Breeding maize for silage.

摘要

In North America maize (Zea mays L.) is grown primarily for grain however, farmers have the option of harvesting the maize crop for either grain or forage, which when ensiled is referred to as silage. The first objective of the research described in this thesis was to determine combining abilities for quality traits of a representative sample of maize inbred lines, and relate these combining abilities to inbred per se evaluations.; There were large and significant differences in early and late inbred trials for forage quality traits. Cross and GCA sources of variation were significant for forage quality traits in both early and late hybrid maturity groups. General combining ability was the primary source of variation for forage quality traits. Phenotypic correlations between the GCA effects and inbred line per se performance were significant for permanganate lignin and in vitro true digestibility in both maturity groups. Suggesting that it would be feasible to conduct preliminary evaluation at the inbred level for these traits.; The second objective of the research described in this thesis was to investigate error and bias associated with prediction of digestibility traits in maize with near infrared reflectance spectroscopy (NIRS). Archived NIR spectra with their associated conventional lab analysis values were obtained from a maize hybrid silage evaluation program. These spectra were used in two ways, first as a database to compare with spectra from unknown samples and identify those samples which are similar, which can be used to estimate the conventional lab analysis values of the unknown. And secondly to develop broad-based prediction equations. The conventional lab analysis values were available for the unknown samples.; Identification of similar samples by matching of spectra was not possible because of the differences between the database and unknown samples spectra. A principal component analysis of the database and unknown spectra suggested that the differences could be attributed to differences in particle size. Because of the differences in spectra considerable error and bias existed in the digestibility traits estimated with the broad-based equations. However highly significant Spearman rank correlations existed for the conventional lab analysis values and broad-based prediction values.