10 AGRICULTURE HANDBOOK 495, U.S. DEPT. OF AGRICULTURE 



extra chromosomes are uncom- 

 mon, and such counts are not re- 

 liable. Aneuploidy, when it occurs, 

 would contribute to the sterility of 

 the species. Extra chromosomes 

 have been attributed to a sex-de- 

 termining mechanism. There is, 

 however, no concrete evidence to 

 suggest an XO sex-determining 

 system. Furthermore, Martin (20) 

 has shown how polypoidy can ex- 

 plain the unusual sex ratios found 

 in most species. Males predominate 

 over females in most controlled 

 crosses, in predictable ratios as- 

 sociated with numbers of sex-con- 

 trolling chromosomes present. 



As previously noted, male and 

 female flowers are usually sterile, 

 and in varietal collections seeds are 

 seldom produced. No efforts to 

 breed the wing-stemmed yam have 

 been reported in the literature. It 

 is difficult to escape the conclu- 

 sion that existing varieties are 

 very old and perhaps have di- 

 verged from their progenitor 

 varieties by somatic mutation. 

 Nevertheless, a sexual system must 

 have been present at one time to 

 have produced the variation ex- 

 hibited by this species. Primitive 

 man presumably practiced selec- 

 tion and gradually produced the 

 fine varieties now known. It ap- 

 pears that this process of improve- 

 ment is no longer possible. 



VARIETIES 



The variation of D. alata has 

 been dealt with in several classifi- 

 cations. Roxburgh (27) divided D. 

 alata into five species (see "Clas- 

 sification") , chiefly based on the 



anthocyanin coloration of tuber 

 and foliage, but also on minor 

 morphological features such as 

 number of wings on the stem. The 

 materials available to him appar- 

 ently did not represent the full 

 variation in the species. Ochse and 

 van den Brink (25) set aside a 

 group of highly improved varieties 

 from a much larger, more hetero- 

 genous primitive group. Burkill, 

 a careful observer who had an ex- 

 tensive Southeast Asian collection, 

 based his classification on four 

 kinds of tubers (3): long and deep- 

 ly buried tubers, half-long tubers, 

 short tubers, and nonburying 

 tubers. He then broke each class 

 into a number of subclasses based 

 on characters such as branching 

 tendency, presence of a neck, and 

 tendency to curve. While his classi- 

 fication groups certain related 

 varieties together, it is an over- 

 simplification of the varietal pic- 

 ture and is useful principally in 

 making rough order out of appar- 

 ent chaos. 



Gooding (15) developed a key to 

 distinguish some West Indian 

 varieties of D. alata. His system, 

 though useful, cannot adequately 

 deal with a large collection. Martin 

 and Rhodes (22) studied the cor- 

 relations among 100 character- 

 istics of 30 D. alata varieties and 

 found that certain tuber and 

 foliage characteristics tended to 

 occur together. Three groups of 

 varieties were delimited: Feo (fig. 

 5) , primitive, and choice. Feo was 

 judged to be a natural grouping; 

 choice, highly artificial; and primi- 



