Systems of Ecological Adaptation
 A limited number of significant data will be quoted only
 among the possible ones.
 Some species are rich in cultivars with a tendency to a perennial
 aerial system. Among D. cayenensis-rotundata, "Yellow yam"
 cultivars, if not or if partially harvested, hold many of their first
 leaves over two years in high rainfall areas. Among D. esculenta
 and D. trifida, the leaves of two successive growing seasons are
 often overlapping. However, the annual disorganization of the
 aerial system is widely prevailing.
 In the northern hemisphere, when plantation is progressively
delayed from July to December, the leafy growth part of the cycle
and the yield are reduced first, then are gone quite haywire in
most cases. But cultivars of D. cayenensis and D. trifida can be
less affected (Arnolin, 1982, 1983).
 Cultivars geographical adaptation appears favoured by archaic
traits and vigour, considering the dispersal of the D. alata
primitive group (Martin, 1974; Degras, 1976) or the experi-
mental behaviour of selected D. trifida hybrids compared to
traditional Guyanese cultivars (Degras, 1980). The correlation
observed by Toure and Ahoussou (1978) between cultivars per-
formances in the centre and the south of Ivory Coast also supports
this idea. The recent world dispersal of some cultivars of D. alata
from the germplasm collected by F. Martin in the seventies,
though highly linked to their low susceptibility to anthracnose, is
not entirely at discrepancy with it (Degras et al., 1983).
 Much has to be described of the different levels of susceptibili-
ty to diseases and pests of the yams among and within species.
While they seem clear for anthracnose, they could appear in-
conspicuous for nematodes: the classification of the same cultivars
of D. rotundata by Adesiyian (1977) and Bridge (1978) is so dif-
ferent that the adequacy of methodologies has to be questioned.
Nevertheless, species rank in susceptibility may be safely
evaluated regarding definite nematodes in definite areas (Hickl-
ing, 1974; Nwauz and Fawole, 1981). The situation could be the
same for virosis susceptibility: in the Caribbean area, decreasing
levels of tolerance to the virus complex are seen from D. esculenta
to D. trifida through D. alata and D. cayenensis (Mohamed and
Mantell, 1976; Marchoux, 1980). Another field of species varia-
tion has recently been discovered in Guadeloupe (Kermarrec,
Febvay and Guerrier, person. comm.): the decreasing level from
D. cayenensis to D. alata or D. bulbifera through D. trifida, of
their leaves cut by the Attines ants.

The Systems of Genetic Variation
 These systems are mainly those which resulted in the different
kinds of variation we have briefly reviewed. But, as everywhere,
the biotechnology era can not only amplify their effects but give
rise to new paths of variation.
 Dioecy (with a higher frequency and/or earliness of males
perhaps) is characteristic of the genus. Monoics, however, are not
unknown (Miege, 1952; Degras, 1957; Burkill, 1960; Sadik,
1975; Toure and Ahoussou, 1978; Abraham and Nair, 1979), but
their fertility remains questionable (Martin and Cabanillas, 1966;
IITA, 1977).
 However, more than dioecy, the deficiencies of the flowering
and seed setting phases are commonly called to mind as the
limiting factors in Dioscorea caynensis-rotundata and at lower
levels in D. trifida. But these cases are far from prevalent among
the hundred edible yam species. Moreover, it is only the poor
status of the research in yam breeding which lets still aside the
fertile circles of wild parent species of any major cultivated ones:
since the interesting hypothesis of Prain and Burkill (1939) the
likely contribution of D. persimilis and D. hamiltonii to D. alata
speciation, no experimental study has tried to utilise them both
as a possible source of D. alata sexuality renewal. The approach of
the relation between the D. cayenensis-rotundata complex and its


supposed circle of wild parents is only beginning (Hamon and
Toure, 1982).
 We consider that, regarding their breeding system, these yams
 must be each used as sympatric species where panmixy has played
 a part during their initial sexual history. Sterility should be secon-
 dary to the higher ecological adaptation of vegetative reproduc-
 tion system in humid tropical zones, the easy accumulation of
 sexually unbalanced genomic and genic combinations and the
 human selection pressure towards juvenile stage for better food
 quality. We must have in mind the successful crossings (Martin
 and Cabanillas, 1966) between diosgenin species pertaining to
 two botanical sections, quite different either in the systematic of
 Uline-Knuth (1924) or the Matuda (1954) one. The preferential
 chimiotropism of the pollinic tube of a D. trfida cultivar towards
 a D. alata cultivar pistil could be an experiment of significant
 value (Bulle-Legrand, 1983).
 That the so-called "sterility" is no more an absolute barrier in
D. alata genetical variation is shown by the breach observed in
the Javanese material and now explored by IITA. Meanwhile,
works are going on in the perennial cultivation, photoperiods
(Miginiac, 1980) or growth substances applications (Bulle-
Legrand, 1982). Cool storage (18" C) has recently brought new
information on D. trifida flowering behaviour (Arnolin, personal
communication).
 The relatively high flowering and fertility level of some paren-
tal combinations in D. trifida, followed by a moderate level of
variation of their progenies for the main characters, will permit to
release "composite" cultivars of clones to the farmers in
Guadeloupe. These composites will benefit for one or more
clonal generation from the natural virus clean state brought by
the sexual reproduction.
 The most serious handicap in sexual breeding of yams will re-
main the absence or the weakness of correlation between seed-
lings and clonal generations: in D. trifida as well as in D.
cayenensis-rotundata (Degras, 1980; Wilson, 1980) many
character variations, among which yield is one, are significant at
the second clonal generation only. However, this delay is not
greater than the time spent in most grain crops. And it can be
used for a vegetative multiplication of a representative sample of
each progeny, enabling thus more valuable tests at the optimum
generation.
 The first contribution of the vegetative system of reproduction
to the enlargement of genetical variations through a better effi-
ciency of sexual selection. It can be done by horticultural tech-
niques (stem cuttings in greenhouse) or in vitro techniques
nodall micropropagation), under diseases or pest selection
pressures.
 But we have to expect more and more direct paths of improve-
ment from in vitro techniques. The development of
micropropagation to improve the exchange and the conservation
of germplasm and to ensure an easier diffusion of selected
cultivars are yet at work. The cleaning of good cultivars from virus
by tip meristemm) culture is currently done in Barbados for D.
alata and is undertaken by in vitro thermotherapy in Guadeloupe
for D. trifida. No consistent permanent variation has yet been
registered from yam micropropagation (Arnolin, pers. comm.).
So, application of gamma irradiation to vitroplants seems well in-
dicated (Marie, INRA-Montpellier). But true tissue cultures of
diosgenin species have shown organs mixoploidy and the
feasibility of cellular selection (Karanova, Shamina, 1978).
 The next steps should be to renew the tentative of
haplomethod, first by androgenesis (Arnolin, 1976), and to
develop the protoplasts and somatic hybridization techniques.
Several laboratories are seemingly at work in these directions.
 To conclude, when comparing the few lines hallowed by
Coursey in his basic monograph to the breeding of the yams with
the present stage of continuously released selected cultivars, of


PROCEEDINGS of the CARIBBEAN FOOD CROPS SOCIETY-VOL. XX


94