Fig. 2: Importance of spectral traits to detect Xf- and Vd-infection symptoms.
From: Divergent abiotic spectral pathways unravel pathogen stress signals across species
a–d Normalised importance of hyperspectral and thermal plant traits retrieved from the pool of spectral plant traits used to detect Verticillium dahliae, Vd- (a) and Xylella fastidiosa, Xf-induced infection symptoms (b–d) across olive (a, b) and almond (c, d) trees. For reference, the full list of spectral plant traits is available in the Supplementary Table 1. The importance analysis was carried out using a balanced training dataset obtained from n = 1878 (a), n = 7296 (b), n = 4048 (c), n = 2680 (d) trees by the permutation of out-of-bag (OOB) predictor methodology. The importance of each spectral trait was normalised by the highest importance obtained for each disease/species within each ML model. e–j Analysis of spectral plant traits measured in the field from asymptomatic vs. Xf-infected olive and almond leaves. e Temperature at midday (t, n = 1797 leaf samples). f Normalised phaeophytinization index-based spectral trait (NPQI, n = 1457 leaf spectral samples). g Anthocyanins (Anth, n = 1318 leaf samples). h Steady-state leaf chlorophyll fluorescence (Ft, n = 2784 leaf samples). i Normalised xanthophyll cycle dynamics index (Photochemical Reflectance Index [PRIn], n = 1457 leaf spectra samples). j leaf chlorophyll content (Ca+b, n = 2584 leaf samples). Statistical analyses were carried out by a one-way non-parametric ANOVA (Kruskal-Wallis test) followed by a Wilcoxon post-hoc test (two-sided) with Bonferroni correction to examine significant differences at p-value < 0.05 between the leaf groups for each species. Severity levels with the same letter are not significantly different (p-value ≥ 0.05). Exact p-values are provided in the repository indicated in the Data availability and Code availability sections that contain the source data and code. The horizontal black line in the boxplots displays the median, and the top and bottom horizontal lines represent the 75th and 25th percentiles, respectively. Whiskers display the lower and upper limits of interquartile ranges (Q ± 1.5xIQR).
