Long-term orchard-floor mulching alters soil fungal assemblages, nutrient status, and peach fruit traits

Orchard-floor mulching is widely used for weed suppression in perennial fruit systems, yet its effects on soil fungal communities and their links to soil nutrient status and fruit traits remain insufficiently understood. Here, we conducted a 6-year field experiment in a peach orchard to compare three orchard-floor management practices: barrier fabric (BF), hairy vetch (HV), and ryegrass (RG). Soil physicochemical properties, fruit traits at harvest, and soil fungal communities were evaluated, and treatment-responsive fungal modules were identified by integrating differential abundance analysis, indicator species analysis, and co-occurrence network analysis. Compared with the other treatments, HV was associated with higher fruit weight and a higher total sugar-to-total acid ratio (TS/TA). Mulching treatments significantly altered soil physicochemical properties and fungal community composition, whereas fungal alpha diversity showed limited overall responses. Three candidate treatment-responsive fungal modules were preferentially enriched under BF, HV, and RG, respectively, and differed in both taxonomic composition and their associations with soil nutrient variables. Fruit weight was positively correlated with soil NO 3 -- N and NH 4 + -N and negatively correlated with soil pH, whereas flavor-related traits were positively correlated with AN, NH 4 + -N, NO 3 -- N, and dissolved organic carbon (DOC). Overall, long-term mulching was associated with shifts in soil fungal assemblages, and these assemblages were linked to soil nutrient conditions and fruit traits. This study provides a fungal community perspective on how orchard-floor management may be associated with soil–plant interactions in peach orchards.

Weed control is a major component of orchard management and often requires substantial labor and material inputs. In perennial fruit systems, orchard-floor mulching practices are widely adopted to suppress weeds and improve soil management. Among these practices, living cover crops are increasingly used as environmentally friendly alternatives to conventional clean tillage because they can reduce erosion, increase organic inputs, and modify soil biological activity (Pannacci et al., 2018; Dong and Zeng, 2024). Both leguminous and non-leguminous cover crops are commonly used in orchards, but they differ in residue quality, nutrient return, and their associations with soil microbial communities (Li et al., 2022b; Dung et al., 2022). Barrier fabric is also widely used in orchard systems because it provides effective weed suppression with lower labor input than repeated tillage (Prosdocimi et al., 2016; Li et al., 2018b). Given these contrasts in management mode and organic matter inputs, orchard-floor treatments may differ in their effects on soil nutrient conditions and belowground microbial communities. However, the effects of barrier fabric relative to living cover-crop systems remain insufficiently explored, particularly in long-term orchard experiments.

Previous studies have shown that orchard-floor mulching and ground-cover management can strongly influence orchard soil functioning and plant performance. Some studies indicate that these practices commonly improve soil structure, water retention, erosion control, and nutrient conservation, while their effects vary with climatic conditions, orchard age, and the type and duration of ground cover used (Giacalone et al., 2021; Fang et al., 2022; Liu et al., 2021). In apple orchards, living grass mulch has been reported to improve several soil physicochemical properties at a regional scale in China (Tang et al., 2022), and cover cropping has been shown to increase soil organic matter, microbial biomass carbon, and total nitrogen while also increasing fungal-bacterial network complexity (Liu et al., 2022). Other orchard studies have further shown that different floor-management systems can alter rhizosphere microbial composition and soil pH or organic matter status (Birkhofer et al., 2019; Bokszczanin et al., 2023). At the same time, orchard-floor management may also influence fruit yield and traits, either positively through improved soil fertility, water availability, and biological activity, or negatively when competition for water and nutrients becomes limiting, especially under resource-poor conditions or in young orchards (Giacalone et al., 2021; Fang et al., 2022).

Most previous studies in orchard soils have focused on overall microbial diversity or shifts in dominant taxa. However, microorganisms associated with management practices may occur as coordinated assemblages rather than as isolated indicators, and identifying such assemblages may provide ecological information beyond diversity-based summaries alone (Chen et al., 2023; Gai et al., 2024; Li et al., 2021b, Li et al., 2022a). In agricultural soils, management practices such as long-term fertilization or organic amendment can reshape fungal co-occurrence network structure, and specific fungal modules have been reported to correlate with soil chemical properties or nutrient-related variables. For example, Xue et al. (2018) found that long-term fertilization altered fungal network assemblage structure and that specific fungal modules were independently associated with soil ions, while Yang et al. (2019) showed that compost amendment modified fungal co-occurrence patterns. In orchard systems, nutrient management and agricultural management have likewise been shown to influence fungal community structure and network characteristics (Muneer et al., 2021; Hulsmans et al., 2022). These findings suggest that a network- or module-based perspective may provide ecological information beyond diversity-based summaries alone (Ding et al., 2025). In the present study, we use the term candidate treatment-responsive fungal modules to refer to groups of fungal ASVs that are both significantly associated with a given mulching treatment and significantly connected within co-occurrence networks under our analytical framework, following the general analytical framework previously used to identify management-responsive fungal assemblages in microbial community studies (Hartman et al., 2018; Wang et al., 2022). This module-based perspective may help reveal structured fungal responses to orchard-floor management and their associations with soil nutrient conditions.

Hairy vetch (Vicia villosa Roth.) and ryegrass (Lolium perenne L.) represent two contrasting cover-crop functional types. Hairy vetch is a winter annual legume with high biomass production and biological N fixation, whereas ryegrass is a cool-season grass with rapid establishment, dense fibrous roots, and strong nutrient-retention capacity (Kucek et al., 2024; Thapa et al., 2018). Ryegrass is a cool-season grass with rapid establishment, dense fibrous roots, strong ground-cover capacity, and high potential for nutrient capture and erosion control. Grass-based covers are commonly used to stabilize orchard soils and retain residual nutrients, but they may also intensify competition for water or available nitrogen under some conditions. These contrasting traits suggest that the two species may differentially influence soil nutrient conditions, fungal community organization, and fruit performance in orchards, yet their long-term effects on fungal community assembly and network structure remain poorly understood. To address this gap, we evaluated a 6-year orchard-floor management experiment in a peach orchard comparing barrier fabric, hairy vetch, and ryegrass. The objectives of this study were to examine how these contrasting floor-management strategies affect (i) soil physicochemical properties, (ii) soil fungal community composition and candidate co-occurrence modules, and (iii) selected fruit traits. We further examined whether treatment-responsive fungal modules were associated with variation in soil nutrient conditions, with the aim of providing an ecological basis for optimizing orchard-floor management in peach production. These candidate modules are interpreted here as exploratory co-occurrence-based assemblage patterns rather than direct evidence of ecological interactions or functional units.

Study site, experimental design, and soil sampling

The field study was conducted in a peach orchard at the National Peach Germplasm Repository, Nanjing, Jiangsu Province, China (32°02′ N, 118°52′ E) at an elevation of 11 m. The region has a humid subtropical monsoon climate, with mean annual precipitation of approximately 1000–1100 mm, a mean annual temperature of 15.7 °C, a mean annual sunshine duration of approximately 1900 hours, a frost-free period of about 220–240 d (Guo, 2024). The experimental orchard is a managed perennial fruit production system located within an agricultural research station, and the surrounding land is dominated by managed orchards and experimental fields rather than natural vegetation. The peach cultivar ‘Zijinhuangcui’ (Prunus persica) was planted at a spacing of 3 m between rows and 2 m within rows.

A long-term orchard-floor management experiment was established in September 2016 and maintained for six consecutive years. Three treatments were compared: barrier fabric (BF), hairy vetch (HV), and ryegrass (RG). The experiment consisted of three replicate plots per treatment. Each plot comprised one tree row containing 10 trees and was considered an independent experimental unit. Each treatment consisted of three plots. Nine plots were established, each measuring 200 m 2. Each year, from late September to early October, the inter-row soil was rotary-tilled and then broadcast-sown with cover crop seeds. Ryegrass was sown at 4.5 g/m 2 and hairy vetch at 7.5 g/m 2. No mowing was conducted in the establishment year after cover crop emergence. In June of the following year, when both species had set seed and started to senesce, all plots were mowed and the residues were evenly retained on the soil surface. The same tillage and sowing procedure were repeated from late September to early October in the following year. A field view of the three mulch treatments is provided in Figure 1A. All plots received the same routine orchard management except for the orchard-floor treatment, and no herbicides were applied during the study period. Other orchard management practices were consistent across treatments. No additional orchard-floor treatment beyond the assigned experimental management was applied during the study period.

In all plots, rapeseed cake fertilizer was applied annually in late September before rotary tillage. Two holes (30 cm depth) were excavated 30 cm from the trunk on opposite sides of each tree, and 1 kg of fertilizer was placed in each hole. The fertilizer contained total nitrogen (TN), 58.77 g kg⁻¹; total phosphorus (TP), 3.31 g kg⁻¹; total potassium (TK), 6.62 g kg⁻¹; soil organic matter (SOM), 204 g kg⁻¹; and nitrate nitrogen (NO 3 ⁻-N), 2.88 g kg⁻¹.

Soil samples were collected in 2022 after six consecutive years of treatment application at two phenological stages: the young-fruit stage (early May) and the fruit maturity stage (late June). These sampling times were selected to represent key stages of fruit development and potential temporal variation in soil nutrient status and fungal communities. For each plot, five soil cores (0–20 cm depth) were collected from the managed inter-row strip in an S-shaped pattern and combined into one composite sample. Surface litter was removed before sampling, and sampling points were selected to avoid the fertilizer placement holes near the tree trunks. Accordingly, the measured soil properties mainly reflected treatment-related variation in the inter-row topsoil rather than in the immediate trunk zone. After collection, samples were transported to the laboratory on ice. Each composite sample was divided into three subsamples: one stored at 4 °C for fresh-soil analyses, one air-dried for physicochemical analyses, and one stored at −80 °C for DNA extraction.

Determination of soil physicochemical properties and fruit traits

Soil physicochemical properties were determined following established protocols. Soil pH was measured in a 1:2.5 soil-to-water suspension (Corwin and Scudiero, 2020). Total nitrogen (TN) was analyzed by Kjeldahl digestion, and available nitrogen (AN) by alkaline hydrolysis diffusion (Kirk, 1950). Ammonium nitrogen (NH 4 -N) and nitrate nitrogen (NO 3 -N) were extracted with 2 M KCl and measured by colorimetry and ultraviolet spectrophotometry, respectively. Total phosphorus (TP) was determined after alkaline fusion digestion (Liu et al., 2024), while available phosphorus (AP) was quantified in NaHCO 3 extracts using the molybdenum blue method (Jin et al., 2023). Total potassium (TK) and available potassium (AK) were measured by flame photometry after alkaline fusion digestion and NH 4 OAc extraction, respectively (Wang et al., 2014). Soil organic matter (SOM) and dissolved organic carbon (DOC) were determined by dichromate oxidation (Liu et al., 2024), and microbial biomass carbon (MBC) was measured by the chloroform fumigation–extraction method (Ao et al., 2023). The data were reanalyzed from Guo et al. (2024) and presented in a new Supplementary Table 1.

Fruit samples were collected at the fruit maturity stage to determine fruit weight (FW), total sugars (TS), and total acids (TA). Total sugars were calculated as the sum of sucrose, glucose, fructose, and sorbitol, whereas total acids were calculated as the sum of malic, quinic, and citric acids. Sugar and acid contents were measured by high-performance liquid chromatography according to Zhang et al. (2018). Figure 1B was newly generated based on previously reported data from Guo et al. (2024).

DNA extraction, sequencing, and taxonomic assignment

Total soil DNA was extracted from 5 g of soil using a TGuide S96 Magnetic Soil/Stool DNA Kit (Tiangen Biotech Co., Ltd., Beijing, China) according to the manufacturer’s instructions. The full-length fungal ITS region was amplified using the primer pair ITS1F/ITS4. Amplicons were quantified using a Qubit dsDNA HS Assay Kit and a Qubit 3.0 fluorometer (Invitrogen, Thermo Fisher Scientific, USA), pooled at equimolar concentrations, and sequenced on a PacBio Sequel II platform. Circular consensus sequences were generated from raw reads, and high-quality non-chimeric sequences were processed using DADA2 to infer amplicon sequence variants (ASVs) (Callahan et al., 2016). Fungal ASVs were taxonomically assigned in QIIME2 using a Naïve Bayes classifier trained on the UNITE v9.0 database with a confidence threshold of 0.70.

Statistical and bioinformatics analyses

All statistical analyses were performed in R version 4.3. Alpha-diversity indices were calculated in QIIME2 and visualized in R using the ggplot2 package. Principal coordinates analysis (PCoA) and non-metric multidimensional scaling (NMDS) based on Bray—Curtis dissimilarities were conducted using the vegan package. Analysis of variance and multiple comparisons were performed using base R and the agricolae package, whereas the Scheirer–Ray–Hare test was performed using the rcompanion and FSA packages. Regression analyses were conducted in base R. Correlations between fruit traits and soil variables were assessed using samples collected at the fruit maturity stage. As an auxiliary functional annotation analysis, fungal ASVs of candidate treatment-responsive modules were submitted to the FUNGuild-based annotation workflow implemented on the Wekemo Bioincloud Platform (Gao et al., 2026). Guild assignments, trophic modes, and confidence rankings were obtained from the platform output according to the underlying FUNGuild annotation rules.

Candidate treatment-responsive fungal modules were identified following the analytical framework described by Hartman et al. (2018) and Wang et al. (2022), in which treatment-associated taxa are further evaluated according to their connectivity within co-occurrence networks.

First, treatment-responsive ASVs were screened by combining differential abundance analysis in edgeR with indicator species analysis in indicspecies. The input data consisted of the raw fungal ASV count table and the sample grouping table defining the three orchard-floor treatments (BF, HV, and RG). Raw ASV counts were normalized in edgeR using the trimmed mean of M values (TMM) method, and normalized abundances were expressed as counts per million (CPM). Differential abundance analysis among treatments was performed in edgeR on the raw count table after TMM normalization, following the standard edgeR workflow. Indicator species analysis was performed on the TMM-normalized ASV table using the multipatt function in the ‘indicspecies’ package with 1,000 permutations. Only ASVs showing significant treatment associations in both analyses (P < 0.05) were retained as candidate treatment-responsive ASVs for subsequent network analysis.

Second, pairwise co-occurrence relationships among ASVs were assessed using Spearman correlation analysis based on the TMM-normalized abundance matrix across all samples with the rcorr function in the ‘Hismic’ package. Correlations were calculated between ASV pairs, and self-correlations were excluded. The interpretation of correlation strength is context-dependent, threshold selection was based on both general statistical convention and prior practice in microbial network studies. To reduce the complexity of the network and focus on robust co-occurrence patterns, we retained only significant correlations with Spearman’s ρ > 0.6 (P < 0.05) or Spearman’s ρ > 0.8 (P < 0.05) have been widely used in microbial co-occurrence network analyses (Barberán et al., 2012; Li et al., 2021a). Community structure within each network was identified using the fast-greedy community detection algorithm implemented in the ‘igraph’ package. Network clusters containing candidate treatment-associated ASVs were then examined, and modules enriched in ASVs associated with a given mulching treatment were designated as candidate treatment-responsive fungal modules. Network graphs were generated separately for the two correlation thresholds. Because correlation P -values were not adjusted for multiple testing, and network topology can be sensitive to threshold selection and sample size, these network analyses were treated as exploratory tools for detecting coordinated occurrence patterns rather than as confirmatory evidence of direct ecological interactions or treatment-specific functional units. Connected clusters containing treatment-responsive ASVs were designated as treatment-responsive fungal modules. Network graphs were generated separately for the two correlation thresholds. Given the modest sample size, these networks analyses were interpreted as exploratory tools for identifying coordinated fungal assemblages rather than as evidence of direct ecological interactions.

Cover crop treatments altered soil physicochemical properties and peach selected fruit traits

The Scheirer–Ray–Hare test showed that most soil physicochemical properties differed significantly among treatments, whereas available potassium (AK) and NH 4 -N did not (P < 0.05; Supplementary Table 1). In general, treatment effects were stronger than growth-period effects for most soil variables, as indicated by larger H values for treatment than for growth period (Supplementary Table 1).

Fruit traits also differed among treatments (Figure 1B). Fruit weight was 1.21-fold higher under HV than under BF and 1.49-fold higher under HV than under RG (both P < 0.05). The TS/TA ratio was highest under HV (Figure 1B). Several soil physicochemical properties were significantly associated with fruit traits. Fruit weight was positively correlated with NH 4 -N, NO 3 -N, and SOM, but negatively correlated with pH (Figure 1C). In addition, AN, NH 4 -N, NO 3 -N, and DOC were positively correlated with the TS/TA ratio (Figure 1D).

Cover crop treatments were associated with shifts in fungal community composition

Fungal alpha diversity showed limited responses to treatment and sampling period. Chao1 was higher under HV than under the other treatments during the young-fruit stage (Figure 2A), whereas no consistent differences in Chao1 or Shannon diversity were detected among treatments overall. The Scheirer–Ray–Hare test further indicated that neither treatment nor growth period significantly affected fungal alpha diversity (Supplementary Table 2).

By contrast, fungal beta diversity differed among treatments. Principal coordinates analysis (PCoA) and non-metric multidimensional scaling (NMDS) based on Bray–Curtis distances both showed clear separation of fungal communities among BF, HV, and RG treatments (PCoA: R² = 0.478, P < 0.01; NMDS: stress = 0.084, P < 0.01; Figures 2B, C). In contrast, sampling period had no significant effect on community clustering (PCoA: R² = 0.10, P > 0.05; NMDS: P > 0.05; Figures 2B, C).

At both sampling periods, the dominant fungal phyla and genera were broadly similar (Figure 2D). However, the relative abundances of several taxa varied among treatments. At the phylum level, Basidiomycota, Glomeromycota, Mortierellomycota, Neocallimastigomycota, Rozellomycota, and Zoopagomycota differed significantly among treatments (Supplementary Table 3). At the genus level, Alternaria, Gibellulopsis, Mortierella, Plectosphaerella, Solicoccozyma, unclassified_Agaricomycetes, and unidentified taxa differed significantly among treatments (Supplementary Table 3). In addition, Chytridiomycota at the phylum level and Cladosporium and Plectosphaerella at the genus level varied significantly between growth periods.

Treatment-responsive fungal modules were associated with soil physicochemical properties

To identify fungal assemblages associated with mulching treatments, co-occurrence networks were constructed using two positive correlation thresholds (0.6 < R < 0.8 and R ≥ 0.8; both P < 0.05) (Figures 3A, B). Five modules were identified within the combined network framework. Among them, M3, M5, and M10 were identified as treatment-responsive fungal modules associated primarily with HV, RG, and BF, respectively. The cumulative relative abundance of M3, M5, and M10, expressed as counts per million (CPM), was significantly higher under HV, RG, and BF, respectively, than under the other treatments (P < 0.05; Figure 3C).

Module composition differed among treatments (Figures 3D, E). More phyla were represented in M5 than in M3 or M10. All phyla in M3 were most abundant under HV, whereas the phyla in M5 were most abundant under RG. In M10, Chytridiomycota and Glomeromycota were most abundant under BF, whereas the remaining phyla showed higher relative abundances under HV or RG (Figures 3D, E).

Regression analyses showed distinct associations between treatment-responsive fungal modules and soil properties (Figure 4). The CPM of M3 was positively correlated with TN, SOM, AN, NO 3 -N, NH 4 -N, and MBC. The CPM of M5 was positively correlated with pH, but negatively correlated with TP and AP. In contrast, the CPM of M10 was positively correlated with TP and AP (P < 0.05).

FUNGuild annotation of ASVs in the treatment-responsive fungal modules yielded no valid functional assignments, as all entries were returned as na across trophic mode, guild, trait, growth form, and confidence ranking (Supplementary Table 4).

Genera within treatment-responsive fungal modules showed distinct associations with soil properties

Genera within M3, M5, and M10 differed in their associations with soil physicochemical properties (Figure 5). In general, genera in M3 were positively associated with TN, AN, NO 3 -N, DOC, and MBC; genera in M5 were positively associated with TK and pH; and genera in M10 were positively associated with TP and AP, but negatively associated with TN, AN, TK, AK, SOM, pH, DOC, and NO 3 -N.

According to variable importance values and Spearman correlations (Figure 5; Supplementary Table 5), Aspergillus, Gibellulopsis, Plectosphaerella, unclassified_Basidiomycota, unclassified_Agaricomycetes, and unidentified taxa in M3 were strong positive predictors of soil carbon- and nitrogen-related variables. Monascus was a positive predictor of nitrogen-related variables and SOM, whereas Clonostachys was positively associated with nitrogen-related variables but negatively associated with TP in M3. In M5, all genera except Heteroconium were positively associated with TK or pH. In particular, Archaeospora, Tetracladium, and unclassified_Agaricomycetes were important predictors of TK, whereas Conocybe, Delfinachytrium, Slopeiomyces, and Thelonectria were important predictors of pH (Figure 5; Supplementary Table 6). In M10, several genera were positively associated with phosphorus-related variables and negatively associated with other soil properties. For example, Tausonia, Scolecobasidium, Saitozyma, Mortierella, and Paracremonium were important predictors of AP, whereas Solicoccozyma, Stachybotrys, Saitozyma, and Mycofalcella were important predictors of TP (Figure 5; Supplementary Table 7).

Phylogenetic analysis further showed that ASVs in M3 and M5 were relatively closely related,whereas no clear clustering pattern was observed for the other modules (Supplementary Figure 1).

Long-term mulching was associated with contrasting shifts in soil nutrient status

Soil management practices can substantially alter soil nutrient availability and overall fertility (Fischer et al., 2020; Galindo et al., 2022). In this study, the two cover crop systems were associated with distinct shifts in soil physicochemical properties. HV was associated with higher TN and AN, which is consistent with the established role of leguminous cover crops in increasing soil N inputs through biological N fixation and residue return (Lee et al., 2021; Pott et al., 2021). By contrast, RG was not associated with a marked increase in inorganic N, but was linked to higher pH and potassium-related properties, in line with previous observations in orchard ground-cover systems (Fu et al., 2022; Zhou et al., 2023).

Both cover crop treatments were also associated with higher DOC, MBC, and SOM, likely reflecting greater organic inputs from residues and roots (Xiang et al., 2018, Xiang et al., 2022). Such increases are consistent with the role of cover crops in promoting soil carbon accumulation and improving soil fertility (Malhi et al., 2011; Li et al., 2018a). In contrast, available phosphorus was lower under HV and RG than under BF, suggesting that long-term cover cropping may alter soil P dynamics not only through plant uptake and P retention in cover-crop biomass, but also through biologically mediated P transformation processes. The size of the available P pool is regulated by multiple pathways, including plant uptake, microbial immobilization and turnover, phosphatase-mediated mineralization of organic P, and the mobilization of sparingly available P fractions (Richardson and Simpson, 2011; Hallama et al., 2019; Yang et al., 2016; Wang et al., 2021). Therefore, the lower available P observed under HV and RG may reflect a combination of greater plant P capture and shifts in rhizosphere processes involved in P cycling. These results suggest that, although long-term cover cropping may benefit overall soil quality, it may also increase competition for available P, indicating that P management should be considered in orchard systems maintained under living mulch for multiple years.

Long-term mulching was associated with distinct fungal assemblages

Fungal community composition differed among the three mulching treatments, indicating that long-term orchard-floor management was associated with distinct fungal assemblages. Among the candidate treatment-responsive fungal modules identified here, M3, M5, and M10 were preferentially associated with HV, RG, and BF, respectively. Ascomycota was the dominant phylum across these modules, indicating that members of this phylum were widely distributed among treatments. Previous studies have reported that some ascomycetous fungi are associated with residue decomposition and soil carbon and nitrogen cycling (Ma et al., 2013; Challacombe et al., 2019). Basidiomycota was more strongly represented in the RG-associated module, M5. This pattern suggests that ryegrass cover was associated with a distinct fungal community, although the underlying ecological processes remain unresolved. Some Basidiomycota have been reported to differ from Ascomycota in enzymatic capabilities related to plant-derived substrates (Manici et al., 2024), but whether such differences contributed to the observed pattern in the present study was not tested directly. Previous work has shown that Ascomycota and Basidiomycota differ in enzyme profiles while sharing many plant polysaccharide-degrading capacities (Manici et al., 2024), suggesting that shifts in their relative representation may be linked to treatment-related differences in substrate inputs. Rozellomycota was relatively more abundant in the HV-associated module. Although this group has been reported to increase under intercropping or organic amendment regimes (Yu et al., 2020; Iqbal et al., 2022), its ecological role remains poorly resolved. Mortierellomycota was enriched in the BF-associated module, M10. Given that some members of this phylum have been linked to mineral phosphorus mobilization (Wang et al., 2020), its enrichment alongside higher soil P under BF is noteworthy, although it should not be interpreted as direct evidence of function. Overall, these results indicate that different mulching practices were associated with distinct fungal community configurations. However, because the module analysis was exploratory and based on correlation patterns, the identified modules should be interpreted cautiously as candidate coordinated assemblages rather than robust evidence of direct ecological interactions.

Candidate fungal modules were associated with variation in soil properties

The overall treatment main effect was not significant, although Chao1 differed among treatments at the young-fruit stage, whereas beta diversity exhibited significant differences among treatments. This result suggests that long-term orchard-floor management altered community composition more than local taxonomic richness, with changes expressed primarily as compositional turnover among treatments rather than substantial shifts in alpha diversity. The relative stability of alpha diversity may reflect a degree of microbial community resistance or resilience to recurrent management disturbance (Shade et al., 2012), whereas the shift in beta diversity is consistent with treatment-dependent community assembly under different environmental conditions (Nemergut et al., 2013). Because soil microbial community composition can influence nutrient cycling and plant–soil interactions (van der Heijden et al., 2008; Bardgett and van der Putten, 2014), these compositional differences may have ecological relevance even in the absence of marked differences in alpha diversity. However, because functional attributes were not directly measured in this study, the functional implications of fungal turnover should be interpreted cautiously.

The candidate treatment-responsive fungal modules showed distinct associations with soil physicochemical properties, indicating that module-based patterns captured variation in the orchard soil environment. In particular, M3 was positively associated with multiple nitrogen-related variables, SOM, and MBC; M5 was mainly associated with pH and potassium-related properties; and M10 was positively associated with phosphorus-related variables. These contrasting associations were broadly consistent with the treatment-level soil patterns, with HV linked to N enrichment, RG linked to pH-related shifts, and BF linked to higher P availability.

At the genus level, the modules contained taxa associated with contrasting ecological roles. Potential opportunistic or pathogenic genera, including Plectosphaerella, Aspergillus, Thelonectria, and Fusarium, were detected in the candidate treatment-responsive fungal modules (Srivastava et al., 2018; Yang et al., 2021; Lutz et al., 2022; Mincuzzi et al., 2023), whereas genera with reported biocontrol potential, such as Clonostachys, Trichoderma, and Tetracladium, were also present (Hilton et al., 2021; Hasan et al., 2022; Tyśkiewicz et al., 2022). However, these assignments were based on taxonomic affiliation rather than on direct functional measurement. The co-occurrence of potentially beneficial and potentially harmful taxa within the same modules therefore highlights the complexity of fungal assemblages in orchard soils rather than demonstrating specific functional outcomes. More broadly, the modules identified here should be interpreted as coordinated occurrence patterns associated with management and soil conditions, not as direct evidence of ecological interactions or causal effects.

Soil conditions under mulching treatments were associated with variation in fruit traits

Fruit weight and the TS/TA ratio differed among mulching treatments, and both traits were associated with soil nutrient status, particularly nitrogen-related variables. These findings are consistent with previous studies showing that nitrogen availability can influence fruit yield and quality traits, including sugar accumulation and acid balance (Kant, 2018; Wang et al., 2018; Rubio Ames et al., 2020). Similar relationships have been reported in citrus and grape, where both the amount and form of nitrogen affected fruit quality (Liao et al., 2019; Yin et al., 2020).

In addition to nitrogen, DOC, phosphorus-related variables, and pH were also associated with fruit traits (He et al., 2022; Yu et al., 2023). The relationship between pH and fruit weight may be indirect, potentially mediated through its effects on nutrient availability and root-zone processes. Taken together, these results suggest that variation in selected fruit trait was associated with multiple interacting soil factors rather than with a single dominant variable. Because the present analyses were correlative, the mechanisms linking orchard-floor management, soil nutrient dynamics, and selected fruit traits remain unresolved. Further work integrating plant physiological or transcriptomic analyses with soil process measurements would help clarify these linkages.

Finally, we acknowledge several limitations in this study. First, the absence of root-related traits restricts our ability to define the precise mechanistic pathways between soil fungal communities and plant physiological responses. Second, our assessment of fruit quality was limited to specific traits; thus, our findings suggest that mulching potentially improves these selected parameters rather than overall fruit quality. Future studies incorporating root architecture measurements and a broader spectrum of metabolic indicators are required to fully elucidate these linkages. An additional limitation is that the experiment did not include a spontaneous-vegetation, natural-weed, or unmulched control. Because the study was designed to compare orchard-floor management options under commercial production conditions, the present results should be interpreted as differences among the three managed systems rather than as contrasts with a natural or unmanaged baseline.

In this 6-year peach orchard experiment, barrier fabric and the two cover crop systems were associated with distinct soil physicochemical conditions, fungal community structures, and fruit traits. Hairy vetch was associated with higher fruit weight and the TS/TA ratio and with increased soil nitrogen status, soil organic matter, and dissolved organic carbon. By integrating indicator species analysis with co-occurrence network analysis, we identified candidate treatment-responsive fungal modules that were differentially associated with nitrogen, phosphorus, and pH related soil properties. These findings suggest that fungal assemblages responsive to orchard-floor management may serve as potential biological indicators of soil condition under long-term mulching. However, because the present study was correlative and the network-based assemblages were inferred from taxonomic and association patterns, the identified modules should be regarded as exploratory rather than confirmatory. Causal relationships among candidate treatment-responsive fungal modules, soil nutrient dynamics, and fruit traits therefore require further validation through longer-term and function-oriented studies.