The objective of this work was to analyse, by means of multivariate variance analysis, the behaviour of 16 leguminous tree species introduced into pastures of Brachiaria decumbens from unprotected young plants and under grazing, in four periods of the year, in Seropédica, RJ, Brazil. Nine variables, related to length and to number of sprouting, before and after animal grazing, were used for the evaluation. The statistical difference of the means of the principal canonical variable, calculated by the Scott-Knott test, indicated the formation of four groups, and the Mimosa tenuiflora group stood out at the 3rd and 4th evaluations. Difference among treatment means for each variable, calculated by Bonferroni confidence intervals, showed that the greatest sprouting length and the highest number of sprouting, after grazing, were found in M. tenuiflora. This leguminous species is indicated to be introduced into B. decumbens pasture of the region, with higher probability of success, without protection of the young plants and under grazing.
canonical variable; acceptability; silvipastural systems
