A comparative analysis of wheat adaptation across international environments using stochastic dominance and pattern analysis

Abstract

A set of ISWYN trial data were analysed using the procedure of stochastic dominance. Wheat lines that are stochastically dominant (“risk-efficient”) exhibit a particular type of environmental adaptation (relatively good performance in poor environments) which makes them suitable for risk-averse farmers. The results of the analysis lend support to the hypothesis that recently bred lines are more risk efficient and thus more suitable for peasant farmers than lines which appeared in the early ISWYNs. Risk-efficiency implies good adaptations to poor environments, although some lines may also perform well in good environments. There was a tendency for risk-efficient lines to be high mean yielders, but many exceptions existed. A combination of high mean yield and risk-efficiency is a virtual guarantee of good performance across all environments. Consequently, stochastic dominance analysis for risk-efficiency would seem to be a useful adjunct to simple measures of mean yield or other more sophisticated analytical methods. Comparisons between the results of stochastic dominance analysis and pattern analysis of the same data set indicated a general tendency for groups formed by pattern analysis to contain lines of the same degree of stochastic dominance. This is not entirely unexpected, since both forms of analysis group together lines whose performance across environments is similar, even though stochastic dominance analysis concentrates on the poor environments. However, there is no guarantee that stochastically dominant (risk-efficient) lines also perform relatively well in good environments and some risk-efficient lines were identified that did not. These were largely from the lower yielding groups.