Previous PageTable Of ContentsNext Page

Limitations of the use of drought susceptibility indices based on relativeyield for selection of drought adapted genotypes

M. Cooper1, D.R. Woodruff2 and D.E. Byth3

1 Department of Agriculture, University of Queensland, St Lucia Q 4067
Queensland Wheat Research Institute, P 0 Box 2282, Toowoomba Q 4350
Queensland Agricultural College, Lawes Q 4343

Selection for adaptation to stress environments is directly or indirectly practised withim breeding programs. Where selection is practised solely for specific stress emvironments, identification of adapted individuals is relatively easy amd based om genotype comparisons withim the characterised stress enviromment. However where adaptation to undefined stress is sought, identification of individuals is more confusing. A number of workers have used a drought susceptibility index based on relative yields, using non.stress environments as a basis of comparison, in an attempt to overcome this problem. Two limitations of this approach are outlined.


Forty.nine wheat genotypes were growm under an irrigated and rainfed treatment at Brookstead, Queensland during 1986. Grain yield data is considered here. A drought susceptibility factor S was calculated for each genotype following the equation1:-

YD = YP(1 -S.D)

where YD and Ye are genotype yields under stress and non.stress conditions respectively and D is defined as 1.0 minus the average relative yield (YD/YP) over all genotypes. The relatiomships between S, YD and Ye were assessed by phenotypic correlation. While this is a correlation between non.independent variates the correlation coefficients are useful for visualising the implications of selection for S in terms of indirect selection on YD and YP.

Results and discussiom

The mean yield of the irrigated trial was 1.98 times that of the rainfed. Sigmificant (P <0.05) genotype and genotype by enviromment (GxE) interaction components of variance were identified and substantial re.ranking of genotypes for yield performamce occurred between the environments. Genotype grain yield under stress was not related to yield under mon.stress conditions (P >0.05) (Table 1). However yield withim both environments was related to S (P < 0.05), YD negatively and Ye positively (Table 1). Therefore selection for low S will improve yield under drought but reduce yield under irrigated conditions, whereas selection for graim yield per se within individual environments will not impose indirect selection within the other environment. In cases such as this where substantial GxE interaction exists and gives rise to re.ranking of genotypes the drought susceptibility indices are of limited value for selection purposes. Informatiom on the relative performance of genotypes within the environments is lost by calculatiom of S. A second problem occurs where variation in genotypic yield performance im the non.stress emvironment (Ye) is greater than that in the stress environment, a situatiom often emcoumtered. Under these circumstances variation in Ye will dominate gemotypic variation in S. Consequently S will be strongly related to Ye and will not represent differences in adaptation to the stress environment. As an example, the three genotypes in table 2 did not differ in YD but differed greatly in Ye; this gave rise to differences in S. This variation in S is solely a function of variation in Y. Selection for reduced S will again reduce Y. Therefore care must be taken if genotype stress adaptation is to be characterised by this index for genotype selection purposes. An alternative approach to this problem is to use a classification technique to summarise the patterns of adaptation expressed by the genotypes2.

Table 1. Phenotypic correlations between YD, YP, and S in an irrigated and rainfed trial comparison

Table 2. Grain yield (t/ha) and S values for three genotypes under rainfed and irrigated conditions

1. Fischer R.A. and Maurer R. (1978). Aust. J. Agric. Res. 29, 897.912

2. Cooper M., Woodruff D.R. and Byth D.E. (1989). An alternative approach for characterising genotype drought adaptation for selection purposes. This volume. 632

Previous PageTop Of PageNext Page