| Genetic mechanism and a rapid testing methods of maize quality are the basic of maize quality improvement. The objectives of this study were (1) to evaluate the feasibility and methods of measuring quality of maize kernel with near infrared reflectance spectroscopy (NIRS) using 82 samples of maize hybrids or crosses. (2)to analyze the genetic effects of grain quality for 13 inbred lines and their 42 crosses of F1, F2 generation in a 6 X 7 incomplete diallel mating design. A genetic model proposed for analysis of quantitative traits by Zhu Jun was used to investigate the relationship of parents with their offspring, and behavior of quality traits in F2 .The results were as follows:1. Different spectral data preprocessing methods have influences on the performance of model established. Calibration equations for quantitatively determination of crude protein, crude starch, crude fat and lysine content were developed by partial least square regression (PLS) of chemical values on NIRS data and tested through both cross and external validation. The result showed that the data protreatments of vector normiliztion, first deravitive with multiplicative scattering correction .constant offset elimination and first deravitive with multiplicative scattering correction achieved the best effects on protein content and starch content, respectively, and gave the ditermination coefficients of cross valadition 88.39%,63.55%,40.39% and 55.29% and the root mean square error of cross valadition 0.312,1.12,0.0228 and 0.466. In addition, 50 progenies of F1,F2 were verified to further evaluate their performance.Resulted indicated that the correlations between the predicted NIRS results and those obtained from traditional chemical methods of protein and starch content were very strong,with the correlation coefficients generally above 0.9, whereas lower coefficients were obtained with models of oil and lysine content from reflectance spetra.2. The results of measuring quality showed that average content of protein,starch, oil and lysine in 13 inbred lines were 10.50%, 70.96%,4.29% and 0.35%,respectively. Variance range were 9.08%~12.50% , 67.83%~74.10%,4.13%~4.54% and 0.31%~ 0.40% .3. The performance of quality traits in F2 generation is complicated and showed much difference between traits and crosses. The heterosis of protein content over parental means, high parent and low parent were all negative In F2 generation, which indicated that protein content of F2 was close to low parent and negative heterosis. Starch content consistently showed positive heterosis with positive heterosis over low and high parents as well as parental means. For oil and lysine content, the heterosis over parental means, high parent... |