Figure 3

Determination of the most discriminating volatiles and their correlation with volatiles collected during the day and at night from wild and cultivated tomato plants (abbreviated as CTDV cultivated tomato plant day volatiles, CTNV cultivated tomato plant night volatiles, WTDV wild tomato plant day volatiles and WTNV wild tomato plant night volatiles). The 25 most predictive volatiles are presented in bar graphs in order of decreasing importance based on: (a) mean decrease in accuracy (MDA) in the random forest analysis; (b) variable importance in the projection (VIP > 1) from the spasce partial least square discriminant analysis (sPLS-DA); and (c) similarity percentage (SIMPER) of the non-metric multidimensional scaling (NMDS). Twenty discriminating VOCs were considered (i.e., a combination of the top 10 of the discriminating VOCs highlighted by the three variable selection methods (RF, sPLSDA, NMDS)). Using the release rates of the volatiles, differentiation of the volatile treatments was analyzed through: (d) a multidimensional scaling (MDS) ordination plot showing the distribution of the volatile treatments, (e) an sPLS-DA biplot and (f) an NMDS biplot, both showing the correlation of the discriminating VOCs with the volatile treatments.