated by G­PhoCS, which infers substantial migration from wolves to the lineage represented by Indian village dogs (and as much as 0.36 migration rate when Indian wolves are included in the tree (Supplementary Methods 12)).

Botigue et al., Ancient European dog genomes reveal continuity since the early Neolithic, bioRxiv, posted August 5, 2016, doi:


Karl_K said...

Obviously those dogs came out of India along with the Indo-Europeans.

That is what I am reading here.

Davidski said...

Hmmm...I may have been wrong about it all.

bellbeakerblogger said...

Interesting how AMY2B corresponds with AMY1 rise in agriculturalists. It's been suggested that something similar to LP exists in some European cats, although the mechanism might be different.

Alberto said...


Not saying that you're wrong, but:

However, SpaceMix additionally inferred around 10% ancestry in CTC (but not HXH or NGD) from the geogenetic space containing Old World wolves (Supplementary Figure S8.2.3). ​f4 statistics of the form ​f4(CTC,HXH,Wolf,Outgroup) suggest that the likely origin of this wolf component is related to contemporary Iranian/Indian wolves (those with the least amount of dog admixture, as revealed using an ​f4­ratio test and SpaceMix, Supplementary Methods 10).

Someone could also argue that Bronze Age steppe wolves migrated to Iran and India following the Indo-Europeans.

So it's not as clear cut as you see it.

ryukendo kendow said...

Interesting that the Basal position of Southeast Asian dogs is once again evinced from the data here, in common with the prior papers.

Agree with Alberto that the situation here looks complex, may well be the case that the paths taken by dogs and humans were quite different. Especially the ADMIXTUREGRAPH makes it look unlikely that this sharing is mediated by similarity of Indian dogs to European dogs. Also there is this:

"This complex pattern of admixture found in CTC is similar to that observed in many modern dog populations in Central Asia (such as Afghanistan) and the Middle East, as shown in our unsupervised clustering analyses (Supplementary Figure S8.3.2). This raises the question of whether CTC and these modern dog populations share a common admixture history and are descended from the same ancestral populations. We performed a MixMapper analysis including HXH in the scaffold tree and observed that CTC draws its European-like component exclusively from this Early Neolithic German population (consistent with the ADMIXTUREGRAPH results).
To the contrary, modern Afghanistan dogs generally demonstrate inferred ancestry from modern European village dogs."

So, after the contact zone between Indian/Iranian and European dog clades produced CTC, a further contact between later European dogs and Central Asian/Indian dogs produced the present-day dogs in Central Asia.

Its also necessary to know the status of Middle Eastern dogs, which doesn't seem well-parsed in the main text; from ADMXITURE naively they look like mixtures of Indian and European dogs already, so they may play some role as well.

Olympus Mons said...

@ all,
See. its always possible to put a spin when is in our interest (pet theory). I love it. ;)

Here, at eurogenes we always follow the DNA...up until it contradicts us! Paper is clear. There were neolithic dogs and then there was the arrival of the CWC with dogs with a considerable amount of India dogs DNA. So... hum, those R1a guys carried dogs that and Indian ancestry? wasn't there something called Out of India regarding R1a?

Olympus Mons said...

Regarding spins: How did it end that issue of the Chalcolithic Iberia component in German Bell beaker?

Olympus Mons said...

You probably don't remember, but I was the guy way back when that only wanted to talk about Grapes, Horses and.... Dogs. Remember.
That was prior to me writing my thesis regarding Shulaveri 2 bell beakers.

to the topic: There is also Mtdna dog haplogroup D. And the only dogs that belong to that haplogroup are the KANGAL Dog (ancestral isolated Eastern Anatolia!) pretty near the Caucasus, the Portuguese SERRA DA ESTRELA (ancestral isolated Portugeses dog) and the SPANISH GALGO.
The spanish Galgo is thought to be the Greyhound dog that was depicted in Ancient Egypt pictures and in the Tibesti paintings of the Tassili Mountains as the Bovidean pastoral II (white man phase).

Funny, I wrote the thesis but never did the chapters I wanted to make from the start regarding Grapes, Horses and Dogs.... those chapters are still blank in my website (

postneo said...

There two resolved wolf species in India
The are diverged and ancestral to grey wolves which in turn are ancestral to dogs

Reportedly they are differentiated from the Israeli and Central Asian wolves so any admix from such a source as well and should be well resolvable

The dendrogram does not show it tho


FrankN said...

A caveat is the following:
"We therefore utilize call set 3, with variants ascertained in New World wolves, as the primary call set for most analyses."
"We note that gene flow between New World wolves and Old World canids could potentially bias the observed genetic variation in this call set. However, previous genomic studies have reported very low to negligible migration rates only between Mexican wolves and Basenji/Dingo ​(Fan et al. 2016)and analysis of one locus under positive selection that confers black coat color suggests a potential old introgression from dogs to North American wolves (though we note that selection on standing variation cannot be ruled out) ​(Anderson et al. 2009) ​. In both cases these admixture events should have minimal impact on our analyses which primarily involve Eurasian dogs."

They seem to have overlooked Ding e.a. (2012) (missing from their literature list) that demonstrated yDNA 6, prevalent in N/C China, having undergone admixture with N. American wolves (suggesting back-migration Out of America suring the mid-holocene).

I don't know how that ascertainment bias affects their overall conclusions related to the tree phylogeny and Asian admixture. However, as in all papers I have read so far, inferred split times are too late to explain the occurrence of domesticated dogs in the Fanco-Iberian Magdalenian 18-15 ky ago by "import" from South/East Asia.

An interesting speculation is whether the Indian signal in CTC actually stems from the Zagros, which has early evidence of dog domestication (Pelagwa, ca. 12.000 BC). Of course, if you are into goat- and shepherding, dogs come in handy. The Gobustan petroglyphs have dogs appearing in Neolithic / Chalcolithic contexts, they obviously spread to the Caspian together with pastoralism.
As such, it may also be envisaged that the difference between HXH and CTC relates to different purposes of the dogs in question.

"CTC was an adult dog​ demonstrating morphological similarity to the so­called Torfhund (​Canis familiaris palustris)"
The Torfhund has been first described for LN Swiss pile dwellings, but ist oldest find was made in a LBK settlement context (Zschernitz, Saxony). It is assumed to have looked similar to the Wolfsspitz/ Keeshound, a traditional watchdog with little hunting instinct.

Balaji said...

It appears that the dog is OIT's best friend.

FrankN said...

Addendum on the "Torfhund": The first description relates to La Neuveville-Schafis on the Lake of Bienne/Biel. The settlement has been dendrodated to the 32/31st cent. BC, and belongs to the Lattrigen complex, a Western Swiss variant of the Horgen Culture. As such, appearance of the "Torfhund" predates CWC/Yamnaya.
Note also that W. Switzerland was never in earnest reached by CWC. Instead, the local early 4th. mBC Auvernier Culture shows continuity with the preceding Lattrigen and L├╝scherz complexes, and influence from the neolithique finale of the French Jura and the Saone plain.

ryukendo kendow said...

Yeah, we've been spinning so hard that we spun Dog DNA into the human genome.

Jokes aside though, this may be evidence of contact between the Indian and European (and East Asian) gene pools *for dogs* in Central Asia, generalisation into human movement is not a given. Either way, these dogs contain East Asian dog ancestry which is not found in S and C Asia either...

@ FrankN
I agree that ascertainment bias can affect stuff at the margins, but its unlikely to copernicanise the entire tree. Nick Patterson, in the paper that introduced the Human Origins snp dataset, demonstrated that running analyses on snps ascertained on different continents produced the same tree IIRC.

Its true agree that other evidence indicates dog domestication may have occurred elsewhere than in E or SE Asia, just that genetic evidence indicated the common ancestor of the presently-existing clade of Dogs probably split into a E/SEAsian clade and a clade containing all other of our waggy friends somewhere close by E/SE Asia. The phylogenetic depth of presently-existing dogs may be much shallower than that for humans, which we know is definitely the case for e.g Africa and Oceania.

ryukendo kendow said...

@ David,

I notice from your ADMIXTURE run that Chamar and UP brahmin are in the same spreadsheet, so I assume the overlap for snps should be quite good? Just to check if we can replicate the strong differential affinity to Andronovo-Sintashta-Srubanaya in the GujaratiA GujaratiD pair, could we try out this pair as well? Once again there are three sets to compare:

Mbuti X Chamar UP_Brahmin
Mbuti X Chamar Chimp
Mbuti X UP_Brahmin Chimp

I also notice another pair: Velamas and Chenchu from AP. The chenchu seem to be hunter gatherers, while the Velamas were zamindari in precolonial India, indicating relatively elevated caste status, nothing like Brahmin though. Such a pair would point out interesting differences too.

Mbuti X Chenchu Velamas
Mbuti X Velamas Chimp
Mbuti X Chenchu Chimp

Jaydeep said...

Indeed this is only one dog sample and it would be unwise to make sweeping arguments for human movement based on it.

The East Asian introgression into European Neolithic dogs is independent and earlier to the Indian dog introgression. It perhaps dates to the period when an East Asian genetic input occurred in European HG groups.

Davidski said...


Yeah, hang on.

Karl_K said...

The comments on this post are so fantastic. It is really great to hear the wisdom.

So from what I have heard, we still have very little knowledge of where and when major dog admixture events occured. We don't know when and where they were domesticated. We don't know when or where admixture with 'wild' polulations of either wolves, or feral dogs, or hybrids occured.

Lots of space for future genomic studies.

So, after reading most if the related literature, it apoears I was wrong in my earlier comment.

Not even close to being enough data, or acurate models of how domesticated species interact with humans and feral and/or wild populations.

Shaikorth said...

At least there's enough data to say the dogs of southern East Asia are the most basal ones, their position relative to other dogs in phylogenic trees (or TreeMix) rooted in various canids is similar to that of SSA relative to Eurasians in human trees. TreeMix with edges indicates this is not a product of admixture from wolves either.

ryukendo kendow said...

David, why were you comparing Iran_N with ANE in the other thread? The paper indicates EHG input into Iran Neols IIRC.

Davidski said...

I'm not seeing any EHG in Iran Neolithic or Iran Hotu. I'm seeing a sister clade to AG3 and a sister clade to Villabruna.

Any stats suggesting EHG in Iran Neolithic/Hotu are an artifact of this combination of ancestries.

Rob said...


"I'm not seeing any EHG in Iran Neolithic or Iran Hotu. I'm seeing a sister clade to AG3 and a sister clade to Villabruna.

Any stats suggesting EHG in Iran Neolithic/Hotu are an artifact of this combination of ancestrie"

One would think that's intuitively most likely scenario.
Did the Lazarides et al. paper analyse using Afontova-Gora or Mal'ta, and/ or Villabruna, or only straight EHG ?

Davidski said...

They went with the West Eurasia = quadrangle angle: Levant_N, WHG, EHG, Iran_N.

But I'm still seeing a triangle: Basal-rich, Villabruna-related, AG3-related.

I'm guessing they didn't have AG3 when they put together the bulk of their paper, and MA1 was a bit useless in this context.

Rob said...

"But I'm still seeing a triangle: Basal-rich, Villabruna-related, AG3-related."

Is there something ASI-related in Hotu ?

Davidski said...

Possibly, but it's more parsimonious to think of it as influence from Central Asian foragers rather than anything from South Asia.

Karl_K said...


"At least there's enough data to say the dogs of southern East Asia are the most basal ones"

There is enough data to say that the modern dogs of southern East Asia have the largest amounts of 'dog' diversity (excluding non-dog modern wolves).

However, we still know very little about extinct wolf variation across Eurasia. The data we do have is suggestive that all modern grey wolves arose from a severe bottleneck. This leaves a lot of area and time for dogs to have been domesticated almost anywhere in Eurasia (but obviously long before ~15,000 years ago, when fully domesticated dogs entered the Americas).

But, that is definitely not the same

Shaikorth said...

I don't think extinct wolves matter all that much. We don't see migration edges from roots of wolf branches or from between dogs and wolves or wolves and coyotes to significant dog populations (dingos and NG singing dogs may have some). Something like modern Middle Eastern wolves admixed into African village dogs and Basenji.

Karl_K said...

"We don't see migration edges from roots of wolf branches"

Of course. And we don't see Neanderthal admixture edges into SSA humans.

By extinct wolves, I mean entire extinct lineages. Not something more related to existing wolves than existing dogs. In fact, they should be more related to existing dogs, and you could only see them as a ghost population.

Shaikorth said...

But those ghost wolf populations aren't showing up as mystery migration edges (from basal roots) in treemix. Various ghostpop stuff does so for humans, so the dog tree seems less complicated.

batman said...

"... the ancestry of present-day dogs is derived from multiple regional wolf populations."

Davidski said...


Looks like Brahmins are from Sintashta.

ryukendo kendow said...

Batman, you are being intentionally misleading. That paper discovered 1.4% admixture from Taymyr wolf into the strongly monophyletic dog clade, that statement by the authors referred to nothing more and nothing less.

Karl, Shaikorth is right, Treemix possesses more than enough ability to discriminate between populations representing a branch tip well vs populations that don't. If a parallel clade having less drift with dogs than dogs have with each other admixed into dogs, it will appear like the admixture edge into Mota. No edge, no evidence of such flow.

And are you sure the 'wisdom' isn't coming from your own head? The data only suggest, and no one is making claims to where dogs originated, that much is pretty clear.

ryukendo kendow said...

Thanks David for the stats.

Here are the most significant stats for Brahmin, most significant at bottom, graphical forms of everything coming later:

Basque_French Basque_French Chamar Brahmin_UP 0.0207 17.493 505534
Norwegian Norwegian Chamar Brahmin_UP 0.0209 17.243 506162
Afanasievo Afanasievo Chamar Brahmin_UP 0.0211 15.107 482906
Poltavka Poltavka Chamar Brahmin_UP 0.0212 15.989 470067
Bell_Beaker_Germany Bell_Beaker_Germany Chamar Brahmin_UP 0.0214 17.323 490494
Lithuanian Lithuanian Chamar Brahmin_UP 0.0214 16.699 505483
Srubnaya Srubnaya Chamar Brahmin_UP 0.0215 17.061 489514
Corded_Ware_Germany Corded_Ware_Germany Chamar Brahmin_UP 0.0216 16.721 489351
Andronovo Andronovo Chamar Brahmin_UP 0.022 16.511 502287
Sintashta Sintashta Chamar Brahmin_UP 0.0227 15.501 429955

The order is very similar with the other pair, with Srubnaya, Sintashta and Andronovo at the top.

batman said...


You're over-reacting. I simply quoted Skoglunds point.

Grey said...

no idea on the dog question except it made me wonder - if you had multiple regional breeds of dog among various categories: hunting dogs, sheep dogs etc then the best one in a particular category may have won i.e. the best sheep herding dog breed at the time displaced all the others.


off topic random thought of the day - i wonder how much more force you need to get a stone spear through a mammoth's hide (fur/hair?) than a deer's? my guess is quite a lot more. if that was so then the guy holding / throwing the spear would probably need to be bigger / stronger than a deer hunter.

capra internetensis said...


Before the introduction of guns, Pygmies were some of the most successful elephant hunters.

Grey said...


interesting - there's even a video of it on youtube

a smaller forest elephant but relatively speaking similar size i guess - guys in the video had steel spear heads but maybe they did it back when they were using stone?

the main thing i was wondering about with regard to mammoth was getting through the hair before you got to the hide itself

anyway, if anyone ever did a test with a stone spear and a simulated mammoth including the hair it would be interesting - especially if thrown


also, says in the video an elephant that size would feed 80 people for two weeks - which i think speaks to how people could survive extreme LGM conditions as long as they had megafauna to hunt

ryukendo kendow said...

@ Davidski

About the parallel clade in Iran_N, how did you demonstrate this? What stats in particular make this more likely?

Also, I went to relook at some plots of EHG vs Ust_Ishim affinity and WHG vs Ami affinity; the anlmalous position of Iran_Neolithic and Satsurblia (shifted off-cline from West Eurasians in U-I and Ami affinity) is very striking. Do you see any evidence of the Central Asian forager influencing Iran_N as well?

Btw Rob thanks for the references.