Hi Ed,

I don't know that there's been a lot of talk specifically about Colmar 239, but if you're basing your origin estimates off Globetrekker, then you're in for a bad time. Globetrekker effectively brings Family Tree DNA up-to-date with the efforts that Rob Spencer, Hunter Provyn and others have been doing for a few years now, but no-one really comes close to an accurate solution because the biases in the data still haven't been correctly accounted for, and the ancient DNA (in my opinion) hasn't been given the right weighting in the calculations.

The problem with assigning origins based on ancient DNA is the same as assigning origins based on modern testers: how much does one sample really tell you? The key here is contemporaneity. A sample like Ploti?tě nad Labem 1, who lived within a few generations of the R-U106 MRCA, says a lot about where R-U106 formed. Colmar 239, living 1000 years or so after the R-A10645 MRCA, says much less about where R-A10645 formed. We also have to be careful with our nomenclature: an ancient DNA sample may not be tested for all the SNPs in a haplogroup, therefore may descend from an intermediate haplogroup that is now either untested or extinct; similarly the sample may be positive for downstream sub-clades, for either not positive for all SNPs, or not tested for any.

If we assume a front speed of a between one and a few km/year for human migrations (https://www.pnas.org/doi/10.1073/pnas.1920051117), we can presume that R-U106 was founded within a few hundred km of where Ploti?tě nad Labem 1 was buried. This puts the origin of R-U106 in the vicinity of Bohemia, although doesn't allow us to definitively narrow it down to a specific country. If we do the same to Colmar 239, we can say that R-A10645 was founded somewhere within between one and a few thousand km of the La Tene culture, i.e., somewhere within Europe. That's not very constraining.

In both cases, we can say that at least some of R-U106 passed through the Bohemian part of the Corded Ware Culture, and that some of R-A10645 passed through the La Tene culture. In all likelihood, proportionally more of R-U106 probably passed through the Bohemian CWC because there had been much less time for R-U106 to diverge into different regions and cultures by that point. So, not only is Ploti?tě nad Labem 1 more constraining geographically regarding the origin of R-U106 than Colmar 239 is of R-A10645, but it also says more specifically about what cultures our contemporary early R-U106 ancestors were doing at the time. By contrast, by 600 BC, R-A10645 could have been spread to many different cultures in many different places, and we only sample one.

To put some numbers on this, imagine a population that diffuses out at a constant rate over time. It might cover 1000 square km after 100 years, 4000 square km after 200 years, 9000 square km after 300 years, and 2.5 million square km (1/4 of Europe) after 5000 years, etc. Imagine a tester in a 5000-year-old haplogroup like R-U106 today. We can draw a circle around him of 2.5 million square km and say that there's a 68% chance that the origin of R-U106 lies within that circle. If we have ancient DNA that's from 1000 years after R-U106 split, then we can draw a circle of 100,000 square km around the burial, thus that ancient DNA is 25 times more precise, and thus "worth" 25 modern testers today. For Ploti?tě nad Labem 1, we are dealing with a sample probably about 100 years or so after R-U106 split, thus the circle is 2500 times smaller, and that ancient DNA is "worth" 2500 modern testers in terms of defining origins.

Of course, populations move and spread and haven't expanded at a constant size, and all sorts of other confounding factors so, actually, each ancient DNA sample is "worth" much more than this, because it strips out all of these systematic trends. But the comparative "worth" of Ploti?tě nad Labem 1 is still more than 100 times that of Colmar 239 by this argument, and Colmar 239 on its own probably doesn't say much more about the origins of R-A10645 than the existing 42 FTDNA testers already did.

The take-home point here, then, is that ancient DNA needs to assigned to a contemporary haplogroup (not a much more ancient one) before it says anything about our ancestors, and it is only samples really close to the TMRCA of the haplogroup that can be so incredibly defining in terms of origins.

Cheers,

Iain.

Hi Randy,

There are plenty people who will give you a more authoritative answer on this, however I'll give it my best shot.

The Germanic language family and, by extension, the Germanic people, can be split into three branches: Northern, Western and Eastern. The Northern Germanic group roughly co-incides with the earliest Germanic people around the shores of the Black Sea and Scandinavia, namely the Norwegian, Swedish, Danish and descendant peoples in the Faeroes, Iceland and Greenland, and parts of Finland. The Western Germanic group comprises modern Germans, but also the countries west of them, including the Dutch, Flemish and English languages. The Eastern Germanic group comprises language families in eastern Europe, which are now extinct but survive in Gothic texts. The Western Germanic group is often sub-divided.

How far back we can place the origins of a distinct Germanic people is debateable. Most references I've seen typically put this definition around 700-500 BC, and the culture didn't fragment into these later branches until the Germanic peoples started expanding, both south-west into Celtic central Europe as the Germanic tribes the Romans knew, and eastwards into what ultimately became the Gothic migrations down towards the Black Sea and into Rome. These migrations and expansions were a gradual process across many centuries, and subsequent migrations (e.g. within Prussia) have erased many differences that there once were.

Part of what we can do with genetic genealogy is use the TMRCA estimates that we create to merge with what is known from an archaeological understanding of this period of history, and help chart the migrations of people that we see evidenced in the archaeology. As you may anticipate, this isn't a straight-forward process.

Cheers,

Iain.

Hi Mike,

For R-FGC910, see R-Z7 and R-FGC902 on this post:

/g/R1b-U106/message/5759

The period of history is very difficult to make predictions about. It's too far back in history for historical sources or modern distributions to accurately reflect origins without complex interpretation, and it's too far forward in history for the known origins of R-U106 to be instructive. Apart from some stand-out cases, we need better modelling in this period before we can be very confident in our results.

Cheers,

Iain.

Hi again folks,

As many of you know, I'm in the process of updating the textbook I'm writing with the Strathclyde team.

We're now looking at the final chapters and would like a short case study describing where someone has had zero useful Y-STR matches, but found out a lot more information about their ancestry when they upgraded to BigY (or YElite, WGS or a similar sequencing test). This is to showcase the value of upgrading to BigY even if Y-STRs haven't been particularly useful to you.

If any of you have a case study that you'd like to put forward, please get in touch with my privately ASAP!

Best regards,

Iain.

Hi folks,

About November time I like to go through the haplotree, looking to see how much different branches and different countries have grown over the previous year. This helps us understand both what the country-level and haplogroup-level biases are, how they change over time, and where we might need to focus our efforts to help get the right people tested. The other reason for analysing these statistics is that it feeds back into the question of origins: where are people of particular haplogroups more or less likely to come from, hence where are specific haplogroups more likely to originate? But that's a question that's going to take me more time to answer than I have right now! Instead, see earlier messages like #5759. Thanks particularly to Ewenn who, late last year, gave me some code to speed up this process hugely.

Here, I'll be comparing two time periods: Nov 2022 - Nov 2023, and Nov 2019 to Nov 2023.

In general, this year continues the trend of previous years. The British Isles bias continues to get slightly worse, but this is more than counterbalanced by the extra information that new testers are bringing in.

Globally, the size of the FTDNA haplotree has increased by 6.6% in the last year, which is a slower rate than the average of the last four years (8.5%). The haplotree contains information from a variety of sources, so we can't infer information about how fast FTDNA's customer base is increasing from this information.

This increase has been disproportionately from people who cannot trace their origins back to Europe. The size of the European testing population has grown by only 6.1%. The British Isles bias now stands as follows: testers from the British Isles now make up 45% of European testers in the database, despite the British Isles making up only 8.6% of the comparative modern population, so we over-sample the British Isles by a factor of about 5.2 compared to the rest of Europe. It should be noted, however, that modern populations are not always indicative of historical populations. If we instead use population estimates from 1800, where the typical person's earliest-known ancestor lived, we find a British Isles bias of 5.9 instead. Some specific country or region-level modern/historical biases are:

England 2.3/1.6

Scotland 0.39/0.57

Wales 1.59/1.41

N.I. 0.71/1.6

Ireland 0.25/0.84

France 11.7/24.0

Germany 5.5/6.4

Netherlands 8.3/4.9

Poland 6.8/6.0

Czechia 6.5/11.5

Austria 9.5/14.1

Denmark 4.8/3.5

Scandinavia+Finland 1.45/1.16

European former USSR 15.0/16.1

Balkans+Turkey 20.9/13.9

Meditteranean 10.8/12.9

The further down we go in the tree, the faster the increase in testing becomes. This is because we get rid of the customers and studies that have only undertaken limited testing, and are increasingly left with only BigY testers. This is therefore a better estimator of the speed at which branches relevant to us are increasing.

The R-U106 testing population has increased by about 11.5% globally, by 10.9% in people with known European ancestry, and by 10.1% in the British Isles. These are very similar rates to those over the last four years, so the growth is fairly constant. These rates are faster than the growth in R-P312 (7.9% globally), but this may reflect the historical depth of testing rather than any inherent behaviour in R-U106! R-U106 testers make up 8.55% the haplotree at FTDNA. We've seen above-average increases in Northern Ireland and among the former Eastern Bloc countries, particularly the Czech Republic, but also Poland. We've seen below-average increases in Scotland, Belgium, Norway, Finland and Russia - the latter largely because FTDNA now offer sub-populations within Russia rather than because of geopolitics.

We can step down further to R-Z2265, which represents almost the whole of R-U106, but ignores the ~1240 testers that have only tested as far as R-U106 with single SNP or SNP-pack testers, i.e. mostly BigY testers. There are 19692 Z2265+ entries globally (an increase of 13.2%) and 8306 in Europe (an increase of 12.1%). This rate is slightly above the average for the last four years (12.4%/11.3%). Given the increase in database size, that shows that the rate of BigY testing is still proportionally increasing. At this rate, the database size doubles about every six years meaning that an average tester will have to wait about six years to receive a match closer to them and a new haplogroup designation (obviously not true for people who have purposefully tested close relatives).

Different haplogroups have grown faster than others. The reasons behind this haven't always been clear! For example, testing in R-Z18 and particularly R-L257 have been growing much faster this year (13.1%, 14.2%) than their average in the previous four years (10.2%, 9.6%). Testing in R-Z156 and particularly R-DF96 have slowed (from 13.9% to 10.8% and 12.6% to 10.3%). R-DF98 continues to out-perform other haplogroups in terms of deep testing (22.0% to 17.4%). R-L47 is below average (9.4% to 10.1%) but R-Z9 have increased testing (13.9% to 16.2%). R-FGC910 in particular has increased in size by a sizeable 20.7% on the year, and R-Z343 by 15.8%. Hopefully many of you will have seen these increases among your matches.

Cheers,

Iain.