Catalog Home Page

Data integration for decision making in wheat breeding

Diepeveen, Dean (2011) Data integration for decision making in wheat breeding. PhD thesis, Murdoch University.

[img]
Preview
PDF - Front Pages
Download (174kB) | Preview
    [img]
    Preview
    PDF - Whole Thesis
    Download (3762kB) | Preview

      Abstract

      Plant breeding is a production process requiring the creation of germplasm through taking existing successful cultivars and crossing them with new parental lines with agronomic and quality attributes of interest. After crossing, F2 generations generally display all possible combinations between the parental lines. The process from this step is to identify elite crossbred lines and backcross these several times to the parental lines in order to generate new elite lines that are predominately equivalent to the cultivar but with specific novel and desirable attributes present.

      Plant breeding continually requires judgements to identify elite plant germplasm containing traits that maximise plant performance. These judgements are often made using incomplete information resulting from the greater complexity in modern plant breeding decision making. Judgements can be improved through the utilisation of new technologies and a stronger scientific basis. This thesis uses decision and information management processes to contribute to:

      • Pioneering the application of unbalanced datasets to wheat breeding. The methodologies were derived from tree and animal breeding experience and successfully applied to data sets from a wheat breeding program.
      • Providing the first integration of molecular data into a decision-matrix framework.
      • Building on the molecular integration in output-2 by establishing a more sophisticated integration of complex NIR spectral data with molecular data.
      • Providing inputs into decision matrices for breeding using the outputs discussed above.

      This thesis establishes the methodology to make use of new technologies to use unbalanced datasets with decision matrix methodology to make better decisions. This thesis has utilised multivariate methodologies more broadly to include complex data such as NIR fingerprint to differentiate flour samples between controls and breeding germplasm. These differences appear to be related to genetic factors as demonstrated after variability relating to the environment had been removed. This thesis first reviews the literature and then addresses this breeding processes through the use of decision and information management processes, and makes significant contributions in using these methodologies.

      Publication Type: Thesis (PhD)
      Murdoch Affiliation: School of Information Technology
      Supervisor: Appels, Rudi and Bellgard, Matthew
      URI: http://researchrepository.murdoch.edu.au/id/eprint/5808
      Item Control Page

      Downloads

      Downloads per month over past year