PDA

View Full Version : The Similar-Sounding Surnames of Haplogroup R-BY140757



IanFitzpatrick
06-25-2021, 12:47 PM
Link to the paper

https://www.fitzpatrickclan.org/The%20Similar-Sounding%20Surnames%20of%20Haplogroup%20R-BY140757.pdf?fbclid=IwAR0raK1mX766JtxXBESfDeq_G6tG v1p0wNbNOux7kGL_9c_4ykGf5nBRBtI


The Fitzpatrick DNA project is one of the oldest DNA surname research projects found by Dr. Colleen Fitzpatrick.

The work continues and this paper shows how the historical annals are disrupted when put under the microscope using y-DNA testing.

The line of the 1st Baron of Ossory's descendants have been examined both by the history records and the DNA and with distinct y-STR markers they have been easily identified. Therefore it was no surprised to find a substantial group of Fitzpatrick men with matching STR results and this was indeed the case with close to 100 Fitzpatrick surnames with the distinct STR markers.

STR testing also linked this line with a large line of men with surnames that appeared to come from the Irish O’Braonáin clan. This made prefect sense and aligned quite well with the historical examinations throughout the centuries.

However, as the results from BigY testing came in, the organization of relationships between these two groups produced no connections to deep rooted Irish clans, which we see typically in other existing Irish clan haplotrees.

Also, both the STR TMRCA calculations and the branching on the Haplotree began to show a convergence of all of the Fitzpatrick men to about the year 1500 which was about the time the 1st Baron of Ossory submitted to Henry VIII

https://en.wikipedia.org/wiki/Barnaby_Fitzpatrick,_1st_Baron_Upper_Ossory

There were simply no Fitzpatrick men who tested BigY that went beyond this branch and it was very distant from the possible O’Braonáin block which appeared to go back to about the time of the Norman Invasion of Ireland and the origin of surname use.

So how could a line that was supposedly descended from the Kings of Ossory not leave one single line from the year 1100 to 1500?

The research into the line of "Bran" variant surnames then showed that they had lineage that never left England but possibly lines that came into Ireland shortly after the invasion. The research pointed to an influential family in both England and Ireland for many years with no connection at all to the Irish O’Braonáin clan at all.

Here is a great example of the power DNA research can bring to the historical records and how the written histories need to always be looked upon with a degree of suspicion, everyone must be prepared to read between the lines and let the science speak the record.

Jon
06-30-2021, 09:32 AM
Very interesting Ian - great paper, and thanks for posting.

We had an exchange on here a year or so ago, regarding L513, which you said played a marginal role in the Fitzpatrick project. Did L513 feature at all in the work for this latest paper?

IanFitzpatrick
06-30-2021, 02:30 PM
L513 did not relate to this line, it is under FGC5494 which itself appeared late in the Isles when compared to the other sons of DF13

Under L21>DF13 you can see we have at least 16 distinct Fitzpatrick lines, and what I mean by that is the lines appear to go back beyond the 15th Century at least.

Under L513 there are two lines and the FGC9811 line has matches with quite a number of McGuire's in Cavan County, but that could mean a number of things and we have just started to expand the testing of this group.

The idea that a clan was made up of lines that are all related and the Chiefly line was made up of unbroken transfers from father to son is being completely blown up by DNA. All of these Fitzpatrick lines, most which go back to the time of the origin of surnames, took on the name and the reason they took on the name in most cases remains a mystery. The best way to solve it, as we did in the paper, is to combine what the DNA tells us and let it guide us through the written history.

Much of the Mac Giolla Phádraig clan history was written by men who had a vested interest in pleasing the clan leaders of the time, it would have been classed as "fake news" today. Most of the modern day descendants, even in the face of overwhelming DNA evidence, hang on to these stories. If you go back and read some of the old threads here you will find that most of these "Kingly Line" stories seem quite far fetched in the face of the DNA evidence we have today. Also, the stories we have today might also appear to be a little laughable as the wealth of DNA information increases.

45385

RobertCasey
06-30-2021, 07:17 PM
Around seven months ago I analyzed your part of haplotree that includes BY140757. Unfortunately, the TMRCA estimates from BigTree set me back for while. FGC5549 was supposed to be around 1550 YBP which is on more recent end of predictable haplogroups which reside in the 1500 YBP to 2500 YBP time frame. It appears that FGC5549 is probably more like 3000 YBP. In fact, YFULL has FGC5549 as being 3700 YBP - probably too old of a TMRCA. Looking at the tree structure of this part of the haplotree, it is pretty obvious that A1487 falls into the 1500 to 2500 YBP time frame. My first round analyzed FGC5549 but I found several signatures indicating that this branch is several hundred years older than 2500 YBP. Once I reduce the scope to A1487, I was able find a ten marker signature and create YSNP model using AcaStat that has 100 % accuracy across 58,000 testers under haplogroup R. I then ran the data that had on hand and manually extracted another 20 or so more recent testers. Here is the last source file for A1487:

http://www.rcasey.net/DNA/Temp/A1487_HG_R_20201117E.xlsx

I also put this data into SAPP and it produced a html version of the chart:

http://www.rcasey.net/DNA/Temp/A1487_SAPP_Output_20201117E.html

As you can see on the dates, this data is somewhat dated but reveals that A1487 has very high accuracy of prediction. The AcaStat worksheet includes both the input and output of this statistical package (available for only $10 plus tax). The model is based on the number of markers that match the A1487 signature (column K has that formula) and the genetic distance from the A1487 signature (column L has that formula). Column DT has the actual binary logistic regression model in another formula. These columns can be copied and pasted into new rows without any editing. I used only Y67 markers testers. As the percentage of Y111 testers becomes higher, you could probably add the markers 68 to 111 to give more YSTR signatures.

I did attempt to manually find new confirmed testers but did not look at nearby possible predictable testers that would add a lot more testers. Feel free to add more new testers as well as updating existing testers. The AcaStat worksheet shows both the input and output for AcaStat for the model constants. The SAPP tab can be copy and pasted into a .txt file for SAPP. These two new worksheets are just copy and pasted from the master worksheet. The statistical model takes the guess work out figuring out which YSTR only testers to include (as well as some testers with ancestral YSNP testing above A1487). Collecting the data as you probably know is 90 % of time but YSNP prediction and SAPP charts are less than 10 % of time. Feel free to add more data to the spreadsheet as you see fit.

IanFitzpatrick
06-30-2021, 10:35 PM
This is not my Haplogroup and as you can see we have done our own dating thank you, it is based on 72 BiGY tests, we do not use 67 markers to calculate age.

Also, the aging is based on historical mile markers that go back into the 1400’s and the emergence of the surname Braham ca. 1200 AD

RobertCasey
06-30-2021, 11:47 PM
Charting uses both YSNP mutations and YSTR mutations. For R-L226 (my predictable haplogroup), 75 % of the branches comes from YSTR only branches. There are 353 Big Y tests in the
L226 files. I am not a fan of YSTR only TMRCA estimates that SAPP produces - but in this case, they are better than the McDonald estimates - but this is not normally the case. Both A1487
and L226 are more fortunate than most haplogroups. Our haplogroup not only has a direct descendant of King Brian Boru (around 1000 AD) but we also have charts of his ancestry back to
around to 500 AD - these document the clan names that branched off of these lines. There is a surprising tracking for many clan names from pre-surname time frame.

http://www.rcasey.net/DNA/R_L226/Haplotrees/L226_Home.pdf#Page=9

Not using YSTRs would be discarding 75 % of your branching and the ratio of YSTR/YSNP branches continues to increase as the resolution of testers grows to Y111 and Y700.
Unfortunately, the 100 testers that belong to the O'Brien surname cluster struggle to map their genealogy to historical documents as there are not many lines traced down
to the present - just the major lines of royalty. We are very fortunate to have a member of this royal line to test - his dedcendancy from King Brian Boru requires proof
for every generation and he must be a male descendant of this royal line to qualify. I try to avoid this mapping of genealogy to historical documents - but there are others
that continue to make this mapping - it works quite well for the few lines that are well documented but these are very few in number. I concentrate on revealing over 25
surname clusters under L226. Again, many are disappointed that they do not belong to the larger surname surname cluster of their surname. There are tons of NPEs
for less royal lines over the last 1000 years.

With over 850 Y67 or higher testers (353 Big Y tested), this sample size allows me to eliminate the statistical variation of the YSNP process. The years per YSNP between
L226 and its 25 surname cluster varies from 48 years to 350 years per YSNP. This reveals as the sample size of testers grow, the years per YSNP branch declines. Here
is my novel approach to determine TMRCA dates using surname clusters as milestone TMRCA event:

https://www.youtube.com/watch?v=sKaxanrxBgs&t=1682s

IanFitzpatrick
07-01-2021, 01:26 AM
We have a Fitzpatrick cluster, 16 Fitzpatrick men, under ZZ31, it goes up to DC194 ca. 1100AD

RobertCasey
07-01-2021, 05:29 AM
These Fitzpatrick testers belong to at least three different surname clusters:

ZZ31_1 > DF194 > FT212775 - this includes H2174, H2238 and 278471
ZZ31_1 > FGC5628 > BY4102 > DC40 > FT62906 > FT159770 - this includes 911728 and H1174 (just moved H1174 to this branch)
ZZ31_1 > FGC5628 > FGC5623 > DC782 > Y5610 - this tester has not joined our project but he shows up as a match with Y5610 - he is a direct descendant of King Brian Boru (which starts with DC782)

The TMRCA of ZZ31_1 is around 646 AD - this is a very accurate TMRCA since this is an intermediate branch between L226 and several surname clusters. It is probably within one generation 25 to 30 years. The TMRCA between L226 (500 AD) and
the 31 surname clusters under L226 (1000 AD) have all the statistical noise filtered out. The years between L226 > FGC5660 > Z17669 is only 48 years per YSNP branch. The years between Z17669 and ZZ31_1 is only 51 years. Here is my latest TMRCA calculations for branches between L226 and its 31 surname clusters:

http://www.rcasey.net/DNA/Temp/L226_TMRCA_Calculations_20210303A.xlsx

Thanks for the heads up that there are so many L226 Fitzpatrick testers. I found eight new YSTR matches that have the surname of Fitzpatrick. They must be still be at the default of private. You have obviously kept your genealogical slant by working on many surname clusters with the Fitzpatrick surname. Around one third of the Casey testers belong to three branches under L226 - the others are many very small clusters that cover many diverse haplogroups. My TMRCA methodology is extremely accurate for YSNP branches between L226 and its 31 surname clusters. For those under surname clusters and those that have no surname clusters below their branches, I just use the average years per YSNP which is currently 70 years per YSNP branch. This is probably too conservative and generates estimates that are older than reality. However, for the O'Brien surname cluster, we now have one branch at 1490 AD which is still a little old but not by much. Accuracy is driven by sample size - recruiting and privacy issues make obtaining good sample sizes pretty labor intensive.

IanFitzpatrick
07-01-2021, 10:50 AM
The TMRCA of ZZ31_1 is around 646 AD - this is a very accurate TMRCA since this is an intermediate branch between L226 and several surname clusters.



I think this is off by quite a bit, the TMRCA for ZZ31 is closer to 1000 AD. Look at all of the deep rooted surname clusters that branch off the sons of ZZ31, that didn't happen before 1000-1100 AD or they would not share surnames back to these SNP's. Your work on A1847 is completely off as well.

Also, this is not a thread about ZZ31, this discussion doesn't belong here.

Here is the A1487 group and its relationship with other surname blocks using the view of looking at FGC5494. On the far right is a huge Maxwell cluster and the red line is approximately the time of surname origin, it lines up very nicely with the emergence of the Braham cluster referenced in this paper. The red line is easy to draw across most of the sons of DF13 when viewed on the Big Tree, it is a very distinct line. Now if you count up the SNP's the line is around 20 SNP's deep, some lines due to the imprecise coverage of BigY testing are missing quite a number of SNP's, this is why you need to have a large overview of the picture in order to calculate the years/SNP. Y-full throws out most of these SNP to do it's calculations but comes up with the exact same age for A1506 as I do in the paper. However, since we had so many samples under A1487 (72 BigY tests and 189 STR tests) it is easy to determine the tree based on the longest SNP lines, when you do this you will find the years/SNP in general is about 40-45 years.

If you find the longest SNP lines under ZZ31 you will also observe it is about 22-24 SNP's deep as well, so it also emerged around the time of the origin of surnames in Ireland, not in the 7th Century.

45395

Mike Magill
07-04-2021, 09:03 PM
Robert your enthusiasm is infectious and your work on A1487 appreciated, but your A1487 tree is a great example why you can't just load data into SAPP, push a button, and hope for the best.

A1487, particularly under A1499, is beset with parallel mutations and so putting anything more than a modicum of confidence in STRs is misplaced. Take, for example, your treatment of of DYS459b. The modal for DYS459b under A1487 is 10. It then becomes, very characteristically, 9 under A1499. Unless it is handled that way you are going to get a mess, and other STRs are just impossible to 'make sense' of. At least you ignored the CDYs; some other STRs need to be ignored as well in certain sub-branches, that is until branching is correctly assigned - then they can be turned on again. All I can say is...thank goodness for the BigY.

SAPP needs to be reigned in with A1487, and that's easy enough once you understand things like: there are NO Brannans who are under A1496, they are ALL under A1506>BY140757. There are NO Daltons who are A1488, they are ALL under A1506>FT12563. And there are NO Fitzpatricks above A1488. In other words, the best way to build the A1487 haplotree is by sub-groups first, and with an intimate knowledge of the individual lines - for example ALL FT70038 Branans descend from a man who settled in Virginia in the late 1600s.

Mike Fitzpatrick PhD
Fitzpatrick DNA Project