Not Available

Not AvailableThere is evidence that high-tillering, small-panicled pearl millet landraces are better adapted to the severe, unpredictable drought stress of the arid zones ofNWIndia than are low-tillering, large-panicled modern varieties, which significantly outyield the landraces under favourable conditions. In this paper, we analyse the relationship of arid zone adaptation with the expression, under optimum conditions, of yield components that determine either the potential sink size or the ability to realise this potential. The objective is to test whether selection under optimal conditions for yield components can identify germplasm with adaptation to arid zones in NW India, as this could potentially improve the efficiency of pearl millet improvement programs targeting arid zones. We use data from an evaluation of over 100 landraces from NW India, conducted for two seasons under both severely drought-stressed and favourable conditions in northwest and south India. Trial average grain yields ranged from 14 g m 2 to 182 g m 2. The landraces were grouped into clusters, based on their phenology and yield components as measured under well-watered conditions in south India. In environments without pre-flowering drought stress, tillering type had no effect on potential sink size, but low-tillering, large-panicled landraces yielded significantly more grain, as they were better able to realise their potential sink size. By contrast, in two low-yielding arid zone environments which experienced pre-anthesis drought stress, low-tillering, large-panicled landraces yielded significantly less grain than high-tillering ones with comparable phenology, because of both a reduced potential sink size and a reduced ability to realise this potential. The results indicate that the high grain yield of low-tillering, large-panicled landraces under favourable conditions is due to improved partitioning, rather than resource capture. However, under severe stress with restricted assimilate supply, high-tillering, small-panicled landraces are better able to produce a reproductive sink than are large-panicled ones. Selection under optimum conditions for yield components representing a resource allocation pattern favouring high yield under severe drought stress, combined with a capability to increase grain yield if assimilates are available, was more effective than direct selection for grain yield in identifying germplasm adapted to arid zones. Incorporating such selection in early generations of variety testing could reduce the reliance on random stress environments. This should improve the efficiency of millet breeding programs targeting arid zones.Not Availabl