MSI99P3059 | |
Badaloo, G. Kearsey, M. J. Domaingue, R. Julien, M. H. R. | |
The use of cross prediction methodologies in the sugar cane breeding programme in Mauritius | |
Abstracts - 4th ISSCT Breeding and Germplasm Workshop: efficiency of sugr cane breeding, 18-22 May 1988, Hotel Coralia Mont Choisy, Mauritius | |
book chapter | |
1998 | |
p. 14 | |
Réduit: Mauritius Sugar Industry Research Institute | |
En | |
En | |
Three univariate and two bivariate statistical methods to predict the potential of families to produce elite progeny were studied, using 20 random biparental crosses involving 29 parents, using the normal probability integral methodology to improve the efficiency of the MSIRI sugar cane (Saccharum hybrids) breeding programme. The studies were carried out over two contrasting environments namely, a subhumid, irrigated area and a superhumid marginal area, over the first three selection stages, i.e. the seedling, the single stool (first clona! generation) and the single line stage (simulated second clonaI generation). The univariate and bivariate prediction methads of a number of agro-morphological characters of economic importance, such as Brix, cane yield, stalk diameter, stalk number, height and weight, leaf-stripping ability and breeder's preference were investigated. ln general, the predicted numbers (PN) and the actual abserved numbers (ON) of genotypes transgressing the set target agree reasonably weIl far most traits at the three selection stages at bath sites. There were highly significant positive correlations between the predicted numbers and observed numbers for most traits at bath sites. For mast traits across selection stages and across both sites, there were highly significant positive correlations between the three statistics, (a) the predicted proportion (PROB), (b) the.observed proportion (OBS) and (c)the family mean (MEAN) in terms of the proportion of elite genotypes that can be derived from the 20 families. The highly significant positive correlations between predicted and observed rankings of crosses canfirmed that univariate cross prediction methodologies are robust enough to unambigously discriminate between crosses in terms of their likelihood of producing superior genotypes that exceed the set targets. Univariate normal probabilities (PROB), based on the normal probability integral, estimated with family means and phenotypic standard deviations generally demonstrated quite satisfactory repeatability among generations and sites. For joint predictions, the sum of ranks (RANK) and the observed frequency of genotypes that exceed the set targets (FREQ) showed highly significant correlations at each generation and site. Results show that very good estimates of statistical parameters can be obtained at the preliminary stages of selection and that predictions could be made on a range of traits of economic importance to the breeder for the very early identification of crosses with the greatest potential of producing superior genotypes | |
SUGARCANE CROSSING CROSS PREDICTION BREEDING METHODOLOGY BREEDING PROGRAMME STATISTICAL METHODS | |
MAURITIUS | |
Cane breeding and genetics | |
Breeding: crossing | |
1998-06-29 | |
En | |
LIB | |
CAT | |
1 | |
PB |