Archiver > GENEALOGY-DNA > 2009-07 > 1247404819

From: Alan R <>
Subject: Re: [DNA] R1b Origins (was OurEuropeangeographicalblock. . .)
Date: Sun, 12 Jul 2009 13:20:19 +0000 (GMT)
References: <><><>
In-Reply-To: <>

That is a good summary of the distinction between TMRCA and a foundation event.  Basically daughtering out and all sorts of chance will normally mean TMRCA is usually younger (sometimes hugely so) than the foundation event.  I suppose the only thing that could prevent there being a large distinction between TMRCA and the foundation event is if there had been huge pioneering early growth from a very tight knit group at the start (i.e. massive proliferation of a small set of lineages) and reasonably healthy evenly spread growth without any major blips since.  Even then, some variance would inevitably be lost I would guess.  Unfortunately ,I think the idea of an even somewhat egalitarian growth after an initial unlimited pioneer growth is not what happened.  Many societies later developed conical lineage structures where the top proliferated and push downwards to squeeze out the bottom.  The Gaelic clan system is a classic example.  That
would surely lead to huge loss of local intraclade variance.  However, does that not just affect intraclade dates?  Surely the interclade variance dates get us around this issue. 

I feel that intraclade TMRCA dates are OK as long as people see their limitations i.e. that the most recent shared ancestor may be much more recent than the founding event.  They still provide a date which is the latest point in time by which an area would have received that lineage and that at least can be compared to the archaeological record and any movements later than the local intraclade TRMCA can be ruled out.  That does at least eliminate some possibilities and therefore is at least worth something (archaeologists call this the terminus anti quem meaning 'the date before which (no idea how long before which) something must have happened.  I suppose the terminus post quem 'date after which something happened' is provided by the interclade TMRCA between a clade and a sibling clade.  In theory this could be hugely before the founding event but in practise with R1b in Europe it seems that the clade defining SNPs were coming thick and
fast so this is not the problem it could have been.  The real foundation date for the appearance of a clade in a given area is sandwiched between the local intraclade date and the interclade date between it and a sibling clade.  . T

That is really the state of play.  Given the limited use of local intraclade dates, it is impossible without further subdivision into subclades to take a clade (say L21) and look at it in one area then another and work out where it existed first and which direction it moved in.  Some people try to infer this from comparing local intraclade dates i.e. looking at the intraclade date for the same clade in two different areas and looking at which is older.  This is worthless other than giving the minimum period of time the clade appeared in each area.  It cannot give the maximum age of a clade's arrival in each area therefore it cannot be used to infer direction or timing of spread.  Perhaps it could 'if all other things were even' but that not the real world.  

The best we can do now to infer direction and timing of spread of a clades:

1. Look at the modern distribution (obviously has its problems).  The core distribition is what matters.  A very thin spread everywhere is inevitable with the passing of 1000s of years and tells us little. 
2. Look at the modern distribution of close sibling clades (as above).
3. Look at modern distribution of the parent (upstream/immediately ancestral) clade. This can be repeated upstream several times to infer the deeper history of movement.
4. Look at the modern distribution of downstream clades.
5. Look at the interclade MRCA dates between a clade and a sibling clades for the terminus post-quem
6. :Look at the intraclade MRCA dates for the clade in each locality for local terminus anti-quems (bearing in mind the major limitations of inference from this). 

That does not provide absolute conclusions but its the best we have and if all the information is looked at logically it can tell us quite a lot. Problem is I feel that its all too common in this hobby to buy into a model or have a preferred outcome.  If you consider the list above and use those resources then often the most likely scenario is clear.  However, I think many have preconceived ideas of what the history of a clade is and a preferred outcome. 


Vince V said:

Using short haplotypes does offer less resolution than longer ones, 
but that's not the primary problem with Zhiv et al. in my mind.

Specifically, Zhiv et al. were not trying to do the thing we are 
trying to do.  That is, they were not trying to estimate TMRCA.  They 
were trying to estimate some other thing, which they don't really give 
a name but which you or I might call a founding event or migration 
event.  They were attempting to use variance and extrapolate it it to 
cover something other than that which it actually represents.

Imagine at some point in the past an island that is uninhabited and 
always has been.  Now imagine some small group of people arriving on 
the island (say 40 men and women).  For whatever reason - size or 
resources or chance - the population of the island remains small for a 
long period of time.  If you visit this island 200 generations after 
the arrival of these people you'd most likely observe two things.

1.  An archaelogical record that would date the arrival to about 200 
generations ago.

2.  A genetic record that shows the TMRCA for the current inhabitants 
to be something less (probably a lot less) than 200 generations ago. 
Not because anything was wrong with the genetic record, but because 
the TMRCA really is less than 200 generations.

Zhiv et al. were looking for a way to reconcile those two data 
points.  It is hard to imagine why they chose the procedure they 
chose, but I can see the logic behind their decision. What I cannot 
see is the logic behind taking an inelegant solution to one problem 
and trying to use it to solve a completely different problem.

Zhiv rates should NEVER be used to estimate TMRCA (that's not what 
they were originally derived to do) and they  should NEVER be used at 
all except in the cases of very small, very isolated populations 
(which is the limited case in which they were derived).  Using them to 
estimate TMRCA for R1b1b2 in Europe violates both prohibitions.


On Jul 12, 2009, at 12:37 AM, Al Aburto wrote:

> I think the fundamental problem from the very beginning is the
> geneticists assuming that a 5 or 6 or 10 Y-STR marker set gives good 
> and
> accurate resolution in time.

To unsubscribe from the list, please send an email to with the word 'unsubscribe' without the quotes in the subject and the body of the message

This thread: