New approach for clustering breeding genotypes using production variables, yield losses, and a double-digit disease scale
1998
Franco, J. | Duveiller, E. | Crossa, J. | Garcia A., I.
Spot blotch resistance was scored in an non-replicated set of 198bread wheat entries in Poza Rica, Mexico. A 00-99 double-digit scale (DD) was used which can be considered as two separate discrete variables with different disease level categories (O, 1, . . ., 9). Continuous production variables were also measured: yield (YLD), 1000-grain weight (TGW), growth stage (GS; Zadoks' decimal code), and percent YLD and TGW loss (after comparison with a fungicide-protected plot in the same field). Genotypes were grouped according to three approaches: (1) using only DD data, (2) using only continuous variables GS, YLD, TGW, and percent loss, and (3) using both types of variables. In the first approach, a cluster analysis was conducted using Ward's minimum variance within groups. In the second method, genotypes were clustered based on the normal mixture method sequential to Ward. The third and new approach used the homogeneous conditional Gaussian (modified) method sequential to Ward. Classifications produced with the three methods were compared in order to choose the best strategy for selecting genotypes based on a simultaneous analysis of discrete disease data, yield, yield loss, and growth stage. In contrast to the two other methods, the third approach presented a balanced distribution of genotypes for both types of variables (discrete and continuous), allowing the identification of three groups of entries. (1) a group (G2) with low values for both disease digits, the highest average value for YLD and TGW under disease pressure, and the lowest average YLD and TGW losses; (2) a group (G4) with intermediate disease scores and the highest average YLD and TWG with fungicide protection; and (3) a group (G6) with the highest disease level, the lowest average YLD and TGW under disease pressure, and the highest average YLD and TGW losses.
Показать больше [+] Меньше [-]