Unlocking almond breeding for nutritional composition with hyperspectral imaging.

Area of Science:

• Plant Breeding
• Agricultural Science
• Nutritional Genomics

Background:

• Traditional phenotyping methods for plant nutritional components are slow and costly.
• High-throughput phenotyping (HTP) offers a solution to accelerate plant breeding.
• Developing efficient HTP platforms is crucial for modern crop improvement.

Purpose of the Study:

• To develop and validate an HTP platform for phenotyping nutritional components in almond ( *Prunus dulcis* ) breeding populations.
• To address the bottleneck in conventional phenotyping for traits like fats, protein, and fatty acids.
• To generate a large-scale dataset for nutritional traits in almonds.

Main Methods:

• Integration of hyperspectral imaging (HSI) with a Python workflow for HTP.
• Analysis of kernel and powder samples from 112 almond genotypes using SWIR range HSI.
• Development of Partial Least Squares (PLS) models to predict nutritional components (fats, protein, fiber, sucrose, fatty acids, phytosterols).
• Estimation of narrow-sense heritability using linear mixed models with pedigree and genomic data (60K Almond SNP array).

Main Results:

• PLS models achieved good predictive accuracy for protein (R²_CV=0.82), fats (R²_CV=0.86), β-sitosterol (R²_CV=0.66), and oleic acid (R²_CV=0.57).
• The platform successfully predicted nutritional components in 528 genotypes and six F₁ populations.
• High narrow-sense heritability (ℎ ² >0.5) was observed for predicted nutritional traits, indicating significant additive genetic effects.
• Generated the largest phenotypic dataset for nutritional components in almonds to date.

Conclusions:

• The developed HTP platform significantly advances nutritional almond breeding by enabling rapid and large-scale phenotyping.
• This approach allows for the assessment of six times more individuals compared to previous studies.
• The findings provide a foundation for selecting and breeding almonds with enhanced nutritional profiles.