Coupling day length data and genomic prediction tools for predicting time-related traits under complex scenarios

Area of Science:

• Plant breeding and genetics
• Agricultural science
• Quantitative genetics

Background:

• Genomic selection (GS) is effective for predicting crop performance but faces challenges with traits like phenology stages.
• Days to heading (DTH) is critical for rice regional adaptability and yield, yet predicting it in new environments is difficult.
• Traditional GS can exhibit bias and reduced accuracy for time-related traits in unobserved environments.

Purpose of the Study:

• To develop and evaluate a novel method for accurately predicting time-related traits, specifically rice DTH, in unobserved environments.
• To incorporate day length (DL) information into GS models to improve predictions for DTH.
• To assess the performance of the new method in two scenarios: predicting tested genotypes in new environments (CV0) and predicting untested genotypes in new environments (CV00).

Main Methods:

• Proposed a genomic selection implementation incorporating day length (DL) data.
• Evaluated the method in two cross-validation scenarios: CV0 (predicting tested genotypes in unobserved environments) and CV00 (predicting untested genotypes in unobserved environments).
• Compared the DL-incorporating methods (C and CB) against conventional GS for predicting Days to Heading (DTH).

Main Results:

• The DL-incorporating methods significantly improved the predictive ability of DTH in unobserved environments.
• Under CV0, the C method achieved a root-mean-square error (RMSE) of 3.9 days and a Pearson correlation (PC) of 0.98.
• Under CV00, the CB method achieved an RMSE of 7.3 days and a PC of 0.93, substantially outperforming conventional GS (RMSE 18.1 days, PC 0.41).

Conclusions:

• Incorporating day length (DL) information is a valuable strategy for enhancing genomic selection accuracy for time-related traits like rice Days to Heading (DTH).
• The proposed methods offer a more reliable approach for predicting crop phenology in unobserved environments compared to conventional genomic selection.
• Day length data offers practical advantages due to its predictability based on location and planting date, facilitating better regional adaptation predictions.