What genetic tool can identify isolated and early-divided populations?

[smd555]

Let's say a population is a separate genetic branch, separated at some stage from others. For example, a population related to EHG (or WHG), but separated from them earlier. Routine analysis of an admixture using standard components like WHG, EEF and Steppe would seem to present it as a mixture of these components - although it is not really a mixture, but simply an early branch. What tool will show that it is an early separated? Admixture with many modern components, or qpAdm, or PCA, or something else?

[vbnetkhio]

Let's say a population is a separate genetic branch, separated at some stage from others. For example, a population related to EHG (or WHG), but separated from them earlier. Routine analysis of an admixture using standard components like WHG, EEF and Steppe would seem to present it as a mixture of these components - although it is not really a mixture, but simply an early branch. What tool will show that it is an early separated? Admixture with many modern components, or qpAdm, or PCA, or something else?

in recent studies "qpGraph" is used for that purpose.

[smd555]

in recent studies "qpGraph" is used for that purpose.

Have you used it? Can you rate its effectiveness?

[vbnetkhio]

Have you used it? Can you rate its effectiveness?

I attemped to use it, but it's slow and complicated compared even to qpadm and admixture.
But it's effective exactly for that issue which you are asking about.

[smd555]

I attemped to use it, but it's slow and complicated compared even to qpadm and admixture.
But it's effective exactly for that issue which you are asking about.

I am interested in samples of Latvian bronze (Kivutkalns) and samples of Estonian bronze (tarand graves) close to them. I have a suspicion that they do not go back to the previous known Baltic populations (Narva, Corded Ware, Pit-Comb Ceramics cultures), but may be descendants of a separate branch of hunters that separated early from the rest.
If this is so, then this radically changes the picture of the genetic origin of the Slavs, Balts, and part of the Finno-Ugric peoples.

[vbnetkhio]

I am interested in samples of Latvian bronze (Kivutkalns) and samples of Estonian bronze (tarand graves) close to them. I have a suspicion that they do not go back to the previous known Baltic populations (Narva, Corded Ware, Pit-Comb Ceramics cultures), but may be descendants of a separate branch of hunters that separated early from the rest.
If this is so, then this radically changes the picture of the genetic origin of the Slavs, Balts, and part of the Finno-Ugric peoples.

that's certainly true, you don't need qpgraph to see that. I guess you saw that new study which claims Baltic BA originates from northern Bulgaria?

[smd555]

that's certainly true, you don't need qpgraph to see that. I guess you saw that new study which claims Baltic BA originates from northern Bulgaria?

Do you mean this?: https://www.theapricity.com/forum/showthread.php?359286-50-HG-relict-population-in-Bronze-Age-Hungary!

[vbnetkhio]

Do you mean this?: https://www.theapricity.com/forum/showthread.php?359286-50-HG-relict-population-in-Bronze-Age-Hungary!

yes

[smd555]

yes

I must admit that this hypothesis arose much earlier, when using conventional calculators. I will try to explain in several posts. To do this, I need to make some introduction.
I will use the G25 calculator here and try to figure out how long (for a given calculator) the genetic population can change significantly. This is to the question of how correct it is to compare ancient and very ancient populations with modern ones.
So, how can we track how much a population might change over time? This is not an easy task, because for this you need to be sure that the population is closed, that no others have mixed in with the passage of time, and changes are only mutations. I thought that the most successful example would be the Native Americans - namely the Amerindians (those who are the first and most numerous branch, NOT Na-Dene). Indeed, they most likely originated from a small population that began to settle, according to various estimates, from 14 to 17 thousand years ago. At the same time, as they settled, groups of Native Americans began to isolate themselves from one another - evidence of early isolation is a large number of language families in the territory of both Americas. So, let's try to compare the samples of Native Americans presented at G25 with each other, also with ancient samples. Take the Mayans, for example this sample:


Mayan:HGDP00856,0.062603,-0.293488,0.112005,0.091732,-0.102788,-0.012271,-0.269557,-0.322833,-0.018203,-0.015308,-0.002111,-0.002098,0,0.014588,-0.012486,0.00305,0.005476,-0.001014,0.001131,-5e-04,0.000873,0.009645,-0.00037,-0.001325,-0.001197

And let's see the distances:

https://i.ibb.co/kgbQkSW/2022-02-05-220322.png (https://ibb.co/SXYRW39)

As you can see, the greatest distances (if we take the Amerindians) to some samples of the Surui people are about 0.11. Next come the Na-Dene Indians - Amerindian North and Chipewyan, with whom they had common ancestors earlier than 14 thousand years ago (and most likely much earlier). Of interest are two samples of the Cree Indians (Alonkin family), which turned out to be very different from all others. But on the single in modern average they show the following picture:

https://i.ibb.co/WnJr9Wt/2022-02-05-221059.png (https://ibb.co/kKCLng1)

So most likely they are simply mestizos, and we do not take them into account. Now let's take averages for comparison - the same Maya, and see what kind of spread is no longer at the individual level, but at the level of the average populations:


Mayan,0.056749,-0.2949387,0.1107659,0.0902554,-0.1048547,-0.0156974,-0.2758356,-0.3286347,-0.0140829,-0.0160369,-0.0017167,-0.0010061,0.001083,0.0199947,-0.008783,0.0037504,0.0079163,-0.0009773,-0.0013466,-0.0019474,-0.0026024,0.0056176,0.0010037,3.44e-05,0.0003594

https://i.ibb.co/dj462Nm/2022-02-05-221419.png (https://imgbb.com/)

As you can see, the greatest distances also go to Surui - about 0.09. Then they go again Na-dene - more than 0.13. With other Amerindians, the Maya generally turn out to be quite close. Even with the same Bolivians and Quechua, although the Mayan ancestors were separated from their ancestors a very long time ago. And now let's see how close the average Maya will be to the ancient samples:

https://i.ibb.co/vX30b9J/2022-02-06-112925.png (https://ibb.co/xF1xyZM)

As you can see, they are quite close, both to samples from Brazil ten thousand years ago, and to other samples from both Americas, from the USA to Tierra del Fuego (Yamana). So, even if we assume that at the beginning of the settlement from Beringia, the Native Americans initially had no internal dispersion and they were a completely homogeneous mass (although this has not yet been proven), and also if we take the minimum time for the start of their wide migration of 14 thousand years (although according to some data, this was even earlier), then for 14 thousand years the maximum dispersion in G25 at the individual level is up to 0.11, at the intergroup level - up to 0.09. More than 0.12 are already Na-dene. And this is the maximum dispersion, given the large genetic drift in some groups, and mostly these are peoples close to the Surui, and in general these distances are much closer, including to ancient samples.

[smd555]

And now let's move on to the genetics of the Balts and Slavs. I will also use the G25 calculator. The main idea is that I’m still paste European and partially Asian samples of the times from the final Paleolithic to the Eneolithic into the source, and I’ll also add Yamnaya culture, and try to model the genetics of modern peoples with the help of Neolithic farmers, hunters and Yamnaya. Take, for example, the Lithuanians:


Lithuanian_PA,0.1337421,0.1241485,0.0796195,0.0746 13,0.0421616,0.0291092,0.0087244,0.0134128,-0.0037324,-0.0263102,-0.0035118,-0.0091044,0.0186756,0.0266472,-0.0118755,0.0034972,0.0042865,0.0010292,0.0039282, 0.0012351,-0.0060519,-0.0032921,0.0093979,-0.006552,-0.0002845
Lithuanian_PZ,0.1356773,0.1255195,0.0883971,0.0871 131,0.044562,0.035447,0.0106459,0.0156685,-0.0030065,-0.0347524,-0.0067879,-0.0129784,0.0230573,0.0315155,-0.0071118,-0.0032219,-0.0034813,-0.0016344,0.000817,0.0034265,-0.0023459,-0.0059477,0.0084426,-0.0090132,0.0005148
Lithuanian_RA,0.135108,0.1240977,0.083306,0.080523 9,0.0431158,0.0294229,0.0123146,0.015207,-0.0044994,-0.0323469,-0.0030203,-0.012469,0.0240532,0.0300018,-0.0152957,0.0017237,0.0070014,-0.0013936,0.00494,0.0030138,-0.0053282,-0.0060342,0.010057,-0.0066392,-0.0003714
Lithuanian_SZ,0.1347664,0.1258242,0.0861344,0.0877 268,0.0448392,0.0315424,0.0128081,0.0185529,-0.0019633,-0.0353537,-0.0039137,-0.0127386,0.0235031,0.0296026,-0.0123914,0.0018563,0.005111,-0.0012796,0.0020866,0.0021511,-0.0029574,-0.0067636,0.0070251,-0.0104229,0.0029337
Lithuanian_VA,0.1344249,0.1257226,0.0824762,0.0750 975,0.0433311,0.027889,0.0099878,0.014884,-0.0020454,-0.0298685,-0.0029553,-0.0138328,0.0200394,0.0280062,-0.0110613,0.0029435,0.0061411,-0.0005702,0.0023129,0.0061155,-0.003731,-0.00481,0.0072346,-0.0078805,-0.0008622
Lithuanian_VZ,0.1372991,0.1261795,0.0850879,0.0823 65,0.0444314,0.0326652,0.0104872,0.0176242,-0.0036048,-0.0315269,-0.0036131,-0.0097039,0.0199762,0.0301394,-0.0104505,-0.0045578,-0.0006029,0.0012828,0.0032525,0.0013445,-0.0043829,-0.0056728,0.00835,-0.005874,-0.001317

Let's try to simulate them:

https://i.ibb.co/yyPNDb1/2022-02-05-230905.png (https://ibb.co/LtR0M2W)

As you can see, the percentage of Yamnaya is quite high - about a third - plus a high percentage of Sredniy Stog. But the total distance of the samples to the proposed mixtures is quite large - about 0.05. And now let's add the Bronze Age samples from Kivutkalns:


Baltic_LVA_BA:Kivutkalns153,0.137726,0.123895,0.11 0873,0.10336,0.041238,0.037929,0.012456,0.013153,-0.010022,-0.051391,0.005521,-0.019333,0.034192,0.046379,-0.013165,-0.011535,-0.00678,-0.005954,-0.007542,0.001126,0.010731,-0.002102,0.003081,-0.017954,-0.010059
Baltic_LVA_BA:Kivutkalns19,0.130897,0.122879,0.108 611,0.110467,0.052933,0.041276,0.015746,0.016384,-0.001636,-0.049204,-0.001137,-0.016785,0.037165,0.037296,-0.017236,0.005834,0.004172,0.007855,0.008045,0.009 505,0.000749,-0.00643,0.019473,-0.020123,0.006586
Baltic_LVA_BA:Kivutkalns194,0.118376,0.13405,0.092 017,0.10756,0.034776,0.038208,0.011281,0.010384,0. 022089,-0.038087,0.004222,-0.007493,0.024232,0.040186,-0.020901,0.009016,0.010561,-0.006841,-0.002137,0.027513,-0.017594,-0.005441,0.013804,-0.022413,0.011735
Baltic_LVA_BA:Kivutkalns207,0.129758,0.12491,0.110 119,0.114666,0.048932,0.047133,0.016686,0.019384,-0.003477,-0.058316,-0.00341,-0.025777,0.036571,0.052709,-0.014658,-0.009546,-0.00678,-0.003294,-0.006913,0.003252,-0.005366,-0.006554,0.008504,-0.025184,0.001796
Baltic_LVA_BA:Kivutkalns209,0.124067,0.117801,0.08 9,0.120157,0.046778,0.041555,0.012926,0.028383,-0.002045,-0.052302,-0.00747,-0.021431,0.037611,0.046104,-0.016422,0.004906,0.0103,-0.008235,-0.002137,0.00963,-0.004492,-0.011252,0.014666,-0.02181,0.002515
Baltic_LVA_BA:Kivutkalns215,0.134311,0.133034,0.09 956,0.108529,0.051702,0.04016,0.013396,0.021461,0. 003477,-0.050844,-0.005034,-0.019633,0.029732,0.039773,-0.012351,-0.003182,-0.015255,0.003294,-0.000754,0.012756,0.003494,-0.005317,-0.000616,-0.020003,0.007664
Baltic_LVA_BA:Kivutkalns222,0.130897,0.123895,0.10 1068,0.120157,0.057549,0.044344,0.020211,0.022384,-0.005522,-0.045923,0.000162,-0.024578,0.031516,0.040736,-0.013979,0.011403,0.011995,-0.007221,0.004022,0.008629,-0.009234,-0.000742,0.007272,-0.0194,0
Baltic_LVA_BA:Kivutkalns25,0.135449,0.125926,0.095 412,0.112405,0.056318,0.033746,0.013396,0.021691,-0.005522,-0.048839,0.001299,-0.016485,0.029732,0.043214,-0.021444,0.009016,0.01695,0.00152,-0.004274,5e-04,-0.000125,-0.005193,0.008011,-0.019521,-0.00012
Baltic_LVA_BA:Kivutkalns42,0.133173,0.135065,0.094 28,0.102391,0.045239,0.034861,0.014806,0.016153,-0.002454,-0.04483,0.002923,-0.018733,0.03553,0.043764,-0.021308,-0.00411,-0.005998,0.005828,0.008799,-0.002876,-0.000624,-0.00371,0.008874,-0.034945,0.002754

What we see:

https://i.ibb.co/XDhvJBt/2022-02-05-231641.png (https://ibb.co/sC0M3Sq)

The picture has changed markedly. Kivutkalns began to occupy the lion's share of the gene pool, and the percentage of Yamnaya decreased to about 10%. However, the distance has decreased significantly. That is, Lithuanians with the help of Kivutkalns are much easier to introduce themselves. So maybe Kivutkalns itself is a mixture of Yamnaya, hunters and farmers? Let's get a look:

https://i.ibb.co/NxKr6y3/2022-02-05-232441.png (https://ibb.co/b1XFgK3)

The distances of Kivutkalns samples to the proposed mixtures are 0.07-0.09, large. This is like the maximum variance between the Native Amerikan groups given in my previous post. And the chronological gap between Kavutkalns and the Yamnaya, late hunters and farmers, is only about 2000 years. What does it say? But the fact that Kivutkalns is a completely separate population, whose ancestors have long since separated from all the others. Personally, my opinion is that this is a population of hunters, most likely without significant admixtures of Yamnaya and farmers, something like a separate EHG cluster. Their ancestors split off from other EHGs somewhere else in the Late Paleolithic. That is, the standard EHG (Karelia, Samara, Yuzhnyi Oleniy Ostrov, Pit-Comb Ware culture, etc.) are much closer to each other than to the Kivutkalns.
Who is close to the people of Kivutkalns? Only the Estonian Bronze Age (tarand grave culture) and some later populations.
In which populations does the Kivutkalns component play a significant role? Mostly Balts, Estonians, Slavs. By the way, Erzya with Moksha also:

https://i.ibb.co/Fgr06pM/2022-02-06-115658.png (https://ibb.co/nzGnrpv)

[vbnetkhio]

And now let's move on to the genetics of the Balts and Slavs. I will also use the G25 calculator. The main idea is that I’m still paste European and partially Asian samples of the times from the final Paleolithic to the Eneolithic into the source, and I’ll also add Yamnaya culture, and try to model the genetics of modern peoples with the help of Neolithic farmers, hunters and Yamnaya. Take, for example, the Lithuanians:



Let's try to simulate them:

https://i.ibb.co/yyPNDb1/2022-02-05-230905.png (https://ibb.co/LtR0M2W)

As you can see, the percentage of Yamnaya is quite high - about a third - plus a high percentage of Sredniy Stog. But the total distance of the samples to the proposed mixtures is quite large - about 0.05. And now let's add the Bronze Age samples from Kivutkalns:



What we see:

https://i.ibb.co/XDhvJBt/2022-02-05-231641.png (https://ibb.co/sC0M3Sq)

The picture has changed markedly. Kivutkalns began to occupy the lion's share of the gene pool, and the percentage of Yamnaya decreased to about 10%. However, the distance has decreased significantly. That is, Lithuanians with the help of Kivutkalns are much easier to introduce themselves. So maybe Kivutkalns itself is a mixture of Yamnaya, hunters and farmers? Let's get a look:

https://i.ibb.co/NxKr6y3/2022-02-05-232441.png (https://ibb.co/b1XFgK3)

The distances of Kivutkalns samples to the proposed mixtures are 0.07-0.09, large. This is like the maximum variance between the Native Amerikan groups given in my previous post. And the chronological gap between Kavutkalns and the Yamnaya, late hunters and farmers, is only about 2000 years. What does it say? But the fact that Kivutkalns is a completely separate population, whose ancestors have long since separated from all the others. Personally, my opinion is that this is a population of hunters, most likely without significant admixtures of Yamnaya and farmers, something like a separate EHG cluster. Their ancestors split off from other EHGs somewhere else in the Late Paleolithic. That is, the standard EHG (Karelia, Samara, Yuzhnyi Oleniy Ostrov, Pit-Comb Ware culture, etc.) are much closer to each other than to the Kivutkalns.
Who is close to the people of Kivutkalns? Only the Estonian Bronze Age (tarand grave culture) and some later populations.
In which populations does the Kivutkalns component play a significant role? Mostly Balts, Estonians, Slavs. By the way, Erzya with Moksha also:

https://i.ibb.co/Fgr06pM/2022-02-06-115658.png (https://ibb.co/nzGnrpv)

I think the problem is just that we lack the proper Hunter Gatherers to model Kivutkalns properly.
Imagine if we didn't have any WHG samples, only EHG, and you tried modelling Europeans with them, the distances would also be high, right?

[smd555]

I think the problem is just that we lack the proper Hunter Gatherers to model Kivutkalns properly.
Imagine if we didn't have any WHG samples, only EHG, and you tried modelling Europeans with them, the distances would also be high, right?

Actually! If we had the sample of hunter with the same genetics - the problem would be solved. In fact, although there are relatively many samples of hunters from that region, the geographic coverage is very small, mainly samples come from Zwejneki and a few other places:

https://i.ibb.co/WDcxmpq/2022-02-11-165507.png (https://ibb.co/wJK05r3)

[smd555]

I attemped to use it, but it's slow and complicated compared even to qpadm and admixture.
But it's effective exactly for that issue which you are asking about.

And what about modern samples? Can qpgraph build an adequate tree of modern peoples, like Slavic?

[vbnetkhio]

And what about modern samples? Can qpgraph build an adequate tree of modern peoples, like Slavic?

I didn't see it used for this, the problem is probably that modern populations are very complex mixes and much more branches would be needed for a good model.