| Doubled haploid (DH) technology has become one of the most important approaches in maize commercial breeding for its advantages of complete homozygosity and short breeding cycle length. Two key issues in this technology are inducer and efficient screening system to identify haploid kernels from the hybrid kernels, respectively. With the fierce competition in the global seed industry, DH technology is widely used by many institutes and companies, and the idea of engineering breeding would lead the rapid development of corn industry. Therefore, to change the traditional manual identification method to automatic identification is the inevitable requirement of haploid identification technique, which urgently call for the guidance of automated identification theory and the realization of automation. Also, it is needed to develop new inducer lines supporting automatic identification, which will provide efficient technical services for the DH technology.The aims of this study were:(1) to construct the haploid discrimination model based onoil content;(2) to develop automatic nuclear magnetic resonance screening system for haploid kernels in maize;(3) to explore new discrimination system of maize haploid seeds from hybrid seeds using Vis spectroscopy and support vector machine method;(4) to breed and select new haploid inducer lines. The major results were as the following:1) Eighty germplasms from five heterotic groups were induced by high oil inducer CAUHOI. The heritability among nine different populations showed high value, less affected by environmental factors.The xenia effect of CAUHOI reached more than0.55. It is feasible that using pattern recognition, linear discriminant function of minimum square error criterion function to establish a monoploid identification of oil rate threshold. Puting forward that flexible adopting three kinds of OC threshold distinguish haploid and hybrid diploid seeds.2) The automatic screening system for maize haploid kernel identification based on oil xenia effect and nuclearmagnetic resonance (NMR) was developed to meet the demand of large-scale application in maize haploid breeding andresearch. The screening system was comprised of a seed feeding module, weighting module, a measurement module of oil content and ascreening module. The results showed that the mean accuracy and speed of this system reach91.4%and4second per single kernel. Even with the shortersignal acquisition time, the accuracy did not show a significant decline. The automatic screening system can be used inpractical application in rapid selection of maize haploid seeds.3) It was established a method with which to distinguish haploid from hybrid maize seeds using a single kernel and the visible spectrum. The original spectral data were preprocessed by smoothing and vector normalization. The feature wavelengths were selected by KS test. The discrimination models were built based on SVM method with the data through PCA. In each time of validation, one of genotypes seeds were used as the test set, and the remaining genotypes samples were reserved as the modeling set for cross-validation. The average Accuracy for modeling cross-validation reached as high as92.06%.An automatic rapid-screening system for haploid maize seeds based on this method will probably be established to meet large-scale DH breeding demands.4) Two populations of cross2262×CAUH-P and cross2267xCAUH-P were constructed to develop new haploid inducer lines. During the selling pedigree selection, the important parameters including Rl-nj marker, purple plumule marker, oil content and HIR were used for deciding the targets of plants. The marker-assisted selection of qhirl and qhir8regions were utilized for validating the homozygosity of the large effect QTL of in vivo induction of maternal haploid. Finally two candidate inducer lines were developed, with OC of average8.39%and8.37%, HIR of average11.38%and10.68, respectively, and superior agromomic performance. |
PDF Full Text Request