Sewall Wright 95lation population “climbs” the higher peak. Although phase 1 depends
on random processes, phase 2 is simple selection and moves relatively
rapidly.
Phase 3.A subpopulation at a higher peak, being more fit, will increase
in numbers relative to other subpopulations and therefore, perhaps
because of crowding, export migrants. These migrants upgrade the
subpopulations into which they have migrated, much as the Shorthorn
bulls exported from a superior herd raised the breed quality. In this
way the entire population is upgraded, and the process can start over.Wright did a great deal of mathematics on phases 1 and 2, but not phase 3.
However more recent work has shown that, at least in some cases, phase 3 can
work as Wright envisioned[Crowet al., 1990]. Repeated directional migration can
be a strong force.
One criticism of Wright’s theory is that it places quite stringent requirements
on the population structure. In the first place there must be several subpopula-
tions, the more of these, the better for the theory. The individual subpopulations
must not lose the variability necessary for random gene-frequency drift, so some
migration from other colonies is required. Yet, too much migration leads to loss
of subdivision; the population becomes a single unit and the process fails. So a
major criticism of Wright’s theory is that the conditions required for it to work
optimally will not occur very often.
A second objection is that phase 1 reduces the population fitness, since random
changes are mainly maladaptive. It is not appealing, at least to some evolutionists,
that a process intended to increase fitness can function only by at least temporarily
reducing it.
A third criticism is that a more appropriate metaphor may not be a rugged
landscape but rather an ocean in which the surface is constantly changing. The
idea is that, given an ever-changing environment, it is very unlikely that a pop-
ulation will ever be in such a state thatnoallele frequency change can increase
fitness.
Finally, mass selection, often aided by artificial insemination, has been strikingly
successful in improving livestock performance. One wonders whether if Wright had
observed modern improvement in milk production of dairy cattle by mass selection,
rather than the process he observed in the history of Shorthorns, he would have
developed the shifting-balance theory.
None of these criticisms deterred Wright. He did not regard his process as nec-
essarily of frequent occurrence, but as a way in which some population somewhere
can bypass the difficulty of changing from a lower fitness peak to a higher one
when the intermediate stage is maladapted. It was his preferred way of generating
evolutionary novelty.
Wright’s theory has long been very popular among biologists, although less so
among theoretical population geneticists. Recently, it has come into more and
more criticism, e.g. [Coyneet al., 2000], so I think it is fair to say that the