The Genetic History of France-updated version

Angela

Elite member
Messages
21,823
Reaction score
12,325
Points
113
Ethnic group
Italian
See:
Aude Saint Pierre: The Genetic History of France

https://www.biorxiv.org/content/10.1101/712497v2

"The study of the genetic structure of different countries within Europe has provided significant insights into their demographic history and their actual stratification. Although France occupies a particular location at the end of the European peninsula and at the crossroads of migration routes, few population genetic studies have been conducted so far with genome-wide data. In this study, we analyzed SNP-chip genetic data from 2 184 individuals born in France who were enrolled in two independent population cohorts. Using FineStructure, six different genetic clusters of individuals were found that were very consistent between the two cohorts. These clusters match extremely well the geography and overlap with historical and linguistic divisions of France. By modeling the relationship between genetics and geography using EEMS software, we were able to detect gene flow barriers that are similar in the two cohorts and corresponds to major French rivers or mountains. Estimations of effective population sizes using IBDNe program also revealed very similar patterns in both cohorts with a rapid increase of effective population sizes over the last 150 generations similar to what was observed in other European countries. A marked bottleneck is also consistently seen in the two datasets starting in the fourteenth century when the Black Death raged in Europe. In conclusion, by performing the first exhaustive study of the genetic structure of France, we fill a gap in the genetic studies in Europe that would be useful to medical geneticists but also historians and archeologists."


"
Gallia est omnis divisa in partes tres [commentarii de bello gallico1 45 ] was one of the earliest46 demographic description of antique France (known as Gaul). These three parts were Aquitania,47 in South West, with Garonne and the Pyrenees mountains as borders; Belgia in North West,48 following the Seine as Southern border; and finally what we know as Celtic Gaul, that spanned49 from the Atlantic Ocean to the Rhine River and Alps. A fourth part of the present-day French50 territory, already part of Romanized territories at this time, was Gallia Transalpina, a strip of51 lands from Italy to Iberia, with Alps and Cevennes mountains as northern border.52 The area that was to be modern France was subject to successive population migrations:53 Western Hunter-Gatherers (15 kya), Neolithic farmers (7 kya) and later steppe Enolithic Agepopulations2,3 54 , Celtic expansion, integration in Roman empire, Barbarian Great migrations,55 whose demographical importance remains to be assessed."

"In this paper, we applied haplotype-based methods that have been shown to provide higherresolution than allele-based approaches13 105 to investigate the pattern of fine-scale population106 stratification in France. We used two independent cohorts, 3C and SUVIMAX with more than107 2 000 individuals whose birthplace covered continental France and genotyped at the genome108 wide level, to assess the genetic structure of the French population and draw inferences on the109 demographic history."

I understand the constraints they're working on in France, so they had to use medical papers which only had location of birth, but how accurate is all of this necessarily without proof all four grandparents, at least, were born in that departement?

"In both datasets, the coarsest level of genetic differentiation (i.e. the assignment into two267 clusters) separates the south-western regions from the rest of France."

"
The clusters are270 geographically stratified and were assigned labels to reflect geographic origin: the South-West271 SW for the dark-red cluster, the South (SO) for the orange cluster, the Centre (CTR) for the272 yellow cluster, the North-West (NW) for the pink cluster, the North (NO) for the blue cluster273 and the South-East (SE) for the cyan cluster. In each dataset, one cluster (labelled “Others” and274 coloured in red) included individuals geographically dispersed over France."

"The broad-scale genetic structure of France in six clusters strikingly aligns with two major287 rivers of France, “La Garonne” and “La Loire” (Figure 2). At a finer-scale, the “Adour” river288 partition the SW to the SO cluster in the 3C dataset. The mean FST between clusters inferred by289 FineSTRUCTURE (Table S1 and S2) are small, confirming subtle differentiation. In both290 datasets, the strongest differentiation is between the SW cluster and all other regions. These FST291 values vary from 0.0016 with the SO cluster to 0.004 with the NW cluster in the 3C dataset and292 from 0.0009 with the CTR cluster to 0.0019 with the NW cluster in SU.VI.MAX."

In Admixture: "In both datasets, we313 identified a genetic barrier around the south-west region (Figure 3). This barrier mirrors the314 first division in the FineSTRUCTURE. The plots also reveal a gene flow barrier around315 Bretagne in the North-West and along the Loire River, which covers the separation of the North cluster. Finally, another barrier is also present on the South-East side that roughly corresponds317 to the location of the Alp Mountains at the border with Northern Italy."

"As expected, the British heritage was more347 marked in the north than in the south of France where, instead, the contribution from southern348 Europe was stronger. The overall contribution from British heritage was substantially higher in349 the NW than in the NO cluster (76% vs 64% in the 3C and 72% vs 63% in SU.VI.MAX). TSI350 was contributing to the SE cluster while IBS was mainly contributing to the SW cluster, which351 again was very coherent with the geographic places of birth of individuals."

The above is a bit overstated, in my opinion. It's not necessarily or even likely that British people settled all of northwestern France, although I'm aware of some migration from there. These supposed "contributions" instead, imo, more likely reflect differing amounts of gene flow from much older migrations, i.e. the same group went to both Britain and Brittany, and isolation means that perhaps southwest England and northwest France received fewer subsequent migrations, which means that they retain more of that original similarity.

"The French genomes were found to map at their expected position in between Nordic (British369 and CEU), Italian and Spanish genomes from 1000 genomes project."

We've been saying the following here for a long time, but nice to see it confirmed finally.

"An important division separates Northern from Southern France. It may coincide with the von375 Wartburg line, which divides France into “Langue d’Oïl” part (influenced by Germanic376 speaking) and “Langue d’Oc” part (closer to Roman speaking) – Figure S15. This border has377 changed through centuries and our North-South limit is close to the limit as it was estimated inthe IXth century34,35 378 . This border also follows the Loire River, which has long been a political379 and cultural border between kingdoms/counties in the North and in the South (Figure 3).380 Regions with strong cultural particularities tend to separate. This is for example the case for381 Aquitaine in the South-West which duchy has long represented a civilization on its own. The382 Brittany region is also detected as a separate entity in both datasets. This could be explained383 both by its position at the end of the continent where it forms a peninsula and, by its history384 since Brittany has been an independent political entity (Kingdom and, later, duchy of Bretagne),with stable borders, for a long time."

As I said, I think the following is only part of the explanation:
"we observe that Brittany is substantially closer to402 British Isles population than North of France, in spite of both being equally geographically403 close. Migration of Britons in what was at the time Armorica (and is now Brittany) may explain404 this closeness. These migrations may have been quite constant during centuries although a twowaves model is generally assumed. A first wave would have occurred in the Xth 405 century when406 soldiers from British Isles were sent to Armorica whereas the second wave consisted of Britonsescaping the Anglo-Saxon invasions37 407 . Additional analyses, on larger datasets may be required408 to discriminate between these various models."

"We also observe, in most cases, a depression during a period spanning from 12412 to 22 generations ago. This may correspond to a period spanning from 1 300 to 1 700. Indeed,413 this period was characterized by a deep depression in population size due to a long series of414 plague events. While the population size in kingdom of France was estimated to be 20 Million...in 1 348, it dropped down to 12 Million in 1 400, followed by an uneven trajectory to recoverthe 20 Million at the end of Louis XIVth reign (1 715)38 416 .417 However, the decrease we observe in the genetic data does seem to affect mainly the Northern418 part of France, and for instance is mainly observed in the NO cluster. We see no reason for this419 trend based on historical records (Figure S16) except perhaps the last plague epidemics in 1 666420 - 1 670 that was limited to the North of France."

This is all very interesting, and in general terms probably would be borne out by a proper study based on people with all four grandparents from one area, but even though France has perhaps not experienced the massive internal migration which has affected modern Italy, I think the fst differences might be bigger had the proper samples been used.

So, the way I feel so at home in southeastern France has a physical as well as cultural basis perhaps. :

4w4kkwk.png


txOZxzw.png


bqYEM0H.png
 
See:
So, the way I feel so at home in southeastern France has a physical as well as cultural basis perhaps. :

I was chatting with an Italian user on Quora once, and he told me a cute anecdote: once he visited Sicily as a child, and somebody asked him if he was French. When he replied he was Italian, from Piedmont, the guy replied: "Ah ok, French". :-D
 
I was chatting with an Italian user on Quora once, and he told me a cute anecdote: once he visited Sicily as a child, and somebody asked him if he was French. When he replied he was Italian, from Piedmont, the guy replied: "Ah ok, French". :-D

I'm used to that sort of thing from Southern Italians. :) When my husband is bored and wants to see some sparks fly he talks about how "German" my father's family is...or he'll see an old picture of one of my cousins and say, yeah, doesn't look Italian. :(

He also brings up how close we are to the French border. Fine with me; like my father I'm a bit of a Francophile. :)

The fact that he gets astronomically high "Italian" scores on 23andme doesn't help (I'm only 72% Italian on there, although that's better than the 55% on ancestry). I keep on telling him it's just because they're using Southern Italians in their reference sample, but he prefers not to absorb it. :)
 
Looks like a La Loire makes a clear boundary, that's interesting.
 
I'm used to that sort of thing from Southern Italians. :) When my husband is bored and wants to see some sparks fly he talks about how "German" my father's family is...or he'll see an old picture of one of my cousins and say, yeah, doesn't look Italian. :(

He also brings up how close we are to the French border. Fine with me; like my father I'm a bit of a Francophile. :)

The fact that he gets astronomically high "Italian" scores on 23andme doesn't help (I'm only 72% Italian on there, although that's better than the 55% on ancestry). I keep on telling him it's just because they're using Southern Italians in their reference sample, but he prefers not to absorb it. :)

In France, when you say Italian, people usually think of the Neapolitan or even the Sicilian stereotype. Blame The Godfather? And also the fact that a lof of Italian immigrants (at least in France) came from the South of Italy.
What other percentages do you get on 23andme, besides your 72% Italian? Have you heard of Ethnogene? I've just tried it, and I must say their ethnicity report is the most accurate that I've got so far - they have +200 population references, and very detailed regional breakdowns. I'm sure you'd get more than 72% Italian with them.
This is my report, and it's really spot on:

Capture-d-cran-2019-08-09-19-46-08.png
 
In France, when you say Italian, people usually think of the Neapolitan or even the Sicilian stereotype. Blame The Godfather? And also the fact that a lof of Italian immigrants (at least in France) came from the South of Italy.
What other percentages do you get on 23andme, besides your 72% Italian? Have you heard of Ethnogene? I've just tried it, and I must say their ethnicity report is the most accurate that I've got so far - they have +200 population references, and very detailed regional breakdowns. I'm sure you'd get more than 72% Italian with them.
This is my report, and it's really spot on:

Capture-d-cran-2019-08-09-19-46-08.png

I've been downgraded! :)

Now I'm only 62% Italian, approximately 10% general Southern European, and the rest mostly Northwest European.

Never heard of this new company. I'll check it out.
 
Bummer! I've been downgraded too: now only 45.5% Italian, for a total of 64% Southern Europe, 23.2% NW Europe and 3.6% Western Asian/NA. But for me it's normal that they should increase my NW Europe and decrease my Italian. Do check out Ethnogene, they work with your raw data; for 30$, they give a shockingly accurate estimate of your recent ancestry. People are generally quite pleased with their reports. Check your results before showing them to your hubby ;)
 
I think one mistake they made in that study was that they checked how samples cluster comparing only regional French samples to each other, without adding non-French (Belgian, German, Spanish, Italian etc.) neighbour populations for comparison.

On the other hand, my map [see below] is based on clustering of entire Europe.

And in my opinion, for example Ile de France is still closer to Auvergne than the former is to Germany or to Belgium.

This is why I obtained a Central French cluster, all regions of which are closer to each other than to Non-French people.

Obviously, what clusters you will obtain depends also on what populations are used in the comparison. So if you make clustering with only French regional samples (without including Iberians, Germans, Belgians, etc.), clusters will be different than when you use both French and Non-French.

In other words - when you use just French samples, Ile-de-France might be closer to Hauts-de-France than to Auvergne & Centre. But when you add Germans, Belgians, etc. for comparison - then Hauts-de-France may "jump" to the Belgian-dominated cluster, while most samples from Ile-de-France may "jump" to the same cluster together with Auvergne & Centre.

=====

Here is what I have obtained when doing clustering with many European populations (not just French regional samples alone):

Important Note:

In Provence, some samples (especially around Nice) are very close to North Italy, but more samples are closer to Iberia/Baleares.

This is why I painted Provence with mixed Orange-Blue instead of mixed Yellow-Blue, but the latter choice would be quite OK too.

wyb6uqJ.png


^^^
Attached image of my clustering exercise, all of these regional averages group together with each other as "Central French":

(I'm showing only the fragment with "Central French" cluster here)

LqOXM2B.png


Poitou-Charentes is mixed for example Vendee Department is Central French but e.g. Charente-Maritime is Aquitanian-shifted.
 
Last edited:
^^^
If there are any French users here who would like to be compared to those regional averages and see how they cluster, PM me about it.

=====

Or if you have French GEDmatch kits, I'm also interested.
 
^^^
Also for example if I examine individual samples one by one instead of averages, I see that:

In case of Provence Marseille (for example):

9 individuals are close to Central French (of which two have Northern shift, one Iberian shift, one Italian shift)
3 individuals are intermediate between North Italy and Central France
2 individuals are intermediate between Iberians and Central France
2 individuals are intermediate between Iberians and North Italy
2 individuals are closest to North-West Italy
2 individuals are closest to Iberia/Baleares

One is a very northern outlier, obviously must have a lot of ancestry from Northern France.

Among samples from Provence Nice, it looks a bit different (and a bit closer to North Italy).
 
I think one mistake they made in that study was that they checked how samples cluster comparing only regional French samples to each other, without adding non-French (Belgian, German, Spanish, Italian etc.) neighbour populations for comparison.

On the other hand, my map [see below] is based on clustering of entire Europe.

And in my opinion, for example Ile de France is still closer to Auvergne than the former is to Germany or to Belgium.

This is why I obtained a Central French cluster, all regions of which are closer to each other than to Non-French people.

Obviously, what clusters you will obtain depends also on what populations are used in the comparison. So if you make clustering with only French regional samples (without including Iberians, Germans, Belgians, etc.), clusters will be different than when you use both French and Non-French.

In other words - when you use just French samples, Ile-de-France might be closer to Hauts-de-France than to Auvergne & Centre. But when you add Germans, Belgians, etc. for comparison - then Hauts-de-France may "jump" to the Belgian-dominated cluster, while most samples from Ile-de-France may "jump" to the same cluster together with Auvergne & Centre.

=====

Here is what I have obtained when doing clustering with many European populations (not just French regional samples alone):

Important Note:

In Provence, some samples (especially around Nice) are very close to North Italy, but more samples are closer to Iberia/Baleares.

This is why I painted Provence with mixed Orange-Blue instead of mixed Yellow-Blue, but the latter choice would be quite OK too.

wyb6uqJ.png


^^^
Attached image of my clustering exercise, all of these regional averages group together with each other as "Central French":

(I'm showing only the fragment with "Central French" cluster here)

LqOXM2B.png


Poitou-Charentes is mixed for example Vendee Department is Central French but e.g. Charente-Maritime is Aquitanian-shifted.

The map you posted makes a lot of sense. In general, the Iberians are closer to the central French than the Italians - I believe it is due to the higher amount of WHG. But it makes sense for Italians in the northwest to be close to the French in the southeast.
 
I think one mistake they made in that study was that they checked how samples cluster comparing only regional French samples to each other, without adding non-French (Belgian, German, Spanish, Italian etc.) neighbour populations for comparison.

On the other hand, my map [see below] is based on clustering of entire Europe.

And in my opinion, for example Ile de France is still closer to Auvergne than the former is to Germany or to Belgium.

This is why I obtained a Central French cluster, all regions of which are closer to each other than to Non-French people.

Obviously, what clusters you will obtain depends also on what populations are used in the comparison. So if you make clustering with only French regional samples (without including Iberians, Germans, Belgians, etc.), clusters will be different than when you use both French and Non-French.

In other words - when you use just French samples, Ile-de-France might be closer to Hauts-de-France than to Auvergne & Centre. But when you add Germans, Belgians, etc. for comparison - then Hauts-de-France may "jump" to the Belgian-dominated cluster, while most samples from Ile-de-France may "jump" to the same cluster together with Auvergne & Centre.

=====

Here is what I have obtained when doing clustering with many European populations (not just French regional samples alone):

Important Note:

In Provence, some samples (especially around Nice) are very close to North Italy, but more samples are closer to Iberia/Baleares.

This is why I painted Provence with mixed Orange-Blue instead of mixed Yellow-Blue, but the latter choice would be quite OK too.

wyb6uqJ.png


^^^
Attached image of my clustering exercise, all of these regional averages group together with each other as "Central French":

(I'm showing only the fragment with "Central French" cluster here)

LqOXM2B.png


Poitou-Charentes is mixed for example Vendee Department is Central French but e.g. Charente-Maritime is Aquitanian-shifted.

There may be something to that. As I mentioned on another thread, I always thought that using the Lyon sample to stand for all the French was misleading. You just had to be familiar with Provence and places like Gascony to know that. In addition, that sample was of university students, so the ancestry in terms of area might have been rather diverse.

As you may know, there are a lot of people from Nice (Nizza) and even further west toward Marseilles and beyond who not that long ago were Italian speakers, and many of the surnames are Italian. Nizza was the birthplace of Garibaldi. Likewise, there are some French surnames in Liguria, but fewer.

Fwiw, I have Ligurian ancestry, and my closest Spanish region is usually Baleares. The people of the Baleares were conquered over and over again, but much of their population was sold into slavery and the islands were settled by the Kingdom of Aragon. Catalan is their "native" language.

The contacts between Liguria, the Mediterranean coastal areas of France, and down into Catalonia go back to the Mesolithic and never really stopped. It was for most of history easier to hug the shore in a boat than to venture very far inland.
 

Nice Map ! It clearly shows the Celtic (European and Atlantic), Mediterranean and Germanic parts of France
!
I am From French Flanders.
 

This thread has been viewed 9016 times.

Back
Top