Modelling and Using Spatial Effects in Nationwide Historical Data Improve Genomic Prediction of Rice Heading Date in...

Area of Science:

• Agricultural Science
• Genetics
• Plant Breeding

Background:

• Genomic prediction models enhance crop breeding efficiency by utilizing historical data.
• Geographically diverse historical data offers rich information but requires effective handling methodologies.
• Spatial structures among field stations can influence prediction accuracy.

Purpose of the Study:

• To improve the prediction accuracy of genomic prediction models using historical breeding and cultivation data.
• To explore methodologies for effectively handling geographically wide historical data.
• To investigate the impact of spatial effects on genomic prediction of rice heading date.

Main Methods:

• Conventional genomic prediction models were constructed using genomic and meteorological elements.
• A spatial model was applied to residual terms to calculate spatial effects of field stations.
• Model performance was evaluated using root mean squared errors (RMSE) under cross-validation and leave-one-line-out schemes.

Main Results:

• Genomic prediction models incorporating genome, meteorological elements, and their interactions performed best.
• The inclusion of spatial effects improved prediction accuracy, reducing RMSE by approximately 15% (10-fold CV) and 9% (LOLO CV).
• Spatial effects were heterogeneous, with regional patterns detected, particularly in the Tohoku region, and identified lines with improved predictions.

Conclusions:

• Spatial effects are crucial for enhancing prediction performance in genomic prediction models.
• Accounting for spatial effects aids in dissecting models and identifying factors contributing to improved predictions.
• This methodology offers a valuable approach for leveraging historical data in crop breeding programs.