Interdisciplinary analyses of Bronze Age communities from Western Hungary reveal complex population histories, Gerber et al.


In this study we report 20 ancient shotgun genomes from present-day Western Hungary (3530 - 1620 cal BCE), mainly from previously understudied Baden, Somogyvar-Vinkovci, Kisapostag, and Encrusted Pottery archaeological cultures. Besides analysing archaeological, anthropological and genetic data, 14C and strontium isotope measurements complemented reconstructing the dynamics of the communities discovered at the site Balatonkeresztur. Our results indicate the appearance of an outstandingly high Mesolithic hunter-gatherer ancestry in the largest proportion (up to ~46%) among Kisapostag associated individuals, despite this component being thought to be highly diluted by the Early Bronze Age. We show that hunter-gatherer ancestry was likely derived from a previously unrecognised source in Eastern Europe that contributed mostly to prehistoric populations in Central Europe and the Baltic region. We revealed a patrilocal residence system and local female exogamy for this Kisapostag population that was also the genetic basis of the succeeding community of the Encrusted Pottery culture, represented by a mass grave that likely resulted from an epidemic. We also created a bioinformatic pipeline dedicated for archaeogenetic data processing. By developing and applying analytical methods for analysing genetic variants we found carriers of aneuploidy and inheritable genetic diseases. Furthermore, based on genetic and anthropological data, we present here the first female facial reconstruction from the Bronze Age Carpathian Basin. 



I'll later update this post. 


Bk-I: Somogyvár - Vinkovci culture S9 2560-2290 35-40 M K1a3a R1a-V2670 None


ambron said…
"We can take into consideration the outgroup f3-statistics results, chronology, timing of HG admixture according to Freilich 2021 between ~3400-2400 BCE, qpAdm results and the geographical distribution of groups and outliers of similar HG makeup. These suggest a dated migration pattern for this undescribed population with dominant HG genetic ancestry from what is present-day Bulgaria to the Baltic through the Eastern borders of the Carpathians (Fig. 3 b). Ancestors of Bk-II likely branched off from this migration route and started to move towards West, by at least around ~2500 BCE, subsequently intermixing with various groups. Interestingly, the phylogeography of mitochondrial haplogroup U4b1b1 (individual S15) perfectly fits this scenario."

The mystery of the Balto-Slavic drift is solved. You talked about it a long time ago.
Arza said…
Not fully solved as there are some weird things in this preprint, but nothing that a mail to the authors can't fix.
Arza said…

Can you do me a favour and correct this piece of disinformation ("Also it seems this result is based on Str results or? so I remain skeptical") on AG?

From the paper ("Bioinformatic analyses" section):

"For Y chromosome haplogroup determination the Yleaf v1 software was applied."
Arza said…
Thanks a lot!
Anthro Survey said…
Well, this is a most excellent development, and it’s super nice to see academic paper corroborate something I noted months ago in G25, almost by accident. Now, prior to that, I never really believed that Baltic_BA is a consequence of pure BA drift simply because of the abysmal distances(~5-6 or so) in attempting to fit it with existing proxies. Or rather, I suspected there to have been drift, but attributed this more to a yet-unsampled HG population. After all, HGs had small population sizes, which is a condition for accelerating drift rate. So, I knew it had to be some divergent HG source, but just automatically assumed it’d be positioned closer to the Baltic sea than the Carpathian chain.

I discovered it when modeling the Mokrin set. After picking out the HG-rich outliers, I made an average out of them. Again, at that time, I just assumed this HG ancestry to be the “usual” Iron Gates WHG or smth. After failing to get a decent fit, I attributed this to presumed poor coverage. But some time later, I tried Baltic_Ba for the heck of it and....voila. [Insert apt meme here] I had to look at it twice. lol

This then prompted me to run other existing Carpathian BA samples and, sure enough, some of them also exhibited this and gave me substantially better fits than I’d achieved previously. When the Slavonia Jag samples came out—-same exact.

Long story short, I’ve had to rethink my prior paradigms, and, just like you guys, I also see Baltic_BA ancestry moving in a south-north direction. For existing Baltic_Ba samples—-Probably some epi-corded derived Mierzanowice-related groups residing on the northern foothills are the direct culprit, if I had to guess? As for the southern foothill(of the northern chain) populations like Encrusted Ware, I think an independent formation occurred. I.e. I don’t think Baltic_BA-like pops crossed the Carpathians from the north to create the Encrusted. They were a local merger of EEF, uncanny HG, and Hyperborean-rich incomers.

When you think about it, it’s unsurprising that an uncanny HG population existed somewhere in the Carpathians or their foothills. That kind of thing provides obvious refuge to older population strata, for one.

And MASSIVE kudos to you guys—-Arza and Ambron——for also noting this. I discovered your blog a couple of weeks ago in an attempt to find other individuals who also made this observation. Always good to level with the like-minded. We are almost there, but we will be vindicated in full one of these years when they dig up our sought-after HGs in the Beskids(?). Heck, more Vatya_outlier samples would be excellent, too.

Dinner is all on Dave lol
Anthro Survey said…
Guys, check out this graphic. I didn’t make it and it’s pertaining to the Komarov culture, which I’m not exactly trying to implicate in this, but the dispersal pattern happens to almost exactly match my mental model of Baltic_BA-like ancestry expanding into/across NE Europe. I would just shift the initial nucleus closer to the 12 o’clock position.

Scroll down a tiny bit and see fig 1:ă/22192f4f8a0d2d1833a0de2d9db39f8c35a1e729
Arza said…

I never understood why anyone would try to explain it as a purely Bronze Age phenomenon. Especially not after the study about the origin of Icelanders was published. If anyone wants to see how a 1000 years of a genetic drift on an isolated island looks like just compare Icelanders to Norwegians. That's nowhere close to the so-called Balto-Slavic drift.

Although there are two layers of this drift, but that's another story - Palaeolithic (practically everyone in Europe has it) and Mesolithic (characteristic to certain BA populations and modern day Balto-Slavs and pops mixed with Balto-Slavs).

IMHO the origin of Baltic_BA and generally the history of Indo-Slavs as EastPole calls them is pretty straightforward - early Indo-Europeans move towards the Carpathians and acquire some additional EEF ancestry. Future Indo-Iranians immediately bounce off the Carpathians (main Nitra cluster is identical to Fatyanovo). Future Balto-Slavs cross the mountains... and now we have two possibilities:
a) Balto-Slavs acquire HG ancestry and to-be-Balts move north, while to-be-Slavs stay near the Carpathians and they start to acquire low-WHG EEF ancestry,
b) Balto-Slavs split, to-be-Balts mix with almost pure HG, while to-be-Slavs mix with a population on an EEF-HG cline.
Later the Balto-Slavic cline (as seen on the North Euro PCA) develops - Proto-Slavs shift towards Baltic_BA while Proto-Balts slowly acquire "southern" admixture that differentiates them from Baltic_BA.

Previously I thought that such HG refuge was north of Ukraine. But now we have plenty samples from the Baltics - no HGs can be the source of the drift, we have plenty samples from Volosovo... not only they lack the drift, but they're EHG-like. Finally we have Fatyanovo that rolled over the whole area and again - no sign of the population we're looking for. On the other hand the Balto-Slavic drift explodes in the BA in the Carpathians and surroundings in a multitude of loosely connected cultures which indicates that the source of this ancestry had to be local.

Kudos also to Matt. Years ago he first noted that the Balto-Slavs deviate from other Europeans in the sixth dimension of the West Eurasian PCA made by Davidski.

The best moment for me was when it became clear that I've indeed discovered the Baltic_BA population. Months before Mittnik et al. published the preprint I already had calculated coordinates of this population. Later, when Davidski added them to the WE PCA, my ghost has landed inside the Baltic_BA cluster. From that point I knew what happened in the BA in this part of Europe, because the solution was only one...

Yep... there's another Vatya outlier. He's close to the Baltic_BA, and he's very similar to VLI051. I suspect that this sample comes from one of the inhumations described here:

That Fig. 1 may be very important. But there was also a lot of interaction across the mountains (e.g. Fuzesabony culture had settlements in South Poland).
Daniel Gerber said…
Hi fellas, i am really surprised that my work was followed since the ISBA conference, and I am really surprised (and also really excited) that others found the exact same outliers to be the carrier of a peculiar HG source. I highly appreciate your work here, and it feels really blessing that you basically found what i have at the end of the day. If any of you got any comments on the paper or have any questions please don't hesitate to ask!
Anthro Survey said…

My origin theory is similar. I also see an initial "Indo-Slavic" nucleus forming in the south of Poland with a composition similar to that of Fatyanovo before ancestors of the latter "bounced off" as you say. However, I don't necessarily see the ancestors of Balto-Slavs *crossing* the chain to acquire the uncanny HG. Instead, I favor the idea of these HGs initially co-existing in tandem with the early "Indo-Slavs", albeit occupying a different phytogeographic niche(Carpathian foothills vs Polish lowlands). Later on, I envision a fusion taking place in southern Poland/W. Ukraine, leading to the creating of Baltic_BA-like, which then expands and blankets much of NE Europe.

As for Slavs---I think their deviation from Baltic_BA is mostly explained by much later---LBA, IA, or even LIA---flux from Pannonia or, ultimately, Balkans proper into what's today Polesia. Probably related to the major introduction of J2b2 and most especially E-V13 into the early Slavic Hg pool. In fact, could be related to the late E-V13 bloom in the Carpatho-Balkan region in general. As a side note, a poster on AG named George once wrote smth about "Thrakoid" toponyms all the way on the Dnieper, but perhaps it's more related more to MBA pre-proto-Thracian populations rather than any reflux from south of the Danube. For what it's worth, Polesians and southern Russians tend to visibly, albeit somewhat equivocally, favor Bgr_IA and/or Nort_Gheg-sans-Slav input over available, chronologically older Pannonian strata like Hrv_MBA or Mokrin. Modern Polish get away with the latter, but then again, West Slavs assimilated "Westish" populations which could be offsetting things.

Btw, we actually have what are likely to be medieval Balts in our sample sets. In the Viking period set, there are ~5 individuals who are almost entirely Baltic_BA by ancestry. I can point these out later on tomorrow when I get to my files and/or share aggregate coords. These certainly are either Balts or some sort of Finnicized Balts. Modern Balts deviate substantially from them, and the most likely reason imho is massive influxes of Slavs into Baltic cities during PL Commonwealth rule from Poland proper or Rus lands. This is historically attested.

Wow, good catch on VLI051! I missed that one. There's also an IA sample like this from SVK or CZE(I forget) from the recent La Tene paper and he's R1a as well. But, in the supplement, it doesn't really say anything about VLI051 being of Vatya provenance. I tend to think of him as just a northern sub-Carpathian fellow somewhere due east of Bohemia.

I also have a question----is the Volosovo and/or actual Nitra data public? I think I vaguely recall a paper on the former(tho Dave didn't make coords), but would be interested on seeing the Fatyanovo-like Nitra!

And, given the overall discussion, what do you think of the general idea in this piece? I'm not sure about the *exact* wording of their conclusion per se, but given what we now (virtually) know about this so-called Balto-Slavic drift, I think it's not unreasonable that pre-proto-Moeso(?)-Albanoid and Balto-Slav ancestors experienced some contact in the circum-Carpathian zone. This isn't something that die-hard Illyricists wanna entertain, ofc.
Anthro Survey said…
Speaking of the Balkans----I think some people are gonna potentially start questioning the extent of Slavic impact there as the idea of Carpathian-rooted HGs gains mainstream traction.

To forestall this, it would be good to point out to folks that Encrusted Ware seems to have been a dead end, a flash in the pan south of the Danube. HRV_MBA and IA from Pannonia virtually lack this signal(the I2a-Din individual who harbors it seems to be mis-dated b/c of age discrepancy of mtDNA on Y-full and he wasn't properly radio-carbon dated either). Even if we create artificial coordinates consisting of diluted Baltic_BA and/or diluted Encrusted and "force" the former to use them, fits suffer visibly. Plus, Bgr_IA lacks this, too, as do Viminacium individuals most likely but we'll soon see for sure when it's published.
Anthro Survey said…
@Daniel Gerber

Will do! Always great to see academics like yourself willing to work with us hobbyists. While you guys clearly have a definitive edge over us in your technical knowledge, sometimes we tend to have a more complete, multidisciplinary picture of things. Really, I wish you the best of luck in your work and hope you publish follow-up data and studies on this!

Global25 is a powerful tool if appropriately used and covers a lot of wavelengths. It's not useful for capturing extremely fine-scale drift(we have IBD-based tools for that, right?) and, on the other extreme, it's not always useful for inferring and/or deducing % of missing, deeply-diverged streams of ancestry(e.g. non-Eurasian ancestry in Iberomaurisians). But other than that, you can learn a LOT with it. Unfortunately, you see a lot of amateurs nowadays without a proper grasp of it nor archaeology/history and bastardize it for the most random calculator schemes(yes, $$$ schemes too) imaginable.

I do have a request for you: if sharing seq data from your paper is not an option at this time due to understandable reasons, can you by any chance project some previously-published Carpatho-Pannonian BA samples onto your PCA? In particular, the Encrusted Ware (Jag) set from this paper(but not limited to it):

Also, has your team or colleagues definitively planned to examine samples from other sites in the region at this point? Should we be looking forward to more? Keep us posted.
Arza said…
@Daniel Gerber

Hello! This was less expected than the Spanish Inquisition!

I'll have a lot of questions and suggestions.

There's really a lot of things to talk about. One is that you can easily widen the scope of the publication (or write a follow-up) to "The origin of European Hunter-Gatherers" or similar.

Another thing is the difference between our "amateurish" tools and qpAdm/f-stats. This whole thing was flying under radar for so long because what is obvious in "our world", is barely visible in the "pro" tools.

Later today I'll go trough the preprint one more time and post some questions.

BTW which email is best to use to contact you?
Daniel Gerber said…
No one expects the spanish inquisition :) so I try to answer the questions before I start the day, I don't want to cliffhang you here, but for detailed explanations I highly suggest that you read not only the main text, but the supplementary as well, it is exhaustive I know but it has a number of explanation for most of your questions. So in a row: Jag almost perfectly overlaps my encrusted pottery group (bk-iii), but while it is clear that bk-ii is the main paternal sourve for that group, the qpadm made 10k possible models, which likely points to a really complex regional population process that yet to be uncovered. We plan to extend the dataset but the rest is yet confidental. The thing is that while I found - truly unexpectedly - what I found, it is only a single site from the BA, so seeing this as a complete HG explanation paper is highly exaggerated, especially in the light of lack of the samples from other regions, but as I said, the show must go on, we will se what we catch in the future. To be honest I have never heard of G25 before, and I dont know if these can be used for "serious" popgen, but according to the suggestions I saw here I believe that the core mechanic behind official softwares and this is basically the same. Anyway I have to go now, but Inwill check back here during the day, and I am waiting for other questions too, if you have some :) cheers
ambron said…
Daniel, we welcome you warmly!

I would like to take this opportunity to ask you, when can we expect the publication of samples from Nitra? We saw in the slides that many of them have the same specific type of HG and that many of them are Z280.
ambron said…
Anthro Survey, I read your post on AG but didn't respond to it as you suggested. We can chat freely on Arza's blog.
EastPole said…
“Future Indo-Iranians immediately bounce off the Carpathians (main Nitra cluster is identical to Fatyanovo). Future Balto-Slavs cross the mountains...”

It looks like Indo-Slavs were one population derived from R1a CWC groups in Southern Poland.

What puzzles me is why Balto-Slavic drift in Europe affected R1a CWC groups and didn’t affect R1b CWC/BB groups?. The general idea of that puzzle is presented here :

One possible answer is that there were no linguistic, cultural, or religious barriers between R1a CWC groups i.e. all those red dots were speaking languages which were at least partially mutually intelligible.

Another interesting problem is cremation. It looks like Somogyvár-Vinkovci culture used cremation. Later cremation became popular also north of the Carpathian mountains and finally reached Sintashta/Andronovo.
Cultural links between Southern Poland, Hungary and Sintashta are known:

Daniel Gerber said…
to ambron: I don't know much about the Nitra project, so sadly I can not inform you about the process of this study. However, to be honest, I believe that the spec HG we discuss here is mostly admix with other pops by female bias, I can not really proove this on the spot, but the thing is that the lot of I2a-L1229 in the Balatonkeresztúr groups likely connected to Megalithic or derived cultures, such as Globular Amphora, same with the Jagodnjak G2a. But to be fair, the existence of this specific HG is yet to be accepted by reviewers, and also yet I am officially the only one who suggested the existence of such HG (not forgetting your ideas, I only talk about the official situation here), thus other suggestions from different studies are (way) less robust, because the population we recovered here has the highest percentage of this ancestry from ANY known populations, thus the actual sequence data for further description is needed. But don't worry, it seems that within a relatively short time the data will be available :)
ambron said…
Daniel, did I understand you correctly? Is there more of this specific HG in the Bk-II population than in the Baltic BA population?
Anthro Survey said…
@Daniel, Arza, and everyone else reading this:

I should point out that Jag_MBA's HG shift is strongly, but not ENTIRELY attributable to the uncanny drift we're all hunting for.

Long story short, G25 is pretty good at picking up relatively diverged streams of ancestry to the extent that if your model is lacking the basics, it will let you know.

Using a setup sufficiently accounting for pretty much all possible relevant ancestral streams(ideally, models should only include the sufficient without duplicate and/or highly similar streams), we can see what happens when we ablate the typical WHG input:


Target: HRV_Jag_MBA
Distance: 1.2166% / 0.01216646 (this is an excellent Euclidean distance)
44.8 HUN_Baden_LCA
31.6 Baltic_LVA_BA
12.8 CWC_Pol_PCW
10.8 ROU_Iron_Gates_HG


Target: HRV_Jag_MBA
Distance: 2.4842% / 0.02484217 (the absolute value is relatively bad and, most importantly, shows a large CHANGE in fit from the above)
47.4 Baltic_LVA_BA
44.2 HUN_Baden_LCA
8.4 CWC_Pol_PCW

When we are dealing with averages of relatively tight convex hulls and good overall genomic coverage per individual, such a change in fit always signifies something.

The result almost hints at there being a HG cline across the Carpathian chain, whereby some groups ended up with somewhat variable proportions of standard WHG vs uncanny HG. Although I did not read the preprint to the T(yet!), I remember it mentioning something along the lines of multiple HG sources being taken up with qpAdm attempts. So, I wonder if the above models would also hold true for Bk-II, and, if so, may explain the multiplicity in part.

To those wondering whether the Iron_Gates_Hg is merely just covering for uncanny HG ancestry and if this is a glorified overfit: Hard Nay. First of all, if that was the case, we wouldn't be left with any spare CWC in the model, as it would all be "sponged up" by Baltic_BA, which is roughly 30% HG-70% CWC. So, if we do the math above 10% would translate into ~30% Baltic_BA. If a population consists of at least 30% Baltic BA-like ancestry, attempted substitutions always result glaring and statistically significant fit worsening. & G25 ultimately operates on the best fit principle.

Plus, this feature is also true for other Carpathian samples---incl some from Mokrin.
Daniel Gerber said…
Yes, the Baltic BA has an average 30 percent or something like that as I recall and the Bk-II an average 42 percent
Arza said…
I've turned the comments moderation off.
Arza said…
Isn't the .xls file with supplementary tables missing from bioRxiv?
Arza said…

Other pops with other haplogroups were affected too (almost all the BA pops from the inner Carpathians). But it seems that the "Celts" and others rolled over them in the IA and this ancestry survived only in the R1a-rich Balto-Slavic related populations in the outer Carpathians and further north.

Unless there was a continuation of it in Slovakia or Hungary, but only Daniel can answer this question.
Arza said…
@Daniel Gerber

Technically you're not the only one. There's another group... but they seem to be not fully aware what they've found.

Demogenomic modeling of the timing and the processes of early European farmers differentiation, Marchi et al.

IMHO "West Meta" is the population we're looking for. They believe that Iron Gates branched off it, but I see this differently. Judging by what G25 shows Iron Gates is rather a 60:40 mix between the Villabruna cluster and West Meta. This is in line with their simulation (Fig 3., panel B):

In fact not only the Iron Gates seems to shifted in this direction, but the whole so-called ANE-WHG cline is bent (or rather consists of a series of clines):

What is even more important to this topic - such hypothetical population is in the almost exact place required to "produce" Baltic_BA cluster.

You can play with that 3D plot here:

My theory is that this unusual HG ancestry comes from another refugium (as in Marchi et al.) and that in the East it merged with ANE forming EHG, and in the Balkans it merged with expanding WHG-proper forming Iron Gates / Narva like population. Then mostly EHG pops with mostly IG-like pops started to form the HG-cline as seen on the West Eurasian PCA.

And such ancestry had to survive somewhere near the Carpathians, because it explodes there in the Bronze Age. Although there's one catch. If you switch PC3 to PC6 in the 3D plot, Baltic_BA lands in a much more extreme position. My explanation is that the "larger" PC3 shows the drift up to the Mesolithic, and the more recent drift that happened when our population X has been already isolated is dumped on the sixth dimension.
EastPole said…

„Other pops with other haplogroups were affected too (almost all the BA pops from the inner Carpathians). But it seems that the "Celts" and others rolled over them in the IA and this ancestry survived only in the R1a-rich Balto-Slavic related populations in the outer Carpathians and further north.”

Yes, but only pops with R1a and drift managed to get out of Hungary and affect all other R1a rich Indo-Slavic pops, except those which earlier migrated east.
R1b pops with drift didn’t get out of Hungary and didn’t affect other R1b rich pops. Like BBs with drift in Hungary didn’t affect other BBs in the west.

This may suggest that R1a pops acquired the drift but didn’t change language and culture and could interact with other R1a pops without the drift passing that drift to them.
Arza said…
@ EastPole

Ah. Now I understand.

But I'm not sure if such spread of this kind of ancestry would be possible without at least partial population replacements (in the Baltics Baltic_BA almost certainly replaced everyone else who lived there before).

Unless the population density in Poland, Belarus etc. was much lower than in Hungary. Do we have any estimates for this period?
Arza said…
@ Anthro

We have (at least) one Volosovo sample and a rumour that the rest looks the same. In case of Nitra and another Vatya outlier we have a quasi-PCA (f3 vs. f3) from EAA presentation made by Anna Szecsenyi-Nagy. It matches West Eurasian PCA very well and has plenty of other samples so it's easy to figure out everything.
ambron said…
Daniel, I'm glad your team will continue to explain the mystery of this specific HG. This is because this specific HG is a characteristic marker of today's Balto-Slavic speakers. It also has a fundamental significance for the genetics of Hungarians, who do not differ genetically from their Slavic neighbors, despite the wonderful history of their ethnolinguistic ancestors - the Magyars.
EastPole said…

“But I'm not sure if such spread of this kind of ancestry would be possible without at least partial population replacements (in the Baltics Baltic_BA almost certainly replaced everyone else who lived there before).

Unless the population density in Poland, Belarus etc. was much lower than in Hungary. Do we have any estimates for this period?”

CWC in Northern Poland, Belarus, Baltics were mostly herders switching to hunter-gatherer lifestyle often, so population density was low.
In Southern Poland and Western Ukraine it was different but there CWC R1a had to compete for good arable land with mixed Globular Amphora, Funnelbeaker, Baden, Tripolye, BB, CWC R1b groups etc.

I see drift expansion as mixed R1a CWC farmers with drift coming from the south and replacing R1a CWC herders and hunter-gatherers without drift (i.e. they spoke similar languages and could mix). And by this they became strong enough to compete with mixed Globular Amphora, Funnelbeaker, Baden, BB and CWC R1b groups. As a result Balto-Slavic became the dominant population in Eastern Europe.
Anthro Survey said…
To add to my last post above—-

Here are other examples of Carpathian basin individuals who show a requirement for a *standard* WHG-rich source(can use Iron Gates).

I highlighted the ones who co-require uncanny HG input, behaving much like Jag.


(Now, despite equivocal need of Baltic_BA for the unhighlighted two, I still think they carry it, albeit to a lesser degree—-not unusual in clinal contexts.)

Interestingly, judging from Scythian period samples, this kind of differentiation persisted for over a millennium.
bce said…
@Anthro Survey
This variation is simply the result of Baltic_BA not being the actual source (too much steppe), and these samples being mixed with local steppe and eef to various degrees.

This ancestry in Pannonia is from a single source, the closest proxy among the available samples is the Mezocsat outlier (and the Vatya outlier, but he's low quality). You should be able to model all other Panonians with the Mezocsat outlier and various local influences.
Anthro Survey said…
Mostly beg to differ...

I mean, you do raise a good point and one I've pondered: we are lucky in the sense that virtually none of our samples really has an excess of HG of discussion in relation to CWC-related ancestry---except for Vatya_o. Were this not the case, then our fits would be quite off, indeed.

But again, this isn't the case in our case. We're looking at a non-artifactual combination of uncanny HG AND "standard" WHG ancestry in a good number of these individuals. Without getting into the specifics of G25's fidelity in discriminating supra-threshold levels of divergent streams, look at my model of Jag average above. As you can see, there is still CWC left over. If Iron Gates was really compensating for what you're contending, we wouldn't see that. And EVEN IF such a compensatory scenario was at hand, the fit would be NOWEHRE NEAR being that crisp at all. Missing 10%HG is like substituting the 30% Baltic_BA in a population like Macedonians for Hyperborean proxies lacking this drift.

We have no real basis on which to assume any single source, especially since while we have smth like Vatya_O, we also have smth like Koros_HG. While an outlier in its local EEF bio-cultural context, it's enriched for standard WHG and is therefore further suggestive of the cline in HG ancestry I make the case for.
bce said…
@Anthro Survey
That leftover CWC is also just a part of the compensation for JAG_MBA.

You have to take into account different components which G25 can tell apart, and their amounts within Baltic_EST_BA and JAG_MBA. In this case it's just Steppe, WHG, EEF and "Balto-Slavic" drift , for both.

Let's look just at the steppe. Baltic_EST_BA is ~60% steppe, and JAG_MBA is ~30%.
There is ~30% Baltic_EST_BA in your model, so it brings less 20% steppe into the model.
So, additional steppe has to be added to compensate.

Koros_HG is from a whole different period. he is from the early neolithic and nothing like him appears afterwards, in the middle neolithic, late neolithic, etc. There are only regular neolithic farmers for a long time. Then these WHG-rich Pannonians appear as newcomers in the EBA.
Anthro Survey said…

I think I'd misunderstood what you were saying and possibly vice versa. I'd initially thought you meant that *WHG* seen in my model is either compensating for cryptic HG ancestry(due to Baltic_BA's ratio of CWC:HG) or an artifact overfit for it. But it looks like we're on the same page that WHG and HG inherent to Baltic_BA are not interchangeable in terms of G25's discrimination(at least, not above a fairly low threshold).

But, no, I'm not saying that a *Baltic_BA*-like was the ACTUAL source in the spread of such drift in Pannonia. I simply use it as a vessel for the yet-to-be characterized HG ancestry. Thanks to sufficient "steppe" ancestry in all samples, things work out.

I do agree the ACTUAL immediate source of Jag_MBA's HG ancestry was probably smth akin to Vatya_outlier, not Baltic_BA-like. But my point is that similar HG-rich populations across the Pannonian basin could have easily varied in the exact composition of their HG. Some were more enriched for WHG proper, others more for "balto-slavic" HG. (As a result, the resultant BA populations also varied) This is a plausible scenario when you consider a hilly/mountainous habitat for such HG-rich populations, differing from the EEF crowd on the Alfold plain and such.

About Koros_HG:

-Sure, but it's likely that people like him contributed to later Chl and EBA populations of the Carpathians. Where else is the massive proper WHG in Jag and Mako_1502 from? And given him being an outlier, again, I'd venture to say that he was from the hill zone. Different phytogeographic zone from the LBK and their successors. This kind of thing can and often does correspond with different ancestral profiles.

-Now, as far as nothing like him appearing afterwards--again, not to belabor things, but we have exactly 0 samples from the mountains and hills. Let's sample Transylvania, Beskids, etc. and then see. It's a question of archaeogenetic visibility.

-But, alas, we actually DO have something similar afterwards from the Circum-Carpathian zone---from Urziceni in Romania. circa 3500 BC, Wallachian plain. And, yet again, such an individual appears to be an outlier amongst the standard EEF crowd. It seems to be of mediocre coverage, but appears mostly WHG, albeit with some unequivocal "balto-slavic" HG as well.
Anthro Survey said…

Scratch that a bit----the date is correct, but not Urziceni---it's Gura Baciului, which is Transylvania. A piedmontese setting which needs more sampling. :-) So, n=1.

I don't wanna make too many assumptions, but *IF* that sample is indicative of mainly WHG that far West in the chain, we're prolly better off sampling the Beskids for the elusive "Baltic" HG.
bce said…
@Anthro Survey
the "proper WHG" in BA Pannonians is also mostly inherited from the "mystery Hunter Gatherers". Baltic_BA carries some "proper WHG" as well.

when we get samples of the "mystery HGs", they will also be modelled with some proper WHG + minor EHG and EEF + a high distance. most of this distance being their specific drift.

I think when we get these samples, they will cover most of the HG ancestry in both Pannonians and Baltic_BA.

Gura Baciului is different. It really does seem to have ancestry from both "Mystery HGs" and Iron_Gates. But it looks like a dead end, because it has too little "Mystery HG" to be ancestral to either the Pannonians or Baltic_BA.
ambron said…
I think that the spread of this specific HG to the north is also confirmed by the close genetic relationship between the Hungarian BA population and contemporary Poles demonstrated by Cassidy.
EastPole said…
Using HUN_EIA_Prescythian_Mezocsat_o1 I made a ghost HG sample HGm:


Using HUN_MBA_Vatya_o I made another HG ghost sample HGv:


HGm and HGv are good for modeling Corded_Ware_Baltic:Spiginas2

Target: Corded_Ware_Baltic:Spiginas2
Distance: 2.4249% / 0.02424892

28.8 HGm

17.6 UKR_Meso:I1763

14.8 RUS_Fatyanovo_Ivanovo_BA:MIL001

12.6 RUS_Fatyanovo_Yaroslavl_BA:NIK007

9.0 HGv

8.8 RUS_Fatyanovo_Yaroslavl_BA:NIK004

5.4 Corded_Ware_CZE_early:VLI076

3.0 Corded_Ware_DEU:I1538

The Baltic_BA is not so good:
ambron said…
EastPole, Baltic BA gives bad fit, because apart from this especial HG, it also contains the usual WHG from another source.
EastPole said…

“EastPole, Baltic BA gives bad fit, because apart from this especial HG, it also contains the usual WHG from another source.”

But I have the usual WHG in my source data. It is not selected.
EastPole said…
I got better fits introducintg Yamnaya to the source data:

Target: Baltic_aLTU_BA:Turlojiske3
Distance: 3.4414% / 0.03441414
29.2 Corded_Ware_CZE_early:VLI007.merged
24.6 HGm
18.4 RUS_Fatyanovo_Ivanovo_BA:MIL001
10.2 CZE_EBA_Unetice:I14192
7.4 Yamnaya_RUS_Kalmykia:RISE546
5.2 HGv
5.0 Corded_Ware_DEU:I1538

Target: Baltic_LTU_BA:Turlojiske1
Distance: 3.5411% / 0.03541069
23.2 HGm
17.8 Yamnaya_RUS_Kalmykia:RISE546
15.4 RUS_Fatyanovo_Ivanovo_BA:MIL001
12.8 HGv
10.6 Corded_Ware_CZE_late:VLI010.A0101
7.4 Corded_Ware_DEU:I1539
6.4 Corded_Ware_CZE_early:VLI007.merged
2.8 Corded_Ware_DEU:I1534
2.0 Yamnaya_RUS_Caucasus:ZO2002
1.6 Corded_Ware_DEU:I1538

Target: Corded_Ware_Baltic:Spiginas2
Distance: 2.4052% / 0.02405177
31.2 HGm
17.2 UKR_Meso:I1763
15.2 RUS_Fatyanovo_Ivanovo_BA:MIL001
10.0 RUS_Fatyanovo_Yaroslavl_BA:NIK004
9.6 RUS_Fatyanovo_Yaroslavl_BA:NIK007
8.8 Yamnaya_RUS_Samara:I0443
7.8 HGv
0.2 Corded_Ware_DEU:I1538
ambron said…
The Baltic BA is on PCA more shifted towards HG than Bk-II, even though it contains less especial HG responsible for the Balto-Slavic drift. The addition of some other HG from another source is probably responsible for it. Maybe Daniel or Arza will explain it to us.
Anthro Survey said…

I do agree that the Mystery_HG----defined as the non-CWC component of Baltic_BA----does contain a proportion of proper WHG in addition to highly drifted/divergent para-WHG stuff. But again, what you're seeing in the Jag model above(as well as a handful of other individuals) is separate. If this was truly an artifact and just an overfit for the Mystery_HG, you would not see free CWC hanging around like that, as it would be "eaten" by the Baltic_BA. I can see it being artifactual if we only got ~1-2% extra WHG and there was no significant change in the fit distance, but not the case here.

So, this leads me to think that G.Baciului-rich groups did also contribute in the long run to our Pannonian BAs, not just Mystery_HGs.

Honestly, it's likely that mixed/clinal groups consisting of Mystery_HG and GB-like ancestry in roughly equal measure were the most common south of the Carpathians. Hence, we see such frequent(but not invariable!) co-scoring of these ancestries. Perhaps on the Polish-facing slopes, they were mainly just Mystery_HG. But again, we won't really be able to infer with precision unless we actually get samples from piedmont and highland settings. It's very hard for me to see communities of such populations on the Hungarian plain during Copper and Eneolithic. Lowlanders=Baden-like agriculturalists imho.
ambron said…
Arza, do you know something more about this statement by David?

"Let's keep this to ourselves, but word is that the mystery hunter-gatherer ancestry in the Kisapostag/Encrusted Pottery pops might be from a very late forager enclave in eastern Romania near the Black Sea coast"

The lolalization of the later source of this especial HG would be similar to the earlier one indicated by Daniel Gerber.
Arza said…
Sorry for a late reply, but I'm practically off-line till tomorrow.

This Romanian group will be the ultimate proof that HG populations survived till the recent times in an unadmixed form but it wasn't the source of the "Balto-Slavic drift".
Anthro Survey said…
Eastern Romania, huh? Hmm. I can see Transylvania, but "near the Black sea coast" just doesn't sound plausible as a source for this "Balto-Slavic" ancestry.
ambron said…
Arza, and has the source population for the Balto-Slavic drift already been found? I mean, of course, a younger source (closer to the CWC period) than the one indicated by Daniel Gerber in Bulgaria. If such a population has been found, from what place and time does it come from?
Arza said…
They were found almost at the mouth of the Danube.

I don't know. There are tons of unpublished data (Reich Lab recently published samples with IDs over 30000) so maybe they have something like that already. What I know is that these Romanian hunter-gatherers can't be the source of the WHG-like ancestry in Baltic_BA.
ambron said…
And we are also waiting for the next results of the work of our Hungarian friends.
ambron said…
Arza, let me remind you of my earlier question:

The Baltic BA is on PCA more shifted towards HG than Bk-II, even though it contains less especial HG responsible for the Balto-Slavic drift. The addition of some other HG from another source is probably responsible for it. Maybe Daniel or Arza will explain it to us.
Daniel Gerber said…
I highly suggest for you to read my supplementary material as well, and I highly recommend to check f3 results in detail + check the heatplot I made, many of your questions will be answered by that analysis. The thing is that I have seen a "void" between Dereivka I (Ukraine), GB (Romania HG) and Malak Preslavets (Bulgaria), it seems that between these regions are the source for a yet unsampled HG type, which time to time left traces in neighbouring groups, in Bulgaria, in Ukraine, NOT in Romania thus ultimately showing the borders of this pop, while suddenly became more visible around ~2400 BCE and ultimately appeared to be group level in the Baltic AND in Western Hungary. As much as I can see, more data and better coverages are needed to discuss this topic further, but we have at least proof for something previously undescribed.
ambron said…
Daniel, thank you for your answer. Of course, I will analyze the supplementary materials in depth.
Anthro Survey said…
@Daniel Gerber

I want to point out that it appears in Eastern Hungary, too. Northeast, to be exact. Notable examples include Mako 1502 and 2 Fuzesabony samples. It also seems to persist into the Iron Age, although this fact in itself may belie more complex migratory/replacement dynamics. Eastern Hungary is notably undersampled in relation to the west, and, with further sampling, I think we would also see such a group level signal there in certain contexts and communities. The truth is, the Pannoanian basin was not a monolith in the BA, but more like a checkerboard with different groups seemingly situated in close geographic proximity.

Consider also that Encrusted Ware’s genesis is supposed to be in the north of the country, before its spread south and subsequent disappearance of the “Balto-Slavic” signal there(MBA Croatia lacks this signal)..
bce said…
@Anthro Survey


this is Baltic_BA with the steppe and eef removed. With these coordinates Pannonians can be modeled with perfect distances, and they all pick up some Iron Gates HG on top of this HG.

So it does look like the Pannonians went trough a step of absorbing Iron_Gates ancestry before coming to Pannonia, it makes sense because we know that there were still Iron Gates-rich populations in the region, like the ROU_C_o.

However, I still believe BA Pannonians all originate from a small group which had a massive founder effect. And the differences in the type of WHG are due to the intra-group variation, and their mixing with various eef and stepep groups which also carrried different types of WHG.
bce said…
Most of the Hungarian BA/IA samples are in fact from the eastern half of Hungary. With them we can follow how the HG ancestry appeared with Mako, and then had constant declines and increases, the last being with the late IA vekerzug Scythians.

from Western Hungary there were practically no BA/IA samples until the recent Celtic study, and these Encrusted pottery which are about to be published. And it looks like here the WHG-rich Pannonians fared much worse troughout the BA/IA and were almost completely replaced by Celtic-related people at one point.
EastPole said…

“this is Baltic_BA with the steppe and eef removed”

How did you do it?
Daniel Gerber said…
That one Mako individual that you refer a lot is actually a misinterpreted Hatvan culture individual, and indeed, it shows high resemblance to the Kisapostag group, but still, its a single individual, and firm conclusions can not be confidently drawn, as the radiocarbon date postdates the appearance of Kisapostag in the Carpathian Basin almost 200 years, meaning that it could be an outlier of Kisapostag ancestry, we do not know YET.
bce said…
first I modelled Baltic BA in vahaduo with Iron_Gates, Barcin and Yamnaya. There was around 46% Yamnaya.

then in Excel I subtracted Yamnaya from each coordinate, with the formula

(Baltic_Ba_coordinate- 0.46*Yamnaya_coordinate) / 0.54

Then I modelled the resulting coordinates with vahaduo, now they scored no steppe, only Iron_Gates and Barcin. Then I removed the Barcin from these coordinates the same way.
bce said…
@Daniel Gerber
are you aware of the samples HUN_MBA_Vatya_o RISE479 from AllentoftNature2015,
and HUN_EIA_Prescythian_Mezocsat_o1 I18241 from Patterson 2021, and how do they compare to the new Encrusted Pottery samples?

because these are the most WHG-rich of the available Pannonian samples, and this PCA from Szeczenyi-Nagy's presentation shows that samples from "Hungary_EBA_MBA2", Serbia_IronGates_BA and Croatia_EBA have even more WHG than RISE479:

from this it looks like these hunter gatherers passed trough the Iron Gates into Serbia and Croatia, and then north to Hungary, the Encrusted pottery samples being one of their early northern offshoots.

does your data suggest a different direction of the migration?
bce said…
I just noticed you did something similar for Mezocsat and Vatya, could you share your methodology?
EastPole said…

I was looking for HG populations which when mixed with Mezocsat and Vatya would produce Mezocsat_o1 and Vatya_o.
Using G25 coordinates and averages for Mezocsat and Vatya I calculated in Excel HGm and HGv coordinates using following formulas:

0.5*(HUN_EIA_Prescythian_Mezocsat + HGm) = HUN_EIA_Prescythian_Mezocsat_o1
0.5*(HUN_MBA_Vatya + HGv) = HUN_MBA_Vatya_o

I am thinking what HG pops would survive for such a long time and have such influence on others. I guess it had to be quite populous. Maybe a Mesolithic pop which switched to farming early like Bug-Dniester culture?
bce said…
a PCA with simulated HGs from Baltic_BA, ROU_C_o, and Pannonians
Daniel Gerber said…
@bce I am aware of these samples, Vatya outlier is technically an Encrusted pottery outlier, check uniparental there as well. Yes there are certain ambiguity on northern introduction, however I still stand with it for the generations older samples from Poland and Germany linked to this special HG type. Southern appearance of this ancestry is on time with the appearance of the Kisapostag in the Transdanubia, which of course could mean southern introduction, but also could signalise rapid spread, which I believe was the actual case, and thus - again, according to older samples from the north, and no older samples from south - I stand with the northern introduction.
Anthro Survey said…

I wonder how your HG ghost compares to one I made earlier. Not near my files right now, but I had a similar approach except that I used a more proximal source to get the "ballpark" modeling of Baltic_BA, as opposed to Yamnaya+Barcin. So, I used a Corded Ware average comprised of PCW and Fatyanovo-like CWC and subtracted this from Baltic_BA. Again, operating on the assumption that "standard" CWCs picked up this cryptic HG from a prolonged stay near the northern Carpathian foothills. In your case, you seem to favor (an early) admixture event of early CWC(or Yamnaya) profiles with Cryptic HGs. Naturally, your ghost has a higher content of WHG proper and EEF.

Regarding it being a founder effect and intragroup variation: that is of course possible, definitely. We just won't know for sure the dominant flow of things until we can get more mountain samples, imo.

Do bear in mind, though, that there are at least two types of BA Pannonians, and this is best seen in the Mokrin set. Some are just "steppe"+EEF balanced mixes, whereas others definitely show the strong Cryptic shift. Given the heterogeneity of the Mokrin set, a high degree of mobility is inferred.

So, if I understand, you basically suggest the following sequence: Baltic_BA-like folk(already harboring cryptic HG) penetrate into Pannonia. While there, they absorb some Rou_C_o-rich populations and also mix with EEF+steppe mixes(who also potentially harbor some Rou_C_o admixture).

My scenario involves steppe-rich groups(CWC or otherwise) moving into Pannonia without any HG tacked on. While there, some absorb varying degrees of Cryptic_HG+WHG from the hillier refuge zones, whilst others mainly absorb Baden-like. Over time, these two profiles mingle and create tertiary mixes in places in this highly-mobile context.

And, I definitely echo what you say about eastern Hungary. "Celtoid" admixture does seem to be important in attenuating Encrusteds' profile in the West. But I also think there was some flow from the Balkans and/or surge in EEF-steppe BA Pannonians(in regards to Hrv MBA/EIA).
bce said…
@Anthro Survey

I don't think anything Baltic_BA-like was involved in Hungary. I think Usatovo was highly drifted and WHG-rich (maybe something like 70%), maybe with some minor eef and steppe.

They were pushed by the Yamnaya expansion, and moved west trough Romania, mixing with ROU_C (eef-rich), ROU_C_o (IronGates_HG-rich), and finally with Yamnaya too. And by the time they reached Hungary, they were already complex mixes like the "Mako" sample.

Then around 2000 BC a mix of CWC and remnants of pure Usatovo (without the extra IronGates, eef, etc.) somehow appeared in the Baltic.

altvred from Anthrogenica ran DATES analyses which show that Mako and Spiginas2 are both a result of a very recent Steppe+WHG mix, we are looking in both cases at members of young populations which had just formed.

the Iron Age Mezocsat outlier turned out to be a very old mix, dating back to the Mako period, which confirms this method.
bce said…
modeling BS-drifted populations in G25 is very tricky. All CWC groups seem to have minor BS-drift, and since G25 is very sensitive to it, CWC might be getting inflated in these models.

I tried with IronGates and Yamnaya because they both have a slight BS affinity, and I was hoping they would cancel each other and none would get inflated.

I'm very satisfied with the resulting ghosts. On a PCA like this, they plot together with WHG/EHG groups, meaning I've succesfully extracted the Steppe/EEF from them, and I didn't remove too much either (in that case they would plot in the empty space, away from both WHG/EHG, Steppe and EEF)
bce said…
@Daniel Gerber
Thanks for the response, it didn't occur to me that Encrusted Pottery could be the core population from which this WHG admixture spread, but now that you pointed it out, it does make sense.

Now it remains to see the exact location and dating, autosomes and uniparentals, and archaeological context of those unpublished samples from Szécsényi-Nagy, to see how exactly they relate to the Encrusted Pottery population.
Matt said…
Hi arza,

Hope all is well with you. Things are quiet on ancient dna at the moment but thought I might point you to this as might be one to watch for future -

"Sam Morris was awarded his PhD (as part of the BBSRC LiDO DTP) for his thesis on "Harnessing haplotype sharing information from low coverage sequencing and sparsely genotyped data". "

"Thesis abstract: The proliferation of DNA from ancient remains is revolutionising the understanding of past population structure and demography. However, these data are often sparse (<1x coverage), making it challenging to extract reliable haplotype information from them, i.e. to model associations among linked SNPs. While this may not be essential when inferring population sub-structure involving genetically diverged groups, larger cohorts of samples from geographically proximate regions are emerging. In such cases, subtle genetic differences may be captured with haplotype information, demonstrated in analyses of cohorts containing geographically nearby present-day individuals.

Among such haplotype-based methods, Chromopainter has been used extensively in ancient DNA papers that analyse higher coverage (>1x) samples. Lower coverage data can be imputed to provide the dense SNPs that haplotype-based techniques require. However, imputation typically uses modern reference panels, and may be sensitive to effects that obscure fine-scale ancestry signals. Although efforts have been made, there has yet to be a detailed characterisation of the effect of imputation bias on Chromopainter analysis of a range of available aDNA populations for different coverages, and whether more power can be gained with new techniques.

In this thesis I propose modifications to the Chromopainter method to assist extracting haplotype information from low-coverage samples. These include accounting for allele probabilities, retaining SNPs adhering to specific criteria, and upweighting the likelihood contribution of genomic regions containing higher coverage SNPs. I explore these in the context of detecting subtle population structure, including measuring the loss in haplotype information from subsetting data to SNPs that overlap among SNP arrays.

I use these insights to analyse two datasets of newly sequenced ancient samples from Bavaria and Slavic-speaking regions."
ambron said…
In new Avar paper there are many genetically Hungarian samples from the 4th and 5th centuries. This proves the genetic continuity of Hungarians from the Bronze Age, through the Iron Age and the Middle Ages, to the present day. And let us recall that before the arrival of the Magyars, the same Hungarian population was Slavic.
ambron said…
Arza, how are you? You haven't written for so long that I'm starting to worry.
ambron said…
Arza, are you planning any entry about Goths from Weklice?
Anthro Survey said…

I would be more cautious in using the term “genetic continuity”. In terms of macro-components? Sure. But it doesn’t at all preclude large population turnover mediated by a genetically similar population.

Iron Age Hungary had a population harboring a “Baltic-BA-like” component of ancestry. But it doesn’t mean that Hungarians more or less descend from them in situ. In fact, even the Roman/late Roman record shows that Hungary was already a checkerboard with ancestry from the south(Balkans) and Germanic zones coming in. And then, as we all know, the Slavs come in a massive way.

Point is, most of the Baltic_BA is from the new Pannonian Slavs+later Polish, Slovak, and Serb infusions. Not from those Iron Age locals. In fact, I think those early modern Slovaks/Poles and Serbs had more a contribution than people realize.
Anthro Survey said…
In fact, check out the models below, without and with a proto-Slav-like proxy(Polesia) which clearly rescues the fit. I account for Western and “urban cosmopolitan” ancestry(with Near East signal) with French_Nord and Collegno_O respectively. As you can see, Iron Age Hungary is not the major source of Baltic_BA like ancestry in the modern population. Slav ancestry is fundamentally different to that of the largely dead-end Iron Age Hungarian population, being comprised of heavy Baltic_BA-like and Iron Age *Balkan*(Dacian?) ancestry. Iron Age Hungary has markedly less Baltic_BA and the rest isn’t exactly modeled as Balkan-like.

Target: Hungarian
Distance: 2.6629% / 0.02662937
82.6 Scythian_HUN
17.4 French_Nord
0.0 ITA_Collegno_MA_o1

Target: Hungarian
Distance: 0.7784% / 0.00778400
52.6 Polesia
37.4 French_Nord
10.0 ITA_Collegno_MA_o1
0.0 Scythian_HUN
ambron said…
Try to use sample I25524 in place of Scythians HUN.
ambron said…
Note that the first model, despite having a worse fit, is more economical as it does not require the addition of an additional, third component.