The experimental design was a randomized complete block with six factors (cvs) and three blocks (replicates). The analysis of variance was performed according to the General Linear Models (GLM) procedure developed by the Statistical Analysis Systems Institute (SAS Inst., V.6.1, Cary, NC, US). Means and standard errors were calculated. Correlations were estimated using Person’s correlation coefficient, P < 0.05. LSD test was applied to establish significant differences between means with a confidence level of 95%.