Skip to main content
. 2019 Oct 25;10:1355. doi: 10.3389/fpls.2019.01355

Figure 9.

Figure 9

(A) Predicted vs. observed Septoria tritici blotch (STB) severity levels of 36 artificially inoculated wheat plots. Mean and standard error of predictions are shown. Data from all experimental plots (n = 72) was used to tune/train the model, but reported performance estimates are based only on artificially inoculated plots (n = 36) in order to avoid overly optimistic performance estimates resulting from a good prediction of disease severity in control plots. STB severity was measured on flag leaves using a combination of visual incidence scorings and scans of flag leaves exhibiting disease symptoms. Predictions were obtained from a cubist regression model based on spectral–temporal features for the same 36 plots and 36 non-inoculated control plots sown with the same genotypes. The broken red line represents the 1:1 line, the blue line represents the least squares line of the linear regression of predicted vs. observed values, and the gray area represents the 95% confidence interval of the least squares line. (B) Spatial distribution of predicted STB severity levels of ∼360 largely disease-free plots of the GABI wheat panel, grown next to the plots used as training dataset. White fields correspond to the plots contained in the training dataset.