Fig. 7.
Multivariate analysis of glucosinolate dynamics in leaf tissues in response to feeding by Pieris rapae caterpillars in plants originating from three wild cabbage populations, a KIM, b OH, and c WIN in Dorset, England. GSL data were subjected to OPLS by means of partial least squares projections. The horizontal axis coincides with increased duration of feeding from left to right. Plants were grown in a greenhouse and sampled repeatedly at t = 0, 4, 8 and 16 days since the introduction of ten first instar P. rapae caterpillars. The full names of aliphatic GSLs compounds are given in Fig. 1 and the “Glucosinolate analysis” section. tot aliphat totals of aliphatic GSLs, tot indole total of indole GSLs, total grand totals of GSLs, % aliphat percentage aliphatic of total, 4Pent 4-pentenyl GSL. Arrows point at variables of which the correlation coefficient with time is significantly different from 0 and at the same time these variables contributed significantly to the separation of samples in relation to time based on model variable importance values (VIP). Variables with VIP > 1 are highly influential (Eriksson et al. 2006). Model statistics for KIM: overall significance statistical model F2,36 = 33.0, P < 0.001, OPLS predictive statistics R2X = 0.347, R2Y = 0.712, Q2 = 0.647; for WIN: overall model significance, F2,40 = 49.3, P < 0.001; OPLS predictive statistics R2X = 0.317, R2Y = 0.743, Q2 = 0.711; for OH: overall model significance, F2,32 = 35.1, P < 0.001; OPLS predictive statistics R2X = 0.325, R2Y = 0.719, Q2 = 0.687