PDA

View Full Version : Interesting P312 Facts



Webb
03-29-2016, 05:15 PM
I was curious the other day and while browsing YFull's tree under P312, I noticed some interesting facts.

There are currently only 7 snps directly below U152. This includes U152*. L2 of course is the largest snp just below U152.
Below L21 there are only 4 snps, this includes L21*. DF13 is the largest with 24 snps at the next lower level.
What really stood out was that there are currently 22 snps just below DF27, including DF27*. If Yfull counted ZZ12 as Alex does on his tree, then 18 snps would be just below ZZ12, leaving four outside ZZ12, one of these Z195/Z196, which is the equivalent of L2 and DF13.

So at first glance, one would make the observation that DF27 has a much wider and quicker spread than the other two large P312 clades. Any thoughts?

Muircheartaigh
03-29-2016, 09:48 PM
An incredibly large number of extinctions across all Haplogroups.

miiser
03-29-2016, 10:09 PM
I was curious the other day and while browsing YFull's tree under P312, I noticed some interesting facts.

There are currently only 7 snps directly below U152. This includes U152*. L2 of course is the largest snp just below U152.
Below L21 there are only 4 snps, this includes L21*. DF13 is the largest with 24 snps at the next lower level.
What really stood out was that there are currently 22 snps just below DF27, including DF27*. If Yfull counted ZZ12 as Alex does on his tree, then 18 snps would be just below ZZ12, leaving four outside ZZ12, one of these Z195/Z196, which is the equivalent of L2 and DF13.

So at first glance, one would make the observation that DF27 has a much wider and quicker spread than the other two large P312 clades. Any thoughts?

In general, the tree exhibits a large amount of variation in both the number of branches per node and the number of phylogenetically equivalent SNPs per node - much greater variation than would be expected from single, randomly distributed SNPs. Also, there are generally more branches per node than would be expected if SNPs are randomly distributed at a rate of 1 per ~144 years. Population growth bursts and bottlenecks may account for some portion of the variation, but they are not able to account for all of it.

The only explanation that is able to statistically make sense of this is if SNPs sometimes occur in runs - in other words, an uneven pace of SNP mutations. Sometimes the gap between SNPs is much greater than ~144 years, and at other times there are many more than a just a single SNP within ~144 years. When the SNPs do occur, they sometimes occur in bursts of not just 1 or 2, but as many as 10 or more.

Below ZZ12, there was probably a stretch of time significantly greater than ~144 years during which no detectable SNPs occurred.

This wide variation in the rate of SNPs can be easily seen in parallel branches that, by definition, must be the same age. One branch may contain 20 SNPs while its sibling branch contains only 5.

ArmandoR1b
03-30-2016, 12:50 AM
We need a lot of high quality P312 ancient specimens with high resolution Y-DNA testing results and reliable radiocarbon dating from whenever it really first appeared and 4,200 years ago or whenever the SNPs 1-step downstream from P312 first appeared. It won't be until then that we know whose hypothesis for the runs of SNPs is correct. Rathlin1 was a high quality specimen and was radiocarbon dated to 2026–1885 BC (3976-3835 ybp) and is positive for DF21 which is estimated to have formed 4300 ybp. Bell Beaker Quedlinburg [I0806/QLB 28] is radiocarbon dated from 2296-2206 BC and is P312+ but I don't think the quality was good enough for high resolution results. So there still hasn't been any high quality P312 specimens with reliable radiocarbon dating from close to the period that the SNPs 1-step downstream from P312 first appeared.

K33
03-30-2016, 02:06 AM
The only explanation that is able to statistically make sense of this is if SNPs sometimes occur in runs - in other words, an uneven pace of SNP mutations. Sometimes the gap between SNPs is much greater than ~144 years, and at other times there are many more than a just a single SNP within ~144 years. When the SNPs do occur, they sometimes occur in bursts of not just 1 or 2, but as many as 10 or more.Is it possible that DF27 diversified so quickly due to natural selection caused by the Iberian climate? As opposed to Central/NW Europe whose client might have better reflected the northern portion of the Pontic Caspian steppe? The 45th parallel appears to be a natural dividing line for red hair frequency, for example-- much lower than the U106 & L21 branches for instance. Obviously y-dna does not tell the whole story, but....

http://cdn.eupedia.com/images/content/red_hair_map_europe.jpg

Webb
03-30-2016, 02:36 AM
I mentioned ZZ12 as a caveat because one well placed snp can change the look of an entire tree.

miiser
03-30-2016, 04:38 AM
I mentioned ZZ12 as a caveat because one well placed snp can change the look of an entire tree.

Alex's tree has a few more branches than YFull, simply because his database is larger and so he's seen branches that YFull hasn't yet seen. The number of testers who participate in YFull is pretty small. And I think Alex is willing to take the time to look at SNPs that YFull and others filter out. But Alex's tree has a large enough sample size that I think it's likely he's discovered well over 90% of branches older than 1000 years, approaching 100%. I don't expect we'll see many additional splits in old haplogroups as more samples are added to the tree. I think the structure that we see in Alex's tree is pretty close to the "true" final structure of the tree for extant lineages. And this structure is unmistakably indicative of an irregular rate of SNP distribution. Single, randomly distributed SNPs just aren't capable of producing the tree structure that we have.

miiser
03-30-2016, 05:35 AM
We need a lot of high quality P312 ancient specimens with high resolution Y-DNA testing results and reliable radiocarbon dating from whenever it really first appeared and 4,200 years ago or whenever the SNPs 1-step downstream from P312 first appeared. It won't be until then that we know whose hypothesis for the runs of SNPs is correct. Rathlin1 was a high quality specimen and was radiocarbon dated to 2026–1885 BC (3976-3835 ybp) and is positive for DF21 which is estimated to have formed 4300 ybp. Bell Beaker Quedlinburg [I0806/QLB 28] is radiocarbon dated from 2296-2206 BC and is P312+ but I don't think the quality was good enough for high resolution results. So there still hasn't been any high quality P312 specimens with reliable radiocarbon dating from close to the period that the SNPs 1-step downstream from P312 first appeared.

Ancient DNA will help nail down the timing of P312's immediate descendants. But even without ancient DNA, the tree structure itself is sufficient to deduce without doubt that the SNP rate has not been uniform. The general tree structure - the large number of branches per node, the large number of phylogenetically equivalent SNPs per node, and the large variation of SNP count between parallel branches - all evident systematically throughout the tree - cannot possibly be generated by single, fully random SNPs.

Given that the tree structure itself clearly indicates the ubiquitous presence of SNP runs (and, conversely, SNP droughts), it is probable that this irregular distribution is also the explanation for the particular case of DF27 (or rather ZZ12). The only other possible interpretation, capable of generating the observed tree structure, is to argue that nearly all the growth of the P312 population over the past thousands of years has occurred within a small number of explosive, very brief expansions. And this position, while being mathematically consistent with the tree structure, is not credible.

ArmandoR1b
03-30-2016, 12:20 PM
Ancient DNA will help nail down the timing of P312's immediate descendants.
It will also possibly break up the large run of phylogenetic equivalents of L21 into smaller blocks of SNPs debunking the large run hypothesis.



But even without ancient DNA, the tree structure itself is sufficient to deduce without doubt that the SNP rate has not been uniform. The general tree structure - the large number of branches per node, the large number of phylogenetically equivalent SNPs per node, and the large variation of SNP count between parallel branches - all evident systematically throughout the tree - cannot possibly be generated by single, fully random SNPs. There is no question that the large runs can be caused by "An incredibly large number of extinctions" as stated by Muircheartaigh.


Given that the tree structure itself clearly indicates the ubiquitous presence of SNP runs (and, conversely, SNP droughts), it is probable that this irregular distribution is also the explanation for the particular case of DF27 (or rather ZZ12). The only other possible interpretation, capable of generating the observed tree structure, is to argue that nearly all the growth of the P312 population over the past thousands of years has occurred within a small number of explosive, very brief expansions. And this position, while being mathematically consistent with the tree structure, is not credible.
Nothing is clear without the evidence from ancient DNA. All else is speculation.

miiser
03-30-2016, 01:18 PM
It will also possibly break up the large run of phylogenetic equivalents of L21 into smaller blocks of SNPs debunking the large run hypothesis.

It's theoretically POSSIBLE for new samples to split phylogenetically equivalent blocks of ancient subclades, but not at all likely. As I pointed out, most branches of old haplogroups have already been discovered. At this point, very few added samples will cause splitting of established ancient nodes. The curve of branch discovery is something like a logarithmic curve, approaching a final value as the sample size increases. The curve has already levelled off and we're near the maximum.

And while there is the slim possibility for new samples to split an ancient phylogenetically equivalent block, new samples are NOT capable of decreasing the number of branches per node. The number of branches per node is already greater than it ought to be for the case of single, randomly distributed SNPs every ~144 years. Additional samples can only have the effect of adding MORE branches to any given node.



There is no question that the large runs can be caused by "An incredibly large number of extinctions" as stated by Muircheartaigh.

Wrong. Large blocks of phylogenetically equivalent SNPs CANNOT be created simply by a large number of extinctions. ALL the branches inevitably contain a large number of extinct lineages, but they DON'T all contain large blocks of equivalent SNPs. The only way you can generate a large block of SNPs from evenly distributed single SNPs is if a haplogroup is whittled down to literally just a SINGLE surviving lineage.

Assuming SNPs occur as singular events once per ~144 years - Unless a haplogroup is whittled down to literally a SINGLE individual due to lineage extinctions, there will still be a branch every few generations. If there are even 2 surviving lineages, then there will likely be 2 distinct branches as the result of a SNP somewhere in one of the lineages. Just TWO surviving men within the haplogroup are sufficient to interrupt the block of phylogenetically equivalent SNPs. Suppose a haplogroup is constrained to a bottleneck of 10 men for a thousand years. After a thousand years, you will not have just a single unbranched group of 10 men, all sharing the same block of SNPs. There would still be a SNP every few generations, even among those 10 men, and the most probable outcome is that you would end up with close to 10 different branches among those 10 individuals.

There may be very rare cases of haplogroups being whittled down to single individual without being whittled down to ZERO individuals. But this is statistically very improbable, and will necessarily be very rare. In most cases, either the haplogroup will retain some number of individuals, even if this number is small, and most of those individuals will be split into separate branches by SNPs occurring every few generations . . . or, the haplogroup will go extinct altogether. The case in which you will see a bottleneck limit a haplogroup to a SINGLE lineage - which is the only case that can generate a large block of unbranched SNPs over hundreds or thousands of years - will be very, very rare.

And the hypothetical bottlenecks are absolutely NOT capable of explaining away the excessive branch-to-branch variation of SNP count. And neither can this variation be explained away by variability of testing coverage. As with the branch discovery, the SNP discovery is near 100% for the older branches. The sample size is in the hundreds for the major subclades of P312. With sample sizes of each branch in the hundreds, SNP and branch discovery doesn't depend on the test coverage of any single sample, but is cumulative of ALL the samples under that branch so that SNP discovery is near 100%. The structure we see now for P312 is the true structure, and the extreme variation of SNP count is a real, undeniable effect.

If one assumes single, randomly distributed SNPs every ~144 years, the only interpretation that fits with the tree structure is to believe that all the population growth of P312 has occurred in fits and starts, with a very small number of explosive expansions separated by long stretches of bottlenecks that constrain haplogroups to SINGLE individuals.

The only reasonable interpretation is that the SNPs are not evenly distributed.

A.D.
03-30-2016, 01:30 PM
There are some studies on the possibly effects of arsenic with regards to mutation. A lot of the subjects being talked about here fall around the time period of the introduction of arsninicated copper working. Toxins, radiation and viruses have all been put forward as causes of mutation and are thought to have occurred in the Copper and Bronze ages. There is also population growth in those periods increasing the statistical chances of increased mutation. There were population slumps too could also effect the whole balance and distribution.

miiser
03-30-2016, 01:49 PM
There are some studies on the possibly effects of arsenic with regards to mutation. A lot of the subjects being talked about here fall around the time period of the introduction of arsninicated copper working. Toxins, radiation and viruses have all been put forward as causes of mutation and are thought to have occurred in the Copper and Bronze ages. There is also population growth in those periods increasing the statistical chances of increased mutation. There were population slumps too could also effect the whole balance and distribution.

Yes, I've heard the arsenic hypothesis. I saw it mentioned by Razyn some time ago, and by others more recently. It makes sense, and I think it's probably a very significant contributing factor for SNP runs. Sickness, vitamin deficiencies, toxic exposure, etc. can all promote DNA damage. Another possible big one is cosmic gamma ray bursts. SNP rate also increases dramatically with the age of the father. So, any time there is a broad societal effect that leads to increasing age of fathers, we should also see an increased rate of SNPs.

There are a variety of known effects that might cause such SNP runs. But it's undeniable, based on the structure of the tree and irregular distribution of SNPs, that they have indeed occurred, and that variations of population growth rate are unable to account for the irregularity.

A.D.
03-30-2016, 02:32 PM
I couldn't asses ' the structure of the tree and irregular distribution of SNPs' I don't know enough. My thinking on a 'Copper scenario' is maybe Guys who owned smelted etc would probably be richer, better fed, have more offspring and live longer. The miners probably died younger and poor etc. but could still have offspring. So there are already compound factors working. So the old rich guy has offspring mutations due to his age and the young miner due to toxic exposure. The point being copper is the common denominator. This is just pulled from thin air but it reasonable and irregular. There was also a vast trade network merchant would have been exposed to diseases they had no or little immunity too, not necessarily fatal ones but left an effect. I'm really just thinking aloud.

MitchellSince1893
03-30-2016, 04:15 PM
Alex's tree has a few more branches than YFull, simply because his database is larger and so he's seen branches that YFull hasn't yet seen...

As Alex got a later start on some branches, this is not always the case. For example Bigtree has 5 branches under U152.

Yfull's Ytree has 6, excluding the separate U152* branches, and 11 branches with them.

For comparison Richard Rocca's latest tree has 9 U152 branches and 17 branches counting the separate U152* branches.

razyn
03-30-2016, 04:19 PM
I just noticed this thread, and agree with miiser pretty much across the spectrum of commentary so far. From past conversations with Hans van Vliet I'm more inclined toward the gamma ray bursts than the arsenical copper theories, but I don't know, and doubt if anybody really does, yet.

Webb, have you (or others monitoring this) given any thought to Alex's suggestion that all of DF27 was originally also ZZ12+; but the Z195 mutation (or something right next to it) took away the ZZ12_1 (or whatever one calls it) from that side of the tree below DF27?

Webb
03-30-2016, 04:56 PM
I just noticed this thread, and agree with miiser pretty much across the spectrum of commentary so far. From past conversations with Hans van Vliet I'm more inclined toward the gamma ray bursts than the arsenical copper theories, but I don't know, and doubt if anybody really does, yet.

Webb, have you (or others monitoring this) given any thought to Alex's suggestion that all of DF27 was originally also ZZ12+; but the Z195 mutation (or something right next to it) took away the ZZ12_1 (or whatever one calls it) from that side of the tree below DF27?

No, because I forgot about his thoughts about the Z195 group losing ZZ12_1. If this is the case, that all of DF27 was and still is, with at least half, this still makes DF27 the bushier tree of the three big P312 clades, with 22 snp's just below DF27. This means to me that DF27 had some success from its inception.

Williamson
03-30-2016, 05:23 PM
For what its worth, I am now less inclined to believe in a possible ZZ12 back mutation for the Z195 branch than I was in the past. Earlier this year I added another variant, Z39893, as a possible equivalent to ZZ12. Its tricky to detect, so I dont yet know if all the ZZ12 branches are positive for it or not. It does however lend support to ZZ12 being a proper downstream branch of DF27.

In the hg19 reference sequence, Z39893 is given by 27417760 TTATATATATATATATATATATA>T. It is an STR, but a fairly extreme one. Its a deletion of 11 copies of (TA) bringing the total at that location down to 13 from 24.

Webb
03-30-2016, 05:54 PM
For what it’s worth, I am now less inclined to believe in a possible ZZ12 back mutation for the Z195 branch than I was in the past. Earlier this year I added another variant, Z39893, as a possible equivalent to ZZ12. It’s tricky to detect, so I don’t yet know if all the ZZ12 branches are positive for it or not. It does however lend support to ZZ12 being a proper downstream branch of DF27.

In the hg19 reference sequence, Z39893 is given by 27417760 TTATATATATATATATATATATA>T. It is an STR, but a fairly extreme one. It’s a deletion of 11 copies of (TA) bringing the total at that location down to 13 from 24.

This is why I do not believe we are past the point of finding new very upstream snps. I am hoping one day see the relationship between all of the P312 clades go through some fine tuning.

ArmandoR1b
03-30-2016, 07:58 PM
It's theoretically POSSIBLE for new samples to split phylogenetically equivalent blocks of ancient subclades, but not at all likely. As I pointed out, most branches of old haplogroups have already been discovered. At this point, very few added samples will cause splitting of established ancient nodes. The curve of branch discovery is something like a logarithmic curve, approaching a final value as the sample size increases. The curve has already levelled off and we're near the maximum.Based on which ancient specimens?


And while there is the slim possibility for new samples to split an ancient phylogenetically equivalent block, new samples are NOT capable of decreasing the number of branches per node. The number of branches per node is already greater than it ought to be for the case of single, randomly distributed SNPs every ~144 years. Additional samples can only have the effect of adding MORE branches to any given node.
If there were people that are positive for some SNPs phylogenetically equivalent to L21 but negative for others that means it wasn't a long run and there were extinctions causing the descendants of the survivors to make it look like a long run. If there hadn't been extinctions then those people who were negative for some of the L21 phylogenetically equivalent SNPs would have created new branches.



Wrong. Large blocks of phylogenetically equivalent SNPs CANNOT be created simply by a large number of extinctions. ALL the branches inevitably contain a large number of extinct lineages, but they DON'T all contain large blocks of equivalent SNPs. The only way you can generate a large block of SNPs from evenly distributed single SNPs is if a haplogroup is whittled down to literally just a SINGLE surviving lineage.
Of course large blocks of phylogenetically equivalent SNPs can be created simply by a large number of extinctions. The descendants of the survivors accumulate SNPs over the generations but extinct lines fail to do that so the branches aren't evident.


Assuming SNPs occur as singular events once per ~144 years - Unless a haplogroup is whittled down to literally a SINGLE individual due to lineage extinctions, there will still be a branch every few generations.
Life was very hard in those periods and it isn't out of the question that a lineage only has one survivor through many centuries but suddenly became successful in creating a large number of surviving descendants.



If there are even 2 surviving lineages, then there will likely be 2 distinct branches as the result of a SNP somewhere in one of the lineages. Just TWO surviving men within the haplogroup are sufficient to interrupt the block of phylogenetically equivalent SNPs.
That's if there are two surviving lineages and if it takes 4-5 generations for a new SNP that's plenty of time for a lineage to die off without creating a new branch.


Suppose a haplogroup is constrained to a bottleneck of 10 men for a thousand years. After a thousand years, you will not have just a single unbranched group of 10 men, all sharing the same block of SNPs. There would still be a SNP every few generations, even among those 10 men, and the most probable outcome is that you would end up with close to 10 different branches among those 10 individuals.Supposition is what all this is based on.


There may be very rare cases of haplogroups being whittled down to single individual without being whittled down to ZERO individuals. But this is statistically very improbable, and will necessarily be very rare.That's another supposition that it is statistically improbable. We don't have time machines to see who went on to have descendants and who didn't.


In most cases, either the haplogroup will retain some number of individuals, even if this number is small, and most of those individuals will be split into separate branches by SNPs occurring every few generations . . . or, the haplogroup will go extinct altogether. The case in which you will see a bottleneck limit a haplogroup to a SINGLE lineage - which is the only case that can generate a large block of unbranched SNPs over hundreds or thousands of years - will be very, very rare. Known to be rare based on which data? Where is this time machine?


And the hypothetical bottlenecks are absolutely NOT capable of explaining away the excessive branch-to-branch variation of SNP count. And neither can this variation be explained away by variability of testing coverage. As with the branch discovery, the SNP discovery is near 100% for the older branches. The sample size is in the hundreds for the major subclades of P312. With sample sizes of each branch in the hundreds, SNP and branch discovery doesn't depend on the test coverage of any single sample, but is cumulative of ALL the samples under that branch so that SNP discovery is near 100%. The structure we see now for P312 is the true structure, and the extreme variation of SNP count is a real, undeniable effect.I just see more speculation being posted without irrefutable data.


If one assumes single, randomly distributed SNPs every ~144 years, the only interpretation that fits with the tree structure is to believe that all the population growth of P312 has occurred in fits and starts, with a very small number of explosive expansions separated by long stretches of bottlenecks that constrain haplogroups to SINGLE individuals.

The only reasonable interpretation is that the SNPs are not evenly distributed.I'm not saying that the SNPs will be exactly evenly distributed but rather that the average is close to the estimation of ~144 years and large runs can be explained by bottlenecks due to extinctions and there isn't any available data that proves it without a doubt. You have the faith it is proven without a doubt but I have yet to see data to back that. Just seeing the results of living people and a very small number of ancient specimens isn't enough. More data is needed.

miiser
03-30-2016, 10:34 PM
I just noticed this thread, and agree with miiser pretty much across the spectrum of commentary so far. From past conversations with Hans van Vliet I'm more inclined toward the gamma ray bursts than the arsenical copper theories, but I don't know, and doubt if anybody really does, yet.

The unique feature of a GRB is that it should be a fairly global phenomenon. They're not able to selectively affect only one haplogroup. Everyone on earth facing that half of the sky would be affected by the GRB. The events were first noticed by geneticists due to the fact that they cause the same observable mutation bonanza in every species at the same time. So, in cases where we see a broad SNP burst across all of P312, I'd say those are good candidates for GRB. The other possible effects, such as arsenic, will tend to affect a limited population of a particular locale or even a single lineage.

miiser
03-30-2016, 10:53 PM
I just see more speculation being posted without irrefutable data.

"Without irrefutable data?" These are directly observable facts! You may just as well ask me to provide irrefutable data to prove that there is a sun in the sky. Just go outside and look at the damn sun!

Some branches have very, very many SNPs while their sibling and cousin branches have very few SNPs over the same time period. Are you claiming this isn't true?

Go ahead and draw a tree with single, randomly distributed SNPs every few generations. Erase a large number of branches for lineage extinctions. Look at the end result and compare it to the actual P312 tree. They won't look anything alike! This is data!

Dubhthach
03-31-2016, 08:46 AM
As a tangent does anyone have any estimate of how many P312+ (all subclades) men are alive today? If we go by Busby 80.65% of his Irish sample was P312 (n=460).

In the 2011 census (new census to be taken end of April 2016) there was 2,272,699 men in Ireland, 271,864 of whom were non-Irish nationals. so there's probably on order of 1.4-1.6 million P312+ men in Ireland today.

I'd imagine given size of UK, French and Spanish populations we are probably looking on at least 25 million plus P312+ men in western Europe today

Webb
03-31-2016, 02:44 PM
At Ytree.net, there are 2339 P312 samples.

L21, 1645 or 70%
DF27, 453 or 19%
U152, 174 or 7%
L238, 26 or 1%
ZZ37, 16 or .7%
DF19, 12 or .5%
DF99, 11 or .47%
A9063, 2 or .08%

So, this begs the question, is this a fair representation of the various P312 clades? Will the numbers change, but the percentages stay roughly the same? As we get more testing in France, Germany, Scandinavia, and Eastern Europe, will the make up of P312 stay fairly static? Is L21 really the juggernaut it appears to be? I would like to think that L21 is a juggernaut in the Isles, but will loose its footing on the continent, but my gut is telling me that if 100 more P312 samples came in from, lets say Germany, the percentages might not be that far off from what we see above. I know that testing bias favors the British Isles.

ArmandoR1b
03-31-2016, 04:54 PM
"Without irrefutable data?" These are directly observable facts!
The only observable fact is that there are groups of SNPs in the living men that have been tested.


You may just as well ask me to provide irrefutable data to prove that there is a sun in the sky.
The analogy is ridiculous.


Just go outside and look at the damn sun!
I see that you are having a hard time keeping your composure. Doesn't do much to support your argument.


Some branches have very, very many SNPs while their sibling and cousin branches have very few SNPs over the same time period. Are you claiming this isn't true?
I am claiming that we don't have enough data to know if it is over the exact same period or if there was a lot of extinctions. I have already stated this.


Go ahead and draw a tree with single, randomly distributed SNPs every few generations. Erase a large number of branches for lineage extinctions. Look at the end result and compare it to the actual P312 tree. They won't look anything alike! This is data!
I've done and I still come to the same conclusion that the runs could be from multiple extinctions.

You should realize by now that you aren't going to change my view on this topic and I will continue to reiterate my points.

Megalophias
03-31-2016, 05:08 PM
"Without irrefutable data?" These are directly observable facts! You may just as well ask me to provide irrefutable data to prove that there is a sun in the sky. Just go outside and look at the damn sun!
Heh, some of us cave-dwellers don't even know where the sky is, i.e. where are you getting the P312 branch length data anyway? Is it publicly accessible?

ArmandoR1b
03-31-2016, 05:18 PM
At Ytree.net, there are 2339 P312 samples.

L21, 1645 or 70%
DF27, 453 or 19%
U152, 174 or 7%
L238, 26 or 1%
ZZ37, 16 or .7%
DF19, 12 or .5%
DF99, 11 or .47%
A9063, 2 or .08%

So, this begs the question, is this a fair representation of the various P312 clades? Will the numbers change, but the percentages stay roughly the same? As we get more testing in France, Germany, Scandinavia, and Eastern Europe, will the make up of P312 stay fairly static? Is L21 really the juggernaut it appears to be? I would like to think that L21 is a juggernaut in the Isles, but will loose its footing on the continent, but my gut is telling me that if 100 more P312 samples came in from, lets say Germany, the percentages might not be that far off from what we see above. I know that testing bias favors the British Isles.

The Isles stats aren't just based on people from the Isles. They are based on the descendants of people from the Isles that have centuries in the U.S. and Canada as well as people still living in the Isles. If all of Latin America and Iberia had the same proportion of NextGen testing with YFull analysis as people from the U.S. Canada, and the Isles, then the amount of DF27 would be very close to L21.

Webb
03-31-2016, 06:00 PM
The Isles stats aren't just based on people from the Isles. They are based on the descendants of people from the Isles that have centuries in the U.S. and Canada as well as people still living in the Isles. If all of Latin America and Iberia had the same proportion of NextGen testing with YFull analysis as people from the U.S. Canada, and the Isles, then the amount of DF27 would be very close to L21.

Yes. That is what I said. Testing bias favors the British Isles. To me this means anyone with British Isles ancestry whether currently living there or abroad. You have mentioned this caveat about Latin America and Spain testing proportions before. At one time I would have agreed with you, but I bet that next to people of British ancestry the next largest ancestral group testing DNA is Spain/Latin America.

corner
03-31-2016, 06:51 PM
This recent thread covers the bias in the hobbyist databases in some detail: http://www.anthrogenica.com/showthread.php?6422-Geographic-Imbalances-in-FTDNA-Database-Some-Comparisons

ADW_1981
03-31-2016, 07:47 PM
How frequently do L21+ guys find YDNA male relatives whose MRCA is within 500-750 years? In some cases I've seen some early American families propagate quite well in the New World, sometimes with the odd NPE thrown in who share different surnames, even though they most certainly share a recent male ancestor. In other circumstances, the male lineage was not successful at all.

I'd be curious to see if there is any pattern to why this might be. I'm wondering how class, profession, or location might factor into migrational history? I am not aware of any YDNA relative (via brothers) moving to the USA, so it's no surprise no matches have turned up. No close matches from England came as a bit of a surprise.

miiser
03-31-2016, 10:02 PM
Heh, some of us cave-dwellers don't even know where the sky is, i.e. where are you getting the P312 branch length data anyway? Is it publicly accessible?

All of this data is directly observable from any public tree, such as Alex's tree. You don't need to know the node-to-node time length of a single branch in order to compare SNP count from one branch to another. You just compare the count along two or more branches from a node to modern time, and it guarantees that all branches are the same length of time. This is publicly accessible data, directly observable from any SNP tree.

If you do this, you will find that the variation of the SNP count exceeds the variation of a random distribution. MJost has in the past provided just such a count and calculated the variance for a large database of samples. When he published the outcome of this exercise, I helpfully pointed out that the variance of his data exceeded the expected variance of a random binomial distribution by a factor of nearly two, which indicates that the distribution is not fully random but instead exhibits structure and clumping. No one seemed to appreciate my helpful data at the time. I suppose some people only like data if it supports their own argument.

Likewise, the count of branches per node and equivalent SNPs per node is directly observable.

To be frank, the non randomness of the SNP distribution is not only directly observable, but OBVIOUS to anyone with any experience in statistics. (I say this not as a formal argument, but just as an observation.) People with experience in statistics know what random distributions look like, and this is not it. It only takes a glance at the tree to realize that the SNP distribution is structured, with large clusters - based on a simple yet informative visual observation that, lo and behold, SNP distribution is structured, with large clusters! No one with any statistics experience could still believe that it's a random distribution after a glance at the tree. Not to be insulting, but it's obvious to me because I have experience and know what I'm talking about. If it's not immediately visually obvious to an observer of the tree, then they have no relevant experience and wouldn't be able to understand any mathematically rigorous argument I might provide, such as comparisons to a random binomial distribution.

Someone who can't tell a triangle from a square by simply looking at them sure as hell won't understand any mathematical argument dealing with the number of sides and vertices or the sum of the angles.

razyn
03-31-2016, 10:45 PM
Likewise, the count of branches per node and equivalent SNPs per node is directly observable.

And if you want to observe it in DF27, an example cited early in this thread that particularly interests me -- look here: http://www.ytree.net/DisplayTree.php?blockID=29

I mean, really look. Try not to look through polarized glasses, that only let you see what reinforces a priori assumptions, e.g. that SNPs occur at a rate (averaging one SNP in 141 years, or whatever); or that groups of them happen sequentially (or in "runs"), as distinguishable from infrequent events producing multiple mutations in one male person -- whose line, subsequent to the multiple mutations, has had great breeding success compared with other lines (many of which die off, or are geographically marginalized).

miiser
03-31-2016, 11:04 PM
And if you want to observe it in DF27, an example cited early in this thread that particularly interests me -- look here: http://www.ytree.net/DisplayTree.php?blockID=29

I mean, really look. Try not to look through polarized glasses, that only let you see what reinforces a priori assumptions, e.g. that SNPs occur at a rate (averaging one SNP in 141 years, or whatever); or that groups of them happen sequentially (or in "runs"), as distinguishable from infrequent events producing multiple mutations in one male person -- whose line, subsequent to the multiple mutations, has had great breeding success compared with other lines (many of which die off, or are geographically marginalized).

Really, ALL of the major subclades of P312, including DF27, have been very successful, having out-reproduced their competitors. This is HOW they became the major subclades. If they hadn't had a large number of offspring, they wouldn't be the dominant subclades that they are today. So having a large number of offspring in a small number of generations is not unreasonable. But what's unreasonable is to imagine a yo-yo of population growth, oscillating wildly from one extreme to the other. Lineages don't struggle along with a single surviving male for a thousand years, and then suddenly become prolific, producing 25 or so offspring in one or two generations without a branch defining SNP. Struggling haplogroups with unhealthy genes that are dying off usually continue to struggle until they die off and go extinct. And healthy prolific lineages continue to prosper and proliferate. Violent swings from prosperity to near extinction and back to prosperity are not credible, especially when they affect only a single lineage and not the entire regional population. Prosperous lineages will generally continue to be prosperous and struggling lineages will generally struggle their way into extinction. The thing about genes, after all, is that they are inherited.

GoldenHind
03-31-2016, 11:21 PM
I'd be curious to see if there is any pattern to why this might be. I'm wondering how class, profession, or location might factor into migrational history? I am not aware of any YDNA relative (via brothers) moving to the USA, so it's no surprise no matches have turned up. No close matches from England came as a bit of a surprise.

Some of the YDNA lineages no longer surviving were not necessarily wiped out. They may well have ended with female children. I have three same surname cousins who have between them a total of seven children. All seven are females. Obviously his has nothing to do with class, profession or location, unless there is something peculiar with the water in California. If so, it doesn't seem to be affecting many people.

miiser
03-31-2016, 11:26 PM
Really, ALL of major subclades of P312 have been very successful, having out-reproduced their competitors. This is HOW they became the major subclades. If they hadn't had a large number of offspring, they wouldn't be the dominant subclades that they are today. So having a large number of offspring in a small number of generations is not unreasonable. But what's unreasonable is to imagine a yo-yo of population growth, oscillating wildly from one extreme to the other. Lineages don't struggle along with a single surviving male for a thousand years, and then suddenly become prolific, producing 25 or so offspring in one or two generations without a branch defining SNP. Struggling haplogroups with unhealthy genes that are dying off usually continue to struggle until they die off and go extinct. And healthy prolific lineages continue to prosper and proliferate. Violent swings from prosperity to near extinction and back to prosperity are not credible, especially when they affect only a single lineage and not the entire regional population. Prosperous lineages will generally continue to be prosperous and struggling lineages will generally struggle their way into extinction. The thing about genes, after all, is that they are inherited.

For example, look at L881 directly below ZZ12. Someone who's committed to the single, random SNP interpretation has to believe that ZZ12 had twenty some male offspring (counting only SURVIVING lineages - the total number of offspring including extinct lineages would had to have been much greater than even this) within a generation or two without a single branch defining SNP during this expansion. And then, immediately following this extraordinary expansion, L881 endured a harrowing stretch of thirty-some SNPs, equating to thousands of years at over a hundred years per SNP, during which time only a single male of this lineage survived into modern times to be sampled. To what desert island was L881 exiled, that this lineage would experience such a harrowing bottleneck while his cousin ZZ41 continued to branch regularly?

Not credible.

A much more credible explanation is that the L881 lineage experienced an exceptionally large number of SNPs within a short time period, and the ZZ12 expansion occupied more than just a generation or two.

GoldenHind
03-31-2016, 11:36 PM
At Ytree.net, there are 2339 P312 samples.

L21, 1645 or 70%
DF27, 453 or 19%
U152, 174 or 7%
L238, 26 or 1%
ZZ37, 16 or .7%
DF19, 12 or .5%
DF99, 11 or .47%
A9063, 2 or .08%

So, this begs the question, is this a fair representation of the various P312 clades? Will the numbers change, but the percentages stay roughly the same? As we get more testing in France, Germany, Scandinavia, and Eastern Europe, will the make up of P312 stay fairly static? Is L21 really the juggernaut it appears to be? I would like to think that L21 is a juggernaut in the Isles, but will loose its footing on the continent, but my gut is telling me that if 100 more P312 samples came in from, lets say Germany, the percentages might not be that far off from what we see above. I know that testing bias favors the British Isles.

Both L238 and DF99 (and possibly DF19 as well) appear to be much more common on the continent than in the British Isles and Ireland, so are likely to be more effected by the FTDNA bias. So far L238 is primarily concentrated in Scandinavia, and DF99 in what I would describe as central Europe. I have no doubt that a sampling in Scandinavia would show a much higher percentage of L238 than 1% of P312, and one in Germany (or at least some regions of it) a higher rate of DF99 than .47%. The Genomes of the Netherlands study found DF99 to be 4.3% of P312 in Holland, nearly 10 times greater than the amount you cite. In the same study DF19 was 10.75% of P312 there, twenty times higher than you amount you cite.

Megalophias
04-01-2016, 04:21 AM
Is it too much to ask for some actual statistical values? Maybe a nice graph? Here, have some P312 branch lengths:8538

As you can see there is clearly a non-random statistical mechanism at work causing there to be exactly 38 SNPs to present. The mutational effects of i-calcium, most likely. :biggrin1:

Megalophias
04-01-2016, 04:38 AM
For example, look at L881 directly below ZZ12. Someone who's committed to the single, random SNP interpretation has to believe that ZZ12 had twenty some male offspring (counting only SURVIVING lineages - the total number of offspring including extinct lineages would had to have been much greater than even this) within a generation or two without a single branch defining SNP during this expansion. And then, immediately following this extraordinary expansion, L881 endured a harrowing stretch of thirty-some SNPs, equating to thousands of years at over a hundred years per SNP, during which time only a single male of this lineage survived into modern times to be sampled. To what desert island was L881 exiled, that this lineage would experience such a harrowing bottleneck while his cousin ZZ41 continued to branch regularly?

Not credible.

A much more credible explanation is that the L881 lineage experienced an exceptionally large number of SNPs within a short time period, and the ZZ12 expansion occupied more than just a generation or two.

ZZ12 is defined by 2 palindromic region SNPs. I can't even guess what that would mean in terms of time, if anything. But this is one branch out of umpteen branches, we *expect* some branches to have very few mutations over a long time and some to have many mutations in a short time as part of the normal random distribution. You can't just point at one branch as if it means anything. Why don't you actually do a statistical analysis? If you are right it ought to be publishable anyway.

And as I'm sure you are aware your personal incredulity about a demographic scenario is not going to convince anyone else. I don't expect you to run simulations of course, but there is a reason we do statistical analyses and run simulations, and don't just eyeball the data and trust to our intuitions - much less to the intuition of some random person on the Internet.

ADW_1981
04-01-2016, 04:42 AM
Some of the YDNA lineages no longer surviving were not necessarily wiped out. They may well have ended with female children. I have three same surname cousins who have between them a total of seven children. All seven are females. Obviously his has nothing to do with class, profession or location, unless there is something peculiar with the water in California. If so, it doesn't seem to be affecting many people.

My speculation is some of the DF27 lineages are actually resident, or more populous in places like France or Belgium which are heavily undertested currently. I'm one of those who think Bronze Age R1b in Britain/Ireland was overwhelmingly a L21 deal.

miiser
04-01-2016, 04:52 AM
ZZ12 is defined by 2 palindromic region SNPs. I can't even guess what that would mean in terms of time, if anything. But this is one branch out of umpteen branches, we *expect* some branches to have very few mutations over a long time and some to have many mutations in a short time as part of the normal random distribution. You can't just point at one branch as if it means anything. Why don't you actually do a statistical analysis? If you are right it ought to be publishable anyway.

And as I'm sure you are aware your personal incredulity about a demographic scenario is not going to convince anyone else. I don't expect you to run simulations of course, but there is a reason we do statistical analyses and run simulations, and don't just eyeball the data and trust to our intuitions - much less to the intuition of some random person on the Internet.

To be perfectly honest, I don't believe that anyone who isn't already convinced will become convinced by a bunch of quantitative analysis that they don't understand. Most of the posters on this forum probably never made it past high school algebra. Providing a single tangible example, such as ZZ12, at least gives people something that they can maybe possibly wrap their heads around. And this example isn't by any means exceptional. Selection of this branch was arbitrary. Similar examples are ubiquitous throughout the tree. Let people look at whatever branch they want to throughout P312. They will find similar inexplicable (by single SNPs) contrasts between huge SNP blocks and explosive branch expansions everywhere in the tree. If they can't understand what the data says for one branch of ZZ12, it doesn't accomplish anything to demonstrate the same pattern of data everywhere else.

People who understand the math well enough to appreciate it can easily do the same calculations on their own. I don't want them to take my word for it. Have them do the math on their own and then they'll believe me. Give a man a fish . . .

For the people who can't understand the math, there's nothing I can possibly do to convince them, so why bother?

What I expect will happen is, after another year or two, the consensus among the respected posters will gradually come to be that SNP runs have occurred. You already see this shift starting to take place in the forums, with increasingly frequent mentions of arsenic poisoning, etc., with people now often speaking of such phenomena as if it was their own original idea. And there are well respected posters on other forums such as the L21 Yahoo project who previously disavowed mutation runs but now acknowledge them.

After some threshold of consensus is passed, other people will notice the tide turning and realize that they're on the losing side of the argument, and then followers will jump ship and pretend that they held this new position all along. (I recently noticed a frequent poster "thank" a popular poster for a comment regarding the effect of arsenic on SNP rate. This same person had vehemently argued against me in the past regarding SNP runs. Nowadays, they still don't approve when I argue for them, but they seem to go along with the flow when their friends argue the case.) But there will always be some stubborn folk who will refuse to adjust their position and would prefer to go down with the ship.

Citizen science really doesn't work so well as some people like to think it does. Publishing mathematical proofs doesn't really gain you any ground in forums such as these. Look at many of the recent discussions involving ancient DNA finds, and you can see many forum readers rejecting the author's formally published work involving analytical methods (such as dental isotope analysis) in favor of their own private interpretation.

Collective citizen political intelligence has gotten us Trump as the leading Republican nominee, for crying out loud. Take a minute to appreciate that many of those same people are contributing to collective citizen science in these forums. I know that political arguing is forbidden on these forums, so please no one take this as a statement for or against Trump, but simply as an example of how the process of group consensus works - whether applied to politics or science.

Dubhthach
04-01-2016, 08:34 AM
But what's unreasonable is to imagine a yo-yo of population growth, oscillating wildly from one extreme to the other.

Population booms and bursts that affect entire population are well documented even within last 700 years in an Irish context with at least 4 of them, the worst leading to 50% population attrition in the 1650's.

Leaving that aside there is also well documented mass expansion of certain lineages in period after christianity, basically from 1500 years onwards. so perhaps a lineage that had for example undergone "run's" (even multiple run's -- say 4-5 SNP's per run) of SNP's in the distant past (Bronze age/Iron age etc.), and had remained relatively small (extended kindred group) could undergo rapid lineage fission once they came into a position of power.

I think a good potential example of this is "Irish Type III" which appears linked to Dl gCais and who only really appear in records from late 9th/early 10th century onwards. Their expansion been driven by their conquest of "Thomond" (North Munster) after collapse of U Fiachrach Aidhne power in Connacht during the 7th/8th centuries.

After this the lineage underwent rapid expansion into sub-kindreds, some of which declined others underwent rapid expansion. I do think it's great that we have a BigY from a man who has proven genealogy to one of these main kindreds. Namely "The O'Brien" aka "the 18th baron Inchiquin" whose direct descendant of Brian Boru (d. 1014).

Webb
04-01-2016, 10:53 AM
Population booms and bursts that affect entire population are well documented even within last 700 years in an Irish context with at least 4 of them, the worst leading to 50% population attrition in the 1650's.

Leaving that aside there is also well documented mass expansion of certain lineages in period after christianity, basically from 1500 years onwards. so perhaps a lineage that had for example undergone "run's" (even multiple run's -- say 4-5 SNP's per run) of SNP's in the distant past (Bronze age/Iron age etc.), and had remained relatively small (extended kindred group) could undergo rapid lineage fission once they came into a position of power.

I think a good potential example of this is "Irish Type III" which appears linked to Dl gCais and who only really appear in records from late 9th/early 10th century onwards. Their expansion been driven by their conquest of "Thomond" (North Munster) after collaspe of U Fiachrach Aidhne power in Connacht during the 7th/8th centuries.

After this the lineage underwent rapid expansion into sub-kindreds, some of which declined others underwent rapid expansion. I do think it's great that we have a BigY from a man who has proven genealogy to one of these main kindreds. Namely "The O'Brien" aka "the 18th baron Inchiquin" whose direct descendant of Brian Boru (d. 1014).

I think a good example of population change is the Rhine Delta. Due to natural causes you have a flood zone that has caused major population replacement countless times. The Irish Famine. How many lineages were lost?

Dubhthach
04-01-2016, 12:02 PM
I think a good example of population change is the Rhine Delta. Due to natural causes you have a flood zone that has caused major population replacement countless times. The Irish Famine. How many lineages were lost?

Well the Great Famine only killed about 12.5% of population, most of major events prior to that killed alot more. What sets Great Famine apart was the demographic collaspe via emigration it spawned. The decline between 1841 and 1851 was on order of 25% of population half of which via emmigration. so to be honest given how the Irish population had doubled by natural growth in the 40 years before than, I'd imagine that the mass demographic growth if anything reduced rate of lineage extinction.

In comparison the Tudor conquest and subsequent Cromwellian "reconquests" not only led to percentage wise lot higher fatalities they also saw the elimination of ruling Irish landed lineages and imposition of a new ruling elite.


http://www.grantonline.com/grant-family-genealogy/Records/population/population-ireland-1100-200.jpg

(the following graph appears to have wrong data for population in 1800 -- still informative of population boom/burst)
http://www.mapspictures.com/ireland/history/imgCF.gif

miiser
04-01-2016, 12:06 PM
I think a good example of population change is the Rhine Delta. Due to natural causes you have a flood zone that has caused major population replacement countless times. The Irish Famine. How many lineages were lost?

Population swings are a reasonable explanation for cases when the explosion or bottleneck is observed universally or uniformly across all or most of the haplogroups in the region. But this is not the case for the majority of such tree structures. Most of the large SNP blocks and multi-branch nodes, supposedly associated with a population bottleneck or explosion, are isolated to just a single branch - not coinciding with a similar structure in any other colocated haplogroups. Environmental events and such are not able to selectively affect only a single lineage. And so environmental effects, societal trends, and such are not able to account for most of the SNP blocks and multi-branch nodes - including the ZZ12 node mentioned in this thread's original post.

Webb
04-01-2016, 02:10 PM
Well the Great Famine only killed about 12.5% of population, most of major events prior to that killed alot more. What sets Great Famine apart was the demographic collaspe via emigration it spawned. The decline between 1841 and 1851 was on order of 25% of population half of which via emmigration. so to be honest given how the Irish population had doubled by natural growth in the 40 years before than, I'd imagine that the mass demographic growth if anything reduced rate of lineage extinction.

In comparison the Tudor conquest and subsequent Cromwellian "reconquests" not only led to percentage wise lot higher fatalities they also saw the elimination of ruling Irish landed lineages and imposition of a new ruling elite.


http://www.grantonline.com/grant-family-genealogy/Records/population/population-ireland-1100-200.jpg

(the following graph appears to have wrong data for population in 1800 -- still informative of population boom/burst)
http://www.mapspictures.com/ireland/history/imgCF.gif

Scratch the famine!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

GoldenHind
04-01-2016, 06:30 PM
Further to my post above about L238 in Scandinavia, I have dug out the only study from that region I know of which tested for L238, a recent survey from Iceland. There were 96 R1b, of which 72 were P312. The P312 portion was further divided as follows:

42 L21
17 L238
7 DF27
4 U152
1 DF19
1 P312* (which I strongly suspect to be DF99, which wasn't tested)

One could speculate how much of the P312 was brought to Iceland from Ireland by the Vikings in one form or another, not necessarily as slaves. However very little L238 has been found in Ireland, and the bulk of that identified to date has been in Norway and Sweden. L238 is 23.6% of P312, and I would expect a similar result in Norway and perhaps a somewhat smaller percentage in Sweden. Note that L238 in Iceland is second only to L21 among the P312, and greater than the combined total of DF27, U152, DF19 and (probable) DF99.

ArmandoR1b
04-02-2016, 04:25 PM
Yes. That is what I said. Testing bias favors the British Isles. To me this means anyone with British Isles ancestry whether currently living there or abroad. You have mentioned this caveat about Latin America and Spain testing proportions before. At one time I would have agreed with you, but I bet that next to people of British ancestry the next largest ancestral group testing DNA is Spain/Latin America.
The customer participation of BigY testing by people from Iberia/Latin American isn't anywhere near the amount of BigY testing by people with Isles ancestry. This is evident with the numbers that I provided in that thread that I mentioned the caveat about Latin America and Spain testing proportions before.

Let's do this again. First compare population totals. U.S. (non-Hispanic white) 197 million, Canada 35.16 million, U.K. (England, Scotland, Wales, N. Ireland) 63.1 million, Ireland 4.2 million. That's a total of 299.46 million people of which almost half are males (149.73 million) and about 55-60%, after considering the non-Isles whites in the U.S., are going to be R1b (149.73 million*0.55=82.35 million). So now it's down to about 82.35 million R1b-P312 males with Isles ancestry and most are L21

Latin America including Brazil has 548 million after subtracting the indigenous population, U.S. Hispanics are 54 million, Spain has 47.27 million, and Portugal has 10.46 million. That's a total of 659.73 million. Again almost half are males. R1b in Latin America and in U.S. Hispanics is about 50% (301 million Latin American and U.S. Hispanic males*0.5 R1b=150.5 million male R1b) and in Spain and Portugal it is about 57% per Busby (57.73 million Iberian males*0.57 R1b=32.9 million R1b Iberian males). About 87% of the Iberian and Latin American R1b males are P312 so now we're down to 163.83 million R1b-P312 males with Iberian ancestry. Even if the R1b and R1b-P312 in Latin America and U.S. is 20% lower that still makes for 107 million R1b-P312 males with Iberian ancestry and most of them are DF27.

Using these populations there should be more R1b-P312 people with Iberian ancestry in Ytree.net than there should be from the Isles. If you were to count the Iberian and Latin American flags and Iberian surnames at Ytree.net I am sure that they won't even come close to the amount of Isles flags and Isles surnames. Just by looking at the amount of L21 and DF27 it's evident that they won't come close. The R1b-P312 people with Iberian ancestry is probably less than 30% of what it should be and if it were 100% then DF27 would be very close to the amount of L21.

Lugus
04-02-2016, 04:41 PM
The customer participation of BigY testing by people from Iberia/Latin American isn't anywhere near the amount of BigY testing by people with Isles ancestry. This is evident with the numbers that I provided in that thread that I mentioned the caveat about Latin America and Spain testing proportions before.

Let's do this again. First compare population totals. U.S. (non-Hispanic white) 197 million, Canada 35.16 million, U.K. (England, Scotland, Wales, N. Ireland) 63.1 million, Ireland 4.2 million. That's a total of 299.46 million people of which almost half are males (149.73 million) and about 55-60%, after considering the non-Isles whites in the U.S., are going to be R1b (149.73 million*0.55=82.35 million). So now it's down to about 82.35 million R1b-P312 males with Isles ancestry and most are L21

Latin America including Brazil has 548 million after subtracting the indigenous population, U.S. Hispanics are 54 million, Spain has 47.27 million, and Portugal has 10.46 million. That's a total of 659.73 million. Again almost half are males. R1b in Latin America and in U.S. Hispanics is about 50% (301 million Latin American and U.S. Hispanic males*0.5 R1b=150.5 million male R1b) and in Spain and Portugal it is about 57% per Busby (57.73 million Iberian males*0.57 R1b=32.9 million R1b Iberian males). About 87% of the Iberian and Latin American R1b males are P312 so now we're down to 163.83 million R1b-P312 males with Iberian ancestry. Even if the R1b and R1b-P312 in Latin America and U.S. is 20% lower that still makes for 107 million R1b-P312 males with Iberian ancestry and most of them are DF27.

Using these populations there should be more R1b-P312 people with Iberian ancestry in Ytree.net than there should be from the Isles. If you were to count the Iberian and Latin American flags and Iberian surnames at Ytree.net I am sure that they won't even come close to the amount of Isles flags and Isles surnames. Just by looking at the amount of L21 and DF27 it's evident that they won't come close. The R1b-P312 people with Iberian ancestry is probably less than 30% of what it should be and if it were 100% then DF27 would be very close to the amount of L21.

And you didn't even mention France and Germany, which also have a lot of DF27, especially France.

razyn
04-02-2016, 05:50 PM
The customer participation of BigY testing by people from Iberia/Latin American isn't anywhere near the amount of BigY testing by people with Isles ancestry. This is evident with the numbers that I provided...

Well, kind of; but if you are complaining about YTree (or YFull), the 1000 Genomes samples are included. And those are rather heavily skewed toward Latin Americans, of Iberian paternal ancestry. FTDNA's customer base isn't the only source of sequenced data -- although the 1000 Genomes samples were of course sequenced at a lower number of passes.

I would agree that DF27 is probably wildly under-represented in most of the extant literature, including some "juried" papers written with ulterior motives (but approved by the jury, who shared those motives). But I don't see much reason to think that was because Iberia was undersampled. Most of it is because much of continental Eurasia was, and remains, barely sampled at all. (Spain and Portugal -- and France -- had American colonies; Poland and Armenia didn't.) And a good bit is just because DF27 was discovered too late to be considered, as such, in papers more than three years old.

By contrast, L21 is over-represented because the Irish, on whatever continent their descendants are being sampled, have historically had large families. It's true that there are a lot of Irish-Americans in the FTDNA database -- but they include the DF27, U106 and other Irish-Americans, proportionately. Those guys are also buying BigY tests.

George Chandler
04-02-2016, 07:01 PM
To be perfectly honest, I don't believe that anyone who isn't already convinced will become convinced by a bunch of quantitative analysis that they don't understand. Most of the posters on this forum probably never made it past high school algebra.

Maybe that's you're problem..you're using algebra instead of stats?

You still have not provided evidence in terms of showing a genealogy list of one or more family surnames (with multiple lines) that "have a proven MRCA verified with STR testing".. then show that those people in the genealogy list that you've chosen have been tested using tests such as Y Elite or Big Y. You need to list the raw SNP results prior to Sanger validation and post Sanger removal of problematic positions. Everyone knows that there are problematic regions on the Y Chromosome prone to more higher mutation rates but you claim (and correct me if I'm misunderstanding) some sort of mechanism for your runs and gaps that are not random. You keep saying it's obvious..and trying to insult everyone and yet you don't show the evidence. Sending a link to a YFull haplogroup isn't evidence. Theories are great but you need to show the exact mechanism that your talking about..show the mutations in a family line..show that multiple Sanger validated SNP's are happening in a single individual. Just because it's a rare, random occurrence that could happen in a single sample doesn't mean there is some sort of mechanism driving it that should be factored into the age estimates. Each time (and they are rare) what you call gaps or runs they can be explained by and adjusted using Sanger validation. You keep saying that Sanger validation doesn't play a role but then provide no data other than the generic quip about how other who disagree with you are beneath you.

Just because you keep saying it's true doesn't make it so. Nice showmanship though!

George

ArmandoR1b
04-02-2016, 09:29 PM
Well, kind of; but if you are complaining about YTree (or YFull), the 1000 Genomes samples are included. And those are rather heavily skewed toward Latin Americans, of Iberian paternal ancestry. FTDNA's customer base isn't the only source of sequenced data -- although the 1000 Genomes samples were of course sequenced at a lower number of passes.I wasn't complaining and there was no reason to use that term to describe it. I was stating facts about the numbers in the Ytree by Alex Williamson that Webb used. The BigY participation by non-Hispanics with Isles ancestry at YFull far outweighs the 1.5:1 ratio of Iberians to non-Iberians in 1000 Genomes and I was well aware that both YFull and Ytree use 1000 Genomes and the passes by 1000 Genomes were enough to identify subclades of P312 as well as P312. There are the very few 1000 Genomes P312 without a subclade defined. There are only 112 Iberian R-M269 in 1000 Genomes and 73 from GBR (British in England and Scotland) and CEU (Utah residents with Northern and Western European) ancestry which provides the 1.5:1 ratio. The 1000 Genomes DF27 at YTree is 92 and the 1000 Genomes L21 is 33 and U152 is 46. So the large difference in DF27 compared to L21 is more to do with a relatively large number of U152 in the GBR and CEU populations and less to do with the 1.5:1 ratio of Iberians to non-Iberians in 1000 Genomes. However, there are 362 non-1000 Genomes DF27 kits compared to 92 1000 Genomes DF27 and 1626 non-1000 Genomes L21 kits (most of which are from the Isles) at Ytree. There are only 37 Iberian and Latin American non-1000 Genomes DF27 kits and only 10 Iberian and Latin American non-1000 Genomes L21 kits. Do you really think that's a fair amount of Iberian and Latin American non-1000 Genomes kits? That's only 47 kits. Do you really think that 1000 Genomes is still a factor? It's obviously not. Everyone should be checking their lists and counting them twice.


I would agree that DF27 is probably wildly under-represented in most of the extant literature, including some "juried" papers written with ulterior motives (but approved by the jury, who shared those motives). But I don't see much reason to think that was because Iberia was undersampled. Most of it is because much of continental Eurasia was, and remains, barely sampled at all. (Spain and Portugal -- and France -- had American colonies; Poland and Armenia didn't.) And a good bit is just because DF27 was discovered too late to be considered, as such, in papers more than three years old.R1b-P312 outside of western Europe exists but at a much lower percentage but regardless including them would increase the ratio of DF27 to L21.


By contrast, L21 is over-represented because the Irish, on whatever continent their descendants are being sampled, have historically had large families. It's true that there are a lot of Irish-Americans in the FTDNA database -- but they include the DF27, U106 and other Irish-Americans, proportionately. Those guys are also buying BigY tests.However, Iberians and Latin Americans also had large families but they are way under-represented in non-1000 Genomes at Ytree.

Webb
04-02-2016, 10:35 PM
The customer participation of BigY testing by people from Iberia/Latin American isn't anywhere near the amount of BigY testing by people with Isles ancestry. This is evident with the numbers that I provided in that thread that I mentioned the caveat about Latin America and Spain testing proportions before.

Let's do this again. First compare population totals. U.S. (non-Hispanic white) 197 million, Canada 35.16 million, U.K. (England, Scotland, Wales, N. Ireland) 63.1 million, Ireland 4.2 million. That's a total of 299.46 million people of which almost half are males (149.73 million) and about 55-60%, after considering the non-Isles whites in the U.S., are going to be R1b (149.73 million*0.55=82.35 million). So now it's down to about 82.35 million R1b-P312 males with Isles ancestry and most are L21

Latin America including Brazil has 548 million after subtracting the indigenous population, U.S. Hispanics are 54 million, Spain has 47.27 million, and Portugal has 10.46 million. That's a total of 659.73 million. Again almost half are males. R1b in Latin America and in U.S. Hispanics is about 50% (301 million Latin American and U.S. Hispanic males*0.5 R1b=150.5 million male R1b) and in Spain and Portugal it is about 57% per Busby (57.73 million Iberian males*0.57 R1b=32.9 million R1b Iberian males). About 87% of the Iberian and Latin American R1b males are P312 so now we're down to 163.83 million R1b-P312 males with Iberian ancestry. Even if the R1b and R1b-P312 in Latin America and U.S. is 20% lower that still makes for 107 million R1b-P312 males with Iberian ancestry and most of them are DF27.

Using these populations there should be more R1b-P312 people with Iberian ancestry in Ytree.net than there should be from the Isles. If you were to count the Iberian and Latin American flags and Iberian surnames at Ytree.net I am sure that they won't even come close to the amount of Isles flags and Isles surnames. Just by looking at the amount of L21 and DF27 it's evident that they won't come close. The R1b-P312 people with Iberian ancestry is probably less than 30% of what it should be and if it were 100% then DF27 would be very close to the amount of L21.

Always up to a challenge I actually counted flags at Ytree.net. I am going to stick with my stance that the testing bias isn't as disparaging as it once was.

DF99: 1 Spanish/Latin, 1 German
U152: 23 Spanish/ Latin, 5 Germans
L21: 21 Spanish/Latin, 14 Germans
L238 and DF19: No Spanish/Latin
DF27: 106 Spanish/Latin, 16 Germans

Notice that DF27 has more Germans with having only a quarter of the samples as L21. Does this indicate that DF27 is the most common P312 clade in Germany? DF27 has the highest rate of Germans and Spanish/Latin testers of the samples given to Williamson. This is very interesting.

GoldenHind
04-02-2016, 10:55 PM
Your flag count for DF99 at ytree is inaccurate. Counting flags only there is one England, one Ireland (an English family from Ulster), one Peru (an anonymous 1kg sample whose ancestry may or may not be from Spain), two Italy and one Germany. However there are only flags for about half of those listed, and I can assure you they are not an accurate cross-section of those who have been identified as DF99+, who are overwhelmingly of German or English origin. The DF99 on Alex's tree are merely the very few who have ordered the Big Y and submitted their raw data to Alex for analysis, plus two FGC and two 1KG samples.

Webb
04-02-2016, 11:07 PM
Your flag count for DF99 at ytree is inaccurate. Counting flags only there is one England, one Ireland (an English family from Ulster), one Peru (an anonymous 1kg sample whose ancestry may or may not be from Spain), two Italy and one Germany. However there are only flags for about half of those listed, and I can assure you they are not an accurate cross-section of those who have been identified as DF99+, who are overwhelmingly of German or English origin. The DF99 on Alex's tree are merely the very few who have ordered the Big Y and submitted their raw data to Alex for analysis, plus two FGC and two 1KG samples.

I know. But I am not going to guess to the ancestry of any kits without flags. So I only counted those kits with flags. My thought is that just like any other testing initiative, the samples at Ytree.net are just a statistical capture. It is the same of any other initiative. My purpose for doing this is to show that you could say that there is a British/Irish bias when comparing any other ethnic group/ or ancestral origin, not just Spanish/Latin. And until someone can show the exact bias ratio, it is just speculation. I admit that even counting flags is speculation, but I find it interesting that DF27 was pegged as being Iberian very early on, yet has the highest ratio of German testers at Ytree compared to L21 and U152, who quickly grabbed titles of being Insular Celtic and Alpine Celtic or what have you, yet here is DF27 that is spread so far and wide in Europe, that I am wondering how this is possible. Is DF27 the first on the scene. If you look at Ytree, you will see a snp spread that is not matched by L21 or U152. 22 snp's just below ZZ12. All successful. Germans and Spanish from one end of the tree to the other.

GoldenHind
04-02-2016, 11:18 PM
I know. But I am not going to guess to the ancestry of any kits without flags. So I only counted those kits with flags. My thought is that just like any other testing initiative, the samples at Ytree.net are just a statistical capture. It is the same of any other initiative.

I certainly understand your reluctance to guess their ancestry. My point is that there are six flags, not two, for the DF99 listed on ytree.

Webb
04-02-2016, 11:32 PM
I certainly understand your reluctance to guess their ancestry. My point is that there are six flags, not two, for the DF99 listed on ytree.

I was attempting to illustrate that the testing bias of British/Irish applies to every ancestral origin, not just Spanish. I just picked Germans, because I think they are under represented as well. So I chose to compare Spanish/Latin flags to German. I suppose I could go back and compare Spanish/Latin flags to Norway flags, then Sweden flags, then French flags, but I have a little bit of a life. Joke.

ArmandoR1b
04-02-2016, 11:52 PM
Always up to a challenge I actually counted flags at Ytree.net. I am going to stick with my stance that the testing bias isn't as disparaging as it once was.

DF99: 1 Spanish/Latin, 1 German
U152: 23 Spanish/ Latin, 5 Germans
L21: 21 Spanish/Latin, 14 Germans
L238 and DF19: No Spanish/Latin
DF27: 106 Spanish/Latin, 16 Germans

Notice that DF27 has more Germans with having only a quarter of the samples as L21. Does this indicate that DF27 is the most common P312 clade in Germany? DF27 has the highest rate of Germans and Spanish/Latin testers of the samples given to Williamson. This is very interesting.

You are sticking to your stance because you are not using a proper comparison. You are missing the counts of the flags from the Isles and the surnames without flags that are obviously from the Isles. Until you get those totals and compare them to the total population of people from the U.S., Canada and the Isles and then compare the Iberian/Latin flag totals and surnames without flags that are obviously from Iberia or Latin America with the total population of Iberia and Latin America you don't have a way to tell if it is disparaging or not.

I'll explain this again.

There are about 82.35 million R1b-P312 males with Isles ancestry and most are L21. YTree has about 1627 non-Iberian L21 kits most of which are from the Isles. It's probably about 1,500 that are from the Isles. That's about a 1.82e-5 representation of Isles people. There are only 21 Iberian/Latin kits in L21.

A lowball calculation of R1b-P312 males with Iberian ancestry is 107 million. There are only 106 Iberian/Latin kits in DF27. That's about a 9.91e-7 representation of Iberian/Latin people. There are about 349 non-Iberian people in DF27.

A 1.97e-5 representation of Isles people in L21 and a 9.91e-7 representation of Iberian/Latin people in DF27 people is definitely disparaging.

If there were an apples to apples comparison based on population totals there would be about 2,114 Iberian/Latin people in DF27 and 1,500 Isles people in L21.

Webb
04-03-2016, 12:04 AM
You are sticking to your stance because you are not using a proper comparison. You are missing the counts of the flags from the Isles and the surnames without flags that are obviously from the Isles. Until you get those totals and compare them to the total population of people from the U.S., Canada and the Isles and then compare the Iberian/Latin flag totals and surnames without flags that are obviously from Iberia or Latin America with the total population of Iberia and Latin America you don't have a way to tell if it is disparaging or not.

I'll explain this again.

There are about 82.35 million R1b-P312 males with Isles ancestry and most are L21. YTree has about 1627 non-Iberian L21 kits most of which are from the Isles. It's probably about 1,500 that are from the Isles. That's about a 1.82e-5 representation of Isles people. There are only 21 Iberian/Latin kits in L21.

A lowball calculation of R1b-P312 males with Iberian ancestry is 107 million. There are only 106 Iberian/Latin kits in DF27. That's about a 9.91e-7 representation of Iberian/Latin people. There are about 349 non-Iberian people in DF27.

A 1.97e-5 representation of Isles people in L21 and a 9.91e-7 representation of Iberian/Latin people in DF27 people is definitely disparaging.

If there were an apples to apples comparison based on population totals there would be about 2,114 Iberian/Latin people in DF27 and 1,500 Isles people in L21.

Could you please illustrate this calculation and use Poland, then Germany, then France, then the Scandinavian countries individually. Then let's compare these finding against one another and see how they rank. If you can do this I will retract my statement if Spain/Latin America is at the bottom.

ArmandoR1b
04-03-2016, 08:44 PM
Could you please illustrate this calculation and use Poland, then Germany, then France, then the Scandinavian countries individually. Then let's compare these finding against one another and see how they rank. If you can do this I will retract my statement if Spain/Latin America is at the bottom.

Per Busby et al. and the FTDNA projects, except for France, we already know that P312 is very low in those countries and I knew those countries wouldn't affect the stats of DF27 vs L21 much. Here goes the populations and rates of Poland, Germany, France, and Scandinavia anyway. I would need for someone to give me the counts of the flags and obvious surnames for those countries, for the ones without flags, in order to find the participation rates.

Germany has a pop. of 80 million with about 40 million males. The following rates are per Busby. P312 is 20%. P312xL21xU152 is 6% (some will be other than DF27 but I'll use it as a max we can adjust this rate later). So that's 2.42 million possible DF27. L21 has a rate of 2.44% making it 987,000 L21. U152 is bigger here. It has a rate of 11.8% so that's 4.76 million U152. So Germany, if it had the same BigY participation rate of the Isles, lowers L21 more than DF27 even if we lower DF27 to 4%.

France has a pop. of 66 million with about 33 million males. The following rates are per Busby. P312 is 46.8%. P312xL21xU152 is 20.4% (some will be other than DF27 but I'll use it as a max we can adjust this rate later). So that's 3.93 million possible DF27. L21 has a rate of 11.3% making it almost 2.18 million L21. U152 is 15% so that's 2.9 million U152. So France, if it had the same BigY participation rate of the Isles, lowers L21 even if we lower DF27 a little due to L238,DF19, DF99 and so on.

Poland only has a pop. of 38.53 million with about 19.26 million males. The following rates are per Busby. P312 is only 6%. P312xL21xU152 is 0.4% so possibly 90,000 DF27. L21 is also 0.4% making it 90,000 L21. U152 is 5% so 975,000 U152. So Poland, if it had the same BigY participation rate of the Isles, lowers L21 as much as DF27 and even if we lower DF27 the amount is almost negligible.

Scandinavia has even a lower pop. 20.65 million with about 10.32 million males. Using the Scandinavian DNA project P312 is about 13.7%. DF27 is 2.22% making it 229,000 male DF27. L21 is 7% making it 726,000 L21. U152 is 1.85% making it 191,000 U152. Here L21 has a higher rate but it's the smallest pop. so if all of the regions had the same BigY participation rate of the Isles this region would affect the overall rate of L21 vs DF27 the least.

Now do you see how the pops and rates of those regions, along with Iberia and Latin America, with equal BigY participation rates as the Isles would lower L21 and increase DF27 at Ytree so much that DF27 and L21 would be almost equal or DF27 would be even higher than L21?

ArmandoR1b
04-04-2016, 10:11 PM
I found a way to pull the flag data from the following SNP groups - DF27, L21, U152, DF99, ZZ37, DF19, L238, A9063. The total Isles flags are 1009 for all of P312 at http://www.ytree.net/

If I use the pop of 82.35 million R1b-P312 males with Isles ancestry that's a participation rate of 1.22524E-05. If I apply that participation rate to the P312 pops we have mentioned for the other regions and countries this is what should be seen for flags at http://www.ytree.net/ based on the Busby rates and the rates of FTDNA for Latin America and Scandinavia.

8599

Megalophias
04-11-2016, 02:03 AM
Regarding mutation runs: Clustered mutations in hominid genome evolution are consistent with APOBEC3G enzymatic activity (http://genome.cshlp.org/content/early/2016/04/01/gr.199240.115.abstract)

This produces clusters of C>A/G/T mutations (in a small stretch of chromosome).

Webb
04-11-2016, 06:49 PM
You are sticking to your stance because you are not using a proper comparison. You are missing the counts of the flags from the Isles and the surnames without flags that are obviously from the Isles. Until you get those totals and compare them to the total population of people from the U.S., Canada and the Isles and then compare the Iberian/Latin flag totals and surnames without flags that are obviously from Iberia or Latin America with the total population of Iberia and Latin America you don't have a way to tell if it is disparaging or not.

You just proved my point. You don't have a way of telling how disparaging the testing bias is or not. Yet I will pull up quotes from you in past threads where you claim it is 10 to 1 between people of British/Irish ancestry and people of Spanish/Latin American ancestry. How do you know it is 10 to 1? I am claiming that there are other countries, namely France, which you used a mathematical calculation to hypothesize that DF27 might be highest amongst the three large P312 clades, which have a testing bias probably greater than that which exists between people of British/Irish ancestry and people of Spanish/Latin American ancestry.

razyn
04-11-2016, 07:31 PM
Not to imply that I think this means more than other self-reported, non-random, geographically imprecise, non-peer-reviewed data from FTDNA projects; but there is a category called Country of Origin Charts, on the Member Reports menu tab of a project's GAP, and here's the current state of the one for the DF27 project:

8753

GoldenHind
04-11-2016, 11:03 PM
FWIW, I have suspected for some time that DF27 will prove not only to be the most numerous (at least in modern times) subclade of R1b-P312, but also the most ubiquitous throughout Europe. Since it is very probably not the oldest P312 subclade, there must be some explanation of why that is the case.

ArmandoR1b
04-12-2016, 12:51 PM
You just proved my point.I absolutely did not.


You don't have a way of telling how disparaging the testing bias is or not.Of course I do. I know what the world populations are and I know what the rates of P312, P312xL21xU152, L21, and U152 in western Europe and the Americas are.


Yet I will pull up quotes from you in past threads where you claim it is 10 to 1 between people of British/Irish ancestry and people of Spanish/Latin American ancestry. How do you know it is 10 to 1?I used a combination of surnames and flags in the past. If we went by surnames and flags and not just flags there would be a 10:1 British/Irish ancestry to Spanish/Latin American ancestry ratio. Using just the flags there should be an equal or higher number of flags from Iberia/Latin America as British Isles for all P312 but there are only 161 for Iberia/Latin America and 1009 for British Isles. That's that is a 6 to 1 ratio of British Isles to Iberia/Latin America flags. Now count the number of people with British Isles surnames and the number of people with Iberian surnames and you will see that the ratio comes close to 1o to 1. The stats support a huge bias whether it is 8:1 or 11:1.


I am claiming that there are other countries, namely France, which you used a mathematical calculation to hypothesize that DF27 might be highest amongst the three large P312 clades, which have a testing bias probably greater than that which exists between people of British/Irish ancestry and people of Spanish/Latin American ancestry.France should have about 208 flags at Ytree and it has 40 so they are only short about 168 flags. Iberia/Latin America should have about 1652 but only has 161. There difference is huge because the population of Latin America is huge but it's participation rate is tiny in comparison to the participation rate of people with ancestry from the British Isles.

ArmandoR1b
04-12-2016, 02:44 PM
Not to imply that I think this means more than other self-reported, non-random, geographically imprecise, non-peer-reviewed data from FTDNA projects; but there is a category called Country of Origin Charts, on the Member Reports menu tab of a project's GAP, and here's the current state of the one for the DF27 project:

8753

If the participation rate of people with Iberian ancestry were equal to the participation rate of people with Isles ancestry the lines for Spain, Portugal, and Latin America would be far higher than the lines for the Isles which is why just a count of results, without taking into consideration source population and the rate of DF27 in all of those regions, provides a very misleading picture. It's very unfortunate that Latin America has so little money and so little interest in DNA testing. Otherwise, that graph would be more reflective of actual rates.

razyn
04-12-2016, 05:22 PM
It's very unfortunate that Latin America has so little money and so little interest in DNA testing. Otherwise, that graph would be more reflective of actual rates.
That's as may be. I've already made such qualifying statements about the FTDNA chart as I feel a need for, and I'm aware of your separate perspective.

I'm actually more interested in the clearly under-tested eastern parts of Europe than the allegedly under-tested Iberian peninsula -- for which we have a sample (FTDNA + 1000 Genomes + a sheaf of papers, primarily targeting the antiquity of Basques in situ) that's plenty large enough to suggest what's there -- including a high percentage of the unfortunately late-blossoming DF27, and other haplogroups in smaller proportions.

brianlm1
04-27-2016, 06:02 AM
I am interested in researching the origin and history of R1b-A9063. At the moment I seem to be one of three men with this as their most recent SNP. I have been in touch with Alex Williamson (the Big Tree) and Terrance MacFarlane (clan MacFarlane project). How is the best way to stay abreast of research for this marker? Many thanks in anticipation for your insights.

razyn
04-27-2016, 01:52 PM
I am interested in researching the origin and history of R1b-A9063. At the moment I seem to be one of three men with this as their most recent SNP.
That may just be because it's only been tested on the P312 SNP Pack since December, and prior to that was just one of the unknown/unnamed SNPs loosely associated as P312**. That's not a large percentage of the population, by now -- but it's more than three guys, and most of them presumably have yet to order the P312 SNP Pack (or whatever equivalent test YSEQ offers, or a BigY, or FGC Elite). Just by way of saying, any group is small when it's just been identified. I was Z196+ when there were only five of us identified as such at FTDNA; just happened to be in the second batch testing for it. That has turned out to be a very big group.

This thread tracks those guys until a SNP under which they can be gathered has been discovered: http://www.anthrogenica.com/showthread.php?1152-P312**-%28the-real-deal%29-%28P312-U152-L21-DF27-DF19-L238-DF99-%29&p=124683&viewfull=1#post124683

Now that A9063 has been discovered (at 9464776, A to T) and named, a separate thread here might be a good place to track it as more tests come in. The P312 Yahoo group is another likely source of new reports, and you already know about the Big Tree.