DNA results for collapsed pedigree

+7 votes
Hello! I'm 18 months into genealogy and have spent the past year carefully cleaning, editing, and rechecking all my work on Ancestry so it will be suitable for transferring over here. I'm also about 2 weeks from Ancestry DNA results coming in, and am hoping my grandmother will do hers this week too.

My question is, in her line, there are a few marriages between close cousins. Let me see if I can explain this in as few words as possible, call me generation 0, grandma is gen 2.  Generation 10, two cousins married each other's sisters, giving generation 9 a double dose of that line's DNA.  Two members of gen 9 had children who then married, doubling the DNA again for Gen 7.

So by my calculations generations 0-7 have about 4x as much DNA in common as one might expect.  (And yes I'm familiar with the broad ranges of how many centimorgans each jump may shared obviously I'm simplifying)  So tell me if these presumptions may be correct:

It's said that DNA matches aren't very useful past 4 steps "removed", but in our case that may be 6 steps "removed" where there's some small chance of getting a DNA match?

My grandmother and another DNA tester had great grandparents who were siblings.  There's been no intermarriage since those siblings. So their DNA match would look like 3rd cousins still, but may carry a greater chance of showing up at all since the ancestor-siblings were each 1/8 "Mathew" instead of 1/32?

Similarly my own DNA matches may be further out on the tree than we'd typically expect, as much as 2 more steps removed?

(And then there's the matter of my great grandmother having not only her quadruple dose of "Mathews" but also a double dose of "Williams" ... and I'm pretty sure I'm going to find that Williams line was itself half-Mathews a few generations back ...)

Can I just sum it up as, wow lucky me I may have more chances than most at using DNA to point the way toward distant generations? I'm sure this isn't all that unusual, with ancestors who emigrated and settled in tightly knit religious sects in the 1700s, and pioneered the west in small tightly knit groups in the 1800s. Hopefully two centuries of limited dating pools makes up for having less documentation and more cousins of the same name and age, in the same place at the same time!
in Genealogy Help by
retagged by Ellen Smith

2 Answers

+10 votes
Best answer

Hiya, anonymous!  :-)  You'll have fun with your genetic genealogy experience. One of the first things you'll want to do when your (and your grandmother's, too) AncestryDNA data come back will be to upload it to a service--like GEDmatch.com or Family Tree DNA--that will allow you to dig into the detail of the matches and look at the specifics of segments along the chromosomes. Ancestry doesn't provide enough information to do the analyses you'll need to do.

Today's DNA Trivia: Ya know where most of our knowledge--and the math to compute it--about pedigree collapse and DNA originated? In the early 20th century from the breeding of farm animals. Kinda stands to reason, though, when you think about it. The generations are relatively short; breeders are trying for very particular traits; and they have to balance achieving those traits with organism frailty due to lack of genetic diversity.

It was a gentleman named Sewall Wright, employed by the USDA at the time, who wrote a brief but important paper in 1921 about inbreeding in animals called "Coefficients of Inbreeding and Relationship." You can view a copy of his paper in the resource archives of my one-name study, if you're curious. It works only for the amount of theoretical autosomal DNA sharing, but it's still a valid yardstick to this day. It can happily work out, for example, that half-1st cousins should be expected to share 66.6% of the amount of DNA that triple-2nd cousins do. <cough, cough>

But unless you get giddy working with summation symbols and superscript "to the power of" variables, a blog post by Jim Bartlett from a couple of years ago may be a good place to start, if you haven't already read it. Jim breaks things down into more genealogist-friendly terms and puts it into example tables. The tables may still take a little study to figure out how he's using them, but you can actually duplicate the format in a spreadsheet for your own use to look at expected, theoretical autosomal DNA sharing among two cousins with one or more shared ancestors, and the sharing in subsequent generations.

Pedigree collapse affects your genealogical use of autosomal DNA through four factors: 

  • The degree of the parent/child relationship (twins having children is the extreme, as Jaimie and Circe Lannister showed us)
  • The distance in generations since the collapse (the original sharing gets more random and breaks up segments with each birth event)
  • Whether the instance(s) of collapse are isolated or multiplicative (approaching actual endogamy)
  • Whether the instance(s) of collapse affect a single family branch or spans across both your maternal and paternal lines

Your matches--certainly on that particular branch of your tree--will show a higher than average amount of total DNA sharing, as you suspect. Although we have to remember two things. First, that the expected shared total decreases by a factor of four at each generation after the pedigree collapse. For example, with no collapse the average sharing with a 2nd cousin is 3.13%, and with a 3rd cousin it's 3.13 divided by 4 = 0.78%. So the gap between the expected average sharing and sharing influenced by pedigree collapse shrinks very quickly. You can have three times more sharing with a 5th cousin than expected, and the total would still be less than 10cM.

Second is that while the shared total in centiMorgans will be higher with pedigree collapse, the length of the largest segments won't necessarily be...at least not significantly so. Your tripling of the total shared with that hypothetical 5th cousin won't result in a tripling of the length of the segments. (Jim points this out in his article, as well.) Segment length is driven mainly by the step during meiosis called crossover. Where the shared segment is located along the chromosome affects this, too. Segments nearer a chromosome's practical center--the spot where the pair of chromatids link together, called the centromere (creative, huh?)--are less likely to shuffle and recombine than are segments farther out toward the telomeres, the ends of the chromosomes. Bottom line, though, is that large segments are not "sticky"; they don't pass down from generation to generation unbroken. The end result here is that you probably won't be able to verify cousins much more, if any, distant than the rest of us since the collapse in your pedigree happened at your 8g- and 7g-grandparents, too far back for larger segments from them to be inherited.

The greater the degree of collapse, the more complex the task in matching the DNA evidence to the paper trail. This is why it's so difficult, if not impossible, to effectively use autosomal DNA to go reliably back more than a couple of generations in endogamous populations like the Ashkenazim or some Polynesian Islanders. The reason it's so complex is that once you suspect impact from pedigree collapse, it doesn't suffice to do what we would consider standard autosomal triangulation: finding multiple tested cousins who all share a meaningfully-sized segment and who also match the paper-trail. With pedigree collapse you can't just follow one trail to an MRCA couple: you have to investigate every possible pathway of genetic inheritance. Autosomal DNA triangulation to a single, shared, ancestral couple can be functionally impossible in truly endogamous populations. There are two reasons I believe the majority of autosomal triangulations I've seen are probably false: A) use of segment sizes that are too small without large numbers of individuals in the triangulation groups to substantiate them (and a corollary, not triangulating sufficiently to the actual MRCA but using shorter legs that stop at more recent generations); and B) use of shared DNA amounts that are far larger than predicted by the theoretical averages and that clearly indicate the likelihood of pedigree collapse, but without doing the hard work--and spending the time looking for and testing additional cousins--necessary to investigate all the possible pathways of the DNA inheritance.

Have fun! And join WikiTree! Most serious genealogists I know have some Sherlock Holmes in them: it isn't just the discoveries that they enjoy, but the detective work that goes into it. And genetic genealogy definitely adds a whole new set of sleuthing tools.  :-)

by Edison Williams G2G6 Pilot (344k points)
selected by Edie Kohutek
Great answer! It will be helpful in my own parsing of possible pedigree collapse. I have to admit I cheated: When I saw the length of the answer, I skipped down to the author's name and --Voila! My guess was correct. It was you, Edison, with a lengthy and very informative answer.

Thanks for the "best answer" star, Edie.


It's a complex subject. My only defense. I actually tried to see if I could enter the equations for the coefficients of inbreeding and relationship using just the text formatting options here on G2G, but very wisely gave up on that.

But guess what? I did find the CoR on Wikipedia, if anyone wants a look. Unfortunately, it uses the coefficient of inbreeding in the very first formula, but never defines the formula for that coefficient, and it doesn't look like it's defined elsewhere on Wikipedia, either. Great. Another to-do for the list: edit Wikipedia article. Aaargh. There are no formulae at all in the related ISOGG Wiki article. Another to-do. Kinda wish I hadn't come back to this question again...  :-)

Thanks Edison!

Please let us know when you're done with the Wikipedia article.

I did the are your parents related at Gedmatch.com yesterday and Unbelievably it came back negative. So I uploaded to the other linked ROH site connected to the Gedmatch.com page and those tests confirmedI am inbred. Lots of segments above 200 and the overall homozygous was 71.823%, heterozygous was 28.177%. 

According to wikitree my parents share 14 common ancestors within 15 generations, 32 within 30 generations. 

Ellen, I tested my Mom to pick up that extra gen back, you'll be fortunate to pick up 2 extra gens by testing your grandmother !


Sorry I should have addressed that to "Anonymous".

And the 14 ancestors in common are gens 4-7.
Pedigree collapse in my experience has significantly less effect after 6, 7, 8 generations than endogamy; I have ancestors on one branch of my tree from an endogamous population; the only instance of pedigree collapse I've found, so far, is that my great-grandparents were 8th cousins, but my mother has many DNA matches on GEDmatch and at Ancestry etc who share DNA in the expected range of 3rd-4th cousin whose closest relationship is 6th cousin...but also 7th cousin, twice, 8th cousin, 3x over, 9th cousin, etc, through multiple lines (the most extreme example I've found is someone who shares no fewer than a dozen separate ancestral couples and matches my mother for 46cM on four segments, each of which may have come from a separate set of ancestors, and some of my DNA matches on those lines are as distant as 7C1R or 8th cousin with a common ancestor born in the 1600's).

Point well taken. But, strictly speaking, endogamy is a social (or geographically mandated) practice; the genealogical result of endogamy is pedigree collapse (the phrase based on the German term Ahnenschwund, roughly translated as "loss of lineage").

Kind of a rectangle and square thing: endogamy always results in pedigree collapse, but pedigree collapse is not always caused by endogamy. Best example is double 1st cousins, where brothers married sisters unrelated to them.

My fantasy is some clever person will write a program that looks at the ancestry and DNA to determine the degree of endogamy and adjust accordingly (to estimate the likely number of generations to the MRCA).
+3 votes
Fascinating situation! Please come back and share the results.

Have  you been able to trace medical issues through such a tightly knit tree? We have a birthmark that floats through the generations as well as a propensity for lypomas.
by Rebecca Peterson G2G6 Mach 1 (17.5k points)
No, interestingly enough, and the same seems to hold true for others who are related.  On the contrary there's a lot of outliving peers of the same era, having a larger percentage of children live to adulthood, catching the disease going through the military barracks but living through it ... A virology research center include me in a study because I'm abnormally resistant to some viruses. A friend of mine (betcha he's a cousin, lol) has a similarly compact Welsh line and he's frequently told he's "ridiculously healthy." Maybe crossing 3 times in 10 generations, with more diversity the rest of the time, evens out to this average I've heard ... that prior to modern times spouses were on average 4th cousins.
to Rebecca,

To this day we have rampant cystic fibrosis in our Martin line. My first cousin died from it and currently there are 2 young girls who attend our Martin reunions that have it. Others are carriers....

My Mitchell line seems to be riddled with various forms of bipolar disorder. Otherwise we're pretty healthy and of course brilliant... HA!
Of Course!

Related questions

+11 votes
1 answer
305 views asked Mar 18, 2018 in WikiTree Tech by Barry Smith G2G6 Pilot (233k points)
+14 votes
2 answers
+5 votes
0 answers
+8 votes
2 answers
+13 votes
1 answer

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright