Archiver > GENEALOGY-DNA > 2005-03 > 1111823492

Subject: (1/3rd) f/s rate - was Re: [DNA] Re: Cryptic Backup Copy ... "Hothead Gene"
Date: Sat, 26 Mar 2005 02:51:32 EST

At this point the simulations appear to indicate "sometimes", and these
occurences *always* have appeared in the 20-40 generations immediately following
a generation which had a dramatic "spike" in ASD due to a dramatic "spike" in
mutation count. ASD can decline for many subsequent generations, indicating
slow, no, or negative growth. (I think I described this in some detail 2 msgs
ago :-). In my simulations it takes 20-40 generations to "recover" to a
steady increase in ASD and corresponding faster mutation rates, eventually
approaching some large fraction of the input f/s rate.

I agree there's no reason to generally adopt 1/3rd the f/s rate as a
"standard" for population mutation rates. Viewed from the perspective of our
simulations, if someone is examing a population's growth in the 1st ~150
generations (say, 4500 years), then there seems to always be a period of 20-40
generations with calculated effective rates 1/3rd or less than the f/s rate.

In this context, the "sometimes" is the 20-40 generations immediately
following a spike in ASD & mutations randomly selected in the simulation - (an
analogous real-world situation could possibly triggered by any number of factors
not characterized by simulation).

So if LV & PU or others happen to to be looking a period of 600-1200 years
immediately following an "explosive" beginning of some
population/tribe/whatever, I can understand, from the perspective of our simulations, how they might
independently come to an estimate of 1/3rd the f/s rate. However, for a
longer period, I currently expect this rate to steadily approach something much
closer to the f/s rate, the rate(s) we see so often.

Can't resist making this point: we genealogists are frequently measuring
rates & TMRCA along single trajectories from survivors into the past. We're not
counting the extinct lines, and generally, most lines go extinct. The
population researchers have a broader scope and are not necessarily measuring
exactly the same phenomena. Methods are understandably different - normally, we
count back only a few generations, max usually ~10, so we need a "wide" tree
for accuracy (lot's of cousins), whereas the population folks have hundreds of
generations & sometime thousands of individual DNA results. (Try assuming
some back or parallel mutations in Charles's YSTR Blog data & watch what
happens to the resultant calculated rates - they'll "zoom" up sometimes 2x or 3x -
John Chandler's "tyranny of small numbers" - quantum leaps :-)

In any case, I currently think the 1/3rd ratio is much too broadly applied.
It's close to true, sometimes, after some identifiable circumstances, and for
a relatively short period of time in a haplogroup/subpop context, but
significant period of family genealogy time.

Mike ...

In a message dated 3/25/2005 4:00:50 PM Pacific Standard Time,

There seems nothing on the horizon that I can see which will allow us to
"always" or "most of the time" divide father/son rates by 1/3 to get a
practical effective rate. Maybe "sometimes" the ratio factor 1/3 is
applies, but if you don't know when that "sometimes" applies then there is
no usefulness to the fudge factor.


Admin: FTDNA Humphrey project
Co-Admin: FTDNA Liles project
multi-family project:Humphrey, Liles, Morton

This thread: