Statistical analyses

The data were subjected to analysis of variance (ANOVA), and means were compared using Tukey’s test at the 5% probability level; regression analysis was also performed. Using the SAS statistical software package 9.2 (SAS Institute Inc., Cary, NC), Levene’s test was used to test the homogeneity of variances in the treatments [54].

The effects of nitrogen on available volatiles were analyzed in relation to the number of molecules in the profile (profile density) in general or based on the group to which the substances belonged (monoterpenes, sesquiterpenes, and fatty acid derivatives) and in relation to the relative abundances of these substances. The profile of the volatiles was also expressed as the phytochemistry diversity index using the Shannon index (H).

The diversity index of the total volatiles (signaling and defense substances (mono- and sesquiterpenes)) was calculated using the following formula: H = −Σ (pi ln pi), where pi = the proportion of substance i in relation to the total volatile substances (relative abundance of the substance i) calculated by ni/N, where ni = the relative percentage of substance i and N = the sum of the percentage of total substances.

The diversity indices (H index) and the number and relative percentages of substances were subjected to ANOVA, and the means were compared using Tukey’s test at the probability level of 5%. Using SigmaPlot 12.0 (Systat Software, Inc. SigmaPlot for Windows), Levene’s test was applied to test the homogeneity of variances across treatments.

Principal component analysis (PCA), Pearson’s correlation coefficient and hierarchical cluster analysis (HCA), and similarity and Pearson’s coefficient (UPGMA) were performed with the relative percentages for substance identification at the different nitrogen concentrations and evaluation time points, and a different hierarchical cluster analysis (Ward’s algorithm, index of dissimilarity of Euclidian distance) was performed using the values of the chemical diversity at the four concentrations of nitrogen and evaluation time points by the software XLSTAT (2017).