Prediction of Rice Plant Height Using Linear Regression Model by Pyramiding Plant Height-Related Alleles.

Although numerous rice plant height-related genes have been cloned and functionally characterized in recent years, a gap between the identified genes and their utilization in breeding still exists. Here, we developed a linear regression model by pyramiding plant height-related alleles to predict rice plant height and confirmed that it can be used in rice breeding. In our study, we firstly identified 22 plant height-associated molecular markers from 218 markers in an association mapping population which consisted of 273 rice varieties. Linear regression analysis revealed a positive correlation between rice plant height and the number of plant height-increasing alleles derived from these 22 molecular markers. Subsequently, linear regression models were developed using 2-10 loci based on the genotype and phenotype data of the association mapping population. The predictive accuracy of the model was tested using a recombinant inbred line (RIL) population consisting of 219 lines, and it revealed the trend that predictive accuracy increased with more loci in a certain range of less than five loci. If the prediction model was built based on 5-10 loci, it yielded an average absolute error from 11.05 to 11.96 cm, which was smaller than absolute error induced by environmental factors (5.72 cm to 12.79 cm). The reliable prediction of rice plant height by this model highlights its value as a practical tool for optimizing rice breeding strategies. Additionally, the linear regression model developed in this study not only can facilitate plant height manipulation but also will inspire other design breeding techniques in other crops or other traits.