Phenolic Acid-Degrading Consortia Increase Fusarium Wilt Disease Resistance of Chrysanthemum

: Soil microbial community changes imposed by the cumulative effects of root-secreted phenolic acids (PAs) promote soil-borne pathogen establishment and invasion under monoculture systems, but the disease-suppressive soil often exhibits less soil-borne pathogens compared with the conducive soil. So far, it remains poorly understood whether soil disease suppressiveness is associated with the alleviated negative effects of PAs, involving microbial degradation . Here, the long-term monoculture particularly shaped the rhizosphere microbial community, for example by the enrichment of beneficial Pseudomonas species in the suppressive soil and thus enhanced disease-suppressive capacity, however this was not observed for the conducive soil. In vitro PA-degradation assays revealed that the antagonistic Pseudomonas species, together with the Xanthomonas and Rhizobium species, significantly increased the efficiency of PA degradation compared to single species, at least partially explaining how the suppressive soil accumulated lower PA levels than the conducive soil. Pot experiments further showed that this consortium harboring the antagonistic Pseudomonas species can not only lower PA accumulation in the 15-year conducive soils, but also confer stronger Fusarium wilt disease suppression compared with a single inoculum with the antagonistic bacteria. Our findings demonstrated that understanding microbial community functions, beyond the single direct antagonism, facilitated the construction of active consortia for preventing soil-borne pathogens under intensive monoculture.


Introduction
Monoculture is the most common agricultural practice that repeatedly cultivated the same plants in soils without rotational cropping, due to limited arable lands and huge demand for foods and other economic plants. However, consecutive monoculture is not sustainable, as soil-borne pathogens rapidly emerge and cause serious yield losses [1,2]. Soil microbial community changes are responsible for pathogen establishment and invasion under monoculture systems [3,4]. Soil microbial diversity and composition are the essential components of soil heath [5]. Therefore, soil microbial community can protect plants from certain phytopathogens, which result in disease-suppressive soils [6,7].
Soil disease suppressiveness primarily relies on the mechanisms of nutrient competition and specific antagonism caused by soil micro flora [8]. This disease-control capability can be abolished by soil sterilization, and is transferred into the disease-conducive soil by mixing the disease-suppressive soil, indicating that the microbial community confers the property of the soil to prevent the build-up of pathogens and its invasion [9,10]. The phenomenon of disease-suppressive soil has been observed disease suppressiveness. Through the prediction of indicator species, a model consortium was constructed on the basis of microbial antagonism and PA-degrading activity. The performances of alternative community assemblies predicted variations in the PA-degrading efficiency and were correlated with their degrading activities. Furthermore, an efficient PA-degrading consortium conferred stronger Fusarium wilt disease suppression as compared to single inoculum with antagonistic bacteria. This function-based assembly of optimized consortia provided an important strategy for overcoming Fusarium wilt disease under monoculture systems.

Soil Sampling
Two typical fields were continuously cropped with monoculture Chrysanthemum plants (C. morifolium cv. Chuju) for more than 15 years in Anhui province, China. Both two fields had similar soil properties and management regimes, but Fusarium wilt disease incidence significantly differed between the two Chrysanthemum plantations. The first site (conducive soil) was situated in the plantations of Shiji (32°17′ N, 118°06′ E), which had over 70% Fusarium disease incidence over the last three years. The second site (suppressive soil) located in the plantations of Jutai (32°13′ N, 118°23′ E), had less than 20% wilt disease incidence over the last three years. In the same regions of the two sites, the soils planted with C. morifolium cv. Chuju for 1 year were used for this study. No significant difference in disease incidence was observed between the 1-year conducive and suppressive soils. The two fields were 19 km apart at geography. For each site, 3 random subplots were chosen and 12 random cores (0-20 cm in depth) from each subplot were pooled to form one sample, leading to 3 samples per site.

Soil Disease Suppressiveness and Fusarium Wilt Incidence in Pot Assays
Soil samples were collected from the conducive and suppressive soils in the 1-year and 15-year monoculture for pot experiments. To examine whether soil disease suppressiveness was attributed to microbial community rather than soil physicochemical traits, soil disease-suppressive capability was evaluated as described by Mendes et al. [9] Four treatments were compared: 15-year conducive soil (Con15), 15-year suppressive soil (Sup15), Con15 amended with 10% (w/w) of Sup15, and Con15 mixed with 10% (w/w) of sterilized (autoclaving at 121 °C for 60 min) Sup15. The soils were poured into the pots, and three-week-old Chrysanthemum seedlings were planted in each pot. Treatments were replicated three times and each replicate contained 20 pots. These plants were cultured at 28 °C/25 °C (light/dark), and were recorded for the incidence of Fusarium wilt disease after three months. Disease incidence was calculated as the percentage of Fusarium-infected plants in total plants. To further verify that soil disease suppressiveness was attributable to induced or natural suppressiveness, disease incidence was calculated in the Con15 mixed with the Sup1 or Sup15 (w/w, 9:1).

Pyrosequencing and Quantification of Rhizomicrobial abundance
Soil DNA samples were isolated from 0.25 g of rhizosphere soils from each of three plants using a DNeasy® PowerSoil® DNA Isolation Kit (Mobio Laboratories Inc., Carlsbad, CA, USA), according to the manufacturer's instructions. DNA libraries were constructed by amplification of bacterial 16S rRNA (515F:′-GTGCCAGCMGCCGCGGTAA-3′; 907R: 5′-CCGTCAATTCMTTTRAGT TT-3′) and fungal ITS regions (ITS5: 5′-GGAAGTAAAAGTCGTAACAAGG-3′; ITS2: 5′-GCTGCG TTCTTCATCGATGC-3′) as described by Guo et al. [33] Moreover, the sequences of Illumina Nextera adapters were ligated to the 5' end of the universal primers for sequencing. The sequencing was carried out by the Magigene Biotechnology Co., Ltd. (Guangdong, China) on the Illumina MiSeq platform. Raw sequencing reads were deposited in the NCBI Sequence Read Archive (SRA) database, under accession number SRP188849 for bacteria and SRP188827 for fungi. All the raw reads were processed for removing low quality sequences using the Quantitative Insights Into Microbial Ecology (QIME) software package, and operational taxonomic units (OTUs) were obtained according to the methods described by Guo et al. [33] Quantitative real-time PCR (qPCR) was performed for quantifying bacterial, fungal, Pseudomonas putida and F. oxysporum abundances, using the SYBR ® Premix ExTaq TM kit (Takara, Shiga, Japan). Overall, 25 μL of qPCR reaction mixture contained 0.5 μL of forward or reverse primers, 1 μL of soil DNA samples, 0.5 μL of ROX Reference Dye II (50 X), 12.5 μL of SYBR ® Premix ExTaq TM and 10 μL of ddH2O. qPCR reaction conditions were conducted in an ABI 7500 machine as follows: 30 s at 95 °C, followed by 40 cycles of 5 s at 95 °C, and then 40 s at 60 °C. Standard curves for quantifying the gene copy number of P. putida and F. oxysporum were obtained based on the previously reported methods, respectively [4,34]. The primers for Xanthomonas sp. NC8 (5'-CGTGCGTAGGTGGTG ATTTA-3' and 5'-ATTCCGCTACCCTCTACCTC-3') and Rhizobium sp. HU16 (5'-CATGTCCTCAT GG CCCTTAC-3' and 5'-TGCAATCCGAACTGAGATGG-3') were designed for qPCR analyses.

Detection of PA Content, Microbial degradation, Mcrobial inoculation and Colony counting
To extract soil PAs, 5 g of rhizosphere soil samples was incubated in a flask with 5 mL of 1 M NaOH for 16 h, and were then shaken at 220 rpm for 1 h. After centrifugation at 8000 g for 10 min, the supernatant was adjusted to pH 2.5 and incubated at room temperature for 2 h, and then were centrifuged at 8000 g for 15 min. Subsequently, the supernatant was filtered through 0.22 μm membrane and detected by high performance liquid chromatography (HPLC), as previously reported by Zhang et al. [23] To assess soil PA-degrading capability, 20 g of fresh soil samples were placed in a glass container containing 0.5 mg/g phenolic acid (ferulic, p-coumaric or phydroxybenzoic acid). The soils were incubated at 28 °C in dark for 50 d, and sterile soils were used as control. Soil samples were taken every 10 d for measuring the remaining PAs.
Different bacteria strains were isolated from rhizosphere soils of three-month-old Chrysanthemum plants cultured in the 15-year suppressive soils by the dilution plating method, and several isolates were identified by sequencing 16s rRNA genes, including P. putida A2 (Genbank No. MK645976), Xanthomonas sp. NC8 (Genbank No. MK645977), Paenibacillus sp. M7 (Genbank No. MK645979), Stenotrophomonas sp. YB9 (Genbank No. MK645981) and Rhizobium sp. HU16 (Genbank No. MK645993). These strains was stored in LB medium at −80°C for the next experiments. For examining the ability of bacteria to degrade PA, 1 mL of bacterial strains at 5 × 10 6 CFUs was added into 100 mL of inorganic medium containing 0.9 g/L PA, and shaken at 150 rpm and incubated at 28°C, according to the method of Zhang et al. [23], with minor modifications. The supernatant of culture solutions was used for assessing the PA degradation, and the precipitate was resuspended in sterile water for measuring the values of OD600. Moreover, the degradation of 0.9 g/L mixed PAs (1:1:1) performed by 16 consortia (random combination of A2 with NC8, M7, YB9 or HU16) were measured by HPLC.
To assess the growth performance of A2 in the consortia, the plasmid pJB861 was firstly transferred into this strain, and A2 harboring the pJB861 with a kanamycin resistance gene was then used to construct different combinations with other strains. As described above, the PA content in the supernatant of culture solutions was measured, and the precipitate was resuspended in sterile water and subjected to serial dilution. Serial dilutions from different combinations were spread on the Luria-Bertani (LB) medium, supplemented with 50 μg/mL of kanamycin and incubated at 28°C to enumerate A2 colonies. To perform soil inoculation, different suspensions of bacterial strains were harvested and diluted in sterile water to obtain indicated inoculum concentrations, and then the bacteria dilution solution was poured into the soils.

Effects of PAs on the Growth of F. oxysporum
PAs including ferulic, p-coumaric and p-hydroxybenzonic acids, or mixed PAs (ferulic, pcoumaric and p-hydroxybenzonic acids, 1:1:1), were prepared at concentrations of 0, 60, 180 and 540 μmol/L, respectively. F. oxysporum was inoculated onto the center of 9 cm Petri dishes, containing PDA medium supplemented with different final concentrations of single PA or mixtures (0, 60, 180 and 540 μmol/L). Each treatment had ten replicates. After 1 week of incubation at 28 °C, the growth of F. oxysporum was measured. Spores of F. oxysporum was collected and then added into an 8-fold dilution of potato-sugar-agar (PSA) broth medium. After 1 week of culture with a shaker with 200 rpm at 28 °C, the spores were counted by a hem cytometer as described by Wu et al. [22]

Bioinformatics and Statistical analyses
The Chao1 richness and Shannon diversity of rhizosphere microbial community were estimated by the PAST software. Linear discriminant analysis (LDA) effect size (LEfSe) (Error! Hyperlink reference not valid.) analyses were applied to characterize the bacterial and fungal taxa significantly related to disease-conducive and -suppressive soils. The Kruskale-Wallis test was employed at α = 0.05, and the logarithmic LDA score (> 2) was used for discriminative features. Taxonomic cladograms illustrated significant differences between the conducive (Con15) and suppressive (Sup15) soils. An analysis of differential OTU abundance was carried out for selecting microbial indicator OTUs, and the Wald test was applied to determine significance (adjusted P value < 0.05) according to the method reported by Guo et al. [33] and these differential OTUs were then visualized by the Manhattan plots. Dominant OTUs (top 30 indicator OTUs based on relative abundance in the Sup15) were selected as indicator OTUs. Representative sequences of these indicator OTUs were used for the phylogenetic analyses through the neighbor-joining tree. The phylogenetic tree was drawn using iTOL (https://itol.emble.de). In addition, analyses of disease incidence, soil pH, richness, and diversity indices in soils were carried out by Tukey's HSD multiple range test (p < 0.05). Statistical analyses were performed in SPSS v16.0 (SPSS Inc., Chicago, IL, USA).

Soil Characteristics and Fusarium Wilt Disease Incidence
Soil physiochemical characteristics in the two fields with one-or 15-year consecutive Chrysanthemum monoculture were examined (Table S1). The one-year suppressive soil (Jutai site) displayed higher OM, P and Fe content than the one-year conducive soil (Shiji site). Soil pH and EC was also higher in the 1-year conducive soil than the one-year suppressive soil. Moreover, the 15year suppressive soil (Jutai site) exhibited higher OM, N, P and Fe content than the 15-year conducive soil (Shiji site). Similar results were also observed for the soil pH and EC. In pot experiments, threeweek-old Chrysanthemum seedlings were transplanted into the soils, and Fusarium wilt incidence of Chrysanthemum was significantly lower in the suppressive soil after three months of culture ( Figure  1a), showing only 17% disease incidence compared to about 68% disease incidence in the conducive soil ( Figure 1b), indicating that the soils from the two fields differed distinctly in the suppressiveness against F. oxysporum.
To examine whether the capability of soil disease suppressiveness could be transferred, the 15year conducive and suppressive soils were mixed in a ratio of 9:1. As shown in Figure 1b, transferring 10% of the suppressive soil (Sup15) into the conducive soil (Con15) markedly inhibited disease incidence, although the Con15 mixed with 10% (w/w) heat-treated suppressive soil (Sup15HT) shared similar disease incidence with the untreated conducive soil. We further assessed a microbial basis of disease suppression that resulted from the induction of crop monoculture or soil natural property. Pot experiments were designed in which the 1-year (Sup1) or 15-year (Sup15) suppressive soil was mixed with the Con15 in a ratio of 1:9, respectively. Then, three-week-old Chrysanthemum seedlings were transplanted into the soils and grown for three months. Transferring 10% of the Sup1 into the Con15 could not reduce Fusarium wilt incidence. However, transferring 10% of the Sup15 into the Con15 obviously suppressed disease outbreak. Consistently, lower F. oxysporum density occurred in the rhizosphere soils of Con15 mixed with the Sup15, than that with the Sup1 (Figure 1c). These data indicated that long-term monoculture may shape specific microbiota in the suppressive soil against pathogen establishment.

Rhizosphere Microbial Community Composition
Microbial community compositions in the Chrysanthemum rhizosphere were analyzed at the phylum level. As for bacterial communities, the conducive and suppressive soils shared the similar compositions with some major phyla, including Proteobacteria, Actinobacteria, Bacteroidetes, Acidobacteria and Chloroflexi (Figure 2a). The most dominant phylum was Proteobacteria, accounting for about 46%-55% of all bacterial sequences, following by Actinobacteria, Acidobacteria, Firmicutes, Bacteroidetes and Chloroflexi. Other minor phyla, such as Cyanobacteria, Gemmatimonadetes and Patescibacteria were identified, accounting for less than 7% of bacterial communities. However, there were a few differences among different soil samples with more Proteobacteria and Firmicutes in the suppressive soils, and the phylum Proteobacteria was most represented in the Sup15 compared with other soils. Concerning fungal communities, dominant OTUs mainly belonged to the two phyla Ascomycota and Basidiomycota, accounting for about 75% of the total sequences (Figure 2b). Ascomycota were more abundant in the suppressive soils than that in the conducive soils, but the opposite trend was observed for Basidiomycota. The relative abundance of Chytridiomycota, Glomeromycota and Zygomycota was less than 1% of fungal communities.

Rhizosphere Microbial Community Abundance and Diversity
The compositions of rhizosphere microbial community were compared among different soil samples by the qPCR and sequencing of 16s rRNA or ITS genes. As shown in Figure S1a,b, the highest bacterial and fungal abundance was present in the one-year conducive soil (Con1), and the microbial abundance was remarkably decreased in the 15-year conducive soil (Con15). A similar changing trend was observed for microbial diversity (Shannon index) and richness (Chao1 index) in the conducive soils, except for fungal diversity (Figure S1c-f). In contrast, the suppressive soils with 1year or 15-year consecutive monoculture harbored lower microbial abundance, richness and diversity (Figure S1a-f). However, the bacterial (Figure S1a,d) and fungal (Figure S1b,f) abundance of the suppressive soils appeared to be less impacted by long-term monoculture compared with the conducive soils. Additionally, non-metric multi-dimensional analysis (NMDS) indicated that different soil samples were observably separated ( Figure S2).
The differences of unique and shared OTUs were further investigated among different soil samples. Venn diagrams showed 802, 227, 190 and 117 unique bacterial OTUs in the Con1, Con15, Sup1 and Sup15, respectively, and 176, 99, 91 and 47 unique fungi OTUs were detected for these soils, respectively. Furthermore, 264 bacterial and 155 fungal OTUs were commonly shared in the conducive and suppressive soils (Figure 3a,b). As for bacteria, the relative abundance of Acidobacteria and Bacteroidetes in the unique OTUs was strikingly lower in the suppressive soils compared with the conducive soils ( Figure 3c). However, the relative abundance of Firmicutes and Patescibacteria was relatively higher in the suppressive soils than the conducive soils. The majority of shared OTUs (average relative abundance > 5%) in these soils belonged to Acidobacteria, Actinobacteria and Proteobacteria, while the Sup15 had the highest Proteobacteria abundance and lowest Actinobacteria abundance, compared with the conducive soils ( Figure 3d).
As for fungi, the majority of unique OTUs (average relative abundance > 5%) in these soils belonged to Ascomycota and Basidiomycota, and the Sup15 had the highest Ascomycota abundance (Figure 3e). In addition, long-term monoculture led to marked decreases in the relative abundance of shared OTUs belonging to Ascomycota in the conducive soils, but no marked changes were observed for the suppressive soils ( Figure 3f). Hence, the conducive and suppressive soils shaped differential microbial communities under monoculture systems, as evidenced by the changes in both the specific (unique OTUs) and core microbiome (shared OTUs).

Bacterial Indicators of Disease Suppression to Fusarium Wilt Disease
LEfSe analyses were employed to explore the relationships of the top 30 abundant bacterial and fungal genera in both the 15-year suppressive and conducive soils. The bacterial genera, including Pseudomonas, Rhizobium, Bryobacter, Occallatibacter, Pseudolabrys and Actinospica, were more abundant in the suppressive soil ( Figure 4). Pseudomonas was the most dominant bacterial phylum in the suppressive soil, accounting for 13.2% of the total sequences in the suppressive soil (only 0.1% in the conducive soil). For the fungi, 19 genera, out of the top 30 fungal genera, were more abundant in the conducive soils. The fungal genera, including Mortierella, Ceratobasidium, Gliocladiopsis and Cylindrocladium, were more abundant in the suppressive soil ( Figure S3). Fusarium was the most dominant genus, comprising 9.3% of total fungal genera in the conducive soil (only 0.5% in suppressive soil). As shown in Figure 5a-d, a total of 98 and 145 depleted bacterial OTUs specifically occurred in the Sup15 compared with the Sup1 and Con15, and 28 depleted bacterial OTUs were commonly shared in the suppressive soil (Table S2). Moreover, a total of 53 and 92 enriched bacterial OTUs (compared with the Sup1 and Con15) specifically occurred in the Sup15, and 37 shared OTUs were defined as indicator OTUs that were highly abundant in the Sup15 (Table S3). These shared OTUs belonged to the phyla Bacteroidetes, Patescibacteria, Chloroflexi, Actinobacteria, Firmicutes and Acidobacteria, respectively (Figure 5e). Moreover, some bacteria strains were isolated from the 15-year suppressive soil, and were identified by sequencing 16s rRNA genes. Among these isolates, five bacterial strains, corresponding to the observed OTUs that were highly enriched in the Sup15, were selected for the next experiments, based on at least 97% similarity for 16S rRNA genes. One strain was identified as P. putida A2, corresponding to OTU2 that was the most abundant in the Sup15. Other strains were Xanthomonas sp. NC8 (OTU1881), Paenibacillus sp. M7 (OTU308), Stenotrophomonas sp. YB9 (OTU5), and Rhizobium sp. HU16 (OTU19), respectively. However, only P. putida A2 exhibited a strong inhibitory effect on the growth of F. oxysporum ( Figure S4).

Functional Selection of Consortia Associated with Soil Disease Suppressiveness
Considering that root-released PAs were the pivotal stimulators of Fusarium wilt incidence, we focused on the selection of potential combinations that could antagonize F. oxysporum and degrade PAs. In this study, the content of PAs in soils was quantified by HPLC. The suppressive soil accumulated less ferulic, p-coumaric and p-hydroxybenzonic acids than the conducive soil under 15-year monoculture, while no significant difference was detected in the 1-year monoculture ( Figure 6). Among the three PAs, ferulic acid displayed the greatest promoting effects on the growth of F. oxysporum. By contrast, the growth-promoting effects of the mixed PAs were stronger than that of single PA ( Figure S5a). Moreover, the mixed PAs markedly stimulated the sporulation of the fungal pathogen, and this effect was further enhanced, followed by the increased concentrations of the mixed PAs ( Figure S5b). Additionally, degradation assays were performed for assessing the efficiency of soils to degrade PAs. As shown in Figure 6b-e, more rapid degradation of PAs occurred in the suppressive soils. By contrast, no or weak degradation was found in both the conducive and sterilized soils, eliminating the possible contribution of abiotic factors to the degradation of PAs. The content of PAs was notably lower in the suppressive soils than the conducive soils at 50 d of culture (Figure 6f). Thus, these results indicated that the suppressive soils may harbor more efficient PAdegrading consortia than the conducive soils. We further examined the capability of the above isolates to degrade PAs. Bacteria culture was inoculated into the medium containing 0.9 g/L ferulic, p-coumaric or p-hydroxybenzonic acid. During 36 h of culture, the content of PAs in the supernatant was measured. All five isolates could consume PAs in the medium, but varied considerably at the degrading levels ( Figure S6). To compare the efficiency of different combinations to degrade the mixture of PAs, we constructed multi-species assemblies, representing all possible alliance that harbored P. putida A2, only an antagonistic bacterium (Figure 7a). The Pseudomonas species was used as a core point, allowing the prediction of positive or negative interactions of other species with P. putida A2. For each community, two core indexes of community performances were examined, including both the PA-degrading efficiency and P. putida A2 growth. As shown in Figure 7b, the performances of different species' assemblies were compared, and variations among different combinations were evaluated for the performance indexes. The control group, containing only P. putida A2, had the least performances, while multi-species assembly improved performances in some cases (Figure 7c,d). All top performing combinations shared three species: P. putida A2, Xanthomonas sp. NC8 and Rhizobium sp. HU16.

Efficient Suppression of Fusarium Wilt Disease by a PA-degrading Consortium
We examined whether P. putida A2 could suppress F. oxysporum-induced wilt disease in Chrysanthemum plants. This strain, at 2 × 10 7 CFU/g of soil, was poured into the base of three-weekold seedlings grown in the 15-year conducive soils (Figure 8a). Three months later, the plants inoculated with P. putida A2 exhibited a significant decrease in disease incidence, compared with non-inoculated (control) plants. However, in the second and third planting (cycle) into the same pots, the suppressiveness of soils to Fusarium wilt was greatly weakened (Figure 8b). The dynamics of P. putida and F. oxysporum in the rhizosphere soils were tracked by qPCR of DNA copy numbers. The population level of P. putida in the soils rapidly decreased for three months and stabilized at 1.6 × 10 5 CFU/g soil. In the second cycle, the population level of P. putida declined to 2.5 × 10 4 CFU/g soil. In the third cycle, the population level of P. putida almost returned to the same level of control soils (Figure 8c). However, the opposite trend of population changes was observed for F. oxysporum (Figure 8d). Figure 8. Effects of P. putida A2 or a PA-degrading consortium (PAC) on disease incidence in the 15year conducive soils (Con15). (a) Experimental scheme for pot designs. P. putida A2 at 2 × 10 7 CFU/g or PAC harboring P. putida A2 (2 × 10 7 CFU/g), Xanthomonas sp. NC8 (5 × 10 6 CFU/g) and Rhizobium sp. HU16 (5 × 10 6 CFU/g) was poured into the base of three-week-old Chrysanthemum seedlings, grown in the 15-year conducive soils. After three months, three-week-old seedlings were subjected to the second and third successive plantings (cycles) into the same pots. Disease incidence (b), A2 growth (c), and Fusarium density (d) were determined in the rhizosphere soils. CP, consecutive planting (nontreated Con15); the A2-or PAC-treated soils were subjected to first (CP1), second (CP2) and third (CP3) three-month planting. Data are means ± SE. Different letters above the bars indicate statistically significant differences by Tukey's HSD test.
We further tested whether a PA-degrading consortium (PAC) conferred increased suppression of conducive soils to Fusarium wilt disease. The 15-year conducive soils were inoculated with a model consortium harboring P. putida A2 (2 × 10 7 CFU/g), Xanthomonas sp. NC8 (5 × 10 6 CFU/g) and Rhizobium sp. HU16 (5 × 10 6 CFU/g). After three months, the PAC-treated soils had lower levels of PAs than the A2-treated soils ( Figure S7). As shown in Figure 8b, the PAC-treated soils also displayed lower disease incidence than the A2-treated soils. In the third cycle, the consortium-treated soils still displayed stronger disease suppressiveness than the A2-treated soils. Consistently, the PAC-treated rhizosphere soils had relatively higher density of A2, compared with the A2-treated soils (Figure 8c). It was also found that both NC8 and HU16 effectively colonized in the PAC-treated rhizosphere soils during three-cycle monoculture ( Figure S8). Furthermore, the PAC-treated rhizosphere soils had a lower density of F. oxysporum than the A2-treated soils during three-cycle monoculture (Figure 8d).

Discussion
Fusarium wilt is the most wide-ranging fungal disease that leads to severe yield loss in many crops such as banana, vanilla and cucumber [2,33,35]. Dissecting characters of rhizosphere microbial communities is essential for understanding the microbial basis of disease-suppressive soil and improving plant fitness [36]. However, most researches on the disease-suppressive soil are confined to one microbial population, ignoring the remaining microbial communities [12]. In this study, the inoculation with P. putida A2 made the soil acquire disease suppressiveness, but this seemed not to be as durable as naturally suppressive soils. We further compared analyses of rhizosphere microbial community changes between the conducive and suppressive soils under long-term monoculture. Several indicator species were screened, which were associated with soil disease suppressiveness. Combined with conventional methods of bacterial isolation, a model PA-degrading consortium was constructed, conferring stronger suppressiveness to Fusarium wilt disease.
Bacterial diversity and abundance are pivotal for maintaining plant health by diverse tactics such as the prevention of pathogen invasion, promotion of plant growth, and activation of host immune responses [37][38][39]. The suppressive soil often harbors strikingly higher bacterial abundance and diversity than the conducive soil [15]. In contrast, our suppressive soil displayed lower bacterial diversity and abundance. These findings were in contrast with a recent study, showing that the diversity of bacterial communities was markedly greater in the suppressive soil of potato, compared with the conducive soil [36]. The lower diversity of bacterial communities in our suppressive soil may have resulted from lower pH conditions, which was similar to the observation that bacterial diversity was distinctly increased in higher pH soils [41][42]. Fungi are also an integral component of microbes in the soil ecosystem, which are important for soil functions and plant health [40]. Consecutive monoculture can shift the diversity and composition of a soil fungal community [17,41,42]. Xiong et al. [17] have shown that vanilla monoculture notably increased soil fungal diversity. Moreover, soil amendment with p-coumaric acid, a key stimulator of Fusarium wit disease incidence, increases the diversity of fungal community [41,42]. The fungal diversity in the conducive soils of Chrysanthemum was obviously increased under long-term monoculture. However, the opposite trend was observed for the suppressive soils. Additionally, the fungal richness in the suppressive soils was markedly lower than that in the conducive soils, but that was less impacted by long-term monoculture. Soil fungal richness has recently been shown to be positively associated with disease incidence and yield decline, indicating that higher fungal richness may be an important driver for Fusarium wilt disease incidence [2,43]. Additionally, soil characteristics are also important indicators of disease suppressiveness. Here, the levels of OM, available N, P and Fe were observably higher in the suppressive soil, compared with the conducive soil. In the potato monoculture system, the abundance of Fusarium pathogens has been found to be negatively related to soil OM [44]. Higher available P in soils is also positively correlated with less incidence of Fusarium wilt disease in the suppressive soil of banana [2]. Furthermore, a higher level of OM and nutrient elements in the soils may also be beneficial to plant growth and health, thereby increasing the capacity of plants to resist disease [2]. In this study, soil properties could not effectively induce soil disease suppression, because pasteurization of the suppressive soils did not transfer soil suppressive capabilities. Therefore, the changes of microbial community in the suppressive soil primarily contributed to higher disease suppression.
Consecutive monoculture adversely impacts plant growth and pathogen defense in many crop plants, which are mainly attributable to root-released autotoxic substances, exacerbation of soil physiochemical properties, and microbial community changes [22,42,45]. The imbalance of soil microbial populations mediated by root exudates is responsible for disease incidence under monoculture systems [22]. In cucumber, root-released p-coumaric acid alters soil microbial communities and promotes the growth of F. oxysporum under continuously mono-cropping patterns [42]. In Rehmannia glutinosa, root-secreted PAs can stimulate the growth and toxin production of F. oxysporum, but inhibit the antagonistic bacteria Pseudomonas species [22]. Among the major components of root exudates, PAs are crucial for inducing the allelopathy and replanting disease, and are prone to modifying soil microbial community, rather than directly inducing autotoxic effects on the mono-cropping plants [1]. It was found here that the levels of PAs were remarkably higher in the conducive soil than in the suppressive soil under long-term monoculture. Concurrently, the abundance of F. oxysporum distinctly differed between the conducive and suppressive soils. This variation between both the soils may at least partially explain the larger discrepancy of disease incidence. This was in line with recent reports that the abundance of F. oxysporum was positively correlated to Fusarium wilt disease of vanilla and banana [2,4]. The differences we observed allowed us to speculate that certain key microbiota existed in the suppressive soil, impeding the growth and invasion of F. oxysporum. As expected, mixing a small quantity of the 15-year suppressive soil led to a relatively low disease incidence in the conducive soil, while transferring the one-year suppressive soil did not lower disease incidence. Hence, long-term monoculture shaped specific microbial communities to hinder the growth of pathogens.
The taxonomic compositions of rhizosphere microbial communities varied among the conducive and suppressive soils of Chrysanthemum. Compared with the conducive soils, Proteobacteria and Firmutes were more abundant in the suppressive soils, which were in concert with previous studies [12,46]. Proteobacteria is the most common phyla in different soil types, but varied with relative abundance [47]. Banana Fusarium wilt disease suppression is negatively related to Proteobacteria in the organic fertilizer-amended soils [48]. It is well documented that Actinobacteria is more abundant in the suppressive soils than the conducive soils [9,39,48]. However, no significant difference in the phyla was observed between our conducive and suppressive soils. Thus, soil disease suppressiveness may not have resulted primarily from the antagonistic effects of Actinobacteria. Moreover, Ascomycota and Basidiomycota were the two most prevalent fungal phyla, in agreement with some studies exploring the microflora of vanilla and peanut continuous-cropping soils [4,49]. Compared with the conducive soils, Ascomycota was the most abundant in the suppressive soils of Chrysanthemum. Similarly, some member of Ascomycota dominated the fungal community in the suppressive soils of vanilla under monoculture systems [4]. Through the analyses of LEfSe and indicator OTUs, some specific species may be tightly related to soil disease suppressiveness. Pseudomonas was one of the most dominant genus belonging to the phylum Proteobacteria in the 15-year suppressive soil. It has previously been indicated that several Pseudomonas species can produce antibiotics against various plant pathogens, which are considered as key indicators of soil disease suppressiveness [12,50]. Accordantly, the 15-year conducive soil inoculated with a Pseudomonas species (P. putida A2) displayed a significant decrease in disease incidence, although this acquired suppressiveness quickly declined during the following monoculture. Moreover, dynamic changes of Pseudomonas in the soils displayed a similar trend. Thus, the survival of soil inoculants in the plant rhizosphere was required for improving soil suppressiveness to soil-borne pathogen.
PAs can trigger disease burst in monoculture soils by inhibiting beneficial soil microbes and promoting pathogen growth [22,51]. In this study, the suppressive soil accumulated less PAs than the conducive soil under long-term monoculture. The results of PA degradation assays indicated that the mixture of PAs obviously inhibited the growth of P. putida A2, although this strain could efficiently degrade single type of PAs, indicating that PA-enriched soils were unsuitable to the growth of the Pseudomonas species. Thus, construction of PA-degrading consortia was conducive to enhancing the colonization of P. putida A2 in the PA-enriched conducive soils. It is becoming clear that the manipulation of microbial communities is a prospective application for environmental and agricultural practices [52][53][54]. The functions of consortia, rather than single species, determine the efficiency of toxic substance degradation [26,27]. However, major obstacles of consortia construction are isolating all the community species and understanding the complex interactions of the microbiome [27]. Here, we tried to develop PA-degrading consortia that can suppress Fusarium pathogen by functional selection methods. Our results showed that long-term monoculture shifted the bacterial community, and the shift was associated with the suppressiveness of soils to Fusarium pathogens, as evidenced by the strong suppressiveness in the 15-year suppressive soil, not in the oneyear suppressive soil. The antagonistic and PA-degrading species may be greatly abundant in the 15year suppressive soil, but the abundance changes could not offer a direct justification for the functional alterations, which were related to interspecies interactions. We thus examined the performance (PA-degrading activity) and valid combinations of the selected isolates that were greatly abundant in the 15-year suppressive soil in different consortia. The increased efficiency of PAdegrading activity and growth performance of P. putida A2 may indicate mutual benefits related to species interactions. According to this, we designed bacterial combinations and validated the improved performances of selected combinations. The enhanced degradation of PAs was not only attributable to the performance of P. putida A2, but also the functions of other members in the community. Pot experiments further revealed that a model consortium harboring P. putida A2 markedly reduced PA levels in the 15-year conducive soils, with lower disease incidence and better P. putida growth during three-cycle monoculture.

Conclusions
Comparative analyses of rhizosphere microbial community characters between the conducive and suppressive soils of Chrysanthemum were conducted, to unravel potential microbiota associated with disease suppressiveness. Our findings demonstrated that the combination of P. putida with Xanthomonas and Rhizobium species greatly enhanced the efficiency of PA degradation. Furthermore, the PA-degrading consortium harboring the antagonistic Pseudomonas species conferred stronger soil suppressiveness to Fusarium wilt disease than the Pseudomonas species alone. Collectively, the functional combination of consortium can assist one to overcome the disease-suppressive instability of single inoculum with antagonistic bacteria in the Chrysanthemum monoculture soils.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1. Table S1 Soil physiochemical characteristics at two cultivated Chrysanthemum orchards that were disease-suppressive (Jutai) and -conducive (Shiji) to Fusarium wilt disease.  Figure S1 Rhizosphere microbial abundance and α-diversity in both the conducive and suppressive soils under 1-or 15-year monoculture systems. Bacterial (a) and fungal (b) abundance was quantified by qPCR. Microbial Shannon diversity (c, bacteria; e, fungi) and Chao1 richness (d, bacteria; f, fungi) were estimated in different soil samples. Data are means ± SE. Different letters above the bars indicate statistically significant differences at p < 0.05 by Tukey's HSD test. Figure S2 NMDS analyses of rhizosphere bacterial (a) and fungal (b) community. The matrix was generated based on the Bray-Curtis similarity algorithm. Con1 indicates 1-year conducive soil; Con15 indicates 15-year conducive soil; Sup1, indicates 1-year suppressive soil; Sup15, indicates 15-year suppressive soil. Figure S3 LEfSe analyses of the most differentially abundant (Top30) taxa in the 15-year suppressive and conducive soils. (a) Taxonomic cladogram showing the phylogenetic distribution of fungal lineages. (b) Histogram of the LDA scores estimated for differential fungal genera between the suppressive and conducive soils. Figure S4 Inhibitory effects of five isolated strains on F. oxysporum. 1 × 10 7 CFU/mL spore suspension of F. oxysporum was mixed in PDA media, and 2 μL of bacterial culture was dipped on the surface of the PDA media. Then, area of inhibition zone was calculated after 48 h of culture at 28 °C. Data are means ± SE. Different letters above the bars indicate statistically significant differences by Tukey's HSD test. Figure S5 Effects of PAs or mixed PAs on the mycelial growth (a) and sporulation (b) of F. oxysporum. Data are means ± SE. Different letters above the bars indicate statistically significant differences by Tukey's HSD test. Figure S6 Degradation of PAs in inorganic media by five bacterial strains. The media were amended with 0.9 g/L PAs. During 36 h of culture, fraction of the remaining ferulic (a), p-coumaric (b) and p-hydroxybenzoic acids (c) in culture solutions was calculated. The growth performances of different strains were examined in the media containing ferulic (d), p-coumaric (e) and p-hydroxybenzoic acids (f). Figure S7 Analyses of the content of PAs in the 15-year conducive soils treated with P. putida A2 or a PA-degrading consortium (PAC). After three months of inoculation, soil samples were harvested for determining the remaining PAs, including ferulic, p-coumaric and p-hydroxybenzonic acids. Data are means ± SE. Different letters above the bars indicate statistically significant differences by Tukey's HSD test. Figure S8 Colonization of Xanthomonas sp. NC8 and Rhizobium sp. HU16 in the PA-degrading consortium (PAC)-treated soils. The 15-year conducive soils were inoculated with the PAC harboring P. putida A2 (2 × 10 7 CFU/g), Xanthomonas sp. NC8 (5 × 10 6 CFU/g) and Rhizobium sp. HU16 (5 × 10 6 CFU/g). After three months, three-week-old seedlings were subjected to the second and third successive plantings (cycles) into the same pots. The density of NC8 9 (a) and HU16 (b) in the rhizosphere soils was quantified by qPCR, respectively. CP, consecutive planting (non-treated Con15); the PAC-treated soils were subjected to first (CP1), second (CP2) and third (CP3) three-month planting. Data are means ± SE. Different letters above the bars indicate statistically significant differences by Tukey's HSD test.

Conflicts of Interest:
The authors declare no conflict of interest.