Selection of promising genotypes based on path and cluster analyses

Publisher Website
Google Scholar
Abstract
SUMMARY The objective of the present paper was to propose a statistical approach to support selection of the most promising genotypes in a breeding programme. The approach is based on applying two state-of-the-art statistical methodologies, likelihood-based path analysis and model-based cluster analysis. The first method is applied to find a causal mechanism lying behind a biological process of development of final crop yield. These results are then used for weighting traits to be used in cluster analysis, which helps select genotypes possessing a desirable level of yield and yield-contributing traits. An application of the approach is presented for a 2-year study on 22 grasspea genotypes, two cultivars (Derek and Krab) and 20 mutants from those cultivars. Seed yield/plant and seven yield-related traits were studied. Among these, plant height, number of branches/plant, pod length and number of seeds/plant determined seed yield; number of pods/plant influenced seed yield only for 2002. These results were used for appropriate weighting in cluster analysis, which indicated that cultivar Krab and its two mutants, K3 and K64, had the best level of the traits and were the most stable genotypes.