Definition of evolution

Stewart Schultz schultz at
Mon Oct 28 15:12:55 EST 1991

In article <1991Oct25.234811.1758 at> arlin at writes:

>In a previous posting I suggested that two major probems with the
>"allele-replacements" definition of evolution are that it does not
>encompass speciation, nor does it circumscribe genomic changes such as
>(for example) "molecular drive."
>First of all, with regard to *speciation*: one population is not the
>same as two populations, and one population cannot split into two
>populations by replacing an allele at a locus with another allele at
>the same locus.  Isolating mechanisms (genetic, behavioral,
>geographic) are necessary for a population to be split into two.
>Geographic isolation and other isolating mechanisms (see discussions
>elsewhere in this newsgroup) are not encompassed in the process of
>allele replacement, and thus a definition of evolution based solely on
>allele replacements would not seem to address the phenomenon of
>speciation, which is of paramount importance in evolution.

Models of sympatric speciation, for example, provide a clear 
demonstration of speciation through simple allele replacement.  Of 
course the conditions necessary for sympatric speciation (or 
reinforcement) are stringent, but the models have done more than 
rhetorical argument in elucidating them.  Far from being some sort of 
paradigm shift, the splitting of a population is easily addressable by 
population genetic theory.  Alleles that would tend to split a 
population include those predisposing individuals to colonize new 
environments, or those promoting assortative mating.

Though diploid sympatric speciation is undoubtedly rare, and external 
(ecological, geographic) forces are usually necessary to start the 
process of speciation, the subsequent dynamic can be driven entirely by 
stepwise allele replacements.  When the two lineages resume contact, 
such replacements can have accumulated sufficiently to cause genetic 
isolation, both prezygotic (behavioral, pheromonal, anatomical) or 
postzygotic (aberrant meiotic pairing, hybrid inviability).  Reciprocal 
silencing during the course of Muller's ratchet may be a widespread 
mechanism whereby hybrid inviability evolves in polyploid plants (Werth 
and Windham, Am. Nat. 1991); this would be hallmark allopatric 
speciation via random accumulation of point mutations.

An allele replacement may not "cause" the initial allopatry, but 
accumulations of allele replacements can cause virtually all other 
speciation phenomena (this is not to deny the importance of 
chromosomal rearrangements in general, although it should be
understood that any type of rearrangement can be the phenotype
of a point mutation).

>Secondly, with regard to *molecular drive and other results of
>non-allelic genomic change*: I hope readers will agree that the
>propagation of transposable elements in genomes and the general
>changes in genome size, form, and chromosome number (more things could
>be listed here) all constitute evolutionary phenomena.  Very simply
>put, the propagation of a transposable element (for instance) cannot
>be modelled as an allele replacement.  Allele replacements involve a
>single pre-defined locus while transpositions (for instance) do not.
>Therefore, a definition of evolution based solely on allele
>replacements does not encompass a particular large (and possibly very
>important) class of evolutionary phenomena.

Occurring within a single organism, all these chromosomal events fall 
under the general heading of "mutation."  I'd claim that the subsequent 
fate of these novelties in populations over time is the process covered 
by "evolution," and that this process can always be modeled by equations 
of population genetics regardless of the form of the "mutation."  The 
modeling may not be analytically easy or have a general formal solution, 
but numerical simulations can always be done. 

Mutation through the action of a transposable element is, physically, an 
allele replacement in precisely the same way any other point mutation is 
an allele replacement.  We can model the evolutionary trajectory of the 
resulting new allele just as for any other allele.  With a few minor 
modifications of the same population genetics equations, we can also 
model the spread of the transposon itself through a population, whether 
or not we call it an "allele."  What we need is an estimate of relative 
survivorship and reproductive output of individuals carrying the 
transposon (perhaps as a function of the number of copies in the 
genome), and the expected frequency of carriers in their offspring.  The 
difficulty here is not in the form of the theory, but our ignorance of 
the fitness parameters.

Of course, some transposable elements (e.g. Ds in maize) cause 
chromosome breakage, and it would be a stretch to think of this as the 
creation of a new allele.  Nevertheless, we can still model the 
trajectory of breaks, inversions, translocations, and other chromosomal 
mutations in much the same way as for simple alleles; again, the only 
requirements are estimates of fitness and frequencies in offspring (and 
perhaps mating probabilities as well).

As for molecular drive, it's easily incorporated into elementary 
population genetics equations, as an example of a mutation rate high 
enough in some cases to outbalance selection.

We don't have good theories that predict the behavior and effects of 
transposable elements within any particular genome, but that's due to 
our lack of understanding of the mechanics, not to shortcomings in the 
concept of gene substitution.  It's not a criticism of population 
genetics that it fails to dish up a predictive algorithm a priori for 
the internal movements of jumping genes, or for the generation of 
genetic variability of any given kind.  The theory isn't intended to 
explain how variation arises, but rather its subsequent demographic fate 
in response to a wide range of conflicting forces.  Mutation/selection 
theory applies to any genetic novelty, regardless of how it arose (via 
transposon footprint, molecular drive, or any other process), even if 
it's a new allele in a new locus.

>In response to the problem of speciation, Larry suggested that he
>was attempting to build a "minimal" definition of evolution, and that
>it did not have to encompass the entire biological universe.  I agree
>with this sentiment.  However, my response to its application here is
>that speciation is of sufficient importance to be addressed in any
>definition of evolution.  Many paleontologists would suggest that most
>major evolutionary transformations are temporally associated with
>speciation, and that anagenetic changes, taking place in a stable
>population, play only a minor role.

This is by no means a consensus.  But even if the question were 
resolved, it would still fail to address the genetic basis of any 
evolutionary change.  The thesis that character evolution is 
concentrated in speciation events is entirely independent of the thesis 
that most character evolution is gradual or due to stepwise accumulation 
of new alleles.  Neither contradicts the other; a geological instant can 
easily be a microevolutionary eternity.  If natural selection is to 
produce any microevolutionary change, it must do so in less than about 
50,000 generations; if it took much more time, the selection would have 
to be so weak as to be swamped by random drift.  Measured in years, 
50,000 generations commonly separate adjacent fossil deposits.

The morphological character involved in one of the first proposed 
examples of punctuated equilibrium, the number of rows of eye facets in 
a trilobite genus, is assumed to be under simple polygenic control and 
to have evolved by stepwise point mutations (albeit during speciation 
events).  Similarly for molar size in the condylarth Hyopsodus (but this 
is disputed as an example of punctuated equilibrium).

I agree that chromosome repatterning can be important in speciation, but 
since chromosomal evolution can affect virtually any trait of an 
organism in addition to its reproductive relationships, I fail to see 
speciation as necessarily different in this regard from any other 
evolutionary change.  Conversely, there's nothing about the speciation 
process that prevents simultaneous divergence due entirely to stepwise 
gene substitutions, and traits important in reproductive isolation are 
often under the control of a few loci (e.g. pheromones, flower color and 
scent, phenology).

>Larry and other readers interested in salvaging genetic definitions of
>evolution would do well to make the following corrections:
>First, avoid limiting the circumscribed phenomena to allele
>replacements.  Its much better to talk about "statistical fluctuations
>in the genetic composition of populations" (Sewall Wright's
>definition), since this will include allele replacements, molecular
>drive, etc., which are legitimate evolutionary phenomena.

The tremendous advances made as a result of the allele's-eye-view of 
population genetics are undeniable.  This of course doesn't mean that 
population geneticists "define" evolution only as change in allele 
frequency, or fail to recognize other evolutionary phenomena (as you 
note with Wright).  Some simple allelic processes are eminently worth 
concentrating on, however, because of the wide-ranging insights they 
yield.  As one of numerous examples, if I want to understand the effect 
of a given mating system on the inbreeding coefficient, I wouldn't miss 
much by focusing entirely on a single intact locus.  The insights gained 
from this can then be combined with a few simple observations to explain 
an enormous range of evolutionary phenomena, e.g. the bimodal 
distribution of outcrossing rate in vascular plants, of polyploidy in 
homosporous ferns, of loss of genetic variability in small populations, 

Evolution may be enormously complex, but in all cases, we're interested
in the demographic trajectory of genetic novelty; whether or not we
call the novelty an "allele" doesn't really change the form of the problem.

Stewart Schultz
Botany Department, U. British Columbia
schultz at

More information about the Mol-evol mailing list