Divergência genética em acessos de melão utilizando redes neurais artificiais

• Ano de defesa: 2015
• Autor(a) principal: Melo, Stefeson Bezerra de
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal Rural do Semi-Árido; BR; Centro de Ciências Agrárias - CCA; UFERSA; Programa de Pós-Graduação em Fitotecnia
• Palavras-chave em Português: Cucumis melo; Germplasm; Genetic divergence; Germoplasma; Divergência genética; CIENCIAS AGRARIAS::AGRONOMIA::FITOTECNIA
• Link de acesso: https://doi.org/10.21708/bdtd.ppgfito.tese.172; https://repositorio.ufersa.edu.br/handle/tede/172

Melon (Cucumis melo L.) is a species economic importance to Brazilian northeast, especially in Mossoró-Assú agropolo. The study of genetic diversity allows in the preliminary selection of individuals with superior characteristics to produce hybrids with the high heterotic in breeding programs. The objective of this study was to evaluate the genetic divergence among 46 melon genotypes for 22 physicochemical and agronomic quantitative variables, evaluated by techniques artificial neural networks. Two experiments were conducted in Horta Experimental Department of Plant Sciences at the Universidade Federal Rural do Semi-Árido - UFERSA in the Mossoró, State of Rio Grande do Norte, in the periods 12/09/2006 to 05/12/2006 and 15/08/2007 to 17/10/2007. Through the techniques of artificial neural networks, was found four groups for both experiments, but also to average of two years. A discriminant analysis was used to check the consistency of groups formed and it was observed that considering the 22 variables, there was 100% hit, that is, for the discriminant function all genotypes were classified correctly. In addition was also observed distances between groups and group 1 was significantly distant from all other groups, more distant, genetically, Group 3. Group 2 are different with respect to group 3 and group similar to 4. The group 3 shows similarity to group 4. And so we suggest possible crosses between accessions 2, 13, 15, 16, 17, 27, 33, 36, 40, 43, 46 that would be most promising for new populations of work. Artificial neural networks have proved viable as a method of analysis of genetic divergence in melon and genetic divergence was found for all groups, and with that you can get new crossings in order to obtain improved populations