Phenotyping a large set of germplasms presents a substantial hurdle in extensive plant breeding programs. Although genomic prediction offers a contemporary solution by leveraging genotypic data to predict intricate traits, it lacks site-specific details and can also be expensive. Remote sensing data emerges as a potential solution, providing detailed in-season data that indirectly incorporates environmental impacts on genotypes.

The stagnation in enhancing the harvest index (HI) in wheat since the 1990s primarily arises from the labor-intensive nature of its assessments. This has impeded the potential advancements in wheat genetic yield. Traits such as spike partitioning index (SPI) and fruiting efficiency (FE) have surfaced as crucial factors associated with HI.

This study aims to develop efficient phenotyping tools for assessing HI and related traits, integrating genomics, and phenomics data from unmanned aerial vehicle (UAV) sensors. Conducted in Citra, Florida, from 2022 to 2024, the prediction model training population encompassed 160, 306, and 306 elite facultative soft wheat lines, respectively.

Furthermore, a separate validation study in 2024 utilized genetically related yet distinct lines to the training set, developed from the University of Florida's world food crop breeding program. These lines underwent comprehensive trait characterization, encompassing HI, grain yield, grain number, SPI, FE, thousand-grain weight, and plant height, alongside UAV-based hyperspectral, multispectral, and thermal data acquisition.

Genotyping utilizing the Genotype by Sequence (GBS) method facilitated the development of SNP markers.

The findings revealed substantial genetic diversity among genotypes across all traits, with significant correlations observed between manual data collection and UAV-based remote sensing data. The single kernel phenomics model exhibited superior performance over the genomics model for most traits, although the model combining both genomic and sensor fusion data (G+H+W) demonstrated the highest efficiency.

In conclusion, this study highlights the utility of UAV-based remote sensing data, either independently or to enhance the predictive ability of genomic prediction model. These advancements are instrumental in estimating the harvest index (HI) and associated biomass partitioning traits in large scale wheat breeding program, ultimately facilitating accelerated genetic yield improvements.