Reason: Preparing Manuscript
until file(s) become available
HYPERSPECTRAL PHENOTYPING OF CROP FUNCIONAL TRAITS OVER VARIATION IN THE ENVIRONMENTAL, ABIOTIC AND BIOTIC STRESS, AND GENETICS
Modern agriculture must address the massive challenge of providing food for the increasing population. The challenge lies in increasing crop yield and reducing losses caused by abiotic and biotic stresses. In fact, for some crops, such as wheat and maize, over 40% of the production is lost due to environmental conditions (abiotic stresses) or pests and pathogens (biotic stresses). Specialists in the area are suggesting a need for a second green revolution to meet the increasing demand in food production. While in the first green revolution was focused on breeding and genetics to produce crops' genetic lines with a higher yield. The second green revolution will utilize cutting-edge technologies to increase yield and reduce crop losses. The development of remote sensing technologies and their applications is the main driving force of modern agricultural practices. Currently, farmers are relying more on automation, data collection, and data analysis to manage farming operations. The reliance on remote sensor technologies is a game-changer for traditional agricultural practices, and it is contributing tremendously to increasing production and avoiding yield losses. Hyperspectral phenotyping is an emerging remote sensing technology that utilizes the light's reflectance to provide insightful information about plant traits. For several years, research groups have been applying hyperspectral phenotyping techniques to detect plant traits information, such as nitrogen content, photosynthesis rates, pests infestation, and abiotic stress detection. Although this is not a novel approach to plant traits detection, this technology application is not mature yet. Several challenges are associated with using hyperspectral information for phenotyping, such as model transferability, data collection scalability, and the heritability of plant traits retrieved using hyperspectral data. In my thesis dissertation, I addressed some of those challenges contributing to advances in hyperspectral phenotyping. My results demonstrate that using full-range hyperspectral reflectance data (400-2400nm) to retrieve nitrogen in winter wheat increases the model transferability across years and genotypes. Predicting nitrogen content using hyperspectral data can be used as a surrogate to calculate nitrogen use efficiency traits. My research highlights the hurdles associated with spectral detection of stresses interaction, such as drought stress, which can mask western corn rootworm detection in maize. Finally, I explored the correlation among spectral, functional, and field traits in a soybean NAM (Nested Association Mapping) population to understand the relationship among those traits' variability and how that information can be used for soybean breeding programs. The outcomes of my thesis dissertation advance the knowledge in the hyperspectral phenotyping field and its application to modern agriculture. Consequently, my study also contributes to food security programs by providing insightful information about the hyperspectral assessment of plant health status, which is essential to increase yield production and reduce crop losses.