PDA

View Full Version : The "Northumbria Problem" - A Short Story About Scottish Roots and Living DNA Tests



sktibo
10-26-2020, 12:43 AM
My primary interest and reason for DNA testing in the first place is in Celtic ancestry, and when the news came that Living DNA had gained the right to use the data-set of the People of the British Isles (POBI) study for a commercial test, I was among the first to sign up. I wanted to verify that I had Gaelic roots of some sort, hoping I would get some amount of Northwest Scotland among my percentages. I got something quite different instead.

My living DNA results came with a surprise: I have always been given a consistent amount of their “Northumbrian” category: initially, 26.2%, then to 29.8%, then to 42.1%, down to 12.4%, and finally back up to 20.3%. I also uploaded a second test, and the pattern continued: The uploads gave me 22.8%, 22.9%, then 20.8%. Whether on the upload or the regular test, Northumbria has always been my highest British percentage.
Looking at my paper trail, this was impossible: the only ancestry I have from this region is one third-great grandparent from the Scottish Borders; ~3%. My initial uploads for both my parents did not show particularly high Northumbrian percentages; dad showing zero, mom showing 9.7%.

The bulk of my Scottish ancestry is from two great-grandparents, my mother’s grandmother, and my father’s grandfather. My father’s grandfather comes from Stirling, but many of his roots come from Perthshire. My mother’s grandmother is of Northern Irish ancestry, from what I can tell, those ancestors came from Scotland, but I’m sure many of you know how tricky it is to determine exact origins for ancestors from Northern Ireland.

Living DNA initially stated that ancestry from Stirlingshire and Perthshire would fall under their Northwest Scotland category. This didn’t line up with my results.

Fortunately, a genetic study focusing on Scotland finally came out: The Genetic Landscape of Scotland and the Isles (GLSI&I). What it found was that much of Perthshire clustered with Fife, and was more closely related to people in Aberdeenshire than to the people of Western Scotland, such as the Hebrideans. Living DNA’s assumption about Stirlingshire and Perthshire was incorrect. The study tells us that their cluster which aligns most closely with many of the locations of my dad’s ancestors from Scotland, showed stronger similarities with English DNA than the other Scottish clusters – at least the ones listed in admixture chart 3A.

Another piece of the puzzle arrived in the form of an update from Living DNA, producing drastically different results for the uploads of my parents, but nothing very different for myself. My mother’s Northumbria percentage dropped to 4.4% (paper trail, 6.25%) and my dad’s results gained a Northumbrian percentage of 11.2%, his second highest British percentage. Although it still didn’t add up, this update finally indicated that my “Northumbrian Problem” might be coming from both of my parents, possibly more from my dad than my mom.

Why, if dad’s Scottish ancestry was from correlated with the Tayside-Fife cluster in the GLS&I, would he be getting Northumbrian?

Those of you who have read the various genetic studies on Britain and Ireland will know that the genetic differences between most of these clusters exists on a cline or a spectrum, ranging from the extremities of Ireland on one end, to England on the other. In terms of England and Scotland, the borders, or Northumbria and Cumbria clusters, lie in between England and the Scottish lowland clusters.

However, in the original POBI study, the Scottish Lowlands were very much under-sampled compared to the GLS&I. The South-West Scotland and Northern Ireland cluster numbered only 74 samples in the POBI, and in the GLS&I their Sco-Ire cluster, corresponding to the same geographic area, numbered 255 samples. The Tayside-Fife cluster didn’t exist, and the GLS&I gathered 177 samples for this cluster. The samples from the North-East of Scotland, “Aberdeenshire” in the POBI number around 52. The GLS&I upped this to 94 for Aberdeenshire, and 24 for Buchan-Moray.

If you look at figure 1 C in the GLS&I, you’ll notice that this Tayside-Fife cluster falls between Aberdeenshire and the Borders.

When sufficient samples are not available, a DNA test generally pulls from the surrounding populations instead. Another consistent category that has featured prominently on all versions of my Living DNA results has been Aberdeenshire, which is something my dad’s results have also featured. He doesn’t have any ancestry from the north-east of Scotland, only central Scotland, Easter-Ross, and Orkney.
Being in-between the Borders and Aberdeenshire makes sense, but the Northumbrian component was still the largest out of all the potential Scottish categories. This indicates that a good chunk of my father’s Scottish ancestry wasn’t really all that Scottish, because the borders regions are as “southern” or "continental" as you can get for a Scottish category. If you look at figure 3 A of the GLS&I, the admixture chart which divides cluster ancestry into Welsh, English, and Scottish, you’ll see that the Tayside-Fife cluster has more English admixture than any other component listed, with the most extreme samples from the cluster being very English-like.

To test this idea, I decided I needed an idea about my father’s overall PCA positioning – Eurogenes Global 25 and the Celtic vs Germanic PCA came into play. I found that my dad plots in a position that isn’t very northern relative to his ancestry on these plots. About 15% of his ancestry is French, but theoretically, his Scottish ancestry should more than counter-balance that and plot him firmly with the English. However, his position looks to be about as southern as you can get for someone of primarily British stock, clustering with Normans and Bretons on the Celtic vs Germanic plot, and on several Global 25 plots. To land in this position, I imagine his Scottish ancestry wouldn’t be all that much different from English ancestry, and then his French ancestry would pull him “southward” on the PCA.

My dad’s family has a very romanticized idea about their Scottish ancestors, imagining tartan-clad, broadsword wielding, red-haired Jacobites. However, the story that our genetics tells is much less exotic, indicating that perhaps many migrants from the nearby continent contributed to the gene pool of my father’s grandfather. Rather than being proud Gaelic speaking folk, our actual Scottish ancestors could have very well contributed to the replacement of the Gaelic language with the Scots language.

My opinion has changed over the years since I got my results from Living DNA, but ultimately, I believe consistency in DNA results appears to be telling us something, even if it takes genetic studies and updated results to be able to figure out what it may be saying. That said, if we are fortunate enough to receive another update from Living DNA, things might change again, throwing my theories out.

Regardless of the accuracy, if you're interested in your British ancestry, Living DNA might be able to give you something to ponder. My Living DNA results have provided me with more entertainment than the other commercial tests.

MitchellSince1893
10-26-2020, 01:02 AM
Why, if dad’s Scottish ancestry was from correlated with the Tayside-Fife cluster in the GLS&I, would he be getting Northumbrian?

Just making sure you know that when you click on the Living DNA map for Northumbria it includes Fife and all of southeast Scotland.

https://i.pinimg.com/originals/05/b2/b2/05b2b2ab9d4d48a654753bd0c469fa8d.png

sktibo
10-26-2020, 01:13 AM
Just making sure you know that when you click on the Living DNA map for Northumbria it includes Fife and all of southeast Scotland.

https://i.pinimg.com/originals/05/b2/b2/05b2b2ab9d4d48a654753bd0c469fa8d.png

Yep, I was aware of that

MitchellSince1893
10-26-2020, 01:19 AM
FWIW, my father should get around 10-13% North Sea facing Scottish. On Living DNA he only gets 4% for Aberdeen & Northumbria, so I'm not really that impressed with it.

Garimund
10-26-2020, 01:34 AM
Very interesting and well thought out post, Sktibo. I hope LivingDNA provides you with some new insight one of these days. It makes me want to read more into these studies myself.

sktibo
10-26-2020, 01:35 AM
FWIW, my father should get around 10-13% North Sea facing Scottish. On Living DNA he only gets 4% for Aberdeen & Northumbria, so I'm not really that impressed with it.

The point of my story isn't about Living DNA being good, I don't think it is for most people. My results and those of my parents are full of problems. The story was about the consistency of an unexpected, arguably inaccurate region that just wouldn't go away, but led me to reconsider my regional ancestry. I think there was truth in it, but it was nowhere near face value, as explained by the years of thinking about the results and trying to compare all the relevant data to come up with an explanation.

I ended my little story by saying "Regardless of the accuracy" this test "might give you something to ponder"

sktibo
10-26-2020, 01:39 AM
Very interesting and well thought out post, Sktibo. I hope LivingDNA provides you with some new insight one of these days. It makes me want to read more into these studies myself.

I wish they would somehow gain access to the data from the Irish DNA Atlas and the GLS&I. That would be amazing. The POBI dataset was great when it came out, but it's been built upon massively since then, with major regions like central Scotland and Ireland left un-sampled.

sktibo
10-26-2020, 04:12 AM
FWIW, my father should get around 10-13% North Sea facing Scottish. On Living DNA he only gets 4% for Aberdeen & Northumbria, so I'm not really that impressed with it.

What did your father get all in all, compared to his known ancestry?

MitchellSince1893
10-26-2020, 04:40 AM
What did your father get all in all, compared to his known ancestry?

A few months ago I talked about it here

https://anthrogenica.com/showthread.php?3987-Help-me-to-solve-a-family-Mystery-My-paternal-line-before-my-great-grandfather&p=643539&viewfull=1#post643539

sktibo
10-26-2020, 04:50 AM
A few months ago I talked about it here

https://anthrogenica.com/showthread.php?3987-Help-me-to-solve-a-family-Mystery-My-paternal-line-before-my-great-grandfather&p=643539&viewfull=1#post643539

Interesting to see what was consistent between your dad's two updates and with his half-sister's upload. Ireland is frequently assigned to what should be Scottish ancestry, because they got a boatload of Irish samples in from many kind people, including forumites here, and their Scottish panels in some areas remain woefully under-sampled.

40639

As a point of comparison, here's my dad, my mom, then my two uploads. None of us were lucky enough to have actual Irish ancestry, but we all get it on Living DNA. I think there's generally truth in the consistent percentages, and the upper percentages, but I don't interpret the percentage amounts assigned literally.

firemonkey
10-26-2020, 08:54 AM
My known Scottish ancestry is from Banffshire,Morayshire,Glasgow,and Midlothian


Original kit (latest\0

Aberdeenshire 39.2
NI + SW S 15.6

23 upload latest

Aberdeenshire 46.6
NI +SW S 17.4


FTDNA upload latest

Aberdeenshire 47
NI +SW S 14.5

Ancestry DNA upload latest

Aberdeenshire 47
NI +SW S 11.8


My heritage upload latest

Aberdeenshire 47.9
NI + SW S 12.8

Those are 1st and 2nd listed matches apart from Ancestry DNA where NI + SW S is listed 3rd .


Perhaps irrelevant/trivial info


335 matches of 15+ cMs at Ancestry DNA seem to have Hebridean ancestry . 17 of those are starred matches .

jadegreg
10-26-2020, 01:52 PM
Perhaps when and IF, we actually see the White Paper, it may shed a great deal of light on this, from reference populations, phasing, smoothing and segment size for ethnicity attribution. As we now know, all of these factors can dramatically effect what you get on these commercial testing enterprises. While I'm generally content with my LDNA results, as they at least identify my main appropriate regions, many of them are not in the ratios one would expect, and there are a few surprises such as my 12% S Yorkshire.......but if they nailed these things consistently, what would we have to talk about? Theorising about, explaining (attempting to) and discussing these inaccurate results keeps these brain cogs whirring, slowing my descent into decrepitude.....barely

sktibo
10-26-2020, 03:52 PM
Perhaps when and IF, we actually see the White Paper, it may shed a great deal of light on this, from reference populations, phasing, smoothing and segment size for ethnicity attribution. As we now know, all of these factors can dramatically effect what you get on these commercial testing enterprises. While I'm generally content with my LDNA results, as they at least identify my main appropriate regions, many of them are not in the ratios one would expect, and there are a few surprises such as my 12% S Yorkshire.......but if they nailed these things consistently, what would we have to talk about? Theorising about, explaining (attempting to) and discussing these inaccurate results keeps these brain cogs whirring, slowing my descent into decrepitude.....barely

Everything.you could hope to know about their British references is documented in the POBI and the materials that accompany that. As for their references outside of Britain.. IMO they've been a bit secretive.

jadegreg
10-26-2020, 04:04 PM
Chatting with customer services, some time ago, they suggested that they had expanded the dataset beyond the initial PoBI study. It's likely that this reference dataset has been continually refined since LDNA's inception, No? Well at least, I would hope that it has, or there is almost no way to trust segmentation of the UK, into its current regional breakdown, particularly the 'beloved Big Red Blob', SE cluster. Unless there are clear regional 'haplotypes' to their prescribed regions contained within the original dataset, which may be difficult to beleive, given that the SE Cluster would have suffered huge attrition beyond the 4 grandparent level (communique with scientist on project), then I don't see how they can be sticking entirely with just the PoBI participants

sktibo
10-26-2020, 05:41 PM
Chatting with customer services, some time ago, they suggested that they had expanded the dataset beyond the initial PoBI study. It's likely that this reference dataset has been continually refined since LDNA's inception, No? Well at least, I would hope that it has, or there is almost no way to trust segmentation of the UK, into its current regional breakdown, particularly the 'beloved Big Red Blob', SE cluster. Unless there are clear regional 'haplotypes' to their prescribed regions contained within the original dataset, which may be difficult to beleive, given that the SE Cluster would have suffered huge attrition beyond the 4 grandparent level (communique with scientist on project), then I don't see how they can be sticking entirely with just the PoBI participants

Did they specifically state that they had expanded their British references beyond the POBI or just that they had expanded their references? they have expanded their non British references quite a bit.

And I do believe there is no way to trust at least England's segmentation, results from people over the years have shown fractured and random percentages of clusters that fall into the big red blob

My experiences asking their customer service questions about their dataset has been frustrating as I've generally received answers which weren't really relevant to my questions.

Here was the info we had on it at the time of launch:



The original POBI map, which formed the centrepiece of the Nature paper, featured 17 regions. For the new Living DNA test the POBI data has been re-analysed using improved methods and the data has been clustered into 21 areas across the British Isles.

The Living DNA test also includes a standard ancestry report providing admixture percentages. This analysis is based on 80 worldwide regions. (Note that the test does not currently report Jewish or Aboriginal ancestry though hopefully these reference populations will be added in the future.)
Source https://cruwys.blogspot.com/2016/09/living-dna-new-genetic-ancestry-test.html

A link to the original Living DNA thread here from 2016 https://anthrogenica.com/showthread.php?8691-Living-DNA-launched-22-Sept-2016-British-Isles-focus

They used just the original POBI dataset for England and formed their other English regions with that at the time of launch

These English regions didn't change as time went on, my original / first results for Britain:
40666

jadegreg
10-26-2020, 06:10 PM
As I was referring to the British results, I assumed that they were referring to expanded dataset for the British populations, but as you rightfully say 'communication' with their customer services has been frustrating at best, with delayed responses, referral to the science team and corresponding half-answers. So yeah, I take your point.........grain of salt.

As I say though, hopefully we will get elaboration at some point with the corresponding White Paper. Whether it will aley any of our concerns/issues.......another story.

sktibo
10-26-2020, 07:42 PM
As I was referring to the British results, I assumed that they were referring to expanded dataset for the British populations, but as you rightfully say 'communication' with their customer services has been frustrating at best, with delayed responses, referral to the science team and corresponding half-answers. So yeah, I take your point.........grain of salt.

As I say though, hopefully we will get elaboration at some point with the corresponding White Paper. Whether it will aley any of our concerns/issues.......another story.

IIRC a few of us have been asking for a white paper for.. years. I'll bet money it's not going to happen. It also seems like they've stopped working on their autosomal updates/projects, though I hope I am wrong.

jadegreg
10-26-2020, 07:49 PM
My last communication, was about said White Paper and it has been delayed, but they did suggest it would be out nect year. As for the autosomal projects, if you mean, the fine scale European breakdown, then that's been put permanently on the back burner, indeed the last person I asked, had no idea it was ever on the cards........so dead in the water.

sktibo
10-26-2020, 07:53 PM
My last communication, was about said White Paper and it has been delayed, but they did suggest it would be out nect year. As for the autosomal projects, if you mean, the fine scale European breakdown, then that's been put permanently on the back burner, indeed the last person I asked, had no idea it was ever on the cards........so dead in the water.

Ah damn, dead in the water it is. They're putting their effort into some matching system that's completely unnecessary as I don't see any way it'll be better than what the other companies with already functional matching offer. If I were them, I'd just buff out my British and Irish regions as much as possible and market it as a test specifically for British and Irish genealogical interest. Overlapping continental regions probably only really serve to conflict with potential British regions anyhow, and for someone like me who has Eastern European and Native american ancestry, I can get that elsewhere anyhow so it doesn't really do me any good.

jadegreg
10-26-2020, 09:08 PM
Ah damn, dead in the water it is. They're putting their effort into some matching system that's completely unnecessary as I don't see any way it'll be better than what the other companies with already functional matching offer. If I were them, I'd just buff out my British and Irish regions as much as possible and market it as a test specifically for British and Irish genealogical interest.

Agreed. And as you say, if they can take advantage of the Byrne et al and Gilbert et al papers to improve the British and Irish breakdowns, that would be all to the good. I still would like them to further refine England and Wales as well, or at least do the work to assess whether it was possible


Overlapping continental regions probably only really serve to conflict with potential British regions anyhow, and for someone like me who has Eastern European and Native american ancestry, I can get that elsewhere anyhow so it doesn't really do me any good.

As it currently stands I think you're right! None of the major companies can differentiate S. England from N. France and the Benelux. But that doesn't mean we shouldn't be hopeful for the future. If LDNA gets to the point of reliably seperating out S. England regions, and I accept that's an IF, then I think the separation of these regions from their continental neighbours 'should' be possible, and obviously this is particularly relevant in my case!

msmarjoribanks
10-27-2020, 01:35 AM
Ah damn, dead in the water it is. They're putting their effort into some matching system that's completely unnecessary as I don't see any way it'll be better than what the other companies with already functional matching offer. If I were them, I'd just buff out my British and Irish regions as much as possible and market it as a test specifically for British and Irish genealogical interest. Overlapping continental regions probably only really serve to conflict with potential British regions anyhow, and for someone like me who has Eastern European and Native american ancestry, I can get that elsewhere anyhow so it doesn't really do me any good.

The problem is that if they want to attract people with mixed ancestry (North Americans), they can't give any reliable B/I results if their German results are way off. Now I get some German, but initially I got 11% Scandinavian (basically right) and some trace stuff, but otherwise English and Welsh results. The fact I knew I had 18% or so German made me think the English results were skewed.

(Unfortunately, it's even more messed up now, IMO.)

sktibo
10-27-2020, 01:40 AM
The problem is that if they want to attract people with mixed ancestry (North Americans), they can't give any reliable B/I results if their German results are way off. Now I get some German, but initially I got 11% Scandinavian (basically right) and some trace stuff, but otherwise English and Welsh results. The fact I knew I had 18% or so German made me think the English results were skewed.

(Unfortunately, it's even more messed up now, IMO.)

Yeah, I think that's the problem though, all these companies competing for the same general diaspora, but tons of people of English heritage chose to test with Living DNA because there was some marketing for that angle, and it really showed how much interest people in Britain and Ireland have in genetic genealogy. I think it could be a better business choice to specialize and target that one audience because none of the other tests offer that. Who, born in Britain, with their roots there, would want to test at 23andme after the recent update where it seems to (accurately) assign British people 90%+ British? (except you jadegreg, you continental-percentage getting traitor) Go for the unique market. the Living DNA facebook group was full of people of entirely British heritage who wanted an estimate that would break it down on a British level.

JerryS.
10-27-2020, 02:12 AM
the North Sea region is a mess on most calculators and from most companies..... if you have some from one place you'll show having some from damn near all places.

msmarjoribanks
10-27-2020, 06:10 PM
Yeah, I think that's the problem though, all these companies competing for the same general diaspora, but tons of people of English heritage chose to test with Living DNA because there was some marketing for that angle, and it really showed how much interest people in Britain and Ireland have in genetic genealogy. I think it could be a better business choice to specialize and target that one audience because none of the other tests offer that. Who, born in Britain, with their roots there, would want to test at 23andme after the recent update where it seems to (accurately) assign British people 90%+ British? (except you jadegreg, you continental-percentage getting traitor) Go for the unique market. the Living DNA facebook group was full of people of entirely British heritage who wanted an estimate that would break it down on a British level.

Yeah, I agree (and would be interested in the results), but it's like US movie-makers going for the Chinese audience -- you go where the numbers are. It's too bad in that my motivation to test was actually the hopes of a different UK-based set of matches, which would be more likely if the test was made to be especially useful to those with longterm UK ancestry.

MacUalraig
10-27-2020, 06:51 PM
Yeah, I agree (and would be interested in the results), but it's like US movie-makers going for the Chinese audience -- you go where the numbers are. It's too bad in that my motivation to test was actually the hopes of a different UK-based set of matches, which would be more likely if the test was made to be especially useful to those with longterm UK ancestry.

That logic never works because invariably tons of other Americans have the same idea and the next thing you know, its another US database ;-)

jadegreg
10-28-2020, 02:43 PM
Yeah, I think that's the problem though, all these companies competing for the same general diaspora, but tons of people of English heritage chose to test with Living DNA because there was some marketing for that angle, and it really showed how much interest people in Britain and Ireland have in genetic genealogy. I think it could be a better business choice to specialize and target that one audience because none of the other tests offer that. Who, born in Britain, with their roots there, would want to test at 23andme after the recent update where it seems to (accurately) assign British people 90%+ British? (except you jadegreg, you continental-percentage getting traitor) Go for the unique market. the Living DNA facebook group was full of people of entirely British heritage who wanted an estimate that would break it down on a British level.

Traitorous soft cheese consumer here. So, yes, They still use the PoBI samples, but have continued to expand and refine the panel, using the same PoBI criteria ( 4 GP, 50 mile radius etc). They are working on the Irish panel, but it sounds like the reason it hasn't 'hit the stores' as of yet, is that the've been trying but still haven't got their mits on any of the 3 Trinity College Datasets. So you'll be waiting until they get enough 100% Irish testers, with identified origins, or they reach a handsome cash figure settlement with the Dubh....

sktibo
10-28-2020, 03:58 PM
Traitorous soft cheese consumer here. So, yes, They still use the PoBI samples, but have continued to expand and refine the panel, using the same PoBI criteria ( 4 GP, 50 mile radius etc). They are working on the Irish panel, but it sounds like the reason it hasn't 'hit the stores' as of yet, is that the've been trying but still haven't got their mits on any of the 3 Trinity College Datasets. So you'll be waiting until they get enough 100% Irish testers, with identified origins, or they reach a handsome cash figure settlement with the Dubh....

Hey Soft Cheese, Chicken Skin here. Thanking you for the info.
They've refined the British part of the panel or did they just say the panel? The Irish and German references have been updated a couple of times IIRC.

jadegreg
10-28-2020, 04:13 PM
Just British. I made clear that is what I wanted to know. I meant refinement in the Irish categories, in line with Byrne et al. But as I say they haven't got their hands on those samples......yet.

sktibo
10-28-2020, 04:15 PM
Just British. I made clear that is what I wanted to know. I meant refinement in the Irish categories, in line with Byrne et al. But as I say they haven't got their hands on those samples......yet.

I wonder which categories samples were added to

jadegreg
10-28-2020, 04:28 PM
I've asked that before, but they only ever deal with generalisms it seems. Still no direct line to the science team, info is always passed on with the heavy filtering of customer services

Webb
10-28-2020, 04:41 PM
My largest grouping at Living DNA is around 20% East Anglia, followed by Yorkshire, then Cumbria. After that I drop into the single digits for the rest of the geographical areas of Britain. My over all British Isles is 72%.

Northman
10-28-2020, 05:45 PM
My primary interest and reason for DNA testing in the first place is in Celtic ancestry, and when the news came that Living DNA had gained the right to use the data-set of the People of the British Isles (POBI) study for a commercial test, I was among the first to sign up. I wanted to verify that I had Gaelic roots of some sort, hoping I would get some amount of Northwest Scotland among my percentages. I got something quite different instead.

My living DNA results came with a surprise: I have always been given a consistent amount of their “Northumbrian” category: initially, 26.2%, then to 29.8%, then to 42.1%, down to 12.4%, and finally back up to 20.3%. I also uploaded a second test, and the pattern continued: The uploads gave me 22.8%, 22.9%, then 20.8%. Whether on the upload or the regular test, Northumbria has always been my highest British percentage.
Looking at my paper trail, this was impossible: the only ancestry I have from this region is one third-great grandparent from the Scottish Borders; ~3%. My initial uploads for both my parents did not show particularly high Northumbrian percentages; dad showing zero, mom showing 9.7%.

The bulk of my Scottish ancestry is from two great-grandparents, my mother’s grandmother, and my father’s grandfather. My father’s grandfather comes from Stirling, but many of his roots come from Perthshire. My mother’s grandmother is of Northern Irish ancestry, from what I can tell, those ancestors came from Scotland, but I’m sure many of you know how tricky it is to determine exact origins for ancestors from Northern Ireland.

Living DNA initially stated that ancestry from Stirlingshire and Perthshire would fall under their Northwest Scotland category. This didn’t line up with my results.

Fortunately, a genetic study focusing on Scotland finally came out: The Genetic Landscape of Scotland and the Isles (GLSI&I). What it found was that much of Perthshire clustered with Fife, and was more closely related to people in Aberdeenshire than to the people of Western Scotland, such as the Hebrideans. Living DNA’s assumption about Stirlingshire and Perthshire was incorrect. The study tells us that their cluster which aligns most closely with many of the locations of my dad’s ancestors from Scotland, showed stronger similarities with English DNA than the other Scottish clusters – at least the ones listed in admixture chart 3A.

Another piece of the puzzle arrived in the form of an update from Living DNA, producing drastically different results for the uploads of my parents, but nothing very different for myself. My mother’s Northumbria percentage dropped to 4.4% (paper trail, 6.25%) and my dad’s results gained a Northumbrian percentage of 11.2%, his second highest British percentage. Although it still didn’t add up, this update finally indicated that my “Northumbrian Problem” might be coming from both of my parents, possibly more from my dad than my mom.

Why, if dad’s Scottish ancestry was from correlated with the Tayside-Fife cluster in the GLS&I, would he be getting Northumbrian?

Those of you who have read the various genetic studies on Britain and Ireland will know that the genetic differences between most of these clusters exists on a cline or a spectrum, ranging from the extremities of Ireland on one end, to England on the other. In terms of England and Scotland, the borders, or Northumbria and Cumbria clusters, lie in between England and the Scottish lowland clusters.

However, in the original POBI study, the Scottish Lowlands were very much under-sampled compared to the GLS&I. The South-West Scotland and Northern Ireland cluster numbered only 74 samples in the POBI, and in the GLS&I their Sco-Ire cluster, corresponding to the same geographic area, numbered 255 samples. The Tayside-Fife cluster didn’t exist, and the GLS&I gathered 177 samples for this cluster. The samples from the North-East of Scotland, “Aberdeenshire” in the POBI number around 52. The GLS&I upped this to 94 for Aberdeenshire, and 24 for Buchan-Moray.

If you look at figure 1 C in the GLS&I, you’ll notice that this Tayside-Fife cluster falls between Aberdeenshire and the Borders.

When sufficient samples are not available, a DNA test generally pulls from the surrounding populations instead. Another consistent category that has featured prominently on all versions of my Living DNA results has been Aberdeenshire, which is something my dad’s results have also featured. He doesn’t have any ancestry from the north-east of Scotland, only central Scotland, Easter-Ross, and Orkney.
Being in-between the Borders and Aberdeenshire makes sense, but the Northumbrian component was still the largest out of all the potential Scottish categories. This indicates that a good chunk of my father’s Scottish ancestry wasn’t really all that Scottish, because the borders regions are as “southern” or "continental" as you can get for a Scottish category. If you look at figure 3 A of the GLS&I, the admixture chart which divides cluster ancestry into Welsh, English, and Scottish, you’ll see that the Tayside-Fife cluster has more English admixture than any other component listed, with the most extreme samples from the cluster being very English-like.

To test this idea, I decided I needed an idea about my father’s overall PCA positioning – Eurogenes Global 25 and the Celtic vs Germanic PCA came into play. I found that my dad plots in a position that isn’t very northern relative to his ancestry on these plots. About 15% of his ancestry is French, but theoretically, his Scottish ancestry should more than counter-balance that and plot him firmly with the English. However, his position looks to be about as southern as you can get for someone of primarily British stock, clustering with Normans and Bretons on the Celtic vs Germanic plot, and on several Global 25 plots. To land in this position, I imagine his Scottish ancestry wouldn’t be all that much different from English ancestry, and then his French ancestry would pull him “southward” on the PCA.

My dad’s family has a very romanticized idea about their Scottish ancestors, imagining tartan-clad, broadsword wielding, red-haired Jacobites. However, the story that our genetics tells is much less exotic, indicating that perhaps many migrants from the nearby continent contributed to the gene pool of my father’s grandfather. Rather than being proud Gaelic speaking folk, our actual Scottish ancestors could have very well contributed to the replacement of the Gaelic language with the Scots language.

My opinion has changed over the years since I got my results from Living DNA, but ultimately, I believe consistency in DNA results appears to be telling us something, even if it takes genetic studies and updated results to be able to figure out what it may be saying. That said, if we are fortunate enough to receive another update from Living DNA, things might change again, throwing my theories out.

Regardless of the accuracy, if you're interested in your British ancestry, Living DNA might be able to give you something to ponder. My Living DNA results have provided me with more entertainment than the other commercial tests.

There's an easy explanation to the Northumbrian problem and that is because their classification of Northumbria includes the region of the ancient Kingdom of Northumbria and not the modern day county of Northumberland. This means, this region correlates to ancestry north of the Humber in south Yorkshire all the way up to Edinburgh. So basically, lowland Scots are very much the same as northern English on a genetic level! I have two grandparents from Scotland (West Lothian and East Lothian), one from Yorkshire and one from Northumberland. Here's my LivingDNA results: 40722

sktibo
10-28-2020, 05:54 PM
There's an easy explanation to the Northumbrian problem and that is because their classification of Northumbria includes the region of the ancient Kingdom of Northumbria and not the modern day county of Northumberland. This means, this region correlates to ancestry north of the Humber in south Yorkshire all the way up to Edinburgh. So basically, lowland Scots are very much the same as northern English on a genetic level! I have two grandparents from Scotland (West Lothian and East Lothian), one from Yorkshire and one from Northumberland. Here's my LivingDNA results: 40722

Truly, the most Northumbrian of us all! Thanks for posting your result.
Why it's a problem in my case is that I have no ancestry from Fife or the ancient kingdom of Northumbria, save 3.125% from Roxburghshire, and so the reasoning in my story had to come into play..

I know I'm long-winded but if I had ancestry from these regions there would not have been a short story, it would have been resolved in 2017.

JerryS.
10-28-2020, 06:03 PM
There's an easy explanation to the Northumbrian problem and that is because their classification of Northumbria includes the region of the ancient Kingdom of Northumbria and not the modern day county of Northumberland. This means, this region correlates to ancestry north of the Humber in south Yorkshire all the way up to Edinburgh. So basically, lowland Scots are very much the same as northern English on a genetic level! I have two grandparents from Scotland (West Lothian and East Lothian), one from Yorkshire and one from Northumberland. Here's my LivingDNA results: 40722

This is interesting. On my Mother's side I had a Scottish grandmother (supposedly "Scot-Irish") and a N. German grandfather (Bremen, Hamburg, and Mecklenburg), If the Scottish is "just Scottish" your map would fit my mother accurately.

Coupled that with my father's paternal line being all Colonial-American English with an I1 haplogroup explains why I get such a heavy Scandinavian pull on all the DIY models.

sktibo
10-28-2020, 06:11 PM
This is interesting. On my Mother's side I had a Scottish grandmother (supposedly "Scot-Irish") and a N. German grandfather (Bremen, Hamburg, and Mecklenburg), If the Scottish is "just Scottish" your map would fit my mother accurately.

Coupled that with my father's paternal line being all Colonial-American English with an I1 haplogroup explains why I get such a heavy Scandinavian pull on all the DIY models.

I don't think there's any doubt that Scottish ancestry can contribute to a Northern or Scandinavian like pull. Scottish ancestry is frequently labelled as Celtic but I think many Scots similarly to the English are a northern European mix

FionnSneachta
10-28-2020, 07:29 PM
Hey Soft Cheese, Chicken Skin here. Thanking you for the info.
They've refined the British part of the panel or did they just say the panel? The Irish and German references have been updated a couple of times IIRC.

I believe that the last update that we had regarding the progress of the project for creating sub-regions within Ireland was on the 25th July 2018 when it was 40% complete. They definitely did an update of the panel around November 2018 when they gave my mum 100% Ireland. The Irish panel is either incomplete for different regions within Ireland or else they've given up on the sub-regional breakdown.

sktibo
10-28-2020, 07:47 PM
I believe that the last update that we had on the project was on the 25th July 2018 when it was 40% complete. They did an update in November 2018 that gave my mum 100% Ireland. They have definitely updated the panels but it's either incomplete or else they've given up on the sub-regional breakdown.

I was under the impression they had updated the Ireland category and the German categories, as well as the other categories specifically mentioned by them in blogs notes ect

sktibo
10-28-2020, 07:49 PM
My largest grouping at Living DNA is around 20% East Anglia, followed by Yorkshire, then Cumbria. After that I drop into the single digits for the rest of the geographical areas of Britain. My over all British Isles is 72%.

Do you have east anglian or yorkshire ancestry? going to guess you have some lowland scots or ulster scots and that's what the Cumbria represents

FionnSneachta
10-28-2020, 07:51 PM
I was under the impression they had updated the Ireland category and the German categories, as well as the other categories specifically mentioned by them in blogs notes ect

The panel has definitely been updated but they either don't have enough samples in certain regions to create the sub-regions or else they're not bothering with that goal anymore and just want to have a solid Ireland region. It does seem like they've given up on that idea since there has no mention of the sub-regions since 2018.

Reggiemercer
10-28-2020, 08:09 PM
How much of this could simply be due to genetic recombination? Given how you get a random portion of your ancestry from either parent, with noise increasing with generational distance. For example, as far as genealogical records go I am 50% French Canadian and 50% Irish/Welsh/English, but my Ancestry results only give me 2% French with my ancestry overwhelmingly being British (69% Gaelic, 9% Welsh, 20% Anglo-Saxon).

It could also be that you just happened to inherit ancestry from a composite of sources and "Northumbria" was the closest reference point to it. For example, the paucity of French in my ancestry results is most likely because northern French is close enough to the English pool that my Celtic ancestry pulls the rest of my French ancestry into the England and NW Europe category.

JerryS.
10-28-2020, 08:15 PM
How much of this could simply be due to genetic recombination? Given how you get a random portion of your ancestry from either parent, with noise increasing with generational distance. For example, as far as genealogical records go I am 50% French Canadian and 50% Irish/Welsh/English, but my Ancestry results only give me 2% French with my ancestry overwhelmingly being British (69% Gaelic, 9% Welsh, 20% Anglo-Saxon).

It could also be that you just happened to inherit ancestry from a composite of sources and "Northumbria" was the closest reference point to it. For example, the paucity of French in my ancestry results is most likely because northern French is close enough to the English pool that my Celtic ancestry pulls the rest of my French ancestry into the England and NW Europe category.

If I remember correctly, French Canadian is from the Normandy and Brittney region of France. Those French regions are very similar to England. That could be why you get such a skewed breakdown.

Reggiemercer
10-28-2020, 08:33 PM
If I remember correctly, French Canadian is from the Normandy and Brittney region of France. Those French regions are very similar to England. That could be why you get such a skewed breakdown.

Yeah that's true. My father's side is from Normandy and there are a good number of Norman and Breton names in my dad's genealogy. Most Quebecois still score at least some significant French ancestry though. I met my paternal half-sister through Ancestry and she scored 36% French, 32% Scottish, 25% English/NWE, and 7% Irish. My best guess is that my father was loaded with Scottish ancestry he wasn't aware of or never mentioned to me, and the most likely candidate was my grandmother who was most likely of Scottish / Irish descent judging by her surname, possibly from Orkney.

sktibo
10-28-2020, 08:34 PM
How much of this could simply be due to genetic recombination? Given how you get a random portion of your ancestry from either parent, with noise increasing with generational distance. For example, as far as genealogical records go I am 50% French Canadian and 50% Irish/Welsh/English, but my Ancestry results only give me 2% French with my ancestry overwhelmingly being British (69% Gaelic, 9% Welsh, 20% Anglo-Saxon).

It could also be that you just happened to inherit ancestry from a composite of sources and "Northumbria" was the closest reference point to it. For example, the paucity of French in my ancestry results is most likely because northern French is close enough to the English pool that my Celtic ancestry pulls the rest of my French ancestry into the England and NW Europe category.

Sounds like you got a ton of Ancestry's Scotland category. I wouldn't cling on to that result from Ancestry too closely, it will change about once a year. I'm 64% "Gaelic" as you put it, which I'm assuming means a combination of Ireland and Scotland.
Why isn't recombination worth considering? because even if it were a factor I'd have no way to tell as my results are pretty different from test to test.
I think your idea about Northumbria being the closest reference to a composite of sources could possibly be part of what's going on. I won't rule it out.

Reggiemercer
10-28-2020, 08:43 PM
Sounds like you got a ton of Ancestry's Scotland category. I wouldn't cling on to that result from Ancestry too closely, it will change about once a year. I'm 64% "Gaelic" as you put it, which I'm assuming means a combination of Ireland and Scotland.
Why isn't recombination worth considering? because even if it were a factor I'd have no way to tell as my results are pretty different from test to test.
I think your idea about Northumbria being the closest reference to a composite of sources could possibly be part of what's going on. I won't rule it out.

Certainly possible, though in my case my mother's side is actually second-generation from England, and Ancestry matched me to the Caithness community in northern Scotland. This matches genealogy my aunt has done on my mother's side, though the properly English ancestry is far lower than it should be. We suspect my maternal grandmother was a NPE from some short Scotsman, based on research my aunt did into my great-grandmother's rather sketchy life. Prior to that I was 65% "Ireland and Scotland" and 35% "England, Wales and Northwestern Europe".

sktibo
10-28-2020, 09:02 PM
Certainly possible, though in my case my mother's side is actually second-generation from England, and Ancestry matched me to the Caithness community in northern Scotland. This matches genealogy my aunt has done on my mother's side, though the properly English ancestry is far lower than it should be. We suspect my maternal grandmother was a NPE from some short Scotsman, based on research my aunt did into my great-grandmother's rather sketchy life. Prior to that I was 65% "Ireland and Scotland" and 35% "England, Wales and Northwestern Europe".

Sure sounds like you have quite an interesting family background. Ancestry currently assigns me a pitiful 13% of their English category, and as you can see in my signature, it should be my highest percentage, in a perfect world. That's a point of comparison for you anyhow

jadegreg
10-28-2020, 09:31 PM
How much of this could simply be due to genetic recombination? Given how you get a random portion of your ancestry from either parent, with noise increasing with generational distance. For example, as far as genealogical records go I am 50% French Canadian and 50% Irish/Welsh/English, but my Ancestry results only give me 2% French with my ancestry overwhelmingly being British (69% Gaelic, 9% Welsh, 20% Anglo-Saxon).

It could also be that you just happened to inherit ancestry from a composite of sources and "Northumbria" was the closest reference point to it. For example, the paucity of French in my ancestry results is most likely because northern French is close enough to the English pool that my Celtic ancestry pulls the rest of my French ancestry into the England and NW Europe category.

Indeed, this was how they explained my South Yorkshire category to me, in a roundabout way, before I knew anything about how these tests work. We still know very little about how they approach the British Isles region assignment, as to whether they use an Ancestry or 23 &me like method, or something slightly different, and all that entails. I mean if their phasing method is a bit of a duffer, it maybe hardly surprising were getting slightly skewed percentages in some of our regional breakdowns.....

Reggiemercer
10-28-2020, 09:35 PM
Sure sounds like you have quite an interesting family background. Ancestry currently assigns me a pitiful 13% of their English category, and as you can see in my signature, it should be my highest percentage, in a perfect world. That's a point of comparison for you anyhow

Its certainly a colorful story, based on how short my grandmother was compared to her siblings (her sister was 5'6 and she was maybe 5'1) and how she looked nothing like her siblings and purported father. We then found out that my great-grandmother's family had hired private investigators to tail her around and watch what she was getting up to.

Given how closely related all British Islanders are I'm amazed geneticists can even tease things as specific as regional ancestry out. Ancestry cheats a bit by using matches to assign communities over actual ancestry, I'm pretty sure thats how I got mapped to Caithness because I have no recent "official" ancestry from anywhere near there. Juries out on how much people knew, because there was old money on both sides of the family at the time and that gives people a big incentive to "let sleeping dogs lie".

Are the %s in your signature from Ancestry itself or are they based on your genealogy? I've found its not uncommon to get "eastern Europe", particularly the Baltic region, because of overlapping ancestry rather than direct descent if you have roots in the British Isles.

sktibo
10-28-2020, 09:46 PM
Its certainly a colorful story, based on how short my grandmother was compared to her siblings (her sister was 5'6 and she was maybe 5'1) and how she looked nothing like her siblings and purported father. We then found out that my great-grandmother's family had hired private investigators to tail her around and watch what she was getting up to.

Given how closely related all British Islanders are I'm amazed geneticists can even tease things as specific as regional ancestry out. Ancestry cheats a bit by using matches to assign communities over actual ancestry, I'm pretty sure thats how I got mapped to Caithness because I have no recent "official" ancestry from anywhere near there. Juries out on how much people knew, because there was old money on both sides of the family at the time and that gives people a big incentive to "let sleeping dogs lie".

Are the %s in your signature from Ancestry itself or are they based on your genealogy? I've found its not uncommon to get "eastern Europe", particularly the Baltic region, because of overlapping ancestry rather than direct descent if you have roots in the British Isles.

It's my paper trail or my genealogy. It's there so that when I post the results of a test, it can be easily compared.

jadegreg
10-28-2020, 09:49 PM
Thanks for sharing. I must say its reassuring to hear of other family shenanigans........it seems none of my grandparents and quite a few of my GGP could barely contain themselves....

sktibo
10-28-2020, 09:53 PM
Thanks for sharing. I must say its reassuring to hear of other family shenanigans........it seems none of my grandparents and quite a few of my GGP could barely contain themselves....

That explains a lot about you

jadegreg
10-28-2020, 10:02 PM
That explains a lot about you

I'm quite sure I don't know what you're talking about...........unless it's the soft cheeses

sktibo
10-28-2020, 10:06 PM
I'm quite sure I don't know what you're talking about...........unless it's the soft cheeses

Quite possibly. may they be soft, smelly and blue

jadegreg
10-28-2020, 10:09 PM
Quite possibly. may they be soft, smelly and blue

And veiny, don't forget the veiny

msmarjoribanks
10-29-2020, 01:06 AM
I was under the impression they had updated the Ireland category and the German categories, as well as the other categories specifically mentioned by them in blogs notes ect

I believe they did. Originally I had no German or Irish (or Scottish), and now I have some (although annoyingly it took away some of my Welsh which looked right on originally).

msmarjoribanks
10-29-2020, 01:11 AM
Sure sounds like you have quite an interesting family background. Ancestry currently assigns me a pitiful 13% of their English category, and as you can see in my signature, it should be my highest percentage, in a perfect world. That's a point of comparison for you anyhow

I'm the same -- get only 12% English, etc. from Ancestry (and way inflated Scots and Welsh). But my sister gets 49%, which is in the ballpark. I get 70% British Isles overall, whereas she gets 81%. Mine is more correct there.

Northman
10-30-2020, 06:35 PM
Truly, the most Northumbrian of us all! Thanks for posting your result.
Why it's a problem in my case is that I have no ancestry from Fife or the ancient kingdom of Northumbria, save 3.125% from Roxburghshire, and so the reasoning in my story had to come into play..

I know I'm long-winded but if I had ancestry from these regions there would not have been a short story, it would have been resolved in 2017.

Here's another comparison for you. This is my great uncle who was born in West Lothian, Scotland (their direct paternal line has ancestry from the Isle of Skye even further back). He's my paternal grandmother's brother. He was the only one living which I could test and his yDNA is Norse in origin:
40815

sktibo
10-30-2020, 06:40 PM
Here's another comparison for you. This is my great uncle who was born in West Lothian, Scotland (their direct paternal line has ancestry from the Isle of Skye even further back). He's my paternal grandmother's brother. He was the only one living which I could test and his yDNA is Norse in origin:
40815

Fantastic! such a comparison makes me proud to score Northumbria as my highest percentage (I'm a big fan of the Last Kingdom series and the Angles are my favorite Germanic group, you see..)

Saetro
10-31-2020, 02:16 AM
the North Sea region is a mess on most calculators and from most companies..... if you have some from one place you'll show having some from damn near all places.

Not to mention the original Elhaik calculator.
That would have averaged my maternal British and paternal German and given my homeland as the Dogger Bank!

Saetro
10-31-2020, 02:50 AM
I wish they would somehow gain access to the data from the Irish DNA Atlas and the GLS&I. That would be amazing. The POBI dataset was great when it came out, but it's been built upon massively since then, with major regions like central Scotland and Ireland left un-sampled.

At the moment I'd give a bottle of single malt on top of the standard DNA price for that extra data to be included.
As it should shed light on both of my Scottish conundrums.
My Scottish line is back a way, and around the 1770s splits into a Perthshire one and a Glasgow? one.
The documented Perth connection is in the lowland part of the county with family lore talking about prior history in the Highlands of Perthshire.
The Glasgow? one is associated with two DNA clusters: one in Ayrshire/Renfrew and one to the surname McKeown, mostly from Northern Ireland.
The surname line it is associated with disappears one generation either way, but there are possible leads.
Walking around St Mirren's in Paisley I saw at least 3 possible spelling variants.
So on the one hand I think the Glasgow? line has one thread clouded by a name variant/change.
And on the other, there was an Irish inflow to Glasgow, maybe preceded by an outflow from Scotland in the 1600s.

JMcB
10-31-2020, 03:13 AM
I'm the same -- get only 12% English, etc. from Ancestry (and way inflated Scots and Welsh). But my sister gets 49%, which is in the ballpark. I get 70% British Isles overall, whereas she gets 81%. Mine is more correct there.

They’ve been all over the ballpark with me. First they gave me 49%, then it jumped to 62% and now it’s 28%. Not counting the Welsh which is 3%.

msmarjoribanks
11-01-2020, 01:26 AM
They’ve been all over the ballpark with me. First they gave me 49%, then it jumped to 62% and now it’s 28%. Not counting the Welsh which is 3%.

Yeah, mine has varied a ton too. At one point it was in the 70s.

Nqp15hhu
03-05-2021, 03:05 AM
.......