Re: Haldane's dilemma

Adrian Teo (
Wed, 30 Apr 1997 22:13:07 -0700

David Campbell wrote:
> By the way, is anyone else interested?
> Adrian wrote:
> >> [most previous material sniped]
> >
> >As far as I know, error catastrophe occurs when the mutation rate gets
> >above one harmful mutation per progeny - .5 harmful mutations per gamete
> >per generation. Since Kimura (1983) estimated that amino-acid altering
> >mutations are ten times more likely to be harmful than neutral, then the
> >estimated rate of neutral mutation would be .05 per generation. That
> >figure seems to be too small to explain every much. In ten millions
> >years, using the example above, you get only about 25000 neutral
> >substitutions.
> >
> Much of the genome does not code for anything actually used by the cell
> and, so far as we know, could vary freely (spacers, exons, pseudogenes,
> etc.). Also, the estimate of harmful to neutral mutations of Kimura sounds
> high relative to the results of mutational studies. A recent (this year)
> Nature (sorry for the vague reference, but it's as much as I remember)
> reviewed a paper that found most mutations had little effect on efficiency
> of the studied proteins.

I am sorry, but I don't get your point. If most mutations have little
effects, then aren't they neutral?

> >> >> In particular, a minor change affecting courtship behavior could
> >> >> be enough to isolate two subpopulations. Also, one mutation may easily
> >> >> involve more than one nucleotide. Gene duplications (e.g., unequal
> >> >> crossing over) and retroviruses can produce large changes in a single
> >>step.
> >> >
> >> >Aren't these events also exceedingly rare? Are they enough to solve the
> >> >problem?
> >>
> >> These events are generally rare (although many plants can easily form
> >> polyploid hybrids that act as genome duplication events, this is unusual in
> >> animals), but the time spans involved are also quite long.
> >
> >Given the above example of ten million years, would it be long enough?
> >For starters, how rare are these events anyway?
> >
> Another mailing list cited 2 million as the time since the most recent
> incorporation of a recognized retrovirus into the human genome, but that
> other mammals have much more recent ones. If this is representative, ten
> million is enough.

If it is representative....

So, it seems to me like the evidence suggest (albeit questionable) that
the time span is long enough, but the mathematics, as calculate by
Haldane says it isn't. Problem is, I don't see anything wrong with
Haldane's argument. Something has to go, and Walter Remine suggested
that perhaps the standard evolutionary model is IT.

> In animals, genome duplications seem to have been quite
> rare, but gene duplication is not all that unusual. Several common genetic
> diseases (e.g. Down's syndrome) represent duplication of sizable portions
> to entire chromosomes; duplication on the scale of genes would rarely be
> noticable unless it were looked for.

I still don't see how this addresses the problem of getting around the
limit imposed by cost of substitution. Perhaps you can elaborate.

Adrian Teo
Institute of Child Development
University of Minnesota