INBRED LINES FOR HETEROSIS TESTS? 335 



with prolificacy, viability, and growth rate in swine, for several reasons. First, 

 the history of selection prior to the beginning of the experiment presumably 

 had not been consistently positive for adult size in mice, as it was for pro- 

 lificacy, viability, and rate of growth in swine. Second, selection for increased 

 size of the organism may be quite different from selection for a further in- 

 crease in efficiency within the same adult body size. Adult size is generally 

 highly heritable but not consistently selected for in either direction in farm 

 animals. The steady decline in effectiveness of selection without reduction in 

 variability for size in MacArthur's study suggests approach to an equilibrium 

 similar to that postulated for total performance in swine. 



Heritability Estimates 



Heritability, the portion of observed variance linearly associated with 

 genotype, ranges from about 10 to 50 per cent for individual characters of 

 economic importance. But heritability is found to be lower for the highly 

 important characters such as prolificacy and viability, for which selection 

 has been appreciable and always in one direction, than for traits such as 

 carcass composition or external dimensions, for which selection has been mild 

 or in opposite directions in different portions of the breed or during different 

 periods of time. Heterozygote superiority is more likely to be important for 

 genetic variability in the highly important characters, since selection would 

 have had greater opportunity to fix those genes whose homozygotes were 

 equal or superior to alternative genotypes at the same locus, leaving at 

 intermediate frequencies (larger qA[i — Qa]) genes exhibiting heterozygote 

 advantage. 



Ineffectiveness of selection for heritable traits suggests that degree of 

 dominance may be higher and heritability lower for total performance than 

 for its individual components. This has been shown for grain yield and its 

 components in corn by Robinson el al. (1949) and by Leng el al. (1949). In 

 swine, Cummings el al. (1947) found heritabilities of 22 per cent for size 

 of litter at birth, 40 per cent for survival from birth to weaning, but only 7 

 per cent for total litter weight at weaning. Heritability of total weaning 

 weight jumped from 7 to 59 per cent when effects of size of litter at birth 

 and of survival were held constant. These results could have arisen from 

 negative genetic-physiological or from high positive environmental correla- 

 tions, or both, between numbers per litter and weight per pig at weaning. 



Positive estimates of heritability for economic characters may be obtained, 

 even though selection is ineffective due to heterozygote advantage. If ^ > 1 

 and rates of reproduction were proportional to phenotypic levels, equilibrium 

 frequency for the more favorable allele would be 9.4 = (1 + k)/2k. At this 

 point, the linear regression of genotype on phenotype in an unselected popu- 

 lation would be zero, and all intra-allelic genetic variability would be due to 



