Friday, August 7, 2020

New Global25 interpretation tools


They're available at Ancestry Calculator and GENOPLOT. Unfortunately, I can't tell you exactly how to get the most out of them. All I can recommend is robust experimentation.


See also...

Getting the most out of the Global25

Genetic ancestry online store

Modeling your ancestry has never been easier

103 comments:

  1. Modern Spain is mix between CeltIberian and North Italy.

    Spain after Bronze age
    https://docs.google.com/spreadsheets/d/1NiQF03JhBMVWZh7MCTVZ2KuwUvDj39xy0yRYoptaoko/edit?usp=sharing

    This could mean, Spanish are largely of Roman Italian origin. If so, it was Romans who did not have any Near Eastern admixture.

    This could be a pretty big historical importance for Roman empire. Because it would mean there was large scale immigration from Italy into Spain during Roman empire.

    France probably also has Italian admixture from Roman era, but much less than Spain. Maybe, when Near East immigrants came to Italy, native Italians emigrated into Gaul and Spain.

    ReplyDelete
  2. @Samuel

    I'm pretty sure that there was some recent Near Eastern admixture in many of the Romans who moved to Iberia.

    As far as I can see, the trajectory of the post-BA/IA genetic shift in Iberia points to this. Take a look at the PCA in this post.

    Open thread: What are the linguistic implications of Olalde et al. 2019?

    ReplyDelete
  3. @Davidski,

    When I include Southern Italy and Near East sources in nMonte, Spain scores 0%. Which is I tend to think they lack Near East admix. But, G25 isn't the final word. I've heard there is Near East Y DNA J2 in Spain which would support the idea of Near Eastern admix.

    Anyways, it probably isn't coincidence Basque don't speak a Romance language and also don't have admixture associated with Roman empire. Which is interesting. It's also interesting they have less Celtic admixture than Spanish do.

    It turns out Basque are pure representatives of a by gone era. But that era was the Iron age not Upper Paleolithic. Lol.

    ReplyDelete
  4. North Italy has a germanic shift but at the same time they have limited west asian influences from South-Central Italy(Greco-Anatolian) etc.I think North Italians and Spaniards clushter because they both share pretty much this EEF+Steppe(BB/Celtic)+WHG.South French are not that far away either.

    ReplyDelete
  5. @Crm

    Georgians from eastern parts are more likely Leyla-Tepe/Kura-Araxes/Trialetii influenced.I agree with you about the Levant N admixture but they are highly CHG admixed and the only way to excuse it is a North caucasus source.The KA culture never reached west caucasus...so i do not see a kartvelian relation tbh.I think Dolmen samples in the future will solve the genetics of Colchians.This paper might help thought.

    https://www.nature.com/articles/s41467-018-08220-8/figures/4


    Dolmen LBA on fig.4 map looks mostly CHG with some Anatolian copper age admixture.This is probably the key of genetics for West Caucasians.The rest admixtures like Anatolian BA brought probably increase in components like ANF,Levant N and Iran N into west caucasus.

    ReplyDelete
  6. @Samuel

    Modern Northern Italians have Germanic admixture, I would not recommend them to be used in modeling ancient Roman colonization.

    ReplyDelete
  7. Also, one should not think of what happened in Iberia as a one-time punctuated admixture. The Latinization facilitated an environment suitable for gene flow with other Romance-speaking regions for centuries. It is that centuries-long process of gene flow that created the modern Iberian genetic structure (we should also take into account the gene flow processes facilitated by the Islamization in Andalus and by the Germanic invasions of course).

    ReplyDelete

  8. @Anatolian Farmer

    “only way to excuse it is a North caucasus source”
    Not quite, for the CHG-rich source in North Caucasus reached it from the South Caucasus, see Darkveti-Meshoko culture, Darkveti being in South Caucasus and Meshoko in North.

    Northwest Caucasians would have been more CHG-rich than Kartvelians had they not mixed with Steppe peoples, because their local Caucasian base was slightly more CHG-rich, but it differentiated from the Kartvelian base by having more Levant ancestry and less Iran_N. The Levant proportions in Meshoko are slightly off for Georgians, but more fitting for NWC, I assume same would apply to Dolmen.

    "The KA culture never reached west caucasus...so i do not see a kartvelian relation tbh."
    Yes, but that's not the point, I'm talking about the opposite. Sioni (a transitional culture between SSC and KAC, with relations to Darkveti) expanded from West Georgia and mixed with Southeast Caucasians to produce KAC. Sioni may have been Kartvelian, because in order to create KAC admixture you need an increase of CHG-related ancestry in SSC and Areni. That CHG-rich source should be like Darkveti but with less Levant_N, and proto Kartvelians are a perfect candidate for that, hence why I assosciated Kartvelians with Sioni.

    Think "Georgian_Imer_without_AnatoliaBA" as a possible proxy for Darkveti/Sioni

    Target: Georgian_Imer_without_AnatoliaBA
    Distance: 2.5700% / 0.02569991
    60.2 GEO_CHG
    35.8 TUR_Tepecik_Ciftlik_N
    2.2 IRN_Wezmeh_N
    1.6 RUS_Progress_En
    0.2 Levant_PPNB

    Target: Kura-Araxes_ARM_Kaps
    Distance: 1.8336% / 0.01833600
    55.6 Georgian_Imer_without_AnatoliaBA
    41.2 IRN_Hajji_Firuz_C
    3.2 RUS_Progress_En
    0.0 ARM_Areni_C
    0.0 AZE_Caucasus_lowlands_LC
    0.0 AZE_Caucasus_lowlands_LN
    0.0 RUS_Darkveti-Meshoko_En

    Target: Kura-Araxes_ARM_Kaps
    Distance: 2.4195% / 0.02419513
    53.4 Georgian_Imer_without_AnatoliaBA
    28.2 AZE_Caucasus_lowlands_LN
    7.6 RUS_Darkveti-Meshoko_En
    7.4 AZE_Caucasus_lowlands_LC
    3.4 RUS_Progress_En
    0.0 ARM_Areni_C


    Now, on the matter of languages, there's a lot of evidence that KAC spoke NEC and even Hurrian, so it's possible that Sioni were NEC speakers, but it's also possible that NEC is pre Sioni, deriving from SSC/Leilatepe/Dalma. Kartvelian languages likely arrived later in East Georgia, after KAC, and is related to the MBA Shida-Kartli culture, (Samtavro necropolis holds layers from that culture), which evidently had ties with West Georgia, which indeed was not a part of KAC horizon. Perhaps it was a second attempt at a Kartvelian expansion, this time being more linguistically impactful?

    ReplyDelete
  9. @ Crm

    I agree that Northwest Caucausians would have been more CHG rich admixed if they didn't have Steppe autosomal but where exactly do you see the big diffrences in Levant N admixture between Kartvelians and Northwest Caucasians???I am playing with them on G25 and their Levant N admixture is pretty much the same(thought some samples from Abkhasians are slightly more).They do not score more than 6% Levant N.The Mesoko samples are also somewhere 5-6% Levant admixed.One sample from RUS_North_Caucasus_MBA is also 6% Levant admixed.Iran N is limited as well.As for the rest they sound okay.Languanges have to do like you mention above,but modern Georgians are less CHG admixed probably because of their BA Anatolian admixture.I am pretty sure both Georgians and Northwest Caucausians received a gene flow from 'Pontus' of North Anatolia hence their decent ANF admixture.

    ReplyDelete
  10. @Anatolian Farmer

    NWC, despite being admixed with Steppe peoples that were rich in both Steppe EBA and East Asian (those being people with little to no Levant_N) still show more Levant_N admixture than Imeretians (who are not mixed with Steppe-rich populations)
    https://i.imgur.com/wHtd36D.png

    Extra Levant_N contribution must have been among the things that set the difference between the proto-NWC tribes from proto-Kartvelian tribes, the latter were probably slightly more Anatolia-rich. It would be interesting to see what Megrels and Svans are like, they should be like Imeretians but with more CHG. I'm curious about their non CHG proportions, especially Iran_N because that's another thing that separates Georgians from NWC, former have more Iran_N, but this may have to do with Imeretians being the easternmost West Georgians, perhaps Svans and Megrels have it less.

    And looking at Darkveti-Meshoko and North Caucasus MBA, you can see two individuals among them that are slightly different from others, both showing more Anatolia_N but less Levant_N, their proportions are more Kartvelian-like.
    https://i.imgur.com/JZGLcAX.png

    Same story with Maykop. Especially Maykop Novosvobodnaya which is very Georgian-like.
    https://i.imgur.com/OQ8dZdx.png

    ReplyDelete
  11. When I tested these samples (scaled; using https://vahaduo.github.io/vahaduo/):

    Iberia_North_IA:I3758
    Iberia_North_IA:I3759
    Iberia_East_IA:I12879
    Iberia_East_IA:I3324
    Iberia_Northeast_Empuries1:I8209

    Using these ancient samples:

    HUN_Koros_N_HG
    TUR_Barcin_N
    Yamnaya_UKR
    CMR_Shum_Laka_8000BP
    MAR_Taforalt
    Levant_Natufian
    IRN_Ganj_Dareh_N
    Levant_PPNB
    CZE_Vestonice16
    BEL_GoyetQ116-1
    Iberia_ElMiron
    RUS_Karelia_HG
    MNG_North_N
    RUS_Baikal_BA
    RUS_Tyumen_HG
    LAO_Hoabinhian
    RUS_Devils_Gate_Cave_N
    JPN_Jomon
    RUS_Khvalynsk_En
    SYR_Ebla_EMBA

    I get these results:

    Target: Iberia_North_IA:I3758
    Distance: 5.2775% / 0.05277543
    50.8 TUR_Barcin_N
    22.6 HUN_Koros_N_HG
    21.4 Yamnaya_UKR
    3.8 RUS_Karelia_HG
    1.4 MAR_Taforalt

    Target: Iberia_North_IA:I3759
    Distance: 3.9347% / 0.03934658
    51.2 TUR_Barcin_N
    23.8 HUN_Koros_N_HG
    23.6 Yamnaya_UKR
    1.4 JPN_Jomon

    Target: Iberia_East_IA:I12879
    Distance: 5.3186% / 0.05318614
    52.2 TUR_Barcin_N
    26.6 Yamnaya_UKR
    21.2 HUN_Koros_N_HG

    Target: Iberia_East_IA:I3324
    Distance: 4.5866% / 0.04586606
    50.4 TUR_Barcin_N
    26.4 Yamnaya_UKR
    20.8 HUN_Koros_N_HG
    2.4 JPN_Jomon

    Target: Iberia_Northeast_Empuries1:I8209
    Distance: 4.6012% / 0.04601185
    51.0 TUR_Barcin_N
    28.0 Yamnaya_UKR
    19.0 HUN_Koros_N_HG
    1.4 RUS_Khvalynsk_En
    0.6 RUS_Tyumen_HG


    That Jomon signal is odd.

    ReplyDelete
  12. @ Crm

    First of all why do you use Iran Wezmeh?Second how accurate Levant PPNB samples are to estimate the Levant N admixture among caucasians?Most people use Iran N and Levant Natufian samples(thought even the Natufian samples are kinda old to estimate Levant N admixture).

    ReplyDelete
  13. @Onur,

    North Italians don't have German/Lombard admixture.

    If you look at West eurasia PCA (which you can create using G25 data), North Italians pull away from Central/South Italy in a northern direction.

    Lombard DNA samples cluster in modern Scandinavia, which is far northEAST of Italy. Lombards are so EASTERN because they had around 53% Steppe ancestry. If Northern Italians had signficant Lombard ancestry, they'd pull northEAST away from the rest of Italy.

    Instead, Northern Italians have the same Steppe ancestry ratios as did Iron age Latins and Etruscans. There's no indication they have any real Lombard ancestry.

    The northern pull of Northern Italy away from Central/South Italy in PCA, isn't due to Northern European admix in Northern Italy, it is due to Near Eastern admix in Central/South Italy which pulled them south of where Iron age Italians clustered.

    My opinon, is Northern Italians are a unique "genetic isolate" in Europe. They are distinct from nearby French and Germans and Slovenians. The Alps has been a strong genetic barrier probably since Bronze age. Modern Northern Italians are probably a near perfect continuation of Beaker, Bronze, iron age populations of Northern Italy.

    ReplyDelete
  14. Many moons ago, I ran all modern Italian samples through Vahaduo using these averages to test for proportions of realistic proximate ancestry:

    Etruscan - IA north/central Italian ancestry
    Italic IA - IA Italian ancestry
    Iberia Empuries1 - in my opinion, a fair proxy for what Cisalpine Gauls were probably like
    Imperial Roman East Med Profile - unresolved historical origin. Seemingly mostly a mix of Mycenaean-like, Anatolian BA-like, and historical Levantine ancestry, creating a profile characteristic of Southern Italians and Aegean Greeks; all outright Near Eastern outliers removed for this average.
    Szolad/Collegno Lombard Northern Profile - proxy for pure Germanic admixture. All southern-admixed samples from the cemeteries removed for this average.
    Berber Tunisia (Sened and Chenini) - proxy for North African ancestry. Substitute with Guanche to taste.

    I posted my findings on AG (which is down right now), but from what I recall pretty much all Italians scored some form of Imperial Roman and "pure" Lombard ancestry using this model. Not saying it's the best out there, but it's a fair start.

    ReplyDelete
  15. @ CRM

    “ That CHG-rich source should be like Darkveti but with less Levant_N”

    Why would Levant N figure here ? The admixing source into Colchian CHG would be Tepecik

    ReplyDelete
  16. If Northern Italians have Lombard and Near East admix as some claim.....

    They should fit well as mix between Iron age Italy, Lombards, and Imperial Rome.

    ReplyDelete
  17. @Anatolian Farmer
    Because there's a preference of Wezmeh over Ganj Dareh/Abdul Hosein. Aren't Natufians from the Mesolithic? Any Natufian ancestry that exists in the Caucasus would have been brought through Levant PPN ancestry.
    https://i.imgur.com/jLvKXPP.png

    However, when subtracting any leftover Steppe and Ikiztepe from Georgians and Adyge it boils down to this.
    https://i.imgur.com/tUtuuiH.png
    I'm not sure the accuracy of the ghosts, but it shows that while there is a difference in Levant ancestry in favour of Adygeans, it may not be that significant in the end.

    It really makes me wonder why do populations with such similar genetic bases speak two unrelated languages. But since they're both a mix of local Caucasian forages and invading Anatolians, one of them could have adopted a non local language.

    ReplyDelete
  18. @Rob

    You can see it with Darkveti-Meshoko, Tepecik covers nicely the Anatolian and Levantine ancestry of I2056, but others are more Levant rich but have less Anatolia.
    https://i.imgur.com/ctyp4x4.png

    Same happens with Kudachurt, KDC001 showing a more NWC profile with more CHG, Levant but less Anatolia and Wezmeh. While KDC002 a more Kartvelian profile with less CHG, Levant but more Anatolia and Iran_N. I thought this had to do with slightly different Anatolian contributors to the Caucasus, some of them being more Levant-rich and mixing with NWC, while others being more Anatolia-rich and mixing with Kartvelians.
    https://i.imgur.com/LFkUQOt.png

    ReplyDelete
  19. Right; something further east from SE Anatolia
    It’s just hard to connect the dots; due the sparsity of Neolithic sites in Northeast Anatolia

    ReplyDelete
  20. @Samuel

    Modern Northern Italians have Germanic admixture, that is obvious. They have more steppe ancestry than all the post-Proto-Villanovan and pre-medieval Germanic migrations Italian ancient populations from which we have genomic data and they have Germanic Y-DNA haplogroups as well. Modern Northeast Italians (some if not all) have Slavic admixture in addition to the Germanic one.

    https://docs.google.com/spreadsheets/d/1a5byEejcptHST6X2L6Exb60uGXnyI6tZYI7HWG-EkkE/edit#gid=0

    https://drive.google.com/file/d/1vLLddJxbMeA1Rm02hrenN9uEvO-OoaL4/view?usp=sharing

    https://drive.google.com/file/d/18RLNqjQPselPnqlX6PwpypkufO4fA30F/view?usp=sharing

    S21 is the other name of U106, the most directly Germanic-associated branch of R1b. In Iberia the Suebi seem to have made a bigger genetic impact than the Visigoths based on the distribution of U106. In Russia, Ukraine and the East Baltic states its distribution aligns well with the known Viking migrations. In Ukraine and Southern Russia also with the East Germanic migrations. No need to explain the situation in the West Slavic countries, Hungary, Romania, France, Finland and Ireland, all of them are in accordance with history. In Northern and Central Italy it also aligns well with the known Germanic migrations, the Lombard migration being the most obvious one.

    When you look at linguistics, you see Germanic phonological (not just vocabulary) influences not just in the Romance languages of France, Belgium and Switzerland, but also of Northern Italy.

    ReplyDelete
  21. What ist Northwest Caucasian? There is a real and old nation which ist called Circassan (Adyghe/Cherkess/Kabardin. All the same people with different named given to them by the Soviets).Is it so difficult to call this people with Just their names as Circassans?

    ReplyDelete
  22. @Oncur,

    I'll show you PCAs and stats later. If Northern Italians do have any Lombard ancestry, I think we can agree it isn't very significant.

    About Y DNA. Amorium 2018 got 17 verifiable Lombard Y DNA sequences: R1b U106=11, I2a2a1b=5, I1=1, R1a Z284.

    Boatanni 2013, shows out of 366 Y DNA in Northern Italy:
    4% I1, 2.6% R1b U106, 2% I2a2a, 1.4% R1a.

    At most this is 10% Lombard/germanic Y DNA. If we assume invading armies always do sex bias admixture, this would mean less than 10% Lombard ancestry. But, of course uniparental markers isn't a way to measure ancestry.

    There's J2a and E1b V13 in Northern Italy than Germanic Y DNA.

    ReplyDelete
  23. @Onur, This is kind the discussion we're supposed to have on this blog. And I don't think either of us care if there is Lombard admix in Northern Italy or not. This is honest debate based on evidence.

    ReplyDelete
  24. @Freeman

    Abkhazians aren't Circassians/Adyghe, but they're still a part of the NWC family.

    ReplyDelete
  25. @ Crm

    Well,languanges varies a lot.The example with Georgians and NW caucasians is pretty much the same with Villanova,Etruscan and Roman-Latins..or even Ligures i would say.They all belong pretty much to the same spectrum genetically but their languanges are not exactly the same.Basques too.Anyway, about Colchian's autosomal i think we need more samples from western caucasus especially from IA period.Maykop culture vanished and KA is definitely not a good source to estimate the CHG to modern West Caucausians.You might be right about the Mesoko relation thought.Their ethnogenesis took place probably during LBA/EIA period with the arrival of Kaskians-Muskis.

    ReplyDelete
  26. @Onur Dincer

    The East Baltic R1b-U106 may even come from Baltic Germans, and the East Germanics probably didn’t leave a legacy considering they seem to have Asian influence and the area was inhabited by Turks until Slavs colonized the region.

    ReplyDelete
  27. @ Samuel

    I think you are wrong about North Italians genetics with all the respect.You can play in G25 with the samples from Lombardy and not only,use some middle aged post Germanic refrences and you can easily realize that they are far northern than BA/IA Italy.Anyway what i would like to mention is that some North Italic and South French samples are very likely to score some MAR_Taforalt admixture.Not much somewhere 1,5% i think it has to do with very ancient migrations(probably Carthage or something similar to Etruscan samples..).


    Target: Italian_Lombardy:BGD31
    Distance: 2.4146% / 0.02414611
    59.2 TUR_Barcin_N
    35.0 Yamnaya_RUS_Samara
    4.8 WHG
    1.0 MAR_Taforalt



    Target: Italian_Liguria:ALP099
    Distance: 1.9630% / 0.01963049
    56.2 TUR_Barcin_N
    35.4 Yamnaya_RUS_Samara
    5.4 WHG
    1.8 MAR_Taforalt


    Target: Italian_Piedmont:ItalyPiedmont136
    Distance: 2.0381% / 0.02038092
    57.4 TUR_Barcin_N
    38.2 Yamnaya_RUS_Samara
    2.2 WHG
    1.6 MAR_Taforalt

    Target: Italian_Piedmont:ItalyPiedmont145
    Distance: 1.9823% / 0.01982301
    57.8 TUR_Barcin_N
    35.8 Yamnaya_RUS_Samara
    3.2 IRN_Ganj_Dareh_N
    2.0 WHG
    1.2 MAR_Taforalt


    Target: Italian_Piedmont:ItalyPiedmont43
    Distance: 1.6763% / 0.01676344
    59.4 TUR_Barcin_N
    36.8 Yamnaya_RUS_Samara
    1.8 WHG
    1.2 MAR_Taforalt
    0.8 IRN_Ganj_Dareh_N


    Target: Italian_Piedmont:ItalyPiedmont63
    Distance: 1.9685% / 0.01968475
    57.6 TUR_Barcin_N
    35.6 Yamnaya_RUS_Samara
    4.0 WHG
    1.4 MAR_Taforalt


    ReplyDelete
  28. Whatever it is; the Lombards and Goth would certainly have been greater; but they were mostly killed during the Byzantine’s reconquests of Italy . They did all the fighting; don’t think many local Italian really cared between Goths or Byzantines

    ReplyDelete
  29. @Samuel

    The pure Lombard ancient DNA samples we have have about 44% Yamnaya-type ancestry on average, barely any different from the modern German average (46%). The post-Proto-Villanovan Italian populations from whom we have ancient genomes have as a rule less than 30% average Yamnaya-type ancestry until the medieval Germanic migrations. Modern Northern Italians have more than 30% average Yamnaya-type ancestry (33% on average). Sure, the Germanic ancestry in modern Northern Italians is relatively minor, but it is by no means negligible.

    ReplyDelete
  30. @Rob

    Whatever it is; the Lombards and Goth would certainly have been greater; but they were mostly killed during the Byzantine’s reconquests of Italy . They did all the fighting; don’t think many local Italian really cared between Goths or Byzantines

    The Byzantines did not recover Italy from the Lombards, but from the Ostrogoths. The Lombards came after the Byzantines and conquered most parts of the Byzantine Italy. It was the Franks, another Germanic people, who would end most of the Lombard rule in Italy, not the Byzantines.

    ReplyDelete
  31. True ; but the point remains - the flower (so to speak) of the Italian-Germanics was spilled during these wars of conquest and “reconquest” .
    Also a couple of Italian programs of gothic families . So this is why their impact is much reduced

    (Of course Franks were also originally Germanic; but by the ~800s they were heavily Rhenicized)

    ReplyDelete
  32. @Rob

    I respectfully disagree. I think the biggest influx of Germanic ancestry to Italy came with the Lombards rather than the Ostrogoths. The Ostrogothic rule in Italy lasted only several decades, not even 100 years, while the Lombard rule in Italy lasted for centuries and was replaced by Frankish rule (including the Frankish-protected papal realms) in Northern and Central Italy and lasted some more in Southern Italy (though in Southern Italy they had to compete with the Byzantines and then with the Normans). The Franks, when they came to Italy, were still all Germanic-speaking even in their western (modern French) territories, they never stopped speaking Germanic in their eastern (modern German) territories.

    ReplyDelete
  33. @Onur,

    Lombards, Northern Italy G25
    https://docs.google.com/spreadsheets/d/1x5grF3fM6w-3W9AS2DzZyeXsCyABTbYMtoQfLabs604/edit#gid=0

    Position of Northern Italy, Iron age Italy, Lombards in West eurasian PCA.
    https://chart-studio.plotly.com/~SamuelAndrews1017/125.embed

    You can hover over the samples to figure out who is who. I include Germans to show modern Germans are diverse and most cluster southwest of Lombards. Germany has not been well sampled so we can't saying definite about exactly how diverse and regional variation. But anyways.

    The Lombards and Saxons (who I include) cluster in Scandinavia quite clearly. Northeast of where any modern Northwest Europeans cluster. They were outside of modern Germanic variation.

    I include Central Italy to show, it isn't clear that Northern Italy pulls northeast towards Lombards. Lombards are really far northeast compared to Italy. I don't see a strong pull towards them in Northern Italy.

    ReplyDelete
  34. Consideirng there's E1b1b, J2a in Northern Italy there's no doubt they have ancestry from Central-South Italy. When, Italian_Calabria is included they score a signifcant amount and they also score 10-20% Lombard. So, you might be right Onur.

    ReplyDelete
  35. @Onur Dincer
    @Samuel Andrews

    I think it is hard to model the exact percentage of Germanic ancestry in North Italy because Germanic ancestry would be on a similar cline like Celtic and early Italic ancestry just on the northern end of this cline. Slavic ancestry on the otherside is easy to detect because of typical Balto-Slavic drift associated from it which can be easily differentiated in Global25 from Germanic and Celtic drift despite having very similar components and ratios of it.

    For example when i tried to model Lombardy Italians with the averages of a Greek-like, Germanic-like, Early Italic-like, Gaul-like and Near Eastern source i got this. But this is obviously an overfit underestimating the early Italic-like ancestry and overestimating Germanic/Gaulic-like ancestry.


    Target: Italian_Lombardy
    Distance: 0.8963% / 0.00896318
    41.0 Iberia_Northeast_Empuries2
    23.0 ITA_Prenestini_tribe_IA
    19.0 FRA_IA_central
    16.8 DEU_MA
    0.2 Levant_LBN_Roman

    When i remove the Greek-like source from Empuries and add Roman_Late_Antiguity i get a more logical model for Lombardy Italians, which indeed seems to point to North Italians sitting on a cline between Imperial/Late Antiguity Romans and early Italic tribes but for other Italians it is not working or creating an overfit which underestimates Italic-like ancestry.


    Target: Italian_Lombardy
    Distance: 1.6536% / 0.01653593
    55.8 ITA_Rome_Late_Antiquity
    38.6 ITA_Prenestini_tribe_IA
    5.6 DEU_MA


    Overall i would say North Italians have around 5-15% Germanic-like dna what would fit with the frequency of Germanic U106 and I1 clades (Germanic Y-dna is likely higher because not all Lombards/Franks would be just belong to U106 and I1)

    ReplyDelete
  36. @Samuel

    Thanks for the analyses. Indeed, on individual level you can find modern Northern Italian-like and even some more northern-shifted results among the post-Proto-Villanovan and pre-medieval Italian ancient genomes. But it is the regional and population averages that matter, especially in a hub like Roman Italy. In such regions outliers are not that uncommon. Davidski did a fine job in grouping the Italian ancient genomes based on some common genetic patterns and making different populations for different outlier groups, so his Italian ancient population averages are useful enough for our purposes. His "pure" ancient Lombard average, for instance, is quite close to the modern German average, only a little southern-shifted compared to it, expectedly. Of course more ancient genomes from different regions and time periods of Italy will allow us to better track the migration patterns and make better estimations. I do not try to make estimations of Germanic ancestry due to the limited data we have, just point to what they indicate currently. Those indications can change with more data, and then we can modify or change our theories and models.

    ReplyDelete
  37. @ Onur

    “I respectfully disagree. I think the biggest influx of Germanic ancestry to Italy came with the Lombards rather than the Ostrogoths”

    It seems you don’t really know what you’re saying or disagreeing about
    Can you present your study of the total number of Lombard settlers compared to Goths ; and if so how it changes the fact that both Goths and Lombards were real Germanic groups coming from Italy’s relative east

    “ The Franks, when they came to Italy, were still all Germanic-speaking even in their western (modern French) territories, they never stopped speaking Germanic in their eastern (modern German) territories.”

    That’s irrelevant because you’ve missed the point ; which was about their genetic constitution in the late 700s ; heavily westernised of an already west Germanic group

    ReplyDelete
  38. In fact; by the 700s Old Frankish was probably extinct. The language of carolingians was Romance

    “ The Ostrogothic rule in Italy lasted only several decades, not even 100 year”

    Actually; the Goths and their east Germanic kinsmen (Heruli; Rugi) had variously dwelt in Italy since the 400s
    So Onur, you’re wrong on all details

    ReplyDelete
  39. @Samuel Andrews @Onur Dincer

    We only have a handful of Italic samples, is it not possible that populations in pre-Roman northern Italy were even more northern-shifted? I know there was a substantial Celtic presence in the area historically.

    Calabria is very southern-shifted it would overestimate the amount of Germanic ancestry imo.

    But I do think (as you mentioned) that it is too early to be certain about this. About 10% Germanic admixture would make sense to me though.

    ReplyDelete
  40. @CrM,

    "It really makes me wonder why do populations with such similar genetic bases speak two unrelated languages. But since they're both a mix of local Caucasian forages and invading Anatolians, one of them could have adopted a non local language."

    I have wondered for a while now if any of the local CHG groups in the Caucasus developed any sort of agriculture or pastoralism on their own before Neolithic Anatolian groups arrived?

    ReplyDelete
  41. The main issue here is that all Italians from North down to Sicily started very western,with high proportions of ANF ancestry and they ended up East med shifted and and 'less' ANF admixed.The WHG admixture is a similar case especially in the south of the country.Immigrations from eastern meditterenean along with the germanic barbarian invasions lead Italy to become a less 'EEF-ANF' state.Sardinians didn't stayed the same as well,i am reading many users in anthroforas mentioning all the time that Sardinians are a pure EEF community,but that's not the case.They have influences too from both North and South Italy.Not to mention the North African admixture and if i have to take as an example the Etruscan samples we are talking probably for a very ancient admixture in the region of Italy.With a few words.North Italy seems to be a combination of 'IE(B.B Italo-Celts)' with influences both from Germanics but also from the South(Greco-Anatolians-Crypto Jews).Central Italy like Umbria,Tuscany,Lazio,Marche,Emilia Romana etc are between the North and the South.I think Central Italy is definitely more Italic with less influences from Germanics and Greco-Anatolians.And South Italy is mostly Eastern Med(Greco-Anatolians)+Italic+maybe some limited Germanic that it depends the region and individual.With exception Ligures(witch i am pretty sure,they were similar to Etruscans and Celts). i think Italy is a really sampled country..along with Spain and Portugal probably the most sampled states in EU.

    ReplyDelete
  42. @Rob

    I will not enter into polemics about who is "real Germanic" and who is not. I already pointed to the West Germanic-like genetics of many of the Lombard ancient genomes we have (they are not genetically homogeneous), I never claimed they were all Scandinavian-like, to the contrary. I also have not written anything contrary to the Franks having West Germanic-like genetics. I already know the pre-Ostrogothic Kingdom existence of various Germanic peoples in Italy, none of them changes the fact that the Lombards were the longest to rule large parts of Italy among the Germanic peoples with a tribal, not just regnal, existence in the territories of Italy.

    The Carolingians still all spoke the Germanic Frankish language as mother tongue when they had conquered all of Northern and Central Italy during the times of Charlemagne (the late 8th and early 9th centuries).

    ReplyDelete
  43. If you have questions or feature requests for the "Ancestry Calculator", just let me know. As a quick summary - it uses gradient descent optimization from the machine learning toolbox to generate a model. The AdaGrad and Adam optimizers are available. Their parameters are exposed to modify the optimization execution. I have a few starting setups available as preset configurations. I think the rest of the setup is somewhat typical, but let me know if something isn't clear or if you have questions.

    ReplyDelete
  44. re: Northern Italy; may I remind you guys of the distribution of blonde hair colour in Italy around 1860 (gathered by Ridolfo Livi)?

    https://justpaste.it/1s5tl

    Large parts of the Emilia-Romagna had less blonde hair than Tuscany or Umbria. Small parts of it (around Forli and Mirandola) had even less blonde hair than large parts of southern Italy. So don't tell me there is significant Germanic admixture everywhere in northern Italy.

    As for North Italian phonetics, check out these audio files of poems in the Cesenate dialect by Walter Galli:

    https://www.homolaicus.com/arte/cesena/storia/poeti/Galli/sommario.htm

    It's not obvious to me what's supposed to be Germanic-like about this. Sure, there is a voiceless and voiced th-sound like in English, but this is also found in Castilian Spanish or Greek for instance. And the vowels show quite some variation, but I'm not aware that this is typically Germanic. Superficially, the dialect reminds me of Rhaeto-Romance and Portuguese.

    ReplyDelete
  45. Judging from Raveane et al. 2019, north Italians do have more steppe (SBA) than central Italians, but at the same time they also have more EEN (EEF), at the expense of Anatolia BA (ABA):

    https://justpaste.it/5ltha

    ReplyDelete
  46. @Samuel Andrews

    The fact that they’re Scandinavian-like means that they’re in the range of modern Germanic variation. And let’s not forget the North Dutch.

    ReplyDelete
  47. @ Onur

    '' will not enter into polemics about who is "real Germanic" and who is not. I already pointed to the West Germanic-like genetics of many of the Lombard ancient genomes we have (they are not genetically homogeneous), I never claimed they were all Scandinavian-like, to the contrary. I also have not written anything contrary to the Franks having West Germanic-like genetics. ''


    This is not about Polemics, but facts. By the time Franks invaded Italy, they had been living in ex-Gaul for ~ 400 years, had become incorporated into Gallic post-Roman infruastructure, and intermarried with local Gallo-Romans.
    Thus, compared to the high point of Germanic demography in Italy with the Goths & Lombards, the Franks would have diminished this




    ''I already know the pre-Ostrogothic Kingdom existence of various Germanic peoples in Italy, none of them changes the fact that the Lombards were the longest to rule large parts of Italy among the Germanic peoples with a tribal, not just regnal, existence in the territories of Italy.''

    But that was not your original contention, which was I think the biggest influx of Germanic ancestry to Italy came with the Lombards
    So you're mis-communicating; & have not supported your claims with any data.




    ''The Carolingians still all spoke the Germanic Frankish language as mother tongue when they had conquered all of Northern and Central Italy during the times of Charlemagne (the late 8th and early 9th centuries).''

    Can you support , qualify & quantify this notion ? Which Carolingians, in what context, in what proportion, & until when

    ReplyDelete
  48. Anyhow, my point was in your discussion with Sam about 'genetic legacies', an understanding of demography (simply - numbers & context) must always be considered to ground one's genetic deductions. Something poorly done even by (certain) professionals

    ReplyDelete
  49. Those who claim that Northern Italy has remained genetically static since the Iron Age should first answer the question why modern Northern Italians have more eastern pull in terms of West Asian ancestry and more steppe ancestry than the Italian Iron Age genomes we have. Until then, all those speculations about thousands of years of static Northern Italian genetics sound like nationalist polemics.

    @Simon_W

    Large parts of the Emilia-Romagna had less blonde hair than Tuscany or Umbria. Small parts of it (around Forli and Mirandola) had even less blonde hair than large parts of southern Italy. So don't tell me there is significant Germanic admixture everywhere in northern Italy.

    Perfect logic! Emilia-Romagna is the southernmost region of Northern Italy and Tuscany and Umbria are among the northernmost regions of Central Italy and they are in fact all geographically adjacent to each other, those correlations with geography are not as strict as you want us to believe. Unsurprisingly, Tuscany has Northern Italian levels of average steppe admixture.

    As for North Italian phonetics, check out these audio files of poems in the Cesenate dialect by Walter Galli:

    https://www.homolaicus.com/arte/cesena/storia/poeti/Galli/sommario.htm

    It's not obvious to me what's supposed to be Germanic-like about this. Sure, there is a voiceless and voiced th-sound like in English, but this is also found in Castilian Spanish or Greek for instance. And the vowels show quite some variation, but I'm not aware that this is typically Germanic. Superficially, the dialect reminds me of Rhaeto-Romance and Portuguese.


    Not all Northern Italian languages have Germanic phonological influences. For instance, Emilian has them, but the closely-related Romagnol has not, Lombard and Ligurian have them too while Venetian does not. Do not tell me that sounds like ö and ü come from any Italic or Celtic substrate.

    ReplyDelete
  50. "Randwulf said...
    If you have questions or feature requests for the "Ancestry Calculator", just let me know..."

    Do you have a help file to explain the what the "L2 Penalty", "Initializer StdDev" and "Adam Beta1" etc. do and what values to try?

    ReplyDelete
    Replies
    1. These parameters allow access to tunable parts of the underlying formula for AdaGrad or Adam. There are many articles that describe the impact of setting these to something other than default values. Here us a link to one example, but you can find others: https://towardsdatascience.com/a-visual-explanation-of-gradient-descent-methods-momentum-adagrad-rmsprop-adam-f898b102325c

      Most of these articles show visually what is happening. It is important to know that in the context of G25, reaching your coordinates as the target value is the goal by producing a mixture from the model. The "Preset Configuration" dropdown will set some of these parameters to values other than defaults, so look at what happens when you run with each of the presets.

      Note that "Initializer StdDev" will introduce some randomness to your result by sampling your model if it is set to something other than zero, rather than using all of the model. It will produce variations each time you press the calculate button and is best used with a large model, like a full modern or ancient G25 spreadsheet. The preset configurations with "alts" in the name demonstrate this. I really like the Adam - Quick Alts configuration to demonstrate this with the Moderns model. Just keep pressing enter repeatedly after each calculation is done to see the variations.

      The L2 penalty works differently than an nMonte penalty. It introduces a penalty to high scoring results, favoring a secondary result, to try to fight overfitting, unless the high scoring result is so much better that it overcomes the penalty. In G25 terms, say you are German, but the unpenalized calculation can get a slightly closer distance by mixing Swedish and North Italian. The L2 penalty can penalize the incorrect Swedish plus North Italian mixture in favor if the more correct German result. That is a simplified explanation, but kind of the gist of the idea.

      The Adam Beta1 and Beta2 values adjust the momentum settings within the formula. Momentum is the idea that a step taken that closes the distance a good amount may be repeated to close it some more. The beta settings control how strongly that is implemented. Most articles I have read recommend keeping the default values, but you can try reducing them from .9 and .999 respectively to see the effect, such as trying values like .8 or .888, for example. It may take more steps to solve the riddle if a good mixture model.

      Gradient descent algorithms can get "stuck" in local minimums that aren't the best solutions. The most annoying one in this context is when it produces a result at a high distance saying you are 100% of a single reference. If this happens, try reducing the number of steps or changing the learning rate (the "size" of each step). Both AdaGrad and Adam will adapt the learning rate as they go, trying to initially take larger steps usually (using your submitted learning rate) and reducing as it gets closer to a solution.

      The graph in the result display shows you the progress it was making as it took steps towards a solution. If you are overfitting badly and you can hover over the graph bubbles to see how many steps were taken, enter that number of steps for the next execution, and see the model that is developed before the distance got squeezed so tightly overfitting.

      I hope that helps. Let me know any other questions.

      Delete
  51. @Rob

    This is not about Polemics, but facts. By the time Franks invaded Italy, they had been living in ex-Gaul for ~ 400 years, had become incorporated into Gallic post-Roman infruastructure, and intermarried with local Gallo-Romans.
    Thus, compared to the high point of Germanic demography in Italy with the Goths & Lombards, the Franks would have diminished this


    Not all of the Franks were living in ex-Gaul during those ~400 years though, most kept living in their core lands in Germania.

    But that was not your original contention, which was I think the biggest influx of Germanic ancestry to Italy came with the Lombards
    So you're mis-communicating; & have not supported your claims with any data.


    My statement of “the biggest influx of Germanic ancestry” was a guess, I did not push it, but it aligns well with the historical record and the genomic data.

    Can you support , qualify & quantify this notion ? Which Carolingians, in what context, in what proportion, & until when

    Charlemagne was certainly Germanic Frankish-speaking, this is a well-known historical fact. There are well enough books written about him, you can check them out. Also Charlemagne ruled his empire from Aachen, a Germanic-speaking city to this day (in what is today Germany).

    Anyhow, my point was in your discussion with Sam about 'genetic legacies', an understanding of demography (simply - numbers & context) must always be considered to ground one's genetic deductions. Something poorly done even by (certain) professionals

    I agree with you on that. But bear in mind that much of what we write here are guesses and I usually try to be explicit when I make guesses using clauses like “I think” and “I guess,” and I do not push them, but rather bring them forward for their open discussion while I could just keep them to myself.

    ReplyDelete
  52. @ Onur

    ''
    Charlemagne was certainly Germanic Frankish-speaking, this is a well-known historical fact. There are well enough books written about him, you can check them out. Also Charlemagne ruled his empire from Aachen, a Germanic-speaking city to this day (in what is today Germany).''

    Yes I know he was a Frank. That's wasn't what I was really asking :)

    ''But bear in mind that much of what we write here are guesses and I usually try to be explicit when I make guesses using clauses like “I think” and “I guess,” and I do not push them, but rather bring them forward for their open discussion while I could just keep them to myself.''

    Of course you should discuss them, indeed you went out of the way to make that point, on which I engaged .
    The estimated Gothic migration was 2-300,000. The Lombards have been deemed to have actually have been smaller, becuase of the demographic attrition of the Byzantine-Gothic war. They essentially walked into Italy
    Here are some discussions on the matter 1 2

    ReplyDelete
  53. That’s 200-300,000 total
    Estimated 40,000 fighting age men

    ReplyDelete
  54. @Onur,

    I'll concede modern Northern Italians shouldn't be seen as perfect reference for Roman Italian ancestry in Spain, considering we don't know what their origins are.

    But, my point stands, that for whatever reason modern Spain fits well as a mix between CeltIberians and Northern Italy.

    ReplyDelete
  55. David,

    Is there a way I can use my own computers to produce G25 std & scaled coords for members of our FTDNA S1194 project ?

    I previously (2018) got my own G25 data from you and have run it through the older download & run tools (R & nMonte) but now find the online PCA and G25 tools to be pretty good and very easy to use.

    Our computing resources are powerful enough (quad i7 proc - with stacks of RAM) plus have both Win PCs and Ubuntu servers.

    Am hoping to process each member of the S1194 project who is willing to participate.

    Restating, are there any tools we can run to convert the various autosomal data files (F/F, Ancestry, 23andme, etc: ) to produce the G25 data ?.

    Thanks DSM

    ReplyDelete
  56. @Onur Dincer

    Can we fully be sure that modern day North Italians have significant more steppe ancestry than ancient Early IA North Italians. I mean using Neolithic components like Yamnaya, Tur_Barcin and Geo_CHG may overestimate direct Yamnaya-like ancestry in modern day South Europe because of the extra layer of CHG-related + very complex IA/BA ancestry

    In the case of Italians we probably have several sources for Steppe ancestry (Bell Beakers, Balkan_BA/IA,Greeks, Celts, Germanics) what makes it hard to model the exact percentage of steppe ancestry out of them. And i think i am not capable to do that.

    Some ancient Italics seem to be almost 40-50% Bell Beaker derived and afaik (correct me if i am wrong) they are from Central Italy so IA North Italy could be even further north shifted.

    Target: ITA_Prenestini_tribe_IA
    Distance: 2.6590% / 0.02658964
    51.8 ITA_Remedello_BA
    47.8 Bell_Beaker_Rhine-Main
    0.4 SYR_Ebla_EMBA

    But this is of course before the big genetic shift towards the East Mediterranean happened and based on the similar or higher amounts of steppe ancestry among modern North Italians a more recent and northern source is definetly needed like you mentioned before.

    ReplyDelete
  57. PS

    Genoplot looks slick. But I could not get it to accept my G25 scaled & unscaled plots as a sample upload.

    It would take my scaled plots and promptly add them into the unscaled input box as well ?

    It is not intuitive and not especially easy to use in this area.

    They offer access to all the good admixture tools. Their prices are not prohibitive but, meh !.

    I sent them a blunt summary & wished them good luck but, am not sure I will go back. The existing tools we have access to seem to work well enough.

    ReplyDelete
  58. @Rob

    Of course you should discuss them, indeed you went out of the way to make that point, on which I engaged .
    The estimated Gothic migration was 2-300,000. The Lombards have been deemed to have actually have been smaller, becuase of the demographic attrition of the Byzantine-Gothic war. They essentially walked into Italy
    Here are some discussions on the matter 1 2

    That’s 200-300,000 total
    Estimated 40,000 fighting age men


    Thanks for the reference. Though I could not find similar estimates for the incoming Lombards in the book. But the book mentions the Lombard migration to Italy as a wholesale migration of the Lombards from Pannonia, which necessitates a very big migration.

    ReplyDelete
  59. @Samuel

    I'll concede modern Northern Italians shouldn't be seen as perfect reference for Roman Italian ancestry in Spain, considering we don't know what their origins are.

    But, my point stands, that for whatever reason modern Spain fits well as a mix between CeltIberians and Northern Italy.


    I would suggest you use the Iron Age mainland Italian samples rather than the modern Northern Italian samples in your analyses. Modern Northern Italians do not just have Germanic admixture, but they also have an eastern-shifted West Asian admixture, which modern Iberians and Iron Age mainland Italians by and large lack. In the case of modern Iberians and typical Iron Age mainland Italians the only West Asian ancestry we see is Barcin Neolithic-type, but modern Northern Italians have more eastern West Asian ancestry in addition to that.

    https://docs.google.com/spreadsheets/d/1a5byEejcptHST6X2L6Exb60uGXnyI6tZYI7HWG-EkkE/edit#gid=0

    ReplyDelete
  60. @CM

    But this is of course before the big genetic shift towards the East Mediterranean happened and based on the similar or higher amounts of steppe ancestry among modern North Italians a more recent and northern source is definetly needed like you mentioned before.

    Yes, like I pointed out, modern Northern Italians do not just have more steppe ancestry but also have more eastern West Asian ancestry than the Iron Age Italian samples we have, add to that the even more West Asian-shifted genetics of the imperial Roman Italian genomes we have, it becomes necessary to use the late ancient and medieval Germanic input to explain the current steppe levels of Northern and even Central Italy.

    ReplyDelete
  61. @dsjm
    "Is there a way I can use my own computers to produce G25 std & scaled coords for members of our FTDNA S1194 project ?"

    I wish davidski would tell us the outgroups and sanples used inside the blackbox that gives us 25 datapoints for each sample, so that they're are no blindspots. Unless it's already public.

    In any case, I would guess you could emulate what published papers do for their pca plots, use f or d stats to create 2d plots. You would need to extend that to 25d.

    ReplyDelete
  62. @Randwulf

    "......"

    Thanks for the info re the Ancestry Calculator. It was fun to play with. The data it produces looked good to me and the options as you outline, offer quite a bit of flexibility.

    Thanks

    DSM

    ReplyDelete
  63. @vAsiSTha

    Tks re comments on generating G25 co-ords.

    The GENOPLOT site look like they can do the G25 data points generation starting with an autosomal raw data file. This suggests it can be done.

    DSM

    ReplyDelete
  64. The Global25 is only available via the Eurogenes store.

    https://eurogenes.blogspot.com/2017/10/genetic-ancestry-online-store-to-be.html

    ReplyDelete
  65. Nice to see new tools. I hope their algorithms are not very different. It is very annoying when sometimes You have different results between Vahaduo and nMonte.

    ReplyDelete
    Replies
    1. The algorithms are different on all of these. You will get different results, but I would say they are in the same ballpark because they are all ways to do optimization and they are operating with the same data.

      Delete
  66. The support folks at GENOPLOT were responsive and helpful when I contacted them initially. I think the site looks good and will likely get better.
    @Randwulf I've been playing around with the Celtic Vs Germanic pca data using the "custom" setting in Ancestry Calculator. It seems to work well. Comparable results to those from Vahaduo. Any reason why this might not be a good idea? Got any tips for getting the most out of the calculator using this data?

    ReplyDelete
    Replies
    1. Using the Celtic vs Germanic PCA data should work great using the custom option the way you did. Right now, it expects exactly 25 coordinates. But, you should be able to try either optimizer and the presets, as well as tweaking the parameters just like with the G25.

      I thought a future modification I could make would be to allow the number of coordinates to vary for the custom as long as the model rows have the same number.

      Delete
  67. This just came out, https://www.biorxiv.org/content/10.1101/2020.08.10.241430v1 I am wondering if that westward migration is truly of Near Eastern origin or not.

    ReplyDelete
  68. Early Alpine human occupation backdates westward human migration in Late Glacial Europe

    Here we generate new genomic data from a human mandible uncovered at the Late Epigravettian site of Riparo Tagliente (Veneto, Italy), that we directly dated to 16,980-16,510 cal BP (2σ).
    MtDNA - U4’9; Y- I2; WHG

    https://www.biorxiv.org/content/10.1101/2020.08.10.241430v1

    ReplyDelete
  69. Looks like another very basal WHG forager has been found in Italy
    "Early Alpine human occupation backdates westward human migration in Late Glacial Europe"
    https://www.biorxiv.org/content/10.1101/2020.08.10.241430v1

    ReplyDelete
  70. Nice! Best confirmation that Western European WHG is from Italy. mtDNA U5b, Y DNA I2a.

    There's an older site in Italy, 19ky, with U5b2b, I've been saying if the DNA is tested he will turn out fully formed WHG.

    ReplyDelete
  71. Those idiots think WHG came to Italy from Balkan/Anatolia refugium? What? Why could it not have evolved in place. WHG ancestry already existed in Spain 20ky, it was clearlly already deep in Western Europe by 20ky.

    ReplyDelete
  72. Figure 3.

    Mesolithic West Europeans are much closer in PCA to Taglennte2 (16ky, Italy) than to Elmiron, confirming they are overwhelmingly of Italy Epigravitean origin. Confirming there was population replacement in Mesolithic transition in Western Europe, but the incoming population came from Italy not Near East.

    ReplyDelete
  73. This idea that the Villabruna cluster represents recent migrants from the Near East is very strange.

    It needs to be put to rest ASAP, because it's based on bad inferences from spotty sampling, just like the idea that the Yamnaya population formed as a result of a migration from Iran.

    At best, there was some gene flow from Anatolia into Southern Europe thousands of years before the Villabruna cluster formed.

    ReplyDelete
  74. Certainly lays to rest the singular, stubborn claims that VB-like introgression only began 12,000 BC.

    However their suggestion that RT lineages are “basal” to all subsequent HGs isn’t quite correct -
    so far 4 Italian HGs are I2a2 ; which can’t be ancestral to north -Central European I2a1 ; whilst the R1b in VB must be relatively recent arrival from elsewhere
    Nor is its mtdna U4-9 basal to the prevalent U5b in WHGs

    ReplyDelete
  75. “This idea that the Villabruna cluster represents recent migrants from the Near East is very strange”

    A very forced idea without much backing

    ReplyDelete
  76. @Samuel Andrews

    "Best confirmation that Western European WHG is from Italy."

    You are clearly written the opposite, in Italy Villabruna cluster (Epigravette culture) came, it replaced the Gravette cluster.

    "Those idiots think WHG came to Italy"
    These people are smarter than you.

    WHG came to Italy from the East. Exactly where it came from the East, they have no data.
    You're wrong, you look at the dates of those tested and see them as dates of origin.

    "mtDNA U5b, Y DNA I2a."
    MtDNA - U4’9; Y- I2

    "WHG ancestry already existed in Spain 20ky, it was clearlly already deep in Western Europe by 20ky."

    You are absolutely wrong. Do not imagine, there were no ancestors of WHG in Spain from the word at all. Proved he wasn't there. Moreover, there was no pure WHG in Spain until the end of Mesolithic. ElMiron is not WHG at all, even LaBrana is not pure WHG yet.

    "Mesolithic West Europeans are much closer in PCA to Taglennte2 (16ky, Italy) than to Elmiron, confirming they are overwhelmingly of Italy Epigravitean origin. Confirming there was population replacement in Mesolithic transition in Western Europe, but the incoming population came from Italy"

    Bullshit, that doesn't prove anything, because no other places in the Epigravette culture have been tested. You confuse the tested area with the appearance of people. Not from Italy, of course.

    ReplyDelete
  77. Well, they are dragging to Anatolia in vain, otherwise they are close to the truth.

    quotes
    "from Early to Late Epigravettian material culture in a vast area ranging from the Rhone river to the Southern Russian plain35. In this broad context, the individual found at Riparo Tagliente denotes thepresence in the region of human groups from Eastern Europe/Anatolia, therefore backdating by at least 3ka the occurrence of genetic components previously reported for the later Villabruna cluster (~14ka ago). At the same time, even earlier migrations into Southern Europe might be envisaged to explain the presence of Villabruna- or Tagliente2-related genetic components in the ~18.7 ka-oldsample found at El Mirón, Spain"

    "Our results rather show that population movements were already in place during the cold phase immediately following the LGM peak. At this stage, Italy, the Balkans, and Eastern Europe/Western Asia were already connected into the same network of potential LGM refugia, and exchanged both genes and cultural information."

    Actually the Epigravette culture goes so far to the south, in the Balkans it's rare and more hypothetical. It is in the Northern Balkans and the Carpathians.

    "The latter scenario stems from interpreting the traces of Villabruna/Tagliente2 genetic components recorded at El Mirón (~19ka ago4,34) as the result of an early phase of this westward expansion, rather than as a proliferation from Western European refugia."

    "Tagliente2, therefore, suggests that cumulative cultural change observed in Southern Europe from the end of LGM to the end of the Younger Dryas(~11.7 ka ago) was at least in part triggered by gene flow from eastern refugia into Northeastern Italy and that this process, in its early stage, was independent of warming events, and contributed to the gradual replacement of pre-LGM ancestry across the Italian peninsula."

    ReplyDelete
  78. A Balkans origin for the Villabruna clusters makes a lot more sense to me than it developing in Italy as Sam suggested. Why? Because the Apulian Gravettian samples Paglicci133 (34-31kya), Ostuni 2 (29-28kya), Paglicci108 (28-27kya), and Ostuni1 (27kya) are all Vestonice cluster individuals. So unless Vestonice-type people hopped over Villabruna-type people on their way to Apulia, it makes more sense to assume Villabruna cluster came to Italy later. They presumably totally replaced Vestonice type people in Italy by the time of Villabruna proper, Grotta Continenza, and maybe this new Alpine sample. I suppose it's possible Villabruna ancestors were living south of Apulia before they took over Europe, but a Balkan origin still makes more sense to me. For one thing, Dzudzuana shares ancestry with the Villabruna cluster. It's easy to imagine a Villabruna cluster sister group migrating from the Balkans into the Anatolia where they would go on to mix with Basal Eurasians. There's also the IJ connection.

    And frankly, I don't see what's so crazy about an ultimate Anatolian origin either. Maybe Basal Eurasian ancestry postdates Villabruna-type ancestry in Anatolia and the Caucasus. How can we know? Get some Balkan and Anatolian DNA from 30 to 40kya (ignoring the weird Romanian samples) and I think any lingering mystery about WHG origins will solve itself.

    ReplyDelete
  79. @Archi
    What is "Epigravette culture"?

    ReplyDelete
  80. If these clowns do not solve the Dzudzuana mystery don't expect to found out what exactly happened with this basal ancestry to Villabruna and later WHG.There is definitely some eastern admixture into EU of 'basal' origins/roots..Btw what Archi wrote is pretty much correct.Many people passing El Miron,La Brana etc as WHG..witch is completely wrong.i dont want to be argumentative but Cro-Magnons were not even exactly human beings from anthropological kind of view.Mesolithic Europeans were definitely more Gracile,less robust and more 'human' lookable.

    ReplyDelete
  81. A problem again lies with traditional terminology. Too many areas are called “epigravettian” when they shouldn’t be; implying a unity which did not exist until the later phases (after ~ 20,000) when wide areas of Europe became reconnected
    Eg The Post ice age industry at Kostenki & other Don sites has nothing to do with the preceding Gravettian, so should be called something different. On the other hand in Italy and parts of the Balkans there is a continuous development of Gravettian traditions into the early epigravettian

    ReplyDelete
  82. @ Michalis

    “ They presumably totally replaced Vestonice type people in Italy by the time of Villabruna proper, Grotta Continenza, and maybe this new Alpine samples”

    There is some cultural continuity ; not to mention mtDNA U2-8 in Italian samples


    “ And frankly, I don't see what's so crazy about an ultimate Anatolian origin either.”

    The Pinarbasi sample - with its YDNA C1 and mtdna U8 has made this less likely
    And generally; there is a paucity of paleolithic sites on Anatolia . Maybe due to poor research but maybe reality

    ReplyDelete
  83. There is no continuous development from the Gravette culture to the Epigravette culture, it's just an archaeological fact. East Gravette is very different from West Gravette, it's different cultural entities. Gravette is defined as a multitude of separate cultures that are united by a common tradition.

    "And generally; there is a paucity of paleolithic sites on Anatolia . Maybe due to poor research but maybe reality"

    Anatolia in Upper Paleolithic was an area uninhabitable for human life.

    "The Pinarbasi sample - with its YDNA C1 and mtdna U8 has made this less likely"
    Pinarbasi is a migrant from Europe, YDNA C1a2 mtdna K2b.
    During the days of Aurignac and especially Gravette, the vector of migration was from Europe from West to East, it reached the Kapova Cave and the Levantine Aurignac. In other times the vector of migrations was from East to West, which to the beginning of Mesolithic became distinct from the Trans-Urals.

    ReplyDelete
  84. @ Archie

    “ There is no continuous development from the Gravette culture to the Epigravette culture, it's just an archaeological fact. ”

    Not quite . It depends where you’re talking about and which phase
    Eg from 30000 BP Italy gravettian to 22000 epigravettian is continuous; but a modest change occurs ; 18000
    But in Russia the change occurs 25000; with absolute no continuity from before

    ReplyDelete
  85. “ East Gravette is very different from West Gravette, it's different cultural entities”

    It’s actually more complex than that. There’s early, middle and late Gravettian phases ; and a number of regional facies; and that’s not even getting into the epiGravettian
    The eastern Gravettian mostly develops in the Late Gravettian (29,000-> ) contains the Kostenki-Avdeevo circle; and other sites which are less close but still broadly related including in Eastern Europe and Central Europe (eg Willendorf)
    Some of the key features are kostenki knives; shouldered points and Venus figurines . But they are seen all the way to Italy and even Portugal

    Now- come 25000 , these disappear from Russia but continue in Italy and Dalmatia until ~19/18000 and even later. In Ukraine and Russia ; 2 new industries emerge with mysterious origins

    ReplyDelete
  86. Isn't AnatolianHG ancestry basically dzudzuana? That's what laziridis says. Pinarbasi = drifted dzudzuana with jo additional ancestry required is his model.

    And what real difference does it make if ancestry of some sample is from Balkans or Italy 17000 years back? They're hardly 1000-2000km away from each other, not 10000km away

    ReplyDelete
  87. @Michalis Moriopoulos

    "Maybe Basal Eurasian ancestry postdates Villabruna-type ancestry in Anatolia and the Caucasus. How can we know? "

    Could you please elaborate on that? I thought WHG did not have BE ancestry

    ReplyDelete
  88. @vAsiSTha

    "Isn't AnatolianHG ancestry basically dzudzuana? That's what laziridis says. Pinarbasi = drifted dzudzuana with jo additional ancestry required is his model."

    Where did Lazaridis say that? I don't think Pinarbasi=Dudzuana albeit Anatolian farmers were consistent with being a clade with Dudzuana in his paper IIRC.

    ReplyDelete
  89. @Michalis Moriopoulos,

    You commented on last thread, saying you dis agree on my opinon WHG is from Italy.

    You make good points. We have many DNA samples from Gravitean Italy dating ~28ky and they are clearly not the ancestors of later WHG people who lived in Italy at least by 17ky. This is undisputable.

    I agree, WHG itself did not originate in Italy. But, I think that the Western European variety of WHG did originate in Italy and that it expanded out of Italy across Western Europe "replacing" the Magdalonian-derived hunter gatherers.

    A unique view I have, that I think others might agree with, is there are two different types of WHG. A Western type and Easter type. The Western type had mtDNA U5b, the Eastern type U5a. The Western type had blue eyes the eastern type had brown eyes.

    I think WHG itself probably originated in Southeast Europe in Upper Paleolithic, but the Western variety (mtDNA U5b) originated in Italy from whence it expanded into France, Germany, Britain, Spain.

    But I think the WHG ancestry in Serbia, Ukraine, Latvia, Russia hunter gatherers is mainly from Balkans not Italy. But WHG ancestry in France, Germany, Britain comes from Italy.

    Which, you would probably agree with. The consensus seems to be Italian WHG replavced Magdalonian HGs but that WHG itself did not originate in Italy.

    ReplyDelete
  90. @Michalis Moriopoulos,

    Looking at mtDNA/Y DNA, the Near East link which geneticists make for WHG, seems unlikely.

    Almost all WHG carried mtDNA U5, if you ignore ANE mtDNA in Eastern Europe. Yet, we see no mtDNA U5 in Near East. But, we do see pre-U5 in Gravitean.

    There is Y DNA J in Near East, but there is Y DNA I* in Gravitean. This places WHG's mtDNA/Y DNA in Europe by 29,000 years ago even if Gravitean isn't their direct ancestor.

    They are smart to point out there is something Near East and WHg share which Gravitean lacks, but this doesn't mean WHG is from Near East. But it could support Southeast European origin of WHg.

    ReplyDelete
  91. "Where did Lazaridis say that? I don't think Pinarbasi=Dudzuana albeit Anatolian farmers were consistent with being a clade with Dudzuana in his paper IIRC."

    Yeah he does say barcin_n not pinarbasi, don't think pinarbasi was published at that point in time but I maybe wrong.

    But i do not find pinarbasi to be different from anatolia_n. In qpAdm, pinarbasi in right pops and anatolia_n as source has given me problems, which says that they are quite similar.

    ReplyDelete
  92. @vAsiSTha

    Yeah Barcin_N seems to be a clade with Pinarbasi (or closely related-maybe Barcin_N has some CHG-like ancestry) though other farmers are apparently not?

    In that Pinarbasi paper they modeled it as a mix of WHG-related and Natufian-related lineages which is a bit different than the model in the Lazaridis paper.

    ReplyDelete
  93. @gamerz
    Yeah because they didn't have dzudzuana sample so they had to do their best with what they had I guess.

    ReplyDelete
  94. Hi everyone!

    New tool - PCA 3D Viewer.

    Global 25 version:

    https://vahaduo.github.io/3d/g25

    West Eurasia version:

    https://vahaduo.github.io/3d/we


    Usage:

    Dots - ancients, circles - moderns.

    Click X, Y, Z or COLOR tab and then click one of the PCx buttons to switch dimensions.

    Click already active X, Y, Z or COLOR tab to temporarily reverse selected dimension. It will be restored to a default state when any of the dimensions will be switched to another one.

    ADD CUSTOM POINTS - self-explanatory. Points will be added as "+". IMPORTANT - G25 version takes NON-SCALED coordinates. This will be true for any new tool dedicated to G25 and coordinates will be scaled automatically when needed or desired.

    "Type parts of names." + TAG button - type parts of names to tag certain samples (try for example "KK1 Afon Pinar"). Search is Case Sensitive. Points will be redrawn as "x".

    Next row - Labels and Annotations. Click the right button to cycle trough:

    CLEAR LABELS + LABELS:AUTO
    CLEAR ANNOTATIONS + ANNOTATIONS:CLICK
    CLEAR ANNOTATIONS + ANNOTATIONS:AUTO
    CLEAR ALL + LABELS AND ANNOTATION OFF
    CLEAR LABELS + LABELS:CLICK

    CLICK - click to add/delete labels/annotations.
    AUTO - same as CLICK plus labels/annotations will be automatically added to newly plotted or tagged samples. Unfortunately adding new labels and annotations becomes very slow when there is too much of them, so there is a limit for the AUTO setting - 250 labels or 20 annotations at once.

    Annotations are editable. They can be dragged to another place and text can be changed. Text can be also wrapped into an HTML SPAN element and some styles can be used, like "font-size" or "color". BR element (new line) works too.

    HIGHLIGHT CLICK/OFF/HOVER - highlight all samples that belong to a single population. Set to HOVER to ignore clicks. Set to OFF to disable this feature and to remove highlight triggered by hover (highlights triggered by clicks will stay until they will be cleared or removed by a click). Click white dot to cycle trough available highlight colors.

    Plotly buttons:

    Default download is set to 1600x1200px PNG.

    Custom Plotly buttons:
    "Toggle projection: orthographic / perspective" - self-explanatory.
    "Toggle background color" - cycles trough dark grey, black, white and light grey. Text color and white highlight will be switched to black when background will be set to white or light grey.
    "Toggle color scheme" - cycles trough several gradients.
    "Reverse color scheme" - reverses all gradients permanently.
    "Download plot as png (custom size)" - default size is the size of the currently displayed plot.

    ReplyDelete
  95. I put up a new blog post about the 3D PCA viewers here...

    https://eurogenes.blogspot.com/2020/08/awesome-new-toys-from-vahaduo.html

    ReplyDelete
  96. Dzudzuana is mentioned often in this thread. Question: are there G25 components available somewhere for Dzudzuana? Did I miss them by another name in the eurogenes DB? Appreciated.....

    ReplyDelete
  97. The Dzudzuana manuscript was never published and the data never released.

    ReplyDelete
  98. Boo, hiss to the tantalizing paper followed by no data. Maybe I'll try a private request for these data. Anyhow it makes sense to me that there would be an epi-paleozoic bi-directional people/gene flow from the Transcaucasus area, through the maze-like valleys of the Karasu-Aras Mountains into the Eastern Taurus Mountains and eventually to the Konya area of Anatolia (as described in the Dzudzuana paper), and then expanding, possibly by sea to the Southern Levant, seeing as there appears to be at least one SNP track which moves a J-M267 sub-tree back and forth along such a pathway (this would explain Natufian genes on the eastern end, etc.), all the way from Mentesh Tepe to Peqi'in. Kura-Araxes seems to exemplify one phase of this pattern, but it looks like the Aras-Taurus highway may have been flowing from a long time before that. Perhaps in its earlier days the complex was driven by obsidian from the eastern end as a raw material before metals came in, and then later perhaps the flow was pushed westward by migrations from folks coming over the Caucasus or north from the Zagros. It would be awfully nice to have some G25 matching results to support these notions. Pleasant and productive researching to all......

    ReplyDelete
  99. @Davidski,

    Can you upload these Khasi samples from Meghalaya from this study into G25 datasheet please?:

    https://academic.oup.com/mbe/article/28/2/1013/1220271

    And also can you upload these Indonesian samples (Sumatra, Mentawai, Nias, Sulawesi, Sumba, Flores, Lembata, Pantar and Alor) from these studies into your G25 spreadsheet please?

    https://www.nature.com/articles/jhg2012154
    https://www.genomicus.com/files/publications/Kusuma_2016_QuatInt_v416_p243.pdf
    https://www.nature.com/articles/ejhg201788?WT.feed_name=subjects_genetics

    Thank you very much. I would greatly appreciated it. I want to see how much West Eurasian ancestry do the Khasis and these Indonesians have?

    Thank you very much.

    ReplyDelete

Read the rules before posting.

Comments by people with the nick "Unknown" are no longer allowed.

See also...


New rules for comments

Banned commentators list