Moots: Ancient Rome Paper

A plausible model could be this. nMonte with G25 is not very precise although and if inappropriate models are used the results can be very misleading.


All the samples from Rome and the surrounding area, from Proto-Villanovan to the Renaissance period.

xPuPsR2.png



The ancient Greek samples. The first two are two Greeks found at Empuries.

ACBdnND.png







What Anatolia? Anatolia BA?

all samples from Bronze Age Anatolia

uG04E59.png

The non-Imperial samples look pretty reasonable, with a few exceptions, don't you think?

I don't know what's up with the Renaissance Roman sample, however: 41% Yamnaya? Maybe it's just an individual aberration, or someone with an origin in more northern places.

That's an awful lot of steppe in Proto-Villanovan. I'm quite close to that sample, and my highest steppe is about 25-30%. I don't know why Villanovan might be pulling so much more WHG. The lower Yamnaya, slightly higher WHG in Spaniards might account for the Spanish matches with some of the Spaniards/Portuguese?

Maybe they're pretty close, maybe they're not. I don't know.

Of course, those are far from proximate sources, but they have the advantage of clarity because of that.

Much better than a lot of the stuff people have been throwing at the wall to see what sticks.

I think there's a real problem using an average for the Imperial samples. It's the same problem I've been pointing out since the first leaks about this paper came out.

The whole point about these Imperial samples is that some of them may have been descendants of people moving up from Cumae to Rome, or Calabria to Rome, etc. and already "perhaps" heavily admixed with "additional" CHG/Iran Neo i.e. in addition to that which arrived with the Italics. The Greek colonization had already taken place by this time, after all, as well as Neolithic and Bronze Age migrations. Until we know what those people looked like this is too much like shooting darts in the dark imo.

Other samples might be directly from Anatolia or places in the Levant who arrived twenty years before. Not all the samples were tested to see if they were born and raised locally. Some of them might be Syrians or Jews, merchants temporarily in the city or not. How the hell would we know???

You can't, imo, average samples from the capital city of what was for its time an international empire, and then make predictions about changes or not in the genomes of people living there 2,000 years later.

Why don't we average out some Chinese and Koreans from Flushing, some Puerto Ricans from the Bronx, some Hasidic Jews from Brooklyn, some Italians from Bay Ridge, some Irish people from Bayside, and some WASPS and JEWS from the upper east side and come up with a prediction for what future inhabitants of this area will look like in 2000 years, and after a near apocalyptic collapse to a population of about 100,000.

It just doesn't work.

The Greek samples are very interesting, especially the "Iranian" percentages of 17%-24%. So, some additional input to the mainland after the Mycenaeans, or are these Greeks from the Islands, Ionia? It will be very interesting to compare them to Greek Colonization Era samples in Southern Italy and Sicily. Very interesting, indeed, Brick.

The Anatolian samples are interesting too. I've been saying for years now that part of Anatolia would have had "Levant" type admixture, and so if migration was from Anatolia, it might pick up Levant-ish admixture as well, and there it is. You can see it in the people of Hatay province in Turkey today, the Cilicia of the Classical World, with its capital of Antioch, even after the "Turkic" migrations. St. Paul knew it well. :)

I hope I can find a lead to that "Tuscan like" sample from the Iron Age Balkans, and maybe you could try it.

Have you ever done this for the more "Italian" like samples from the Langobard paper?

It's cheeky of me, I know, but it would be interesting.


Maybe they're pretty close, maybe they're not. I don't know.
 
Even well intentioned, honest analysis of ancient samples using these tools has to be undertaken and interpreted cautiously.

Let's take a look at one analysis published upthread.

ITA_Prenestini_tribe_IA_o

Greek_Central_Anatolia,42
Sardinian,19.8
Spanish_La_Rioja,13.8
Italian_Piedmont_o,12.2
Greek,8.6
Italian_Umbria,3.6

This does clear up the absurd assertion that the sample is "half Phoenician", but then what are we to conclude from this about the "Latin" Prenestini? In terms of modern populations were they adjusted percentages of Sardinian, Spanish La Rioja, Piemontese, Greeks and Umbrians?

Yes, these tools are certainly not a complete replacement for the more rigorous methods used in the professional studies. That said, they're pretty good as far as the amateur world goes; much better than the more binary "oracles" in my opinion. The model itself simply indicates that the Latin side of this individual was something like a mixture of those populations but it's never going to be perfect. The Latin samples are quite varied in general but the pattern is firmly Western Mediterranean i.e. Spanish, Northern Italian and French populations. We already know that Neolithic and Copper Age Italians are close to Sardinians so this model is something of a crude calculation with some underlying logic. The best course of action for samples that are mixed is to find two points of ancient ancestry and model them as simply as possible.

[1] "distance%=2.4179"

ITA_Prenestini_tribe_IA_o

Anatolia_Kaman-Kalehoyuk_MLBA,60.6
ITA_Prenestini_tribe_IA,39.4

Keep in mind that ITA_Prenestini_tribe_IA is the most northern shifted of all the Iron Age & Republic samples and is closest to French_South. This may result in the outlier coming across as more Anatolian than they would if ran against some of the other Latins, but it made sense to use the other non-outlying Prenestini.

As for Rome Imperial, I'm assuming this is an average?

ITA_Rome_Imperial

Cypriot,40.6
Greek_Crete,33.4
Anatolia_Kaman-Kalehoyuk_MLBA,15.6
ITA_Latini_IA,10.4

Right off the bat, whether this is an average or a particular sample, I think it's difficult to determine how "accurate" it is when it's a mixture of modern and ancient samples.

What we really need to understand these changes is not only Bronze Age but Iron Age and post Greek colonization samples from southern Italy.

Even going with the samples we have, did anyone try to model this group using the Sicily non Beaker "Beaker" sample, for example, or samples from Sardinia?


This is indeed the average. The heterogeneity of these Imperial Roman samples makes it hard to model the average, or come to any concrete conclusions about their genetic structure. They have such a genetic range that it's obvious which ones are native to Italy (those that cluster like Italians) and those that are from the Middle East (those that cluster like Iraqi Jews/Lebanese). The problem lies in the fact that the majority of them have a centroid of clustering around Greek_Kos. I imagine we will need to "organize" them into further distinguishable clusters; not only between Romans and foreigners, but also Aegean, Anatolian and Middle Eastern foreigners. It's a complete mess, in other words.

The samples we currently have of Ancient Greeks indicate a strong continuation from the Bronze Age. The Mycenaean samples are all homogeneous and strongly overlap with the Empuries samples who were themselves Ionian Greeks from Western Anatolia. So we know that Greeks were still much like their Bronze Age ancestors both in Greece and Western Anatolia throughout the Archaic, Classical and Hellenistic Periods, and according to Olalde et al. also well into the Roman Empire. T
he colonists who settled Southern Italy from the Greek world would have almost certainly clustered like Mycenaeans, and by extension Southern Italians, as a result of this (for the most part). I would expect some samples to look like Cypriots/Bronze Age Anatolians because those regions were part of the Hellenic world but I doubt they formed a majority in Southern Italy. Perhaps they were the natural merchant class among the Greek population of Rome; that would explain why the Isola Sacra necropolis has Greek inscriptions but the people seem to cluster tightly around Cypriots rather than Mycenaeans. There are many possibilities; I focus on the Republic and Late Antiquity because port towns during the height of the Imperial Era are simply too messy to deal with using our current tools.

The Sicilian Beakers were removed from the datasheet a while ago for being "low quality".


 
LTG;591548]Yes, these tools are certainly not a complete replacement for the more rigorous methods used in the professional studies. That said, they're pretty good as far as the amateur world goes; much better than the more binary "oracles" in my opinion. The model itself simply indicates that the Latin side of this individual was something like a mixture of those populations but it's never going to be perfect. The Latin samples are quite varied in general but the pattern is firmly Western Mediterranean i.e. Spanish, Northern Italian and French populations. We already know that Neolithic and Copper Age Italians are close to Sardinians so this model is something of a crude calculation with some underlying logic. The best course of action for samples that are mixed is to find two points of ancient ancestry and model them as simply as possible.

[1] "distance%=2.4179"

ITA_Prenestini_tribe_IA_o

Anatolia_Kaman-Kalehoyuk_MLBA,60.6
ITA_Prenestini_tribe_IA,39.4

Keep in mind that ITA_Prenestini_tribe_IA is the most northern shifted of all the Iron Age & Republic samples and is closest to French_South. This may result in the outlier coming across as more Anatolian than they would if ran against some of the other Latins, but it made sense to use the other non-outlying Prenestini.



This is indeed the average. The heterogeneity of these Imperial Roman samples makes it hard to model the average, or come to any concrete conclusions about their genetic structure. They have such a genetic range that it's obvious which ones are native to Italy (those that cluster like Italians) and those that are from the Middle East (those that cluster like Iraqi Jews/Lebanese). The problem lies in the fact that the majority of them have a centroid of clustering around Greek_Kos. I imagine we will need to "organize" them into further distinguishable clusters; not only between Romans and foreigners, but also Aegean, Anatolian and Middle Eastern foreigners. It's a complete mess, in other words.

The samples we currently have of Ancient Greeks indicate a strong continuation from the Bronze Age. The Mycenaean samples are all homogeneous and strongly overlap with the Empuries samples who were themselves Ionian Greeks from Western Anatolia. So we know that Greeks were still much like their Bronze Age ancestors both in Greece and Western Anatolia throughout the Archaic, Classical and Hellenistic Periods, and according to Olalde et al. also well into the Roman Empire. T
he colonists who settled Southern Italy from the Greek world would have almost certainly clustered like Mycenaeans, and by extension Southern Italians, as a result of this (for the most part). I would expect some samples to look like Cypriots/Bronze Age Anatolians because those regions were part of the Hellenic world but I doubt they formed a majority in Southern Italy. Perhaps they were the natural merchant class among the Greek population of Rome; that would explain why the Isola Sacra necropolis has Greek inscriptions but the people seem to cluster tightly around Cypriots rather than Mycenaeans. There are many possibilities; I focus on the Republic and Late Antiquity because port towns during the height of the Imperial Era are simply too messy to deal with using our current tools.

The Sicilian Beakers were removed from the datasheet a while ago for being "low quality".



The Sicilian Beakers were removed from the datasheet a while ago for being "low quality".


How "low quality"? Are there other samples on the datasheet of similar or "lower" quality?

Is there access to them for independent modeling?

The problem lies in the fact that the majority of them have a centroid of clustering around Greek_Kos. I imagine we will need to "organize" them into further distinguishable clusters; not only between Romans and foreigners, but also Aegean, Anatolian and Middle Eastern foreigners. It's a complete mess, in other words.

So I've been saying until I'm blue in the face.

Still, it would be informative, imo, to list the Imperial Era samples by sample number in clusters: Aegean, Anatolian, Middle Eastern, Italian Peninsula. If they have been uploaded to gedmatch people can compare their own data to the samples.

As to the Greeks from Empuries, the post from LTG using G25 seems to indicate that one of the Empuries' samples has more "Iranian" type ancestry than the Mycenaeans, doesn't it? Unless you mean that it's still roughly in the same range. I had already speculated that the more Iranian heavy sample might be from Ionian Greece.
 
I hope I can find a lead to that "Tuscan like" sample from the Iron Age Balkans, and maybe you could try it.


Bulgaria_IA? Croatia_IA?

That's an awful lot of steppe in Proto-Villanovan. I'm quite close to that sample, and my highest steppe is about 25-30%. I don't know why Villanovan might be pulling so much more WHG. The lower Yamnaya, slightly higher WHG in Spaniards might account for the Spanish matches with some of the Spaniards/Portuguese?

Maybe nMonte/G25 inflates WHG, I don't know. However yes, the higher WHG is the reason why the Latins and some Etruscans go in the direction of Iberia/South France.


The whole point about these Imperial samples is that some of them may have been descendants of people moving up from Cumae to Rome, or Calabria to Rome, etc. and already "perhaps" heavily admixed with "additional" CHG/Iran Neo i.e. in addition to that which arrived with the Italics. The Greek colonization had already taken place by this time, after all, as well as Neolithic and Bronze Age migrations. Until we know what those people looked like this is too much like shooting darts in the dark imo.

Other samples might be directly from Anatolia or places in the Levant who arrived twenty years before. Not all the samples were tested to see if they were born and raised locally. Some of them might be Syrians or Jews, merchants temporarily in the city or not. How the hell would we know???

You can't, imo, average samples from the capital city of what was for its time an international empire, and then make predictions about changes or not in the genomes of people living there 2,000 years later.

Why don't we average out some Chinese and Koreans from Flushing, some Puerto Ricans from the Bronx, some Hasidic Jews from Brooklyn, some Italians from Bay Ridge, some Irish people from Bayside, and some WASPS and JEWS from the upper east side and come up with a prediction for what future inhabitants of this area will look like in 2000 years, and after a near apocalyptic collapse to a population of about 100,000.

It just doesn't work.


Later I post all the individual results from Imperial samples.


Have you ever done this for the more "Italian" like samples from the Langobard paper?

It's cheeky of me, I know, but it would be interesting.

Maybe they're pretty close, maybe they're not. I don't know.


The more "Italian" like samples from the Langobard paper


Bv8ufiL.png



The G25 Italian averages

5Xket97.png



The G25 Balkans averages (including Greece and Crete)

7O8kAfo.png





PCA

SZ43 plots in the Tuscan cluster (the yellow one), Z36 more with Marche, CL23 with Lombardy/Bergamo.


Evi7wO8.jpg
 
These models are some of the most logical models I've seen on Ancient Greeks and Romans using this tool.
And I knew the Greek Empuries had more Iran than the Mycenaean average bc it seemed to plot a bit eastward in
that study and it shows in its high Armenia score of 25 percent. Some Greeks may have had more Caucasus
than others (and most or all of it from Bronze Age/copper age Anatolian ancestry)

Agreed. The ancient Greeks had colonies in Anatolia and will have mixed very much with the local Anatolia_BA-like population. Probably the ancient Greeks are among the main sources of CHG/Iran_N in Italy.
 
Bulgaria_IA? Croatia_IA?



Maybe nMonte/G25 inflates WHG, I don't know. However yes, the higher WHG is the reason why the Latins and some Etruscans go in the direction of Iberia/South France.





Later I post all the individual results from Imperial samples.





The more "Italian" like samples from the Langobard paper


Bv8ufiL.png



The G25 Italian averages

5Xket97.png



The G25 Balkans averages (including Greece and Crete)

7O8kAfo.png



PCA

SZ43 plots in the Tuscan cluster (the yellow one), Z36 more with Marche, CL23 with Lombardy/Bergamo.


Evi7wO8.jpg

It was Bulgaria in the Iron Age, but only one of the samples was "Tuscan like". I'll try to find the paper, but who the heck knows if he put it on the list.

Where are you estimating placement for CL36? It looks a little "north" of Toscana, but not Lombardia.

Thank you to both LTG and Brick for sharing all of this data. These definitely go in my files.
 
It was Bulgaria in the Iron Age, but only one of the samples was "Tuscan like". I'll try to find the paper, but who the heck knows if he put it on the list.


HRV_IA:I3313 might be Croatia IA, and BGR_IA:I5769 might be Bulgaria Iron Age.

DQqUITw.png



Where are you estimating placement for CL36? It looks a little "north" of Toscana, but not Lombardia.

Most likely Liguria, even if CL36 plots in the Piedmont cluster that is currently the northern Italian sample closest to the Tuscany. It's doesn't look so distant from Tuscany and there is no Emilia-Romagna and Liguria has only one individual (ALP099).

For a complete picture, G25 Spanish and Portuguese averages.

DPY3l9Q.png


What Iberomaurusian is.

KLZvM2X.png
 
There is a displacement, something fails, none are correct. It's the only thing I can say: Oracle
 
HRV_IA:I3313 might be Croatia IA, and BGR_IA:I5769 might be Bulgaria Iron Age.

DQqUITw.png





Most likely Liguria, even if CL36 plots in the Piedmont cluster that is currently the northern Italian sample closest to the Tuscany. It's doesn't look so distant from Tuscany and there is no Emilia-Romagna and Liguria has only one individual (ALP099).

For a complete picture, G25 Spanish and Portuguese averages.

DPY3l9Q.png


What Iberomaurusian is.

KLZvM2X.png

How is Levant_Natufian modeled in this program? I know they already had some Anatolain_N in them.
 
HRV_IA:I3313 might be Croatia IA, and BGR_IA:I5769 might be Bulgaria Iron Age.

DQqUITw.png





Most likely Liguria, even if CL36 plots in the Piedmont cluster that is currently the northern Italian sample closest to the Tuscany. It's doesn't look so distant from Tuscany and there is no Emilia-Romagna and Liguria has only one individual (ALP099).

For a complete picture, G25 Spanish and Portuguese averages.

DPY3l9Q.png


What Iberomaurusian is.

KLZvM2X.png

The "Piedmont" cluster and the Ligurian cluster are almost identical because those "Piedmont" samples are mountain Ligures. They only became "part" of Piemonte very recently, they speak a Ligurian dialect, and every town has "Ligure" in its name. They're also pretty close to Apennine Emilians. So, yes, CL36 is probably closest to mountain Ligures, and therefore to eastern Liguria. It makes absolute sense I'd get a 4.5 similarity to CL36, since my father's family comes from that adjoining area of Apennine Emilia. I get almost the same "fit" on calculators to that Piedmont sample when it's on calculators, although actually a bit higher.

As for SZ43, my highest match at 3.4, it has absolutely no Iran related ancestry, and Marche with its 12%, plus 5% Levant/Natufian seems too "southern" to me, so I don't think Marche is quite right, and it might be closer to Romagna, but I won't quibble because we have no sample from there, and there's the whole Northern Marche vs Southern Marche thing as well. :)

It doesn't look to me like he included the "Tuscan like" Iron Age sample, of course.

As for the Spanish samples, they may plot near the Northern Italians, but the mix is different: less Indo-European (about 30%, down to 26% among the Basques), but more WHG. It makes perfect sense: it's always been clear from the ancient samples that there was more surviving WHG there. That probably explains the affinity to the Etruscan samples they have, because their WHG is a bit elevated as well.

It still amazes me how ancient dna can just put to bed questions that have generated debate for so long.

From a personal point of view I'm amazed that I can end up with a mix so close to an ancient sample.

The miracle of ancient genomics.
 
How is Levant_Natufian modeled in this program? I know they already had some Anatolain_N in them.

Yes, excellent question.
 
How is Levant_Natufian modeled in this program? I know they already had some Anatolain_N in them.

It's the average of two Levant_Natufian samples, Levant_Natufian:I0861 and Levant_Natufian:I1072. It might be not completely accurate.

MAR_Iberomaurusian is instead the average of five ancient samples.


As for SZ43, my highest match at 3.4, it has absolutely no Iran related ancestry, and Marche with its 12%, plus 5% Levant/Natufian seems too "southern" to me, so I don't think Marche is quite right, and it might be closer to Romagna, but I won't quibble because we have no sample from there, and there's the whole Northern Marche vs Southern Marche thing as well.


The position in a 2D PCA may depend on many reasons, not necessarily an ancient sample that plots in the cluster of a modern sample has exactly the same percentages of ancestral components.

The average of Marche is based on 16 individuals.
 
rYytQv8.png
cC9Gcn0.png

dCdPVMr.png


Here are figures from the Raveane et al 2019 paper, that was referenced in the Moots paper.

One of the Anatolian Bronze Age samples plots right on top of Southern Italians. Does anyone know which sample that one is? There seems to be a noticeable difference from the others, on the PCA. It is also close to the Mycenaean samples.

Also, according to Antonio et al, 850 can form a clad with an Anatolian Copper Age individual. I recall from the qpAdm modeling from de Barros Damgaard et al 2018, that Copper Age Anatolian are modeled as 60% Anatolian_N + 40% CHG

PCA (Fig. 2B) indicates that all the Anatolian genome sequences from the Early Bronze Age (~2200 BCE) and Late Bronze Age (~1600 BCE) cluster with a previously sequenced Copper Age (~3900 to 3700 BCE) individual from Northwestern Anatolia and lie between Anatolian Neolithic (Anatolia_N) samples and CHG samples but not between Anatolia_N and EHG samples. A test of the form D(CHG, Mbuti; Anatolia_EBA, Anatolia_N) shows that these individuals share more alleles with CHG than Neolithic Anatolians do (Z = 3.95), and we are not able to reject a two-population qpAdm model in which these groups derive ~60% of their ancestry from Anatolian farmers and ~40% from CHG-related ancestry (P = 0.5). This signal is not driven by Neolithic Iranian ancestry, because the result of a similar test of the form D(Iran_N, Mbuti; Anatolia_EBA, Anatolia_N) does not deviate from zero (Z = 1.02). Taken together with recent findings of CHG ancestry on Crete (58), our results support a widespread CHG-related gene flow, not only into Central Anatolia but also into the areas surrounding the Black Sea and Crete. The latter are not believed to have been influenced by steppe-related migrations and may thus correspond to a shared archaeological horizon of trade and innovation in metallurgy (59).

https://science.sciencemag.org/content/360/6396/eaar7711
 
One of the Anatolian Bronze Age samples plots right on top of Southern Italians. Does anyone know which sample that one is? There seems to be a noticeable difference from the others, on the PCA. It is also close to the Mycenaean samples.


In Raveane 2019 the ancient samples labeled as Anatolia_BA are I2495, I2499, I2683 from Lazaridis 2017 on Mycenaeans and Minoans.
 
In Raveane 2019 the ancient samples labeled as Anatolia_BA are I2495, I2499, I2683 from Lazaridis 2017 on Mycenaeans and Minoans.

This is my affinity to those samples:

78. Anatolia Bronze Age (1625 BC) ..... 18.73 - I2495 -
Top 98% match vs all users
81. Bronze Age Anatolia (2050 BC) ..... 18.98 - I2683 -
Top 98% match vs all users
89. Bronze Age Anatolia (2650 BC) ..... 20.18 - I2499 -
Top 98% match vs all users
 
This is my affinity to those samples:

78. Anatolia Bronze Age (1625 BC) ..... 18.73 - I2495 -
Top 98% match vs all users
81. Bronze Age Anatolia (2050 BC) ..... 18.98 - I2683 -
Top 98% match vs all users
89. Bronze Age Anatolia (2650 BC) ..... 20.18 - I2499 -
Top 98% match vs all users

50. Copper Age Anatolia (3800 BC) ..... 15.88 - I0184 [FONT=&quot][/FONT]
Top
[FONT=&quot]​
99
% match vs all users

However, I get a closer affinity to the copper age sample.[/FONT]
 
It's the average of two Levant_Natufian samples, Levant_Natufian:I0861 and Levant_Natufian:I1072. It might be not completely accurate.

MAR_Iberomaurusian is instead the average of five ancient samples.





The position in a 2D PCA may depend on many reasons, not necessarily an ancient sample that plots in the cluster of a modern sample has exactly the same percentages of ancestral components.

The average of Marche is based on 16 individuals.

Well, I highly doubt any of them are zero "CHG/Iran Neo" like admixture.

We'll have to agree to disagree.
 
50. Copper Age Anatolia (3800 BC) ..... 15.88 - I0184
Top
99
% match vs all users

However, I get a closer affinity to the copper age sample.

Those Anatolian Bronze Age individuals don't appear on my husband's list of 60, but he's still shy of a fit of 16 at that point, so I'm sure they'd show up at some point. I just don't know the exact fit.

His fit with the Copper Age Anatolian:
49. Copper Age Anatolia (3800 BC) ..... 14.79 - I0184 [FONT=&quot][/FONT]
Top
[FONT=&quot]​
99
% match vs all users[/FONT]
 
Eurogenes' analysis of these samples is also dead in the water. Time to fish that out too.

What's that old saying? "After three days guests and fish start to stink"? This one stunk to high heaven the FIRST day. :)
 

This thread has been viewed 357975 times.

Back
Top