Archiver > GENEALOGY-DNA > 2007-03 > 1172890321

From: "John McEwan" <>
Subject: Re: [DNA] Lactose tolerance evolved recently
Date: Sat, 3 Mar 2007 15:52:01 +1300
In-Reply-To: <BAY111-DAV157626EE0FA04E5CB18A0AB1870@phx.gbl>

Lawrence said
> While John C is correct each individual piece of evidence has
> only so much value, the two combined do suggest that the best
> explanation is that the variant was most likely at a low
> frequency prior to "milk drinking".

No, just the opposite. If the mutation occurred only 9300 years ago,
its frequency at 7500 years ago was still almost zero, it almost
had not spread beyond a single tribe. You are forgetting that absent a
strong natural selection, most mutations die away quickly, and those
hang around at all have no reason to spread beyond a tribe, much less
two continents.


I had better be explicit in what I meant by low frequency i.e. most
likely very much less than 0.01 or 0.02. I assume the allele would have
minimal selective value prior to that time. If it was lethal it would be
in mutation selection balance and its frequency would be about ~10-6, so
it was "probably" some where in that range.

Burger et al. 2007 using a deterministic model of population growth and
estimated selection pressure values found.
" By applying a simple deterministic model (see Materials and Methods)
we would expect the -13.910*T allele frequency 7,000 years ago to be
~0.089 (when S = 0.014)and ~0 (when S = 0.15)"

Without a lot more work we will not know the ancient frequency or
distribution with any more accuracy as John C has said.

Then you commented
> R1b1c7, R1b1c10, the Scots R1b
> cluster, R1b1c* and R1b1c9 all have distinctive geographical
> patterns, almost certainly these mutations and distribution
> predate extensive adoption of "milk drinking" in Europe on a
> large scale.

This is a totally unproven assumption. Indeed, just the opposite:
Calculations using Chandler's mutation rates give rather small age
for some of these subclades.

I assume you mean low or young age estimates for these groups, and yes
we get low estimates if we use observed father son, or John C, mutation
rates and use the ASD measure of diversity to measure TMRCA. However,
then we have issues about these estimates due to the effect of
population size fluctuations amongst other things. All we know is any
such estimated ages are lower bound (as in younger) values.

You may not know this, but microsatellites in linkage disequilbrium with
the lactase variant have also been used to age the mutation (Coelho et
al 2005) as well as the more commonly quoted estimate based on the
physical decay of the length of the haplotype (Bersaglieri et al. 2004).
This was done by looking at the accumulated variation in the variant
Lactase haplotype and using ASD and associated methods very much like
what we commonly do on Y, but allowing for a very low rate of
recombination as well. They got an estimates of 7400-12300 yrs bp, close
to that used by using haplotype length decay (which is independent of
mutation rate estimates and depends only on generation length) ****using
a STR mutation estimate of 0.001****.... the value of which is close to
the "3 fold lower" estimate of Zhivotovsky et al 2004. So in this case
it is clearly "what is sauce for the goose is sauce for the gander". You
cannot use one STR mutation rate estimate for the lactase variant age
calculation and then another to estimate Y haplogroup age and then say
the results differ.

If using Chandler's rates was appropriate then younger rates would also
be obtained for lactase mutation age which would conflict with an
independent method (though perhaps not significantly). However, if it
was true the mutation would have occurred *AFTER* dairying commenced in
Northern Europe, and selection pressures would have to have been much
higher, perhaps unbelievably high, and I would have some concerns that
the variant could have possibly increased in frequency and spread far
and wide in such a short period.

Perhaps the major finding to emerge from this paper is that studies on
ancient DNA are getting tantalizingly close to also testing for Y SNPs
and STRs. Hopefully, debates about ages of various haplogroups will then
become much more circumscribed.


John McEwan

This thread: