Variation and yield prediction analyses of some morphological traits in six Kenyan landrace populations of spiderflower (gynandropsis gynandra (L) briq
The objective of this study was to investigate the presence of heritable variation in Kenyan landrace populations of spiderflower (Gynandropsis gynandra (L) Briq) and assess the utility of yield components in the improvement of yield and duration of .harvesting through the c'~ponent breeding approach. Genetic componenCS of variat10n and expected gain from selection as a percentage of the population mean were estimated from an S famdly mating design. The experiment was conducted during the short rains of 1988 and the long rains of 1989 at the Field Station of the University of Nairobi, Kabete Campus and the National Horticultural Research Station, Thika. Twenty S families from each population were planted in a three-replicate compact family block design at both sites. The data was collected for days to flowering, plant hei9ht, number of primary leaves, leaf length, leaf breadth, fresh leaf weight and dry leaf weight. The analysis was based on family means. The results indicated that there was significant variation among populations for leaf length and dry leaf weight at Thika and number of primary leaves at Kabete. This reflected a variation in the yield potential of the populations. Significant variation was detected within population E for days to flowering and population M for both fresh leaf yield and dry leaf yield. Yield prediction studies revealed that duration of harvesting could be prolonged by selecting late flowering genotypes in population E. Dry leaf yield and fresh yield could be improved by direct selection in population M. There was no inherent variation for any of the traits studied in populations P, Q, D and I. Therefore.the improvement of these population would necessitate the creation and maintenance of genetic variability by incorporation of genes from other populations and intermating within the populations.