Tese - Universidade Federal do Acre, Programa de Pós-Graduação em Agronomia, 2018

Abstract: The species Panicum maximum stands out in the Brazilian livestock industry in the formation of productive pastures with good nutritional quality. The development of new cultivars in Brazil is carried out by the Embrapa Panicum maximum Genetic Improvement Program, which has a germplasm bank of this species and promotes intraspecific crosses of apomictic plants with tetraploidized sexual plants. One of the stages of the breeding program is the regional trials in the different Brazilian biomes and Embrapa Acre is responsible for the evaluation in the Amazon biome. In this study, the objectives were to estimate the genetic parameters, select superior genotypes, characterize morphologically and know the genetic diversity of P. maximum hybrids evaluated under cuts in the Amazon biome. The experiment was installed in December 2012 in the experimental field of Embrapa Acre, in Distrophic Red Latosol, in a completely randomized block design with four replicates. Twenty intraspecific hybrids and the controls cultivars Mombaça and Tanzânia-1 were evaluated. Sixteen harvests were evaluated between May 2013 and March 2015. Ten agronomic variables were evaluated, considering the 16 evaluations four bromatologic, in the first seven cuts and eight of morphological characterization evaluated before the 13th cut. The genetic parameters of the agronomic and bromatological variables were estimated, and the genotypic values were predicted by the mixed models methodology via REML/BLUP, considering all the evaluations and separately for the rainy and dry seasons. Pearson correlation was also obtained based on genotypic values. From the selection indexes by sum of ranks and multiplicative, the best hybrids were selected considering four variables. For the study of diversity, Tocher optimization grouping and principal component analysis were performed based on the agronomic variables and the Scott-Knott grouping of the morphological characterization variables after analysis of variance. There was significant genetic variability for all variables except crude protein, which allows for genetic gains. The individual broad sense heritabilities were of low to moderate magnitude. There were no significant differences in the estimates of the parameters among the analyses, allowing selection in the dry season. Repeatabilities were low to moderate, but allowed to infer in the minimum number of harvests with 80% reliability for the agronomic variables within the evaluations done. There were significant and high magnitude correlations that allow for the indirect selection between regrowth and leaf percentage in the joint analysis and dry matter and leaf dry matterproduction in the dry season. The indexes by sum of ranks and the free from weights and parameters were concordant in the choice of the seven best hybrids that made possible to obtain genetic gains for the agronomic variables. Genetic variability was observed for all morphological characteristics, especially for plant height with a greater number of groups. The Tocher method and principal components analysis showed consistency in the formation of groups by agronomic variables, with partial influence of kinship. The estimated parameters indicate success in this phase of Panicum maximum breeding with selection for the next stages of evaluation the genotypes C12, C55, B97, C10, B126, B44 and C53. There is genetic diversity among the genotypes, even among complete siblings, with different morphological and agronomic characteristics that allow the selection of hybrids superior to cv. Mombaça.

Orientadora Giselle Mariano Lessa de Assis