18 November 1991
From: G. A. Dover
Subject: Molecular Drive
On 6.11.91 Mark Yandell states that 'it is pretty obvious' how biased
gene conversion can lead to homogenization and fixation but not unbiased
conversion. He also admits to 'most confusion' on how 'fixation of a mutation
can act in the ABSENCE OF BOTH SELECTION AND GENETIC DRIFT!!', and pleads for
help on this from 'anybody out there'.
On 8.11.91 Joe Felsenstein attempts to come to his rescue by denying
molecular drive as a process altogether on the grounds that unbiased conversion
(he doesn't say what he thinks the consequences of biased conversation are)
could lead to one chromosome being A1-A1-A1-A1 and the other A2-A2-A2-A2 such
that 'homogenization within a chromosome happens but there is no homogenization
of the population'. He then argues that 'the rate of substitution at any one
site is EXACTLY the same as if there was no multigene family! This fixation is
due to drift, as usual'. He concludes that 'one understands the process better
if it were called concerted evolution'. Earlier on he argues that concerted
evolution is one of the 'central processes' of a 'series of processes' under
the name molecular drive, 'that occurs when unequal crossingover and gene
conversion spread the same mutations through a gene family'.
Marc, has this cleared up your problem? It would surprise me if it did.
Let me try (from the Horse's Mouth so-to-speak). I'll ignore all single and
double exclamation marks and upper case bold lettering in both messages by
sticking to a straight scientific description. First, let me say that the
difference between concerted evolution and molecular drive is one of observed
pattern versus inferred process. The former is a pattern of mutant
distribution in multigene families such that two genes sampled at random from
within a species have a higher identity coefficient than from two different
species. Note this does not imply blanket homogeneity within a species. This
pattern can be seen often whether genes are distributed on one pair chromosomes
in each individual or on several pairs. What is the process behind this
pattern? For large non-coding families it is unlikely to be selection (and for
other arguments see Trends in Genetics 1986 2 159-165; Genetics 1989 122
245-252). Ohta & Kimura invoked a process of 'double diffusion' i.e. first
diffusion of a variant gene through a single chromosomal array (as a
consequence of stochastic gain-and-loss due to say unequal crossingover or
unbiased conversion) i.e. leading to Felsenstein's A1-A1-A1 and A2-A2-A2;
followed by diffusion (genetic drift) of one homogenised array in the
population. This could well be the case for chromosomally restricted families
in small populations, but is not satisfactory when considering the simultaneous
homogenisation and fixation within tens of families all of different copy-no
and chromosomal distribution in each plant and animal species of widely
different sizes, breeding etc. To achieve independent family evolution (i.e.
not relying in the first instance on whole chromosomal drift) I suggested that
genomic mechanisms of gain-and-loss (whether biased or unbiased) need to
operate between homologous (or where necessary non-homologous) chromosomes, so
that in a sexual species, homogenisation and fixation occur gradually and
concomitantly. This is molecular drive. Without interchromosomal
non-reciprocal exchanges there is no molecular drive. There are no
intellectual Brownie-points to be gained in describing a scenario in which
molecular drive does not operate (Felsenstein's A1-A1-A1 evolving separately
from A2-A2-A2) in order to say that molecular drive does not operate.
Now there are many caveats re: relative rates of intra versus
interchromosome diffusion, size of family, population size etc. that influence
the extent to which the concerted evolution pattern is observed i.e. partially
or fully fixed species-diagnostic mutations. Nevertheless, this internally
produced process (based on known genomic mechanisms of turnover) is
operationally distinct from the other two evolutionary processes of selection
and drift, but is known to interact with them in a variety of circumstances.
(See Genetics loc. cit.).
(1) unbiased mechanisms (within and between chromosomes) can lead to
concerted evolution patterns, just like biased mechanisms. They take longer,
and in the early stages are more likely to throw out a rare variant. But if
you understand diffusion as a consequence of fluctuating populations of gametes
and individuals (i.e. sampling error) then it also applies at the level of the
(2) We can't call both process and outcome (pattern) by the same name i.e.
concerted evolution. Otherwise we end up like that old chestnut of adaptation
being used for both the process of natural selection and its product.
(3) The process of molecular drive relies on two sets of well documented
observations: interchromosomal non-reciprocal exchanges and the concerted
evolution pattern. It is not the same (sorry) as drift or selection as
traditionally understood at the level of populations of gametes or individuals.
Naturally, the extent to which selection acts depends on the variance in the
degree of homogenisation per individual, at any given generation, which in turn
depends on the relative rates an intra versus interchrom. exchanges. (See Ohta
& Dover 1984 Genetics 108 501-521 on this; then get back to me, if necessary).
(4) Finally, I think that it's acceptance is more a psychological rather
than scientific issue - but the electronic board is hardly the place for group
therapy. Just think how long it took (on largely irrational grounds) for
selection and neutral drift to be accepted. Funny thing the human intellect.
Deep-down I don't care what we call the process (concomitant homogenization and
fixation in each separately evolving family) that underlies concerted evolution
patterns, so long as it's recognised operationally for what it is, and we get
on with the important business of quantifying it's fascinating interactions
with selection/drift in real situations. I welcome you to this challenging and
objective enterprise, in the study of the natural world, Joe.
For further details see my forthcoming book 'A Molecular Drive through
Evolution', (for the under 5's to advanced players!).