How to deal with a tricky triangulation with lots of pedigree collapse?

+6 votes
I'm starting to work on my first segment triangulation and of course I picked one that's complicated.  I've got a solid group of matches (two segments, on Chr 1 and Chr 16) that can confirm a connection between my ancestor Hugh Lavender of South Carolina and his brother Robert, a Tory who moved from South Carolina to Nova Scotia after the Revolutionary War.  So far all of the DNA matches on Robert's side descend from his daughter Sarah.  Here's the hitch.  Sarah had nine children from two marriages (Godfrey and McLean).  Just to make things interesting, four of the McLean kids married four MacDonald siblings.  The McLeans, Godfreys, and MacDonalds then intermarried like crazy for the next two or three generations.  The end result is that so far all of the DNA matches that I've found are descended from Sarah in at least two or three different ways.  I therefore have no way to know which of those pathways should be used for WikiTree DNA confirmation.  I'll keep looking at matches and hope that I find one that is only connected to Sarah one way, but does anyone have any other suggestions for how I can move forward?  I'm considering setting up a freespace page to lay out all the evidence since it's more complex than a typical triangulation, but that doesn't solve the pathway problem.
WikiTree profile: Robert Lavender
in WikiTree Help by Lisa Hazard G2G6 Pilot (181k points)
I'm really not being sarcastic when I say it might be a good idea to pick an easier case for your first try at triangulation.

It would be interesting to see your free-space page.  Right now it's hard to know what kind of analyses and charts you've made so far.

One thing you could try is to use the GEDmatch Segment Search (Tier 1 only) to find other people who share the segments.
I'm using DNAPainter's chromosome mapper to aggregate segments from across all the major sites (GEDMatch, FTDNA, MyHeritage, 23andMe), so I have a quick way to see all the people who match at these two locations.  Most of them don't have good trees, of course, so it's hard to find the connections.  But there are a lot of matches left to investigate, so I may yet find one connected through a single path.

When I do find a connection, I add it to a WATO tree to see the path(s) back to Robert and Hugh's father.  (I'm not using the WATO function itself; this is just a nice quick way to build a tree down from a common ancestor.)  Eventually I'll have to get the critical profiles up on WikiTree with proper documentation; I'm just using this to get organized.

These are really solid matches, too... I'm not chasing tiny segments.  Some of Robert's descendants match me on Chr 1 (up to 67 cM), some on Chr 16 (up to about 30 cM), and some on both.  I know those seem like large matches for this distant a connection and I've checked to make sure we're not also related through some other line.  These appear to be what Blaine Bettinger has referred to as "lucky" segments that have been passed on intact through many generations.

The connections are clear to me and I'm confident that the relationships are valid; I'm just trying to figure out how to document it within WikiTree's framework.  Maybe it's time to either work on a freespace page or switch over to something a little more straightforward for now.

Oh, maybe I misunderstood your question.  There is a DNA Confirmation Citation Maker app here.  (I haven't used it.)

You probably know that you can be related to someone through ancestors you've identified, and you can share DNA with that person, but that doesn't prove the DNA came from those ancestors.  So how you sort out the various pathways you mentioned, I don't know.

I can do pieces of it by "walking the segments back."  On Hugh's side, I share these segments with a known Lavender 4C and a ½ 2C1R.  Those narrow the path on my side down to our shared ancestor, Hugh's grandson Daniel.  On Robert's side, my matches are 3C/4C or so to each other, which also narrows things down.  With this number of matches all tracing back to a single common ancestor (Hugh and Robert's father), the odds of us all sharing these segments by coincidence through other pathways is unlikely, though I can't completely rule it out.  My issue is that (for example) one match's grandfather's parents were first cousins- Sarah was their grandmother.  So I don't know whether this segment was passed down from Sarah through his mother's side or his father's side, which means I don't know which parent would get the WikiTree DNA confirmation statement.
I could be mistaken about this, but I wouldn't worry about their parents, only your parents, grandparents, and so on... I'm trying to get this right too, because I have done a lot of DNA triangulation that only recently has become easier to add to wikitree... but I would think that you have discovered a DNA confirmation to Sarah, your ancestor, through triangulating her other descendants, with yourself and each other, and in the process discovered they have a double ancestor... well, this wouldn't really affect you other than probably having a higher amount of DNA with that cousin than you do with other cousins in other lineages or even within the same one, of the same degree of kinship to you. If you triangulated the DNA to Sarah, then I believe you can say on your page that you found the connection, and then list the cM %, what company, etc. with your match, and then you can confirm your DNA trail, but I'm not so sure about theirs, especially if you aren't sure: I would leave that up to them, or maybe shoot them a message and they can probably tell you more about it, and what their suggestions. preferences, and conclusions are.
I've done a lot of DNA confirmation at the 3C level (no triangulation needed), and I'm pretty sure that the expectation is that you should be able to add the statement and check the DNA confirmed box all the way up and down the chain, from one match to the MRCA to the other match.  Right now I've got a break because I don't know which path the segment followed.  I could flip a coin and just pick one, but then I'm confirming a couple of links that may or may not actually be confirmed.  (Worst case scenario: I randomly pick path A and confirm it, but path B is actually correct, and there's an unknown misattributed parentage event on path A.  I've made it hard for someone else to sort out path A later.)
That is a good question.  I've skipped over generations with DNA (not so much on WT but in general when identifying matches).  If you are uncertain of a parent, I certainly wouldn't mark that one as confirmed.

There are times when it makes more sense, and would be just as useful to other researchers, to simply add a "DNA Information" section to the profile without trying to achieve "confirmation."

But I love my little yellow check boxes!  laugh  Seriously, though, I think I'll just work on putting a freespace page together for now.  I can link the profile pages to that and sort out the WikiTree protocol later.  I still have a lot of profiles to make on Robert's side before I'm even ready to do DNA confirmation anyways.

While I too see that on wikitree it says up to 3c no triangulation needed... that is not to assume it wouldn't be especially beneficial for you given your conundrum. I'm kinda confused, have you found at least a couple of matches all share the same segment on the chromosome 1 and/or 16 or not? If you have, and they match on the side you expect it to and not on the side you don't: YOUR maternal side, it looks like, whichever Hugh's descendent is relative to you, that IS triangulation, because you could only arrive at that conclusion by comparing yourself with both these people and with them with each other, all at the same time... Like, I have DNA tested my parents and confirm them with me, and in turn, my dad can confirm he is the son of his father, and also his mother, because of the known cousin relationships found, as well as my grandfather's great grandfather, but just because they are "technically" within the "3c" relationship doesn't mean suddenly triangulation is irrelevant: in fact I think of it as a type of triangulation: if a "3c" cousin matches you but not your parent they should also relate to, or a first cousin, or aunt/uncle in that family, then it's likely not a match to the MRCA you think it is. If they do, then triangulation is not really necessary given that you already found this "third party" connector... so I can confirm my great grandfather because not only do I match with his son's son and daughter, but my dad and his dad does as well, in the appropriate relationships as expected. I don't think you should be able to add "DNA Confirmation" beyond your great great grandparents without some serious triangulation proof.. (and the profile you attached is from 1751! this is out of typical autosomal range) If you haven't done so already, this would be where some answers are likely to be found. Besides, with triangulation, you are eliminating the "pick a side, flip a coin" perspective and it may or may not be right, when triangulation SHOWS you whether you are right or wrong, with the third person to consider (yes, this IS really tedious work, especially the further back you go, within autosomal range, respectively) I love the yellow boxes too I might add, and I might be missing something here... but this is my two cents, hope it makes a lick o sense to somebody outthere lol
I am in the process of trying the DNA Confirmation Citation app for the first time (I included a link above).  It illustrates very well how tricky triangulation is.  You can't take just any three matches that share the segment.
Just to clarify for Stephanie, the DNA triangulation is valid (multiple matches on Chr 1 and 16); the issue is the paper trail connecting the matches.  I can't be sure of the exact pathway on Robert's side because so far all of his descendants are connected to him through multiple paths.  

Julie, maybe I'll try the citation app with a simpler case first, just as practice.  This one is really cool but messier than it looked at first.

Just to clarify your clarification laugh, you cannot use DNA Painter for the triangulation.  It wasn't clear if that's what you intended from your earlier comment.  DNA Painter can highly suggest that a segment will triangulate, but it has to be confirmed on the same DNA site for all three testers, either GEDmatch, 23andMe, MyHeritage, or sometimes FTDNA (with hoop jumping).

As far as your original question, I've dealt with that situation.  As you noted, you cannot know which pathway the DNA traversed, so just don't mark that portion of the relationship pathway as confirmed.  Mark only the relationships that are part of a single path from tester to MRCAs.  You may wish to put an explanatory note on those profiles that are part of the divergent pathway as to why they are not DNA confirmed.

Kerry, thanks.  I wasn't sure about that.  I've got multiple matches at multiple sites (and a handful who have their kits at more that one site), so hopefully I'll be able to come up with a combination of matches at a single site that works well.  (I just found a GEDMatch/Ancestry match who appears to be a descendant of Robert through just one path, but I'm still working to verify that.  If so, with luck I can find another match there.)  But if I can work around the problem by just leaving the DNA confirmation out of a couple of links in the chain, that would be great.

First pass at organizing all my info is here.  Comments welcome, but keep in mind it's a work in progress.  It's already out of date; I need to add in this GEDMatch match I just found (and test company info for each of the matches).

1 Answer

+4 votes
I have been trying to figure out that one myself: similar situation, where there's a pedigree collapse due to two siblings ([[Richardson-18778|Barnett Richardson (1760-abt.1840)]]) marrying another set of siblings ([[Martin-12613|Nancy (Martin) Richardson (abt.1776-)]] [[Martin-12611|Elizabeth Betsy (Martin) Richardson (1772-1832)]]) and it goes on and on because not only do I descend from the above mentioned group, I also descend from a third Martin/Richardson union, as well as the Teels who are also connected to this family, etc., which makes things look crazy on the DNA side of things, and I'm talking like 123 matches whoa for these connections, AND because this ISN'T my main paternal Y-DNA "Martin" line, it would appear that a lot of the matches are assumed to be in this line creating an "overcast" in comparison to all my other matches. so the good news is that I know I definitely descend from this family* but trying to triangulate matches, I have found a few keys:

1. You have to keep all your lines in mind. That is to say it isn't enough to have a match on a certain segment on a certain chromosome, but also to be able to make sure it's the "correct" chromosome you have triangulated, and I do that by triangulating my closest known matches so if it's my paternal line, I compare the match with my father to make sure it matches him as well, and also with my mother, to make sure it DOESN'T match her on that same chromosome.

2. The closest, shortest pathway is the one to use, that is confirmed: that is to say kind of ignore the ones where you have found multiple pathways, and/or in the process of adding their tree to yours, it's probably the shortest most direct pathway, as in you can start to see that the MRCA is just that, regardless of multiple connections to and around those people; in fact, because of this, you likely have "more" DNA from this couple as a result, and "more" cousins, too. They don't all match exactly where you want them to, however: at this point you need to consider the "rules" of endogamous DNA which can look very similar due to the overlaps in connections, but are not as "accurate" in cM/relationship as other families tend to be due to this. So, as a result, you may think you are looking at a close cousin who is in reality further away in the tree than another cousin in this family group that was the known child of your great grand aunt or whatever, and in this case this person is the one to anchor the DNA relationship, and not the more "overlapped" cousin... this is to say why triangulation is so important. You may discover the MRCA is a generation further back than the results would guess, or where they are expected to be in the tree, but with "way too high" of shared DNA to be "there"...

3. If you can say with relative certainty where 3 people match on a chromosome segment, then that is the person to build from, checking how the other matches in the family not only compare to you, but also your certain relationship with them, repeating the process, and that is to say kind of ignoring the "extra" DNA and "overlap" and getting right to the MRCA and triangulating that with certainty, making note of the earlier MRCA "double ancestor" attributing to the additional "boost" of that DNA lineage.

So for example, I have a lot of overlap in the Richardson/Martin family, and as a result, the Martin/Caffery lines... I see that I have a ton of matches, but only a handful have been proven to be super easy for me to find our MRCA despite the overlap, and as a result have been key matches when looking at unknown ones, but certainly descending from this same very large family.

I hope this will help you and others, and myself too: I think it's really cool that we attempting to master such new technology and science and making meaning out of it. It's difficult, but not impossible and I love this community you all are very motivating in my own research!!
by Stephanie White G2G4 (4.4k points)

Related questions

+14 votes
2 answers
+11 votes
1 answer
275 views asked Mar 18, 2018 in WikiTree Tech by Barry Smith G2G6 Pilot (220k points)
+7 votes
2 answers
+13 votes
1 answer
+7 votes
1 answer

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

disclaimer - terms - copyright