Alfalfa breeding for cultivation in the tropics: insights on genetic diversity and yield persistence

• Ano de defesa: 2021
• Autor(a) principal: Santos, Iara Gonçalves dos
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Viçosa
• Palavras-chave em Português: Medicago sativa - Melhoramento genético; Análise de regressão; Redes neurais (Neurobiologia); Genética Quantitativa
• Link de acesso: https://locus.ufv.br//handle/123456789/28196

Alfalfa (Medicago sativa L.) is considered “the queen of forages” and plays a key role in highly specialized dairy herds. Alfalfa has the potential to be grown in different edaphoclimatic regions, though its cultivation in tropical regions is still limited. Because yield persistence is one of the bottlenecks of alfalfa breeding in tropical regions, efforts should be done to overcome this problem. This study aimed to investigate whether the alfalfa germplasm held by Embrapa Southeast Livestock has satisfactory genetic diversity regarding bromatological and agronomic traits. The investigation also looked into yield persistence, how to access it, and how to select persistent accessions based on random regression models and artificial neural networks (ANN). Best linear unbiased predictors (BLUPs) of nine traits of seventy-seven alfalfa accessions from a temperate genetic background evaluated in eight harvests were used to estimate the phenotypic diversity. Microsatellite markers assessed the molecular diversity. Phenotypic data analyses revealed the presence of genetic diversity. The genetic variability obtained by both phenotypic and molecular information indicated the potential of the germplasm for developing base populations adapted to tropical conditions. Dry matter yield taken from 24 cuttings was used to assess the persistence. A random regression model was used to build trajectory curves of the accessions. The fitted curves showed a great amplitude regarding dry matter yield over time, which suggested a high variability regarding persistence. The three-step method for accessing persistence presented in this study included (1) a random regression model to obtain persistence trends, (2) a k-means method to define different persistence clusters, as well as (3) an ANN to perform classification of persistent accessions in an automated way. The upside of this method is to evaluate different alfalfa accessions using the same ANN. Basically, when new accessions are evaluated, they will be classified according to their genetic value scores using the same ANN previously fitted, with no need for a new clustering step. The persistence method jumps down from three to two steps and can help alfalfa breeders in the decision- making process. Keywords: Medicago sativa. Self-organizing maps. Random regression models. Artificial neural network.