Uralians of the Sargat horizon

Many years ago, well before the start of the ancient DNA revolution, someone made the very clever inference that the N-Tat Y-chromosome marker was closely associated with the expansion of Uralic languages.

Since then, N-Tat has been renamed several times over, to the point that I no longer know what it's called, but the aforementioned inference has turned into a very solid consensus backed up by a wide range of studies focusing on modern and ancient DNA.

Nowadays, Y-haplogroup N-L1026, a subclade of N-Tat, is seen as the main genetic signal of the Uralic expansions, along, of course, with Nganasan-related genome-wide genetic ancestry.

A recent paper at Science Advances by Gnecchi-Ruscone et al. featured the first ever genome-wide samples from the Sargat horizon, which is an Iron Age archeological formation in western Siberia normally associated with the Ugric branch of the Uralic language family. Surprisingly, and disappointingly, the authors failed to investigate this widely accepted connection.

If we go by the Y-haplogroup classifications in the paper, which may or may not be the smart thing to do, at least two of the Sargat horizon males belong to N-L1026, and one also to the more derived N-Z1936 subclade, which has been found in the remains of Hungarian Conquerers from Medieval Hungary. Of course, Hungarian is an Ugric language generally thought to have been introduced into the Carpathian Basin by the Hungarian Conquerers who originally came from western Siberia.

That's probably enough to corroborate the association between the Sargat horizon and the spread of Ugric/Uralic languages, but let's also take a quick look at the autosomal DNA of these Sargat individuals. Firstly, here's a Principal Component Analysis (PCA), based on Global25 data and produced with the Vahaduo G25 Views online tool. The results are self-explanatory.

Interestingly, I can't get a decent statistical fit when I try to reproduce the four-way qpWave/qpAdm model done by Gnecchi-Ruscone et al., probably mostly because my right pops or outgroups are different. This suggests to me that there's something important missing in their model.

Sargat_IA
MNG_Khovsgol_LBA 0.203±0.045
RUS_Ekven_IA 0.183±0.044
RUS_Sintashta_MLBA 0.545±0.014
TKM_Gonur1_BA 0.068±0.013
chisq 16.805
tail prob 0.0186971
Full output

So how about if I replace RUS_Ekven_IA with kra001, the oldest Nganasan-like individual in the ancient DNA record (see here), and MNG_Khovsgol_LBA with KAZ_Mereke_MBA, to add a more local stream of ancestry?

Sargat_IA
KAZ_Mereke_MBA 0.135±0.017
kra001 0.301±0.007
RUS_Sintashta_MLBA 0.499±0.023
TKM_Gonur1_BA 0.066±0.015
chisq 8.872
tail prob 0.262001
Full output

That's a better statistical fit and also, I'd say, a more realistic model, at least in terms of distal ancestry proportions. Note that Nganasan-related ancestry makes up 30% of the genome-wide genetic structure of the Sargat samples, which again corroborates the view that Uralic languages were spoken within the Sargat horizon.

Update 28/04/21: This is the best qpAdm model that I could find for Sargat_IA, at least in terms of the chisq and tail prob. It shows that the Sargat population was in large part very similar to that of KAZ_Pazyryk_IA.

Sargat_IA
KAZ_Mereke_MBA 0.032±0.016
KAZ_Pazyryk_IA 0.698±0.016
RUS_Sintashta_MLBA 0.236±0.021
TKM_Gonur1_BA 0.034±0.014
chisq 2.023
tail prob 0.958561
Full output

It's missing kra001, because KAZ_Pazyryk_IA packs enough kra001-related ancestry for the job.

KAZ_Pazyryk_IA
KAZ_Mereke_MBA 0.144±0.018
kra001 0.429±0.008
RUS_Sintashta_MLBA 0.378±0.026
TKM_Gonur1_BA 0.049±0.018
chisq 8.899
tail prob 0.259983
Full output

The fact that KAZ_Pazyryk_IA can be modeled with significant kra001-related ancestry isn't surprising, considering that its territory was located in Siberia. However, my model doesn't necessarily prove that the Sargat population was largely or even partly of Pazyryk origin. Indeed, N-L1026 hasn't yet appeared in any Pazyryk remains.

See also...

The Uralic cline with kra001 - no projection this time

First taste of Early Medieval DNA from the Ural region

Hungarian Conquerors were rich in Y-haplogroup N

More on the association between Uralic expansions and Y-haplogroup N

It was always going to be this way

On the association between Uralic expansions and Y-haplogroup N

The BOO people: earliest Uralic speakers in the ancient DNA record?

N-L1026 is the Y-chromosome haplogroup most closely associated with the speakers of Uralic languages. Thus far, the oldest published instances of N-L1026 are in two Siberian-like samples dating to 1473±87 calBCE from the site of Bolshoy Oleni Ostrov (BOO), located within the Arctic Circle in the Kola Peninsula, northern Russia.

So does this mean that the BOO people were Uralic speakers? I'm now thinking that it probably does, even though, as the scientists who published the BOO samples a year ago pointed out, they predate most estimates of the spread of extant Uralic languages into the Kola Peninsula (see Lamnidis et al. here).

Hundreds of ancient human samples from across Eurasia have been sequenced since last year. In fact, thousands if we count unpublished data. But only a handful of them belong to N-L1026.

Indeed, as far as I know, the next oldest instance of N-L1026 from Europe after those at BOO is still in an Iron Age sample from what is now Estonia published earlier this year as 0LS10. Of course, this individual was in all likelihood an early west Uralic (Finnic) speaker (see Saag et al. here).

Moreover, consider these comments by Murashkin et al. in regards to the BOO site (referred to as KOG in their paper, available here):

Most of the bodies had been buried in wooden, boat-shaped, lidded caskets, which looked like small boats or traditional Sámi sledges (Ru. kerezhka).

...

The morphological characteristics of the skull series of the KOG are not like those of any other ancient or modern series from the Kola Peninsula, including the Sámi people. Instead, the series shows closer biological affinities with ancient Altai Neolithic and modern, Ugric-speaking Siberian groups (Moiseyev & Khartanovich 2012). It has earlier been suggested that modern Ugric-speaking Siberians, together with Samoyeds and Volga Finnic populations, share some common morphological characteristics that indicate their common origin (Alekseyev 1974; Bunak 1956; Gokhman 1992).

...

Based on the materials from the grave field, we can argue that there were direct or indirect contacts between the inhabitants of the Kola Peninsula and southern and western Scandinavia (Murashkin & Tarasov 2013).

Thus, the BOO people may have spoken an early west Uralic language related to Sami languages. It's also possible that they are in part ancestral to the N-L1026-rich Sami people.

Another intriguing thing about these mysterious ancients is that individual BOO003 belongs to the rare mitochondrial haplogroup T2d1b1. Now, this clearly is not a lineage native to Europe or indeed any part of North Eurasia. Its ultimate source is probably West or Central Asia. So how did this pioneer polar explorer end up with such an unusual and exotic mtDNA marker, and might the answer be an important clue about the origins of the BOO people?

The most plausible explanation is that the ancestors of BOO003 were associated with the Seima-Turbino phenomenon, which stretched from the taiga zone to the oases of what is now western China along the Ob-Irtysh river system, and probably facilitated cultural, linguistic and genetic exchanges between the populations of North Eurasia and Central Asia.

In other words, considering all of the clues, it would seem that the BOO people came from some part of the Ob-Irtysh basin, which might thus be the best place to look for the population with the oldest and phylogenetically most basal N-L1026 lineages. And if we find that, then we've probably found the proto-Uralians and their homeland.

Below is a Principal Component Analysis (PCA) based on Global25 data featuring the earliest likely Uralic speakers in the ancient DNA record. It was produced with an online PCA runner freely available here. EST_IA includes the above mentioned 0LS10, while FIN_Levanluhta_IA is largely made up of Saami-related samples from western Finland. See anything interesting? Feel free to let me know about it in the comments below.

See also...

Big deal of 2019: ancient DNA confirms the link between Y-haplogroup N and Uralic expansions

It was always going to be this way

More on the association between Uralic expansions and Y-haplogroup N

Big deal of 2019: ancient DNA confirms the link between Y-haplogroup N and Uralic expansions

The academic consensus is that Indo-European languages first spread into the Baltic region from the Eastern European steppes along with the Corded Ware culture (CWC) and its people during the Late Neolithic, well before the expansion of Uralic speakers into Fennoscandia and surrounds, probably from somewhere around the Ural Mountains.

On the other hand, the views that the Uralic language family is native to Northern Europe and/or closely associated with the CWC are fringe theories usually espoused by people not familiar with the topic or, unfortunately it has to be said, mentally unstable trolls.

The likely close relationship between the CWC expansion and the early spread of Indo-European languages was discussed in several papers in recent years (for instance, see here). This year, we saw the first ancient DNA paper focusing on the transition from the Bronze Age to the Iron Age in the East Baltic, including the likely first arrival of Uralic speech in what is now Estonia.

Published in Current Biology courtesy of Saag et al., the paper showed that the genetic structure of present-day East Baltic populations largely formed in the Iron Age (see here). It was during this time, the authors revealed, that the region experienced a sudden influx of Y-chromosome haplogroup N, which is today common in many Uralic speaking populations and often referred to as a Proto-Uralic marker. Little wonder then that Saag et al. linked this genetic shift in the East Baltic to the westward migrations of early Uralic speakers.

The table below, based on data from the Saag et al. paper, surely doesn't leave much to the imagination about what happened.

Unfortunately, I have to say that the genome-wide analysis in the paper was less informative than it could have been. The authors focused their attention on rather broad genetic components, and, as a result, missed an interesting fine scale distinction between their Bronze Age and Iron Age samples. The spatial maps below, based on my Global25 data for most of the ancients from Saag et al., show what I mean. The hotter the color the higher the genetic similarity between them and present-day West Eurasian populations.

Note that the Bronze Age (Baltic_EST_BA) samples are most similar to the Baltic-speaking, and thus also Indo-European-speaking, Latvians and Lithuanians, rather than the Uralic-speaking Estonians, even though they're from burial sites in Estonia. On the other hand, the Iron Age (Baltic_EST_IA) samples show strong similarity to a wider range of populations, including Estonians and many other Uralic-speaking groups.

See also...

It was always going to be this way

Fresh off the sledge

More on the association between Uralic expansions and Y-haplogroup N