GENETIC DRIFT AND NATURAL SELECTION IN EXPERIMENTAL POPULATIONS OF DROSOPHILA PSEUDOOBSCURA

An interesting indeterminacy of outcomes is observed in experiments with certain laboratory populations of Drosophila. Natural populations of Drosophila pseudoobscura and of many other species are polymorphic with respect to the gene arrangements in their chromosomes. The polymorphism is balanced; the heterokaryotypes with two chromosomes of a pair differing in gene arrangement are, in most environments, superior in Darwinian fitness to the corresponding homokaryotypes. The relative frequencies of the different karyotypes in experimental populations change from generation to generation until quasistable equilibria are attained. The results of the experiments depend, however, on whether the populations are uniform or mixed in geographic origins, i.e., on the chromosomes having been derived from the same natural population or from populations of different localities. In the former populations, the results are generally determinate; if the environment is well controlled, the selection rates and the equilibrium points are constant and reproducible. Not so in populations of geographically mixed origins; replicate experiments often give disparate results, and the selection rates and the equilibrium points are unpredictable for a given population. 1-4 A Working Hypothesis.-The indeterminacy of the outcomes of the selection processes in populations of geographically mixed origins may be due to our experimental populations containing far too few flies to have more than a fraction of the possible genotypes actually realized. Supposing that the geographic populations, or races, crossed differ in 50 -to 100 genes, between 350 and 3100 genotypes will be potentially possible in the progenies; on the other hand, the populations fluctuate in size roughly between 1,000 and 4,000 adult individuals. Now, the relative fitness of the homoand heterokaryotypes depends upon the genetic system as a whole. Which of the possible recombination genotypes arise first in a given population, and which appear later or not at all, is a matter of chance. Natural selection in the experimental populations will, however, work with the genotypes which will happen to be available. The dissimilar results found in experimental populations of geographically mixed origins arise, then, owing to the divergent action of natural selection; the divergence is due, in turn, to the operation of the sampling processes which bring some of the potentially possible genotypes to realization and leave others unrealized. This working hypothesis permits some predictions to be made. In populations large enough to have all the potentially possible genotypes actually appear, the results should always be determinate. Experimental populations of 350 or 3100 individuals obviously cannot be obtained. There must, however, be some relation between the population size and the degree of the indeterminacy, the variance of the results presumably increasing as the population size decreases. To test this prediction, Dobzhansky and Pavlovsky5 have crossed 12 strains of D. pseudoobscura