The Development of Population Genetics 331was simply not an option.
Without invoking Wright’s adaptive landscape theory we can characterise the
two approaches to evolution as atomistic vs. holistic [Gayon, 1999]. In Fisher’s case
the gene itself is the primary element in adaptive evolution while for Wright it is
gene interaction. It is important to stress that this is not simply an issue about the
importance of selection in different contexts, but rather about its actual operation.
For Fisher, selection is a straightforward affair. Once the genetic raw material has
been furnished by mutation, natural selection is regarded as the sole agency in
shaping genetic evolution. Because complexes of genes at various loci tend to be
broken up by recombination, a stronger emphasis is placed on genes at single loci
than on gene complexes. And, the total additive genetic variance is defined as the
sum of the constituent one-locus marginal values [1918, 405]. Moreover, an allele
having a net selective advantage is destined for fixation as long as that advantage
persists; hence, even a minute selective force can have evolutionary consequences.
Contrast this with Wright’s view and its emphasis on the entire gene complex.
As I mentioned above, selection is, in many senses, a deterministic matter on
Fisher’s account but for Wright the importance of relatively small population
sizes is that such deterministic behaviour does not occur. Both random drift and
random changes in selective values can move gene frequencies to higher selective
peaks. Migration between these small populations also results in interpopulational
selection. So, the unit of selection here is the entire gene complex and not the
individual alleles which are often seen as having no absolute selective advantage.
Most characters were thought to be affected by genes at many loci with most genes
influencing several characters.
Although Fisher and Wright were indeed asking the same kinds of questions re-
garding the role of selection in producing adaptation, each had different answers.
Fisher’s view was that selection operated best in large randomly breeding popula-
tions. Wright, on the other hand, argued that there would be a rapid differentiation
of local strains which was in itself non-adaptive, but permitted selective increase
or decrease of the numbers in different strains. This would result in a relatively
rapid adaptive advance of the species as a whole. For Wright the problem with
large populations is that they were subject to slow and often reversible changes
with selection while small ones were too likely to decline or become extinct as a
result of random fixation of disadvantageous genes. In an intermediate population
or one with small subgroups there tends to be partly adaptive and partly non-
adaptive radiation among the subgroups causing the successful genes to spread
and the others to decline. This, he thought, produced changes in the mean gene
frequency of the species as a whole with a much faster rate of evolution [1931,
151]. Both random drift and inbreeding were significant in this model because the
latter allowed for complex interactive gene systems to be held together and the
former allowed the population to come into contact with genes of greater selective
advantage. In a small inbred population where only selection operates the result
would be stagnation once a certain level of adaptation was reached.
What the synthesis of Darwinian selection and Mendelian genetics produced