Evolution should not be equated with mutation, etc

arlin at ac.dal.ca arlin at ac.dal.ca
Sat Jul 4 21:16:58 EST 1992

Two terminological muddles that warrant clarification:

1.  PAM stands simply for "percent accepted mutations" and not "point accepted
mutations per 100 residues," as Gonnet and colleagues suggest (p. 1444, top
left). I don't know where this interpretation started, but I've seen it many
times before.  Obviously, since "percent" is Latin for "per 100", both
statements say exactly the same thing-- the only difference is that Dayhoff's
original coinage conforms smoothly to the acronym "PAM," while the other
coinage does not.

2.  But there is a more serious problem with terminology, which may reflect an
underlying misunderstanding of the evolutionary process (I hope not!). The
log-odds matrix that appeared in Figure 2 of Gonnet et al. (Science 256: 1444)
is in no correct sense a "mutation" matrix; also "mutation" is mis-used to
describe site differences between two sequences (middle right of p. 1444, and

A genetic mutation is a cellular event in which a nucleic acid genomic
sequence changes slightly.  We also use the term "mutation" to describe the
phenotypic effects of such a genetic mutation: for instance, the sudden
appearance of a novel alcohol dehydrogenase allele or of a white-eyed fly in a
drosophila population would represent "mutation".  My own personal way to
remain precise and consistent in making distinctions is to reserve
"mutation" to describe the molecular mutational process, and use "mutant allele"
and "mutant phenotype (or trait)" to describe the effect of this process on a
"mutant individual's" genotype and phenotype.

Whether or not one chooses to adopt the above set of distinctions, there is no
sense in which a "mutation" is an ALLELE REPLACEMENT, the unit step in genetic
evolution.  Genetic evolution may begin with the process of mutation, but it
does not stop there.  Most mutant alleles that arise by mutation are neutral or
deleterious.  Deleterious mutant alleles tend to be removed by selection; only
a few rare neutral or advantageous mutant alleles go on to be fixed by
selection or drift: the vast majority of the neutral and advantageous mutant
alleles that arise by mutation are quickly lost by genetic drift (for
reference, see Tables and in Crow and Kimura, _An Introduction
to Population Genetics Theory_, p. 422-423).  That is, "mutation" is only the
first step in the process of genetic evolution, and the vast majority of
mutant alleles arising from mutation play no lasting role in evolution.

Thus, saying that evolution is a bunch of mutations is in some ways like
saying that a garden is a handful of seeds: most seeds are lost, and even
those that bear fruit have been carefully planted and tended-- both seeds and
gardening are necessary and creative steps in the process.  Equating an
evolutionary change with a "mutation" recalls a non-Darwinian hypothesis
called "mutationism," which died out over 50 years with the advent of
evolutionary population genetics.  Of course, no one here subscribes to 
that outmoded view-- but that's the way it sounds, unfortunately.  

The "A" in "PAM" acknowledges that evolution involves more than mutation by
implying that some mutations are "accepted" by a vague sort of evolutionary
filtering process, while others are not.  Although "accepted mutations" is
better than just "mutations", it does not do adequate justice to 70 years of
population genetics theory.  A consistent and valuable set of terms is as
follows: a difference between two amino acid sequences is not a "mutation" but
a "difference," or "site difference."  We may infer that a difference resulted
from a process of "amino acid replacement," which is fundamentally a result of
"allele replacement."  The number of amino acid replacements necessary to
account for the observed level of difference between the sequence is
fundamentally a distance measure (see my earlier post on "distance" and
"difference").  The sort of probability matrix pioneered by Dayhoff is an "log
odds AMINO ACID REPLACEMENT probability matrix", not any kind of "mutational"
probability matrix.  

Many readers will recognize that these suggestions were not invented by me, 
since Walter Fitch, editor of Molecular Biology and Evolution,
used to make prescriptions regarding these terms, insisting that manuscripts
refer to observed differences as "differences" rather than as "mutations", and
the inferred allele replacements as "nucleotide substitutions" (for genes)
or "amino acid replacements" (for proteins), rather than as "mutations" or

Arlin Stoltzfus
Department of Biochemistry
Dalhousie University
Halifax, Nova Scotia
arlin at ac.dal.ca 

More information about the Bio-soft mailing list