Suppression of Banana Fusarium Wilt Disease With Soil Microbial Mechanisms Via Pineapple Rotation and Residue Amendment

Aims The large outbreak of banana Fusarium wilt has become a bottleneck limiting the industry’s development, and crop rotation is a cost-effective and essential measure to overcome the obstacles of banana crop monoculture. The present work was carried out to explore the mechanisms of how changes in physicochemical properties and the reestablishment of soil microorganisms in high-incidence soils are affected by crop rotation and plant residue. Methods In this study, pineapple-banana crop rotation and pineapple residue amendment were used to alleviate banana Fusarium wilt, and their effects on bacterial and fungal communities were studied using the MiSeq Illumina sequencing platform. Results Both pineapple-banana rotation and residue addition signicantly reduced disease incidence. Moreover, pineapple rotation and residue amendment altered the bacterial and fungal community composition. The taxonomic and phylogenetic alpha diversity of bacteria and fungi signicantly increased against disease suppression and nutrition competition. The relative abundances of the Burkholderia, Pseudomonas, Elaphocordyceps, Penicillium, and Talaromyces genera were higher, and the number of Fusarium was signicantly lower in rotational soil than in banana monoculture soil. Finally, linear models (LM) was used to show that the Burkholderia and Talaromyces in crop rotation, and Aspergillus in residue amendment have signicant negative relationship to disease incidence, which plays a key role in Fusarium reduction. Conclusions To consider the economic benets and protect the vitality of the soil, this study suggested that pineapple-banana rotation and pineapple residue amendment both could be considered for the sustainable management of banana wilt. hypothesise that pineapple-banana crop rotation and pineapple residue addition provides greater suppression of Fusarium wilt due to improved soil physicochemical properties that stimulate the growth of indigenous benecial microorganisms, resulting in a signicant reduction in disease incidence in bananas compared to that under monoculture. The aims of this study were (1) to evaluate the effect of the pineapple-banana crop rotation system on wilt disease in banana seedlings during the seedling period, (2) to estimate the changes in microbial community structure from the bulk soil to the rhizosphere soil zone, (3) to analyse the diversity of soil microorganisms, and (4) to study the relationship between the soil microbial community composition and environmental factors.


Introduction
Currently, soil-borne diseases are a serious threat to soil health and crop productivity (Dita et al. 2010; Mendes et al. 2013) and have become one of the major problems for the sustainable development of intensive agriculture (Butler 2013). Banana (Musa spp.) is among the crops more severely affected by succession disorders caused by multiple abiotic (poor soil fertility, drought, high temperature and salinity) and biotic (fungi, bacteria, viruses and pests) factors ( Su et al. 2017). However, currently, in high-incidence banana soil, the soil microbiota is severely imbalanced, and the short-term reduction in the pathogenic population does not fundamentally improve the imbalance of the soil microbiota (Ploetz 2015). Therefore, in high-incidence elds, to gradually restore a healthy soil microbiota, crop rotation may be an important measure for the gradual modi cation of soil ecology. Studies show that soil pathogens can be reduced by using speci c agricultural practices, for example, the incorporation of plant residues in the soil and crop rotation (Larkin 2015;Bonanomi et al. 2018).
Crop rotation is an important measure for gradually changing the soil ecology ). Different crop rotation systems have different effects on the suppression of different diseases, such as obstructing the inbreak of pathogens, chemosensitive substances, and antagonistic microorganisms, improving soil fertility through root exudates and residues, and increasing soil microbial biomass and activity (Christen and  The soil microbiome plays an important role in promoting plant growth and development, nutrient uptake, disease resistance and adaptation to environmental stresses. Their abundance, composition and activity largely determine the sustainability of agriculture. Recent studies have shown that modulation of the soil microbiome can reduce the number of pathogens in the soil and enhance rhizosphere immunity, thereby reducing or suppressing disease occurrence. For example, Wei et al. (2019) showed that phages can "target" tomato cyanobacteria to disrupt their ability to survive while structuring the rhizosphere soil ora to restore community diversity and increasing the abundance of bene cial microbes. In addition, the abundance and activity of Pseudomonas, Sphingomonas, Penicillium and Trichoderma spp. were found to be signi cantly higher than the Fusarium number in the soil ). The soil microbiome can enhance rhizosphere immunity, improve soil fertility and crop yield, maintain the persistence of plants, animals and humans, and protect soil health (Bender et al. 2016). Additionally, there are many protozoans that protect crop health by feeding on or producing bene cial antibacterial substances to defend against the invasion of other Fusarium pathogens ). More interestingly, the interactions between soil microbial communities are also closely related to the ability of Fusarium to invade plants and affect their health. Studies have shown that immune rhizosphere microbial communities are more diverse and have a more complex microbial network (Ge et al. 2021). It was shown that rhizosphere competitive intercropping communities could produce more bacterial inhibitory substances or fully occupy limited ecological niches, thus effectively suppressing pathogen invasion and reducing widespread outbreaks of soil-borne diseases.
Furthermore, studies have shown that plant root exudates and residues in soil can have a long-lasting effect on pathogen control as a result of crop rotation and relieve soil pressure by improving soil properties (Hu et al. 2018;Yuan et al. 2018;Zhou et al. 2020). As such, it is clear that crop rotation and plant residue manipulation in the eld are important strategies for soil-borne disease management (Mawar and Lodha 2015). However, these strategies have received relatively little attention, as efforts have mostly been given to investigate how such approaches relate to overall aspects of soil quality, nutrient cycle, and crop performance Martinez-Feria et al. 2018;Chen et al. 2018). Therefore, understanding how these approaches impact the soil microbiome composition and its ecological function remains elusive (Afshan et al. 2015). This opens up potential opportunities to explore bene cial outcomes of agricultural management associated with soil-borne disease control (De et al. 2020).
In our previous work, banana-pineapple rotation was picked out for its high-e ciency in banana Fusarium wilt disease prevention and control . In normal farming operations, pineapple residue is returned to the eld after rotation. Therefore, the effect of crop rotation consists of two parts: simple crop rotation and residue return. We hypothesize that both crop rotation and residue amendment have inhibitory effects on banana Fusarium wilt, while, how they work was still unknown. In this study, two pot experiments, crop rotation and residue amendment, were designed to study the improvement of soil physicochemical properties and their effect on soil microbiota.
Further, we hypothesise that pineapple-banana crop rotation and pineapple residue addition provides greater suppression of Fusarium wilt due to improved soil physicochemical properties that stimulate the growth of indigenous bene cial microorganisms, resulting in a signi cant reduction in disease incidence in bananas compared to that under monoculture. The aims of this study were (1) to evaluate the effect of the pineapple-banana crop rotation system on wilt disease in banana seedlings during the seedling period, (2) to estimate the changes in microbial community structure from the bulk soil to the rhizosphere soil zone, (3) to analyse the diversity of soil microorganisms, and (4) to study the relationship between the soil microbial community composition and environmental factors.

Materials And Methods
Crop residues, root exudates and soil sampling Preparation of banana and pineapple residues: Banana and pineapple residues were collected from a eld located at Paigou village in Zhongyuan town, Qionghai City, Hainan Province, China (110°30'E, 19°5'N). Brie y, the whole plants were put into plastic packaging bags, kept on ice, and transported to the laboratory (< 6 h). These crop residues were carefully washed ve times with sterile deionized water in the laboratory and divided into root, stem leaf and fruit plant parts. CK: no residue, B_L: banana stem leaf residue, B_R: banana root residue, P_L: pineapple stem leaf residue, P_R: pineapple root residue, and P_F: pineapple fruit residue. These crop residues were chopped into tiny pieces and ground down to powder. Each residue type was sieved through a 4-mm mesh, and the total nutrient content was measured. Fresh plant samples were dried, ground and extracted at a ratio of 1:5 plant residue: deionised water for 48 hours and aseptically ltered to obtain the extract master batch.
Preparation of banana and pineapple root exudates: The plants were removed from the soil, rinsed ve times with tap water and ve times with deionised water, and incubated in a plastic cup containing 300 mL of deionised water for 24 h. The solution containing the root exudate was combined and then slowly ltered ( lter membrane pore size: 0.45 μm, size: Φ25 mm), and the ltrate was freeze-dried to dryness. All roots of pineapple plants involved in root exudate collection were then dried and weighed separately, the freeze-dried material was dissolved separately in an appropriate amount of deionized water, and the volume was xed to one gram of dry weight root exudate per mL of root exudate master batch (i.e., 1 g of dry root exudate per mL) and stored at -20 °C.
Plates were poured according to different volume ratios (1:10, 1:100, 1:1000) of residue extracts to water agar. Pathogen cakes were inoculated into the centre of the prepared plates with a 0.8-cm-diameter hole punch to obtain uniform growth of Fusarium acuminatum cakes of uniform media thickness. Effect of mycelia growth (%) = (diameter of treated colonies -diameter of control colonies) / (diameter of control colonies -0.8 cm) × 100.
Soil with high banana Fusarium wilt disease incidence (> 65%) was collected from the eld trial site (Lingao Xinxing Farm, Hainan Province, China (109°77'E, 19°77'N), where bananas were continuously cropped for 11 years. The eld trial site has a tropical monsoon climate with an average annual temperature of 23.8 °C, an average precipitation of 1,786 mm, and an annual average of 2,059 hours of sunshine. The soil is a dry red soil with a pH of 5.31, an organic matter content of 3.64 g·kg −1 , an available phosphorus content of 15.64 mg·kg −1 , an available potassium content of 411.67 mg·kg −1 , and an available nitrogen content of 91.23 mg·kg −1 . The excavated eld soil was collected within the drip line next to the banana plant approximately 20 cm deep, thoroughly mixed and immediately transferred to a greenhouse (average temperature of 36 °C and 37% humidity) at the College of Agriculture, Hainan University, Haikou, Hainan Province, China (110°34'E , 20°06'N).

Pot rotation experimental design
A pot experiment was conducted from May 2018 to February 2019. It was completely randomized and included three treatments (i.e., Treatments I-III): fallow treatment (Treatment I), banana monoculture (Treatment II), and pineapple rotation (Treatment III). Each polypropylene pot (35 × 25 × 25 cm, length × width × height) had 10 kg of soil per pot, with six replicates of 10 pots per replicate. Before transplanting, commercial bio-organic fertilizer (a pure plant-derived bio-organic fertilizer containing 1.3% nitrogen (N), 1.1% phosphorus (P 2 O 5 ), 1.0% potassium (K 2 O), organic matter ≥ 40%, and moisture ≤ 30%) was applied at a rate of 2% soil as a base fertilizer in the pots.
After 5 days of transplanting, evenly grown pineapple and banana seedlings were transplanted into the pots. After 6 months of cultivation, planted banana seedlings exhibited mild Fusarium wilt symptoms, such as yellow leaves and scabs at the base of the stem. Subsequently, the next crop of banana seedlings was continued by pulling out pineapple and banana seedlings and planting the banana seedlings. When the maximum incidence of treatment banana (B) was greater than 80% (4 months), the incidence of banana seedlings was counted for all treatments.
At this time, the whole banana plant was removed, and the bulk and rhizosphere soil were collected with sterile tweezers or brushes. The bulk soils of fallow, banana and pineapple were de ned as CBf, BBm and PBr, respectively, while the rhizosphere soils of fallow, banana and pineapple were de ned as CRf, BRm and PRr, respectively (Fig. 1A).

Pineapple residue amendment pot experimental design
For the experiment (September 2018 to March 2019) on the effects of soil amendment with crop residue on disease severity, eld experiment available crop residue, banana and pineapple (Qionghai, Hainan) were chopped and nely ground using an Oster blender and were then passed through a 2-mm mesh sieve. Crop residue was then thoroughly mixed with the pasteurized soil at a rate of 3.0% (v/v). There were three treatments (i.e., Treatments VI-IV) designed for soil collected from the eld with the same potting speci cations as above. Treatment VI was the control: no residue added, Treatment V was banana residue added, Treatment VI was pineapple residue added. The experiment ended when the incidence of banana seedlings treated with added banana residues was over 60% (at six months). At this time, the whole banana plant was removed, and the bulk and rhizosphere soil were collected with sterile tweezers or brushes. The bulk soils of fallow, banana and pineapple were de ned as CBn, BBb and PBp, respectively, while the rhizosphere soils of fallow, banana and pineapple were de ned as CRn, BRb and PRp, respectively (Fig. 1B).

Disease incidence determination
Banana plants with typical wilt symptoms, such as yellow leaves, wilted leaves, and root rot, were counted as diseased plants, and the incidence rate was expressed as the percentage of the number of diseased banana plants to the total number of planted banana plants. In other words, the incidence rate (%) = banana plants with incidence/total number of banana plants × 100%.

Soil sampling collection and DNA extraction
Bulk soil samples were collected by removing banana plants from soil cores to a depth of 10 cm using a soil auger. The soil was mixed, homogenized and sieved through a 2-mm nylon sieve to remove plant debris and was then divided equally into three subsamples. One subsample was air-dried for physicochemical analysis, and the other two were stored at 4 °C and -80 °C for microbiological analysis and DNA extraction, respectively. At this time, the whole plant was removed, and the bulk soil was removed by careful shaking. Soil still adhering to the roots was collected with sterile tweezers and de ned as rhizosphere soil. Total soil DNA was extracted using a PowerSoil DNA Isolation Kit (MoBio Laboratories Inc., Carlsbad, CA, USA) following the manufacturer's protocol. Bulk soil and rhizosphere soil samples of banana and pineapple collected from pot planting and residue addition tests were accurately weighed to 0.7000 g and 0.4000 g, respectively. Soil DNA microorganisms were extracted using a kit (MoBio products, USA: PowerSoil® DNA Isolation Kit, 12888) ). The extraction results were veri ed by nucleic acid assay and stored at -20 °C in a refrigerator. A volume of 20 μL of DNA samples from each treatment was selected and sent to "Nanjing aomike Bio" for Illumina MiSeq sequencing.
Quantitative PCR of Fusarium oxysporum in bulk and rhizosphere soil Quantitative real-time PCR ampli cation (qPCR) was performed to determine the number of F.oxysporum using total bulk and rhizosphere soil DNA as a template. The primers used were FocSc-1 and FocSc-2 (Table S1). The system for real-time uorescence absolute quanti cation of pathogens at 50 μL was 2 μL of DNA template sample, 25 μL of premix reagent, 1 μL of primer 1 (FocSc-1) (10 pmol/L), 1 μL of primer 2 (FocSc-2) (10 pmol/L), and 21 μL of ddH 2 O. All soil DNA samples and the standard curve were analysed using a 7500 Real-Time PCR System (Applied Biosystems™, Foster City, USA). The melting curve and ampli cation e ciency were con rmed as standards for ampli cation. Six replicates of each sample were estimated, and the results were converted to log10 copy numbers (copy numbers g −1 soil) (Xiong et al. 2017a).
Soil chemical analysis and culturable microorganism determination Soil pH was assayed using a soil-water ratio of 1:5 (w/v) suspensions. The organic matter (OM) content was measured using the K 2 Cr 2 O 7 -H 2 SO 4 oxidation method. Available potassium (AK) was extracted with ammonium acetate and analysed by ame photometry. Available phosphorus (AP) was determined by molybdenum antimony colorimetry, and the available nitrogen (AN) content was determined following

Illumina MiSeq sequencing and data processing
The V4-V5 region of bacterial 16S rDNA and the fungal ITS region were ampli ed using the individual barcoded primers 520F/802R and ITS5F/ITS2R, respectively (Table S1)  Raw sequences based on unique barcodes were assigned to soil samples following the QIIME software package (version 1.9.1) tutorials after elimination of the adaptors and primer sequences (Caporaso et al. 2010), and pairs of reads were merged using the FLASH software tool (version 1.2.7). The paired sequences were processed by the UPARSE pipeline to produce an operational taxonomic unit (OTU) table with USEARCH 11 and Perl scripts (Edgar et al. 2011). Brie y, sequences with a quality score < 0.5 or length < 200 bp were eliminated. The retained sequences were assembled to identify OTUs at 97% similarity, and chimaeras were removed based on the UCHIME method. Then, high-quality OTUs were classi ed with the RDP Bacterial 16S database and the UNITE Fungal ITS database of the RDP classi er procedure for bacteria and fungi, respectively (Wang et al. 2007; Kõljalg et al. 2013).
The raw sequence data have been deposited in the NCBI Sequence Read Archive (SRA) database under accession numbers PRJNA745388 and PRJNA746047.

Statistical analysis
Microsoft Excel 2010 was used to process and graph the obtained data. Data were analysed with SPSS 20.0 software, and one-way analysis of variance (ANOVA) was used to compare the data. The signi cance of differences between treatments was tested using the new complex polar difference method (Duncan's test). R (3.6.0) language (vegan, mvpart, ggplot2 and graphics packages) was used to perform principal coordinate analysis (PCoA) and Pearson and Spearman correlation analyses. Permutational multivariate analysis of variance (PERMANOVA) and variance partitioning analysis (VPA) were based on the Bray-Curtis distance using the adonis and varpart functions within the R package vegan, respectively (Oksanen et al. 2012).

Results
Effects of pineapple rotation and residue amendment on banana Fusarium wilt disease incidence and relative abundance of Fusarium The effects of different concentrations of pineapple and banana residues (0.1%, 1.0% and 10.0%) on spore germination were evaluated ( Fig. 2A). The infusion of all parts of banana had a promoting effect on spore germination of the pathogen, while the infusion of all parts of pineapple had an inhibiting effect on spore germination, with the fruit infusion of pineapple having a highly signi cant effect on spore germination. The effect of pineapple and banana root exudates on spore germination, compared to banana root exudates, pineapple root exudates were signi cantly less able to promote spore germination of the pathogen. The fungus signi cantly increased the number of pathogenic spores compared to the control (Fig. 2B). We observed an overall signi cantly reduced Fusarium wilt disease incidence in both the pineapple rotation and residue amendment systems (Duncan's t-test, p < 0.001) (Pr and Pp), which was signi cantly lower than those in the fallow and monoculture treatments (p < 0.001) (Fig. 2C). Moreover, the abundance of rhizosphere Fusarium (rotation treatment) (p = 0.001), as indicated by the residue treatment bulk Fusarium (p = 0.118), and the relative abundance of rhizosphere Fusarium (p < 0.001) were both positively correlated with banana Fusarium wilt DI (Fig. 2D), demonstrating disease suppression ability after rotation and residue addition with pineapple in the banana orchard. In the rhizosphere soil, the relative abundance of Fusarium was signi cantly lower in the rotation than in the monoculture treatments (p < 0.05, Fig. 2C). The qPCR results showed that signi cantly fewer Fusarium oxysporum were detected in the rotation rhizosphere (Duncan's t-test, p < 0.05, Fig. S1).
Effects of pineapple rotation and residue amendment on soil microbial community structure The PCoA plots (Fig. 3A, B, C and D) showed signi cant differences in the bacterial and fungal community composition in the rotation and residue amendment systems (PERMANOVA, p < 0.001), in the rotation system, there was a signi cant difference in bacteria (p (bulk) < 0.01, p (rhizosphere) < 0.01), moreover, the different rhizosphere soil microbial communities changed signi cantly from the corresponding initial soil microbial communities, and the microbial communities of different plants grown under the same soil type also differed signi cantly (p < 0.001) (Fig. S2). The sequencing results were analysed and detailed, and the microbial community richness and diversity sequencing results are shown in the Supplementary Material (Table S3). PERMANOVA and VPA analyses revealed that soil rhizosphere microbial communities were signi cantly in uenced by soil type (p < 0.001) and soil pH (p < 0.001). The relative importance of soil type and its soil pH on soil and rhizosphere microbial communities was greater than that of plant species (Fig. 3 E, F, G and H).

Effect of pineapple rotation and residue amendment on taxonomic composition
Volcano plot analysis of the sequence results revealed rotation and residue amendment bacterial and fungal community compositions with speci c respective sets of OTUs ( Fig. 4A and B). With red indicating bulk soil and black indicating rhizosphere soil, we found both that the pineapple rotation residue amendment treatment had a higher OTU number in the bulk soil, and selecting an OTU with a relative abundance greater than 0.1% for Venn analysis revealed that rotation and residue treatments had more bacteria, while fungi in the bulk soil had signi cantly higher OTU numbers than those in the rhizosphere soil ( Fig. 4C and D).
Analysis of the variability of the top 30 genera of bacteria and fungi showed that the variability of genera signi cantly increased in the soil.

Bulk and rhizosphere network construction through effects on speci c microbial taxa
In this study, we constructed co-occurrence networks using random matrix theory (RMT) to determine the differences in bacterial and fungal assemblages (OTU relative abundance > 0.1%) in bulk and rhizosphere soils of the different treatments. All values of the calculated modularity index were larger than 0.4 (Table S9), suggesting typical module structures (Chen et al., 2020). Overall, pineapple rotation and residue amendment showed marked effects on the soil microbial network: the average path distance (GD), the average clustering coe cient (avgCC) and the modularity of the empirical networks were higher than those of the corresponding, identically sized random networks (Table S9). Here, we found that residue assemblages (in Fig. 6 C and D) formed more connected and more complex networks with fewer nodes but more connections (edges) between nodes compared with the bulk soil. There were many keystone taxa in the microbial communities whose removal could cause a dramatic shift in microbiome structure and function. Keystone taxa in network analysis can be computationally identi ed as hubs with a high within-module degree Zi (Zi ≥ 0.5 indicates that the nodes are "well connected" to other nodes in the module). The PBr and PRr treatments had some keystone taxa, such as Burkholderia and Pseudomonas, and no hub was found in the bulk or rhizosphere soil (Cf, Cn, Bm and Bb) (Fig. S3).
Based on the above results, a conceptual model illustrating potentially path with important suppression predictors in intercropping system was constructed (Fig. 7). The conceptual mode indicated that two ways in banana plantations reduced the relative abundance of Fusarium by the soil characteristics and microbial community structure regulation. Among all the suppression predictors, the key physicochemical factors AP (in rotation system) and OM (in residue system) contents are signi cantly affected by pineapple rotation and rseidue amendment, respectively. And they were signi cant leading to changes of fungal community and bene cial microorganism. And the signi cant increases in bacterial genera (Burkholderia) and fungal genera (Talaromyces and Aspergillus) can directly affect the relative abundance of Fusarium, thereby reduce the incidence of banana (Fig. 7).

Discussion
Banana with different crop rotations can effectively reduce banana Fusarium wilt Fan et al. 2020). In this study, the results of pot experiments revealed that pineapple-banana rotation and residue amendment systems could signi cantly enhance the suppression of Fusarium wilt disease compared with banana monocropping and residue addition. This pattern is consistent with the results reported in a previous eld study . Notably, through the study of microbial community changes, we found that the reduction in disease incidence in crop rotation and residue amendment treatments could be attributed to a reduction in the number of Fusarium as well as an increase in bene cial genera and changes in soil physicochemistry. We propose two systems in which rotation and residue amendment could inhibit fungal infections. Similar results have been shown for strawberry soil disease (Fang et al. 2012), likely due to the interruption of the host pathogen cycle in the root. Both high-throughput sequencing and qPCR showed that the Fusarium number was smaller in the pineapple rotation and residue amendment treatments than in the banana monoculture and residue addition and exhibited signi cant positive correlations with disease incidence in our study ( Fig. 2 and Fig. S1). The incorporation of both pineapple residues (P_L, P_R and P_F) into the soil was found to signi cantly decrease the pathogen density in the soil and the incidence of Fusarium wilt disease. This nding aligns with our previous studies showing that using a pineapple-banana crop rotation system can effectively minimize the incidence of this pathogen . This study thus provides strong evidence that pineapple-banana rotation and residue amendment are effective approaches for suppressing disease.
In our study, pineapple rotation may have resulted in a signi cant decline in pathogen number via increased abundance of some bene cial microorganisms and disease suppression by antagonistic and nutritional competition with harmful microorganisms, parasitism, predation, induction of plant resistance, and interference with pathogenic signals, leading to a decrease in nutrient availability to pathogens. In addition, it seems possible that pineapple crops can stimulate bene cial bacterial microbiomes through root exudates to suppress pathogens during the crop rotation season (Hao et al. 2010).
The composition of the rhizosphere fungal community was clearly divided into different groups based on the PCoA results of the crop rotation and residue amendments, the fungal communities were clearly clustered between the pineapple-banana rotation and residue pineapple addition treatments and the banana continuous crop and banana residue addition treatments (Fig. 3). We hypothesize that soil fungi are less redundant than bacteria when exposed to environmental stresses of crop rotation and under the effective nutrients of residue addition. Additionally, VPA analysis showed that the variations in bacterial and fungal communities were best explained by the type of soil, followed by the pattern of cultivation. It has been previously reported that cultivated crops have a greater impact on microbial structure than cultivation patterns. Future research elucidating the mechanisms of plant species-root exudate-soil microbial community interactions may be an effective way to study the suppression of pathogens. Speci cally, in our experiment, this suppressive status emerged as a function of changes in the soil fungal communities rather than one that was mediated by changes in soil physicochemical properties. This was also further corroborated by Linear models (LM), which traced the relationships among microbial communities, pathogen density and banana disease incidence (Table S10).
Consistent with previous ndings, at the phylum level of bacteria and fungi, the major phyla were Proteobacteria, Acidobacteria, Bacteroidetes, Firmicutes, Gemmatimonadetes, Actinobacteria, Verrucomicrobia, Chloro exi, Planctomycetes, Cyanobacteria/Chloroplast, and Thaumarchaeota, and the fungal phyla were Ascomycota and Basidiomycetes ( Fig. 5A and B). In this study, only the relative abundance of Firmicutes and Ascomycota increased after pineapple rotation and residue addition and showed a strong negative correlation with F. oxysporum abundance (Table S5). This nding suggests that Firmicutes and Ascomycota may be involved in disease suppression through the production of biocontrol agents.
The pineapple-banana rotation signi cantly stimulated the relative abundance of the Burkholderia, Pseudomonas and Talaromyces genera in the bulk and rhizosphere soils ( Fig. 5C and Fig. S2). These microorganisms showed signi cantly negative correlations with the abundance of F. oxysporum, and there is a large body of work demonstrating the stable suppression of disease by these bene cial microorganisms. For example, Burkholderia spp. has been shown to be abundant in golden pineapple-banana rotations , and Pseudomonas spp. plays a key role in pathogen antagonism by stimulating the synthesis of a bene cial microbiome common to the native bene cial microorganisms in the soil, after the application of bioorganic fertilizers, to reduce the number of Fusarium in bananas ). Thus, they are widely reported as biological control agents. These microorganisms are capable of producing antifungal metabolites and can colonize the roots of plants, making them targets for biocontrol. Additionally, studies have shown that some fungal microorganisms pre-empt the microecological environment with pathogens, such as the effect of the mass production of Talaromyces spp. on pathogens populations (Bahramian et al. 2016). This result is the same as that reported by Xiong et al. (2017b), who studied two treatments of vanilla wilt, where the crop rotation treatment signi cantly reduced the number of Fusarium oxysporum in the soil. In addition, both pineapple crop rotation and residue addition signi cantly reduced the incidence of banana seedling wilt, closely related to the decrease in the number of Fusarium oxysporum in the soil after pineapple cultivation.
According to Wang et al. (2015), golden pineapple rotation can increase the organic matter and available phosphorus content in the soil, with a decrease in the soil pH, the results of this experiment after planting pineapple were consistent with previous studies. PCoA, volcano plot and VPA analyses revealed signi cant differences in microbial community structure after pineapple rotation, which is consistent with soil bacterial and fungal genus levels, the pineapple rotation treatment signi cantly increased the numbers of Burkholderia, Pseudomonas, Penicillium, and Talaromyces, in the residue amendment, Pseudomonas, Elaphocordyceps, and Aspergillus signi cantly reduced the relative abundance of Fusarium oxysporum in the soil. Wang et al. (2015) and Steensels et al. (2019) showed that the numbers of culturable bacteria and Burkholderia increased signi cantly, and the number of Fusarium oxysporum decreased signi cantly in pineapple rotation with bioorganic fertilizer. In herb and "maize-potato" rotations, we found that the Pseudomonas genus was abundant and signi cantly suppressed Fusarium oxysporum populations (Xiong et Moreno et al. 2005). Although Talaromyces and Aspergillus are fungal diseases in some crops, a signi cantly negative correlation was found between these two genera and Fusarium oxysporum in this experiment, which may be because the fungi Talaromyces and Aspergillus compete with Fusarium oxysporum for nutrients in the soil, resulting in a decrease in the number of Fusarium oxysporum, the exact cause of which needs to be further investigated.
The incidence of banana seedlings replanted after pineapple residue addition was highly signi cant and positively correlated with Fusarium oxysporum and culturable Fusarium oxysporum, except that it was signi cantly and negatively correlated with soil pH, Pseudomonas, and Aspergillus and signi cantly and negatively correlated with culturable bacteria. Penicillium and Burkholderia were abundantly present in the bulk and rhizosphere soils after pineapple residue addition, while Talaromyces and Aspergillus could pass through the pineapple rhizosphere to the soil and then survive in the next banana soil, and Talaromyces could colonize the next banana soil. Therefore, they are also potential key microorganisms (Baileya and Lazarovitsb 2003). In summary, these results provide some directions for our next work and further research on the mechanism of banana-pineapple crop rotation and residue amendment.

Conclusion
Our rotation pot experiment further demonstrated the effectiveness of banana-pineapple rotation in alleviating banana wilt ). In addition, our study contributes to this body of research by showing that the addition of pineapple plant residues in the soil can trigger soil suppression of a serious soil-borne pathogen (Fusarium spp.). Here, we described that this state of suppression is modulated by changes in the soil bacterial and fungal communities and highlighted one bacterial taxa (Burkholderia spp.) and two fungal taxa (Aspergillus spp. and Talaromyces spp.) that may be directly involved in pathogen suppression. We provided further evidence for their suggested different and complementary modes of action and subsequently validated their suppressive potential in well-controlled indoor experiments ( Fig. 2A, B), which offers great promise for the search for bene cial substances in root exudates and residues in the future.
Taken together, our research may provide new avenues for exploring agricultural practices with a focus on bene cial outcomes that directly affect soil health and crop productivity in a viable and sustainable manner.

Declarations Con ict of interest
The authors declare no con icts of interest.
. Zhou G, Gao S, Lu Y, Liao Y, Nie J, Cao W (2020) Co-incorporation of green manure and rice straw improves rice production, soil chemical, biochemical and microbiological properties in a typical paddy eld in southern China. Soil Till Res 197:104499 Tables   Table 1 Linear models (LM) for the relationships of microbial indicators with disease incidence and the relative importance of each indicators in the crop rotation and residue amendment system. P was results of ANOVAs.    PCoA based on Bray-Curtis distance of rotation (bacteria (A), fungi (B)) and residue (bacteria (C), fungi (D)) of soil community structure.
Contributions of soil type, planting method, and soil chemicals (pH) to the assembly of soil bacterial (E,G) and fungal (F,H) communities were calculated based on variance partitioning analyses (VPAss) in rotation and residue, respectively, and the p value was determined by PERMANOVA. The treatmentabbreviations are de ned in Fig. 1.

Figure 4
The effects of soil type (bulk and rhizosphere soils) in the pineapple rotation and residue identity (following banana and pineapple cultivation) on microbial compositions. The rotation bacterial (A) and fungal OTUs with relative abundances that were signi cantly (log2fold change > |1.5| and FDR adjusted p value < 0.05) different between the bulk (red) and rhizosphere (black) soils are coloured in the volcano plot. Venn diagrams C and D represent the numbers of bacterial and fungal OTUs in the bulk and rhizosphere soils originating from the rice soil, respectively. Venn diagrams e and f represent the numbers of bacterial and fungal OTUs in rhizosphere soils originating from the forest soil, respectively. The treatment abbreviations are de ned in Fig. 1.

Figure 6
Plant rhizosphere and the corresponding bulk soil networks in the rotation and residue addition systems. Networks represent random matrix theory co-occurrence models derived from 6 biological replicates at each site, where nodes represent OTUs, and the edges between the nodes indicate signi cant correlations. A green edge indicates a negative covariation between two individual nodes, while a red edge indicates a positive covariation. The treatment abbreviations are de ned in Fig. 1.