GRS is one of the most concerned ginseng diseases at present. In recent years, GRS has appeared in large areas in some ginseng producing areas in China, seriously damaging the ginseng industry. As a root disease, a thorough understanding of its rhizosphere characteristics is essential [18]. In this study, to provide basis for soil improvement and disease control, we made a more detailed division of the disease (from HG to GRS4) to comprehensively reveal the pattern of rhizosphere changes in this ginseng disease's background.
Changes of microbial diversity and richness in rhizosphere
Through the statistics of the Shannon index and Simpson index of each group, we found that with the increase of ginseng disease degree, the change of bacterial community diversity in rhizosphere was firstly increased and then decreased. Observed species and Chao1 index indicated that the richness of the rhizosphere bacterial community fluctuated in these five degrees, and the highest richness was found in the GRS2 group (Table 1). This result seems to imply that GRS2 (red skin area 25 to 50%) is a grade of concern because of the turning point of changes in the ginseng rhizosphere's bacterial diversity.
For fungi, the fungal diversity in the severe groups (GRS3 and GRS4) was significantly lower than that in the HG group, while the richness did not change significantly (Table 1). Overall, the fungal diversity in the GRS's rhizosphere was affected, especially in severe groups.
Changes of microbial community composition in rhizosphere
To better reflect the nonlinear structure of the ecological data and the inter- and intra-group differences in the samples, we performed NMDS analysis based on OTUs level (Fig. 1). The bacteria in HG and GRS1 groups were separated from the other three groups, which means that with the development of the disease, the bacteria in rhizosphere had a continuous change process and tended to be stable at the later stages (from GRS2 to GRS4) of development (Fig. 1A). For the NMDS analysis of fungi, we still found that the HG group was significantly separated from the GRS3 group and the GRS4 group. Interestingly, the groups were not separated by the degree of disease development (Fig. 1B). The possible reason is that the fungal community develop fluctuating rather than gradual as the disease progresses. Further, by clustering analysis, we also found that the bacterial HG group was individually clustered; meanwhile, the other GRS groups were not incredibly significantly clustered, confirming the bacterial community's instability in the rhizosphere of the diseased ginseng (Fig. 2A). In terms of fungi, in addition to the HG group, the GRS3 group and GRS4 group also had obvious clustering, which seemed to indicate that the fungal microbial community in the severe groups again formed the stable state different from the HG group. (Fig. 2B). Based on the above, the bacterial community of ginseng rhizosphere appears to be more sensitive than the fungal community at the beginning of the disease. On the contrary, the fungal community has more stability compared to the bacterial community.
Characteristics of microbial community taxa variation in rhizosphere
Proteobacteria is the most dominant phylum among bacteria, and its abundance increases with the degree of disease (Fig. 2A and Fig. S2). The increase of Bacteroidetes abundance in GRS3 and GRS4 groups (Fig. S2) may be due to its good degradation ability, which is more suitable for survival in the fiercely competitive soil[19]. The abundance of Actinobacteria, Chloroflexi, and Gemmatimonadetes was reduced in severe groups (GRS3 and GRS4) (Fig. S2). Actinobacteria contribute to OM's the decomposition [20], but the ecological functions of Chloroflexi and Gemmatimonadetes were unclear. In the LEfSe analysis, 24 biomarkers were obtained and distributed in four groups (Fig. 3A). We also identified Acidobacteria as a potential biomarker for the HG group, which has been reported to degrade plant-derived OM specifically and is more abundant in soils of enriched plants [21]. In the GRS1 group, Rhizobiales, a clade of Alphaproteobacteria, were screened out and reported to fix nitrogen or as a pathogen[22]. Most biomarkers in the GRS3 group belongs to Proteobacteria, and the other belong to Bacteroidetes. Although described as phytopathogenic, the GRS4 group biomarker, Rhizobiaceae, is not known whether it is related to rust roots [23].
The fungal community's dominant phyla are Ascomycota, Mortierellomycota, and Basidiomycota (Fig. 2B). In severe groups, the dominance of Mortierellomycota decreased (Fig. S3). Besides, there are Glomeromycota and Mucoromycota that decrease in abundance as the disease progresses. In the present study, we still found changes in the abundance of Ilyonectria in different groups at the genus level. We also found the fungi genus (Trichosporon and Paraglomus), whose abundance decreases with disease development (Fig. S3). We also obtained specific biomarkers for each group through LEfSe analysis, such as Glomeromycota, Fusarium, and Chaetomium of the HG group (Fig. 3C). In the GRS3 group, biomarkers all belong to Ascomycota, the soil ecosystem's main fungal decomposer (Fig. 3D) [24, 25]. Among them, except Ilyonectria, the other biomarkers were not associated with GRS, although they have been reported to be associated with plant diseases [26, 27].
In general, we used statistical methods to look for changes in the composition of rhizosphere microbial communities with different disease degrees. Although taxa were found to be correlated (positively or negatively) with disease severity, their ecological function was unknown.
Changes of the microbial interactions in the networks
Co-occurrence networks can be constructed to analyze the interaction and co-existence patterns among different microorganisms, which is crucial to our further understanding of the changes in rhizosphere with GRS development [28].
Fig. 4 shows that the phylum with the highest connectivity in the five groups' bacterial community is Proteobacteria, which is consistent with previous study[29]. Therefore, with the disease's development, the keystone phylum has not changed in ginseng rhizosphere. However, it should be noted that Proteobacteria accounts for a larger proportion in GRS3 and GRS4 group. Since Proteobacteria could exploit labile carbon sources and produce extracellular polysaccharides to bond sand particles, this may mean that the rhizosphere of the severe groups is more solid, while has a stronger nutrient metabolism [30, 31]. Further, from the obtained topology parameters, we found that links and average degree were fluctuating (from HG to GRS4). The average path length of the GRS4 group was significantly lower than the other four groups (Table S3). Thus, with the disease's development, the bacterial community in the ginseng rhizosphere underwent multiple changes from complex and stable to simple and unstable. In addition, by counting the ratio of positive to negative links, we can also predict that there may be more interspecific competition and ecological niche separation in the GRS groups than in the HG group (Table S3) [32].
For the fungal networks, we can find that the keystone phylum of rhizosphere was Ascomycota, unchanged at all disease development stages. It can be seen that in ginseng rhizosphere, Ascomycota was essential for resisting the harsh environment and maintaining system stability [33]. Through the statistics of topology parameters, we found that from HG to GRS4, the number of links, clustering coefficient, and the average degree were the lowest in the GRS1 group, and then gradually increased (Table S3). Therefore, in the early stage (GRS1) of ginseng disease, fungal communities' interaction and stability decreased. Interestingly, after this, the soil's fungi gradually formed a complex and stable community (GRS4). From the change in the ratio of positively and negatively links, it appears that the competitive relationships and the taxa in the same ecological niche were severely affected in the GRS1 group (Table S3). Furthermore, the negative correlation in GRS groups was lower than that in the HG group. In previous studies, GRS was closely related to fungi [8] [34]. Here, based on the interaction network, we identified microorganisms that interacted closely with the reported potentially pathogenic fungi. Of these, Simplicillium, which is extremely negatively correlated with Ilyonectria, is reported as a biological control agent [35]. In addition, although these microorganisms, such as Musicillium and Arachnotheca, have not been fully studied, the establishment of the networks also provides a basis for the future search for microorganisms that have antagonistic interactions with pathogens and, thus, for biocontrol development.
Changes of physicochemical properties and their relationship to microbial communities
Changes in the soil's physicochemical properties directly determine the structural composition of the microbial community [36, 37]. In this study, pH, nutrient composition, and enzyme activity of rhizosphere were measured from HG to GRS4 group. From the GRS2 group, the pH of rhizosphere decreased significantly (Fig. S4). Compared with other groups, AP and TP in rhizosphere of the severe groups (GRS3 and GRS4) were reduced considerably. Also, AK's content was different among the groups, while there was no significant difference in other nutrients (Fig. S5). In addition, the activities of three enzymes (CAT, INV, and PHO) were significantly decreased in the rhizosphere of the severe groups (GRS3 and/or GRS4), while the activity of the URE was not significantly changed (Fig. S6). The results indicated that the development of GRS was correlated with the physicochemical properties of rhizosphere. The reasons for the above changes in physicochemical properties may be related to the occurrence of diseases affects the production of enzymes and the conversion of nutrients in the soil or increases their consumption[38, 39]. It is also associated with the release and accumulation of secretions from the root of diseased ginseng [2].
To explore the microbial-environmental linkages and gain insight into the key factors influencing microbial communities' changes, we used dbRDA to examine microbial communities and environmental factors [40, 41]. In the dbRDA results for bacteria, TP, AK, CAT, and AP were the main factors determining the distribution of bacterial community composition and taxa changes (Fig. 6A). In the dbRDA analysis of fungi, INV, CAT, and AP were significant factors affecting the structure change of the fungal community (Fig. 6B). This study suggests that AP and CAT may focus on future research on soil improvement due to their close correlation with both bacterial and fungal communities.