Parâmetros genéticos e seleção genotípica de cacaueiro na Amazônia brasileira

Abstract

This research aimed at the estimation of genetic and statistical parameters as well as the determination of genetic and genotypic values utilized for selection among and within hybrid populations of Cacao (Theobroma cacao L.). The data utilized were taken of individual plants in two field experiments of competition of hybrid combinations installed in proper experimental design and conducted at the experimental farm Paulo Dias Morelli (CEPLAC) located at the main producing region of the Brazilian Amazon basin. Due to the unbalance of the original set of data, a peculiarity of the perennial habit of the species, it was used the REML/BLUP, an mixed model. The variable in study was the total number of harvested fruits per plant taken for three consecutive years from one year before through one year immediately after the climax. The parameters repeatability and heritability were estimated based on the estimates of phenotypic and genotypic variances. Similarly the breeding values of the hybrids and of the individuals were estimated. The repeatability values found allowed to infer, with 95% precision, on the number of measurements needed to discriminate among hybrids at plot level (seven for experiment 1 and six for experiment 2). Such numbers are smaller than the ones really utilized in both experiments. Furthermore, four to five seasons were enough to discriminate amongst hybrids. With the level of precision of 80%, both the number of measurements and seasons were three, which is practically acceptable. The analyses for heritability, genetic and genotypic values (BLUP) were performed for both the individual experiments and the combined trials. Heritabilities estimates were of low magnitude (0.20 to 0.21 for hybrids and 0.13 to 0.19 on individual plant basis. High significant estimates were found, however, for all the estimates of genetic and genotypic variances, regardless of the type of analysis. The results obtained for both analyses individual and combined, on discrimination by hybrids and individual plant BLUP were similar as to the hybrids selected and their rank, as well as to the number of selected families regarding both breeding simulations on the short and on the medium range. With regard to the discrimination among individual plants there was a slight discrepancy among the different analyses. According with the combined analysis, three genotypes would be excluded (one in experiment 2) and fourteen extra genotypes would be selected in experiment 1 (two of those of the same hybrid that had their genotypes excluded). It was concluded that the combined analysis was adequate as well as the REML/BLUP procedure for the selections of hybrids per se and genotypes. These results as far as selection of hybrids per are is concerned, indicate the strong necessity of performing evaluations in all the field experiments that have in their composition the hybrid combinations recommended for commercial plantations applying the REML/BLUP methods with data taken on an individual plant basis and with number of measurements adequately repeated (plots and seasons). In the case of non-selected hybrids on the present study, that in turn are presently recommended, they should be eliminated of the hybrid mixtures offered to the growers. They should be also substituted by seeds of the two synthetic varieties which may be immediately produced from the selected individual plants in their respective experiments.