Evaluation of agronomic and physiological traits of soybean, submitted to different iron concentrations and soil waterlogging, by means of artificial neural networks methodology: deep learning

• Grant number: 21/03935-7
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: July 01, 2021
• End date: June 30, 2022
• Field of knowledge: Agronomical Sciences - Agronomy - Crop Science
• Principal Investigator: Rafael Simões Tomaz
• Grantee: Arthur Lopes Ribeiro
• Host Institution: Faculdade de Ciências Agrárias e Tecnológicas. Universidade Estadual Paulista (UNESP). Campus de Dracena. Dracena , SP, Brazil

Abstract

Poor or iron-rich soils present problems for crop cultivation. High concentrations of iron associated with soil waterlogging make this mineral phytotoxic to the plant. In this context, another harmful factor is the filling of the pore spaces of the soil by water, restricting the supply of O2 to the root and nodules. In view of the above, this work aims to evaluate the agronomic and physiological characteristics of soybean varieties, in order to investigate the plant behavior when subjected to increasing concentrations of iron under optimal water and soil waterlogging conditions. We will consider two experiments, already executed, which we propose to evaluate by means of traditional methodology (REML BLUP) and by means of Artificial Intelligence methodology - Artificial Neural Networks. The experiments were conducted in the agricultural years 2018-2019 and 2020-2021, under entirely randomized design, in an agricultural greenhouse. The first with three repetitions, in a 4x3x3 factorial scheme; four varieties, three levels of soil waterlogging and three Fe concentrations; and the second, with four repetitions, in a 5x4x5 factorial scheme, five varieties, four levels of waterlogging and five Fe concentrations. Additionally, images of the experiments were collected in order to make associations between the phenotypes and the characteristics measured through the Neural Networks methodology. The identification of varieties tolerant to the toxic effect of this combination of factors is of great importance to mitigate the deleterious impacts of this type of stress. An additional challenge is to associate images with the phenotypes measured by means of Artificial Intelligence tools, which have proven to be very promising in the process of data modeling.(AU)