No abstract

A breeding program aimed at improving the seed yield in bean for processing was started. Six parents divergent in a range of agronomical characters were crossed according to diallel mating design. Cluster analysis performed on the composite cross population evidenced divergence between Fl hybrids and to maximize genetic recombination, the most different Fl hybrids were intercrossed. A Factor analysis was performed on 7 morphological and yield-determining traits of 15 hybrids Fl and parents. Three Factors, representing patterns of variables interpreted as productivity per plant, growth factor, and branchingproductivity, were extracted. These results evidenced the inverse relationship between seed yield components (seeds/pod, pods/plant, and hundred seed weight): i.e. the selection for either variable is detrimental to the other. It may be possible to break this association if the plant construct assures high sink to satisfy all plant requirements: tall and large plants bearing numerous nodes, leaves and reproductive structures. Moreover, the selection practiced simultaneously on the yield components positively affected the seed yield trait. An index of selection and response to selection were estimated on the segregant populations. The realized gain in seed/yield reached after one cycle of selection (Cl) was 6.5 g, about 20% over the base population (CO). The expected gain from selection based on the superior 5% plants in C2 families was 25.5% and approached the realized gain. The result of this paper gives evidence to support that the seed yield can be improved by an adequate selection intensity and selection index can be very useful when it is faced with the improvement of several traits at the same time.