The Cree Y-DNA Project - detailed analysis of test results (July 2013)
by Mike Spathaky The scope of the Cree Y-DNA Surname Project is outlined in the two-part Review referenced on the right, together with some broad conclusions drawn from the Y-DNA test results available up to April 2013. This article attempts to provide more detailed conclusions through a technical analysis of the results. The full Y-DNA profiles of participants in the Cree Surname Y-DNA Project are available as a downloadable spreadsheet. This is a modified version of FTDNA's chart for the Cree Project (with some data from their Humphrey Project chart). Y-DNA profiles are identified in the spreadsheet by upper case letters A, B, C, etc. which are also used in the table and text of this article. The identification numbers (kit #) used by the testing company, FTDNA, are also given. Cree ID numbers in the spreadsheet refer to person details in the Cree On-line Genealogy Database. All the test results were analysed using the Y-DNA Comparison Utility developed by Dean McGee. An additional non-Cree result has been added to the analysis because it is a close match with some of the Cree results and rare within the whole FTDNA database. Matching each of fifteen or so Y-DNA profiles with every other profile gives rise to over a hundred pairings. Table 1 shows, for every pair of test subjects, the number of steps of difference between their 37-marker profiles. (A step is a difference of one step in one marker, so two steps of difference can be one step in each of two markers, or two steps difference in one marker.) The number of steps of difference between two people's profiles is a measure of how distantly they are related. A difference of three steps means that the two people are related through a common male ancestor for whom there is a 50% probability that he lived more recently than the year 1500. A difference of two steps takes us to the year 1650, again with a 50% chance that he lived more recently than that (and therefore of course a 50% chance that he lived before 1650.). A one-step difference takes us to the year 1770 while an exact 37/37 match means that the two people are have a 50-50 chance of being related through a common ancestor born less than 90 years ago. So they are probably first or second cousins. However the time periods stated are calculated from gene mutation rates. Mutations are random events so the estimates are very approximate indeed which is why we can only state them as statistical probabilities. |
This article aims to provide a more detailed analysis of Cree Y-DNA tests than is contained in the two-part Review article
Related articles:
Download file: |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Table 1. Genetic distances for all pairs of test subjects |
| A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| A | 2 | 3 | 2 | 3 | 15 | 15 | 15 | 14 | 14 | 14 | 25 | 15 | 17 | 17 | |
| B | 2 | 1 | 2 | 2 | 16 | 16 | 15 | 15 | 15 | 15 | 26 | 16 | 17 | 15 | |
| C | 3 | 1 | 1 | 3 | 16 | 16 | 15 | 15 | 16 | 15 | 26 | 16 | 18 | 16 | |
| D | 2 | 2 | 1 | 4 | 16 | 16 | 16 | 15 | 16 | 15 | 26 | 16 | 18 | 17 | |
| E | 3 | 2 | 3 | 4 | 15 | 15 | 14 | 14 | 14 | 14 | 25 | 17 | 17 | 15 | |
| F | 15 | 16 | 16 | 16 | 15 | 0 | 1 | 1 | 1 | 2 | 20 | 12 | 15 | 19 | |
| G | 15 | 16 | 16 | 16 | 15 | 0 | 2 | 1 | 1 | 2 | 20 | 12 | 15 | 19 | |
| H | 15 | 15 | 15 | 16 | 14 | 1 | 1 | 2 | 2 | 3 | 20 | 12 | 15 | 18 | |
| I | 14 | 15 | 15 | 15 | 14 | 1 | 1 | 2 | 2 | 1 | 20 | 11 | 14 | 18 | |
| J | 14 | 15 | 16 | 16 | 14 | 1 | 1 | 2 | 2 | 3 | 21 | 13 | 14 | 18 | |
| K | 14 | 15 | 15 | 15 | 14 | 2 | 2 | 3 | 1 | 3 | 21 | 10 | 15 | 18 | |
| L | 25 | 26 | 26 | 26 | 25 | 20 | 20 | 20 | 20 | 21 | 21 | 23 | 23 | 25 | |
| M | 15 | 16 | 16 | 16 | 17 | 12 | 12 | 12 | 11 | 13 | 10 | 23 | 14 | 21 | |
| N | 17 | 17 | 18 | 18 | 17 | 15 | 15 | 15 | 14 | 14 | 15 | 23 | 14 | 19 | |
| O | 17 | 15 | 16 | 17 | 15 | 19 | 19 | 18 | 18 | 18 | 18 | 25 | 21 | 19 |
The table shows a number of pairs whose profiles are different by four steps or less. These fall into two groups - profiles A to E and profiles F to O (now F to S see Table 3). From the known ancestries of the test subjects we have called these groups the Pennsylvania/New Jersey (PA/NJ) group and the Scotland group. There are further test subjects who are not in either group.
The remaining pairs show genetic distances ranging from 10 to 26 steps. In effect we can say that these pairs are not related. Their male-line ancestries have no link more recent than 1620 years ago and for most the MRCA calculatons give dates back in the Bronze Age.
The conclusions are as follows:
1. All members of the PA/NJ group show a degree of matching to all other members of that group, although only two pairs show a match as close as 8 generations from the test subjects at 50% probability level. The other pairs in this group show typical MRCA distances of 12 to 17 generations at that probability level.
2. In the Scotland group, there are ten pair-combinations of results that show matches as close as 8 generations with one of these being at 3 generations (an exact 37-marker match in fact).
3. Pairings containing one member of each of the two groups and pairings involving the ungrouped test subjects show no significant matches, with MRCAs being between 1620 and 2850 years ago at 50% probability level.
In view of the last conclusion it is useful to deal with the two groups separately and construct a chart of the pairings for each one. Here we will add into the mix the mode profile for each group. This is the hypothetical test result that would match most closely all the actual results in that group. The mode might represent a hypothetical profile of the MRCA of the whole group. The mode profile is calculated by Dean McGee's Y-DNA Comparison Utility.
All the profiles of this group gave at least a 35/37 match to the mode. Profile B was identical to the mode profile across all 37 markers. Profile C had only one marker difference (36/37) from the mode and two markers (35/37) from Profile A. The remaining two profiles, A and E, each had just two markers difference from the mode and three (34/37) from each other. Profile E had four markers different from profile D, the only pairing with a difference as large as this (33/37).
To protect their right to privacy, test subjects are identified in this article by the letters A to S. They can identify themselves by reference to their FTDNA kit number. The Cree "line" or "branch" to which they belong is also shown here. The lines and branches are listed on the Lines and Charts page of the Cree Genelaogy Database Section.
| ID | Kit# | Line# |
|---|---|---|
| A | 130457 | non* |
| B | 269865 | 19 |
| C | 45002 | 21a |
| D | 37881 | 18 |
| E | 37466 | 21b |
| F | 225403 | 10b |
| G | 264144 | 14 |
| H | 266051 | 10b |
| I | 208191 | 17 |
| J | U3541 | 10c |
| K | 38270 | 17 |
| L | 243951 | 1 |
| M | 266775 | 1 |
| N | 266215 | 2 |
| O | 175159 | 6 |
| Q | 267873 | 23 |
| R | 267874 | 16 |
| S | 280003 | 7 |
| mode | A | B | C | D | E | |
|---|---|---|---|---|---|---|
| mode | 360 | 90 | 240 | 360 | 360 | |
| A | 360 | 360 | 510 | 360 | 510 | |
| B | 90 | 360 | 240 | 360 | 360 | |
| C | 240 | 510 | 240 | 240 | 510 | |
| D | 360 | 360 | 360 | 240 | 660 | |
| E | 360 | 510 | 360 | 510 | 660 |
The mode profile is identical to profile B from which it is six generations back to Robert Cree Sr. There is firm evidence that Robert migrated from County Down, Ireland.
With one marker difference from the mode (36/37) is profile C. Four generations back from profile C also takes us back to an original known migrant from Ireland, John Cree of Huntingdon County, PA. There is a 50% probability that the MRCA of these two subjects is 8 generations back from the test subjects, or about two to four generations back from Robert Cree Sr and John Cree. The Cree genealogy in County Down is only conjectural that far back but is consistent with this result.
With one marker difference from profile C and two markers from profile B and the mode is test subject D, a descendant of Joseph Cree of the NJ line whose origin is uncertain from genealogical records. This test subject has a difference of two markers to the mode.
Test subject A is the non-Cree whom we have added to the group because of his matches with its members. He has a two-marker difference from each of test subjects B and D, who themselves have a two-marker difference. He has a three-marker difference with each of C and our final subject E.
Finally test subject E is descended from David Cree of Huntingdon County, PA. He is no closer than a two-marker difference from any other member of the group and two markers different from the mode. This is surprising as there is good family history evidence that he shared a common ancestor with test subject C only about 6 generations back. The descendants of both David and John Cree, both of Huntingdon County, have recounted family traditions that they were brothers. Yet their Y-DNA profiles suggest that David Cree was more closely related to Robert Cree Sr. than to his supposed brother John Cree.
The most distant pair-relationship in the group is a four-marker difference between test subject D of the Joseph Cree line and E of the David Cree branch, each having a two-marker difference from the mode. These two test subjects are predicted to have a MRCA 22 generations away at the 50% probability level.
The closeness of the relationships between these test subjects supports in general terms the hypothesis that the lines they represent are related in the proposed time frame of the County Down Crees. In detail however, the suggested time-frames of the relationships seem to contradict the hypotheses of the actual relationships. In particular the lines of David and John Cree seem to be placed too far apart in view of the strong evidence that they were in fact brothers. I am reluctant to abandon this genealogical evidence and believe the discrepancies can be explained in terms of gene mutations between test subject E and his ancestor David Cree of Huntingdon County. In any case the level of uncertainty inherent in determining the time-frames of matching Y-DNA profiles could also account for the discrepancies.
Gary Maher, the Administrator of the Cree Surname Y-DNA Project, has written a more detailed analysis, shown below, of the Pennsylvania/New Jersey group results. This takes account of a recent (September 2013) re-testing of all the samples by the testing company FTDNA at the higher resolution 67-marker level and provides more detailed insight into the relationships of the lines within this group.
In the Scotland group the first observation is that matches are generally much closer than they are in the PA/NJ group.
Technical details:
- Infinite allele mutation model is used
- Average mutation rate varies: 0.0031 to 0.0031 (rates derived by Doug McDonald from the Sorenson database)
- Probability is 50% that the TMRCA is no longer than indicated
- Average generaton: 30 years
| mode | F | G | H | I | J | K | Q | S | |
|---|---|---|---|---|---|---|---|---|---|
| mode | 90 | 90 | 240 | 240 | 240 | 360 | 360 | 240 | |
| F | 90 | 90 | 240 | 240 | 240 | 360 | 360 | 240 | |
| G | 90 | 90 | 240 | 240 | 240 | 360 | 360 | 240 | |
| H | 240 | 240 | 240 | 360 | 360 | 510 | 240 | 90 | |
| I | 240 | 240 | 240 | 360 | 360 | 240 | 240 | 360 | |
| J | 240 | 240 | 240 | 360 | 360 | 510 | 510 | 360 | |
| K | 360 | 360 | 360 | 510 | 240 | 510 | 360 | 510 | |
| Q | 360 | 360 | 360 | 240 | 240 | 510 | 360 | 240 | |
| S | 240 | 240 | 240 | 90 | 360 | 360 | 510 | 240 |
Whereas in the PA/NJ group the two closest pair-relationships each showed 8 generations to MRCA, here in the Scotland group there are two pairs of results, those of F and G, and those of H and S, that each show an exact match over the full 37 markers. Profiles of the first pair are only one step in one marker different from the profiles of the other pair. Two other results, I and J, are each only one step change in one marker from the F and G pair (36/37). Two further profiles, K and Q, still have only a two-marker difference from that pair (35/37).
The F and G pair of identical results are identical to the mode profile of the Scotland group. F is a member of the Aberdalgie, Perthshire line and G of the Sprotbrough line of Yorkshire, England. With DNA matching this close there is normally a 50% probability that the MCRA is no more than 3 generations (say 90 years) back from the test subjects. Here it is the closeness of the match that is surprising. Our documented genealogies reach back, respectively, six generations (Test subject G to William Cree of Sprotbrough) and seven generations (Test subject F to Thomas Crie in Aberdalgie). So the actual MRCA must be further back than three generations. But we should remember that genetic distance calculations from Y-DNA profiles are always probabilities only, never certainities.
Similarly the H and S pair of identical results are one step of one marker different to the mode profile of the Scotland group. H is a member of the same Aberdalgie, Perthshire line as F, while S is of the line of Cree glovers of Perth. Again it is the closeness of the match that is surprising. Our documented genealogies reach back, respectively, eight generations (Test subject H to Thomas Crie in Aberdalgie) and 13 generations (Test subject S to William Cree whose estimated birth years is about 1574). So the actual MRCA should be further back than seven generations. But we should remember that genetic distance calculations from Y-DNA profiles are always probabilities only, never certainities.
The closeness of the matches within each of these pairs, together with the fact that participants F and H were thought to be of the same line is prompting an urgent re-evaluatrion of the documentation behind their genealogies. For example our current hypothesis that William was from the Abernethy family of John Crie and Marjory Lean needs closely looking at as this family was not thought to be closely related to the Aberdalgie line. On the other hand the F and G match confirms the theory that William Cree of Sprotbrough (married 1744) was of Scottish descent.
Test subject J is descended from William Cray of the Fife and Devon line. His Y-DNA profile also differs by only one step from the mode. This points to close relationships between William Cray of Fife and both Thomas Crie in Aberdalgie and William Cree of Sprotbrough. This should encourage genealogical research to establish where the links occur.
Profiles I and Q are both of descendants of Irish lines and the close match between them (36/37) shows their close relationship. I is also only one step of one marker different to F and G (the Scotland group mode) while Q is only one step from the other identically matching pair H and S. So the long hypothesised descent of these lines from Scotland is confirmed. A real possibility arises of the actual genealogical links being found.
Technical details as above
The profile of test subject Q differs by two steps from the mode but only by one step from each of H, J and S. Q is the first Ulster-born Cree to participate in the Project and is a descendant of the Lisburn Cree line (Branch 23b, Lisburn weavers). This confirms that his Ulster line is closely related to the Scottish lines and that he has a most recent common ancestor with Line 10 who was born around the year 1700.
Participant R is a descendant of Hugh Cree, weaver, of Bangor, County Down. His profile shows no connection to any other Cree and is unique across all surnames in the FTDNA database. There are two possible situations that could account for this result. First is the possibility that the Cree line of Hugh Cree, weaver of Bangor, is genetically completely separate from other Cree lines that have been tested. This seems unlikely but could result from the surname Cree having been a variation of a different surname such as McCrea. Secondly it is possible that the surname Cree was passed down to Hugh Cree and on to test subject R from one of our known Cree lines but the male DNA profile has not been passed down by the same route. In other words there may have been a non-paternity event the ancestry of this Cree line.
The closest match with R in the entire FTDNA database is a 33 out of 37 marker match with a Daniel Frederick Nesbitt. The common ancestor of Daniel Nesbitt and Cree test subject R is likely to be about 17 generations (say 500 years) ago.
The two test subjects I and K are descendants of Robert Cree of Newtownards, County Down, Ireland. One of these, test subject I, is also only a step away from test subject F of the Aberdalgie line. This is of great significance since it supports the theory that Robert Cree's line originated in Scotland. The second test subject of this Newtownards line, with profile K, is a close family member of the first. Surprisingly therefore, the two members have a difference of one marker from each other. This cannot be due to a non-paternity event within the family as such a event would give rise to a Y-DNA profile completely different to any other Cree line. It must be due to a mutation occurring in Y-DNA of K himself. The line of test subjects I and K traces back to County Down through South Wales and County Cork.
There is strong genealogical evidence that most of the PA/NJ lines originated from known Cree lines in County Down, Ireland, and circumstantial evidence that all of them did. As for the link from County Down back to Scotland, although there is in this case no genealogical evidence, the circumstantial evidence is overwhelmingly in favour of such a link.
It is therefore very surprising that the Y-DNA test results show no matches between cross-group pairings, that is, that there were no matches between any members of the MA/NJ groups and any members of the Scotland group. All the cross-group pairings show MRCAs of 81 to 95 generations. That takes us back to the Iron Age!
It is particularly surprising that the two participants discussed in the previous section, I and K, show no link to the PA/NJ group. Their earliest known ancestor Robert Cree of Newtownards of County Down was thought to be the father of Robert Cree of Cumberland Township, PA, the progenitor of the line of PA/NJ test subject B. Clearly there is an inconsistency to be resolved here.
The genealogical and historical evidence strongly suggests that the Cree families in County Down Ireland resulted from the settlement of that area by Scottish migrants as part of the Ulster Plantations of the Seventeenth Century. There is also good evidence for most if not all of the PA/NJ lines that they are descended from the County Down Crees through the so-called Scotch-Irish migrations of the Eighteenth Century. So why is there a zero match between the Scottish Cree lines and the PA/NJ lines?
Of the sixteen Y-DNA 37-marker test results, five show no significant match to any of the others. The remaining eleven results have fallen into two groups of five and six members respectively. Within each group all the profiles give matches of at least 35/37 markers with their respective group modes. All members of both groups therefore have a most recent ancestor in common with their group mode who lived, wih a 50% probablity level, more recently than the year 1650.
A firm conclusion can now be drawn that the the County Down Cree line of Robert Cree of Newtownards (Line 17); the Sprotbroguh, Yorkshire, Cree line (Line 14); the Aberdalgie line (Branch 10b), the Fife and Devon branch (now Branch 10c) and the Lisburn weavers line (Line 23b) have been shown to be closely related.
The tests support the genealogical evidence that Lines 18, 19, 21a and 21b from PA/NJ Cree lines are descendants in the male line of ancestors in County Down. The hypothesis that the County Down ancestors of these same lines are descended further back from Scottish Cree lines is NOT confirmed.
The lack of a match between any member of the Scotland group and any member of the PA/NJ group is a surprise and we have set out a scenario to explain it. We clearly need further Y-DNA test subjects to come forward in order to calrify this issue.
Test results suggest that the Derbyshire Cree line (Line 1), the Huguenot Cree line (Line 2) and the Massachusetts Cree line (Line 6) are separate from each other and from the PA/NJ and Scotland lines. However more test participants are needed from these lines to confirm these conclusions, since we have only one test result for each line and non-paternity events could be involved. Indeed results have strongly suggested that one branch of the Derbyshire Cree line resulted from a non-paternity event, probably in the late eighteenth century.
by Gary Maher, Administrator of the Cree Surname Y-DNA Project
We recently commissioned an upgrade of all of the results for the distinct group of New Jersey and Pennsylvania lines from 37 markers to 67 markers. The results are now in, and here are our conclusions based on the new information:
1. The closest match is between Joseph (Line 18**) and John (Line 21a). We know the common ancestor cannot be more recent than 7 generations back. There is a 50% chance he was 8 generations back, and 75% at 10 generations. So there is a 50% chance that Joseph and John were 1st cousins or closer, and a 75% chance that they were 3rd cousins or closer.
2. The McCreerys are still in the game. The McCreery DNA is closer to Joseph's line than David's line (Line 21b) is to Joseph's. Statistically, the common ancestor of Joseph and the McCreerys is only one generation back of the common ancestor of Joseph and John.
3. We still can't say at this time that any two of these lines have a common ancestor more recent than any other of them. We can say that there seem to be more differences between the McCreerys and the other lines, which is unsurprising given the different surname. But Joseph and John share one mutation, while Robert (Line 19), John and David seem to share another. So there must have been at least one situation where separate lines underwent the same mutation.
3a. Nevertheless, if we hypothesize that the split with the McCreery line was earlier than the splits between the NJ/PA Cree lines, and that the match between Joseph and McCreery at CDYb is a coincidence because both mutated to that result, then the common mutation at DYS460 in the Joseph and John lines suggests that Joseph and John have a common ancestor who is not an ancestor of any of the other lines.
4. Only the McCreery DNA has other non-Cree families which are close enough matches to be worth investigating. For David and John's lines, there are no 67 marker matches outside of our project. Joseph's line has a few close matches, but only two are as close as the Cree project matches. The McCreery DNA has several matches which are likely to have been more closely related than some of the Crees (although the closest match is Joseph's DNA). We suspect that means that the McCreery DNA most resembles the DNA of the earliest common ancestor (who would have been a closer match to other families) and that the other lines mutated away from that profile.
We do not believe further test upgrades would yield any more information at this time. The best way to make more progress here (aside from traditional genealogical research) would be to have distant cousins within known lines tested so we could start to determine the age of some of these mutations. If a mutation is present in more than one descendant of a common ancestor, it is likely that their ancestor also had that mutation.
Related articles:
Review Part 1
Review Part 2
My Results
Appendix
Download file:
Spreadsheet of Cree Y-DNA profiles