Nonlinear genomic selection index accelerates multi-trait crop improvement

Area of Science:

• Plant breeding
• Quantitative genetics
• Genomics

Background:

• Linear selection indices limit exploiting complex, nonlinear trait relationships.
• Genomic selection has advanced breeding but often relies on additive models.

Purpose of the Study:

• Introduce the Quadratic Genomic Selection Index (QGSI) to integrate nonlinear genomic information.
• Enable phenotype-free, rapid-cycle, multi-trait selection using genomic estimated breeding values (GEBVs).

Main Methods:

• Developed QGSI by extending the quadratic phenotypic selection index (QPSI) to a genomic framework.
• Integrated additive, squared, and cross-product terms of GEBVs.
• Evaluated QGSI using maximum-likelihood additive and nonlinear Gaussian kernel genomic prediction models.

Main Results:

• QGSI demonstrated superior performance across simulated and real datasets (maize and wheat).
• Achieved higher selection response and lower prediction error variance compared to linear and quadratic indices.
• Effectively captured genome-wide nonlinear relationships and epistatic signals.

Conclusions:

• Combining nonlinear genomic prediction with quadratic selection indices accelerates multi-trait crop improvement.
• QGSI offers a general strategy for enhancing breeding efficiency by leveraging complex genetic architectures.