Best linear unbiased prediction as a method of estimating breeding value of sugarcane parental clones

Stringer, Joanne Kay. (1993). Best linear unbiased prediction as a method of estimating breeding value of sugarcane parental clones Master's Thesis, School of Land, Crop and Food Sciences, The University of Queensland.

       
Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
THE8671.pdf Full text application/pdf 3.00MB 1
Author Stringer, Joanne Kay.
Thesis Title Best linear unbiased prediction as a method of estimating breeding value of sugarcane parental clones
School, Centre or Institute School of Land, Crop and Food Sciences
Institution The University of Queensland
Publication date 1993
Thesis type Master's Thesis
Total pages 59
Language eng
Subjects 0701 Agriculture, Land and Farm Management
Formatted abstract The Bureau of Sugar Experiment Stations (BSES) use an empirically based index to assess the breeding potential of sugarcane (Saccharum spp.) parents and this may take many years to provide reliable estimates. A rapid and efficient method of assessing the breeding value of parents in early stage families is needed to increase the rate of population improvement.

Early stage family selection trials are typically highly unbalanced and so data analysis cannot be undertaken by ordinary least squares approaches. Statistical techniques such as Best Linear Unbiased Prediction (BLUP) were specifically developed to allow prediction of breeding values from messy dairy cattle data sets. Although the theory could be adapted to other breeding programs, there have been few applications in plants. Hence, it was necessary to review the theory of BLUP to determine its suitability for application to sugarcane data. A discussion on BLUP necessitates a discussion on the related technique of Best Linear Prediction (BLP). These techniques are compared to several fixed effects approaches and the advantages of BLUP and BLP are discussed.

Once the suitability of BLUP for analysing sugarcane data was established, it was necessary to determine if the predicted breeding values based on BLUP differed from current BSES methods. The current empirical method used by BSES is outlined and reasons for investigating alternative methods are given. Family selection data from the southern Queensland breeding program were then used to study three univariate methods of predicting breeding value: the family mean, the BSES empirical method and the Best Linear Unbiased Prediction (BLUP).

Initially data from the 1989 Southern Queensland series were analysed assuming a North Carolina design I. BLUP estimates were correlated with the arithmetic mean and all correlations were high regardless of the narrow sense heritabilities.

In subsequent analyses data were analysed assuming an incomplete diallel mating design. Using data from the 1990-1991 series, BLUP estimates of breeding value for 234 parents were calculated. For cane yield, sucrose content and net merit grade (selection index) the BLUP estimates for each parent were used to predict the performance of 79 families in the 1992 series. Predicted performance from the BLUP estimates was correlated with performance in the 1992 series. The correlations were strong, particularly for relative sucrose content (0.64).

For net merit grade only, breeding values based on the empirical formula were correlated with the BLUP estimates for the 79 jaJ1lilies in the 1992 season. The correlation was moderate (0.48).

The BLUP analyses provided BSES plant breeders with a simple and rapid means of combining data from a wide range of sources to identify superior parents. BLUP appears to be as effective as the BSES el1zpirical method or family mean for identifying superior clones. Given that the empirical method incorporates information from up to ten years whereas the BLUP results were based on few years data, it is probable that BLUPs will increase the rate of population improvement which would have major benefits for the plant breeding program and the sugar industry.

As BLUP can accommodate complicating factors such as inbreeding, it was important to quantify the level present in BSES family selection data. Inbreeding coefficients for all families except three were less than 0.3. The level of inbreeding was correlated with agronomic performance. Correlations were negligible although future studies on inbreeding are suggested.
Keyword Sugarcane -- Breeding -- Mathematical models
Additional Notes

Variant title: Best linear unbiased prediction applied to sugarcane data.

 
Citation counts: Google Scholar Search Google Scholar
Access Statistics: 171 Abstract Views, 1 File Downloads  -  Detailed Statistics
Created: Fri, 21 Oct 2011, 15:41:36 EST by Miss Wendy James on behalf of Scholarly Publishing and Digitisation Service