I am not an expert on I2, I2-Y3120 or a subject matter expert on ancient tribes. If anything, I'm an expert on geographically peripheral J2b-L283. I am only looking at I2-Y3120 through the prism of geolocated samples on the YFull tree and the estimated TMRCAs.
I can't believe that 1.5 years have passed since I first wrote about this much discussed lineage I2-Y3120.
I decided to take another look at this interesting lineage using some of my newly developed tools, some still under development and soon to be released.
I now feel that in my original post, I may have overlooked the possibility of a Dacian tribal origin for this lineage because I was unfamiliar with Dacian history.
There are different ways diversity can be measured, and I explain the methodology used by Diversity Heatmap in the FAQ. Diversity Heatmap takes regional sample rate into account.
Interestingly, I2-Y3120 exhibits diversity over a relatively large area. The central-most areas exhibiting high diversity form a roughly boomerang-shaped region with one peak in the Carpathians of Poland and Slovakia and another in the Dinaric Alps of Bosnia, Serbia, Montenegro and Kosovo.
Diversity Heatmap, based on the geolocated samples on the YFull YTree, does not compute Ukraine as an area of high diversity for I2-Y3120. I mention this because I have heard that some older studies based on STRs indicate Ukraine as having higher diversity.
Centroids Computed Around Romania
Most of I2-Y3120's six child lineages' centroids are computed to within the aforementioned area of greatest diversity. I2-Y3120 itself has a centroid computed to the northwest of modern Argeș county. This is because it is a central location among the centroids of its children.
I include screenshots from the Haplogroup Research Analytical Suite I am currently developing together with Thomas Krahn. This tool allows you to designate countries as outliers and then recomputes the centroids on-the-fly.
Distant outliers usually pull the centroid in their direction to some extent, even though my algorithm computes a centroid not in the strictly traditional sense (i.e. average of positions), but as the average of the weighted centroid (strictu senso) and the position from the set of downstream nodes which minimizes distance to all other nodes weighted by downstream node TMRCA and giving additional weight to ancient samples. I will wait until the tool is released to provide the exact details, as they may change.
I will note that these screenshots use a version of centroid calculation that does not use regional sampling rate as an additional factor. I plan to make the algorithm take this into account, if not by the time the tool is initially released, then in a future update.
In the case of I2-Y3120, I decided to exclude samples from Russia, Finland, Sweden, Estonia, Latvia and Lithuania because these samples are far from the centers of diversity. The effect of removing these outliers was not drastic but I have more confidence in the result.
That being said, neither Diversity Heatmap nor these centroid computations are guaranteed in any particular case to indicate the origin of a haplogroup.
These tools are meant to be used to visualize geospatial sample data as part of a multi-disciplinary approach to investigate haplogroup origins.
I now include the screenshot for I2-PH908 to show that about 84% of the peak I2-Y3120 frequency in Bosnia and Serbia comes from the offspring of a single man, I2-PH908 who lived about 1700 years ago.
I2-Y3120's Sibling - Ancient I18719
I'm not aware of any older ancient samples than the 1200 ybp samples from South Moravia, Czechia and Bursa, Turkiye.
In that case we should look at where I2-Y3120's siblings are found.
There is an ancient sample on the YFull YTree with low coverage that is designated as either I2-CTS10228 or a kind of I2-Y3120 (though his age significantly predates the TMRCA). The coverage is very low at 2x but the age makes the sample interesting.
YFull has the sample as 3200 ybp. I will note that his female line is in subclade HV0a1a1b with a much more recently computed TMRCA of 225 years according to MTree 1.02.19113. I have heard from another researcher that people doubt the age of this sample is really 3200 years but I do not think this young MTree estimate alone is proof of some sort of mistake, because MTree TMRCAs are in beta and not as reliable as those on the YTree.
Please feel free to contact me or post as a comment a link to some opinion regarding the provenance of this sample.
All I know is that, in my experience with J2b-L283 research, the people who doubt the provenance of ancient samples are generally motivated reasoners.
Assuming that the sample really is 3200 years old, then the true geographic origin of I2-CTS10228 (or whatever subclade he may actually be, were his coverage high enough) may have been near Vrhovine, Croatia given that he lived only about two hundred years after the I2-CTS10228 TMRCA.
I2-Y3120 Lacks Roman-mediated Migration to England or the Rhine
I am however skeptical that I2-Y3120 itself may have arisen in Croatia. The modern distribution lacks the pattern of likely Roman-mediated migration to the Rhine or England that J2b-L283 lineages from this area exhibit.
Of the two oldest lineages that settled within areas controlled by Rome:
- I2-PH908 in Bosnia / Serbia - the TMRCA is 1700 ybp so they may not have been living there during the period that Roman was in control and recruiting there
- I2-Y18331 in Greece - I have not found many references of Romans recruiting Greeks to serve as auxiliaries or legionaries.
I2-Y3120's Sibling - I2-Y81696
The TMRCA of 1600 ybp and being found in Baden-Wuerttemberg and Bas-Rhin, in my opinion, is not a reliable indicator that the deeper origin of this line came from the area. I say this based on the research of my own haplogroup J2b-L283 that has many lineages found in that area now that are proven by ancient DNA to not have originated there.
This line could also possibly have migrated to the Upper Rhine area due to a Roman-mediated migration or some other Migration Period population movement.
If I2-Y3120 really did originate in a Dacian dynasty, and I'll get to them next, then we might consider I2-CTS10228 as a Balkan remnant whose line barely survived the J2b-L283 expansion (and other regional challenges) and who did reasonably well in reproducing around 1400 BCE.
I find that some of the circumstantial information regarding I2-Y3120 matches aspects of Dacian history. Modern historians appear to prefer the term Geto-Dacian but I have just used the term Dacian for simplicity here.
I recommend reading the Wikipedia article on Dacians, it is concise and interesting.
To paraphrase, there were Dacian kings before Burebista, dating back to 450 BCE, but only after subsequent contact and influence from the Celts was their new La Tene-based economy able to facilitate the consolidation of political power through tribal unions.
The estimated TMRCA of 100 BCE is very close to this period of consolidation of power of the Dacian monarchy.
Burebista's accession came with the expulsion of Celts around 60 BC when his forces moved through to the middle Danube region, and with the support of the religious establishment and leaders in Dacia which brought around a stricter moral code in the Dacian kingdom. Around this time the pottery of the Dacian style began appearing in Celtic settlements in Central Europe, including the area covered by the former Yugoslavia, especially in Gomolava, Yugoslavia, and Budapest, Hungary. - Burebista on Wikipedia
The Dacians are believed to have defeated the Scordisci, who were the dominant tribe of the northwestern Balkans at that time.
So if I2-Y3120 were Dacian, some of them may have migrated to these centers of modern diversity and higher frequency as far back as the reign of Burebista 60 BCE.
Burebista became Rome's enemy by siding with Pompey (spoiler alert - he lost) but before a war broke out he was assassinated by his own noblemen.
After his death, the kingdom was split into four and then five parts, with the original power center confined to the Orăştie Mountains (Bunson 2014)
I2-S17250 Through a Dacian Prism would be Decebalus
I'll copy an excerpt from my 2021 article that is still valid:
If you look at how many SNPs comprise each of I2-Y3120 child lineages, you can infer the following:
- He had three sons* that were the most successful in terms of being prolific, I2-Y18331, I2-Z17855, I2-Y4460
- A fourth son was not as prolific himself, but several generations later, one of his line founded a new dynasty, I2-S17250.
- Through this dynasty, the fourth son would ultimately have more descendants than all the rest of his siblings put together
(note two lines I am not mentioning here with longer bottlenecks followed by less growth - I2-BY154615 and I2-FT76511)
The fourth son, fourth in enumerating as we do not know the sequence of birth or whether each SNP in fact represents several successive generations during which only a single SNP accumulated, had five child lineages.
It is worth noting that some of the oldest I2-Y3120 found in South Moravia, however unfortunately not old enough to preclude various theories of when it arrived there, is only found in two lines of I2-Y3120, namely I2-Y5596 and I2-Y4882 which are children of I2-S17250
So if our attributions of I2-Y3120 to Burebista and I2-S17250 to Decebalus are correct, then this would indicate that the 1200 year old ancient samples in South Moravia were not migrants during the period of Burebista's conquest of the Boii and Taurisci, but may have migrated there later.
What about Celts and Slavs?
The interesting thing about a Dacian theory is that it doesn't shut the door on either of:
- Possibility that I2-Y3120's deeper origins are Celtic from Central Europe
- Possibility that some lines joined migrating Slavs and Slavicized before migrating to their final destination
To learn more about the former, we would need to break the I2-Y3120 bottleneck from 3400 to 2100 ybp. However if the 3200 year old ancient I18719 is legit (and as I said I have no evidence that it is not), then I think it makes more sense to view I2-CTS10228 as having been a Proto-Illyrian/Dacian/Thracian lineage in 1400 BCE.
To definitively learn more about the latter, we would need ancient samples dating to before the Slavic migrations to the Balkans. Again I18719 is key because it shows that the lineage's closest sibling was living in future Slavic Croatia nearly two millennia before the arrival of Slavs.
I2-Y18331 Overwhelming Diversity in Greece
The diversity of this one particular line of I2-Y3120 in Greece is so great, that it even contributes significantly toward Greek diversity of parent I2-Y18331, the calculation of which takes five other sibling lines into account, which have significantly less representation in Greece and those lines in Greece having younger TMRCAs.
The samples of some of these other lines of I2-Y3120 represented in Greece contain hints on the YFull YTree of a more recent migration to Greece from somewhere else, for instance "mkd" Macedonian, "blg" Bulgarian, and "pnt" Pontic language codes.
However there is overwhelming circumstantial evidence that I2-Y18331 originated in Greece. Otherwise it was a massive later co-migration almost exclusively to Greece proper leaving no trace of where it came from and siring no children along the way.
One of the few child lineages with non-Greece modern distribution, I2-Y23115, has a TMRCA of 1100 years ago. I was told by another researcher that these men all have oral traditions of being eastern Ashkenazi.
Given the context of overwhelming diversity in Greece dating to 100 BCE, this Ashkenazi line may have originally been Greek Jews. History of Jewish settlement in Thessaloniki may stretch back 2000 years, according to Wikipedia.
Looking at I2-Y18331 through a Dacian prism, this line may represent a son or descendant of Burebista who came into contact with the Greeks. Burebista conquered and annexed the Greek fortresses along the Black Sea coast from Olbia to Apollonia but I find no mention of raids or conquests into Greece proper.
There is an interesting Greek figure named Akornion, an important citizen of the Ionian Greek colony of Dionysupolis (modern Balchik). An inscription there dating to 48 BCE describes Akornion as the "first and greatest friend" of Burebista (Oltean 2007). As Burebista's ambassador, Akornion convinced Pompey to accept Burebista's support.
I find this reference to the meeting place by googling from jstor (F Millar 1984):
When the king took him as a confidant and sent him as an ambassador to Pompey, encamped at. Heraclea Lyncestis in northern Greece in 49-8 B.C
So some of what I consider most likely explanations for this line becoming Greeks now living in Greece proper include:
- Demic diffusion of Hellenized Dacians from conquered Black Sea coastal colonies
- Dacian bodyguard of ambassador Akornion may have accompanied him to meet Pompey at then Greek Heraclea Lyncestis (modern Bitola, N Macedonia)
- A son of Burebista may have sealed a political alliance with a friend or family member of Akornion through marriage to a Greek woman. A higher social status can help explain the higher reproductive fitness of this line in 100 BCE.
Rgarding Heraclea Lyncestis, this town was later abandoned by Greeks following an earthquake and Slavic invasions.
What about the Russian Flag Samples?
There is a lot of diversity of geography and ethnicity in the samples who have marked paternal descent to somewhere in Russia.
Thankfully, 72 of the 76 samples with Russia country code have a more specific regional code.
I calculated some quick statistics comparing the Russian regional code distribution of I2-Y3120 with R1a-L1029, the latter being an examplar of what we would expect for a Proto-Slavic tribe.
I found a similar proportion of Russia samples (about 1/8) tracing descent to Tatarstan in both. I share this information just out of curiousity. It didn't help me reach any conclusions.
It is easier to see differences in regional distribution in the heatmap.
Two things are clear by comparing the relative frequency maps.
- Much more of R1a-L1029 ended up within Russia vs elsewhere as copmared to I2-Y3120.
- There is much less representation of I2-Y3120 within the Grand Duchy of Moscow relative to R1a-L1029.
Did the Romans and/or R1a-L1029 displace/replace I2-Y3120 in Romania?
R1a-L1029 having displaced/replaced I2-Y3120 in Romania could help to explain why:
- I2-Y3120 is found less in Russia
- R1a-L1029 has a peak in Romania
- I2-Y3120 has a relative hole in Romania
Note that Romania and Moldova are still undersampled relative to their neighbors. There is more that we can learn.
I left out the terms for ethnic groups Dacian or Slav here because by the time of the Slavic migrations these older haplogroups may have taken on any number of ethnic identities, dictated by their geography and the shifting political alliances.
Another reason for the I2-Y3120 hole in Romania is destruction of Dacian men and power centers by Trajan followed by Roman colonization.
One hundred thousand male slaves were sent back to Rome; and to discourage future revolts, legions XIII Gemina and V Macedonica were permanently posted in Dacia. - from Trajan's Dacian Wars but not attributed
By the way only one sample from Italy on YFull could possibly descend from a Dacian slave sent to Rome at that time, in this branch I2-FT277965.
The other three samples from Italy are all too closely related (within 1000 years) to men from Bosnia, that such an early migration to Italy can be ruled out.
If I2-Y3120 were actually Dacians, we might eventually expect to find some relatives, perhaps splitting the bottleneck, in the burials of the neighboring Thracian Odrysian kingdom.
Many elite burials with golden funeral masks have been recovered by archaeologists, yet I find no Iron Age male samples from Bulgaria on the YFull tree yet.
I would not be surprised to find some ancient J2b-L283 in samples from the Odrysian kingdom.
Regarding the ancient sample I18719, I will be willing to send a link to my contacts at YFull to a subject matter expert's best arguments for why this sample's 3200 ybp age should not be accepted on the YFull tree.
The argument should not consist of theories about the origin of this lineage because the theories should proceed from the samples, not vice versa.
Now for some additional maps and diversity metrics per country.
Diversity of I2-Y3120 by Country
Some countries are disproportionately represented by the most prolific child of I2-Y3120, I2-S17250.
Bosnia, Serbia, Montenegro and Croatia have the least diversity in this sense, and therefore likely do not represent the geographic origin of the I2-Y3120 Most Recent Common Ancestor.
Ukraine and Russia also exhibit low I2-Y3120 diversity, especially considering that the only samples from these countries that are not either I2-S17250 or Carpathian I2-Y4460 (see below) are in I2-Y18331 and I2-Z17855, subclades with very consistent diversity well south of the Carpathians.
The amount of the modern diversity of Poland and Belarus in I2-Y18331 that could reflect the ancient origin is a bit reduced by the fact that 4/4 of the Belarus samples and 2/3 of the Poland samples are below I2-Y23115, which someone anecdotally told me is a Jewish line, and if this is true, might reflect a more recent migration from the more likely ancient origin of Greece. I have not had any contact with any men in this line. I note that there are no Yiddish or Hebrew language codes on any of these 11 samples on the YFull YTree. If you, the reader, manage any of these samples, please consider applying a language code in the event that the sample has an oral tradition of having spoken a different language.
Czechia is least diverse because all the samples are found downstream of just one child, I2-S17250. Slovakia is also not diverse.
Poland and Belarus have a bit higher computed diversity compared to their neighbors Ukraine, Russia, Czechia and Slovakia because one sample in I2-FT76511 is from Poland and one sample at the I2-Y3120 level is from Belarus. Note that the code I have written that computes pie charts does not take into account that this sample is not proven to be basal, hovering over the sample you can see he could be positive for a number of subclades for which he has no call. The Diversity Heatmap does however take 'incomplete-basality' into account.
North Macedonia is interesting because for a country with only five samples of I2-Y3120, four major subclades are represented. Greece and Turkey are also relatively more diverse than some of the remaining countries I have not mentioned.
Greece and Turkey have relatively high diversity, however the samples from these countries positive for I2-Y4460 and some lineages of I2-S17250 probably reflect a later migration from the Carpathians, if modern diversity of I2-Y4460 reflects a deeper origin there (below).
I leave it to subject matter experts to make any further conclusions on their own.
NE Carpathian I2-Y4460
This is the second child line of I2-Y3120 with highest diversity in the Carpathians. The other is I2-S17250 (see above in original post).
Because it has a more eastern distribution, I did not mark samples from Russia as outliers to be excluded from the computation of the centroids.
Interestingly this lineage did not end up migrating in great numbers to the Balkans where there are now Slavic speakers.
However, one line did migrate to Greece, Montenegro, Syria and Turkey. The YFull TMRCA estimate for this line is 1650 ybp, however the 95% CI is quite large, at 2200-1200 ybp. So this is a bit shaky footing to conclude for sure whether the migration was before or with the Slavs.
I find it noteworthy, however, that the newest sample from Greece, YF115696, ends up splitting the subclade and establishing greater diversity in Greece vs Montenegro for this lineage - I2-FTD23572. These samples don't have accompanying language codes that in many cases make it obvious that that the ancestors were Slavic speakers that we see in many of the other branches of I2-Y3120 that are widespread in the Balkans.
Given that the sole sample from Syria in all of I2-Y3120 is most closely related to these Greece and Turkey samples, I would consider a late Roman Empire migration of Greeks (whose ancestor had Hellenized prior to the TMRCA) as a possible vector by which it became established in Turkey and Syria.
Because the currently computed TMRCA is 1650 ybp, I am open to the possibility that this lineage I2-Y4460>BY37317 may have originally migrated to Greece from the north before the migration of the Slavs, though Slavic migration is also possible if the true TMRCA is later or we are dealing with a co-migration where all descendants migrated together to Greece.
Another factor that I consider makes a pre-Slavic-mediated migration plausible for I2-Y4460>BY37317 is that I2-Y4460 isn't found in any other neighboring Slavic speakers of the Balkans, except for a sole Bulgarian sample.
Another type of migration that could move a Slav to Greece from a population of Slavs that didn't invade the Balkans is through being enslaved by Greek colonies of the Black Sea.
Maybe this explanation is the most parsimonious way to bridge the geographic gulf between Ukraine and Greece. However how likely were offspring of an enslaved Slav in Greece to successful diversify over the subsequent several generations?
So this is the second line in Greece of I2-Y3320 whose diversity in Greece, based on current age estimates, appears to predate documented Slavic migrations - though it could have still been a proto-Slav who migrated in this case (and maybe in the case of I2-Y18331, I didn't think of this possibility when I wrote the original post).
South of Carpathians I2-Z17855
This line's centroid is computed between Bulgaria and Romania, taking all samples into account.
However, setting a country exclusion on Russia, which is an outlier, the computed origin shifts southwest to the tri-state border of Serbia, Romania and Bulgaria.
The oldest and most prolific subclade of I2-Z17855 is I2-Y230195, with TMRCA about 1700 years ago.
Because I2>Z17855>Y230195's TMRCA is older than its siblings, it contributes more weight toward western Bulgaria for the centroid of parent I2-Z17855.
So I think the TMRCA of I2-Z17855 was most likely living in western Bulgaria or the southern Carpathians.
Given that the 95% Confidence Interval for the TMRCA of I2-Z17855 is 2000-1550 ybp, I think we should be open to the possibility that this lineage, as some of the others of I2-Y3120, may have predated a Slavic migration to the region.
We will require ancient samples to be found in clear archaeological contexts to know for certain which of these lineages at which times may have been Celts, Dacians, Thracians, Greeks, Slavs or other groups.
While two lineages of I2-Y3120 have greater diversity in the NE Carpathians, other lineages appear to have originated south of the Carpathians and would likely represent populations, portions of which, in some cases, later underwent Hellenization, Romanization or Slavicization.