Using Aerial Images and Canopy Spectral Reflectance for High-Throughput Phenotyping of White Clover

Plant breeders are demanding high-throughput phenotyping methodologies to complement the abundant genomic information currently available. Remote-sensing technologies offer new tools for high-throughput phenotyping in field conditions, and many remote sensors have shown high capacity for describing plant physiological behavior. The objective of this study was to evaluate the genotypic relationship between high-throughput phenotyping based on image analysis and canopy reflectance estimated traits and dry matter (DM) production, the most important trait in forage species. An experiment of a white clover (Trifolium repens L.) association-mapping population was established in three locations. Plant DM production was evaluated during two growing seasons. The plant area (PA), normalized difference vegetation index (NDVI), and plant growth were estimated from multispectral aerial images collected with an unmanned aerial vehicle. Additionally, canopy reflectance was evaluated with a spectroradiometer (350-1075 nm) and 10 spectral reflectance indices (SRIs) were calculated, including NDVI. The image-derived PA trait showed the highest genetic correlation with DM production (r(g) = 0.88, P < 0.001) with a broad-sense heritability (H-2) value of 0.56. All the SRIs showed highly significant genetic correlation with DM production with r(g) absolute values between 0.54 and 0.72 (P < 0.001). However, the popular NDVI index showed one of the lowest DM correlations using both systems. The results indicate that aerial-image-derived traits and SRIs could be used together as a high-throughput proxy to estimate genotypic variation of white clover DM production. Use of these variables could contribute to alleviating phenotypic bottleneck in discovering genes or predicting yield using genomic data.