Origins

July 2016

My initial interest in DNA was to find out something about my racial origins. This is an attempt to pull together what I’ve found so far from DNA tests. It is written from my point of view and that of my brothers, Stephen, Peter and Edwin. Most of this paper is concerned with autosomal DNA (atDNA). Y chromosome results are detailed here and mitochondrial here. The overall results for all tests and my conclusions are summarised at the end of this paper.

Genealogy

All of our sixteen great-great-grandparents are known and all were born in England. They come from a range of places scattered over western, southern, eastern and central England. Further back on father’s side, we have one German immigrant, Lyon Asher, and his name suggests he was Jewish. From their names, it’s possible that the Nottingham families he married into also had Jewish connections. On mother’s side we have people who were Flemish, from what is now Southern Netherlands, Belgium and Northern France. We also have a Scottish connection through the Dunmores.

DNA

All of the methods available for estimating one’s origins have the same basic approach which is to take a set of DNA samples, where the ancestry of the donors is known to a greater or lesser extent, and group them into presumed ancestral populations. The grouping is done using a mixture of statistical methods (primarily principal component analysis) and subjective ranking. The groups are then used in a program, admixture, which uses a maximum likelihood method to classify one’s sample. Admixture is public domain and can be freely downloaded.

Family Tree DNA

MyOrigins

Family Tree DNA has a feature they call “MyOrigins” which aims to find one’s more distant origins. We have the same parents (confirmed by our genetic closeness) and consequently exactly the same ancestors. However, our percentage ancestry, as estimated by the MyOrigins analysis differs between us as in the following table:

So, Peter’s the most “British,” Edwin’s the most “Scandinavian and “Southern European” and has no  “Western & Central European” component, whilst I’m the only one with the “Middle Eastern” connection. There is a “white paper” describing the MyOrigins methodology. Their explanation is far from crystal-clear, but they have 36 reference populations representing several hundred individuals. They have grouped these into 18 broad geographical regions.

I also have MyOrigins results for Brian, Janet, Margaret and Margaret’s cousin David. They are very similar to ours with David being the most British at 60% and Edwin’s the least at 14%. This great variability in the results, particularly for the four of us leads me to doubt MyOrigins' usefulness.

Gedmatch

Gedmatch also has an ancestry feature using admixture. In this case Gedmatch just hosts calculators developed by different people. There are 5 different developers and of these, two target Europeans. Each has several variant sets of reference samples. Results for the two European calculators, Eurogenes and MDLP, follow.

Eurogenes

This table is for the K13 reference set which is the latest “standard” version. The columns don’t add up to 100 because I have rounded all the percentages and there are other minor contributors that I haven’t shown. These are much more consistent between us, particularly the main components, but they are also more inclusive groupings. West Asia is what today is the Middle East and South Asia is Southern India.

Associated with admixture is another program, oracle, which estimates one’s genetic distance from the underlying reference populations. I’m not sure what the units of the distance are, but the program compares one’s estimated proportions with the proportions in the reference populations. For our K13 results the distances are all over the place! The following are the ranks for the top eight matches for each of us, closest first:

Eurogenes also has a reference set aimed more specifically at hunter-gatherers and farmers:

Anatolia is approximately the Asian part of modern day Turkey and the “Anatolian Hypothesis” (see wikipedia) is that farming Indo-European people spread from that area into Europe starting around 7,000 years ago.  

As with MyOrigins, I have also tried both of these on other people and they all produce similar results to us, even Patrick’s wife Jean whose ancestry is all Irish.

MDLP

MDLP’s default reference set is K23b which is also targeted at hunter-gatherers and farmers:

I think Caucasian here equates more-or-less to Anatolia above. South Central Asia is India and Altaic derives from the Altai Mountains in Central Asia where Russia, China, Mongolia and Kazakhstan come together.

MDLP also have another reference set, World-22, that looks further back in time:

For World-22, there are maps available. In these North-East European has a focus around Poland and extending into the Ukraine; Atlantic-Mediterranean Neolithic covers Spain and Southern France; North European Mesolithic is centred on Finland; West Asian has a strong focus in the Caucasus and extends into Iran; Near East as one would expect is the Arabian Peninsula and Indo-Iranian is self-explanatory. Neolithic and mesolithic refer to technology rather than time periods and the transition from mesolithic to neolithic marks the change from hunting and gathering to farming for at least some of people’s food supply.

MDLP results are quite consistent between us for the main components and some commenters  have suggested ignoring anything that is 4% or less. Again, results for the other people I have access to are very similar to ours.

Ancient DNA

Gedmatch also has a section on ancient DNA. This matches one’s DNA with a set of samples from various archaeological sites varying in age from 1,000 to 50,000 years ago. My results are

shown in the figure at the end of this document. Stephen and Peter look much the same as do all the other people to whom I have access! The first two, which are also the oldest, are from Neanderthals, I, along with all Europeans, have a touch of Neanderthal. The ones that match most closely are all from Hungary and I match “BR2” (3,200 years old from the Ludas site, the nearest town is Gyöngyös, 10 km to the west;  Gyöngyös is 80 km east of Budapest). My match is close enough for me to be considered a direct descendant, Stephen and Peter, not quite but close; Edwin doesn't match at all. Other close matches in the figure are in Germany, Luxembourg and Siberia. Perhaps surprisingly, we are very distant from the samples from Hinxton which is near Cambridge!

Discussion

The groups listed in the tables above do not match each other exactly – or not at all in some cases – even when they have been given the same name. In the “white paper” on their methodology (see URL above), MyOrigins describe their groups as geographical areas with changing population composition as time progresses which to my mind seems inconsistent with the aim of the procedure. I think that the Eurogenes and MDLP groups are also essentially geographical with a more explicit time component.

In the following table I have ordered the top four groups for each of the Eurogenes and MDLP reference sets. The ranking is the same for all four of us and the number in brackets is the percentage for me:

There is some consistency across some of these rows but not enough to draw any firm conclusions. One of my aims in all of this was, initially, to see how our ancestry fitted into the theory of how Britain and Northern Europe were repopulated as the ice retreated at the end of the last ice age around 11,000-12,000 years ago. That is, there was a population of hunter-gatherers who spread, following the coast, from a western refuge in what is now Spain and Southern France, up the west coast of France to the British Isles and thence up the western seaboard of Britain. There was another population, mainly of farmers, who spread over Central Europe from an eastern refuge in what is now the Ukraine. A recent paper also shows flows from the Italian peninsula and the Near East (Eastern Mediterranean: e.g. Syria, Lebanon etc.)

DNA support for this theory is based on mtDNA and Y-DNA, but there is also support from other fields in particular linguistic traits and archaeological finds. Without the ability to time events in the evolution of atDNA, it’s hard to see how one can accurately infer historical migration patterns. With this in mind, all of these methods look at one's origins primarily in terms of geography. In the light of the Eurogenes and MDLP results, MyOrigins seems suspect, particularly given the variation between the four of us when we know that we have the same ancestry.

MDLP W22 seems to me to be the most allied with the re-population theory which would make us about 50% from the eastern refuge, 36% from the western refuge and most of the rest from what is now Turkey.

None of these estimates are very detailed and the MDLP researcher says that much larger samples of people are needed to unravel detailed British and European ancestry. Almost all the people I have tried these tests on give similar results and that includes some of the “Ancient DNA” samples including both BR2, which one might expect, and Hinxton which I didn’t expect. The only different result is for a person with a Jewish name that I nevertheless match distantly on Gedmatch. For MDLP World-22 he is only 4% North-East European a similar 28% Atlantic-Mediterranean Neolithic, much higher West Asian, 25%, and his Near East is 38% which is only 2% in my profile.

Clearly there has been some differentiation (evolution/mutation) between groups originating from the ancestral populations. In the British Isles there are, for example, red-haired, fair skinned Scots and Irish, the short dark-haired Welsh of Snowdonia and the dark-haired blue-eyed Irish. In looks, Patrick's wife Jean is a good example of the last but as I've noted above, her ancestry mix from MDLP and Eurogenes is the same as ours.  All of these characteristics must have evolved quite recently.

Summary & Conclusions

From the atDNA we clearly come from a mixture of ancestral populations over the last 10,000 years, as do all the other people I have data on who have recent origins in the British Isles. Given the historical invasions starting with the Romans, followed by the Angles, Saxons, Jutes (all of whom came from Eastern-Central Europe) and later the Vikings, it’s hardly surprising to find we have Central European and Scandinavian ancestry.

Our (my siblings and me) distant Y-DNA and mtDNA ancestors both lived in the Middle East during the last ice age (20 to 25 thousand years ago). As the ice retreated around 10,000 years ago, they migrated first north over the Caucasus to the Pontic Steppes (the vast plains in Russia north of the Caspian and Black Seas). It is possible that at this time both male and female lines (R1b1a2 and H2a1) were part of the same population. However, it seems unlikely to me that that continued to be true over succeeding millennia.

Our terminal mitochondrial mutation happened on the Pontic Steppes 6,000 to 7,000 years ago but our ancestors clearly migrated west into eastern Europe and eventually to Britain where they may have arrived with any of the waves of immigrants from the Bronze age to those seeking refuge from Catholic persecution in the 16th and 17th centuries.

We have much more detail for our Y-DNA forebears and the mutations track movement through eastern and western Europe and into Britain between 3 and 4 thousand years ago and the terminal mutation (R-A16) probably occurred in Britain. So, after Stonehenge but a millennium before the Romans.

This exercise has given me different view of our ancestry than I had when I started. In particular the  “re-population from two refuges” theory is very much idealised and what actually happened was more complex with people migrating to Britain from the rest of Europe over many thousands of years. I think it likely that most or all of our ancestry is from the east and they lived in what is now the Middle East during the last ice age.