Tuesday, April 10, 2012

DNA Portraits: Second Cousins


On the PBS program “Finding your Roots” a few weeks ago, Henry Lewis Gates, Jr. set out to discover the identity of Newark, NJ Mayor Cory Booker’s white great-grandfather. It seems that when Booker’s grandfather was 13, his mother took him to the doctor. As they were leaving, his mother told the boy that the doctor was his real father. Gates narrowed the possibilities to Stephen H. Brown, MD. In order determine if Brown was Booker’s great-grandfather, a living descendent of Brown could be tested for matching autosomal DNA.

They were in luck as a living descendent who consented to be tested was located. On the show, 23andMe scientist and Senior Director of Research Dr. Joanna Mountain stated, “Virtually all pairs of second cousins share long segments of DNA that are matching – usually several of those segments. We’ve done simulations to figure out what to expect there and if they are second cousins we should be able to see that in the DNA.”

2nd Cousins Booker & Hislop
The test confirmed that Cory Booker and Michael Hislop were descended from the same man and that they were second cousins with matches on the 1st, 4th, and 10th chromosomes. While termed as second cousins, the two are actually “half second cousins.” As the name indicates, they only share one great-grandparent and they share half the DNA that full second cousins would normally share.

According to the program, we should be able to determine if a suspected second cousin is just that. So genetically, what do second cousin pairs look like? I turned to data from 23andMe to find out the answer. Over a year ago, genetic genealogist Andrea Badger asked members of the 23andMe community to provide family member statistics. All sorts of relationship data was generated that included 33 second cousin pairs.

These were full second cousins only and therefore second cousins, once removed; second cousins, twice removed, double second cousins, and half second cousins were not included in the 33. Of that number, the highest percentage share was 6.32% and the lowest was .93%. The mean score was 2.94% while the median score was 2.90%. Generally speaking, it is expected that second cousins would share on average 3.13% of their DNA; however, the sample relationships skewed lower.

My own extended family provided nine second cousin pairs. Interestingly enough, the second (5.22%) and third (4.45%) highest percentages, the lowest percentage (.93%), and the median score (2.90%) all were found within my family. Seven of those pairs descended from one couple while the remaining two were from unique couples. Our numbers skewed slightly higher than the average expected DNA percentage for second cousins at 3.22%.

Couple 1: Newton French Owston & Mary Emma Merriman

Relationship 1-A

Not counting my great grandfather's two sons to his second wife, my great grandparents gave birth to a total of five children; four of the five attained adulthood and three (Martha, George, & Ovington) produced issue. While there are numerous second cousins from this couple, five have been tested via 23andMe and represent each of the three children. Among these five, there are a total of seven second cousin relationships.

Three of the participants are brothers and provide an interesting look at segments that are passed down from their father compared to those from the children of his first cousins. The brothers include me (in the light blue) and my two older brothers in represented dark blue and green. The comparison person is our second cousin through Ovington French Owston – the brother of our grandfather George Hood Owston. The following illustration shows the randomness of recombination and gives us an indication of what we inherited from our father from his father and mother.


While the segments are somewhat long, take notice of the segments on chromosome 4 and 10 – they are the two longest segments. My oldest brother shares a very long segment on chromosome 4 that is 114cM in length. He received this segment from my father and my grandfather – the same with our second cousin who received this long segment from his father and grandfather.

It is likely that this segment (as it has remained intact via two different lines) probably came from one or the other of our great-grandparents. Comparing our matches with a second-cousin once removed from the Merriman family (and who is related to all five of us) further suggests that this long segment came from our great-grandfather Newton French Owston’s ancestry; however, without further proof this supposition is inconclusive.

When comparing this segment with a number of fourth cousins from the Owston family, none match on this area of chromosome 4. While this may indicate that the match comes from Newton French Owston’s mother (Martha Newton French), it does not completely rule out his father’s ancestry either. Some of the others matching us on this segment can be attributed to Newton French Owston’s great-grandmother’s surname of Fraser. Other portions of this segment appear to be Irish in origin and may be connected to Martha Newton French’s mother of Rebecca McAnulty who had Irish forebears. Time may indicate if this is correct.

The sections of chromosome 4 that my other brother and I do not match with our second cousin came from our father’s mother’s side of our family. Comparing these segments with two of my grandmother’s relatives confirms this hypothesis.

On chromosome 10, I match this same second cousin with a segment that is 112cM in length. While my middle brother does not match him at this location, my oldest brother shares a portion of the segment that is 56.9cM long. This segment is a little more difficult to place. By checking with 23andMe’s Ancestry Finder tool, none of the public individuals who match my cousin and me have any surnames or locations that are congruent with our known family lines.

In addition, our Merriman relative does not match us on this segment; however, it is impossible to conclude that this match does not come from that particular family. While my Merriman relative is a second cousin to me and my brothers via my Day family, she is also my second cousin, once removed through the Merrimans – a relationship she has with all of the Owston second cousins. Her grandmother was my grandmother’s sister (Day family) and her grandfather was my great-grandmother’s brother (Merriman family).

Those segments that can be identified to the Merriman family are as follows:

  • The segment on chromosome 3
  • The segment on chromosome 8
  • The first segment on chromosome 9

Relationship 1-B

During fall 2011, another second cousin in my father’s line had her DNA tested with 23andMe. Being the granddaughter of my grandfather’s oldest sister, Subject 5 as she is called here was probably the best choice in our genetic genealogy study, not only does she match her second cousins recorded here, she matches all of the Owston fourth cousins and fourth cousins, once removed as well as our two Merriman cousins. Because she matched so many people, her results were perfect for providing triangulation.

While my brothers matched my second cousin with nearly twice as much DNA as me, the following illustration shows the number of long segments to which Dr. Mountain alluded in the television broadcast.


I am particularly interested in three long segments on chromosome 1 that are shared by my oldest brother (Subject 1) and our second cousin (Subject 5). It appears that these three segments may have been part of one long segment passed down from one ancestor and that these were split due to recombination.

When comparing this same chromosome with our mutual second cousin (Subject 4), it indicated that there was probably an even longer original segment. Our Merriman second cousin, once removed did not match any of the five of us in this region, so it may be that this particular segment came from our great-grandfather’s lineage. None of our other Owston fourth cousins were matching to this section of chromosome 1 either.


I checked with the public matches on 23andMe’s Ancestry Finder (AF) tool and found a match to a gentleman whose paternal ancestry was unknown as his father had been adopted. His father’s birth and adoption occurred in Detroit and this segment may suggest a common ancestor in the personage of our second great grandparents. From as early as 1859 to as late as 1863, my second great grandfather lived in Detroit.

The 1860 census lists the family as containing my second great grandparents, my second great grandmother’s maternal aunt, and the couple’s two children: Fanny (AKA Fannie and Frances) and French (or Newton French) Owston. The surname was misidentified as Austin, but it was clearly our family from the other data. I dealt with finding this census entry in a previous post.


The daughter, Frances Owston, is the mystery. I have only found three references to her – the first was a sample of her hair that was dated 1859 and found within a family bible I was given in 1977. Another reference was her birth record in 1852; it was the first birth to be recorded in Allegheny County, Pennsylvania. The final reference is the 1860 Detroit census.


While I have searched for her in all of the cities where her family lived – Detroit, Saginaw, and Pittsburgh, I have never been able to find her marriage or death. It may be possible that our AF match is descended from the elusive Frances Owston.  (In 2014, we found Frances and her family in Western Pennsylvania.  This relationship is probably not through her.  See our post on this at http://linealarboretum.blogspot.com/2014/11/its-just-another-brick-from-wall.html).

As far as the other sections, we are able to confirm that some of these came from our great grandmother’s Merriman family roots. By comparing with our Merriman second cousin, once removed, we can conclude that following segments are related to our Merriman line:

  • The second segment on chromosome 3
  • The second segment on chromosome 7
  • The second segment on chromosome 11
  • The segment on chromosome 17

The other segments are not as easy to track. An AF match on the first segment on chromosome 3 has roots in Western Pennsylvania and may be indicative of either the Merriman or French families. The other segments cannot be triangulated at the present.

Couple 2: Thomas Wesley Day & Amy Alice Champlin


My double second cousin is also a second cousin to my half cousin – the son of my father’s half sister. My half cousin and my brothers and I share a grandmother. Our second cousin’s grandmother was our grandmother’s older sister and my half cousin’s matches are confined to the Day/Champlin lineage; however, since our grandfather was her grandfather’s nephew, my brothers and I share additional segments via our Merriman ancestry.

At the present, it is difficult for us to determine whether the segments were from the Day line or the Champlin line as both had colonial ancestry.


Notice the long (122cM) segment on the X chromosome. Because of the unique transmission of the X chromosomes, these two second cousins share a segment which neither my brothers nor I share with her – although, my father should have matched as well as he had an X chromosome passed from his mother as did my aunt.

The only difference is that my aunt could pass an X chromosome onto all of her children and my father, who only had sons, could not. Had we had a sister, she would have probably had a match on the X chromosome as well. The matching segment could come from either one of our great-grand parents or a combination of both.

Couple 3: Charles William Owston & Emma Lydia Morton Shephard


Although, I am not directly descended from this couple, Charles William Owston and my great grandfather, Newton French Owston, were first cousins. While both of these individuals match others in the Owston clan, none of these segments match any of our tested fourth cousins.


At the present, we cannot determine from which great-grandparent these segments can be traced. This particular second cousin relationship is also the smallest (at .93%) of 33 recorded second cousin pairs on 23andMe. Compare the above with our family’s largest second cousin share (at 5.22%) depicted below.

Can We Determine Second Cousins with DNA?

While I’ve met several second cousins, I haven’t found one that looks like me; although, I did meet a second cousin, once removed who looked more like my father than I do. Genetically, we should be able to confirm whether a person is a second cousin or not.

There is one word of caution - recombination sometimes allows more or less shared DNA than the average prediction; therefore, a second cousin, once removed or a third cousin may share as much as the average second cousin. A second cousin sharing less may appear to be a second cousin, once removed or a third cousin - so while segment size matters, it is not always conclusive by the amount shared.

In our families, we have 39 relationships that are not strict second cousin relationships, but that have individuals that share within the same range as the second cousins. I have also included 23andMe's predicted relationships. These relationships include the following:

Half Cousin – Average Share 6.25%
5.95%; 446cM; 13 segments; Predicted: Second Cousin
5.76%; 432cM; 15 segments; Predicted: Second Cousin
5.07%; 380cM; 14 segments; Predicted: Second Cousin

First Cousin, Once Removed – Average Share 6.25%
8.23%; 617cM; 27 segments; Predicted: First Cousin
7.67%; 575cM; 20 segments; Predicted: First to Second Cousin
6.95%; 521cM; 22 segments; Predicted: First to Second Cousin
6.56%; 491cM; 25 segments; Predicted: First Cousin
5.29%; 397cM; 20 segments; Predicted: Second Cousin
4.47%; 334cM; 18 segments; Predicted: Second Cousin

Second Cousin / Second Cousin, Once Removed –
Average Share 4.69% (Double Relationship)

5.79%; 434cM; 20 segments; Predicted: First to Second Cousin
5.34%; 401cM; 15 segments; Predicted: Second Cousin
4.98%; 373cM; 14 segments; Predicted: Second Cousin

First Cousin, Twice Removed – Average Share 3.13%
3.64%; 273cM; 12 segments; Predicted: Second Cousin

Half Cousin, once removed – Average Share 3.13%
3.93%; 295cM; 9 segments; Predicted: Second Cousin
3.83%; 287cM; 10 segments; Predicted: Second Cousin
2.80%; 209cM; 9 segments; Predicted: Second to Third Cousin

Second Cousin, Once Removed / Second Cousin, Twice Removed –
Average Share 2.34% (Double Relationship)

3.41%; 255cM; 10 segments; Predicted: Second Cousin
3.04%; 228cM; 10 segments; Predicted: Second Cousin
2.79%; 209cM; 8 segments; Predicted: Second to Third Cousin

Second Cousin, Once Removed – Average Share 1.56%
2.47%; 185cM; 4 segments; Predicted: Second to Fourth Cousin
2.22%; 167cM; 10 segments; Predicted: Second to Third Cousin
2.14%; 160cM; 6 segments; Predicted: Second to Third Cousin
1.82%; 137cM; 8 segments; Predicted: Second to Third Cousin
1.51%; 113cM; 5 segments; Predicted: Second to Third Cousin
1.47%; 110cM; 4 segments; Predicted: Second to Fourth Cousin
1.30%; 97cM; 7 segments; Predicted: Second to Third Cousin
1.29%; 97cM; 5 segments; Predicted: Second to Fourth Cousin
1.17%; 88cM; 5 segments; Predicted: Second to Fourth Cousin
1.15%; 87cM; 4 segments; Predicted: Third to Fourth Cousin
0.92%; 69cM; 2 segments; Predicted: Third to Fifth Cousin
0.83%; 62cM; 4 segments; Predicted: Third to Fourth Cousin
0.30%; 23cM; 2 segments; Predicted: Third to Sixth Cousin

Second Cousin, Twice Removed / Third Cousin, Once Removed – Average Share 1.17%
2.68%; 268cM; 10 segments; Predicted: Second Cousin
2.36%; 177cM; 9 segments; Predicted: Second to Third Cousin
0.76%; 57cM; 5 segments; Predicted: Third to Fourth Cousin

Second Cousin, Twice Removed – Average Share 0.78%
1.56%; 117cM; 6 segments; Predicted: Second to Third Cousin

Third Cousin, Once Removed / Fourth Cousin
Average Share 0.76% (Double Relationship)

1.73%; 130cM; 8 segments; Predicted: Second to Third Cousin
0.75%; 56cM; 3 segments; Predicted: Third to Fourth Cousin
0.23%; 17cM; 2 segments; Predicted: Third to Sixth Cousin

For comparison, here are the 9 second cousin relationships

Second Cousin – Average Share 3.13%
5.22%; 329cM; 11 segments; Predicted: Second Cousin
4.45%; 334cM; 10 segments; Predicted: Second Cousin
3.59% 269cM; 15 segments; Predicted: Second Cousin
3.57%; 268cM; 15 segments; Predicted: Second Cousin
3.28%; 246cM; 8 segments; Predicted: Second to Third Cousin
3.27%; 245cm; 13 segments; Predicted: Second Cousin
2.90%; 218cM; 9 segments; Predicted: Second to Third Cousin
1.85%; 139cM; 8 segments; Predicted: Second to Third Cousin
0.93% 70cM; 6 segments; Predicted: Second to Fourth Cousin

According to Family Tree DNA, there is a 99% probability that we will share DNA with those related to us at the second cousin level and closer. As Dr. Mountain intimated, the longer segments should be the proof of the suspected relationship and a confirmation of a known relationship.

Finding your Roots Video

I have embedded the episode of “Finding Your Roots” below.


Watch John Lewis and Cory Booker on PBS. See more from Finding Your Roots.

6 comments:

  1. Dear Jim

    I enjoyed reading this blog - and it was the firs time I had seen these two stories. I can actually remember some of the Selma to Montgomery march seen over here on BBC TV when I would have been about 9. Thse are two amazing stories - and just shows what autosomal DNA testing can do - perhaps especially for African-Americans.

    ReplyDelete
  2. Thanks Brian. I appreciate your comments. It was a trying time in US history, but right prevailed. They are amazing stories. The use of autosomal DNA for genealogy has yet to be fully tapped.

    ReplyDelete
  3. Thanks, very much Jim! I was curious to get more detail on that comparison from the show and this clarified quite a lot for me.

    ReplyDelete
  4. So if a suspected 2nd cousin and I were to get tested, if the match is between 6.92% and .93% would that be proof that we're related and that our family tree research that linked us is valid?

    ReplyDelete
    Replies
    1. Logan: Sorry I missed this originally. If you suspect that you are second cousins and are matching at a level consistent with second cousins, it would indirectly confirm your suspicions. Outside of parents and full siblings, the amount shared may represent several possible relationships. It becomes difficult to determine a relationship strictly by the numbers; however, the closer the relationship, the more likely you are to predict the supposed relationship.

      Delete