January 13, 2014

Paternal and maternal demographic histories (Lippold et al. 2014)

A new preprint on the bioRxiv on the different male/female demographic history of humans.

Red=female, blue=male.

This is probably related to the new paper on selection on the Y chromosome which interprets reduced diversity as evidence for selection.

doi: 10.1101/001792

Human paternal and maternal demographic histories: insights from high-resolution Y chromosome and mtDNA sequences 

Sebastian Lippold et al.

To investigate in detail the paternal and maternal demographic histories of humans, we obtained ~500 kb of non-recombining Y chromosome (NRY) sequences and complete mtDNA genome sequences from 623 males from 51 populations in the CEPH Human Genome Diversity Panel (HGDP). Our results: confirm the controversial assertion that genetic differences between human populations on a global scale are bigger for the NRY than for mtDNA; suggest very small ancestral effective population sizes (less than 100) for the out-of-Africa migration as well as for many human populations; and indicate that the ratio of female effective population size to male effective population size (Nf/Nm) has been greater than one throughout the history of modern humans, and has recently increased due to faster growth in Nf. However, we also find substantial differences in patterns of mtDNA vs. NRY variation in different regional groups; thus, global patterns of variation are not necessarily representative of specific geographic regions.

Link

55 comments:

Annie Mouse said...

"Figure 4 Pictorial representation of the simulation results estimating divergence times
and female and male effective population sizes. Red numbers reflect Nf (with
ancestral Nf at the point of the triangle and current Nf at the base of the red triangle),
blue numbers reflect ancestral and current Nm, the numbers in the black oval indicate
22  the founding effective sizes for the intial out-of-Africa migration, and dates on arrows
indicate divergence times based on the tree in Supplementary Figure 14. Arrows are
meant to indicate the schematic direction of migrations and should not be taken as
indicating literal migration pathways, e.g. the results indicate divergence of the
Downloaded from on January 13, 201415 
 
ancestors of Oceanians 61,000 years ago, but not the route(s) people took to get to
Oceania. "

Daniel Szelkey said...

Does this disprove the existance of the MP and possibly even the LT clades? Does K need to be redsigned.

Annie Mouse said...

Hmm. Not sure about this data. Are they really suggesting that only 1 in 5 chinese men (22k:100k) effectively get to pass their genes on "currently"? Or does "current" mean the last few 1000 years (when there were also more women then men to tilt the balance).

And there seem to be very big regional differences. Is this culturally to do with the relationships between men and women in different regions I wonder.

barakobama said...

Knowledge on genetics is growing extremely fast. It seems like every month something is discovered in ancient DNA or a new hypothesis is created. I wish they gave specifics on haplogroups. There is an obvious difference between mtDNA and Y DNA distribution, I am just assuming it has alot to do with polygamy(only for males)and that females immigrate through marriage into foreign tribes more often.

eurologist said...

We should keep in mind that diversity is an entire different concept for uniparental genes than for recombining DNA. For the latter, even small migrating groups can harbor enormous genetic diversity (if that was their recent background), while subsequent migrating subgroups are mostly distinguished by what parts of their diversity they lost - not by new mutations. And that is lost "forever."

Conversely, within the non-recombining DNA, a local migrating group will likely be highly biased towards only a few and mostly related haplogroups from the get-go, and that will further dwindle down very quickly in isolation. On the other hand, mixing of several such groups after several tens of thousands of years will lead to an apparently large diversity, because each (smallish) population accumulated a substantial number of new mutations that did not get sorted away as it does in recombining DNA (this is visible in the often very long branch lengths). So, in mixed groups, uniparental diversity as measured by "mean number of pairwise differences" is bound to be rather large, because the mutations previously had kept accumulating - there is no loss process.

On the flip side, this diversity "on paper" may not mean much in real life: you may have, e.g., three very specific haplogroups such as one specific y-DNA subgroup each of E, R, and I in some small population. But if due to selection or by chance two are lost, your on-paper high diversity just did an extreme nosedive to almost nothing. The same does not happen, with similar population sizes, with autosomal DNA, for which just a short period of mixing re-shuffles the DNA enough to make the "on paper" high diversity a reality.

eurologist said...

However, the slow NRY mutation rate gives an estimated age for the intial out-of-Africa divergence of about 100 ky, and an age for the divergence of Amerindian-specific haplogroup Q lineages of about 20 ky, while the fast rate gives corresponding estimates of about 60 ky for out-of-Africa and about 12.5 ky for Amerindian haplogroup Q lineages, in better agreement with the mtDNA and other evidence for these events.

So, when they say "better," the authors mean worse? ;)

1Marcelo said...

barakobama's picture featuring bigot Phil Robertson is in bad taste. Please take it down.

terryt said...

"Does this disprove the existance of the MP and possibly even the LT clades? Does K need to be redsigned".

Probably not because, as Annnie says, 'ancestors of Oceanians 61,000 years ago, but not the route(s) people took to get to
Oceania'. The diagram in the paper is very much simplified. In fact it is difficult to make sense of it.

"On the other hand, mixing of several such groups after several tens of thousands of years will lead to an apparently large diversity, because each (smallish) population accumulated a substantial number of new mutations that did not get sorted away as it does in recombining DNA"

But it follows that in a population of consistent size (not expanding) haplogroup diversity will contract. Not all paeents will leave an equal number of descendants so not all parent haplogroups will survive. I see that phenomenon in practice in dairy farms. These days we have records going back many generations for herds. If a farmer is not expanding his herd numbers the number of mother lines decreases, sometime quite rapidly, as the poorer milkers leave far fewer descendants in the herd. At the same time autosomal diversity can be maintained through bulls kept for breeding.

"you may have, e.g., three very specific haplogroups such as one specific y-DNA subgroup each of E, R, and I in some small population. But if due to selection or by chance two are lost, your on-paper high diversity just did an extreme nosedive to almost nothing".

In general minority haplogroups will be lost and the majority haplogroup will in time take over. Other factors can influence things of course.

Average Joe said...

1Marcelo is apparently not a big believer in free speech.

Grey said...

"Are they really suggesting that only 1 in 5 chinese men (22k:100k) effectively get to pass their genes on"

Isn't there a risk of confusion here. A woman passes her mtDNA on to both sons and daughters while a man only passes his y DNA on to his sons so any man who only has daughters won't pass on his y DNA but he still passes on half of his genes.

"I am just assuming it has a lot to do with polygamy"

Is polygamy needed to explain it? Wouldn't random chance - a lineage having all daughters at some link in the chain - make a decline in the diversity of male y DNA more or less inevitable.

"But it follows that in a population of consistent size (not expanding) haplogroup diversity will contract."

Exactly.

Mark Moore (Moderator) said...

If founding populations stayed that small that late, it could help answer some questions I have had about certain late-arriving genes showing up in such a large portion of mankind.

As for the Phil Robertson pic, I should think that his world view would be much more amenable to insuring that his DNA is well-represented in populations 1,000 years from now than the traits of which he is critical. It seems like for that reason at least his pic should be tolerated here.

Chad Rohlfsen said...

They may need to account for y-DNA C-V20 entering Europe early now. If anyone hasn't heard, the rumor is that La Brana 1, is C-V20.

Annie Mouse said...

It is possible that I am mis-interpreting this data. But this is what I think I am seeing. The ratio appears to represent the ratio of effectively breeding women to men.

Africa
Ancient (Nf/Nm)=1.78
Modern=1.74
Change =2.18% (no change)

Australasia
Ancient =1.73
Modern = 1.67
Change = 3.85% (no change)

Europe
Ancient 6.56
Modern 2.11
Change 67.87% (shift towards balance)

Central Asia
Ancient 22.47
Modern 3.43
Change 84.74% (bigger shift towards balance)

South Asia
Ancient = 0.96
Modern = 4.55
Change = -373.89% (massive shift to imbalance)

Americas
Ancient = 4.29
Modern = 2.57
Change = 60% (shift towards balance

So Africa and Australia started off with about 1.7 women breeding to men and that has not changed. Probably reflecting mostly a loss of men to accident/war? Or the loss of women in child birth? I favour the second explanation. I saw a lot of second marriages in my genealogical research due to the death of the first wife. I suspect this must be close to the natural ratio for human animals.

Europe, Central Asia and the Americas start out with way more breeding women than men. Doesn't really fit with packs of men with a few women travelling out into the wilderness. Either a few males male had breeding privileges or most of the men did not survive/connect. I remember a documentary on the collecting of the remnants of an Australian aboriginal tribe in a nuclear testing zone. Huge pack of women and kids, the few remaining men had disappeared a long time ago. Perhaps this was typical for struggling populations? The poor fellow responsible for looking after them until they could be relocated was amusingly terrified by their interest. :)

China appears to have started off with more slightly more breeding men than women! Effectively ratio =1 or apparent balance. I assume women were not immune from child birth death in China. So. Polyandry? Scarcity of women? Patrilocal advantage? Then the it shifted massively in the other direction towards great male disadvantage. Prolonged feudalism? The Horde?

Clearly different things happened in different areas.


1Marcelo said...

@Mark Moore
Mark, do you have any factual evidence that the frequency of homosexuality or the genes linked to homosexual behavior is decreasing over time? I didn't think so.
Anyway, thanks for letting us know that this blog will publish any extremist comment just because the author is likely to have more children. Makes a lot of sense.

terryt said...

"They may need to account for y-DNA C-V20 entering Europe early now".

That is by no means impossible IF true. It is becoming accepted that European C6-V20 is closest to Japanese C1. Less accepted, but just as likely, is that the two of them are in turn close to South Asian C5:

http://www.phylotree.org/Y/tree/

If these connections prove to be correct we have basically a single branch of C spread through South/Central Eurasia from Japan to Europe. We also have the recent evidence that some sort of R was present at Mal'ta some 24,000 years ago. Its origin is most probably South/Central Asia as well, and so we have a set of Y-DNAs spread through Eurasia at an early period.

andrew said...

"Isn't there a risk of confusion here. A woman passes her mtDNA on to both sons and daughters while a man only passes his y DNA on to his sons so any man who only has daughters won't pass on his y DNA but he still passes on half of his genes."

Yes, lineage loss can occur that way. Indeed, for any given individual at a time when survival rates are close to replacement rate, the probability of that man's lineage being lost is about 25% per generation.

But, the probability of lineage loss due to random chance dramatically declines when the population is expanding due to more than replacement rate fertility, and when you get a few generations out with surviving lineages present in more than one man in the community. So, lineage loss is heavily front loaded and strongly concentrated in population bottle-neck eras.

Average Joe said...

the genes linked to homosexual behavior

There is no evidence that genes are linked to homosexual behavior.

Average Joe said...

this blog will publish any extremist comment

What "extremist comment" are you talking about?

Grey said...

@Marnie

"Doesn't really fit with packs of men with a few women travelling out into the wilderness."

That's only liable to happen at a frontier e.g. oil rigs, gold rushes, lumberjacks, fur trappers etc.

.

Things that could shift the ratio:

There are some things that you'd imagine would be constant e.g. a percentage of men who have only daughters.

My guess at the most important variable factors:

1. Relative male vs female mortality.

2. Polygamy.

3. Female infanticide.

So looking at the ratios listed by Marnie I'd guess the M/F ratio goes (from lowest to highest)

1. Foragers with low excess male mortality.

(Both Ancient and Modern Africa and Australia and Ancient South Asia).

2. Foragers with high excess male mortality.

(Ancient Europe and Americas.)

3. Pastoralist Polygamy.

(Central Asia)

4. Female Infanticide Farmers

(South Asia)

Grey said...

Previous was meant to be @Annie Mouse, (reading too many blogs at once.)

Chad Rohlfsen said...

Homosexuality is pretty much linked to epi-genetic triggers in the womb(as they are also able to do the same thing to animals in a lab). It is likely inherited from the opposite sex parent, also showing signs of running in families. Either way, it is not a choice for those who are born that way. Not that there aren't people who make the choice to go that way, or those who are bi-sexual and choose one. Either way, religious views of sexuality and the world around us hold no scientific value. If you choose to pick science out of a book of Bronze Age, highly in-bred, ethnocentric, misogynists then so be it. Keep it to yourself unless you bring scientific data to back it up.

Mark for Summit/Sunnoco said...

"religious views of sexuality and the world around us hold no scientific value."

Which even if true, is not to say they hold no value at all. But it should not be true. Of all possible social arrangements concerning sexuality one particular form has become so dominant that all other arrangements have practically disappeared from the earth.

Oh, maybe some of the anthropologists on this board could name a few mothers-brothers cultures, but whether one believes children are ordained by God to be nurtured in a man-woman traditional family or whether one believes we evolved that way, one arrangement has clearly been the most successful way of spreading ones genes for a long time. That fact ought to have some value, scientific and otherwise.

Maybe this next comment would be better on the post below, where it shows the map, but has anyone else noticed the correlation between proximity to the equator and the age at which the various splits are supposed to have occurred? I do wonder if mutation rates are higher for populations which spend many generations in the intense solar radiation near the equator, which gives a false impression of relative age?

Annie Mouse said...

"anyone else noticed the correlation between proximity to the equator and the age at which the various splits are supposed to have occurred?"

It is climate change in my opinion. The equator was the most amenable to life and the other areas did not open up until later.

terryt said...

"has anyone else noticed the correlation between proximity to the equator and the age at which the various splits are supposed to have occurred? I do wonder if mutation rates are higher for populations which spend many generations in the intense solar radiation near the equator, which gives a false impression of relative age?"

That is more likely to be a result of regions nearer the equator being able to support larger populations than those regions nearer Arctica rather than being a product of increased mutation rates.

1Marcelo said...

Mark,
Homosexuality is not a social arrangement, it is a natural variation of sexual behavior. Its predisposition might have a genetic component. Epigenetic and congenital factors also seem to play a role but we don't know enough about all these mechanisms.

Homosexuality hasn't been wiped out by natural selection in hundreds of species for millions of years. These are the facts. Anything else is superstition.

Mark Moore (Moderator) said...

TO Annie M: That makes sense, but I would feel better if there were proper studies on the question. X-rays in particular should be partially able to get through near the equator but virtually all blocked higher north or south.

To Terryt: That one I am a bit less sure about, unless you mean the finds give us a larger sample size near the equator so we are more likely to find the oldest. I would think mutation RATE would be the same for a population of 1000 or 100,000. They might even get fixed more easily in the small populations.

To Marcelo: Yes, like cleft pallets, homosexuality is still around. But one man's "bigotry" is another man's reasonable opinion. I really came to this site to discuss ancient human populations rather than be subjected to PC harassment. So unless you want to postulate some more specific 'drone hypothesis' or something as to why your favorite fitness-for-reproduction-reducing behavior is still around, I'd just as soon leave you be and ask you to do the same to me.

German Dziebel said...

It looks like nobody has noticed that the revised mtDNA phylogeny presented in this paper made macrohaplogroup N an upstream lineage from what's now a new L3'M node.

eurologist said...

"It looks like nobody has noticed that the revised mtDNA phylogeny presented in this paper made macrohaplogroup N an upstream lineage from what's now a new L3'M node. "

That is an incorrect representation of the results of the paper. There is now an L3MNUR group, which is upstream of L3M and NUR. However, the LRM-NUR and L3-M splits are only 5,000 years apart (or likely ~1/3 more, IMO) - well withing the ~20,000 years of an advantageous ooA ~125,000 - 105,000 ya time period.

L3 then simply is the surviving NE African portion of L3MNUR.

terryt said...

"There is now an L3MNUR group, which is upstream of L3M and NUR. However, the LRM-NUR and L3-M splits are only 5,000 years apart (or likely ~1/3 more, IMO) - well withing the ~20,000 years of an advantageous ooA ~125,000 - 105,000 ya time period".

Thanks for that clarification Eurologist. It does show that the M/N split is more ancient (by perhaps 5000 years) than the simple split shown previously. And U being the product of a very basal split in the N/R sequence. In fact U may not really be part of R at all.

Rokus said...

'That is an incorrect representation of the results of the paper. There is now an L3MNUR group, which is upstream of L3M and NUR.'

Wrong figure. Lippold's investigation on the paternal and maternal demographic histories of humans results in the full mtDNA phylogeny as provided in Supplementary Figure 14. Here mtDNA L and mtDNA M emerge as a single subclade of mtDNA N. However, the approach is very global and without comment on the details. I figure this could serve as a guideline to identify mutations with more security as ancestral or derived, since obviously the current tree is just a guess.

I already illustrated the arbitrarity of the current phylogeny referring to MA-1's purported A16399G mutation, while 16399G is ancestral in primates, Denisova, Sima and the Insert of chr.11. This 'mutation' is recurrent in L3, L4, M and P and defines U5a1 - the latter situation would thus require an intermediate flip in U to sustain the current phylogeny.
Relevant to and in favor of the 'new phylogeny' is the A15301G! backmutation in N that is actually ancestral now we can compare rCRS with the genomes of Sima/Denisova, what means 'more basal' than M and L3. N, L3 and M are currently all defined by A769G, while this 'mutation' is actually ancestral as well - what may be important to redefine the phylogeny of other L-clades as well.

Ebizur said...

Rokus wrote,

"Wrong figure. Lippold's investigation on the paternal and maternal demographic histories of humans results in the full mtDNA phylogeny as provided in Supplementary Figure 14. Here mtDNA L and mtDNA M emerge as a single subclade of mtDNA N."

Actually, this tree has N and M'L3 as two coordinate branches of a stem that ultimately derives from the same trunk as L2. In other words, according to this phylogeny, former L3 would be better named as M1 (or L2m1) with former M being renamed M2 (or L2m2). This African-Asian linkage in the mtDNA phylogeny seems to be paralleled in the Y-DNA phylogeny by haplogroup DE-YAP.

However, please note that this study's Y-DNA phylogeny is also somewhat different from the currently "standard" tree of ISOGG; it does not recognize a CF node, but rather places C as a basal branch of DE (i.e. {C + {D + E}}) or as a basal branch of CDEF (i.e. {C + {DE + F}}).

Rokus said...

'Actually, this tree has N and M'L3 as two coordinate branches of a stem that ultimately derives from the same trunk as L2.'
First, figure 14 doesn't show two branches, it shows M and L3 as derived from N. This means, with your proposal there won't be any L2M1 or L2M2, only L2NM1 and L2NM2.
Second, of L2's defining mutations only C16311T may be presented as 'inherited by L3'. In other words, L2 was not really designed as ancestral to L3, very unlike the defining mutations of L3 that were (erroneously) designed as inherited by N and M. Indeed, only this single mutation separates a common L2'N clade from the 'more ancestral' African lineages L0 and L1. I figure that to make a really good mtDNA match with YDNA, Lippold et al. could as well have proposed an Eurasian L2-clade that backmigrated to Africa, analogous with the E clade of CFDE. Why did they scare away from this result, if C16311T was also found in the very divergent mtDNA of Lake Mungo 3?

terryt said...

"First, figure 14 doesn't show two branches, it shows M and L3 as derived from N".

I am unable to see the supplementary material for some reason and so I have to rely on your collective contributions.

"In other words, L2 was not really designed as ancestral to L3"

I don't think L2 has ever actually been considered 'as ancestral to L3'. Rather 'sister to L3'4'6'. Have either L4 or L6 been considered in the figure? Both have been found outside Africa, but not too far removed. Perhaps N was part of that exit but managed to move further than either L4 or L6.

"Actually, this tree has N and M'L3 as two coordinate branches of a stem that ultimately derives from the same trunk as L2".

Possibly L2 remains sister haplogroup to N and M'L3. And L2 is certainly African, and so we still have no reason to claim L3 as having originated outside that continent.

"There is now an L3MNUR group, which is upstream of L3M and NUR".

Does the paper change the phylogeny that went: N-R-U? In other words do we now have three new separate basal haplogroups within N before the other N and R haplogroups form?

Ebizur said...

Rokus wrote,

"First, figure 14 doesn't show two branches, it shows M and L3 as derived from N. This means, with your proposal there won't be any L2M1 or L2M2, only L2NM1 and L2NM2."

Are we both looking at the present study's Supplementary Figure 14?

I should not need to explicate how to read a phylogenetic tree.

I repeat: Supplementary Figure 14 has L2a'b'c as a sister clade to N'M'L3 (the most recent common ancestor of N, M, and L3), with the two sister clades, L2a'b'c and N'M'L3, sharing a most recent common ancestor approximately 105,000 years before present. L2a'b'c yields L2a and L2b'c approximately 81,000 years before present, whereas N'M'L3 yields N and M'L3 approximately 76,000 years before present. M'L3 yields M and L3 approximately 71,000 YBP, whereas N yields R and N(xR) approximately 66,500 YBP. M'L3 is not descended from N; M'L3 and N are sister clades, both descended from a common ancestor, and that common ancestor is a sister clade of L2a'b'c.

eurologist said...

However, please note that this study's Y-DNA phylogeny is also somewhat different from the currently "standard" tree of ISOGG; it does not recognize a CF node, but rather places C as a basal branch of DE (i.e. {C + {D + E}}) or as a basal branch of CDEF (i.e. {C + {DE + F}}).

I trust their new mtDNA tree better than their y-DNA trees - especially since those are dependent on the mutation rate. And yes, I know -I can't wrap my mind around that.

Rokus said...

'I should not need to explicate how to read a phylogenetic tree.'

'M'L3 is not descended from N; M'L3 and N are sister clades, both descended from a common ancestor, and that common ancestor is a sister clade of L2a'b'c.'

Reading a phylogenetic tree is not the same as interpreting the phylogenetic tree. Earlier you said 'this tree has N and M'L3 as two coordinate branches of a stem that ultimately derives from the same trunk as L2', what may be an elaborate way to express that current groupings carry the load of tens of thousands of years of divergence and hence can't be attributed any ancestral status per definition. It isn't my intention to deny you the pleasure of this precaution, though this is mere semantics that has all appearance to obfuscate some current misunderstandings on the origin. Especially since L3 'sensu stricto' was always explicitly understoood as the very trunk of mtDNA N, M and L+ (ie. Out of Africa). According to the defining mutations of L3, that were fully shared by N and M, the opposite applied here in the sense that mtDNA and mtDNA M were always understood as mtDNA L3 'sensu lato'. Now, with this study, we may agree this isn't the case, not even implicitly by renaming the trunk.
Instead, reading the tree 'as is' it doesn't elaborate on the renaming of trunks as you do, what indeed may foster similar suggestions the other way round (ie. Into Africa). However, the ancestor of mtDNA N is still N back in time when it ceased to be L2 'sensu stricto', but at least we may agree that mtDNA N is not mtDNA L2 'sensu lato': per definition it never was. This may conflict with the divergence of L3 and M from a point in time where only a few defining mutations of current N were in place, such as the A15301G! backmutation in N that is actually ancestral and even conflicts with L2 being an ancestral grouping. As I already say, Lippold et al. don't elaborate on these issues. In a way all branches are sisters and at the same distance to e.g. the Neanderthal outgroup. However, this is not the way how people think nor how any phylogenetic tree is represented.

terryt said...

"this is mere semantics that has all appearance to obfuscate some current misunderstandings on the origin".

Not so. I agree with Ebizur.

"Supplementary Figure 14 has L2a'b'c as a sister clade to N'M'L3 (the most recent common ancestor of N, M, and L3), with the two sister clades, L2a'b'c and N'M'L3, sharing a most recent common ancestor approximately 105,000 years before present. L2a'b'c yields L2a and L2b'c approximately 81,000 years before present"

So far that fits the 'old' tree':

http://www.phylotree.org/tree/subtree_L.htm

According to that tree basal L2 shares a number of mutations with basal L3'4'6 before the two separate, presumably because they became isolated from each other in some way. With L2 undergoing a further series of mutations until expanding as L2a and L2b'c, completely independent of any L3'4'6 haplogroup's expansion. Similarly basal L3'4'6 undergoes a couple of mutations before forming L6 and L3'4. The problem I have is will someone tell me where N fits into this diversity?

"whereas N'M'L3 yields N and M'L3 approximately 76,000 years before present".

That is where the 'new' tree is different from the 'old' tree. N forms a sister clade to L3'M sometime about the stage L4 and L6 branch off L3'M.

"figure 14 doesn't show two branches, it shows M and L3 as derived from N".

In fact the figure seems to show (as near as I can judge through comments here) that L3'M and N, like L2 and L3'4'6, are 'sisters'. One doesn't 'descend' from the other. To sum up:

"M'L3 is not descended from N; M'L3 and N are sister clades, both descended from a common ancestor, and that common ancestor is a sister clade of L2a'b'c".

terryt said...

I've managed to see the supplementary figures, and I see what German is on about. M is not simply a branch within L3. But, unfortunately for him, neither is L3 a branch within M. L3'M splits in two: an African branch and a Eurasian branch. Likewise we have a simple split between N and L3'M. That gives no idea as to direction of movement but the ancestral form is L2, a definitely African haplogroup.

Interestingly N still shows a simple east/west split, which is difficult to explain without extinction. And U is still part of R but an early split from it. The authors include just one 'South Asian' R haplogroup unfortunately, and so we can't determine if South Asian R entered from the east or the west. They have also ignored the numerous South Asian M haplogroups. Their tree is by no means complete although they do indicate an interesting sequence of M's divergence. We will have to wait and see if Phylotree adopts any of their suggestions.

German Dziebel said...

@Rokus

"I already illustrated the arbitrarity of the current phylogeny referring to MA-1's purported A16399G mutation, while 16399G is ancestral in primates, Denisova, Sima and the Insert of chr.11. This 'mutation' is recurrent in L3, L4, M and P and defines U5a1 - the latter situation would thus require an intermediate flip in U to sustain the current phylogeny. Relevant to and in favor of the 'new phylogeny' is the A15301G! backmutation in N that is actually ancestral now we can compare rCRS with the genomes of Sima/Denisova, what means 'more basal' than M and L3. N, L3 and M are currently all defined by A769G, while this 'mutation' is actually ancestral as well - what may be important to redefine the phylogeny of other L-clades as well."

Yes, that's the most important point. Although technically the new phylogeny didn't explicitly make NUR ancestral to L3'M, it's still fundamentally a problematic phylogeny because ancestral and derived states have not all been determined.

@Ebizur

"M'L3 is not descended from N; M'L3 and N are sister clades, both descended from a common ancestor, and that common ancestor is a sister clade of L2a'b'c."

There's a problem here. M and L3 are sister clades. But, if this is so, M'L3 cannot be a sister clade to N without N becoming a sister clade to M and L3. So N or NUR must be a common ancestor to M'L3 and in turn a sister clade to L2a'b'c.

terryt said...

"There's a problem here. M and L3 are sister clades. But, if this is so, M'L3 cannot be a sister clade to N without N becoming a sister clade to M and L3".

You seem to be confused about divergence here. I thought it was simple: M and L3 are sister clades to each other, and their ancestral clade, L3'M, is sister clade to N. Yes, N is sister clade to L3'M, but not to L3 and M individually. How come you see a problem?

"So N or NUR must be a common ancestor to M'L3 and in turn a sister clade to L2a'b'c".

Neither N nor NUR are ancestor to M'L3 (we may as well forget the UR part of NUR as U and R both descend from N). The ancestor of both N and L3'M is L3'M'N. Agreed that L3'M'N is sister clade to L2a'b'c. But L3, M, N, L2a, L2b, and L2c are not all sister clades with each other. We have a progressive divergence.

terryt said...

"There's a problem here. M and L3 are sister clades. But, if this is so, M'L3 cannot be a sister clade to N without N becoming a sister clade to M and L3. So N or NUR must be a common ancestor to M'L3 and in turn a sister clade to L2a'b'c"

I've just thought of a way to explain it in a way you can perhaps understand things. Yes, M and L3 are sisters. Their mother is M'L3. M'L3 is sister to N. Therefore M'L3 and N are cousins. Their grandmother is ML3NUR. And that means their great grandmother is L2Ml3NUR, or something like L2'3'4'6 in the 'old' nomenclature, although the position of L4 and L6 remains problematical at this stage.

Ebizur said...

terryt wrote,

"I've managed to see the supplementary figures, and I see what German is on about. M is not simply a branch within L3. But, unfortunately for him, neither is L3 a branch within M. L3'M splits in two: an African branch and a Eurasian branch. Likewise we have a simple split between N and L3'M. That gives no idea as to direction of movement but the ancestral form is L2, a definitely African haplogroup."

Yes, just as I have already stated...twice.

terryt wrote,

"Interestingly N still shows a simple east/west split, which is difficult to explain without extinction. And U is still part of R but an early split from it. The authors include just one 'South Asian' R haplogroup unfortunately, and so we can't determine if South Asian R entered from the east or the west."

Please be careful here. N does not show a simple east-west dichotomy; N(xR) does so (Eastern A'N9 vs. Western W'X'N1). Then again, X is sort of ambiguous, since it is found in both the extreme east (America) and the extreme west (Europe, Southwest Asia, North Africa).

As for the R half of haplogroup N, this study has placed U as the most basal subclade of R, sharing a most recent common ancestor with the rest of R at about 62,000 YBP, only about 5,000 years after the split of R from N(xR). The next haplogroups to split from the main body of R are R9'F and R1 at about 60,000 YBP. The next to split off is R2'JT at about 57,000 YBP. This is followed almost immediately by the derivation of R5. The rest of R trifurcates into P, B (including R11 as a sister clade of the B5 branch), and R0'HV at about 55,000 YBP. One of these clades (P) is Australasian, one (B) is East Asian (with a southerly tendency), and one (R0'HV) is Western Eurasian (but with an apparent origin in southwestern Asia between the Mediterranean and Pakistan).

In other words, unlike N(xR), haplogroup R in the present study's phylogeny does not exhibit a simple east-west dichotomy. The first branch, U, is Western. Of the second branches, R9'F is Eastern and R1 is North-Central. The next branch, R2'JT, is South-Central > Western. The next branch, R5, is South-Central. Thus, R0HV'B'P most likely has originated somewhere around the center of the southern part of Asia, i.e. in the region that is generally known as "South Asia" or somewhere between Iran and Bangladesh.

terryt wrote, "They have also ignored the numerous South Asian M haplogroups. Their tree is by no means complete although they do indicate an interesting sequence of M's divergence. We will have to wait and see if Phylotree adopts any of their suggestions."

This study actually has fairly good coverage of South Asian M haplogroups (better than their coverage of South Asian R, at least). They find that all the South Asian M haplogroups except M2 (i.e. M5, M3, M4, M18, M30, and M65) plus the Cambodian M24 form a monophyletic sister clade to East Asian M11, separating from M11 at about 44,000 YBP. (South Asian M2 is instead positioned as a sister clade of Northeast Asian D, sharing a MRCA with D at approximately 53,000 YBP.) All the South Asian M(xM2) clades (including Cambodian M24) coalesce to a MRCA at approximately 37,000 YBP. The nearest outgroup to the {M11 + M3'4'5'18'24'30'65} group is Cambodian M51.

In other words, most of South Asian haplogroup M appears to have invaded South Asia from East Asia via southern East Asia/Southeast Asia well after the diversification and dispersal of haplogroup R from South Asia. Haplogroup R (or at least the R5'R0HV'B'P subclade) probably represents the genetic legacy of the real aborigines of South Asia. Western Eurasian/African M1 and South Asian M2 seem to have a different history from the main body of South Asian M, and may reflect more closely the path of early members of haplogroup M after separating from their L3 relatives.

Ebizur said...

German Dziebel wrote,

"There's a problem here. M and L3 are sister clades. But, if this is so, M'L3 cannot be a sister clade to N without N becoming a sister clade to M and L3. So N or NUR must be a common ancestor to M'L3 and in turn a sister clade to L2a'b'c."

What you have written here makes no sense at all. According to the present paper's phylogeny, M'L3 is a single haplogroup, and this haplogroup is a sister clade to Haplogroup N. N'M'L3 is also described as a single haplogroup, which is the ancestor of both Haplogroup N and Haplogroup M'L3, and a sister clade to Haplogroup L2a'b'c. Why is that so hard for you and Rokus to comprehend?

I understand that you have some doubts about the rooting of the human mtDNA phylogeny. I am sympathetic in that regard, but please refrain from misrepresenting the phylogeny as it has been presented by any particular author(s). You are free to argue why you think any particular published phylogeny is inaccurate, but you are not free to misrepresent others' work.

Rokus said...

TerryT,

"First, figure 14 doesn't show two branches, it shows M and L3 as derived from N".

I am unable to see the supplementary material for some reason

In fact the figure seems to show (as near as I can judge through comments here)

Not 'in fact', in your fantasies. Please acquire a copy of the docs before commenting.

"this is mere semantics that has all appearance to obfuscate some current misunderstandings on the origin".

Not so. I agree with Ebizur.

You agree to obfuscate. Without a copy. So now what?

terryt said...

"Please acquire a copy of the docs before commenting".

I have and have now commented on what I see. I still agree with Ebizur and see he was correct in his interpretation.

"Please be careful here. N does not show a simple east-west dichotomy; N(xR) does so (Eastern A'N9 vs. Western W'X'N1)".

True. We have to explain that dichotomy. Unless we're going to claim R's presence through South Asia prevented any other basal N haplogroups establishing a presence there.

"Then again, X is sort of ambiguous, since it is found in both the extreme east (America) and the extreme west (Europe, Southwest Asia, North Africa)".

And may have survived in between the eastern and western branches. However it looks most likely to have spread from the west. A is also actually ambiguous as well, and even more likely to represent a survival somewhere between east and west.

"This study actually has fairly good coverage of South Asian M haplogroups (better than their coverage of South Asian R, at least)".

Their ommission of South Asian R is unfortunate.

"most of South Asian haplogroup M appears to have invaded South Asia from East Asia via southern East Asia/Southeast Asia well after the diversification and dispersal of haplogroup R from South Asia".

I have long accepted the M distribution reflects a major spread from the hill region between northeast India and southwest China. Maju has posted some information on Burmese haplogroups and notes two new M clades from there.

"Western Eurasian/African M1 and South Asian M2 seem to have a different history from the main body of South Asian M, and may reflect more closely the path of early members of haplogroup M after separating from their L3 relatives'.

Which suggests to me that M's original route east was via South Asia, but members of the haplogroup later re-entered from the east.

"Of the second branches, R9'F is Eastern and R1 is North-Central. The next branch, R2'JT, is South-Central > Western. The next branch, R5, is South-Central. Thus, R0HV'B'P most likely has originated somewhere around the center of the southern part of Asia, i.e. in the region that is generally known as 'South Asia' or somewhere between Iran and Bangladesh"

To me the South Asian Rs look to be either eastern or western in origin, not indigenous. R0 is definitely 'SW Asian' rather than 'South Asian', as is R2'JT and probably R1. R5 is the only unambiguously South Asia haplogroup represented in the study, and may have entered from the northwest as we can be sure members of the above R haplogroups as well as representatives of several N did. As may have R30 and R31.

"most of South Asian haplogroup M appears to have invaded South Asia from East Asia via southern East Asia/Southeast Asia well after the diversification and dispersal of haplogroup R from South Asia".

But we have the obviously indigenous M haplogroups presnt in South Asia before any likely eastward movement from East/SE Asia. Besides which the other three South Asian R haplogroups such as R7, R8, R6 look very likely to have expanded from along the eastern coast of South Asia rather than be ancient indigenous to Greater India. Other R haplogroups present in South Asia are definitely eastern in origin.

"As for the R half of haplogroup N, this study has placed U as the most basal subclade of R"

And that is extremely revealing. U looks to have formed at the southern margin of the steppe, which tends to indicate R too may have a split distribution. The pattern of R's distribution may suggest R's original route east was also through Central Asia rather than through South Asia.

Rokus said...

'According to the present paper's phylogeny, M'L3 is a single haplogroup, and this haplogroup is a sister clade to Haplogroup N. N'M'L3 is also described as a single haplogroup, which is the ancestor of both Haplogroup N and Haplogroup M'L3, and a sister clade to Haplogroup L2a'b'c. Why is that so hard for you and Rokus to comprehend?'

'[...] please refrain from misrepresenting the phylogeny as it has been presented by any particular author(s).'

Lippold's investigation on the paternal and maternal demographic histories of humans results in the full mtDNA phylogeny as provided in Supplementary Figure 14. Do you agree with this?

Well, no matter what you infer about sister clades, stems and trunks: here mtDNA L and mtDNA M emerge as members of a single subclade of mtDNA N. You don't have to agree with this, but this is what all people with eyes can see on the figure.

The problem seems to be that you (and others) have difficulties to take Lippold's new phylogeny as is. The authors don't comment on it, so why should you attempt to see the new phylogeny using older insights and shuffle existing branch orders - while actually current stem definitions have to be redefined entirely?

Sure, N has several stem mutations in the current phylogeny that are not shared by most M and L3+ subclades. Hence, you think that N can't be ancestral to L3 and M per definition. Your refusal to consider that current definitions should be changed accordingy and your shuffling of branches instead may be interpreted as misrepresenting the impact of the new phylogeny.

My proposals are not misrepresenting figure 14 in mtDNA N being ancestral to L3+ and M, only free of the need to please current dogmas - that explicitly are not part of the new phylogeny.

Literally, if you don't want to misrepresent the tree in figure 14, you should be prepared to consider the defining mutations of mtDNA N to be potentially all 'ancestral' and those of L3+ and M all derived. Are you ready for this?

N is defined by the following mutations: G8701A, C9540T, G10398A, C10873T, and A15301G!
. All the sequences before the N* point have C10873T and C9540T. Ancestral or derived? Since N is older than M and L3 this mutation must be ancestral downstream L2.
. A15301G! is most probably simply not a backmutation in N, but a mutation in U5a2b4 and the subclade for L3+ and M.
. N's defining G8701A mutation exists in L3e1c as well (found in South Africa and Oman) and currently has 'backmutations' in N21, R0a2e, U5b3e. If indeed not ancestral in N, this may also be interpreted thus that N21, R0a2e, U5b3e are actually more basal within N since they lack the derived mutation.
. N's defining G10398A mutation exists in L3e1a3 as well (found in Chad and Syria), backmutations in N8, S3, Y, N1a'e'I, P4, R11, R12'21, R0a2k1. If indeed not ancestral in N, this may also be interpreted thus that N8, S3, Y, N1a'e'I, P4, R11, R12'21, R0a2k1 are actually more basal within N since they lack the derived mutation.

Hence the point of divergence for L3+ and M may eg. be imagined being after C10873T and C9540T, halfway N and nearest to L3e1c/L3eia3. However you may object to this interpretation, this configuration simply follows from the figure. Does this possible outcome still look like a sister grouping to you?

Interesting is that, despite L3e's current west-African connotation, Majo already came to the conclusion that branches of the L3e1 lineage maybe coalesced much more to the east very long ago.

terryt said...

"the full mtDNA phylogeny as provided in Supplementary Figure 14. Do you agree with this?"

There are huge conflicts in the M phylogeny as compared to Phylotree. I have long believed the M tree is capable of greater definition however. It remains to be seen how it is all resolved. And the tree omits many important basal branches and so I think we cannot yet assume the new tree is 100% accurate.

"Hence the point of divergence for L3+ and M may eg. be imagined being after C10873T and C9540T, halfway N and nearest to L3e1c/L3eia3. However you may object to this interpretation, this configuration simply follows from the figure. Does this possible outcome still look like a sister grouping to you?".

You may have something but more peple need to do more work shuffling the haplogroups around.

German Dziebel said...

@Ebizur

"What you have written here makes no sense at all. According to the present paper's phylogeny, M'L3 is a single haplogroup, and this haplogroup is a sister clade to Haplogroup N. N'M'L3 is also described as a single haplogroup, which is the ancestor of both Haplogroup N and Haplogroup M'L3, and a sister clade to Haplogroup L2a'b'c. Why is that so hard for you and Rokus to comprehend?"

I think we are talking about the same thing. I didn't realize you were referring to M'L3 as the mother of M and L3. If M and L3 are sister lineages within a single haplogroup, it means they are descendants of the "mother" M'L3 superhaplogroup. It's the mother M'L3 superhaplogroup that's a sister to haplogroup N. It means N is hierarchically above M'L3. This is the most important thing. I'm glad we got there.

"You are free to argue why you think any particular published phylogeny is inaccurate, but you are not free to misrepresent others' work."

I didn't. I was referring to what you wrote, which I misunderstood.

Ebizur said...

German Dziebel wrote,

"It's the mother M'L3 superhaplogroup that's a sister to haplogroup N. It means N is hierarchically above M'L3. This is the most important thing. I'm glad we got there."

No, it does not mean that N is hierarchically above M'L3. They are coordinate clades; neither is superordinate nor subordinate to the other.

German Dziebel said...

@Ebizur

"No, it does not mean that N is hierarchically above M'L3. They are coordinate clades; neither is superordinate nor subordinate to the other."

A small correction: N is hierarchically above M and L3. There're no sequences that are M'L3*. The rest of my original statement stays the same.

Ebizur said...

German Dziebel wrote,

"A small correction: N is hierarchically above M and L3. There're no sequences that are M'L3*. The rest of my original statement stays the same."

I suppose you must also believe that Y-DNA haplogroup N is "hierarchically above" O3 or R1. It is meaningless to compare across different lines of descent in that manner; the phylogeny is not constructed in such a way that the letters or numbers (or the number of letters or numbers) in an alphanumeric appellation for a haplogroup has any regular, measurable chronological significance.

You may say, for example, that mtDNA haplogroup M is superordinate to mtDNA haplogroup D, haplogroup D is superordinate to haplogroup D5, or haplogroup U is subordinate to haplogroup R (or, one order further removed, haplogroup N). Saying that haplogroup D5b is subordinate to haplogroup R, or that haplogroup N is superordinate to haplogroup M or L3, is nonsensical and incorrect; you cannot make such a statement about two haplogroups that belong to separate lines of descent.

(By the way, in reality, the MRCA of all extant haplogroup N Y-chromosomes is much more recent than the MRCA of all O3 Y-chromosomes or the MRCA of all R1 Y-chromosomes.)

German Dziebel said...

@Ebizur

"I suppose you must also believe that Y-DNA haplogroup N is "hierarchically above" O3 or R1."

Let's not switch gears. Although elephants' trunks do look like snakes, I'm not willing to discuss snakes at this point.

"haplogroup N is superordinate to haplogroup M or L3, is nonsensical and incorrect; you cannot make such a statement about two haplogroups that belong to separate lines of descent."

Oh yes I can. Although in the current rendition of the tree M and L3 are not 'daughter' haplogroups to N, N is an "aunt" to them. And aunts and nieces belong to different generations. It's true that some aunts are younger than nieces but it's rare and the analogy is irrelevant for present purposes.

terryt said...

"Let's not switch gears".

Ebizur has raised a valid point here, and you cannot glibbly dismiss it without appearing to use different criteria for different haplogroups.

German Dziebel said...

@Ebizur

"Ebizur has raised a valid point here, and you cannot glibbly dismiss it without appearing to use different criteria for different haplogroups."

No he hasn't. If we are talking about a general principle and the terminology we use, it doesn't matter what haplogroups we're talking about. Plus he mistakenly assumed that I follow the alphabetical order of haplogroup names.

terryt said...

"Plus he mistakenly assumed that I follow the alphabetical order of haplogroup names".

I doubt that is the case, although I'm sure he has not a clue as to what reasoning you do actually use to follow the order of haplogroup names. Nor do I for that matter.