Concepts – Endogamy and DNA Segments

Members of endogamous populations intermarry for generations, creating many segments that match, especially at small centiMorgan levels. These matching segments occur because they are members of the same population – not because they are genealogically related in a recent or genealogical time-frame.

Said another way, endogamous people are all related to each other in some way because they descend from a small original population whose descendants continued to intermarry without introducing people outside of the community into the genetic line. In other words, the DNA segments of the original population simply keep getting passed around, because there are no new segments being introduced.

If you only have 10 segments at a specific genetic location to begin with, in the original population – then the descendants of those original people can only have some combination of the DNA of those original people until another person is introduced into the mix.

Examples of endogamous populations are Ashkenazi Jews, Native Americans, Acadians, Mennonite, Amish and so forth.

If you have some family lines from an endogamous population, you’ll match with many members of that group. If you are fully endogamous, you will have significantly more matches than people from non-endogamous groups.

I suggest that you read my article, Concepts: The Faces of Endogamy to set the stage for this article.

In this article, I want to provide you with a visual example of what endogamy looks like in a chromosome browser. It doesn’t matter which vendor you use so long as you can drop the cM count to 1, so I’m using FamilyTreeDNA for this example.

I’ve used three people as examples:

  • Non-endogamous European
  • Ashkenazi Jewish
  • Native American (Sioux)

For all testers, I selected their closest match above 200 cM total plus the following 4 for a total of 5 people to compare in the chromosome browser. I have only shown chromosomes 1-8 because I’m trying to convey the concept, not exact details of each chromosome, and 8 chromosomes fit into one screen shot.

If you’re not familiar with the terminology, you can read about cM, centiMorgans, in the article “Concepts – CentiMorgans, SNPs, and Pickin’Crab.”

Let’s take a look at our 3 examples, one at a time.

Non-Endogamous European Individual

The tester is non-endogamous. Four of the 5 individuals are known family members, although none were target tested by the tester.

Endogamy non-endogamous.png

The tester’s matches at 1 cM are shown below:

Endogamy non-endogamous 1cM.png

Note that the grey hashed regions are regions not reported, so no one matches there.

Below, the same 5 matches shown at 7 cM where roughly half of the matches will be identical by chance. Identical by descent segments include identical by population. You can read about the various types of “identical by” segments in the article, “Concepts – Identical by…Descent, State, Population and Chance”.

Endogamy non-endogamous 7cM.png

Ashkenazi Jewish Individual

The tester, along with both of their parents have tested. None of the matches are known or identified relatives.

Endogamy Jewish.png

Even though none of these individuals can be identified, two are related on both sides, maternal and paternal, of the person who tested.

In the chromosome browser, at 1cM, we see the following:

Endogamy Jewish 1cM.png

At 7cM, the following:

Endogamy 7cM.png

Native American Individual

The tester is 15/16 Native from the Sioux tribe. It’s unlikely that their matches are entirely Native, meaning they are not entirely endogamous. None of the matches are known or identified family members.

Endogamy Native.png

At 1 cM shown below:

Endogamy Native 1cM.png

At 7 cM, below:

Endogamy Native 7cM.png

Side by Side

I’ve placed the three 1 cM charts side by side with the non-endogamous to the left, the Jewish in the center and the Native, at right.

endogamy side by side.png

It’s easy to see that the Jewish tester has more 1 cM segments than the non-endogamous tester, and the Native tester more than both of the others.

Summary Comparison Chart

The chart below shows the difference in total number of segments, number of segments between 1 and 6.99 cM, and number of segments at 7 cM or larger. I downloaded these results into a spreadsheet and counted the rows.

Total Segments Total segments at 1 – 6.99 cM Total at 7 or > cM % 7 or >
Non-Endogamous 98 70 28 29
Jewish 168 139 29 17
Native American 310 295 15 5

You’ll note that the non-endogamous individual only has 58% of the number of total segments compared to the Jewish individual, and 32% compared to the Native American individual. The Jewish individual has 54% of the number of segments that the Native person has.

I was initially surprised by the magnitude of this difference, but after thinking about it, I realized that the Native people have been endogamous for a lot longer in the Americas than the Ashkenazi Jewish people in Europe. At least 12,000 years compared to roughly 2000 years, or approximately (at least) 6 times longer. Furthermore, the Native people in the Americans were entirely isolated until the 1400s, with no possibility of outside admixture. Isolation lasted even longer in the tribes that were not coastal, such as the Sioux in the Dakotas.

Note that the Jewish person and non-endogamous person have almost as many 7cM segments as each other, but the Native person has roughly half as many when compared to the other two. That means that because I made my selection starting point based on total cM, and the Native person has a LOT more 1-6.99 cM segments than the others, at that level, there are fewer strong segment matches for the Native individual.

The Native person’s percentage of 7 cM or greater segments is a much smaller percentage of the total segments.

As a percentage, the 7 or greater cM segments are 29% of the non-endogamous person’s total, 17% of the Jewish person’s, but only 5% of the Native person’s total.

Endogamy not only makes a difference when comparing results, but the specific endogamous population along with their history, how heavily endogamous they are, and how long they have been endogamous appears to factor heavily into the comparison as well.

49 thoughts on “Concepts – Endogamy and DNA Segments

  1. I was curious as to the results I’d get. My paternal grandparents were 3rd or 4th cousins (according to my father’s results on Gedmatch, but not verified yet by research) and there are a few other endogamous matches on my tree. My results: total segments: 106; segments from 1-6.99, 84; segments 7 or greater, 22; percent greater than 7: 21%. Which puts me between “non-endogamous” and Ashkenazi Jewish, except for the total segments 7 or greater, which is between Ashekenazi and Native American. I guess this pretty much makes sense? Interesting exercise!

  2. Thank you for that. Information on kits from endogamous populations is always welcome. They are quite a challenge. I manage a kit this is from an Eastern European Romany population. Among the top 5 matches on FTDNA is a match sharing 249 cM, with the longest block being 9 cM. When I sort by longest block there are five kits in the 20-25 cM range and the rest of the matches have less than that as the longest block. Endogamy is also noted in the number of matches on any given ICW list. There are only 217 matches in total. When I select the first kit and choose !CW, I get 181 matches. That’s about 83% of the list. I suspect poverty and other socioeconomic related factors figure into the low number of matches. However, there has been so much growth in this field in such a short time that I think more inroads will be made in the ability to make good use of the data. Even now, adjusting the parameters on the new clustering tools is helping to tease out a few clusters from the one large block.

    • I don’t know if you have transferred to MyHeritage but they have improved the Jewish clusters. Might work with yours too.

  3. As a person who found out through DNA testing that I am half Askenazi Jewish (with a different father than I expected), I sincerely appreciate your explanation of endogamy here. I started following your blog several months ago while trying to find out who my father was and have really enjoyed reading your many articles on this subject, as well as others.
    This is probably the best comparison/explanation that I have seen, so a big thank you for your efforts!
    P.S. I have found out who my father was as well as finding 3 half-brothers, who are wonderful! Quite a surprise for someone in their 60’s who was an only child!

  4. Endogamy is definitely something I have to consider when I check out my mother’s DNA matches. Her lines go into Quebec and Acadia. When I was filling out my family tree, I noticed a lot of my French Canadian lines were being duplicated. It wasn’t a lot. But, I’ve noticed several connections to a guy in the 1600s named Etienne Racine. There were a few others too. It was pretty interesting. I swear I have half of the town of Trois Rivieres in my tree.

    But, it’s okay. These things happen and everything was sorted out on WikiTree. I think I might have some on my Italian lines back that far. I have to check. I know I have some on my paternal grandmother’s side.

    With my mom’s tree the way it is, I suspect endogamy is the reason why some people (not all) appear closer than they are. Like if one guy is from my 3x great-grandparents and is in the same cM range as someone else, that other person may be further back. That’s what it seems like. Dunno if that makes sense. =)

  5. Perhaps you covered this question in your previous posts (to which you referred in this article) but I will ask anyway . . .

    Many somewhat isolated rural North American (and likely other) populations, particularly from the colonial period through the mid-19th century, had limited marriage opportunities for several generations. Examples I am familiar with include the early settlers of western Massachusetts and the Connecticut River Valley or even their eventual descendants, the New England Planter emigrants to Nova Scotia (those who replaced the Acadian population prior to the Revolution). While these populations eventually became exposed to a wider genetic pool, the genealogical record of repeated intermarriage is obvious. But at what point can these populations be considered endogamous. Or perhaps the question should be, at what point do they cease to be considered endogamous?

  6. I loved they way you outlined this ‘teaching moment’, and how effective it is. It was me playing with chr Browser, using the ‘lesser’ cM that helped me figure out our unique distant match. Later, I found I also was not related to my ‘step’ father, and that my 51.4% Ashkenazi was from one male, my paternal father (who I have never found). But using this process, I have been able to sort out some of my Ashkenazi questions. I recently ran my Cousin in law’s last name, and came up with a match to her 1st C 1 X R. It showed me that I am 1st – 5th C, which I thought odd. Using the lesser cM’s, I could see so much more, that this particular post of yours, is right on. I have paid for a kit at FTDNA for my cousin in law, where I can then compare all three of us to figure out if our match is on the paternal side or grandmother’s side. This process is helpful in more ways than I can say, and it works. Thank you, Roberta!m – A

  7. Roberta, ALL Jewish communities are endogamous. Not only Ashkenazi, but also Sephardic and Mizrahi. The only one that is not is the Ethiopian which has extremely strict taboos on marrying any where within numerous generations. So many genealogists and genetic genealogists reference Ashkenazi endogamy – when it should be JUST Jewish.

  8. I am looking at a ancestryDNA matches of a woman searching for her birth father. Her mother is of Polish descent.
    The relationship between her mother and father was not endogamy.

    Her closest paternal matches come from a region in Mexico that has a lot of endogamy. Of her top 5 matches, EACH match has maternal and paternal lines that lead back to the same couple in the mid-1850’s.

    Her closest match is 459cM. The match’s maternal great-grandparents and paternal great-great grandparents are the same couple. Would this match fit into the predicted relationship range or not?

  9. Wonderful visuals for endogamy and the comparative stats help clarify. It would be interesting to use the data from Lara Diamond’s survey
    I have worked with a couple of clients whose roots are in the German immigrants to the U.S. in the 1800s who chain migrated from intermarried populations of small villages in Germany and settled and often remained in the same towns and surrounding areas in the U.S; Iowa and Illinois specifically. I had not expected to find endogamy. I’ll be doing this exercise to see the impact, thanks for the inspiration!

  10. Not sure why anyone would want to see anything below 5cM. I remember when Bennett first looked at my mother’s top match (back in 2014) and first thing he did was ask if it was my mother’s aunt/niece as that was the predicted relationship 1st cousin aunt/niece 620cM, longest block 10cM. I said no, and that match is Maori, and we’re Hawaiian, and we don’t share any recent connection except that which occurred 800 yrs. ago.

    But today, comparing both my mother & her 2nd cousin (who isn’t admixed)

    Predicted Relationship 1st Cousin Aunt/Niece
    749cM (longest block 16cM)

    2C’s TOP MAORI MATCH (same top Maori match as my mother)
    Predicted Relationship 1st Cousin Half-Sister, Grandmother/Granddaughter, Aunt/Niece
    1033cM (longest block 18cM)

    That same person matched to me (but not my top match)
    Predicted Relationship 1st Cousin – 2nd Cousin
    403cM (longest block 13cM)

    This link shows my mother’s top 5 MAORI matches. I purposely picked people (Maoris) we know that we have no RECENT connection, at least not within the last 8 centuries.

    And of course we have pages of these very high matches. And dropping it down to 1cM makes it look like missing teeth on a comb, as many of the segments are very close to each other.

    • Hi Kalani. I wasn’t suggesting that people use 5cM and below and I hope no one interpreted it that way. Just using this as an educational tool. Your Hawaiian results are fascinating. Thanks for sharing.

  11. Another population dealing with endogamy are Romanichal. In my own paternal lines I have basically 4 families that married into each other over and over again, with first cousin marriages being common. It makes for some headaches for sure when doing the tree.

  12. This was very useful. I am descended from a few European-North American communities that are small and rural and composed of the same families intermarrying over a couple of hundred years, so I wondered if I needed to be concerned about endogamy. I ran the same test you did on several DNA matches from one of these communities, and found percentages very close to those you got for the non-endogamous European. I was happy to lay that concern to rest. Now I have some idea how to evaluate other ancestral lines for which I might suspect endogamy. Thank you for sharing this.

  13. I just checked a group of relatives against my father. Using the chromosome browser at FTDNA, I selected seven first-second and second-third cousins related to a pair of my dad’s grandparents. It is interesting because his grandmother’s father was from Ireland and mother of French Canadian ancestry, but his grandfather’s people had been in New Brunswick since the end of the American Revolution. They met and married in an area of New Brunswick that does have a lot of endogamy. This time the chromosome painting looks a lot more like the Ashkenazi Jewish chart with a setting of 1 cM than my earlier test using myself did. I don’t know exactly how to interpret it, but it is a fascinating way to think about DNA segments, even very small ones. This comparison has been helpful too because I have been able to identify some X DNA and DNA specific to my great grandmother’s parents to map on DNA Painter. Just wanted to share.

  14. My birth-father’s family is about 80-90% French Canadian, and I’ve been able to trace over 300 lines of descent back to the original emigrants. The endogamy is absolutely incredible ! This results, as you mentioned to ‘Cferra’ above, is that Ancestry reports cousin matches closer than they actually are. One purported 4th cousin of mine shares an amazing 66 surnames with me, and 4 “shared ancestors”, but he is fact no closer than a 7th cousin. I believe all French Canadians are cousins, usually several times over. And we all know about Acadian endogomy 🙂

  15. If the closest match is 72 cM, does this comparison still work for determining the amount of endogamy?

  16. Am curious, what does it show when comparing known relatives in the endogamous population vs non-endogamous? Graphically, this may not be possible as you want to not include the large segments of close relatives. So may have to do it via analysis of the data download. And then maybe adjust figures for amount removed by the large segments. (For example, if 20% of area due to removed large segments, then may have 10% less small segments to count as they got masked?) Does the pattern still hold (after adjusting for large segments with known relatives)? Did my explanation make sense?

    Didn’t Leah have a rule of thumb based on size of segments? How does that compare in your samples?

    I have long seen patterns of only small segments in the Parsi population matches and will run this analysis on them to see if similar. In Parsi’s (most of which have not married outside the community yet), almost everyone is a 100-200cM total match. But known relatives show expected match strengths (above and inside this noise range) and larger segments.

  17. This article was very interesting! All four of my grandparents were Ashkenazi Jews from Eastern Europe. I just compared my results with my two sisters at 1 cm. I saw very few really short segments. However, if I compared my results with my other top matches who are either distant cousins or not known relatives, then I see several short segments at 1 cm, as you described in your article.

  18. Another interesting finding. One of my 3rd cousins, once removed shows up as my 4th best match at 204 cm and 29 for the longest block. However, her brother appears way down in the list as a 4th to distant cousin, matching me at 135 cm and 11 for the longest block. Guess this is the throw of the genetic dice in ONE family.

  19. Roberta,

    Thanks so much again for you helpful DNA analysis. My father’s family were Mennonites
    who originated in Holland and then migrated to what is now Poland and then at the
    invitation of Katherine the Great moved to Ukraine to grow wheat for the Russians.
    Each move was basically a village relocating to a new site and the same was true
    when the village in Ukraine relocated to Nebraska. (the czar went nuts and the land that the railroad had received for building the RR from Omaha to Sacramento was available
    in NE) For those folks who are connected to the Mennonite Tradition, I would recommend
    Tim Penner’s Mennonite DNA project, Tim has been
    able to research most of the Mennonite surnames via Y DNA and in my case suggest a village in Holland as an ancestral home. Most of my cousins in Ancestry are from the
    Mennonite tradition and most of them show up as being much more closely related
    than they actually are. Thanks again, Bill

  20. Can I assume that an endogamous Jew who shows matches with relatively high cM numbers is not especially closer on a typical tree than a Non-endogamous European with lower cM matches? That is, do high cM values fool us into thinking a match is closer than it appears if the group is endogamous?

    • Yes. To some extent, Family Tree DNA has an algorithm that adjusts for that in Jewish populations, but not in other populations.

      • Do they do this based on self-identification? I wonder if this algorithm would work on a eastern European Romany kit. It would be interesting to see the impact that might have on the results.

    • In my case, sometimes the match for a real cousin is closer OR more distant than is the actual relationship.

      If the person is half Jewish, do the algorithms adjust for that?

  21. Pingback: Honoring Veteran USMC William Tully Brown, Navajo Code Talker | DNAeXplained – Genetic Genealogy

  22. Fascinating stuff. I have one Jewish grandfather, and FamilyTreeDNA rates me as 28% Jewish, with the rest other Europeans. I come out with 10500 “matches” and lots of endogamy when I look on a 1cM, but the closest relationships are 90 cM with the longest block 19 (others have one or the other higher). None of the “closest” relatives in the rank have ancestral surnames that match, but most don’t list many surnames and I only know two surnames on my Jewish line. My wife took the same test and came essentially entirely European, with no Jewish ancestry (or any other known endogamy). She has about 2600 matches but has real first and second cousins she knows and that she can match by surname. Is that the kind of difference you would expect between endogamous and non-endogamous individuals?

  23. Incredibly clever! We had turned off matching for all but immediate relatives because I was sick of trying to explain “no, you are not a close cousin, FTDNA cannot do correct relationship estimates” to people over and over. Most people were nice, but a few were hostile, demanding personal information. Even phoning. We turned matching back on temporarily to do this.

    As we have family members from 2 completely different indigenous groups I picked one from each. Why two groups with no geographical or linguistic proximity – coincidence. We have an eclectic family.

    One sample is from a Central American ethnic group, the other from a South American group. Ethnicity estimates are from FTDNA, but uploads to other sites produced similar estimates, neither is very much admixed. Haplogroups from Yfull.

    Central America New World 99% Trace Siberia 7 cM |% >7
    C.A. 100% indigenous 515 502 13 2.52%
    S.A. 100% indigenous 438 430 8 1.83%

    no segment was exactly 7 cM

    7/13 of the larger segments for the first sample were from the 5th match who is likely a genuine relative, at least he is from the country.

    The breakdowns:
    Central America New World 99% Trace Siberia <1% D1* Q-Y128700
    1st Cousin – 3rd Cousin 391 9 97% indigenous
    1st Cousin – 3rd Cousin 345 9 100%
    1st Cousin – 3rd Cousin 302 6 100%
    1st Cousin – 3rd Cousin 274 9 85%
    1st Cousin – 3rd Cousin 273 34 45% (probably an actual relative)

    South America New World 100% (FTDNA estimate) B2b-a1a8a* (Q-M902)
    1st Cousin – 3rd Cousin 251 8 unk
    1st Cousin – 3rd Cousin 251 5 unk
    2nd Cousin – 3rd Cousin 235 9 97%
    2nd Cousin – 3rd Cousin 216 8 91%
    2nd Cousin – 3rd Cousin 191 9 unk

    I have the segment screenshots if you want to see them. Very dramatic.

    • Segments from complete chromosome sets including X

      1 Central America – New World 99% Trace Siberia D1* Q-Y128700
      2 South America – New World 100% (FTDNA estimate) B2b-a1a8a* (female)
      3 US Deep South Colonial – European 93% T2b4b R-L21
      4 US Mid-Atlantic Colonial – European 96%T2b46 (female)
      5 US African American Upper South – African 72%; European 23% L3d1b2 E-Z1720*
      6 US African-American – Southeast – African 77%; European 21% mt unk (female)

      total segments || 1-6.99 cM || >7 cM |||| % >7
      1 CA 100% indigenous 515 502 13 |||| 2.52%
      2 SA 100% indigenous 438 430 8 |||| 1.83%
      3 US 93% european 114 58 56 |||| 49.12%
      4 US 96% european 87 71 16 |||| 18.39%
      5 US 72% african 22 14 8 |||| 36.36%
      6 US 72% african 51 17 34 |||| 66.67%

      7/13 of the larger segments for the first sample were from a likely cousin, all matches to the “Europeans” were to likely cousins
      No segments were exactly 7 cM

      I’ve added 2 unrelated (biologically) “Europeans” and two African-Americans. Except for one more unrelated “European”, I’ve run out of independent samples within our family. Everyone else is a bio-relative of one or more of these 6. I’ve not done any statistics tests to see if the apparent differences, quite amazing in the screen shots, are real.

  24. I needed to proofread that.
    sample 1 is Central American estimated New World 99% Trace Siberia <1%
    Sample 2 is South America estimated New World 100%

    in both cases, few of the 5 "matches" are even from the same country.

    • I began to think — was the “total number of segments” for all chromosomes or only 1 through 8? The above figures are totals for all chromosomes. For 1 – 8 only:
      —Central American Indigenous:
      segments total 288
      over 7 cM 5 (4/5 are from the probable real cousin)
      less than 7 283
      % over 7 1.74%
      —South American Indigenous:
      segments total 222
      over 7 cM 3
      less than 7 219
      % over 7 1.35%

  25. Roberta, Thank you for sharing this wonderful, visual explanation! I do have a question… you started out by saying “For all testers, I selected their closets match above 200 cM total plus the following 4…” For the first two, this is what it appears you did. But, for the Native American individual, you chose the following: 209, 205, 205, 201, and 199. Can you explain? Thanks!

    • I had to go back and look. When I selected their matches, they had a lot near 200, unlike the other testers, but right above 209 was a big jump, so this was the group that was the closest to 200.

    • I’m a member of an endogamous Jewish population. I used the AutoCluster feature in My Heritage and 99 people were grouped into several differently colored clusters. My two second cousins were in the second biggest cluster but my 3rd cousin was excluded perhaps because she didn’t match the other people. I assume that AutoCluster excludes really close family members from the analysis as they would match much more closely.

  26. so what is the summary findings of endogamy regarding CM matches down the line; how does endogamy affect future generations? In my example, which was not indicated above, there is some pockets of endogamy from Colombia, South America. I do not know my full family tree nor do I have many matches over 150 from any dna site.

  27. Hi. I have a problem here in myheritage. Long message, ok? I really need help and I hope you have the patience and the willing to answer to this. I’m writing a book about my family and i’m in the middle of it. But i was using just parish documentation.

    This book is about immigrants with the surname Blumer from my city (in Limeira, São Paulo, Brazil). My great-great-grandfather, Rudolf Blumer was one of the first to arrive here. He came from the town of Engi/Matt (they are almost the same, they are 3km apart), in the canton of Glarus, Switzerland. I am documenting (with real parish documents) my genealogy from my great-great-grandfather to the year 1600. All the information I got from a 32-volume work by a Swiss genealogist named Kubly-Müller. He completed this work in 1930, which meticulously encompasses all births, marriages and deaths in the canton of Glarus.

    After having the records taken from Kubly-Müller’s work, I personally went to Matt/Engi parish and took a picture of all the books, from 1550 until 1850, and followed the dates (I can read German Sütterlin). All of his work perfectly matches the original documents as far as my ancestry is concerned.

    The father of Rudolf Blumer is called Josef Blumer and Euphemia Stauffacher. There was JUST ONE marriage between these two persons, and in this register I have as well the information that the parents of Rudolf Blumer was named Josef and the father of Euphemia was called Rudolf Stauffacher. And look, the name Rudolf Stauffacher is very rare in this region. And I checked if there was others as well. There’s as well no notation about “illegitimate son” or “adoption”. When it happens, the priest writes a notation because it was not something wishable.

    Now talking about genetics… from the MyHeritage, I got the information there that Anna Blumer (my grandmother) should be 1st cousin twice removed – 2nd cousin twice removed of a user that I have knowledage about all the genealogy as well (using the work of Kubly-Müller) and the parish documents. The problem is that these two users could not have that closer degree of familiarity ( 2nd cousin twice removed). According to the documents, their relation is of “a daughter of a 5th cousin”, “a 6th cousin”, a 7th cousin, a daughter of 7th cousin, 3 times 8th cousin.. there is a mixture relationship in this small cities surrounded by alps. But the result was that they are 2nd cousin twice removed with 1,7% shared dna, 117 cM, and the longest 34cM.

    It would be impossible that the great-grandmother (2th 18x) of my grandmother would be the same as one of the great-great-grandmother (3th 32x) of this user. They were different families. The side of the m other of Rudolf Blumer is the side i’m studying and writing about in the book. But I have all the registers of the side of the father of Rudolf Blumer as well.

    • Enter that cM amount into DNAPainter Shared cM Project tool and view the relationship ranges there. They maybe related in a way you’re not aware of, and is not reflected in the parish registers.

Leave a Reply