PATH ANALYSES OF YIELD AND YIELD-RELATED TRAITS OF 15 DIVERSE RICE GENOTYPES

Literature on the path analyses of grain yield and at least 14 yield-r elated traits in a path diagram that is organized with at least second -order variables has been lacking. The objectives of this study were t o obtain and interpret information on the nature of interrelationships between first-, second-, and third-order yield-related traits and ric e (Oryza sativa L.) grain yield. Fifteen rice genotypes were used in t his study to represent the combinations of low and high levels of four traits that were identified as important yield determinants - maximum number of tillers, grain size, panicle node number, and panicle size. 'Lemont' and 'Teqing' were two of these genotypes. The remaining geno types were Fg lines from a Lemont x Teqing cross. Field experiments we re conducted during the 1994 and 1995 cropping seasons at the Texas A& M University Agricultural Research and Extension Center near Beaumont, TX. The 1994 path coefficient (p) of panicle weight on grain yield (p = 0.72; r(2) = 0.93) was used to predict the 1995 grain yield (r(2) = 0.90). Based on a path analysis of the combined 1994 and 1995 data, t he following traits had positive path coefficients on grain yield: pan icle weight (p = 0.84), number of filled grains per panicle (p = 0.67) , panicle density (p = 0.52), maximum tiller density (p = 0.34), numbe r of spikelets per panicle (p = 0.34), and 100-grain weight (p = 0.23) . The panicle node number has a negative path coefficient on grain yie ld (p = -0.23). These results may be useful to rice breeders for the i ndirect selection of grain yield during the early segregating generati ons when yield tests are not yet being conducted.