INBRED LINES FOR HETEROSIS TESTS? 337 



AasLndAA will be magnified and those between. 1.1 and aa minimized, mak- 

 ing k' larger than k and equilibrium (ja smaller than (1 + k)/2k. The larger 

 the number of genes controlling genetic variation in the basis of selection, 

 the less difference intensity of culling will introduce between k' and k. 



Estimates of heritability from regression of offspring on parent will in- 

 crease positively as equilibrium q becomes larger than (1 + k)/2k, and as- 

 sume larger negative value as q becomes smaller than (1 + k)/2k. Larger 

 positive heritability estimates based on paternal |-sib correlation will be 

 obtained as q becomes either larger or smaller than (1 + k)/2k, since this 

 method estimates fraction of the phenotypic variance linearly associated 

 with genotype regardless of the sign of the regression of ofTs})ring on parent 

 (Fig. 21.2). 



It seems clear that positive estimates for heritability of individual char- 

 acters do not rule out the possibilities (1) that heterozygote advantage ob- 

 tains, especially for net selective advantage or total performance; and (2) 

 that effectiveness of selection may be only a small fraction of that indicated 

 by the estimates of heritability for individual characters. More attention 

 needs to be given estimates of heritability for total performance indices and 

 their components. 



Negative Genetic Correlations between Components 

 of Total Performance 



Existence of negative genetic correlations would correspond to hetero- 

 zygote superiority. This is in the sense that an increase in frequency of 

 genes with partially or completely dominant favorable effects on one char- 

 acter would amount to a decrease in frequency of their alleles having partial- 

 ly or completely dominant favorable effects on one or more other characters. 

 This involves the reasonable assumptions that genes have manifold end 

 effects and that selection maintains at intermediate frequencies where 

 contribution to genetic variability is larger — only those genes having domi- 

 nant favorable but recessive unfavorable effects on performance. Mac- 

 Arthur's (1949) experiment provided ample evidence that selection for a 

 single character (adult size) produces many important changes in other 

 characters. 



Direct evidence for negative genetic correlations is not plentiful. Much 

 data must be analyzed to estimate genetic correlation with precision, par- 

 ticularly when the traits correlated are of low heritability. Also, it is difficult 

 to avoid bias from environmental correlations. If leaner hog carcasses are 

 considered desirable, the genetic correlations of .3, .7, and —.7 for ratio of 

 fat to lean cuts with 180-day weight, daily gain, and feed requirements per 

 unit of weight gain found in a study of Iowa Record of Performance data 

 (Dickerson, 1947) need to be considered. In the same and in another study 

 (Dickerson and Grimes, 1947) evidence for genetic antagonism between 



