Abstract
More than 95% of Brazil nut production in Brazil comes from the extractive collection, with few plantations for commercial production. To contribute with Brazil nut tree genetic improvement, this study aimed to estimate number of fruits (NFP) repeatability coefficients; determine the appropriate number of measurements for efficient selection; and select Brazil nut promising trees from two Brazil nut stands in the state of Acre. The study was carried out in permanent plots in Filipinas (Resex Chico Mendes, Epitaciolândia-AC) and Cachoeira (PAE Chico Mendes, Xapuri-AC). The number of fruits per plant (NFP) was estimated at 140 Brazil nut trees from 2002 to 2014 (13 years), while in Cachoeira there were 175 Brazil nut trees from 2009 to 2014 (six years). Genetic-statistical analyzes were performed based on the mixed model methodology, using the Selegen-REML/BLUP Model 63. Twenty matrices were selected, 08 in Filipinas and 12 in Cachoeira. In both Brazil nuts stands, the permanent phenotypic variance (Vfp) represented the largest percentage of the individual phenotypic variance (Vf). The repeatability coefficient was considered moderate for Cachoeira and high for Filipinas. Considering a 90% coefficient of determination, it would take seven years of NFP assessment for Cachoeira and six years for Filipinas. The existing database for the Brazil nut production is sufficient to make selection for matrices with high fruit production. Based on the estimated values we expect the new population to have an average yield of 465.3 fruits.arv-1, indicating a gain in fruit production of about 2.4 times for Cachoeira and 4.2 for Filipinas.
Keywords:
Crop; Genetic improvement; Brazil nut