Motoo Kimura 105generated almost exactly cancels the epistatic variance. Thus, contrary to the
wisdom of the time, a long-time selection program can be predicted more accu-
rately by ignoring epistasis. It vindicated his earlier lack of interest in epistatic
components while he was at Iowa State University. For a discussion, see Crow and
Kimura[Crow and Kimura, 1970], pages 195-225.
This is only a sample of Kimura’s enormous number of accomplishments in
mathematical population genetics. The best way to key into Kimura’s work is
a volume of his major papers that have been reprinted with explanatory intro-
ductions by N. Takahata, published in 1994 by the University of Chicago Press
[Kimura, 1994].
3 THE NEUTRAL THEORYIn 1968, Kimura presented his block-buster[Kimura, 1968]. This happened in
the beginning days of molecular evolution and the pattern of amino acid changes
in various evolving proteins was becoming known. The conventional wisdom was
that natural selection was the main driving force. Kimura dared to say that most
molecular evolution — evolutionary changes at the DNA level — were selectively
neutral, driven by the mutation rate and random allele frequency drift. A conse-
quence is that, viewed over a long time period, the rate of neutral evolutionary
change is determined by the mutation rate alone.
Kimura’s first arguments were not very convincing, but at about the same time
King and Jukes[King and Jukes, 1969]reached the same conclusion. There was
a great deal of debate, some quite vociferous. To some traditionalists, the idea
of neutral evolution was simply unthinkable. Nevertheless, additional evidence in
favor of neutrality accumulated.
It should be emphasized that Kimura did not deny that evolution of form and
function is driven by natural selection. What he asserted is that this was a small
part of change at the DNA level, most of which was evolving neutrally. With
time it became clear that in mammals only a small fraction of the DNA codes for
proteins; the great bulk is “junk” with no known function. It is reasonable that
much of the evolutionary change in this junk is neutral while changes in amino
acid composition are more often selected. The jury is still out on the question
of just how much DNA change is neutral. Although the neutral theory remains
controversial, one point is not in doubt. Being concrete and mathematically simple
it leads to testable predictions, and this has been one of its greatest merits. It has
had great heuristic value and a neutral prediction is often used as a null hypotheses
for testing various models of selection.
It is common in the history of science for mathematics that was developed for
its own sake or for a specific subject turns out to have a totally unexpected use.
Examples abound, but usually the application comes long after the theory was first
developed. Kimura had the good fortune that much of the mathematical theory
that he developed in the early days of his career was almost made to order for the
neutral theory. Such problems as the probability of fixation of a neutral mutation,