Genetic Systems II \ 191 



havior is variable. For example, in Raphanobrassica, there seem to 

 be so many small differences between the chromosomes of the 

 parental species that pairing does not take place and only univalents 

 are found at meiosis. The same is true of the cross between Allium 

 fisfulosum (scallion) and A. cepa (onion). The univalents are dis- 

 tributed at random, and only rarely will cells with balanced chromo- 

 some complements be produced. At the opposite extreme, in the 

 Primula example and in the cross between Festuca prafensis (mead- 

 ow fescue) and Lolium perenne (perennial rye), bivalents are 

 formed in the diploids. This suggests that there are fewer differ- 

 ences between the genomes involved. Nevertheless, following dis- 

 junction, the diploid hybrids are sterile. Doubling of the chromo- 

 some number provides the chromosomes of each genome with the 

 appropriate homologues at meiosis. Bivalents are formed and fertil- 

 ity is restored. 



The resulting allotetraploids are sometimes referred to as amphi- 

 diploid since, in effect, they are diploid for each parent. That is, 

 2n = 2xi + 2xo. Allotetraploids of this sort occur in nature. Observa- 

 tion of the chromosomes of artificial hybrids between the suspected 

 tetraploid and its possible parents will suggest the relationship in- 

 volved. In a backcross to a genuine parent, a mixture of bivalents 

 and univalents occurs. It is apparent, however, that there is an evo- 

 lutionary disadvantage inherent in the make-up of a strict allotetra- 

 ploid. The genomes may be so differentiated that, when they are 

 combined in the fertile tetraploid hybrid, genome recombination 

 does not occur. The parental genomes presumably continue to ex- 

 hibit the same amount of recombination as they did in the diploid. 

 Because pairing is perfectly normal, genes of different genomes can- 

 not be recombined and segregated. 



Probably more common in nature is a tetraploid in which the 

 genomes are partly differentiated but are still sufficiently similar 

 that multivalents are found in which crossing-over between chro- 

 mosomes of the two parents may occur to produce genetic recombi- 

 nation. This is referred to by Stebbins as segmental allopolyploidy, 

 because only some segments of the chromosomes are different. 

 While segmental allopolyploids are of the greatest significance for 

 evolution, they are difficult to recognize in nature. They may result 

 from hybridization of morphologically similar parents, and because 

 of multi\'alent formation, they simulate autopolyploids. Further- 

 more, as a result of their partial sterility, they are unstable. Selec- 

 tion, acting to increase fertility, may tend to favor further chro- 

 mosome differentiation. Bivalents may be formed and a strict allo- 

 polyploid result. 



