2.4. Experimental design
The experimental design was completely randomized with the
factors nitrogen blanket in headspace (with or without) and storage
time (0, 60, 120 and 180 days), resulting in eight treatments with
three replicates each. The instrumental data and acceptance data of
BO samples were analyzed by ANOVA and when significant, Fisher
test (p < 0.05) was applied. Free Choice Profile data were analyzed
by Generalized Procrustes Analysis (GPA), with Euclidean transformations
of the data by rotation, translation and auto scaling,
followed by Principal Component Analysis (PCA) of the consensus
configuration, with 5% significance.