Spatial kernel models capturing field heterogeneity for accurate estimation of genetic potential.

Area of Science:

• Agricultural Science
• Genetics
• Biometrics

Background:

• Traditional field trials use block designs for local control, assuming within-block homogeneity.
• Large experimental fields often exhibit heterogeneity, biasing genetic potential estimations.
• Accurate genotypic value estimation is crucial for effective plant breeding.

Purpose of the Study:

• To develop and validate spatial kernel models for accurate genotypic value estimation in heterogeneous field trials.
• To assess the performance of spatial kernel models compared to classical block models.
• To apply spatial kernel models for quantitative trait locus (QTL) mapping.

Main Methods:

• Development of spatial kernel models incorporating genome-wide markers.
• Simulation studies to evaluate model robustness under various conditions.
• Application of models to field experimental data of bioenergy sorghum lines.

Main Results:

• Spatial kernel models demonstrated robustness and superior performance over classical block models.
• Model accuracy was influenced by heritability, spatial autocorrelation, replication number, and missing plots.
• Validation with sorghum data confirmed the effectiveness of spatial kernel models in heterogeneous environments.

Conclusions:

• Spatial kernel models effectively account for continuous field heterogeneity.
• These models provide more accurate genotypic value estimations in plant breeding.
• Spatial kernel models are valuable tools for genetic potential evaluation in challenging field conditions.