The viability of high-throughput phenotyping via remote sensing of tropical maize inbred lines

• Grant number: 18/19572-8
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: December 01, 2018
• End date: November 30, 2019
• Field of knowledge: Agronomical Sciences - Agronomy
• Principal Investigator: Roberto Fritsche Neto
• Grantee: Leonardo Fioravante Gotardi
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil

Abstract

Phenotyping can be performed at different phenological crop stages, which allows the understanding of the relationships between yield and physiological traits, as well as their dynamics throughout the growth and development process. However, the traditional evaluation is a labor-intensive, poorly automated, highly sensitive to environmental features, and, in some cases, subjective. On the other hand, the high-performance phenotyping can circumvent these barriers. These are usually based on remote sensing and imaging, in which conventional (red, green and blue - RGB), multispectral and hyperspectral bands are used to capture reflection/emission profiles of the visible spectrum (VIS), near infrared (NIR) and distant (thermal). In this context, the objective is to evaluate the feasibility of the use of high-throughput phenotyping by remote sensing in the evaluation of tropical maize inbred lines. Therefore, 361 inbred tropical maize lines representing the germplasm of breeding programs of ESALQ-USP, IAC, IAPAR, and CIMMYT will be phenotyped (by the traditional way and via multispectral cameras) in two locations. Based on the data and relations between the methods, we aim to: I) develop high-throughput phenotyping protocols for tropical maize; II) identify the ideal phenological stage for early selection; III) evaluate the effect of genotype by environment interaction on the accuracy of remote sensing evaluations; IV) verify the minimum number of observations for each phenotype in order to maximize the correlations between the conventional phenotype and high-throughput phenotyping.