338 GORDON E. DICKERSON 



good milking ability and rapid, economical fattening in swine is presented. 



Other reasons for expecting negative genetic correlations that might be 

 mentioned are (1) lower heritability for total performance than for its com- 

 ponents, as outlined previously, and (2) approach to some physiological 

 maximum, where increase in one function must necessarily reduce others, as 

 in division of nutrient energy available between milk production and fleshing. 



Negative genetic correlation, in some degree, is maintained by the 

 process of selection itself and would disappear if selection were relaxed. 

 Animals mediocre in any one respect are retained as breeders only if superior 

 in several other characters. Thus selection leads to a negative correlation be- 

 tween characters among the animals selected as parents. To a much lesser 

 degree, these negative relationships would appear among the progeny, where 

 fresh selection would magnify them again. Such negative character relation- 

 ships may explain in part the discrepancy between rates of improvement 

 "expected" and obtained, and could exist quite apart from any real heterozy- 

 gote advantage. 



Analogy between Results with Corn and with Swine 



In both corn and swine, (1) inbreeding has been slight during domesticated 

 history, until recently at least, (2) degree of heterozygosity exerts a major 

 influence on performance, (3) effectiveness of continued phenotypic selection 

 is questionable in stocks with a long history of selection for the same complex 

 of characters in which further improvement is sought. 



Hull (1945) has postulated overdominance or heterozygote superiority, 

 with additive interaction of non-alleles, to explain corn breeding results. He 

 does so on the basis that (1) yields of parental, Fi, Fa, and backcross popula- 

 tions are linearly related to proportion of loci heterozygous (Neal, 1935), (2) 

 yields are usually less than one half as large for homozygous lines as for 

 their Fi crosses, (3) regression of Fi yield on parental inbred yield among Fi 

 crosses having one parent in common often is zero or negative for the higher 

 yield levels of the common parent. Robinson et al. (1949) have obtained esti- 

 mates indicating heterozygote advantage {k = 1.64) for grain yield but only 

 partial to complete dominance for components of yield. Crow (1948) has 

 shown that under complete dominance {k — 1 ) of favorable genes combining 

 additively, average superiority of maximum hybrid over the variety at 

 equilibrium gene frequency would be the product of mutation rate and 

 number of loci, or less than 5 per cent, whereas potential hybrid advantage 

 under some degree of heterozygote advantage (^ > 1) at even a small pro- 

 portion of loci could be many times greater, in agreement with results al- 

 ready obtained 



The impossibility of accounting for the 15 to 25 per cent advantage of 

 better corn hybrids over open-pollinated varieties through complete domi- 

 nance of favorable genes combining additively can be demonstrated (Dicker- 



