Fig. 1: Hypotheses and experimental design.

A Scenarios of disease incidence in the pathogen-only experiment. Plants are expected to die after pathogen inoculation, while (generally) not all plants show disease symptoms, and some plants survived up to the end of the experiment (binary disease outcome). B Scenarios of pathogen abundance in the rhizosphere of the pathogen-only experiment. Although plant individuals are exposed to the same initial abundance of the pathogen in the rhizosphere, compared to healthy plants, the diseased plants often have higher pathogen loads in the rhizosphere. Changes in pathogen abundance between healthy and diseased plants are detectable prior to the onset of disease symptoms. C Schematic figure of the hypotheses. Soil community state gradually changes as plants grow. We hypothesize that the microbiome can diverge toward different configurations out of an initial constant species pool, later affecting plant protection against disease. D Schematic figure of the experimental design. We used the rhizobox system [6] for nondestructive repeated sampling of individual plants at different growth stages. Soils associated with 54 plant individuals at the initial stage (IS; 2 days after planting (dap)), vegetative stage 1 (VS1; 15 dap), vegetative stage 2 (VS2; 27 dap), reproductive stage 1 (RS1; flowering stage; 40 dap), and reproductive stage 2 (RS2; fruiting stage; 60 dap) were collected. E We recorded the disease incidence, sampled rhizosphere soils, measured soil physicochemical properties, quantified pathogen abundance, determined bacterial community composition, and identified the discriminating OTUs associated with healthy and diseased plants.