Spatial variation is often encountered when large scale field trials are conducted which can result in biased estimation or prediction of treatment (i.e. genotype) values. An effective removal of spatial variation is ...Spatial variation is often encountered when large scale field trials are conducted which can result in biased estimation or prediction of treatment (i.e. genotype) values. An effective removal of spatial variation is needed to ensure unbiased estimation or prediction and thus increase the accuracy of field data evaluation. A moving grid adjustment (MGA) method, which was proposed by Technow, was evaluated through Monte Carlo simulation for its statistical properties regarding field spatial variation control. Our simulation results showed that the MGA method can effectively account for field spatial variation if it does exist;however, this method will not change phenotype results if field spatial variation does not exist. The MGA method was applied to a large-scale cotton field trial data set with two representative agronomic traits: lint yield (strong field spatial pattern) and lint percentage (no field spatial pattern). The results suggested that the MGA method was able to effectively separate the spatial variation including blocking effects from random error variation for lint yield while the adjusted data remained almost identical to the original phenotypic data. With application of the MGA method, the estimated variance for residuals was significantly reduced (62.2% decrease) for lint yield while more genetic variation (29.7% increase) was detected compared to the original data analysis subject to the conventional randomized complete block design analysis. On the other hand, the results were almost identical for lint percentage with and without the application of the MGA method. Therefore, the MGA method can be a useful addition to enhance data analysis when field spatial pattern exists.展开更多
Crop seeds are important sources of protein, oil, and carbohydrates for food, animal feeds, and industrial products. Recently, much attention has been paid to quality and functional properties of crop seeds. However, ...Crop seeds are important sources of protein, oil, and carbohydrates for food, animal feeds, and industrial products. Recently, much attention has been paid to quality and functional properties of crop seeds. However, seed traits possess some distinct genetic characteristics in comparison with plant traits, which increase the difficulty of genetically improving these traits. In this study, diallel analysis for seed models with genotype by environment interaction （GE） effect was applied to estimate the variance-covariance components of seed traits. Mixed linear model approaches were used to estimate the genetic covariances between pair-wise seed and plant traits. The breeding values （BV） were divided into two categories for the seed models. The first category of BV was defined as the combination of direct additive, cytoplasmic, and maternal additive effects, which should be utilized for selecting stable cultivars over multi-environments. The three genetic effects, together with their GE interaction, were included in the second category of BV for selecting special lines to be grown in specific ecosystems. Accordingly, two types of selection indices for seed traits, i.e., general selection index and interaction selection index, were developed and constructed on the first and the second category BV, respectively. These proposed selection indices can be applied to solve the difficult task of simultaneously improving multiple seed traits in various environments. Data of crop seeds with regard to four seed traits and four yield traits based on the modified diallel crosses in Upland cotton （Gossypium hirsutum L.） were used as an example for demonstrating the proposed methodology.展开更多
文摘Spatial variation is often encountered when large scale field trials are conducted which can result in biased estimation or prediction of treatment (i.e. genotype) values. An effective removal of spatial variation is needed to ensure unbiased estimation or prediction and thus increase the accuracy of field data evaluation. A moving grid adjustment (MGA) method, which was proposed by Technow, was evaluated through Monte Carlo simulation for its statistical properties regarding field spatial variation control. Our simulation results showed that the MGA method can effectively account for field spatial variation if it does exist;however, this method will not change phenotype results if field spatial variation does not exist. The MGA method was applied to a large-scale cotton field trial data set with two representative agronomic traits: lint yield (strong field spatial pattern) and lint percentage (no field spatial pattern). The results suggested that the MGA method was able to effectively separate the spatial variation including blocking effects from random error variation for lint yield while the adjusted data remained almost identical to the original phenotypic data. With application of the MGA method, the estimated variance for residuals was significantly reduced (62.2% decrease) for lint yield while more genetic variation (29.7% increase) was detected compared to the original data analysis subject to the conventional randomized complete block design analysis. On the other hand, the results were almost identical for lint percentage with and without the application of the MGA method. Therefore, the MGA method can be a useful addition to enhance data analysis when field spatial pattern exists.
基金supported by the National Basic Research Program of China (No. 2006CB101708)National Science and Technology Supporting Item of China (No. 2006BAD10A00).
文摘Crop seeds are important sources of protein, oil, and carbohydrates for food, animal feeds, and industrial products. Recently, much attention has been paid to quality and functional properties of crop seeds. However, seed traits possess some distinct genetic characteristics in comparison with plant traits, which increase the difficulty of genetically improving these traits. In this study, diallel analysis for seed models with genotype by environment interaction （GE） effect was applied to estimate the variance-covariance components of seed traits. Mixed linear model approaches were used to estimate the genetic covariances between pair-wise seed and plant traits. The breeding values （BV） were divided into two categories for the seed models. The first category of BV was defined as the combination of direct additive, cytoplasmic, and maternal additive effects, which should be utilized for selecting stable cultivars over multi-environments. The three genetic effects, together with their GE interaction, were included in the second category of BV for selecting special lines to be grown in specific ecosystems. Accordingly, two types of selection indices for seed traits, i.e., general selection index and interaction selection index, were developed and constructed on the first and the second category BV, respectively. These proposed selection indices can be applied to solve the difficult task of simultaneously improving multiple seed traits in various environments. Data of crop seeds with regard to four seed traits and four yield traits based on the modified diallel crosses in Upland cotton （Gossypium hirsutum L.） were used as an example for demonstrating the proposed methodology.
