In <3raglt$etm at service1.uky.edu> "Christopher L. Schardl"
<clscha00 at ukcc.uky.edu> writes:
>Your comments seem entirely reasonable. Would you not expect that
>the degree to which this is a problem would depend on the degree of
>genetic redundancy; that is, in groups of organisms that typically
>have multiple genes for necessary functions, particularly 'bad'
>mutation barriers may be relatively easy to cross, whereas those
>eukaryotes (of which there are some) and bacteria (most) that tend
>to have single copies of most genes would very rarely be expected to
>cross such barriers.
Individual bacterial species have multiple copies of some genes.
Sometimes this comes up when people try to find mutants that lose
function and can't do so -- they have to knock out 2 different genes.
They can sometimes put copies of a gene onto a plasmid, and then one
copy can mutate. Sometimes when a gene product is in short supply
then "amplification" is selected -- a plasmid carrying the gene
replicates many times and gives many gene copies.
This leads to a potential problem for evolution: When there are
multiple copies of a gene, a mutation for improved function will have
a diluted effect. Say the improved version would result in 10%
faster growth. But there are 10 copies and 9 of them make the old
version. And recombination can destroy the new version (even while
it makes further variations). So a copy-number mutation or a control
mutation can have a large effect -- twice as much product say, while
a structural mutation that produces a 10% better product may have less
Another thing that can make a difference: Often when mutations occur
it's because DNA repair mechanisms have broken down locally at some
time and place. This is likely to produce multiple mutations close
to each other. So if the problem is to produce multiple mutations in
a single gene, where any one of them is selected against, that is a
partial solution. They'll happen more often than you'd expect by
multiplying the individual probabilities.
Finally, this may simply be an unsolved problem. Maybe there are large
numbers of favorable mutations lying around, utterly unavailable because
they must be reached through multiple unfavorable intermediates. If one
unfavorable intermediate happens at a frequency of 10^-8, and you can
completely avoid selection, then it will on average be present in one
copy later. (It might be washed out immediately or might drift to high
numbers and only later be washed out.) So on average the second
mutation (also at frequency 10^-8) should occur within 10^8 generations
or so. Getting a particular pair of mutations when neither of them is
favorably selected, looks hard. It looks like it takes large
populations and long times. Maybe almost all of adaptive evolution
follows other pathways.