220 TH. DOBZHANSKY 



tion will consequently tend to lower the frequency, or to eliminate deleteri- 

 ous mutants. 



Selection against a dominant deleterious mutant is, however, a far more 

 efficient process than that against a recessive mutant. This is because dele- 

 terious recessive mutant genes are sheltered from selection by normal domi- 

 nant alleles in heterozygotes. Deleterious dominants are eliminated by selec- 

 tion within relatively few generations after their origin. Deleterious reces- 

 sives accumulate in heterozygotes until their frequencies become so high that 

 recessive homozygotes are produced. Dominant alleles are not intrinsically 

 beneficial, and recessives are not necessarily deleterious. But at any one time, 

 we find in cross-fertilizing populations more deleterious recessives than dele- 

 terious dominants, because the former are not eliminated by selection as 

 promptly as the latter. 



Analysis of wild populations of several species of Drosophila has revealed 

 extensive infestation of the germ plasm by deleterious recessive mutant genes. 

 According to the unpublished data of Pavan and collaborators, 41 per cent 

 of the second chromosomes in Brazilian populations of Drosophila willistvni 

 are lethal or semilethal when homozygous. Among the remainder, 57 per 

 cent are sublethal when homozygous. Furthermore, 31 per cent of the second 

 chromosomes make the homozygotes completely sterile in at least one sex, 

 32 per cent retard the development, and 16 per cent cause various visible 

 abnormalities. Comparable figures for the third chromosomes are 32 per 

 cent of lethals and semilethals, 49 per cent subvitals, 28 per cent steriles, 

 36 per cent retarded, and 16 per cent containing visible mutants. Since 

 every fly has two second and two third chromosomes, it is easily seen that a 

 great majority of individuals in Brazilian populations carry several deleteri- 

 ous variants in heterozygous condition. 



The mass of deleterious recessives carried in normally breeding natural 

 populations has no disastrous effects on the average fitness of members of 

 such populations. This is because the frequency of recessive homozygotes 

 found in a population at equilibrium is equal to the number of the corre- 

 sponding recessive mutants that arise in every generation. The loss of fitness 

 caused in a normally breeding population by dominant and by recessive mu- 

 tants is thus proportional to the frequency of the origin of these mutants by 

 mutation. 



The situation changes completely if a normally crossbred population is 

 subjected to inbreeding. For inbreeding renders homozygous many reces- 

 sives that would remain sheltered in heterozygotes under normal crossbreed- y 

 ing. These recessives become suddenly exposed to natural, or to artificial, 

 selection. The loss of fitness in inbred lines of normally cross-fertilized species 

 is the consequence. Conversely, the heterosis observed in the progeny of 

 intercrossed inbred lines is the outcome of restoring the normal reproductive 

 biology and the normal population structure of the species. 



