One hundred and eight soil samples were collected with a hand auger from a 5–20 cm soil layer around the crown of plants in Jiande (loquat-rice), Chun’an (mulberry-rice), Tonglu (blueberry-rice), Fuyang (grape-rice), Lin’an (bamboo-rice), and Yuhang (seedling-rice), Zhejiang province, China. For each conversed mode, six samples were taken from each field (such as loquat-rice converted mode in Jiande, every six samples were collected from loquat garden, paddy field converted from loquat garden, and perennial paddy field, respectively) using the five-point sampling method (Panagos et al., 2014), and a total of eighteen samples were collected. Then, each sample was packed individually on-site into a sterilized and sealed polythene bag, and transported to the laboratory using a portable cooler for further analysis. Indeed, after dissolving 10 g fresh soil in 90 mL sterile water, mixing well by vortex, and serially diluting with 9 mL of sterile water, 100 μL of 10^–3 dilution was inoculated on potato dextrose agar (PDA, potato extract 200 g, dextrose 20 g, agar 20 g, ddH₂O 1000 mL) medium and then incubated at 25°C for 3–5 days. After three times of repetitive purification on PDA medium, the colonies with different morphological characteristics and higher growth rate were selected to further purify via single-spore isolation, and then stored in 20% glycerol at -70°C.

The fungal isolates were identified as described before (Samson et al., 2014; Tian et al., 2021), which was carried out by incubating them on PDA, Czapek yeast autolysate agar (CYA, czapek concentrate 10 mL, sucrose 30 g, yeast extract 5 g, K₂HPO₄1 g, CuSO₄·5H₂O 0.005 g, ZnSO₄·7H₂O 0.01 g, agar 20 g, ddH₂O 1000 mL), and male extract autolysate (MEA, malt extract 50 g, CuSO₄·5H₂O 0.005 g, ZnSO₄·7H₂O 0.01 g, agar 20 g, ddH₂O 1000 mL) medium for 7 days at 25°C, and observing colony morphology of each isolate. Molecular identification of the fungal isolates was carried out using the ITS, TUB, CaM, and RPB2 combined dataset with the following steps: mycelia of each test isolate were harvested from the surface of PDA, frozen in liquid nitrogen, and extracted for genomic DNA using the Rapid Fungi Genomic DNA Isolation Kit (Sangon Biotech Co., Ltd., Shanghai, China).

Primer sets ITS5/ITS4 (5’-TCC TCC GCT TAT TGA TAT GC-3’ and 5’-GGA AGT AAA AGT CGT AAC AAG G-3’), Bt2a/Bt2b (5’-GGT AAC CAA ATC GGT GCT GCT TTC-3’ and 5’-ACC CTC AGT GTA GTG ACC CTT GGC-3’), CMD5/CMD6 (5’-CCG AGT ACA AGG ARG CCT TC-3’ and 5’-CCG ATR GAG GTC ATR ACG TGG-3’), and 5F/7CR (5’-GAY GAY MGW GAT CAY TTY GG-3’ and 5’-CCC ATR GCT TGY TTR CCC AT-3’) were used for amplification of rDNA ITS, TUB, CaM, and RPB2 genes, respectively (Hong et al., 2005; Nasri et al., 2015; Tomaha et al., 2020; Haituk et al., 2021). PCR amplification reaction mixture (50 μL) included ddH₂O (18 μL), 2 × Hieff^® PCR Master Mix (25 μL), 10 μM each primer (2 μL), and DNA (3 μL). Then, PCR was carried out as follow: for ITS gene, 94°C 5 minutes; 94°C 45 s, 52°C 45 s, 72°C 60 s, 35 cycles; 72°C 10 minutes; for TUB and CaM genes, 94°C 5 minutes; 94°C 60 s, 55°C 60 s, 72°C 90 s, 35 cycles; 72°C 10 minutes; for RPB2 gene, 94°C 5 minutes; 94°C 30 s, 51°C 30 s, 72°C 60 s, 5 cycles; 94°C 30 s, 49°C 30 s, 72°C 60 s, 5 cycles; 94°C 30 s, 47°C 30 s, 72°C 60 s, 30 cycles; 72°C 10 minutes.

PCR products were visualized on 1.0% agarose gels, then purified and submitted to Tsingke Biotechnology Co., Ltd. (Hangzhou, China) for sequencing in both directions. Sequences for each region were assembled and edited using DNASTAR Lasergene (v7.0.1) (DNAStar Inc., Madison, USA) and BioEdit (v7.0.9) (North Carolina St. University, Raleigh, USA) (Tippmann, 2004), and then analyzed by BLAST search in the GenBank database. After the reference sequences of closely related species for test strains were downloaded from the GenBank database, maximum likelihood (ML) phylogenetic trees were constructed using Mega 7.0 (Kumar et al., 2016). Bootstrap replicates were performed 1000 times, and bootstrap values above 50% were indicated on the cladogram to indicate the significance of the separation.

The ability of the selected isolates to solubilize phosphate was evaluated in National Botanical Research Institute’s phosphate (NBRIP, glucose 10 g, Ca₃(PO₄)₂5 g, MgCl₂5 g, KCl 0.2 g, MgSO₄·7H₂O 0.25 g, (NH₄)₂·SO₄ 0.1 g, agar 25 g, ddH₂O 1000 mL, pH 7.0) medium (Abdallah et al., 2019). In brief, a 6-mm agar disc of test fungus was placed on NBRIP medium at 25°C for 3 days, the phosphate solubility was assessed by observing the formation of a clear halo zone surrounding the colony, and measuring the size of the phosphate-solubilizing zone. Each fungal isolate was performed in triplicate.

Fungal isolates with high phosphate solubility (producing a clear halo zone surrounding the colony) were further used to evaluate their siderophore-producing ability, which was carried out using the chrome azurol S (CAS) assay as previously described (Murugappan et al., 2011; Li et al., 2021). In brief, a 6-mm agar disc of the test fungus was placed on CAS-agar medium and incubated at 25°C for 3 days. Colonies of siderophore-producing fungus were surrounded by an orange halo in the CAS plate. Each fungal isolate was performed in triplicate.

Fungal isolates with high phosphate solubility (producing a clear halo zone surrounding the colony) and siderophore-producing (producing an orange halo surrounding the colony) ability were further analyzed for indole acetic acid (IAA) production, which was performed as described before (Li et al., 2021). In brief, a fungal piece (6 mm in diameter) was separately added to a test tube containing 10 mL PDB (PDA without agar) containing 0.1% or 1.0% tryptophan, respectively, while PDB without tryptophan was used as a control. After shaking at 25°C, 150 rpm for 7 days, approximately 2 mL of culture solution was collected and centrifuged at 4°C, 8000 rpm for 5 minutes, respectively. Finally, 1 mL of supernatant for each isolate was taken and separately mixed with 4 mL Salkowski’s reagent (0.5 M FeCl₃, 15 mL, H₂SO₄, 300 mL, ddH₂O, 500 mL). After the mixtures were incubated at room temperature for 20 minutes in the dark, the observation of pink–red color changes indicated IAA production. Absorbance was read at 530 nm using a spectrophotometer (Perkin Elmer Lambda35, Waltham, MA, USA), and IAA concentrations were calculated on the basis of the standard curve prepared by standard IAA solutions (0, 10, 20, 30, 40, 50, 100 μg/mL). Each experiment was performed in triplicate.

Plant growth promotion (PGP) activity of the selected fungal isolates was determined by evaluating their effect on growth of rice cv. Xiushui131 (purchased from Jiaxing Academy of Agricultural Sciences, China). In details, after sterilization with 75% alcohol for 1 minutes, rinsing three times with sterile water, and pre-germination on wet sterile filter papers at 25°C for two days, rice seeds were sowed into planting pots (8 cm × 8 cm × 12.5 cm) containing soil previously planted with vegetables, and then put in a greenhouse with a relative humidity of 70% and temperature of 25°C. Two days after sowing, 10 mL conidial suspension of each tested isolate (prepared by scraping the conidial masses on a 7-day-old culture grown on PDA at 25°C to sterile distilled water, and diluting to 10⁶ spores/mL) was poured into the soil surrounding the plants. Plants irrigated with 10 mL sterile water were used as controls. After 55 days of inoculation, the PGP ability was determined by measuring the seedlings height, root length, fresh and dry weight (dried in an oven at 65°C for two days), and the growth promotion efficacy (GPE) was calculated using the formula: GPE% = (treatment – control)/control × 100%. Each treatment consisted of three replicates with four pots (six plants per pot) per replicate (total 72 plants per treatment).

At harvest of rice, about 1.0 kg of soil from each treatment was used to detect soil pH, SOM, TP, and AP as previously described. Indeed, after drying at room temperature and passing through a 0.45-mm sieve, soil pH was measured using a pH meter (FE28, Mettle-Toledo, Zurich, Switzerland); SOM content was measured by the potassium dichromate volumetric method; TP was determined by automated colorimetric analysis after persulfate digestion; AP was measured by molybdenum-antimony anti-spectrophotometry following extraction with hydrochloric acid ammonium fluoride (Jackson, 1973; Brookes et al., 1985; Vance et al., 1987; Baran et al., 2019).

Simultaneously, 10 g of root-zone soil around rice plants was sampled from each treatment and stored it at -80°C for further genome sequencing. In detail, genomic DNA was extracted from 1 g of soil samples using the E.Z.N.ATM Mag–Bind Soil DNA Kit (Omega, USA) following the manufacturer’s instructions, and then the DNA quality was assessed using a Qubit 4.0 (Thermo, USA). Here, the V3-V4 region of bacterial 16S rRNA genes and the ITS1 region of fungal ITS genes were amplified using the universal primers 341F (5′−CCT ACG GGN GGC WGC AG−3′) and 806R (5′−GGA CTA CHV GGG TWT CTA AT−3′) (Wu et al., 2015), and ITS1F (5′−CTT GGT CAT TTA GGA AGT AA−3′) and ITS2 (5′−GCT GCG TTC TTC ATC GAT GC−3′) (Adams et al., 2013), respectively. PCR mixture consisted of 2×Hieff^® Robust PCR Master Mix (15 μL), 10 μM universal primer (1 μL of each primer), DNA template (1 μL), and ddH₂O (12 μL). PCR thermal protocol consisted of an initial 3 minutes denaturation step at 94°C, 25 amplification cycles of 94°C for 30 s, 55°C for 30 s, 72°C for 30 s, and a final extension step of 5 minutes at 72°C. PCR products were purified using Hieff NGS™ DNA selection beads (Yeasen, China), and then pooled in equimolar concentrations and sequenced using the 2 × 250 bp pair-end sequencing protocol on an Illumina MiSeq system (Sangon Biotechnology Co., Ltd., Shanghai, China).

After sequencing, raw data were assembled using PEAR (v0.9.8) and preprocessed using PRINSEQ to remove low-quality reads (average quality score < 20) to ensure high data quality (Schmieder and Edwards, 2011; Zhang et al., 2014). After primers were trimmed with Cutadapt (v1.18), clean reads were clustered into operational taxonomic units (OTUs) of ≥97% similarity using Usearch (v11.0.667) (Martin, 2011; Edgar, 2013, 2016). After selection of the representative read of each OTU using the QIIME package (v2020.06), all bacterial and fungal OTU representative sequences were classified taxonomically by blasting against the RDP database and UNITE database, respectively (Altschul et al., 1997; Wang et al., 2007; Caporaso et al., 2010; Quast et al., 2013; Urmas et al., 2013). Among them, relative abundances (RAs) of Aspergillus and Talaromyces species, and their relationships with environmental factors were especially focused using Origin software (v2023, Hampton, USA) and redundancy discriminant analysis (RDA), as well as correlation network.

One-way variance analysis (ANOVA) was performed using SPSS (v16, SPSS, USA). OTUs and alpha diversity indices (Chao1 and Shannon index) were analyzed by Origin software. The difference of microbial communities among different samples was measured by principal component analysis (PCA) based on beta diversity metrics from the Bray-Curtis metrics (Ramette, 2007), and the significant difference was tested using permutational multivariate ANOVA (PERMANOVA) with 999 permutations used to calculate p-values (Dixon, 2003). RAs (at the phylum and genus level, respectively) and heat map (at the family level) of the dominant microbes were calculated using Origin software. Differential biomarkers between groups were discovered by linear discriminant analysis effect size (LEfSe) (Segata et al., 2011). Different environmental factors on microbial community structure were calculated using RDA. Diagrams of the correlation network between soil properties and microbial taxa were performed using R software (v4.1.3).

The isolated fungi were screened to determine their plant growth-promoting traits. Results showed that 15 isolates displayed the ability of phosphate solubilization, siderophore production, and IAA production ( Figures 1 , 2 ). Indeed, as shown in Table 1 , the diameter of the phosphate-solubilizing halo ranged from 11.91 to 31.65 mm, among which, isolates JD-L31f (31.65 mm), TL-B21f (28.96 mm), and FY-R71f (28.43 mm) exhibited the highest phosphate solubilization ability ( Figure 1 upper). Moreover, the siderophore production ability was tested by growth of fungal isolates on CAS media. Results showed that all 15 isolates could secrete iron carriers. Among them, isolates YH-R21f, TL-B31f, LA-R21f, and YH-R31f displayed the siderophore-containing area of 24.66, 24.41, 24.16, and 23.66 mm, respectively, which were significantly (p < 0.05) higher than those of the other isolates ( Figure 1 lower). Finally, all isolates had the ability to produce IAA in PDB medium amended with 0.1% or 1.0% tryptophan. At 0.1% tryptophan, the highest IAA production was obtained in isolates TL-B21f (50.23 μg/mL) and YH-R21f (31.18 μg/mL), while at 1.0% tryptophan, the highest IAA production was achieved in isolates FY-G-R31f (96.50 μg/mL) and TL-B31f (95.29 μg/mL) ( Figure 2 ).

All 15 fungal isolates were further examined for their growth-promoting activity in rice seedlings. After 55 days of inoculation, all 15 fungal isolates could cause a noticeable increase in rice growth compared to the control based on the phenotypic observation ( Figure 3 ). The measured data further demonstrated that the fungal 15 isolates significantly (p < 0.05) affected the growth and biomass accumulation of rice seedlings compared to the control ( Table 2 ). Indeed, among all treatments, rice seedlings inoculated with various fungal isolates exhibited different increases in plant height. Isolate TL-B31f exhibited the highest increase in plant height (359.59 mm), which was about 1.15 times that of the control. Isolates FY-R11f and FY-R31f followed, with 14.58% and 13.90% increases in plant height, respectively, as compared to that of control seedlings. Moreover, the fungal strains exhibited significantly (p < 0.05) different growth-promoting effects on root length of seedlings. Isolate FY-R41f showed the highest increase in root length (205.47 mm), with an increase of 43.31% compared to that of the control, followed by isolates FY-G-R31f and TL-B11f with an increase of 37.41% and 33.98% compared to that of the control, respectively. Furthermore, compared to the control, the fresh weights of seedlings were significantly (p < 0.05) increased by isolates CA-R11f, FY-R41f, and TL-B31f. In particular, the highest fresh weight of seedlings was achieved by isolate CA-R11f, with an increase of 95.09% with respect to that of the control. In addition, the dry weight of seedlings treated by isolates FY-R41f, FY-R31f, and FY-R11f was significantly (p < 0.05) higher than that of the control, with increases of 28.12%, 19.58%, and 16.39%, respectively. In general, the quality of rice seedlings treated with different fungal isolates was higher than that of the control.

After being cultured on PDA, CYA, and MEA medium at 25°C for 7 days, it was revealed that 15 fungal isolates belong to Aspergillus sp., Penicillium sp., and Talaromyces sp. based on the obvious distinctions of colony morphology ( Figure 4 ). Indeed, eight isolates including JD-L31f, YH-S11f, TL-B21f, YH-R21f, YH-R31f, TL-B31f, FY-R71f, and FY-R11f were attributed to the genus Aspergillus due to the coffee brown to black colonies. Three isolates including FY-G-R31f, TL-B11f, and CA-R11f were attributed to the genus Penicillium in view of the irregular, green, velvety texture of colonies. Four isolates including LA-R21f, FY-R41f, FY-R31f, and JD-R21f were attributed to the genus Talaromyces due to the yellow-green colonies and a characteristic red pigment (Tsang et al., 2018; Houbraken et al., 2020). To further identify the species of all isolates, phylogenetic analysis was conducted by using the maximum likelihood method based on sequences of several conserved genes including ITS, TUB, CaM, and RPB2 of the tested isolates and closely related species ( Table 3 ). The phylogenetic trees with the highest log likelihood were drawn, with branch lengths reflecting evolutionary distance and bootstrap proportions beside the branches indicating credibility.

The phylogenetic trees showed that all 15 isolates were separated into six distant clades ( Figure 5 ). Indeed, isolates JD-L31f, YH-S11f, and TL-B21f were clustered with A. brunneoviolaceus isolate NRRL4912 as a distinct clade with a bootstrap of 96%, suggesting the three isolates should be considered as A. brunneoviolaceus. Isolates YH-R21f, YH-R31f, TL-B31f, FY-R71f, and FY-R11f were clustered with A. tubingensis isolate NRRL62644 as a distinct clade with a bootstrap of 100%, suggesting the five isolates should be considered as A. tubingensis ( Figure 5a ). Isolate FY-G-R31f was clustered with Penicillium oxalicum isolate CBS219.30 as a distinct clade with a bootstrap of 100%, suggesting this isolate should be considered as P. oxalicum. Isolates TL-B11f and CA-R11f were clustered with Penicillium limosum isolate CBS339.97 as a distinct clade with a bootstrap of 100%, suggesting the two isolates should be considered as P. limosum ( Figure 5b ). Isolate LA-R21f clustered with Talaromyces purpureogenus isolate B195_0419 as a distinct clade with a bootstrap of 100%, suggesting this isolate should be considered as T. purpureogenus. Isolates FY-R41f, FY-R31f, and JD-R21f were clustered with T. veerkampii isolate NRRL6095 as a distinct clade with a bootstrap of 100%, suggesting the three isolates should be considered as T. veerkampii ( Figure 5c ).

Two PGPF isolates TL-B31f and FY-R41f, with the best growth-promoting properties, were selected for analyzing their effects on soil quality, microbial community composition, and function after 55 days of inoculation. As shown in Table 4 , compared with the control, the soil pH and TP were generally unaffected by At-TLB31f (inoculated with isolate TL-B31f) and Tv-FYR41f (inoculated with isolate FY-R41f) treatments. Meanwhile, At-TLB31f treatment caused a 4.38% decrease (p < 0.05) in SOM compared to the control, while there was no significant difference between Tv-FYR41f treatment and the control. Furthermore, AP was significantly (p < 0.05) increased by At-TLB31f (42.52%) and Tv-FYR41f (48.51%) treatments compared to the control.

Based on high-throughput amplicon sequencing of DNA extracted from soil samples, bacterial and fungal taxonomic diversity (α-diversity) and composition (β-diversity) were compared between treatments inoculated by isolate TL-B31f or FY-R41f and the control. Indeed, the average bacterial Chao1 index was 2,087 (1,977–2,169), 2,123 (2,000–2,262), 2,133 (2,043–2,228), and the average bacterial Shannon index was 7.07 (7.00–7.16), 7.14 (7.02–7.26), and 7.07 (6.97–7.13) in At-TLB31f, Tv-FYR41f treatments, and the control, respectively ( Figures 6A1, A2 ). In general, the bacterial Chao1 index was lower (2.17% and 0.48%) in At-TLB31f and Tv-FYR41f treatments, along with a higher Shannon index (0.13% and 1.00%) than that in the control. Meanwhile, the average fungal Chao1 index was 223 (199–243), 195 (187–202), 191 (175–207), and the average fungal Shannon index was 3.89 (3.71–4.03), 3.70 (3.53–3.83), and 3.85 (3.82–3.88) in At-TLB31f, Tv-FYR41f treatments, and the control, respectively ( Figures 6B1, B2 ). In other words, isolate TL-B31f inoculation caused a 16.49% (p < 0.05) and 1.00% increase in the fungal Chao1 and Shannon index of soil, while isolate FY-R41f inoculation caused a 2.21% increase and 4.00% reduction, respectively, compared to the control. Overall, there was no significant (p < 0.05) difference in the α-diversity of soil bacteria between two PGPF isolates treatments and the control, but the α-diversity of soil fungi was significantly (p < 0.05) changed by inoculation of isolates TL-B31f and FY-R41f.

Results of PCA revealed that the OTU abundance from 12 soil samples of At-TLB31f, Tv-FYR41f treatments, and the control comprised three different groups, but there was noticeable overlap among all three different treatments, regardless of bacterial or fungal data ( Figures 6A3, B3 ). Indeed, the two principal components accounted for 77.98% (PC1 68.94%, PC2 9.04%, respectively) and 63.56% (PC1 54.35%, PC2 9.21%, respectively) of the total variance in the bacterial and fungal communities, respectively. PERMANOVA performed on all samples also indicated that different PGPF explained 16.6% (p = 0.054) and 23.2% (p = 0.030) of the variation, respectively. These findings showed that the bacterial and fungal communities in soil inoculated by isolate TL-B31f or FY-R41f were generally similar to the control, albeit with some differences.

The compositional differences of soil bacterial and fungal communities among the three different treatments were compared at the phylum and genus levels. Indeed, in the bacterial community, Pseudomonadota, Chloroflexota, Acidobacteriota, Bacteroidota, Bacillota, Actinomycetota, Cyanobacteriota, Gemmatimonadota, Myxococcota, and Thermodesulfobact were the top 10 dominant phyla ( Figure 7A1 ), and OLB13, Luteitalea, Aggregatilinea, Subgroup 10, Vibrionimonas, MND1, Gemmatimonas, Sphingomonas, Vicinamibacter, and Lysobacter were the top 10 dominant genera ( Figure 7A2 ). Compared with the control, the RAs of Vibrionimonas (22.12%) and Vicinamibacter (10.13%) were increased, while Lysobacter (62.56%), Gemmatimonas (11.76%), Sphingomonas (8.17%, p < 0.05), and MND1 (7.64%) were decreased in At-TLB31f treatment; the RAs of Vibrionimonas (30.15%), OLB13 (17.32%), Aggregatilinea (11.83%), Lysobacter (10.29%), and MND1 (6.79%) were decreased, while Sphingomonas (19.68%, p < 0.05), Vicinamibacter (7.92%), and Luteitalea (5.73%) were increased in Tv-FYR41f treatment ( Figure 7A2 ).

Similarity, according to the distribution and RAs of fungi among all three different treatments at the phylum level, Ascomycota, Basidiomycota, Mucoromycota, and Blastocladiomycota were the main fungal phyla, with abundance of 73.80–76.11%, 8.12–14.09%, 9.20–13.03%, and 0.40–2.93%, respectively ( Figure 7B1 ). Furthermore, at the genus level, Botryotrichum, Mortierella, Fusarium, Triangularia, Phlebiopsis, Curvularia, Hyphodermella, Acrophialophora, Podospora, and Mariannaea were the top 10 dominant genera ( Figure 7B2 ). Compared with the control, the At-TLB31f treatment caused a 40.29%, 36.68%, 34.39%, 33.16%, 24.37%, and 9.45% reduction in Curvularia, Acrophialophora (p < 0.05), Botryotrichum, Mortierella, Mariannaea, and Triangularia, respectively, while a 641.74%, 98.87%, 49.74%, and 6.07% increase in Podospora, Phlebiopsis, Hyphodermella, and Fusarium, respectively ( Figure 7B2 ). In addition, the RAs of Aspergillus were significantly increased by 264.68% in At-TLB31f treatment, while Talaromyces were significantly increased by 1328.35% in Tv-FYR41f treatment ( Figure 7B3 ). In other words, compared to the control, the microbial community composition of the two PGPF isolates treatments was reconstructed with an increase or decrease of some specific microbes.

Furthermore, LEfSe was carried out to identify specific bacterial and fungal biomarkers distinguishing soil microbial communities among At-TLB31f, Tv-FYR41f treatments, and the control ( Figure 8A ). Results showed that 14 bacterial biomarkers (LDA > 3.0) and 18 fungal biomarkers (LDA > 3.5) were present in all treatments. In fact, communities of At-TLB31f treatment were enriched with Agaricales, Aspergillus, Aspergillaceae, Fragosphaeria, Pezizales, Pezizomycetes, Ophiostomataceae, and Ophiostomatales. Communities of Tv-FYR41f treatment were enriched with Caldilineaceae, Caldilineales, Desulfobulbia, Desulfuromonadia, Ellin6067, Geobacteraceae, Geobacterales, Geomonas, Lysobacteraceae, Pseudolabrys, Ramlibacter, Thermodesulfobacteriota, and Cystobasidiomycetes, Eurotiales, Eurotiomycetes, Trichocomaceae, Talaromyces. The controls were enriched with Hydrogenophaga, Lysobacter, and Blastocladiomycetes, Blastocladiales, Boudiera, Catenariaceae, and Catenaria. Overall, these microbial taxa might play important roles in modulating rice growth.

RDA was performed to analyze how soil microbial community composition correlated with soil environmental factors. Indeed, SOM (r² = 0.74, p = 0.004) and AP (r² = 0.38, p = 0.117) were the most important factors in explaining the variation in the composition of the bacterial community, with axes 1 (19.30%) and axes 2 (11.02%) accounting for 30.32% of total variations ( Figure 8B1 ). Meanwhile, AP (r² = 0.80, p = 0.004), TP (r² = 0.41, p = 0.080), pH (r² = 0.37, p = 0.135), and SOM (r² = 0.26, p = 0.281) were vital factors in explaining the variation in the composition of the fungal community, with axes 1 (16.05%) and axes 2 (11.96%) accounting for 28.01% of total variations ( Figure 8B2 ). Simultaneously, correlation network analysis also indicated that soil properties obviously influence the composition of microbial communities at the genus level ( Figure 8C ). Results showed that Ellin6067, Lysobacter, and Phaeodactylibacter were significantly (p < 0.05) positively correlated with SOM. Acidiferrimicrobium was significantly (p < 0.05) positively correlated with AP, while Curvularia was significantly (p < 0.05) negatively correlated with AP ( Figure 8C1 ). Mortierella was significantly (p < 0.05) negatively correlated with pH and AP. Phlebiopsis and Hyphodermella were significantly (p < 0.05) positively correlated with TP. Talaromyces was significantly (p < 0.05) positively correlated with pH, while Aspergillus was significantly (p < 0.05) negatively correlated with pH ( Figure 8C2 ).

In order to further enhance the productivity of land converted from non-grain to grain cultivation and facilitate the growth of food crops on such soils, we conducted an isolation of soil fungi by collecting soil samples from six sites where land use had been converted from non-grain to grain cultivation, yielding a total of 215 fungal isolates from 108 soil samples. These obtained fungal isolates were further screened and characterized based on their ability to solubilize phosphate, to produce siderophores, and to synthesize IAA, which has been widely reported to play an important role in growth promotion of plants ( Figure 9 ).

Phosphorus is a key component of many vital substances and structures within plant cells, such as nucleoproteins, phospholipids, and nucleic acids. It plays a critical role in the physiological and biochemical processes throughout the plant’s life cycle and is indispensable for plant growth and development (Han et al., 2022). However, most phosphorus in the soil exists in insoluble inorganic or organic forms, making it difficult for plants to absorb and utilize (Wang et al., 2021). In this study, initial screening showed that 15 fungal isolates were capable of producing clear phosphate solubilization zones on NBRIP medium, demonstrating a broad ability to solubilize phosphate. Phosphate-solubilizing PGPF can dissolve insoluble forms of phosphorus or mineralize organic phosphorus through mechanisms such as the secretion of organic acids, excretion of protons, or enzymatic production, thereby making phosphorus available for plant uptake (Chen et al., 2019). In agreement with the result of this study, some fungal species including A. brunneoviolaceus, Aspergillus niger, Nigrospora sphaerica, Penicillium corylophilum, and Penicillium chrysogenum have been reported to enhance plant growth by participating in phosphate solubilization (Hassan, 2017; Li et al., 2023).

Additionally, previous studies showed that PGPF that produce siderophores also play a significant role in plant growth and development by improving plant resilience to stress and iron limitation (Sharma et al., 2024). Similarly, our study confirmed that all 15 fungal isolates possess the ability to produce siderophores. PGPF not only promotes crop growth by producing siderophores but also enhances the iron content of food crops, hence addressing nutritional deficiencies in human diets (Kaur et al., 2020). In addition, various phytohormones (particularly IAA) could be produced from PGPF, which could either directly enhance plant growth through regulating root growth, or indirectly induce tolerance to stress situation by modifying growth inhibitors (Ansabayeva et al., 2025). For instance, fungi-produced IAA has been reported to positively influence the root development of eggplant plants (Li et al., 2021). In agreement with previous results, our study indicated that IAA can be produced by the selected PGP fungal isolates in the presence of 0.1% and 1.0% tryptophan, indicating that the fungi isolated from the converted farmland have the potential to stimulate root growth and development. However, the IAA production of these fungal isolates can be affected by the concentration of tryptophan.

Interestingly, the bioassays indicated that all 15 fungal isolates screened in this study differed in their performance in promoting rice growth. For instance, isolates FY-R41f and FY-R31f did not display particularly strong capabilities in phosphate solubilization, siderophore production, or IAA synthesis during in vitro assays, yet they significantly enhanced rice seedling growth in bioassays, substantially increasing seedling height, root length, fresh and dry weight. The discrepancies between in vitro and in vivo results may be mainly due to differences in fungal metabolites or colonization abilities of fungal species in soil. Additionally, several plant growth-promoting isolates, including TL-B31f, FY-R11f, TL-B11f, and CA-R11f were non-pathogenic to the host rice, but significantly increased rice seedling height, root length, fresh weight, and dry weight. The PGP effect on rice seedlings depends on fungal isolates.

Based on a combination of morphological characteristics and phylogenetic analysis of multiple fungal gene sequences, all 15 fungal isolates were classified into six species: A. brunneoviolaceus, A. tubingensis, P. oxalicum, P. limosum, T. veerkampii, and T. purpureogenus. Previous studies have highlighted the potential applications of several of these isolates in agriculture through various mechanisms ( Table 5 ). For example, P. oxalicum isolate UOM PGPF 16 had been used as PGPF to enhance plant growth and induce resistance in pearl millet against downy mildew disease (Murali and Amruthesh, 2015). P. limosum AK-7, isolated from soil, has demonstrated significant antifungal activity, making it a promising resource for developing biological pesticides (Basavarajappa et al., 2023). T. purpureogenus isolate M13026-2, isolated from mushroom substrate, could colonize in the rhizosphere soil stably, and significantly improve the growth of cucumber seedlings (Zhao et al., 2021). A. tubingensis demonstrated the capacity of increasing the bioavailability of phosphorus and potassium in the rhizosphere, and the ability of controlling Fusarium spp (Zapata et al., 2024). Similarly, in our earlier research, we isolated A. tubingensis HZ123, which showed potential as a PGPF by improving eggplant growth and soil fertility (Li et al., 2021).

The novelty in this study is the significant enhancement of rice growth and yield by T. veerkampii. Notably, there is limited literature on the agricultural application of T. veerkampii. In other words, our research further confirms the plant growth-promoting capabilities of A. tubingensis and P. limosum, isolated from non-grain soils, and highlights the potential of T. veerkampii to improve non-grain soils and promote crop growth. These findings provide valuable fungal isolates for the development of novel fungi-based products, such as bio-organic fertilizers and bio-amendments, grounded in PGPF. Furthermore, this study lays a theoretical foundation for improving non-grain soil conditions and enhancing crop growth. Further research should be conducted to elucidate the growth-promoting mechanisms of these fungal isolates.

Among all 15 PGPF isolates, isolates TL-B31f (identified as A. tubingensis) and FY-R41f (identified as T. veerkampii) with the best growth-promoting properties were selected for further study. In fact, results showed that the soil nutrient elements could be significantly improved by inoculation of isolates TL-B31f and FY-R41f, in particular, the AP of 27.33 and 28.47 mg/kg, respectively, were greater than that of the control (19.17 mg/kg). Aligning with this result, our previous studies indicated that the application of A. brunneoviolaceus isolate HZ23 caused a significant increase in AP compared to the control (854.09 mg/kg vs 38.82 mg/kg) (Li et al., 2023). Phosphate application in farmland could enhance crop yield, while to increase phosphate use efficiency, and improve the fertility of low phosphate soils, which makes it a promising approach for sustainable agriculture (Li et al., 2015; Wang et al., 2024). Therefore, it can be inferred that isolates TL-B31f and FY-R41f may have a great effect on the improvement of the soil quality of non-grain cultivated land.

According to previous research, the effect of PGPF on various aspects of plant growth was variable, depending mostly on their successful survival, colonization, growth, and efficient adaptation to environmental conditions. Root colonization was an important strategy of PGPF for plant growth promotion. Upon colonization of the host plant, PGPF can change the microbial community features in the soil, thereby improving the plants’ well-being and productivity through a beneficial association with plants (Adedayo and Babalola, 2023). After 55 days of inoculation, isolates TL-B31f and FY-R41f had an obvious effect in increasing the richness of fungal communities, while the bacterial communities were not significantly changed. In all treatments, Pseudomonadota, Chloroflexota, Acidobacteriota, Bacteroidota, OLB13, Luteitalea, Aggregatilinea, Subgroup 10, Vibrionimonas, MND1, Gemmatimonas, and Sphingomonas were the main bacteria, while Ascomycota, Basidiomycota, Mucoromycota, Botryotrichum, Mortierella, Fusarium, Triangularia, and Phlebiopsis were the main fungi. In fact, more attention should be paid to Pseudomonadota, Chloroflexota, Acidobacteriota, Sphingomonas, Ascomycota, Basidiomycota, Botryotrichum, Mortierella, Fusarium, Triangularia, and Phlebiopsis. Pseudomonadota can mediate nutritional and growth-promotional activities for sustainable food security (Sah et al., 2021). Chloroflexota members may play a role in the nitrogen cycle, thereby improving the nitrogen removal performance in anammox bioreactors and activated sludge systems (Bovio-Winkler et al., 2024). Acidobacteriota is often an important contributor to the nutrient cycling system in soil microhabitats (Zhu et al., 2022a). Sphingomonas possess multifaceted functions ranging from the remediation of environmental contaminants to the production of highly beneficial phytohormones (Asaf et al., 2020). Ascomycota and Basidiomycota have the ability to degrade soil organic matter by producing cellulolytic enzymes (Manici et al., 2024). Mucoromycota can produce lipids, ethanol, organic acids, pigments, and enzymes. Thus, it has been considered a powerful cell factory for modern biorefinery (Dzurendova et al., 2022). Botryotrichum can produce some secondary metabolites that are unfavorable for plant growth (Elkhateeb and Daba, 2022). Mortierella has the ability to increase nutrient uptake efficiency, which has a positive effect on crop protection against adverse conditions, and reduces the use of chemical fertilizers and pesticides (Ozimek and Hanaka, 2020). Fusarium can cause severe economic damage in different agricultural productions such as rice, wheat, potato, etc (Ekwomadu and Mwanza, 2023). Triangularia often grows as saprophytes in the ground among leaf litter or in association with plant roots (Huang et al., 2021). Phlebiopsis has been used as a biocontrol agent against Heterobasidion annosum (Terhonen et al., 2013).

Compared with the control, the RAs of all these microbes were changed by the application of isolates TL-B31f and FY-R41f to potentially increase soil health and decrease rice morbidity. It was particularly worth mentioning that the RAs of Aspergillus were significantly increased by 264.68% in At-TLB31f treatments, while Talaromyces were significantly increased by 1328.35% in Tv-FYR41f treatments. In addition, RDA and correlation network analysis of microbes with soil properties indicated that SOM was significantly positively correlated with Ellin6067, Lysobacter, and Phaeodactylibacter; AP was significantly positively correlated with Acidiferrimicrobium and Mortierella, but was significantly negatively correlated with Curvularia; TP was significantly positively correlated with Phlebiopsis and Hyphodermella; pH was significantly negatively correlated with Mortierella and Aspergillus, but was significantly positively correlated with Talaromyces. In agreement with the results of this study, Guo et al. (2021) showed that SOC, TN, and TP were the most important factors explaining variations in the bacterial community structure. Li et al. (2023) reported that the composition of bacterial communities in packchoi rhizosphere soil was affected significantly by AP, pH, OMC, and TN. Arunrat et al. (2023) found that soil bacterial community was strongly influenced by organic matter and organic carbon in Maize field. The growth of microbes in bayberry rhizosphere soil was affected by many environmental factors, including pH, OMC, AP, and exchangeable magnesium (Ren et al., 2021). Taken overall, compared to the control, the microbial community composition of the two PGPF isolates treatments was reconstructed with an increase or decrease of some specific microbes, and soil properties might play important roles in influencing the growth of microbial communities to modulate rice growth.