Breeding for Grain Yield Stability Using Full-Sib Family Recurrent Selection in Sorghum
- M. Chisi,
- P. J. Bramel-Cox ,
- M. D. Witt,
- M. M. Claassen and
- D. J. Andrews
- M ount Makulu Res. Stn, P/B7, Chilanga, Zambia
D ep. of Agronomy, Throckmorton Hall, Kansas State Univ., Manhattan, KS 66506
S outhwest Kansas Res. Ext. Ctr., Kansas State Univ., Garden City, KS 67846
H arvey County Exp. Field, Kansas State Univ., Hesston, KS 67062
D ep. of Agron., Univ. of Nebraska, Lincoln, NE 68583
Sorghum [Sorghum bicolor (L.) Moench] is generally grown unpredictable environments, where the genotype × environment (GE) interaction is large. Recurrent selection can be used to select genotypes adapted to all these environments. This study evaluated the potential gain from selection of full-sib families for grain yield stability across diverse environments using various selection criteria. These criteria were mean productivity, a rank summation index, a selection index, and scores from the principal component analysis of the G × E matrix. One-hundred full-sib families were derived from a broad-based population of sorghum, KP9B. The full-sib families were evaluated at three locations, Garden City and Hesston, KS, and Mead, NE, in 1991 and 1992. The experiment involved two replications of a block-in-replication design with five blocks of 20 entries each. The interaction of the full-sib families with the individual environments was high. The use of mean productivity, rank summation, and selection index using a subset of the environments to select for genotypes with a consistent, high yield across the environments or stability was predicted to be ineffective. The GE interaction matrix was partitioned into three significant (P < 0.05) principal components. Cluster analysis of the genotype's three principal component scores grouped the full-sib families into five groups that differed in their response to the six environments as characterized by differences in their mean standardized grain yield and bloom date in each individual environment and overall. This technique described the types of full-sib family responses to the test environments and identified a yield response that could be considered as stable.
Copyright © . .