search this blog

Friday, August 12, 2022

Mediterranean PCA update


I updated my Principal Component Analysis (PCA) of Mediterranean populations with most of the ancient Jewish samples from the Waldman et al. preprint. To view intereactive versions of the plots paste the data from here into the PCA DATA field here and press PLOT PCA. The ancient Jews are labeled DEU_MA_Erfurt.

See anything interesting? I'm again seeing more complexity than claimed by Waldman et al., but what would I know anyway?

See also...

My take on the Erfurt Jews

Greeks in a Longobard cemetery

224 comments:

«Oldest   ‹Older   201 – 224 of 224
Copper Axe said...

@Davidski

I was about to go to bed but alright. Focusing on the bronze age:

- The Yassitepe samples are a bit to the south of Troy.For sure an Anatolian presence but idk if it would be a Luwic language or Lydian. There is a big debate in general how far west Luwian proper was spoken and how prominent it would've been.

- Amasya as far as I can tell is on the "border zone" between Palaic (Anatolian) and Kaskian (unclassified).

- Isparta is squarely within the zone of Luwic language diversity. Lycian maybe?

- Kilis is a bit tricky it depends on the dates. It is right where the Syro-Hittite states were after the Hittite empire collapsed. If the samples from here were within the Hittite period, and a bit on the earlier side, they might be relevant for Hittite.

Despite the Syro-Hittite name Luwian was the most prominent Anatolian language there, as Hittite was getting replaced and ultimately went the way of the dodo after the empire collapse. However this definitely does not guarantee that the samples actually were from Hittites or Luwians or people with genetic interactions with them. Same goes for Gaziantep.

These are the sites that stood out to me. Some Anatolian languages persisted throughout the iron age, into the common era, but that won't be helpful for the genetic debate ofc. I might have missed some hopefully someone will point them out.

Its kinda funny how they talked such a big game about Anatolian languages but provided a relatively meagre amount of relevant period anatolian genomes. I get that the cremations don't help but still...

But I honestly dont really see a lot of sites I'd associate with core hittite territory. I'd also say their importance in the IE homeland debate is a bit overstated. The other Anatolian languages have more weight in the debate imo. Hittites were cool and all, but demographically they were a non-factor and culturally they were quite far removed from their linguistic relatives. Luwian was more widely spoken in the Hittite empire than Hittite itself, and Hittite as a language went extinct not long after the Hittite empire collapsed because it was only kept alive as an official language.

Davidski said...

@All

Stop quoting large chunks of the paper here.

For one, it's not allowed, and secondly I don't care.

The paper is crap, and I'm about to explain why it's crap in a new blog post.

Davidski said...

@Copper Axe

Thanks. Sweet dreams.

Sam Elliott said...

Re; EAA abstracts, this one:

"New radiocarbon dates from the Usatovo culture suggest a re-evaluation of the culture’s chronology"

"Usatovo is one of the kurgan cultures of Eneolithic southeast Europe, occupying a strategic crossroads between the eastern Balkans and the Ponto-Caspian steppe. Usatovo is considered to be related to the farming cultures of southeastern Europe as well as to the pastoralist groups of the North Pontic. Usatovo currently dates to the second half of the 4th millennium BCE, based on C14 measurements obtained on pottery and charcoal using the scintillation method, as well as through ceramic typologies. Recent AMS radiocarbon dates obtained on human remains from the Usatovo site of Mayaki fall within the second half of the 5th millennium BCE. Stable isotope values (δ13C and δ15N) obtained on human remains are atypical for farming or hunter-gatherer communities of southeast Europe and imply a heavy input from freshwater and/or marine resources in diet. Stable isotope ratios suggest the presence of Reservoir Effect (RE) potentially offsetting the dates from the human remains, although its quantification is difficult to establish due to the lack of contextual faunal remains. Even with the RE taken into account, the discrepancy between the scintillation C14 dates/ceramic typology and absolute AMS dating may still remain. Both anthropological and ancient DNA data are consistent with Usatovo being an admixture of farmer and steppe ancestries. The timing of this admixture may help refine the Usatovo chronology."

Question: Any estimate on when this Usatovo study might be published? There are several samples from the western steppe that I thought would have been included in the Southern Arc study that are conspicuously missing. As a matter of fact, one of the authors had previously indicated that these samples were intended for the Southern Arc study. I’m assuming because of their age and burial location they were instead relegated for this Usatovo study?

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

Funny story, I went to the UK and got married this month. We went to visit Stonehenge on our honeymoon and went on one of the highest recommended tours. The tour guide explained that it was built by Celts who first arrived in Britain after the last Ice Age, best exemplified by Cheddar Man who had dark skin and blue eyes. Did you know Cheddar man was a Celt and he built Stonehenge? I didn't even say anything it would have been hopeless. All I could do was laugh.

old europe said...

@copper axe

I remember Rob some months ago provided a different model for the Smydovo sample with the proportions of Ukraine Neo and steppe reversed. IIRC it was something like that

60 EEF
27 Ukraine neolithic
13 steppe eneolithic ( Yamnaya)

ambron said...

Matt, Arza has returned to activity on his blog and is currently dealing with these Middle-European samples with Balto-Slavic drift.

Gaska said...

Harvard is trying to resurrect Yamnaya and put an end to Max Planck's CWC-Bohemia theory (famous Narva outliers-R1b-L151 signal).

Glad to speak again about Smyadovo, because it is the oldest M269 we have and remember that everybody can consult the archaeological data of its tomb. He is a Gumelnita culture-Neolithic farmer, nothing to do with the steppes(at least archaeologically speaking). Harvard acknowledges that finding out the origin of this lineage is a challenge (i.e. they only have suspicions, but no evidence to link it with EHG). I remember the conversation with Arza two years ago when everyone was denying that it was even M269 (even after Lazaridis acknowledged it via twitter).

I understand that Lazaridis did not want to include it in his conclusions about southern arc's male markers because its dating is very old (4,500 BC) and it complicates his theory about the origin of M269 and of course totally disassociates it from the Yamnaya culture. Good luck Iosif, you have a hard work ahead of you.

Regarding Anatolia and Greece, what everybody expected, i.e. total absence of steppe markers and irrelevant and insignificant results of autosomal Yamnaya signal (EHG). Their attempt to link R1b-PF7562 with Yamnaya is ridiculous because the two samples found in the Palace of Nestor are after the collapse of the Bronze Age (1.150 BC)-The Mycenaean were overwhelmingly G2a and J2a and at the moment there is not even Z2103 in Mycenaean territory. The Balkans have nothing to do with the male markers supposedly involved in the spread of IE. There is no trace of L51 in the western Balkans, Greece, Anatolia, Armenia, Iran, Levant etc etc, which confirms its absolutely marginal role (if it had any).

BTW the effort is appreciated because there are dozens of interesting samples.

LGK said...

@Gaska

"Their attempt to link R1b-PF7562 with Yamnaya is ridiculous because the two samples found in the Palace of Nestor are after the collapse of the Bronze Age (1.150 BC)"

Who else do you suppose these people are and why is the timing an issue? They are either Mycenaeans or (somewhat early maybe) Dorians, both are linguistic and cultural Greeks and ultimately descended of Yamnaya males

epoch said...

@David

"Stop quoting large chunks of the paper here.

For one, it's not allowed"

Reich lab has a free copy already online.

@All

The gist of the paper is that WHG admixture in Yamnaya is rejected and the model needed Levant admixture. So they state that they reject the model of Wang et al. That is the new info.

I looked at the S5 excel sheet but almost half of the Yamnaya samples seem to score 0% Levant. Don't know if I miss something, though.

Davidski said...

This paper hasn't rejected the Eneolithic steppe/European farmer model.

It couldn't because it used a very broad, distal model that missed basic details.

John Smith said...

Hi Davidski,

In the file you shared yesterday with all the Global25 coordinates, there are about 1268 samples that do not belong to this Southern Arc study but rather are old samples, possibly re-sequenced.

You decide whether to add / update them to the spreadsheet.

https://pastebin.com/raw/Y22BRS3V

Dospaises said...

I2181 Smyadovo has possible deamination reads for CTS9018/PF6484 (C>T) and PF6452/YSC0000167 (G>A) which are R1b1a1 and R1b1a1b respectively. This calls into question the accuracy of the results although there is only one definite false positive due to deamination (FGC37690). Higher resolution samples are needed to determine the accuracy.

Arza said...

@ Matt

Yeah, S2949 is Dzudzuana. But there's another Georgia UP sample - S11857.

Re: I5243
Balto-Slavic drift without any doubts.

Arza said...

@ Dospaises

The final SNP is in the middle of the read sequence where the probability of deamination drops significantly.

Simon Stevens said...

@Arza

What’s a good resource for uncovering which areas suffer from high rates of deamination and which do not? And what programs do you guys use to analyze Y-DNA?

Arza said...

@ Simon Stevin

General rule is that deamination is most likely at the ends of the read. See e.g. https://www.pnas.org/doi/10.1073/pnas.0704665104

Call on the M269 level is outside of the region with a high deamination risk.

Yleaf + LibreOffice Calc for manual inspection, samtools if you want to check position of an SNP in the read.

Simon Stevens said...

@Arza

Much appreciated sir, thank you for answering and helping me.

gsueso said...

Noob question... What coordinates does this PCA use, and how can I add more samples to it?

Unknown said...

I think you need to update your Albanian averages, they are more East than any Albanian average I have seen and it does not say where they are from. Here I got over 100 Albanian samples broken down into different averages

Albanian_Korçë,0.12302,0.14544,0.01538,-0.01564,0.02552,-0.00124,0.00026,-0.00072,0.00242,0.01266,-0.00032,0.0021,-0.00588,0.00424,-0.01294,0.00004,0.00876,0.00042,0.00424,-0.00482,-0.00664,0.00108,0.00068,-0.00048,0.00232
Albanian_Labëria,0.11954615,0.14466154,0.0129,-0.019907692,0.022992308,-0.0051538462,-0.000084615385,-0.00081538462,0.004,0.015184615,0.00076923077,0.0033769231,-0.0067769231,0.00077692308,-0.010015385,0.00097692308,0.0073384615,0.00022307692,0.0031461538,-0.0038615385,-0.0038230769,0.00053076923,0.00037692308,-0.00092307692,0.00026923077
Albanian_Tropojë-Gjakovë,0.1227,0.14686,0.01738,-0.01518,0.02732,-0.00336,-0.00012,0.00064,0.00334,0.01646,0,0.004,-0.0092,0.00056,-0.00988,0.00074,0.00806,0.0002,0.00318,-0.00284,-0.00548,0.00092,-0.00042,-0.0002,0.00048
Albanian_Dibra,0.1220716,0.1464068,0.0146516,-0.0200808,0.0259236,-0.0061182,0.001511,0.0002814,0.0015192,0.0169838,0.0007522,0.003908,-0.0085242,0.0052854,-0.0155376,0.0011286,0.0154266,-0.0000666,0.0057018,-0.0049814,-0.007322,0.0003786,0.0008056,-0.0010482,-0.0013252
Albanian_Central_Albania,0.11976,0.1461,0.0129,-0.02186,0.02414,-0.00558,0.00028,-0.00128,0.0046,0.01858,0.00032,0.00286,-0.00684,0.00124,-0.0102,0.00174,0.008,0.0012,0.00302,-0.00248,-0.00334,0.0016,-0.00072,-0.00046,0.00102
Albanian_Mirditë,0.120524,0.14547375,0.0159095,-0.020123,0.0244255,-0.00776375,0.00094125,-0.00101525,0.0012,0.02152425,-0.00089,0.002936,-0.0118215,-0.0031395,-0.01339875,0.0006915,0.0127645,0.00101675,0.0046275,-0.004877,-0.00644825,0.002372,0.00307425,0.0040085,-0.003198
Albanian_Pukë,0.120325,0.143725,0.01315,-0.016575,0.024275,-0.00425,-0.000625,0.0004,0.003925,0.01555,-0.0004,0.00225,-0.00675,0.001475,-0.009375,0.0021,0.008,-0.000775,0.002425,-0.0046,-0.005,0.0015,-0.000175,0.000575,0.00055
Albanian_Himarë,0.11856667,0.14563333,0.0139,-0.022333333,0.023966667,-0.0074666667,0.0005,-0.0018333333,0.0050333333,0.016166667,0.0017333333,0.0042333333,-0.0063666667,0.00016666667,-0.0090666667,0.00023333333,0.0060666667,0.00063333333,0.0031333333,-0.0042333333,-0.0019,-0.00046666667,0.0003,-0.0026,-0.0012333333
Albanian_Northwestern_Albania,0.12250414,0.14578543,0.018467571,-0.014054286,0.027054714,-0.0033674286,0.0013371429,0.0019691429,0.0038122857,0.015923429,-0.00065328571,0.0041131429,-0.0071024286,0.001901,-0.011136857,-0.0013804286,0.0070728571,0.00078385714,0.0039077143,-0.004279,-0.0048454286,0.0014961429,-0.00058242857,0.000877,0.00073685714
Albanian_Montenegro,0.124484,0.1439887,0.0213772,-0.0100424,0.0286827,-0.0047728,0.00146,0.0017577,0.002775,0.0146583,-0.0002061,0.0014157,-0.0071503,0.0037379,-0.0132531,-0.0009117,0.0074787,0.0011261,0.0040591,-0.0053111,-0.0064995,0.001472,0.0003751,-0.0011684,0.0003354
Albanian_Kosovo,0.12295462,0.14478773,0.017853885,-0.011466462,0.026755269,-0.0027497692,0.00088576923,0.0013136154,0.0022273077,0.012821846,0.00018711538,0.0022654231,-0.0073647692,0.0029782692,-0.011526423,-0.00073811538,0.0084285769,0.00013642308,0.0035097692,-0.0056040769,-0.0053803462,0.0016953077,0.000376,-0.0017949231,0.0010002308

My G25:

Albanian_Skenderaj,0.1241,0.1448,0.0232,-0.0118,0.0284,-0.0023,0.0027,0.0009,0.0021,0.0149,-0.0012,0.0031,-0.0089,0.0014,-0.0115,-0.0023,0.0057,-0.0013,0.0069,-0.0047,-0.0043,0.0048,0.0013,-0.0038,-0.0007



Greeks don't shift more west than Albanians.


These averages were made from over 100 individual samples.

Mohamed Amerkanti said...

Hello David,
Can you share all the samples you used to make this map ?

Davidski said...

It's based on quite a few samples that aren't available to the public, so I can't do that.

But you should be able to produce a similar plot just by using the same populations as those that are visible in the PCA.

AlbCoder said...

How are those samples labeled as 'Albania_modern_o_Caucasus' in your ancient spreadsheet supposed to be outliers or from the Caucasus ? These samples plot similar to medieval samples from Albania actually and Roman Illyrians.. It's those two samples you have labeled as 'Modern_Albania' or so that shift North-East and are atypical for their time frame

AlbCoder said...

I was wondering btw is there any special reason you removed
the medieval sample from Albania I13834 from your list ? Because I needed it's G25

«Oldest ‹Older   201 – 224 of 224   Newer› Newest»