Evaluation of parent selection methods for yield improvement of cowpea (Vigna unguiculata (L.) Walp.)

Abstract

The mean aim of this study was to identify an effective method of predicting cross-potential in respect of yield improvement of the cowpea (Vigna unguiculata (L.) Walp.) through hybridization and subsequent selection. Three prediction tests based on data collected from experiments with ten cowpea lines and the F1 and F2 generations of all possible crosses between these lines were evaluated. The three prediction tests were based on: i Yields of prospective parents. ii. General combining abilities estimated from F1 and parental data iii. Frequencies of superior plants in F2 populations Each of the three procedures was used to identify the crosses with the lowest or highest yield potential, i.e. those crosses most likely to produce the greatest number of lines with yields higher or lower than those of either parent. Crosses selected on the basis of these criteria were evaluated in the following growing season on the basis of biometrical analysis of F2 and backcross populations and the performance of the respective F3 and F4 generations. Six different crosses were selected from the 45 tested on the basis of the three prediction tests. Analyses of F2 and backeross populations from each cross provided estimates of [d] the difference in parental means and 1/2D the additive genetic component which, in turn, were used to predict the frequency of transgressive inbred lines obtained by single-seed descent from the F2. Both crosses selected on the basis of F2 data were expected to produce a much higher frequency of transgressive lines than those selected on the basis of parental yields. The high potential cross selected on the basis of general combining ability was expected to produce a higher frequency of transgressive lines than the corresponding low yield potential cross. Replicated trails of F3 bulks of the selected crosses showed that high yield potential crosses were significantly higher yielding than low yield potential crosses for all prediction criteria. In a trial with 22F4 lines of each of the two crosses selected on the basis of parental yields and frequency of high yielding F2 plants, the high yield potential cross in each case had a greater mean and variance for seed yield than the corresponding low yield potential cross. It was concluded that parental yields provide a sound basis for an initial screening of prospective parents. This screening should identify a high yielding line which can be crossed with a number of contrasting lines to produce F2 populations for the final screening phase.