*2.5. Statistical Analysis*

Analyses of variance were performed using the Statgraphics Centurion ver. XVI software package (Statpoint Technologies, Inc., Warrenton, VA, USA). Prior to ANOVA, the residuals (standardized) of the data were visually tested with qq-plots, as well as with Shapiro–Wilk tests, using SPSS (IBM SPSS Statistics for Windows, Version 22.0, IBM Corp. Armonk, New York, NY, USA). Percentage values concerning disease severity were arcsine transformed prior to ANOVA. Significant di fferences between treatment means were compared by the protected least significant di fference (LSD) procedure at *p* < 0.05. Commonality analysis was performed in the R environment (version 3.4.3) using the "yhat" package (version 2.0–0) as described by Nimon et al. [46]. For a number k of predictors, CA returns a table of (2k-1) commonality coe fficients (or commonalities), including both unique and common effects [47]. In the case where the dependent variable y is explained by two predictors *i* and *j*, the unique effects are:

$$\begin{array}{l} \mathcal{U}(i) = \begin{array}{l} \mathcal{R}^2\_{y,ij} - \mathcal{R}^2\_{y,j} \\ \mathcal{U}(j) = \mathcal{R}^2\_{y,ij} - \mathcal{R}^2\_{y,i} \end{array} \tag{8} $$

and the common contribution I is:

$$\mathcal{L}(ij) = \, \prescript{\,}{R}\_{yjj}^2 - \, \prescript{\,}{\mathcal{U}}(i) - \mathcal{U}(j) \tag{9}$$
