The experiments revealed that direct selection for grain yield
under stress is effective under lowland stress and confirmed the
earlier findings of its effectiveness under upland stress. Indirect
selection for grain yield under non-stress was less effective as
compared to the direct selection for grain yield under stress for
both lowland and upland drought stress. None of the secondary
traits studied here were more effective for selection under stress
than grain yield per se. The study indicated a positive response
to selection under severe drought stress that reduced the mean
yield of the stress trial by more than 65% compared with the
non-stress yield in both lowland and upland situations.
Finally, the study has also indicated that it is possible to
combine drought tolerance with high-yield potential. Crosses
involving drought-tolerant donors yielded a much higher
frequency of drought-tolerant progenies than did crosses
between two high-yielding but drought-susceptible lines. Some
lowland-adapted lines produced from crosses between donors of
tolerance and lines with high-yield potential combined yield
levels as high as those of elite cultivars with yield levels under
stress as high as those of the most tolerant donors. These results
point the way to an effective strategy for the development of
cultivars for drought-prone areas. The principal elements of this
strategy are: (i) crosses between tolerant parents and lines with
high-yield potential; (ii) direct selection of progeny in replicated
trials for yield under optimal conditions and managed stress; and
(iii) achievement of stress levels in managed-stress screens that
reduce yield by 65–85% relative to unstressed controls.