INBRED LINES FOR HETEROSIS TESTS? 347 



then can the slow-starting reciprocal selection begin moving qi and 52 toward 

 opposite extremes. 



Use of Partially Inbred Tester Lines 



In large animals, or even in poultry, discussion of selection utilizing homo- 

 zygous tester lines is still largely academic. Few very highly inbred and 

 usable lines of swine and chickens exist. However, there are many partially 

 inbred lines of swine and poultry whose average cross performance has been 

 or is now being tested. These partially inbred lines should be extremely use- 

 ful in overcoming the initial disadvantage of reciprocal selection, because in- 

 breeding will have pushed frequencies of individual genes in these lines much 

 further away from equilibrium than in non-inbred stocks. Since effectiveness 

 of reciprocal selection (equation 8) increases with 





even a moderately inbred line used as one of two populations under reciprocal 

 selection would materially increase initial progress per cycle from Aqi. 



Of course, further selection within the inbred line itself on the basis of 

 cross performance would be relatively ineffective (A92 small) until the selec- 

 tion on cross performance has had time to shift qi at individual loci in the 

 non-inbred population away from (1 + k)/2k in the opposite direction from qi. 

 It might be wise to ignore cross performance in selecting replacements 

 within the inbred line for a number of cycles to allow time for qi to make this 

 shift at loci where initial ^'i and 92 chance to deviate from (1 + k)/2k in the 

 same direction. Beyond this point, progress from reciprocal selection between 

 the partially inbred and the non-inbred populations should approach and 

 finally exceed that from selection for cross performance with a homozygous 

 tester. 



In selecting a partially inbred line for use in reciprocal selection, one in- 

 stinctively would choose a line known to be superior in its average cross per- 

 formance and in its usability as an inbred strain. This seems desirable to 

 assure that the line carries at high frequencies any genes whose favorable 

 effects on total performance are incompletely dominant. In addition, it 

 would be helpful to try a number of different partially inbred lines in crosses 

 with a given non-inbred stock, choosing finally for reciprocal selection the 

 line showing best initial cross performance. Diversity of origin and previous 

 crossing data would of course aid in selecting the lines more likely to be 

 initially complementary to a given non-inbred stock. 



Presumably, both initial cross performance and rate of progress from 

 reciprocal selection are likely to be greater if the two populations are of dif- 

 ferent breeds. However, Dobzhansky's (1949) finding of greater heterozygote 

 advantage from alternative homologous chromosome segments within a 

 single population than in crosses between non-interbreeding populations of 



