Biometric and Environmental Models for Predicting the Productive Stability of Soybeans under Multiple Climatic and Geographic Scenarios

• Grant number: 24/01868-9
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: August 01, 2024
• End date: July 31, 2026
• Field of knowledge: Biological Sciences - Genetics - Quantitative Genetics
• Principal Investigator: José Baldin Pinheiro
• Grantee: Maurício dos Santos Araújo
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil

Abstract

Soybean, due to its significant socioeconomic importance, stands out as an agronomic and nutritional crop. However, prolonged global climatic events have negatively impacted seed yield. The assessment of predictive scenarios and understanding environmental traits at cultivation sites becomes crucial to optimize genetic selection. This study aims to develop an innovative predictive breeding method, employing environmental features and genomic data to assess the productive stability of soybean genotypes in untested environments. This involves the integration of big data and machine learning. To achieve this goal, two soybean datasets will be utilized. The first comprises information on 290 purelines, including genomic data, in the São Paulo region over four years of improvement. The second dataset covers 406 purelines evaluated in 120 environments in the state of Mato Grosso do Sul, with traits such as plant height, maturation degrees, days to R8 maturation, and seed yield. Initially, a multivariate analysis will be conducted using factor analytical mixed models (FA) and the assessment of similarity between tested and untested environments. Subsequently, various prediction scenarios will be tested, incorporating information from environmental features. Based on these analyses, agroclimatic breeding zones for soybean cultivation will be defined. After predicting genotypes for untested environments, the results will be presented in thematic recommendation maps. Thus, the intention is to develop a predictive tool to assist breeders in decision-making, directing genotypes to specific environments.