| Summary: | 博士 === 國立中興大學 === 農藝學系 === 90 === Solar Radiation Use Efficiency and Its Discriminant Functions of the Rice Varieties Resealed in Taiwan
Abstract
This study aimed to explore a set of canonical discriminant functions, i.e., linear combinations of certain quantitative traits, that best reveal the difference of solar radiation use efficiency of grain yield (RUEy) among rice varieties. A total of 25 varieties randomly selected from the collection of rice varieties released in Taiwan during the past 60 years were used to conducted field experiments with various nitrogen rates and/or planting densities in the first and the second crop seasons through the years from 1997to1999. Plant height, tiller number, leaf area index , leaf inclination angle, light interception rate, and total shoot dry weight were periodically measured throughout the growing season, and grain yield and its components, i.e., panicle number, spikelet number per panicle, ripen rate and 1000-grain weight were also recorded at harvest. RUEy was calculated as the ratio of grain yield to the total intercepted solar radiation (LIsum), which in turn was the product of the total incoming radiation and the mean light interception rate (MLI).
The MLI value of rice canopy, although significantly different among the test varieties, was unstable across the range of the experiments. Both the estimates of heritability (intra-variety correlation) and genetic gain of MLI were low from a combined analysis on the pooled data. When data were analyzed separately for each of the two crop seasons, however, the estimates of these two genetic parameters of MLI raised remarkably. It revealed that varieties with high MLI value might best be identified under the conditions of applying 180 Kg nitrogen per hectare in the second crop season or applying no nitrogen in the first crop season. As expected, many characteristics of the rice canopy were significantly correlated with MLI. Among them, the LAI at heading, which showed high heritability and genetic gain, was correlated with MLI very closely. This suggested that selection for variety with high MLI could be accomplished by the way of selection for LAI at heading.
When grown with no nitrogen fertilizer or with 180 Kg nitrogen per hectare in the second crop season, the MLI of the test varieties showed a clear decreasing trend over the time the varieties were released. Such a time trend was not detected in the first crop season. Besides, no significant evolution pattern in the mean light interception efficiency (LIE), i.e., the ratio of MLI to LAI, was observed in the range of the experiments. In other words, the older varieties showed higher MLI but not higher LIE.
All the test varieties gave higher values of RUEy in the first crop season than in the second crop season. The varieties released in recent years showed higher RUEy when they were fertilized with more nitrogen. But, the reverse was true for the old varieties. No significant GxE interaction was found in the analysis of variance of RUEy, which indicated that RUEy was a rather stable character of the rice varieties released in Taiwan. However, higher estimates of the heritability and genetic gain of RUEy were obtained when these varieties were fertilized with nitrogen in the second crop season. This suggested an efficient environmental condition for the selection of varieties with high RUEy. No significant correlation was detected between RUEy and other canopy characteristics investigated in this study. In general, the test varieties showed an increasing time trend of RUEy in respect to their breeding era . This trend was absent when these varieties were grown with no nitrogen fertilizer.
To elucidate the varietal variation of grain yield and RUEy, grain yield was expressed as the product of LIsum and RUEy, which in turn was expressed as the product of radiation use efficiency of total dry matter production (RUEtdw) and harvest index (i.e., ratio of grain yield to total dry matter production, HI). The variance of grain yield was mainly accounted for by LIsum when the varieties were fertilized with 90 Kg nitrogen per hectare. However, RUEy took the role in determining the variation of grain yield when these varieties were fertilized with 180 Kg nitrogen per hectare or no nitrogen fertilization. On the other hand, contribution of RUEtdw to the varietal variance of RUEy was much higher than HI across the treatment levels of nitrogen rate. In sum, grain yield of a variety was mainly determined by its RUEy, which in turn was affected greatly by the RUEtdw of the variety when it was fertilized with certain amount of nitrogen.
As a provision to develop the discriminant functions for RUEy, the test varieties were clustered, by a nonhierarchical clustering algorithm, into three to five groups according to their RUEy values evaluated in each of the six treatment combinations of the two crop seasons by the three levels of nitrogen rate. The optimal number of variety group was determined so as to maximize the among-group variance and to minimize the within-group variance of RUEy simultaneously. The result showed that the test varieties could adequately be separated into three groups except the treatments with 90 Kg nitrogen per hectare, which a four- group clustering seemed optimal. A set of two discriminat functions for RUEy were established for each of the six treatment combinations. Owing to their significant ability in identifying the varieties with low RUEy, a breeding program would be much more efficient to improve the RUEy of rice varieties if these discriminant functions were employed.
|
|---|
