Alternativas para atenuar o efeito da perda de plantas em parcelas experimentais de cafeeiros arábica

Abstract

There are many questions about the methodology used in the evaluation of coffee (Coffea arabica) progenies, and one of them is how the occurrence of plant loss on experimental plots affects selection. The compensatory effect of crop cultivation is another issue that needs to be considered, because in the event of plant failure in the plot, the plant adjacent the failure usually shows a higher production yield, owing to less competition. Thus, this study aimed to determine the magnitude of compensatory effect on coffee crop, verify the need for correction for the number of failures in coffee tree experiments, and determine the best to perform the correction. Production data from 6 crops (2001–2006) from 11 coffee experiments conducted by the Agricultural Research Company of Minas Gerais (EPAMIG), Brazil, were used for analysis. In these experiments, progeny in segregated generations and of the ‘Mundo Novo,’ ‘Catuaí,’ and ‘Icatu’ groups were evaluated. The experiments were implemented in a randomized block design with four replicates, and the experimental plots consisted of 6 plants with no failures. The methods for evaluating stand adjustment were as follows: no correction, rule of three, Zuber (1942), Vencovsky and Cruz (1991), covariance by the ideal or medium stand methods, and the methodology proposed by Cruz (1971). The estimates of the mean, standard deviation of the mean, p value, Bayesian information criterion, accuracy, genetic correlation, and index of coincidence were used to verify the magnitude of the effect of each method in relation to the data without correction. There is a compensatory effect on the cultivation of coffee tree and it varies between genotype groups, sites, and years. Stand correction is advantageous in experiments of genotype competition in coffee cultivation, and in this study, covariance methodologies using a medium or ideal stand showed the best results, followed by the methodology proposed by Cruz (1971). The rule of three and Zuber (1942) methods did not provide satisfactory results and thus are not recommended for stand correction in experiments of genotype competition in coffee cultivation.