"Tim Tyler"  wrote in message
news:c9fpqp$19pp$1{at}darwin.ediacara.org...
> I've written an essay about genetic drift and population size.
>
> This essay is intended to dispell some of the myths (recently
> expressed on other threads in this forum) suggesting that the
> effects of genetic drift are independent of the population size.
>
> Supposedly this is common knowledge - but in fact it's nonsense.
>
> The mix-up seems to arise from getting genetic drift muddled
> up with neutral evolution :-(
>
> ``Genetic drift and population size''
>
>  - http://alife.co.uk/essays/genetic_drift_and_population_size/
>
> This page shows the results of experimental modelling of the
> probability of fixation of near neutral alleles by genetic
> drift in populations of various sizes.
>
> It illustrates pretty clearly that the rate of fixation of
> alleles due to genetic drift is usually a function of the
> population size.
>
> The fixation rate of slightly-deleterious alleles by genetic drift is
> *dramatically* reduced in large populations - since there the effects
> of genetic drift on near-neutral alleles are more powerfully and
> effectively opposed by selective forces.
>
> Enjoy,
> -- 

Excellent!  The only thing missing from your essay is a
definition of "near neutral".  It is my understanding that
a mutation is "effectively neutral" if the selection
coefficient is less than the reciprocal of the population
size.  So, the borderline between neutrality and
non-neutrality for a population size of 100 would be a
fitness of .99.  For a population size of 50, the borderline
would be at a fitness of .98.

As can be seen from your graphs, right at the borderline
there is a balance between NS and drift.  Drift "wants"
the fixation percent to be 50%.  NS "wants" 0%.  From your
simulation results, it can be seen that an excellent
compromise is reached at around 37%.  I expect that this
37% figure will be true for all borderline cases, no
matter what population size you choose.

If you want to go from a sharp border between selection and
effective neutrality to a border region of "near neutrality",
I would suggest that you simply apply a factor of two on
either side of the nominal border.  That is, a fitness "w"
can be:

Neutral
  1 > w > (1 - 1/2N)

Near-neutral
  (1 - 1/2N) > w > (1 - 2/N)

Selective
  (1 - 2/N) > w

Of course, the fraction of all genes that fall into each of
these three regimes is an empirical question, not a
theoretical one.  Or is it?  I am currently playing around
with some ideas that try to predict the distribution of
w (or s) values that will be found in a typical genome.
It seems to me that ideas from the Haldane's Dilemma debate
can be combined with an assumption that a steady state is
reached between mutation and selection, and with the idea that
mutation is random across genes.  Taken together, these
ideas may somehow force the distribution to make most
non-deleterious gene mutations to be effectively neutral.
If I come up with anything, I'll let you know.
---
þ RIMEGate(tm)/RGXPost V1.14 at BBSWORLD * Info{at}bbsworld.com

---
 * RIMEGate(tm)V10.2áÿ* RelayNet(tm) NNTP Gateway * MoonDog BBS
 * RgateImp.MoonDog.BBS at 6/1/04 6:50:24 AM