Saturday, January 9, 2016
DNA: Triangulation, Pileups & Endogamy
There has been some debate at the ISOGG group about whether triangulation is possible beyond 4 generations. For triangulation to work the segments we are comparing would need to go back to a common ancestors within the genealogical time frame. The DNA testing companies estimate the regions of the genome they are comparing contain uncommon SNP's. They estimate when you share segments with matches that generally the relationship isn't much farther back than 6 or 7 generations. If this is the case then triangulation is possible if both you and your match have a tree that is fairly complete to 6 generations. Even if it isn't complete you can reasonably draw inferences about what the rest of the tree might look like. If someone is half Italian and you haven't found any Italian ancestors you can easily eliminate half that persons tree. In other cases in the US, for example. you can reasonably eliminate possible ancestor matches based on the region of the country they were from.
Some are questioning the age of the SNP's we inherit. Are they 200 or 300 years old or are they ancient, 500 years old or older? We generally share very little DNA with ancestors who lived 200 years ago. It's hard to believe that we would continue to share SNP's from 500 years. If we do it sounds like it would be a very small number and the amount of DNA would be very small, and would not be considered a match by the testing companies.
Some cite endogamy as the reason 500 year old and older SNP's persist. There is a high degree of interrelatedness among those of us who have Colonial American ancestors. Americans whose families remained in the same eastern seaboard areas since Colonial times tend to have problems with endogamy when they DNA test. Although those living in urban eastern seaboard areas tend to be more ethnically mixed as waves of immigrants settled these areas. The amount of interrelatedness among Americans varies. Even if someones ancestors lived in the same rural area for hundreds of year it doesn't mean they are highly genetically related to their neighbors. You might also see more recent immigrant groups, like the Italians, coming in and adding to the gene pool in rural areas. Many Scandinavians settled in the Midwest adding their own genes to the mix. Many of us on the West Coast of the US have Hispanic or Asian genes. This dilutes our Colonial American gene percentage.
Most of my Colonial American ancestors were Scot-Irish and German. I can pinpoint exactly when they came to the US in the 1700's. As for some of the others it's possible some of these lines go back to the first settlement of Jamestown? Could I be mistaken and some of the segments I've named actually go back to another ancestor who settled early in Jamestown? Or even go back to an ancestor back in England? I would think the odds are low considering the odds of actually still having a measurable amount of DNA, from that far, which would be enough to signal a match.
So why do we have so many matches piling up on one segment. Would sound like these are old SNP's that many people inherited and are common to certain ethnic population? Or maybe there are other reasons? Most all of my matches at AncestryDNA are from the same family group Roller/Zirkle/Roush. These families tended to marry close cousins because they lived in the isolated Shenandoah Valley, and I'm sure there were language, religious and cultural differences. This endogamy means that their descendants potentially have retained more of their DNA. My family never stayed in the same area for more than a generation or two. They didn't marry close cousins. Since we have inherited small amounts of DNA from our German ethnicity Roller/Zirkle/ Roush families we tend to match this family group more than any other group. We tend to get a match with one of those families once a week. We have 5 DNA Circles for these families. The likely reason for this is that our matches have ancestors who married cousins in this family group. Often I will see, for instance, Zirkle and Roush on their tree a couple times at the very least. These are our ancestors from around 250 years ago. We match so often because many of these families lived in the Shenandoah Valley for generations, and continue to live there, so these genes continue to cycle through the population. They have more DNA from these ancestors to potentially match with.
Another reason for pileups is large numbers of descendants. In America families tended to be large before urbanization. The survival of children into adulthood tended higher than in Europe. American couples living in 19th Century America have large numbers of descendants living today.
America was settled during the genealogical time frame so this should mean that triangulation is possible. All of these facts I mentioned mean you need to build your tree out as far as possible, and compare with as many cousins as possible. The odds of sharing the same segments with the descendants of the same ancestors may not be statistically high. Considering the number of descendants some ancestors left I think it is statistically possible. The major problem I have is the lack of records dating back to the 1600's in the Mid-Atlantic and Southern States. Otherwise I believe triangulation is useful and accurate if other parties have reasonably complete trees. Odds are reasonably good the segments don't go back to the 1600's. Plus, in my case, only 31% of my ancestry goes back to Colonial America. Much of that ancestry is already traced back to the immigrants.
Could large proportions of the early population of American have shared recent ancestors because they came in a mass migration? I believe the early population of the Mid-Atlantic states and South was more varied? New England may have had a more homogeneous population coming from the same stock in England.
I'm a believer in Triangulation. The more testers we have the more opportunities we will have to make connections through Triangulation.
Without triangulation DNA testing will be useless for Americans with a high degrees of interrelatedness. How will they separate their lines?