504 MUTATION AND PLANT BREEDING 



associated with certain mutagens are already approaching a degree 

 of differentiation which would be of optimum usefulness to the 

 plant breeder. These are: 



(a) Chemical mutagens capable of producing a high yield of 

 visible mutants and a low yield of concomitant breaks. The 

 problems of recurrent screening for fertility and the pres- 

 ervation of unthrifty "carriers" of mutants, which have been 

 described by Nilan, Gaul, and others, should be reduced 

 proportionately with a reduction in chromosome breakage. 



(b)Agents capable of producing chromosomal breakage unac- 

 companied by visible mutations. These would seem to be 

 of potential use in interspecific breeding where the problem 

 involves the transference of a chromosomal segment from 

 one genome to another, e.g.. Sears' transference of rust 

 resistance from Aegilops umbellulata to wheat. 

 It is possible that continued research will lead to the discovery 

 of more efficient mutagens and breakage agents, but, beyond cer- 

 tain limits, the more restricted their action, the less generally use- 

 ful will they be for plant breeding purposes. 



In her review of chemical mutagens, Auerbach concluded that, 

 "Avhile there is good evidence that a common mechanism is respon- 

 sible for the production of gene mutations, chromosome breaks and 

 minute re-arrangements by chemicals, crossing-over seems to be 

 induced in some different way". From the plant breeding point of 

 view this is an unfortunate situation because methods of controlling 

 chromosomal recombination have at least as great a potential value 

 as methods of inducing mutations. In any heterozygous population 

 of finite size, linkage imposes stringent restrictions on the genetic 

 variation which can be expressed. Its effects only become negligible 

 as the population approaches homozygosity through selection and/or 

 close inbreeding. Following hybridization, linkage favors the recov- 

 ery of a desirable parental complex, but makes more difficult the 

 insertion of specific improvements. Conversely, the successful recom- 

 bination of certain selected characteristics can only be gained at 

 the price of disrupting other desirable parental combinations result- 

 ing from uncontrolled crossing over. The problem can be alleviated 

 though not removed by the use of recurrent backcross techniques. 

 The latter have usually been limited to the transfer of a few simply 

 inherited characters and in wide crosses are often complicated by 

 close linkages and/or apparent pleiotropic effects. Theoretically, 



