Genomic selection performs as effectively as phenotypic selection for increasing seed yield in soybean.

Area of Science:

• Plant breeding
• Genetics
• Agricultural science

Background:

• Increasing genetic gain for soybean seed yield is a primary objective in breeding programs.
• Genomic selection (GS) shows potential for accelerating genetic gain in soybean.
• Limited studies have validated GS accuracy and compared it with phenotypic selection (PS) in soybean.

Purpose of the Study:

• To empirically validate genomic selection (GS) for increasing soybean seed yield.
• To compare the effectiveness of GS against phenotypic selection (PS) and random selection (RS).
• To assess GS prediction accuracy in diverse validation sets within a real-world breeding program.

Main Methods:

• Utilized over 1,500 soybean lines and 7 years of replicated experimental data (2010-2016).
• Trained the genomic selection model on a large population and validated it on three distinct sets: two biparental populations and one advanced line population.
• Compared prediction accuracy from validation experiments with cross-validation results.

Main Results:

• Realized prediction accuracy in validation experiments (.54) was comparable to cross-validation accuracy (.64).
• Both GS and PS improved population mean performance over random selection (RS).
• GS demonstrated a selection advantage over PS, maximizing genetic gain and top-line identification at 10-20% selection intensity.
• GS resulted in minor increases in genetic similarity compared to PS and RS, potentially shifting allelic frequencies.

Conclusions:

• Genomic selection (GS) can be as effective as phenotypic selection (PS) for improving soybean seed yield.
• GS may accelerate genetic gain and identify superior lines more efficiently.
• Consideration must be given to potential erosion of genetic diversity when implementing GS, necessitating protective measures.