Abstract: Rice sheath blight caused by Rhizoctonia solani is one of the most devastative diseases worldwide, which causes serious yield loss under favorable environmental conditions. R. solani has a very wide host range and the rice germplasm with complete resistance to this disease is zezo. Sheath blight resistance is generally controlled by polygene, with a wide variation among rice cultivers. It is necessary to evaluate probably partial resistance to sheath blight from germplasms for rice breeding program. A total of 166 indica hybrid rice combinations collected from 11 provinces in southern China were inoculated by using five isolates with different pathogenicities at seedling stage in the greenhouse. No combination was found to be completely immune or highly resistant, and the resistance to sheath blight varied significantly among different combinations. Average disease index of these combinations were in accordance with normal distribution N (5.27, 0.68²) and ranged from 2.84 to 7.64. The results showed that this inoculation method at seedling stage was simple, rapid, uniform, and allowed to distinguish minor differences in susceptibility, so it can be used for a large-scale screening or primary resistance identification. According to the disease index of each isolate, 166 combinations were classified into five types based on dynamic clustering analysis (DCA), namely resistant (R), moderately resistant (MR), moderately susceptible (MS), susceptible (S) and highly susceptible (HS), accounting for 1.20%, 13.86%, 36.14%, 43.37%, and 5.42%, respectively. The discriminant functions for each type were calculated based on Bayes method, and the accuracy rate for discrimination reached 96.39%. The results indicated that most of the combinations were susceptible. Only two combinations, K you 88 and Zhongyou 9801, were resistant, but should be verified at the adult stage in paddy field. Synthetic disease index (SDI) of all combinations was also calculated by average disease index of the five isolates, the combinations were graded as R, MR, MS, S, and HS based on SDI, accounting for 1.20%, 13.25%, 63.25%, 21.69%, and 0.60%, respectively. There was significant correlation between the results of DCA and SDI (r = 0.81, P < 0.01), showing that both methods can be used for evaluating the disease resistance. But the former method is suitable for seedling screening under the uniform growth conditions; while the latter is independent of time, place and batch of the identification, and more versatile than the former. Only 26 combinations with resistance or moderate resistance to sheath blight were gained by DCA and SDI, the genetic distances ranged from 0.04 to 0.71. Twenty six combinations were divided into two groups by UPGMA cluster analysis, and 22 out of them were grouped into a group with the narrow genetic background at genetic distance of 0.36. There was a close relationship among the parents of the 26 combinations according to the pedigree tracing. Therefore, the practical strategy for rice resistance breeding to sheath blight is further to broaden genetic bases of germplasm, pyramid quantitative resistance genes, construct hybrid combinations with parents having desirable traits and resistance to sheath blight.

Key words: Indica hybrid rice, Sheath blight, Resistance evaluation, Dynamic clustering analysis (DCA), Synthetic disease index (SDI)