European genealogy FAQ

These FAQs represent some of the non-technical questions that we’ve been asked about our paper
Ralph P, Coop G (2013) The Geography of Recent Genetic Ancestry across Europe. doi:10.1371/journal.pbio.1001555 PLOS biology
there’s a synopsis of the paper here.

Peter Ralph and Graham Coop

1. How did you learn about genealogical ancestry from genetics?
2. How did you find out when these shared ancestors lived?
3. What did you find?

Questions about the interpretation of genealogical ancestry and our results.
4. If all Europeans share the same set of common ancestors 1000 years ago, how can there be variation in the number of shared ancestors?
5. If you and I share all of our common ancestors 1000 years ago, why are we not genetically (almost) identical?
6. I know from the work on mtDNA that we all share a common ancestor tens of thousands of years ago, and yet you say that we all share a common ancestor only a few thousand years ago. Is one of these facts wrong?
7. How can it be that a person from the UK has more distant cousins in Ireland than in the UK?
8. What about recent immigrants to Europe?
9. How long ago did the most recent common ancestor of all modern humans live?

Questions about historical interpretations
10. Why do people on the Italian and Iberian peninsulas have relatively fewer common ancestors with other populations than other European populations?
11. Could population movements in the past 100 years explain the higher levels of sharing in some parts of the world?
12. Can your results about (insert population) be explained by (insert historical fact)?

Questions about our work and personal genomics results.
13. How does this relate to who my mitochondrial/Y-chromosomal haplogroup says I’m related to?
14. How can personal genomics companies (e.g. 23andme) place a European on the map of the Europe, if all Europeans are related to each other just a thousand years ago?
15. Personal genomics companies identify genetic connections (via shared genomic regions) between distant cousins. How long ago does this connections date too, and what do they mean in light of your results?

How did you learn about genealogical ancestry from genetics?

In a nutshell, people are relatives if they share ancestors; and distant relatives sometimes (not always!) share long chunks of genome that they’ve both inherited from their shared ancestors. So, looking at genome sharing tells us about shared genealogical ancestry.

For example, first cousins share about 1/8 of their genome due to inheritance from their shared pair of grandparents. The shared genome is not scattered — it comes in long “chunks” — the more recent the ancestor, the longer. We call these chunks “identical by descent” (IBD).

In this cartoon we show three generations of a family tree relating two cousins. We’ve colored all six grandparents’ chromosomes so that we can track how those chromosomes have been passed down to the cousins at the bottom of the figure. You can see that at most parts of the chromosome, the cousins have usually inherited from different grandparental chromosomes, but in one region they share a region inherited from the red grandparental chromosome.

More distantly related individuals usually share less material. For example, second cousins only share 1/32th of their genome, since they only share one pair of great grandparents. More distant relatives share even less of their genome, and people who share an ancestor longer ago than eight generations usually haven’t inherited any genetic material in common from that ancestor. You have a lot of eighth cousins, though, so there’s still a lot of them you do share genome with.

If distant relatives do share a chunk of genome, we can hope to detect it, since they look more similar over that chunk than usual. So, we looked for long blocks of shared genome in a sample of about 2000 people from all over Europe.

How did you find out when these shared ancestors lived?

Chunks inherited from more distant ancestors tend to be smaller than those inherited from recent ones (since there’s been more time for recombination to whittle them down). Once we found these shared chunks (about a million of them), we used their lengths to estimate how long ago they came from (in other words, how long ago the shared ancestors lived). Because of this, we’re quite certain that the chunks of shared genome we find are all from ancestors living in the last 4,000 years, and mostly from the last 3,000 years.

What did you find?

There is a lot of information in the data. The main conclusions we came to are that: everyone is related, surprisingly recently; and there are regional differences in patterns of relatedness due to historical events.

Ubiquitous shared ancestry: We found that even people living on opposite sides of Europe are genealogically closely related to each other over the past thousand years. Even pairs of people as far apart as the UK and Turkey share a chunk of genomic material 20% of the time. Since the chance that two people inherit genetic material from any one shared ancestor from 1,000 years ago is incredibly unlikely (<10^-10), to explain such sharing we need these pairs of individuals to share many ancestors. In fact, they need to share a number of ancestors that is far larger than the size of European population, indicating that any pair of individuals share as ancestors all of the individuals alive back at the time in Europe, each many times over.

This strange idea that everyone is everyone’s ancestor was actually predicted about ten years ago by Joseph Chang (and collaborators) using maths and simulations. In hindsight this is intuitively clear, due to the rapidly expanding number of ancestors you have as you go back further and further in time. You have 2 parents, 4 grand-parents, 8 great-grandparents, and so on doubling every generation. After k generations you have 2^k ancestors, and this number grows so quickly that just a thousand years back (~30 generations) you have roughly 1 billion ancestors, which is far larger than the population size of the Earth (let alone Europe) back then. The consequence is that anyone alive 1,000 years ago who left any descendants will be an ancestor of every European. While the world population is larger than the European population, the rate of growth of number of ancestors quickly dwarfs this difference, and so every human is likely related genealogically to every other human over only a slightly longer time period.

Striking regional variation in relatedness:

The most obvious pattern is that the number of ancestors shared by two people decreases the farther away the two people live. This makes a lot of sense: someone from the UK is on average more related to someone else from the UK than to a German, and is more closely related to a German than someone from Greece.

Patterns of relatedness reflect historical population demography (population sizes and migration rates). Two people from a historically small population will tend to be more closely related to each other than two people from a larger population. This is because in a smaller population, there are fewer possible ancestors, so two people’s ancestors are more likely to overlap. Migration also affects things: people from populations that have exchanged many migrants will be more closely related than if there was a lower level of migration. Therefore, we should, and do, see differences across populations.

For example, we find that Italians are more distantly related to each other, than individuals in most other regions of Europe. We think that this reflects larger and more stable population sizes, perhaps drawn from a larger geographic area, over the past 2000 years.

We also see that people in Western Europe are somewhat less related to each other than are people in Eastern Europe. Two people from across Eastern Europe share more ancestors with each other, than is typically seen even within a Western European country. This higher rate of shared ancestry seems to come from ancestors living 1,000-2,000 years ago. We think this may reflect the expansion of various groups of people (perhaps the Slavs) in Eastern Europe during the migration period of European history. But this is just a hypothesis, there are other (non-mutually exclusive) explanations.

If all Europeans share the same set of common ancestors 1000 years ago (and likely many shared ancestors more recently), how can there be variation in the number of shared ancestors?

You can be related to the same ancestor multiple times. For instance, someone could be your great, great, great, great, great grandfather on your mother’s side and also on your father’s side. Because of this, you and I can share the same ancestral individual as a common ancestor many times over. People who share more common ancestors have more overlap this degree of relatedness. We can measure this difference through degree of shared genome, since even when everyone is a common genealogical ancestor, not everyone is a common genetic ancestor.

If you and I share all of our common ancestors 1000 years ago, why are we not genetically (almost) identical? (or, How can it be that things like blonde hair and blue eye color occur with different frequencies in different European populations?)

We can share all of our common ancestors and yet not be genetically identical due to the randomness of how the genome is inherited from parent to child (the process of Mendelian segregation that happens in meiosis). This is why sisters, who share both their parents (and all the rest of their ancestors, too), are not genetically identical. They can –and do– inherit different genetic combinations from their shared parents.

The same idea can play out over much larger time-scales. Two Europeans share all of their ancestors from only 1000 years ago, but have each inherited most of their genomes from distinct ancestors (the same is true for individuals worldwide over slightly longer time-scales). Someone from Spain and someone from Sweden might share all of their ancestors from 1,000 years ago, but the Spanish person has probably got most of their genome from ancestors living near modern-day Spain, far from where the Swedish person got most of their genome.

What about mitochondrial Eve? I know from the work on mtDNA (or the Y chromosome) that we all share a common ancestor tens of thousands of years ago, and yet you say that we all share a common ancestor only a few thousand years ago. Is one of these facts wrong?

Both facts are correct, they are just telling us about different aspects of shared genealogy. Mitochondrial are inherited from the mother, so their inheritance only traces up one of the many possible branches of someone’s genealogy (the maternal line). You and I can have all of our common ancestors in common at a particular time, but we can (and likely will) have different maternal lineage ancestors at that time and so not share a mtDNA ancestor from then.

How can it be that a person from the UK has more distant cousins in Ireland than in the UK?

In general individuals share (slightly) more recent common ancestors with other individuals from the same country than with individuals from another country. But there are some exceptions, such as the UK, where individuals share more recent common ancestors with individuals in Ireland than with other individuals in the UK. (An Irish person is more likely to be related to another Irish than to someone from the UK, though.). Another example is that a German individual has more distant cousins in Poland than in Germany.

We were quite surprised by this result. However, we think it can be explained by migration rates and population size changes between countries. For example, over the past few centuries there has been a great deal of migration between the UK and Ireland, particularly of people of Irish ancestry into the UK. Because of this, many UK individuals have recent Irish ancestors (see for example). This was happening at the same time as population growth in both places. Since the population size of Ireland was smaller than the UK, an Irish ancestor is more likely to be related to a modern-day Irish person than an ancestor in the UK is to be related to a modern-day person in the UK.

What about recent immigrants to Europe?
As we explain in the paper, our sample of “Europeans” are individuals with all 4 grandparents born in the same country, and individuals who don’t fall too far outside the range of European genetic variation. This is obviously a narrow view of what it means to be a European. However, this was a necessary step to allow us to explore the impact of historical events happening within Europe in the last 3,000 years (to explore other geographic regions would other samples from those regions). We used the convenient shorthand “Europeans” as it avoided the circumlocutions about the ancestors of people who lived in Europe during…. In the discussion of the paper we talk about the issues of sampling more carefully.

Obviously, immigrants from all over the world are an important part of what it means to be European. In fact, these sorts of results make it clear why that should be true.
Even immigrants from across the world in the last generation will [likely] be related to everyone in Europe from not too much longer ago. Rohde,
Olsen, and Chang determined that probably even someone of Native American descent would share ancestors with modern Europeans from just ~3000 years ago.

How long ago did the most recent common ancestor of all modern humans live? (or, could native Australians, Amazonian tribes people, and Europeans really share ancestors in the last few thousand years?)

Rohde, Olsen, and Chang have approached this question better than we can, and come to the answer of “sometime in the past 3,500 years”, and maybe more recently. Their work depends on (weak) assumptions about human migration, and our data support their conclusion.

Many people object to this, arguing that modern-day isolated populations must be an exception. But it doesn’t take very much migration: just one “outsider” who bore a child into an isolated group sometime in the last few thousand years would suffice. Another example from Rohde, Olsen, and Chang’s paper: their simulation assumes 10 migrants across the Bering strait every generation; but decreasing this to one migrant every 10 generations (300 years!) only increases their estimate by a few hundred years.

Why do people on the Italian and Iberian peninsulas have relatively fewer shared common ancestors than other European populations?

We’re not totally sure, but we have some ideas. First, geography has probably contributed to both cases: both are peninsulas partially separated by mountain ranges, and less migration results in less shared ancestry. This might explain the Iberian peninsula, but it doesn’t seems sufficient to explain the level of common ancestry with Italians (who also share less ancestors with each other as well as with other Europeans).

We also know that Italians share relatively fewer common ancestors with each other than European populations over the past 2,000 years ago. Those past 2,000 years have seen a number of population expansions within Europe, and after a population expansion you expect to see a lot of shared ancestors, since a large group of people then trace their ancestry back to a small one. This suggests that Italians might also have less sharing of common ancestors with other Europeans because the peninsula was less demographically affected by those population expansions of the last 2,000 years (i.e. their population has been large and relatively stable over a long time period).

One initially plausible explanation of the somewhat lower relatedness of people on the Italian and Iberian peninsulas to other Europeans due to migration from Africa. While the African contribution into Iberia and Italy has likely reduced the amount of sharing with the rest of Europe, as it contributed only a few percent of the ancestry of modern genomes (Moorjani et al., 2011), it cannot explain the size of the reduction that we see.

Could population movements in the past 100 years explain the higher levels of sharing in some parts of the world?

Probably not. The mobility of individuals has increased greatly over the last century; in addition, there were also a number of population transfers that could conceivably increase relatedness in some areas of Europe more than in others. While increased mobility would “blur” any geographic patterns, it is unlikely to increase or decrease levels of relatedness. This is because although it is much more likely today than it was 100 years ago that someone from Poland will marry someone from Portugal, for this to have a noticeable effect in our data, we’d have lots of people related on this time scale — second and third cousins, say. But, the chance of finding third cousins in a random sample of this size is far too small. (There are a few pairs of siblings, but we assume that wasn’t random chance, and removed one of each pair.) Because of this, we think that our results about increased relatedness in particular areas are due with events much deeper in history.

Can your results about (insert population) be explained by (insert historical fact)?

Well, maybe. Europe has seen a lot of events over the past three thousand years that could have impacted demography. In the paper we do discuss a few possible explanations of some broad patterns. However, in general we haven’t applied our results to more historical questions in part because it is too daunting: there are just too many plausible hypotheses, many of which are not mutually exclusive and are layered in complicated ways. Also the degree of uncertainty on the dates of shared ancestors makes resolving these things hard. Obviously these types of data could shed some light on specific hypotheses, but this will need a lot of careful work in collaboration with historians and anthropologists.

how does this relate to who my mitochondrial/Y-chromosomal haplogroup says I’m related to?

Your mitochondria and your Y chromosome (if you are male) are inherited through your maternal and paternal line respectively. Because of this, they tell you about only a small fraction of your ancestry — one or two ancestors at any time at most! — and your genealogical connections to a small set of people. A a simple example, if you and I are first cousins because my mum and your dad are siblings, we will share neither our mtDNA (nor Y-chromosome haplogroup) but we are still closely related. Our results demonstrate a similar idea on a much deeper time scale. We show that Europeans are related to each other in many ways from a thousand years ago, but a European will only be related to a much smaller group of people through their maternal lineage over that time-scale (the people they share their mtDNA haplogroup with; and similar for the Y chromosome).

As such while results about your Mitochondrial/Y-chromosomal haplogroup are interesting, but they tell you about only a small slice of your family history, when in fact the family of a modern European will be drawn from all over Europe over the past one thousand years. There’s a very good discussion of this here at Sense About Science

Personal genomics companies (e.g. 23andme) can use genome-wide data to place the genetic data of a European on the map of the Europe. How can they do this if all Europeans are related to each other just a thousand years ago?

A European individual is related to everyone in Europe by ancestors in the past thousand years ago or so (and likely to everyone in the world within the past three thousand years). However, a particular individual is not related to everyone equally. Some of these individuals will be cousins many times over, through many different routes through their family tree, while some will be a cousins fewer times over. When these companies position a person on a map (usually a principal components map), they are looking at your average genetic similarity, which summarizes the average of these relationships. For example, if you have relatively more close relatives in the north of Europe than the south you will be positioned more in the north of Europe.