Presented here are ten new phylogenetic charts based on the data from the 333 high-resolution (HVR1 plus HVR2) mtDNA haplogroup K entries on FTDNA’s MitoSearch when it reached a total of 750 entries. Earlier discussion of the material may be seen here. Nine of the diagrams were produced using Fluxus-Engineering software with data input from Tom Glad’s mtDNAtool. For most of them I used the Median Joining Network Calculations method. An exception will be noted below. The tenth chart was created using PowerPoint. Note that there are now 775 K entries on MitoSearch.


In addition, there is one new map showing the European origins of five K subclades, also created in PowerPoint from the same MitoSearch data.


Referred to often below is the 2006 paper on Ashkenazi Jewish mtDNA by Dr. Doron Behar, now the Chief mtDNA Scientist at FTDNA. That paper has a chart which defines the subclades of haplogroup K. Most subclades are defined by coding-region (CR) mutations which are outside the HVR regions found on MitoSearch. However, some subclades are defined at least partially by HVR mutations or may be predicted from those mutations.


The Fluxus diagrams are based on various combinations of five HVR2 mutations, 114T, 146C, 152C, 195C and 497T, which were used by Behar in his definitions. Most of these are recurrent mutations; that is, they occur in more than one place on the K chart. They are not usually recent random mutations; they are usually thousands of years old. The exception to the above is 497T, which is very close to being similar to Y-DNA’s UEPs (Unique Event Polymorphisms). But in general, even mutations which define subclades may show up in another subclade as a random or personal mutation. I will point those out. In most cases where Fluxus has suggested alternate routes, I have deleted the least likely one.


For each diagram, you should start with the node marked “KROOT,” which represents the original K’s “basic six mutations”: 16224C, 16311C, 16519C, 73G, 263G, and 315.1C. On some diagrams I have helpfully colored that node blue. It is always found on either the left side or in a corner. The other nodes are proportional to the number of exact entries represented and are labeled by the first MitoSearch entry involved. The order of the nodes from the Kroot is meant to be in chronological order. However, there are often alternate ways to draw the charts. Fluxus has no magical way to determine the order or age of the mutations or nodes. The list of mutations between nodes is not in chronological order.


Most of groupings on the chart are not officially subclades (also called subhaplogroups), so I call them “haplotype clusters.” Each cluster usually consists of a “perfect haplotype” – the basic six mutations plus the mutations specific to that chart – and haplotypes with additional mutations.


I have discussed the HVR2 position 524 insertions previously. I recently learned that the Sorenson Molecular Genealogy Foundation (SMGF) always uses the form 524.1A, 524.2C, etc., based on a Wilson paper. The majority of FTDNA entries use the reverse, 524.1C, 524.2A, etc. On the several charts where both versions occur I have normalized them by simply deleting the letters, using 524.1, 524.2, etc. In at least one case, a perceived genetic difference of eight becomes an exact match! I have also discussed previously my practice of counting each pair of these insertions as a single mutation, especially in determining genetic distance. Behar did not use the 524 insertions when creating his K chart, probably for two reasons. One, being very recurrent they make a phylogenetic chart much more complicated. Two, they don’t occur in the three Ashkenazi subclades, at least not that I have found on MitoSearch or in the K Project. They also don’t occur (with one known exception) in the K1c/K1c2 and some other subclades. As I have said: “Recurrent but not random.” However, 21% of K haplotypes contain at least one pair of these insertions. According to the SMGF Top 50 Mutations list, they only appear in 5% of mtDNA haplotypes overall.




497T: This haplotype cluster contains the basic six K mutations plus 497T, but not HVR2 mutations 114T, 146C, 152C or 195C. Mutation 497T, which defines subclade K1a, was probably the first to occur after the founding of K. This complex diagram features a star pattern, with several subsidiary stars and branches. There are three examples of the perfect haplotype on MitoSearch, from Germany, Poland and France. The largest node 5GZKD, with 11 entries, adds 309.1C. An alternate route for node CEPYQ might be drawn from there. In all, the cluster has 13 ancestral originations from Western Europe – nine from Germany, eight from the Britain, three from Ireland, three from Portugal, two from Italy (Sicily), and only four from (not very far) Eastern Europe – two from the Czech Republic, and one each from Slovakia and Poland. K1a with all its branches contains almost 60% of K’s members. For this diagram I used the Reduced Median method. One entry has a rare back mutation on 16519C. One person with several additional mutations (at node 2QNRE) has been designated a K1a* by a full-sequence test, while another one at 5QZKD has been designated a K1a1a. The difference is due to CR mutations. HPM3U has the 16320T mutation which usually defines K1c2, but here it is a random personal mutation.


146C: Behar uses 146C, probably the next oldest mutation after 497T, to define subclade K2 and as part of the definitions of K1c and K1b2. This small, branching cluster contains 146C without any of the other four; so most of these entries might be classified as K2* without benefit of CR mutation testing. Of the three perfect examples, one traces back to Ireland and the other to Lithuania. Of the rest, three are British (all but one Irish) and one is German. 16270T seems to be forming a subcluster; there are other examples in the K Project. One entry includes 16234T, whose main role otherwise is to define the largest Ashkenazi subclade K1a1b1a. Here it is just a random mutation.


146C-497T: This cluster combines the two earliest mutations and also shows a branching pattern. There are no perfect examples. Interestingly, while there are 13 examples of this combination on MitoSearch, it does not appear on Behar’s chart. Three entries trace back to Ireland, one to England, and one to Sweden. A large subcluster is defined by 16245T. The node with 114T is also included in the diagram for that mutation.


152C: Standing alone, this mutation starts a small cluster with almost as many branching points as nodes. There are no perfect examples. I have deliberately left in a pair of alternate routes, although I would lean toward eliminating the line from 38M62 to XJXJ3. Only two of the four list countries of origin: Ireland and the Czech Republic (specifically Bohemia while part of Austria).


146C-152C: Behar uses this combination, with some CR mutations, to define K2a, and with 498- to define K1c. I discussed this either/or situation recently and published a combined Fluxus chart based on K Project members. This new diagram omits all the entries in K1c and K1c2 (defined by adding 16320T). I did a PowerPoint chart for those subclades back in May. This large cluster exhibits a star pattern radiating from a large node representing a dozen perfect examples, well-distributed back to England, Ireland, Poland and Sweden. Entries with additional mutations trace back to Spain, France, Switzerland, Lithuania, Scotland and the Czech Republic. It includes a back mutation on 16519C, not marked as such on the diagram. One of the perfect examples here has been designated a K2a by a full-sequence test. [On October 15, 2006, I created a revised version of this chart by adding 13 entries with HVR2 mutation 512C which partly defines Ashkenazi subclade K2a2a. I had previously excluded those and the K1c/K1c2 subclade entries to simplify the chart. On the revised version I have excluded insertions 309.1C and 309.2C to eliminate reticulations or alternate paths. Those heteroplasmic mutations are very recurrent and would not necessarily appear in the proper chronological order. Only one of the 13 probable subclade K2a2a entries has the 309.1C insertion. These newly added probable Ashkenazi entries trace back to Belarus (4), Lithuania (2), Russia, Poland, Germany, and England.]


152C-497T: A fairly large branching cluster with no perfect examples, this is another branch of the K1a subclade. Members trace back to Jordan, England, Austria, Spain, France and Ireland (2). The last three also have 195C. Most of the entries have 524 insertions; one has three pairs. Two entries have very rare back mutations at 16311C.


146C-195C: A fairly large branching cluster with no perfect examples, there are three tracing back to Germany, two to Scotland, and one each to Finland, Belgium, Sweden and Ireland. One contains 497T and really looks out of place here. Another has 16320T, which otherwise defines K1c2. Yet another has 16524G, usually the definer for K1a9. These last two mutations are very likely random here. One person at NH5N3 has been designated a K1b2 by a full-sequence test.


195C-497T: I have left this diagram pretty much unedited to show how complicated a large diagram can look straight out of Fluxus. Shown are several alternate routes and one set of lines that cross but do not connect. There are five perfect examples, one tracing back to Poland; but the others are brickwalled in the USA. One person at CS8F3 has been designated a K1a* by a full-sequence test. The line marked in orange or red is the preferred main route for Ashkenazi subclade K1a9. That line is of course mostly Eastern European, with examples from Russia, Poland (2), Ukraine (2), Moldova, Romania, Hungary, but also Germany, Austria, England and Ireland. The order of the mutations, including the one with 152C, tracks Behar’s chart perfectly. The blue line is the main route for the 16048A haplotype cluster discussed below, although there are two alternate routes to some with that mutation at node FSVS8. Among the others, three trace to Ireland, with one each to Scotland, Germany and Portugal. I note that another one has a suspicious 512G, rather than the common 512C.


114T: Actually all entries in this cluster also contain 497T, but it does not include those in the largest Ashkenazi subclade K1a1b1a. Most 114T mutations are found there, where it is shown below mutation 16234T, which defines the subclade. Most entries in that subclade on MitoSearch have 114T. Two members of the K Project have been designated as K1a1b1a; both have 114T. Also, Behar’s chart has 114T in subclade K1a1a in conjunction with several CR mutations and 16093C and in subclade K1a1b along with 152C and several CR mutations. Here there are examples of the first of those combinations and others, with origins scattered around the British Isles, Germany and the USA. The only examples of it combined with 152C were not included in the diagram. One was in K1a1b1a. The other was missing 497T; but that may be explained by that person being tested by “Other.” Clearly 114T is a complicated mutation deserving of further study. It must not be confused with HVR1 mutation 16114T, which appears in MitoSearch K’s once to 114T’s 47 times. By contrast, in the largest Haplogroup H at 668 entries, 114T was outnumbered 20 to 2 by 16114T. Ron Scott provided data downloaded from MitoSearch in July for all haplogroups, with a comparison table for HVR1 mutations.


16048A: As a change-up, this haplotype cluster is defined by an HVR1 mutation. It is included in the diagram for 195C-497T above, but it appears distinctive enough and large enough to be entitled to its own cluster. Also, the Fluxus software resists what I think is the proper alignment. So I went back to PowerPoint, which I used for such charts before Tom Glad made available his utility to create Fluxus import files. This cluster has two main branches. Sticking with the HVR1 theme, I call them the 16291T and 16047A branches, with the former being the larger. So far there is only one example at the first node GAEWF. Since that entry traces back to Ireland, I’ll assume until proven wrong that the 16048A mutation occurred there. The other examples go back to Ireland, England and Germany, as shown on the diagram. Two known HVR1-only examples go back to Sweden and Norway. Behar has an unnamed group headed by 16093C next to K1a9 which might include the larger branch but not the smaller. As I’ve stated before, Behar’s data did not include significant samples from the British Isles. I would nominate this cluster to be subclade “K1a10,” but that would depend on one or more full-sequence tests being taken. Again looking at Ron Scott’s data, in July there were a total of 23 16048A’s in MitoSearch; fully 16 of them were in K. This cluster actually has a very active study group which has been encouraging those with 16048A to upgrade to HVR2 results, upload their data to MitoSearch, and join the K Project. They are also trying to find genealogical connections. So far two of their members have tracked their ancestors down to 15 miles apart in one Irish county. For this chart only, I added one entry from the K Project to the MitoSearch entries.


Ashkenazi-K1c/K1c2 Map: Now for something completely different, this is a map showing the European origins of the three Ashkenazi subclades – K1a1b1a, K1a9, and K2a2a - contrasted with those of the K1c and K1c2 subclades. The East-West division is clear, with Germany in the middle as the only country with all five subclades represented. England and Poland each have four of the subclades. I’m still standing by my theory that the 16320T mutation which defines K1c2 occurred in the British Isles.


Finally, all K’s are encouraged to join the K Project by clicking on the blue Join button on their FTDNA personal page.


William R. Hurst

Group Administrator, mtDNA Haplogroup K Project.