Main content

Using Advanced Proximal Sensing and Genotyping Tools Combined with Bigdata Analysis Methods to Improve Soybean Yield

Show full item record

Title: Using Advanced Proximal Sensing and Genotyping Tools Combined with Bigdata Analysis Methods to Improve Soybean Yield
Author: Yoosefzadeh Najafabadi, Mohsen
Department: Department of Plant Agriculture
Program: Plant Agriculture
Advisor: Eskandari, Milad
Abstract: Improving yield potential in major food-grade crops such as soybean (Glycine max L.) is the most sustainable way to address the growing global food demand and its security concerns. Selections for high-yielding cultivars have been mainly focused on the yield performance per se but not necessarily on secondary related-traits associated with yield. Recent substantial advances in proximal sensing have provided plant breeders with affordable and efficient tools for evaluating a large number of genotypes for important agronomic traits, including yield, at early growth stages. Nevertheless, the implementation of large datasets generated by proximal sensing such as hyperspectral reflectance in cultivar development programs is still challenging due to the essential need for intensive knowledge in computational and statistical analyses. Therefore, this thesis was aimed to: (1) investigate the potential use of soybean hyperspectral reflectance, hyperspectral reflectance indices (HVI), and yield components such as number of nodes (NP), number of non-reproductive nodes (NRNP), number of reproductive nodes (RNP), and number of pods (PP) per plant for predicting the final seed yield using different machine learning (ML) algorithms, (2) select the top-ranking hyperspectral reflectance and HVI in predicting soybean yield and fresh biomass (FBIO) using recursive feature elimination (RFE) strategy, (3) implement genetic optimization algorithm and the improved version of the strength Pareto evolutionary algorithm 2 (SPEA2) to optimize yield components and HVI for maximizing soybean seed yield and FBIO, and (4) study the genetics of soybean yield and its secondary related-traits in order to discover genomic regions underlying the traits by using genome-wide association study (GWAS). In this study, different ML algorithms such as ensemble stacking (E-S), ensemble bagging (EB), and deep neural network (DNN) were tested to evaluate their efficiency in predicting soybean yield and FBIO production using a panel of 250 genotypes evaluated in four environments. Also, for the first time, we implemented ML algorithms in GWAS to detect the associated QTL with soybean yield components. The results of this study may provide a perspective for geneticists and breeders regarding the use of ML algorithms in phenomics and genomics that will result in the selection of superior soybean genotypes.
URI: https://hdl.handle.net/10214/26229
Date: 2021-07-27
Terms of Use: All items in the Atrium are protected by copyright with all rights reserved unless otherwise indicated.


Files in this item

Files Size Format View
YoosefzadehNajafabadi_Mohsen_202107_PhD.pdf 5.991Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record