FRED H. HULL 



Florida Agriculfural Experimenf Sfation 



Chapter 28 



Recurrent Selection 

 and Overdominonce 



For many breeders, in considering problems that lie ahead and methods 

 of meeting them, the main problem is whether to continue with varieties or 

 breeds, or to work with inbred lines and Fi crosses. Behind this question are 

 the problems dealing with the relative importance of general and specific com- 

 binability, or of prepotency and nicking: 



Is the yield gain of hybrid corn due mainly to selection within and among inbred lines, 

 or to selection among Fi crosses of inbred lines? 



Is it due to improved frequencies of dominant favorable genes in elite inbred lines which 

 are parents of elite-yield hybrids? 



Is selection within and among inbred hnes to accumulate higher frequencies of domi- 

 nant favorable genes many times more powerful in one cycle without recurrence, than is 

 selection without inbreeding through many recurring cycles? 



To what extent may higher levels of specific combinabihty be reached by recurrent 

 selection? 



How may heritability of specific combinability be evaluated? 



Why have the less favorable alleles of vigor genes been retained in such high frequencies? 



May selection for general combinability and selection for specific combinability some- 

 times have counter effects on gene frequencies? 



Does superiority of Fi crosses of inbred lines over varieties or breeds necessarily depend 

 on overdominance? 



If this choice of problems is approximately correct, the research emphasis 

 may begin to shift from effects of inbreeding to effects of selection. 



EARLY EXPLANATIONS FOR HYBRID CORN 

 East and Emerson in an early paper considered the theoretical problem of 

 recovering two traits together from a crossbreeding po]Hilation in which the 

 frequency of each trait was 1/1000, and the two were independent. The au- 

 thors offered two solutions: first to select at the rate of one per million in one 

 generation, and second at the rate of one per thousand in two generations re- 

 currently, first for one trait and then for the other. It is clear now that selec- 



451 



