Endogamy and DNA

I really enjoyed the presentation Paul Woodbury of Legacy Tree Genealogists did about Endogamy last year at the SCGS Jamboree so I am delighted to present this guest post from him on that topic.

Personally I struggle with the endogamy in my mother’s family tree. My jewish grandad was an only child so there are not that many close cousins on his side. However due to endogamy my maternal aunts have hundreds of 2nd-3rd/4th cousin matches. No wonder I spend more time working with the easier Norwegian DNA from my Dad!

Endogamy and DNA By Paul Woodbury

Autosomal DNA testing is a valuable resource for genealogists seeking to overcome recent brick walls in their family history, particularly in instances where traditional historical research is limited or unavailing. At Legacy Tree Genealogists, we frequently use autosomal DNA test results to answer questions regarding adoption, unknown paternity, or ancestors who are difficult to trace. However, there are some factors that can complicate the use of autosomal DNA in tracing ancestors. One of those factors, which is what we will be discussing in the article, is Endogamy.

Endogamy is the custom of marrying only within the limits of a local community, clan, or tribe over the course of many generations. The reasons for this genetic isolation could be cultural or religious (as with Ashkenazi Jews and Low-German Mennonites) or geographic (as with island and tribal populations). Members of endogamous populations may descend from a limited pool of “founder” ancestors who represented the initial genetic makeup of their population. After many generations and hundreds of years of isolation from outside pedigrees, genetic profiles of population members can easily be distinguished from the DNA of outside populations. However, this can also cause pedigree collapse.

Pedigree collapse occurs when two closely related individuals produce offspring. As a result, the number of unique individuals occupying locations in a pedigree decreases or “collapses”. For example, while most people have eight unique great-grandparents, a child of two first cousins will only have six unique great-grandparents. They will also have inherited a larger portion of their DNA from the ancestors held in common between their parents. This makes it more difficult to determine exact relationships between genetic matches.

Here we explore three keys for dealing with endogamy in autosomal DNA test results: 1) Exploring the exact amounts of shared DNA between relatives; 2) Testing multiple relatives; and 3) Using direct line tests to narrow results.

1) Exploring Exact Amounts of DNA

Runs of Homozygosity

In order to accurately evaluate DNA test results in endogamous populations, consider exactly how much DNA two individuals share in common. Most genetic matches will only share DNA segments on one copy of their chromosomes – either maternal or paternal. However, if both of an individual’s parents are from the same endogamous population, or are known close relatives to each other, then they may have a “Run of Homozygosity” (a region of their DNA where the maternal copy is identical to the paternal copy). In these cases, the subject is a genetic match to themselves. If another genetic cousin overlaps in this same region of DNA, then the amount of DNA that person shares in common with the test subject should be doubled for that particular segment since they match the maternal copy and the paternal copy.

None of the DNA testing companies report total amounts of shared DNA which take into account runs of homozygosity. However, these regions can be discovered through analysis at Gedmatch.com and through David Pike’s Utilities, and might be used to confirm and refine the total amounts of shared DNA between two individuals. However, the researcher must have access to test results for both the subject and the genetic match in order to perform these comparisons.

Applying Different Centimorgan Thresholds

In endogamous populations, much of the population shares extremely small segments of DNA from many distant ancestors. If these small segments of DNA are included as part of the total shared DNA between two individuals, it can skew the estimates for how closely they might actually be related. At
Family Tree DNA, all segments larger than 1 cM (centimorgan) are included as part of the total shared DNA. At Ancestry, all segments larger than 5 cMs are included; however, some larger segments may be excluded based on their matching algorithms. At 23andMe all segments larger than 5 cMs are included as part of the total. Depending on the nature of the endogamous population, it may be beneficial to recalculate the total amounts of shared DNA between matches through comparison at Gedmatch and through application of higher centimorgan thresholds. For example an individual may share 120 centimorgans at Family Tree DNA, but when segments smaller than 7 cMs are excluded, this total may drop to 60 or 70 and may be more representative of the nature of their closest relationship. The appropriate threshold to apply in any given case will depend on the amount of endogamy within a population and whether the test subject is a full member of that population or has recent admixture from outside populations. Consider calculating several totals using different thresholds to give a better indication and overview of the shared DNA.

2) Testing Multiple Relatives

Since different descendants of an ancestral couple inherit different segments of DNA, it is always a good idea to invite additional relatives to test when exploring a brick wall problem. Other relatives will have different segments and will therefore have different matches as well. When dealing with endogamous populations it is particularly important to test as many known relatives as possible. When dealing with a case of unknown parentage or adoption, collaborate with genetic cousins and work with them to test more of their known relatives, focusing first on the older generations of their living family. Even if you know that specific relatives may be related through multiple family lines, test them anyway; their test results will greatly assist in filtering and organizing your matches.

Create Complete Trees

In endogamous populations it is quite likely that close relatives will be related through multiple family lines. Therefore, it is essential to have trees that are as complete and far back as possible for each relative tested. Since shared segments of DNA may (and often do) come from more distant relatives in endogamous populations, consider extending family trees 7 generations or more. This will not only permit researchers to identify the multiple ancestors shared in common with the relative, but will also aid in the exploration of the multiple relationships that the relative has with other relatives and genetic cousins.

In some cases, it may not be possible to build extensive family trees for relatives. Record loss or limited availability may prevent the construction of detailed multi-generation family trees. In these cases, consider documenting and focusing on the ancestral origins rather than ancestral couples. For example, in Ashkenazi Jewish research, try to identify the ancestral villages in Eastern Europe for genetic matches. This may help to tie brick wall ancestors to specific localities and regions.

Obtain Shared Segment Data for All Relatives

Though each of the DNA testing companies offers information regarding relatives that a subject holds “in common with” genetic cousins (also known as ICW), in order to be useful in the investigations of endogamy, researchers must also have access to the actual test results of other tested relatives. It is not enough to know that a genetic cousin matches a known relative. In an endogamous population we can expect that most members of the population match many other members of the population. We also need to know how much DNA they share with that relative. Some people may not be willing to share their specific results, but if you explain what you are trying to do and offer to share your findings, many people will be happy to help.

Exploring Averages and Assigning Segments

Testing multiple relatives and obtaining access to the test results for multiple relatives allows researchers to evaluate relationships between a match and a larger group of known relatives. By calculating the amounts of DNA shared between a genetic cousin and multiple relatives, it is possible to more accurately estimate the nature of relationships. Knowing that a subject shares 200 cMs with a genetic cousin, and that three first cousins of the subject share 69, 139, and 150 cMs (average of 139 cMs) with the same genetic cousin might lead us to estimate that the match is related at the level of second cousins once removed (or some equivalent relationship rather than at the level of full second cousins). By averaging between multiple relatives, researchers can identify the most likely level of relationship.

Access to the test results of relatives also permits the investigation of unique shared segments of DNA which could be representative of how much DNA a match might have shared with the common ancestor of multiple known relatives. Performing this type of research with multiple close relatives can also aid in chromosome mapping, which in turn can identify the segments of DNA that were inherited from a specific brick wall ancestor. Assigning segments of DNA to particular ancestral lines and sifting through the relationships between multiple family members enables researchers to identify and focus on the most pertinent matches to a particular research question.

3) Using Direct Line Tests to Narrow Results

Autosomal DNA analysis in endogamous populations can get messy quickly, so testing additional relatives with direct line tests (Y-DNA and mtDNA) and encouraging genetic cousins to do the same can help to narrow the possible relationships and can give increased structure to the investigation. Wherever possible, attempt to perform DNA testing of specific relatives in order to determine the Y-DNA signatures of each great-grandfather and the mtDNA signatures of each great-grandparent. Encourage genetic cousins to do the same for their own family. Direct line tests can serve as anchors and guides for interpretation of autosomal DNA test results and can identify family groups within a single pedigree that are likely related to each other.

Paul Woodbury is a Senior Genealogist with Legacy Tree Genealogists, a genealogy research firm with extensive expertise in genetic genealogy and DNA analysis. To learn more about Legacy Tree services and its research team, visit the Legacy Tree website at https://www.legacytree.com


14 thoughts on “Endogamy and DNA

Click here to add your thoughts at the end of the comments
  1. In cases where only one grandparent is Jewish, can that be used to do a rough phase of DNA results? Can we expect that all of the AJ segments will appear only on one copy of every pair of chromosomes?

  2. Can I get your views on the following instance I am working to resolve for my family?

    Four full siblings tested on Ancestry. All four siblings have between them from 13-28 percent AJ. We strongly suspect it is the four sibling’s unknown grandfather who was AJ. Let’s call them Group A.

    Four other people who are known second cousins to each other (who are not siblings) also tested – three on Ancestry and one on MyHeritage. Let’s call them Group B.

    Luckily, there is a well-respected family tree website created by two academic genealogists which allows me to confirm the relationships of Group B members to each other, ie, they are all second cousins to each other. There has also been other family research and family reunions that confirm the relationship.

    Following my research, I propose Group A and Group B are also second cousins to each other.

    The range of shared DNA from Group A members to Group B members is quite large – from very low 95 in MyHeritage, so very high 426 in Gedmatch!

    I have managed to copy all of Group A profiles and two of Group B profiles to Gedmatch. I can now see that the two people in Group B (R.T and H.G) are also low in shared cMs to each other:

    Largest segment = 27.4 cM
    Total of segments > 7 cM = 95.6 cM
    6 matching segments
    Estimated number of generations to MRCA = 3.6

    Does my hypothesis still stand given that there seems to be a precedence in the family for a result like this?

    I would like to add, just to complicate matters further, that there is a 4C1R that is also showing up just as strongly as the above proposed 2Cs. But again I can verify that they are 4C1R.

    Thanks for your insights!

    • This is too long for a blog post comment so I will send you an email.
      However I doubt they are full 2nd cousins. Look like half 2nds to me

      In my experience true 2nd cousins will share many large segments (> 10) including at least one > 30cM, particularly in endogamous populations.

  3. I am trying to solve a mystery of how I am related to a gentleman with only one known ancestor from the same area as my own ancestors, namely Pomerania (or somewhere else in Germany). He’s usually my number one match on most sites (e.g. GedMatch, My Heritage) and yet I cannot connect his Jewish G grandfather to anyone on my own family tree (ALL German). We both do have relatives who lived in the Stettin area circa 1871. And there was a rumour (according to my mother) that we had a Jewish ancestor. (Both my maternal grandparents looked rather “dark” for people with significant Frisian ancestry, and they both had ancestors with surnames that could be either Jewish or German. This is very common it seems.) We are also about 10 years apart in age; so we may be off by one generation, I’m guessing. His “Stettin” G grandfather was almost a teenager, for example, when my “Stettin” G grandfather was born.

    Here’s what I see when I compare my closest cousin match to my DNA kit at GEDmatch:

    Total CM: 71.5 Largest segment: 38.1 Estimated Generation: 3.8

    Incidentally, further down on most lists, I also see people with Jewish surnames and/or who show large amounts of “Ashkenazi” as their ethnicity. Many of these matches ALL share the same DNA with me on an overlapping bit of Chromosome 9. Yet my number one match does not share this same DNA.

    So, finally the questions:

    1) For any of these matches, is there anything that could be inflating our shared DNA results. Otherwise, it seems that someone changed or concealed records or there was some exchange of DNA outside of marriage.

    2) Is there something about endogamy among Ashkenazi Jews that is bringing a very distant ancestor closer? I must say the DNA hints at a Jewish ancestor. It’s very confusing to a novice but I don’t see how I could be related to all these people who have much, much more Jewish ancestry, especially with that one overlapping bit. Wouldn’t it be quite a stretch to guess that we all have one common European ancestor, who got one common Ashkenazi pregnant?

    I’ve tried to exclude matches under 15 CM on the largest strand/bit and I still have some of these Ashkenazi matches (some still with another bit of other DNA).

    Yet I do NOT have Ashkenazi as part of my identified ethnicity at various sites (Ancestry/FTDNA/MyHeritage/DNALand), except once, under >1%. I do get things like 3% West Asian, some combo of south Italian and west Med (about 5%) or just variations of Northern Europe stuff (English???, Scandinavian???, Eastern European).

    Any thoughts? Thanks for your time. It’s been my summer project but I’m about to give up. My cousin match is not. He’s been doing this genealogy stuff for years now. We’ve tried contacting the one other shared match we have in common. She only knows that she’s English and Scottish and doesn’t want to dig. She’s his second cousin apparently. Too bad, eh.

  4. Heidi
    There are very few Germans who have tested so that may explain why such a distant match, at best a 3rd cousin, perhaps a 4th, is your top match.
    Most likely a 2nd/3rd/4th grandparent of yours from Stettin married into the Jewish community. Or as you suggest, midattributed fatherhood … if you do not see any Jewish in your DNA or your cousin’s (try the eurogenes Jtest calculator at GEDmatch) that is the most likely explanation.
    The frequent intermarriage among family members in the jewish community can also make a distant match seem closer when DNA is coming down from the same ancestor more than once.
    Finally, it is unlikely that you will solve this puzzle unless you both get alot of cousins to test and you can find records back to the 1600s for both of your families in Pomerania.
    Interestingly my 100% father also has a pile up area on chromosome 9 which matches many russians and poles and some ashkenazim!
    Enjoy this journey!

  5. Thanks so much!
    Yes, my Ancestry.com cousin has some “European Jewish” as does our shared match and a shared match of that match… whom I already knew about before DNA genealogy, etc. I think someone assimilated in my family. There are just too many hints. I’ll keep hunting and delicatedly ask the only “German” cousin in this group. (I also have “Ashkenazi” on the Jtest but thought it was invalid.)

    I’ve really enjoyed reading through your many interesting blog entries.

  6. Hi!

    Trying to figure out the relationship with a DNA match on 23&Me.

    16.2% shared DNA
    Half identical
    1207 cM
    26 segments

    Any thoughts?

    • You really can’t tell for sure from those numbers although it looks like a 12.5% relationship with perhapsd more relatedness (this is an endogamy post, is that also an issue?)
      The calculator at https://dnapainter.com/tools/sharedcmv4 gives various options: most likely a half aunt/uncle/niece/nephew but can also be a high 1st cousin. Ages of the matches and the amount shared by other matching relatives can help sort this out.

      • Kitty,

        This is a match between my father, who was born in 1948 and a woman who was born in 1975. They matched on 23&Me. She does not know who her biological father is. My paternal half-brother also matched with her on Ancestry DNA with 596 cM and 20 shared segments. I just got a kit and need to send it off so I can compare my own DNA to hers.

        The shared segments between this woman and my father are VERY large… Here are some of the larger shared segments:
        #1 – 104.60cM (& 35cM)
        #2 – 226.20cM
        #4 – 95.35cM & 62.86cM
        #11 – 73.12cM (& 8.94cM)
        #12 – 56.64cM
        #13 – 72.55cM (& 13.74cM)
        #16 – 84.81cM
        #20 – 55.22cM

        There are other, smaller shared segments, but those are the larger ones.

        Not sure if endogamy is an issue or not. There seem to be some odd stories regarding my grandfather and his siblings’ youth. My paternal grandfather, as well as his siblings, did not have much of a relationship with their father as adults (my dad did not know his grandfather lived near him until after he passed away). My dad did not know he even had as many 1st cousins or aunts and uncles as he does.

        Due to the size of the segments shared though and the total amount of cM shared I would imagine that this woman’s father would be a brother or half brother to my father. We are unaware of a brother as of now though. My father does have a sister who is 9 years his senior and their parents did separate for a while at some point between his and her births. My grandfather would have been too old to be this woman’s father so I feel like there must be a missing brother.

        I am working on getting other cousins with known genealogical relationships to upload their raw DNA data to GEDMatch so I can view a more complete comparison since you can’t really see much on Ancestry.

        Your blog has been SO informative and helpful and I am learning so much!

        Here are the results from the DNA Painter:

        Relationship probabilities (based on stats from The DNA Geek)

        82.36% Great-Grandparent Great-Aunt / Uncle Half Aunt / Uncle 1C Half Niece / Nephew Great-Niece / Nephew Great-Grandchild

        17.64% Half Sibling † Aunt / Uncle † Niece / Nephew † Grandparent Grandchild

        † this relationship has a positive probability for 1207cM in thednageek’s table of probabilities, but falls outside the bounds of the recorded cM range (99th percentile)

        Any guidance or thoughts?

  7. Men have fathered children into their 90s … The numbers suggest your grandad fathered her dad or less likely fathered her .. So she is most likely your and your half brother’s half first cousin and your Dad’s half niece
    Endogamy is not a factor in close family matches.
    You are doing the right things by getting more relatives to test like yourself. Are there other cousins who would test? Will your dad’s sister test? More important to test her than yourself or anyone else; If she shares X with the match that can be a big clue, the entire X would suggest half sisters.
    Keep up the good work!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.