GENEALOGY-DNA-L ArchivesArchiver > GENEALOGY-DNA > 2009-01 > 1232199340
From: Alan R <>
Subject: Re: [DNA] More L21 results
Date: Sat, 17 Jan 2009 13:35:40 +0000 (GMT)
Tim that's a very fair post. I take your point that you are measuring variance essentially and not age. So many factors can effect variance that reliably using variance to infer real age intraclade-style seems to me to be a near impossible. Personally, I feel that there are so many real life factors to do with the growth or extinction of lines that could effect MRCA in each area and make the intraclade MRCA potentially so different from the real root of a line that I think this makes them of little real value in dating a mutation. The variability of the same real life phenomena from place to place means they surely are of little use in identifying which area was earliest or route of spread.
I was thinking, does an intraclade MRCA date in each area really even tell us a date when L21 was present in each locality? If a group of migrants arrived, they could already all have a common MRCA who lived somewhere else, possibly a lot earlier making the the local MRCA date for an area a lot older than the actual local arrival. Conversely, after arriving, many lines could die out making the local intraclade MRCA date much younger than the local arrival of the lineage. At the same time, back near the real source of the mutation you could have had death of lines there which make the MRCA date far younger at the source point than its real date of origin in that area. To be honest I have to put my hands up and say I have only thought really hard about this because the result seemed so weird compared to archaeological evidence but now I have thought about it it I am convinced that intraclades can tell us very little.
I totally accept that interclade is not an option when looking at L21 but I think essentially we have to accept that leaves us in a position where we have no tools to date real arrival of it in each area. So, this may be a problem that simply cant be solved given that intraclades are all we can calculate.
I think the closest an intraclade can come to providing a useful date of any sort is by pooling every L21 from every area (as wide as possible) you can get your hands on together in a hope that enough of the variations of L21 are captured and calculate an overall intraclade date for the clade. It does seems when a maximum sample is pooled from everywhere that some of the pooled inraclades are not too far off the interclade dates. So we need to thank Tim for demostrating that L21 was in existence by the mid Bronze Age somewhere. That in itself when combined with the interclade date for it and S116 sandwiches its origin between two dates (although its a matter of opinoin if the formulas used need tweaked and the dates will move). However, on a localised basis too many demographic phenomena will make inferring dates of arrival impossible.
Dear Alan, David, and others,
Please keep in mind that the TMRCA estimates I have given are
estimates only at this point. They are probably the best that can be
generated using the intraclade method at this point. As I (and others) have
mentioned in the past, I suspect that the TMRCA estimates this intraclade
program generates are lower than the actual ages, possibly as much as 25-75%
lower than the actual ages. The ages are not the crucial thing here. What
is important is the fact that the variances we are seeing in continental
Europe for the L21+ samples tested to date are quite a bit lower than the
variances we are seeing for Scotland and Ireland. Let's not get too hung up
on the ages at this point in time. Alan, I know that as an archeologist you
would like to draw conclusions from date estimates. I suggest that it is
too early to do that until we are all more certain about accuracy of the
date estimates. Maybe the ages will turn out to be correct and maybe they
won't. Even if we can improve our estimates, they will still be simply
estimates unless we can extract Y DNA from ancient bones from this region
that can be carbon dated. It is still possible that L21 originated in
western Europe and that the oldest lines in western Europe all daughtered
out. It is also possible that the sample size is still too low. I suspect
that current trends will continue, however, as more data emerges. I suspect
that the L21 data can't be solely explained by descendants of a few L21
nobles being spread over a wide area on the continent unless we say that
those L21 nobles were Marco Polo types who immigrated back to continental
Europe from the British Isles 300-500 years after the L21 SNP originated in
the British Isles.
I personally believe that the sample size for Scotland and Ireland
is adequate at this point in time. As I have pointed out in previous
messages on this list last year, the TMRCA estimates go DOWN as the sample
size increases in many cases. This happened with the latest L21 data for
Scotland and Ireland as well. See my posting below from a month ago. Note
that the TMRCA estimate for the L21 haplotypes from Ireland and Scotland
combined decreased from 4236 years using 15 67-marker haplotypes using Ken's
method to 3575 years using 40 67-marker haplotypes using Ken's method. More
samples from Ireland and Scotland will likely continue to skew the TMRCA
I agree with Ken that interclade estimates are better for dating
nodes than using the intraclade method. I have always believed that to be
the case and have posted interclade estimates for nodes previously on the
list. However, the interclade method is not appropriate in a situation like
L21 where we are simply trying to determine the probable area of origin of
the SNP. You must use the intraclade method for this situation.