Friday, 25 July 2014

My letter in Family Tree Magazine about "genetic homeland" stories

The August 2014 issue of Family Tree Magazine includes a letter I wrote in response to two stories in their July issue which claimed that it is possible to identify someone's "genetic homeland" from a DNA test. There was not space in the magazine to include the full text of my letter and the editor has kindly given me permission to reproduce the full letter here.
In your July issue you published two articles which claimed to show that it is possible to pinpoint your "genetic homeland" a thousand or more years ago by taking a simple DNA test. It would indeed be wonderful if a DNA test could give us this information but sadly it is not possible. 
The first story “DNA: find your ancestral home” (p9) referred to a new autosomal DNA test developed by a company called Prosapia Genetics. They claim that they are able to pinpoint someone’s genetic homeland one thousand years ago and that their test is “accurate to home village with a time resolution of the past 1,000 years”. The underlying research on which this test is based was published in a scientific journal (Elhaik et el 2014, Nature Communications 5: 3513). However, the researchers were only able to place 50% of people within 450 kilometres of their country of origin, which is hardly the level of precision claimed. Furthermore the research focused solely on the present-day country of origin and made no attempt to determine an ancestral origin one thousand years ago. Such a method is best thought of as human provenancing – finding the location where an individual’s genotype is most likely to be found – not a method for inferring ancestry. Indeed, if you go back one thousand years you have in theory about 35 billion ancestors, although you actually inherited DNA from only a small, random subset of those ancestors. Identifying a single location as the “genetic homeland” of either all your pedigree ancestors or just the DNA ancestors would be a meaningless exercise. The International Society of Genetic Genealogy received a number of reports from dissatisfied customers of Prosapia Genetics, many of whom had received bizarre results placing their "genetic homeland" in the middle of a river or ocean. Fortunately they were able to get refunds from PayPal. 
Dr Tyrone Bowes, the author of the second article "Routes to roots" (pp14-18), claims to be able to tell his customers when their ancestors arrived in Britain and where they came from based on recurring surname matches received as part of a commercial Y-chromosome DNA test. DNA testing is a very useful tool for the genealogist, and Y-DNA matches can often provide clues about our recent origins. However, Y-DNA results should always be interpreted in combination with genealogical and historical records. While a Y-DNA test is very good at indicating whether or not two people share a recent male-line ancestor, it is much more difficult to determine precisely when or where that ancestor might have lived. For example, a match on 34 out of 37 markers could indicate a shared ancestor who lived 200 years ago or 2000 years ago, and there is no way of determining the precise timeframe. Furthermore, surnames did not become common in Britain one thousand years ago, as is claimed. They were introduced into Britain with the Norman Conquest but the adoption of surnames was a gradual process. While most English people had acquired surnames by the fifteenth century, surnames were not adopted until the nineteenth century in some parts of Wales. In the Highlands of Scotland the clan system survived until the eighteenth century, and people adopted the name of the clan rather than using an hereditary surname. Even when surnames are passed on through the fatherline the link between the Y-chromosome and the surname is often broken as a result of illegitimacy, cuckoldry, adoption or name changes. In addition, there is an inherent American bias in the commercial databases. Consequently Y-DNA matches will often tell us more about non-paternity events in Colonial America in the last 400 years, rather than a person’s origins in the British Isles. All these factors need to be taken into account when interpreting DNA results. 
The mutations that determine haplogroups (the deep-rooted branches of the human Y-DNA tree) did indeed occur at a specific time and place several thousand years ago, but determining when and where that happened is a different matter entirely. We are reliant on making inferences from the DNA of living people. However, the current distribution of haplogroups differs from each of the distributions at different times in the past. The changes occur due to migration and the randomness of genetic drift. As more ancient DNA samples become available it might one day be possible to provide some answers, but we are not there yet and it is unlikely that we will ever be able to "precisely reconstruct our ancestral journey". 
Readers wishing to understand more about the legitimate uses of DNA testing for genealogy and the limited inferences that can be made from deep ancestry tests might like to refer to the new “Debunking genetic astrology” website that I have worked on with my colleagues at University College London: In particular readers might like to look at the page on dubious commercial claims ( where the genetic homeland stories are discussed in greater detail.
The editor also included in the August issue of Family Tree Magazine responses to my letter from Vladimir Makarov of Prosapia Genetics and Tyrone Bowes, the creator of the Irish, Scottish and English Origenes websites. Unfortunately, neither letter addressed any of the issues I'd raised. Vladimir Makarov argued that his reports do reflect ancient origins simply because the reference populations he uses are "totally agnostic of political boundaries". He failed to recognise that these reference populations are themselves possibly admixed and do not necessarily represent the population in the same location one thousand years ago. Tyrone Bowes went into great detail about a case report that he'd compiled for a Mr Valentine which purported to show a link with the McGregors. It seems that his entire dating method has been based on this single case. Numerous scientific studies have of course demonstrated that mutations occur at random, and any attempt to investigate the validity of matches must take into account the uncertainties in the TMRCA (time to the most recent common ancestor) calculations. My points about differing times of surname adoption and non-paternity events remained unanswered.

For further information about Prosapia Genetics and the case reports offered by the Irish, Scottish and English Origenes websites see my two earlier blog posts:

- Driving in the wrong direction with a dodgy DNA satnav (a critique of the test from Prosapia Genetics)
- A look at the genetic homeland case reports from English Origenes, Irish Origenes and Scottish Origenes

© 2014 Debbie Kennett


Unknown said...

Dear Editor,
When one takes a painless commercial ancestral Y-DNA test what one will receive are the names of people (who have also tested) with whom one shares a common male ancestor. BUT the surnames of those people are not random, some surnames appear frequently as genetic matches and what one has to ask is ‘how one can share a common male ancestor with people with different surnames?’ Unless one’s shared male ancestor lived prior to the appearance of surnames (in Britain and Ireland paternally inherited surnames appeared among related males an estimated 1000 years ago). Since surnames arose in an agrarian society, farmers in early census data could still be found farming the lands where their ancestor lived when he first inherited his surname. So if you examine your Y-DNA results, you can identify the surnames that reappear as genetic matches and examine where farmers with those surnames were found in early census data and you will discover an area common to all. The logical conclusion is that that identified area will be where your direct male ancestor lived when he first inherited his surname (approximately 1000 years ago).
It is possible to put a timeframe to Y-DNA STR results and I have based mine on a ‘Valentine’ Case study. Upon Y-DNA testing Mr Valentine was a close genetic match to lots and lots of people called MacGregor (in fact he matched the current chief of the MacGregors at 36/37 Y-DNA markers). The MacGregors are known as the Clan that ‘dare not speak its name’ as that surname was banned in 1608AD after the MacGregors had massacred the Colquhouns (and some spectating priests) at the battle of Fruin Glen. It was illegal to have the surname MacGregor, this meant that anyone who bore the name had to renounce it or suffer death (many were hanged). As a result the Clan was scattered with many taking other names like Valentine. But crucially this gives me a date for a change in surname (approximately 400 years ago). Using this date it means that anyone who is an exact match at 37 markers will have a shared ancestor anywhere within the last 200 years, a distance of -1 at 37 or 36/37 markers (like Mr Valentine and the MacGregor Clan chief) corresponds to a shared ancestor between 200 and 400 years ago. Likewise 35/37 markers corresponds to 400-600 years, and 34/37 markers corresponds to a shared ancestor between 600-800 years ago.
This Valentine Case Study is also an example of a non-paternal event (a simple surname change that has arisen in one’s ancestral past), but that is no hindrance in pinpointing an ancestral origin. For example although Valentine farmers cluster in Angus in Northeast Scotland where Mr Valentine has his roots, the MacGregors, Buchanans, Drummonds and Campbells which are the surnames that appear as Mr Valentines closest and most frequent genetic matches all cluster in the area surrounding Loch Lomond in the Scottish Highlands which is the traditional homeland of Clan MacGregor. With the DNA and historical evidence it is therefore possible to discover what happened in one’s paternal ancestral past. In Mr Valentines case, his MacGregor ancestor based near Loch Lomond was a close relative of the Clan Chief (remember Mr Valentine is a close genetic match to the current Clan Chief) and he may have fought at Fruin Glen. At some point after the ban he decided to go into exile (many of his clansmen were killed for defying the ban). Even today a fugitive flees to a new location and changes his name! This is not ‘genealogical astrology, it is simply putting the pieces together, after all the DNA does not lie.
I would argue that next time when you examine your Y-DNA results rather than look at what particular SNP you are, look at the surnames of the people with whom you share a common ancestor. The full Valentine DNA Case Study can be obtained by emailing me
Dr Tyrone Bowes
Scientist and creator of the Irish, Scottish and English Origenes websites

Debbie Kennett said...

Tyrone, Thank you for sharing your letter with my readers. As I explained in my blog post you cannot extrapolate from the results of a single genealogical case study and apply those mutation rates universally. All the published studies show that mutations occur at random and there is considerable uncertainty in the calculation of TMRCAs. For example, two people who match on 35/37 markers could be a father and son or they could be two men who share a common ancestor 2000 or more years ago.

Unknown said...


I'm puzzled. The leaders of both firms are scientists - at least scientist work there; how can they be so "off" on this issue? I know this is a developing and new science, but from reading what you and others have written, they are missing the A, B, C's of this issue.

Debbie Kennett said...

I don't think either of the two scientists are still actively working in science, and have not been for a few years. However, I'm also puzzled as to how scientists cannot grasp the basics about testing hypotheses, the importance of large sample sizes, the null hypothesis, etc. There have also been a number of peer-reviewed scientific papers where the authors similarly haven't grasped all the basic concepts. I think when it comes to ancestry people get carried away with their own subjective views, and they try and make the evidence fit their theories (what is known as confirmation bias).

Caro said...

As a professional genealogist, currently working very hard to establish the paper trail which links two men with a Y-DNA match, the breathtaking assumptions, sweeping statements and complete lack of historical rigour in Tyrone Bowes' comment make me angry. The current Chief of Clan Gregor is not descended from the clan chief at the time of the battle of Glen Fruin. The proscription of the names Gregor and McGregor was lifted by Charles II in 1654 and not reimposed until 1693. These two facts alone make a nonsense of the fantasy family history Tyrone has constructed for poor Mr Valentine. Not just genetic astrology but voodoo history as well.

Debbie Kennett said...

Thank you Caroline for setting the record straight about the MacGregor surname. You would have thought that for the amount of money being charged for these reports it would be possible to get the basic facts right and to provide citations to back up the statements.

Interpreting DNA matches needs to be done with caution. A father and son can match on 35/37 markers, whereas two people with the same surname who share common ancestors back in the 1600s can match on 37/37 markers. Even when you do match someone with the same surname it's not always possible to find the paper trail link, as the match could well precede the availability of parish registers.

Unknown said...

In his letter of response Dr Bowes advances the fanciful notion that "Since surnames arose in an agrarian society, farmers in early census data could still be found farming the lands where their ancestor lived when he first inherited his surname". 

The concept is fundamentally flawed. In the case of England, historical events led to the transformation of the economy and would have largely broken this genetic link to the past. Dr. Tony Wrigley, noted historical demographer suggests that medieval peasants in the 12th and 13th century may have constituted as much as 70% of the English population. By the mid 19th century, the period from which the Origenes study selects it's "farmer data", the primary sector of the British economy had fallen to between 20% and 30% of the population. This included non agricultural workers, fishermen miners etc. Farmers were a distinct minority.

Where did all the farmers go? They became agricultural labourers as a consequence of the enclosure movement. By the mid 19th century agricultural labourers out numbered farmers in the south of England by a factor of 10 to 1. In the 13th and 14th centuries they became weavers as the cloth trades emerged as an economic powerhouse. They became shop keepers and members of guilds in the burgeoning cities towns and villages as England urbanized, and they became factory workers during the industrial revolution. Importantly they carried with them the genetic signatures that YDNA test takers find in their FamilyTree DNA reports.

Why is this important? The vast majority of Origenes case studies involve widely distributed surnames, ones with multiple, in fact scores of unique geographic origins. In most of these areas the man/land relationship had been severed for generations. These areas are not taken into account as they are not populated with farmers bearing the surname of the test taker. Dr Bowes falls into the trap of evidence by exclusion.

A case in point, the English Bennett case study. Dr. Bowes Identifies two farmer clusters of the Bennett surname where matching surnames overlap. Somerset and York/Lincolnshire. The latter is dismissed as it "is only in Somerset that farmers with these surnames cluster in close proximity". There is always a high risk associated with superficial spatial analysis. A  detailed study would identify 58 distinct clusters of the Bennett surname plus surname matches scattered from one end of the country to the other. Any one of these areas might include the ancestors of the Bennett YDNA test taker. The adage "the absence of evidence is not evidence of absence" is appropriate in this context. Farmers constituted approximately 5% of the population. By the mid 19th century the vast majority of genetic links to the past thru farmers had been extinguished. However the remaining population carry the genetic signature of their peasant ancestors and it is thru them that one would be most likely to trace a genetic homeland.


Debbie Kennett said...

Thank you Howard for that very useful summary of the historical situation. You make some very important points.