Validation of models developed for sexage of papaya using near red spectroscopy (NIR)

• Grant number: 21/02633-7
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: May 01, 2021
• End date: April 30, 2022
• Field of knowledge: Agronomical Sciences - Agronomy - Crop Science
• Principal Investigator: Gustavo Henrique de Almeida Teixeira
• Grantee: Gabriel Ramos
• Host Institution: Faculdade de Ciências Agrárias e Veterinárias (FCAV). Universidade Estadual Paulista (UNESP). Campus de Jaboticabal. Jaboticabal , SP, Brazil

Abstract

The Master Science study of the student Thiago Feliph Silva Fernandes at the Universidade Estadual Paulista (UNESP), Faculdade de Ciências Agrárias e Veterinárias(FCAV), Jaboticabal Campus, was related to the development of classification models for seed and seedlings (leaves) of five papaya cultivars according to the sexual type using near infrared spectroscopy (NIR). When seeds were used, it was possible to obtain an accuracy of 0.80 for the external validation group using PCA-QDA. For leaves, the accuracy was slightly lower, that is, 0.77 using PCA-LDA. Despite the promising results, it is necessary to validate these models with a new sample group in order to confirm the results and incorporate this new data set to improve the robustness of the models. Thus, the general objective of this project is to validate the models of classification of seeds / leaves according to the sexual type of papaya (female and hermaphrodite) and, for specific objectives: i. improve the robustness of the models by incorporating a new source of variation. For this, seeds of cultivars 'T2', 'Formosa' and 'Calimosa', from the Formosa group, and 'THB' and 'Gold', from the Solo group, will be used. In addition to the seeds, the other plant material will consist of the leaves of the seedlings originating from them. After obtaining the NIR spectra of the seeds and leaves of the seedlings, the sexual types will be predicted and then confirmed in field conditions. Spectral and reference data (sexual types) will also be incorporated into the database and new models will be developed using principal component analysis (PCA), linear discriminant analysis (LDA) and discriminant analysis by partial least squares (PLS-DA) . We expect to develop a robust alternative method for the early determination of the papaya's sexual type to be used by industries and growers. (AU)