Structural equation modelling and factor analysis of the relationship between agronomic traits and vegetation indices in corn

Abstract

Corn has great relevance worldwide due to its economic and social importance. Its production grows every harvest, and determining traits directly related to the grain yield is essential for crop management. The hypothesis of this study is that structural equation modelling (SEM) and factor analysis allow us to identify agronomic variables and vegetation indices (VIs) that most contribute to grain yield. Thus, the objective was to identify which agronomic traits and VIs are most associated with grain yield in corn using SEM and factor analysis. Field experiments were carried out during the 2017/2018 and 2018/2019 crop seasons. The experimental design was in randomized blocks with three replicates in a factorial scheme. The following agronomic traits were evaluated: ear length (EL), number of rows per ear (NRE), number of grains per row (NGR), plant height, ear insertion height, stem diameter, and grain yield. The VIs evaluated were: Normalized Difference Vegetation Index (NDVI), Normalized Difference Red Edge (NDRE), Green Normalized Difference Vegetation Index, Soil-Adjusted Vegetation Index (SAVI). Data were subjected to factor analysis and structural equation modelling. The EL, NGR, and NRE were the agronomic traits that most related to corn yield. The VIs most related to yield were NDVI, NDRE, and SAVI. By identifying which agronomic variables and VIs are the most related to grain yield in corn, this study contributes to the decision-making process on HTP in corn and consequently to a higher efficiency of corn breeding programs aiming at selecting high-yielding genotypes.