Abstract: Statistical method of multivariate genetic analysis were used to study relationships among 18 sorghum characters.The result showed that phenotypic coeffecients of correlation of seed weight/ head with most of other characters,1000-seed weight with flag leaf width,seeds/ head with diameters of head,peduncle and stalk,seeds/ head with primary and secondary branches in head,were significant at 5% or 1% level.The genetic coefficients of correlation between characters demonst rated the same trends as the phenotypic ones did,but were generally lager than the latter.Four factors(Fs)were loaded with all the 18 characters through principle component analysis and factor analysis,F1,called head factor,with 8 characters related mostly to head structure,F2,called growth factor,with 6 characters related to days from seedling emergence to heading,flowering and maturing,and stalk diameter,F3,called flag leaf factor with 3 characters related to flag leaf,and F4 with only one character,head length,called head length factor.By cluster analysis 18 characters were clustered into 3 main groups(G1,G2 and G3),and group 3(G3)could be further divided into 2 subgroups(G3-1,G3-2),each group having the same characters as those of Fsloaded.The results of cluster analysis and factor analysis were completely the same.The relationships among the quantitative characters studied reflected the properties of / one factor with more effect and mult-i factor with the same effect and the coordinated variations in correlated characters.

Key words: Sorghum, Principle component, Factor analysis, Cluster analysis

CLC Number: 

• S514.03