Re: Randomness-chance-accidents

Date: Wed Sep 27 2000 - 15:57:59 EDT

• Next message: bivalve: "Teaching evolution"

>When David (bivalve) said "Probably much more important is that natural selection makes evolution highly non-random", it appears that he was referring to a state of order rather than disorder.<

My main point there was that natural selection can be highly directional, so that a probabilistic model assuming equal probability of change in any direction is inaccurate.

For a relatively large population and a relatively small relative advantage of a mutation, the probability that a new advantageous mutation spreads throughout the population is approximately twice its relative advantage. A more precise formula for the probability of a new mutation replacing the old one in a population, assuming that the effect of the mutation in a heterozygote is half of the effect on a homozygote, is P=(1-exp(-4*Ne*s*q))/(1-exp(-4*Ne*s)), where exp is the natural antilog, Ne is the effective population size (a function of the number of breeding individuals), s is the relative advantage (or disadvantage) of the mutation (for neutral mutations, s=0), and q is the frequency of the mutation at the starting time. For a brand-new mutation, q=1/N, where N is the population size. (Original derivation was by Kimura, 1962; any population genetics text will have this.) Thus, a mutation that is highly advantageous is almost certain to be established in a population!
. For example, a mutation that gives a 50 percent improvement over the existing one will prevail over 63 percent of the time, if the entire population is reproductively active.

This provides an effective ratchet to establish new mutations in a population.

Dr. David Campbell
"Old Seashells"
Biology Department
Saint Mary's College of Maryland