A New Approach For Assessing Traits In Wheat Breeding Programs And Its Application

Many traits concerning morphology, yield, quality, resistance, adaption and others need to be investigated in wheat breeding program. The relative importance of these traits is not same for different breeding objectives. Intensive recording of all traits is not feasible and necessary. The strength of investigation for different traits being connected with breeding target accordingly is most economically effective. This paper put forward a new statistical method to assess reasonably and objectively the importance of many traits. It would be beneficial for plant breeders who might focus on main characters according to different breeding objectives. Making intensive records for main characters, neglecting those trivial traits, breeders could reduce workload, raise the efficiency in large-scale breeding program.In this paper, yield and cooking quality of wheat were taken as two different breeding objectives. Two experiments were conducted in LIXIAHE Agricultural Institute in previous years. The first experiment included 1413 records of 580 wheat lines from 2006 to 2013, involving the traits as yield, bulk density,1000 grain weight, plant height, heading date, maturity periods, disease resistance (powdery mildew, scab), with yield as breeding objective. The second experiment had 488 records of 14 wheat varieties, recording traits as yield,1000 grain weight, bulk density, hardness, flour yield, flour protein, wet gluten, SDS, falling number, whiteness (L*, a*, b*), powder mass spectrometer parameters (formation time, stability time, RVA (peak viscosity, dilute creek value), solvent retention capacity (water SRC, sucrose SRC, sodium carbonate SRC, lactic acid SRC), decorative pattern, ratio of diameter to thickness, texture analyzer parameters, taste test of biscuits according to China GB (score summary of biscuits tested by 7 people, three-point bending texture analyzer parameters) with baking quality ratio of diameter to thickness as breeding objective. It is essential to understand the correlation of characters before evaluating the lines or varieties. Firstly, the error variance and covariance matrix, the underlying correlation of the traits, was calculated by matrix of total sum of squares and products eliminating the matrix of year and that of lines, and then divided by degree of freedom of error. Then, a standard (E) was determined by taking average of the best lines (or varieties). The quantitative relationships value (and similarity) between lines (variety) and standard (E) was calculated through a nonlinear function, which was established though Monte Carlo simulation test for original data. Numerical scoring and ranking of each line was used as evaluation value. The evaluation value was used as the dependent variable Y, and character records were used as X, the respective partial coefficient of determination was calculated through multivariate polynomial regression analyses. The partial coefficients of determination could be the criteria to clarify the relative importance of each character and effect to breed selection.The study showed that when breeding targets was yield output, the linear effects of wheat yield, bulk density, plant height and maturity periods had large numerical value in the partial coefficient of determination; the quadratic effects of wheat yield, bulk density,1000 grain weight, plant height and growth period had large numerical value in the partial coefficient of determination, indicating that these traits had a range of the most appropriate value. Some traits, such as yield and bulk density, bulk density and 1000 grain weight, growth period and heading date, have large interaction effects to display the selection of these traits should be taken into consideration simultaneously.