Department of Plant Breeding and Genetics, J. N. Agricultural University, Jabalpur 482004.
* Present Address: International Rice Research Institute, P. O. Box 933, Manila, Philippines.
Beyond a certain panicle number perunitarea, further increase in rice yield would depend on increase in mean panicle yield. To identify panicle yield characters 18 elite, early maturing, semidwarf rice varieties were grown under high fertility (N100:P50:K50) and low fertility (N50:P25:K25) levels. Results from stepwise regression analysis showed that the number of spikelets per panicle was the most important determinant of panicle yield which accounted 65% and 45% of the total variation in panicle yield under high and low fertility levels, respectively. The number of sterile spikelets/primary branch and 1000-grain weight were next in order of importance. Together, these three characters accounted 82% and 86% of the total variation in panicle yield at high and low fertility levels, respectively. Thus, if selection for panicle yield is to be based on spikelet number/panicle alone, high fertility should be preferred. However, for maximum gains, selection should be based on increased number of spikelets/panicle, reduced number of sterile spikelets/primary branch, and higher 1000-grain weight. If selection for panicle yield is based on all these three determinants of panicle yield, the fertility levels would make little difference in the effectiveness of selection.
Correlation, fertility, panide yield, rice, stepwise regression