Biometric features and oil contents of macauba (Acrocomia aculeata) palms natively growing in sites representatives of the Brazilian Cerrados (Campo Grande-CG, and Sao Gabriel do Oeste-SGO), and Pantanal (Corumba-CO, and Aquidauana-AQ) biomes were studied. The objective was to obtain data that could assist better planning for the exploitation of natural populations, and knowledge to help selection and breeding programs. Plant (spines presence; stem circumference; number and length of bunches) and fruits characteristics (fruits per bunch; size; mass; proportion of components-husk, pulp, shell and kernel; fruits moisture at ripening; pulp and kernel oil content) were measured from ten native palms from each site. There was variability in biometrics, biomass production potential and oil yield in the palms within each site and among the sites. No correlation was found between biometrics features and oil contents. On average, plants from CG showed better general productive characteristics than any other of the studied sites, but a large variability in fruits size. Palms from CO presented smaller fruits, but they were more homogeneous in size and with higher kernel oil contents. Fruits from SGO show bigger mass and size, but it did not result in higher fruit biomass production or oil yield. Although oil yields potential were almost half of what is usually measured in other areas of Brazil, results provided interesting data for a better planning to rational exploitation of native macauba groves in these areas. Variability within plants from all sites indicate that they could be a reserve of genetic materials for further development of cropping systems aiming production of oil and other useful products.

A tropical forage breeding program contains several peculiarities, especially when it involves polyploid species and facultative apomixis. Urochloa spp. are excellent perennial forages, and the identification of superior genotypes depends on the selection of many characteristics under complex genetic control, with high cost and time‐consuming evaluation. Therefore, the use of tools such as multivariate analysis and diallel analyses could contribute to improving the efficiency of breeding programs. Thus, the objectives were to estimate (i) the contribution of additive and nonadditive effects on agronomical and nutritional traits in a population of interspecific hybrids of Urochloa spp., originated from a partial diallel between five apomictic and four sexual parents, and (ii) the accuracy of multivariate index selection efficiency. Genetic variability was detected between the parents, crosses, and hybrids for all the traits. There was no clear trend of the importance of the additive and nonadditive genetic effects on agronomical and nutritional traits. Furthermore, the predominant component of genetic variance changed depending on the characteristic. Moreover, there was no parent or cross that was outstanding for all traits simultaneously, showing the high variability generated from these crosses. The Mulamba and Mock index associated with principal components analysis allowed a more significant gain only for agronomic characteristics. However, the per se index, at the univariate level, promoted a more balanced response to selection for all traits.