PROJET ANCESTRA

[poilus]

page 226
https://www.isbarch.org/assets/documents...t_Book.pdf

 

[okshtunas]

page 226
https://www.isbarch.org/assets/documents...t_Book.pdf

 

Any Y-DNA?

[Sephesakueu]

E1b and I2 in neolithic France, the video soesn't show iron age Y-DNA, it seems.

[poilus]

page 226
https://www.isbarch.org/assets/documents...t_Book.pdf

 

Any Y-DNA?

Could not find any other information and still awiting for the publication.

[Megalophias]

Nos ancetres les Gaulois

[GHurier]

Looking at clades cross-correlation of the Y-tree is quite interesting for France.

Here is the R-M269 clade-correlation for France vs the rest of the world (using YFULL tree v11.05 data)



How to read :
-The blue curve shows the R-M269 diversity outside of France as a function of time (diversity is a function of time, when a clade diversifies a region didn't necessarly host all the subclades diversifying from the parent clade, and thus the total amount of diversity decrease), "FRA bar" means "anything but FRA".
-The red curve shows the R-M269 diversity observed in modern France.
-The black curve shows the amount of shared diversity between France and the rest of the world as a function of time.

By construction, such curve (when diversity-weighted) can only decrease with time.
When clades are "decoupling" the black curve drop significantly. Such signal indicates a demographic event implying that the two populations (here "FRA" and "FRA bar") are not mixing together anymore (while it is usually caused by geographical reasons/migrations, it is also possible that such thing might occur for "social-reasons").
Importantly, this probe tells us when the sub-populations are "separating" (not "where" they are when separating) but is less clear regarding the "direction" of the population movement.


Here, we observe that the major effect for France is around 1000 BCE, which has to be linked with the Gauls ethnogenesis.
Some decoupling keeps going after 1000 BCE, it indicates that movement of peoples toward or outside of France were still significantly and regularly occuring at that time.

Importantly, if we "zoom" on some specific subclades/countries it can be shown that clades decoupling already starts as early as Bell-Beaker times for R-M269.

Around 0 CE, the lack of significant clade-decoupling speaks volume about the limited impact that Roman DNA had on France inhabitants.
Similarly, 1st millenium CE shows some limited decoupling likely to be related with the Germanic invasions (the Franks).
While the impact of these Germani invasions is visible on the Y-lineage, it remains fairly small compared to the main 1000 BCE event.

A legitimate question about this figure, would be to know if other haplogroups are telling the same story ? Main haplogroups related to Romans are usually considered to be E-V13 and J2. Here are the same curves for these two haplogroups :





The two clades are showing a similar main decoupling around 1000 BCE and no decoupling around Roman times, indicating again that the Roman genetic influence on Gaule (traced here using modern French frontiers) was limited regarding Y-lineages.

Thus the results of the study presented in this thread is consistent with the analysis of the Y-lineages from YFULL tree.
The key part to understand about "Gauls" is the ~1000 BCE epoch to identify if the signal we see in the above curves is tracing clades expanding from France or migrants arriving in France (in that case it would be from central Europe).

To test that, a carefull (and deep) sampling of France MBA-LBA (and France EIA optionnally) would be helpfull.

[ChrisR]

How to read :
-The blue curve shows the ... diversity outside of France as a function of time (diversity is a function of time, when a clade diversifies a region didn't necessarly host all the subclades diversifying from the parent clade, and thus the total amount of diversity decrease), "FRA bar" means "anything but FRA".
-The red curve shows the ... diversity observed in modern France.
-The black curve shows the amount of shared diversity between France and the rest of the world as a function of time.

By construction, such curve (when diversity-weighted) can only decrease with time.
When clades are "decoupling" the black curve drop significantly. Such signal indicates a demographic event implying that the two populations (here "FRA" and "FRA bar") are not mixing together anymore (while it is usually caused by geographical reasons/migrations, it is also possible that such thing might occur for "social-reasons").
Importantly, this probe tells us when the sub-populations are "separating" (not "where" they are when separating) but is less clear regarding the "direction" of the population movement.

Here, we observe that the major effect for France is around 1000 BCE, which has to be linked with the Gauls ethnogenesis.
Some decoupling keeps going after 1000 BCE, it indicates that movement of peoples toward or outside of France were still significantly and regularly occuring at that time.

Importantly, if we "zoom" on some specific subclades/countries it can be shown that clades decoupling already starts as early as Bell-Beaker times for R-M269.

Around 0 CE, the lack of significant clade-decoupling speaks volume about the limited impact that Roman DNA had on France inhabitants.
Similarly, 1st millenium CE shows some limited decoupling likely to be related with the Germanic invasions (the Franks).
While the impact of these Germani invasions is visible on the Y-lineage, it remains fairly small compared to the main 1000 BCE event.

A legitimate question about this figure, would be to know if other haplogroups are telling the same story ? Main haplogroups related to Romans are usually considered to be E-V13 and J2. Here are the same curves for these two haplogroups :
The two clades are showing a similar main decoupling around 1000 BCE and no decoupling around Roman times, indicating again that the Roman genetic influence on Gaule (traced here using modern French frontiers) was limited regarding Y-lineages.

Thus the results of the study presented in this thread is consistent with the analysis of the Y-lineages from YFULL tree.
The key part to understand about "Gauls" is the ~1000 BCE epoch to identify if the signal we see in the above curves is tracing clades expanding from France or migrants arriving in France (in that case it would be from central Europe).

To test that, a carefull (and deep) sampling of France MBA-LBA (and France EIA optionnally) would be helpfull.

I still need to fully understand the analysis/conclusions, but I find this concept interesting.
As I'm interested in some Alpine lineages which came from the Upper Rhine area and are at least there since middle/late IA but possibly are there since LBA or even MBA they fall into the " Gauls ethnogenesis".
May I ask why the black curve for E-V13 does not continue down to the X-Axis and why for J2 the graph does not continue to 1000 AD/CE?
I think this would be interesting (for me).

[GHurier]

I still need to fully understand the analysis/conclusions, but I find this concept interesting.
As I'm interested in some Alpine lineages which came from the Upper Rhine area and are at least there since middle/late IA but possibly are there since LBA or even MBA they fall into the " Gauls ethnogenesis".
May I ask why the black curve for E-V13 does not continue down to the X-Axis and why for J2 the graph does not continue to 1000 AD/CE?
I think this would be interesting (for me).

If you have an idea of the "starting region" and the "destination region" of a given clade, I can produce you a plot for the concerned regions and the concerned haplogroup. Maybe we will see some interesting stuff.

About the curves being "cutted", it is because I used a log-scale (if not, due to diversity-weighting, it might be hard to see stuff occuring significantly after TMRCA and only involving a limited amount of subclades).
The software I use for display fails to plot Zero-values when in log-scale. I could solve that by replacing 0 by i.e., 10^-10, but when the curves are "vanishing" it bascially indicates that there is no coupling anymore between the two regions for the concerned clades, which is a valuable information.
In the case of France, some degree of correlation clearly exists in real life at later dates for E-V13 and J2, but these two lineages are not that common in France, and for some "legal" reasons the sampling depth of modern peoples is just garbage ...
Thus, the problem is the amount of available samples.


To explain a bit more the rationals behind this approach:
1- It start from the hypothesis that populations sharing an ancestral source population will also share Y-lineages. Whereas some lineages can be lost by drift, if a source populations carries a significant amount of lineages, it is highly unlikely that all those lineages will get lost by drift.
2- Then, when the source population is "spliting" (to form two daughter populations), some brothers/cousins/near-relatives will also be separated. It creates a correlation of clades between the two populations.

Therefore, let imagine that you have a population (pop-A) living in a regions R1 that are mainly populated by some Y-haplogroup lineages (Y1). Then, at a given time "t0", pop-A is sourcing two other populations, pop-B and pop-C, living respectively in regions R2 and R3.

Under the assumptions (1) and (2) above, if you compute the clade cross-correlation for the haplogroup Y1 between regions R2 and R3 you'll see that Y1-descendents with TMRCA > t0 will often appear in both regions R2 and R3, whereas Y1-descendents with TMRCA < t0 will only appear in R2 or R3, but almost never in both.

Translated into cross-correlation signal, it means that before "t0" you'll see a correlation of clades between R2 and R3 (positive signal, presented as a black curve on my plots), whereas after "t0" you won't see any correlation (or a drop in the correlation rate).
In a real cases, we deal with many populations movements across history and we only see the "integrated" history on the actual Y-tree, therefore many movements are stacked over each other on the curves, thats why the correlation didn't disapear in a single event.

I compute the curves I presented before, D_X, as follow (I already appologise for the following equations in a dirty format) :

D_X = \sum_{clades with presence in X} W_{clade}

Where "X" can be geographical regions "R2", "R3" or "R2 & R3" if we use the notations defined above.
Here "W_{clade}" is a diversity-weight asigned to each clade such that at a given time "t1",

\sum_{clades with TMRCA < T1 & TF > T1} W_{clade} = 1

\sum_{Y's direct descendent clades} W_{clade} = W_{Y}

Basal branches are considered as their own individual clades with TMRCA = 0 ybp.

*To avoid weird behavior it is also helpfull to "prune" the Y-tree from any branch that have no presence in either R2 or R3. In the case I showed for "FRA" and "FRA-bar", the union of the two regions is covering the whole world, therefore no branch was pruned.

**Also, to avoid a too big impact of the deep structure of Y-tree on the diversity-weights, it is helpfull to reduce the time resolution of the Y-tree. This can be achieved by connecting to a single node cascading diversification events happening in a short amount of time. It significantly reduces the noise on diversity estimator aswell, at the cost of the hypothesis that two diversification events occurring successively in a short time window are tracing a single population migration/expansion.
Ex : clade Y1 diversifies at "t0", one of the subclade of Y1, Y2, also diversifies at "t0", a suclade of Y2, Y3, diversifies at "t0+50 y", and another subclade of Y2, Y4, diversifies at "t0 + 400 y". When degrading the time resolution, all Y2 and Y3 childs will be considered as Y1 direct childs for diversity weights computation whereas Y4 childs are too far in time and won't be merged with Y1-diversification.


This probe is not without limitations, it is hard to get accurate things when we look at the last centuries (due to not enough sampling depth of the modern population and the growth of the world population), but for IA (or before) migrations/expansions it works amazingly well.

Do not hesitate if you have any questions.

[ChrisR]

If you have an idea of the "starting region" and the "destination region" of a given clade, I can produce you a plot for the concerned regions and the concerned haplogroup. Maybe we will see some interesting stuff.

About the curves being "cutted", it is because I used a log-scale (if not, due to diversity-weighting, it might be hard to see stuff occuring significantly after TMRCA and only involving a limited amount of subclades).
The software I use for display fails to plot Zero-values when in log-scale. I could solve that by replacing 0 by i.e., 10^-10, but when the curves are "vanishing" it bascially indicates that there is no coupling anymore between the two regions for the concerned clades, which is a valuable information.
In the case of France, some degree of correlation clearly exists in real life at later dates for E-V13 and J2, but these two lineages are not that common in France, and for some "legal" reasons the sampling depth of modern peoples is just garbage ...

Thank you very much for the kind extensive explanation. May I ask which software you use? My guess is R and some custom script to pull the YFull Tree.
Regarding the poor sampling depth and Gaul source area might it be an idea to to look for the slightly older pop-A and region R1?
How difficult is it to diversify to regions from YFull? I ask since I think it would be interesting to have the Core Hallstatt area (LBA) analyzed in this way:

• From France: Bourgogne-Franche-Comté, Grand Est
  
• From Germany: Baden-Württemberg, Bavaria,
  
• Switzerland, Austria, Czech R.
  

Not sure if from Western Hungary counties should also be included, similar question for Hesse, Rhineland-Palatinate, Saarland or even more from Upper Rhine?
Maybe also limiting to R1b-P312 to exclude the more "Northern/Germanic" R1b-U106 with big late influence could result in further insights?
I think doing this also for E-V13 is certainly interesting given the North Balkanic expansion center. I'm also interested in J2 for this area but will expect less surprises except for the Upper Rhine area, as said there seems to be an early "melting pot" even if with low frequencies from more exotic origins.

[poilus]

If you have an idea of the "starting region" and the "destination region" of a given clade, I can produce you a plot for the concerned regions and the concerned haplogroup. Maybe we will see some interesting stuff.
About the curves being "cutted", it is because I used a log-scale (if not, due to diversity-weighting, it might be hard to see stuff occuring significantly after TMRCA and only involving a limited amount of subclades).
The software I use for display fails to plot Zero-values when in log-scale. I could solve that by replacing 0 by i.e., 10^-10, but when the curves are "vanishing" it bascially indicates that there is no coupling anymore between the two regions for the concerned clades, which is a valuable information.
In the case of France, some degree of correlation clearly exists in real life at later dates for E-V13 and J2, but these two lineages are not that common in France, and for some "legal" reasons the sampling depth of modern peoples is just garbage ...

Thank you very much for the kind extensive explanation. May I ask which software you use? My guess is R and some custom script to pull the YFull Tree.
Regarding the poor sampling depth and Gaul source area might it be an idea to to look for the slightly older pop-A and region R1?
How difficult is it to diversify to regions from YFull? I ask since I think it would be interesting to have the Core Hallstatt area (LBA) analyzed in this way:

• From France: Bourgogne-Franche-Comté, Grand Est
  
• From Germany: Baden-Württemberg, Bavaria,
  
• Switzerland, Austria, Czech R.
  

Not sure if from Western Hungary counties should also be included, similar question for Hesse, Rhineland-Palatinate, Saarland or even more from Upper Rhine?
Maybe also limiting to R1b-P312 to exclude the more "Northern/Germanic" R1b-U106 with big late influence could result in further insights?
I think doing this also for E-V13 is certainly interesting given the North Balkanic expansion center. I'm also interested in J2 for this area but will expect less surprises except for the Upper Rhine area, as said there seems to be an early "melting pot" even if with low frequencies from more exotic origins.

 
 


EAA 2023 pdf abstract. Wait and see.

[NewEnglander]

Were these new samples? Any Frankish or otherwise migration era?

[GHurier]

Thank you very much for the kind extensive explanation. May I ask which software you use? My guess is R and some custom script to pull the YFull Tree.
Regarding the poor sampling depth and Gaul source area might it be an idea to to look for the slightly older pop-A and region R1?
How difficult is it to diversify to regions from YFull? I ask since I think it would be interesting to have the Core Hallstatt area (LBA) analyzed in this way:

• From France: Bourgogne-Franche-Comté, Grand Est
  
• From Germany: Baden-Württemberg, Bavaria,
  
• Switzerland, Austria, Czech R.
  

Not sure if from Western Hungary counties should also be included, similar question for Hesse, Rhineland-Palatinate, Saarland or even more from Upper Rhine?
Maybe also limiting to R1b-P312 to exclude the more "Northern/Germanic" R1b-U106 with big late influence could result in further insights?
I think doing this also for E-V13 is certainly interesting given the North Balkanic expansion center. I'm also interested in J2 for this area but will expect less surprises except for the Upper Rhine area, as said there seems to be an early "melting pot" even if with low frequencies from more exotic origins.

I'm not using R, I'm working in IDL ... "remants" from my field of research.
I could do it in Python, but I'm way more efficient in IDL.

Fine regions splitting for YFULL might get tricky and depends on what information the users provided.
For France, YFULL allows to provide departement numbers, but not all samples are localised with such precision, which decrease the effective sampling depth. Usually I prefer to stay at a "country" level (but then for big countries is starts to be a bit problematic).
For France, we clearly know that some stuff are significantly different for the North of the Loire, inbetween Loire and Garonne, Basque, and South-Eastern part around Nice.

Let look at R-P312, E-V13, J-M410 for Germany/Austria/Czechia :





Interestingly, R-P312 shows signal linked with Hallstatt period, whereas it didn't seems to have significantly contributed to later Germanic tribes expansion when Roman empire collapsed.
Interestingly, Germany-Austria-Czechia didn't seems to have significant "exclusive" R-P312 diversity until Iron Age (still, some decoupling occurs smoothly during BA).
More test is needed of course, but it would go in the direction that most R-P312 arrived in this area by Iron-Age.

E-V13 here is fairly interesting, as it stack multiple events,
The Hallstatt related movements, concerning E-V13 central-european subclades, then the Roman-Era period which shows a big decoupling related to Thracian-E-V13, and finally some decoupling around Roman-Empire collapse.

J2a mainly shows signal around ~200 BCE, it could be related with J2a absorbed in Italy by Celts and then retreating with Celts across the Alps around that time.

Then, looking at the relation of the three regions you proposed : FRA, DEU, CZE+AUT+CHE, for R-P312





The main decoupling for these three regions is arond 1000 BCE.
France and Germany interestingly are showing significant BB-related interaction, before contacts get significantly reduced by 1900/1800 BCE ("flattening" of the decorrelation curve).

[J Man]

Indeed it will be interesting to see which Y-DNA haplogroups show up from the Roman era samples. Hopefully some J2a has been found among them.