Making the Genotypic Variation Visible: Hyperspectral Phenotyping in Scots Pine Seedlings.

Area of Science:

• Plant Science
• Remote Sensing
• Forest Genetics

Background:

• Hyperspectral reflectance reveals plant physiological status through leaf functional traits.
• High-throughput phenotyping using hyperspectral data can aid in selecting trees with adaptive potential.

Purpose of the Study:

• To evaluate two non-destructive hyperspectral reflectance methods for phenotyping Scots pine (Pinus sylvestris) seedlings.
• To assess the potential for distinguishing between lowland and upland ecotypes from different local populations.

Main Methods:

• Compared leaf-level and proximal/canopy hyperspectral reflectance measurements (350-2500 nm) on 1,788 Scots pine seedlings.
• Utilized spectroradiometer with contact probe for leaf measurements and fiber optics for canopy measurements.
• Applied random forest and support vector machine algorithms for population prediction.

Main Results:

• Both spectral datasets showed significant differences among Scots pine populations across the entire spectral range.
• Proximal canopy measurements achieved up to 83% accuracy in predicting three distinct Scots pine populations.
• Leaf-level measurements also provided valuable phenotypic data.

Conclusions:

• Both leaf-level and proximal/canopy hyperspectral phenotyping are viable for Scots pine.
• These methods can effectively distinguish phenotypic and underlying genetic variation within seedling populations.
• This approach supports tree breeding and nursery selection for environmental adaptation.