Fig. 1

a Schematic picture for the location of rhizosphere soils sampling. b Contributions of deterministic and stochastic processes in community assembly within collected rhizosphere soil samples. βNTI calculation of phylogenetic turnover among diseased and healthy samples indicates that variable selection was more consistent in diseased soils. c The relative influence of each community assembly process among diseased and healthy samples was defined by the percentage of site pairs governed by each process. BD, diseased banana; BH, healthy banana (from Hainan); CD, diseased cucumber; CH, healthy cucumber (from Guangdong); WD, diseased watermelon; WH, healthy watermelon (from Beijing); LD, diseased lily; LH, healthy lily (from Hunan). e Contributions of deterministic and stochastic processes on community assembly within diseased and healthy soils of collected metadata. βNTI calculations of phylogenetic turnover between diseased and healthy soils indicate that variable selection has greater effects on disease than health. f The relative influence of each community assembly process between diseased and healthy soils as defined by the percentage of site pairs governed by each process. g Disease incidence of the first, fifth, and eighth generation. h Contributions of deterministic and stochastic processes on community assembly within pot experiment. i The relative influence of each community assembly process of rhizosphere soil samples from pot experiment