Dating with DNA? Problems and Current Approaches

Apart from that, I'd be grateful if you could post a comprehensive
summary of how genetic dating is done. What are the tolerances? Against
what other measurement is it calibrated ?
To use an example : If you had to date Oetzi without any hint from other
sources, how would you place him and date him ?

I have actually done this. Dating something like Oetzi is particularly
problematic and I do have to bring in an eastern example. Ad*** et al did
a sequence of LM3 and claimed it was as divergent from humans as
Neandertals. I looked at the sequence and there appears to be examples
of cytosine deamination. The core eastern/east african mtDNA sequence is
present in LM3 sequence but many other substitutions are present.
In general mtDNA, particularly HVR1 cannot be used to date migrations
unless a very large sampling is done.

The problem starts with the following assumptions
1. There was a C/H LCA
2. We can define when that C/H LCA lived.
3. We can use the number of substitutions per length of DNA to define rate.

Its not simple, transitions are muddled up by backmutations and
transversions are of no help in intraspecific investigations.

Paabo's results translated to deal with lHVR1 which I use requires the
half-time between new mutations to be 9000 years. I state that the range
is between 9000 and 18000 years based on uncertainty in the human
evolution (uncertainty in LCA, apparent slowing of poly-nt evolution)
Lets apply this to Oetzi, what can we learn, nothing about dating
it. If Oetzis mtDNA went extinct and you could show that other similar
mtDNA went extinct you might have something.

Let us say our rate is 0.5/12,000 years or so. That represents a
situation whereby on that sequence of 260 nt we expect one mutation
per 24,000 years. This is not the way evolution works, and the continuity
can be exploited.

For example.
24,000 years/24 years per generation = 1000 generations so that
0.001 mutation is expected per generation. So let us say you had
1,000,000 people all with the same original mtDNA you could
assess the age by looking at the number of mutations.

In europe this did happen¹, apparently there was a constriction
that lead to the almost fixation of the CRS over large areas, certainly
we could find a length of DNA where everybody over a large region had
CRS version of the DNA segement. Such that as the population expanded we
can begin to assess the mutations. This has been done for western europe.
I shows a population similarity that runs from iberia to NW europe.

The issue is timing. Paabos timing is based on a late exit time, this
has been recently discussed by Paul Mellars in Science Magazine. Paul
glosses over the Skhul/Qahfez problem and completely ignores the
Liujiang controversies. Even if he is correct, the level of
HLA A*02 relative to surrounding sites in southern Iran is astounding
and suggest a regional isolation prior to spread, what is particularly
interesting in the HLA A*02 variants in arabia are different from those
in S. Iran, different from those in east africa, and no other people, in
india or europe show any where near the level of A*02 diversity. So
relatively speaking either the Arabian/Iranian populations were massive
(where is the evidence?) and central asia population small (probably) and
somehow only A2402 reached India during the early settlement period.
There is some A*02 diversity in India but is it small relative to S. Iran
and Arabia. So my considered judgement on this is that LM3 and other asian
finds except Liu Jiang maybe late but that does not make entry into eurasia
neccesarily late, there may have been a long period of acclimatization
before people spread. The climate may have been most acceptable about 124
to 118 kya as sea levels were low, but rose and increasing rainfall in
desert areas would have facilitated the migration into persia and
expansions.
The complicating issue however is mtDNA and the way it behaves in human
population for molecular anthropology. mtDNA apparently has a large
settlement bias, the first to settle tend to be overrepresented in mtDNA.
Sort of like a game of hot potatoe. The issue is this, did M and N
lineages branch in africa, and if so where, or did they branch in eurasia.
Since the basal branch point were used to determine the exit times.
I want to say with confidence that 52 kya spread times are way to late, we
can practically eliminate any exit time before 65 kya, and the Toba
event may have allowed the last appropriate opportunity to leave africa.
If so we can use the 70kya and 120 kya to recalibrate the mutation rate.
The refines the accuracy a little.
Then there is a final problem, sequencing and sequence errors.
For instance in the region that is typically sequenced there is the
7 nt stretch of Cs This region appears to be both a sequencing
nightmare and anomolously evolving. Humans appear based on parsimony
to have the original sized peice, but the replacement of one or
two alternative Ts with Cs results in either expansions or contractions
and more importantly increased numbers of mutations on either side
of the sequence, probably these are sequencing errors. None the less
if you are detecting sequencing errors as nascent mutations that also
is going to affect your rate.

So the bottom line is that it is difficult to get an exact dating.

The other problem is with Y. mtDNA shows by my calculations a TMRCA
from africa of about 170 to 320 ky. Many have agreed that a date
about 200 to 250 is plausible. The Y chromosome shows a TMRCA between
25 and 140kya with many suggesting a central time about 40 to 60 kya.
By mtDNA and archaeology humans are in eurasia by the time Y
begins to expand from africa or pakistan (depending on the theory
of the week).
In almost every study were males of one origin mixed with females
of another the effective mixing size was 1:2 males to females, and
in cases this is backed up by HLA. (Maori, Jomon/Japan [hah-hah])
Spencer Wells promotes the idea that humans left africa but Y spread from
central asia, I don't know what his timing is but seems to me
like he favors the SW african theory so 25 kya entering eurasia and
spreading from central asia 20 kya or so?
That is quite a difference from the archaeological expectation,
don't you think? In theory we should see genetic evidence before
archaeological since humans tend to have a greater preference for breeding
relative to preserving materials for archaeologist to uncover.

What is the advantage of the HLA. It is a massive and highly variable
data set, this means lots of probes. Lots of probes means different
handles on different problems, potential of people specific probes.
What is the problem with HLA, recombination in absense of past
hap frequencies means estimating rates toward equilibration is hard.
In new world for instance, or any highly tribalized people, biasing
of frequencies is rapid, and recombinations effects are sporadic.
In my recent post on HLA I cut out the section on comparison
with mesoamericans because of the complaints, but to make a short
study every tribe had its own nodal alleles and very dissimilar gradations.
The problem is that almost all the remant tribes of mexico, now, were
not part of the core agrarian complexes as these were sites the
spanished focused on conquest and the natives did better in the
woodlands and highlands, as a result we bias our
analysis by targeting tribes by autonomous status or language.

Dating HLA or any technique will eventually need to be verified
with lots of ancient DNA sequencing, mtDNA has begun, very few
Y has been sequenced. Massive amounts of sequencing over as
many sites as possible will be required in the long run.

¹Am J. Hum Genet. 75 (2004):693-702. McEvoy et al.

.

Loading