GENIC INTERACTION 185 



studies as those of Sumner 1302, 1303 on the genetics of variability within and between 

 subspecies of the deermouse Peromyscus maniculatus. More recently this general view- 

 point has received increasingly massive support in species of Drosophila under the 

 leadership of Dobzhansky, 283 and we have an enormously richer concept of what the 

 genetics of such wild species is actually like. 



We consider first the case of a species (or subspecies) that is essentially homo- 

 geneous throughout its range. Because of the conditions of balance that maintain the 

 array of alleles at each locus, this would consist largely of isoalleles. The virtually 

 infinite array of possible recombinations would insure the genetic uniqueness of every 

 individual even though there would be little conspicuous variability. Since, however, 

 the recombinant genotypes are broken up in each generation by the reduction division 

 (with qualifications that are unimportant unless linkage is nearly complete and selection 

 very strong), selection is based only on the net effects of the genes. If the absolute 

 selective values of genotypes are independent of the genie frequencies, Fisher's 373 

 fundamental theorem of natural selection holds: "The rate of increase of fitness of any 

 organism at any time is equal to its genetic variance in fitness at that time." 



Genetic variance was here defined as merely the additive component. Thus 

 favorable interaction effects of genes with unfavorable net effects cannot be utilized 

 any more than under the first theory. There is no way by which a species can work its 

 way from a lower to a higher selective peak with respect to mean selective value. 

 Once the population has arrived at a selective peak, further evolution can occur only 

 by a change of conditions or a wholly favorable mutation, both of which change the 

 whole system of selective values. If, however, such a change occurs, the heterallelic 

 character of the population permits an extensive readjustment on the basis of interaction 

 effects until a new peak is arrived at, a process that cannot occur under the first theory. 



In the third theory, it is assumed that there is sufficient isolation of small local 

 populations (denies) in at least some part of the range of the species to permit significant 

 genetic differentiation, but not so much that a successful deme cannot modify the genetic 

 composition of less successful neighboring demes by emigration and crossbreeding. 1422 

 The introduction of immigration pressure into the conditions of balance at each locus 

 makes for much more strongly heterallelic arrays, locally as well as in the species as a 

 whole, although these conditions are still such that these arrays consist largely of 

 isoalleles and minor modifiers. In such a species, population is continually welling 

 up in some places, falling off in others, but the sites of the population sources and sinks 

 may be changing from time to time. This process of interdemic selection supplements 

 the continuous process of intrademic change of the sort considered in the second 

 theory. 



With a suitable balance between selection and immigration on the one hand, and 

 between the resulting tendency toward and away from equilibrium from the cumulative 

 effects of random processes on the other, the array of genie frequencies may occasionally 

 pass from control by one selective peak to control by another. Figure 32 shows the 

 paths which the array of genie frequencies tends to take under selection alone in two 



