The aim of this study was to investigate the use of phenotypic and genotypic values in the path analysis involving agronomic characteristics in sugarcane at the stages of plant cane and ratoon cane. Under these two approaches it was calculated the direct and indirect effects of yield components - number of stalks per plot, diameter of stalk and average height of stalk - on sugarcane yield per hectare. To ensure a better consistency of results, data were obtained in two experiments in randomized blocks with five replications. The experiments were carried out at the Center of Sugarcane Experimentation of the UFV, Oratórios, MG, Brazil. Each experiment consisted of 22 full-sib families of sugarcane. The coefficients of determination were high in all path analysis, indicating that the evaluated components explain most of the variation in straw yield. Practically there was no difference in interpretations and conclusions from the analysis based on arrays of genotypic or phenotypic correlations. From the analysis of the direct phenotypic and genotypic effects, number of stalks was the variable that best correlated with sugarcane yield per hectare in both experiments and harvest stages, demonstrating the possibility of obtaining significant gains by means of indirect selection of sugarcane yield per hectare using number of stalks. There was variation in the estimates of correlation in the comparison between experiments, probably due to different origin of the families surveyed in each experiment.
correlation; early selection; indirect selection
