The soil was collected from a long-term experimental field at the Quzhou Experimental Station (36.87°N, 115.02°E) in Quzhou County, Hebei Province, North China. The soil is classified as a Cambisol with a silt loam texture. The background soil properties were as follows: pH 7.24 (H₂O), 7.95 g kg⁻¹ organic carbon (SOC), 0.90 g kg⁻¹ total N (TN), 12.01 mg kg⁻¹ Olsen-P (AP), and 176.2 mg kg⁻¹ available K (AK) (Zhang et al., 2016). The collected soil was sieved through a 2-mm sieve, air-dried, and mixed with sand to a soil-to-sand ratio of 1:1 (m/m) as the cultivated medium. The medium was gamma-irradiated with a maximum dose of 32 kGy to eliminate indigenous AMF. To ensure nutrient supply, 100 mg kg⁻¹ of N (calcium nitrate tetrahydrate), 20 mg kg⁻¹ of P (monopotassium phosphate), and 100 mg kg⁻¹ of K (monopotassium phosphate) were added to the medium prior to the experiment.

The experiment followed a full factorial design with two factors: 1) patch types: organic patch and control patch and 2) inoculation treatments: inoculated with AMF (+AMF) and without AMF (−AMF).

Mesocosm units were constructed by two joining compartments (one host compartment, 3 × 10 × 15 cm³; one hyphal compartment, 10 × 10 × 15 cm³) via a 30-μm mesh that allowed hyphae, but not plant roots, access to the hyphal compartment ( Supplementary Figure 1 ). The host compartment contained 450 g of the sterilized medium and 50 g of inoculum of Funneliformis mosseae in the +AMF treatment or 50 g of the sterilized inoculum with microbial filtrate from the unsterilized inoculum in the −AMF treatment. Funneliformis mosseae is a common AMF in Chinese agricultural soils. The inoculum consisted of spores, mycelium, and fine root segments and was propagated using the same cultivated medium and planted with Zea mays L. for 4 months in a greenhouse. The hyphal compartment contained 1,500 g of the sterilized medium. A sterile PVC pipe (3.3 cm wide, 11 cm high) was inserted into the hyphal compartment 3 cm from the host compartment and 4 cm from the base of the hyphal compartment. Two opposing aperture windows (2.5 cm wide, 4 cm high) were opened 1 cm from the base of the PVC pipe. These windows and the base of the pipe were covered with a 30-μm mesh ( Supplementary Figure 1 ).

Two pregerminated maize seeds were grown in each host compartment and thinned to one plant after emergence. The patch was added to the PVC pipe 29 days after maize planting. The patch materials contained 14 g of the sterilized medium mixed with 7 g of milled dried roots of maize plants (total carbon 40.62%, total N 1.43%, C:N ratio 28.41) as the organic patch or 50 g of the sterilized medium as the control patch (to ensure equal volume with the organic patch). Then, 5 mL of microbial filtrate was pipetted into each patch to equalize the initial microbial communities. The microbial filtrate was produced from the same fresh field soil according to the previously described method (Koide and Li, 1989; van der Heijden et al., 2006). The mesocosm experiment was conducted in a greenhouse at China Agricultural University, Beijing. Soil moisture was maintained at approximately 50%–60% of water-filled pore space (WFPS) with deionized water.

Thirty days after patch addition, the PVC pipe was sealed at the top with a rubber plug that included a syringe needle and a three-way valve to form a gas sampling port (5 cm high). The total internal volume of the gas port was approximately 35 cm³ ( Supplementary Figure 1 ). Two days after sealing the pipe, the gas within the pipe was sampled with a syringe through the three-way valve. A 10-cm³ gas sample was slowly removed from the pipe, waiting for an additional 5 s to allow the sample to mix inside the syringes before removing the syringes. The N₂O concentrations of the gas samples were determined using a model 7890A gas chromatograph (Agilent, Santa Clara, CA, USA).

Mesocosms were destructively sampled after gas sampling. Samples of patches in the hyphal compartment were collected and divided into three portions: one for the determination of soil gravimetric water content, mineral nitrogen, and DOC and total dissolved nitrogen (TDN) contents was stored at 4°C; one for the analysis of total carbon (TC) and TN contents and mycorrhizal hyphal length density (HLD) was stored at room temperature; and the remaining soil for molecular analysis was immediately frozen with liquid nitrogen and stored at −80°C.

In the host compartment, the shoots of the maize plants were removed from the soil surface. The roots of the plants were extracted from the medium, washed, and divided into three portions: a fraction for the estimation of AMF root colonization was soaked in 50% ethanol, and the rest of the roots and shoots were dried (65°C; 48 h) for the determination of N contents and dry weights.

The gravimetric water content of the patches was measured using the oven-dried method. The ammonium (NH₄ ⁺-N) and nitrate (NO₃ ⁻-N) contents of the patches were measured using a continuous flow analysis (TRAACS 2000, Bran and Luebbe, Norderstedt, Germany) after the fresh patch samples were extracted with 0.01 mol L⁻¹ of CaCl₂ solution (Peng et al., 2013). DOC and TDN in the extracts were measured by infrared spectrometry after combustion at 850°C. The total N and C contents of the plant shoots, roots, and patch samples were determined using a FLASH Elemental Analyzer 1112 (Thermo Finnigan, Waltham, MA, USA) after the samples were ground in a ball mill.

The subsamples of roots were cleared with 10% KOH, acidified with 1% HCl, and stained with 0.05% trypan blue (Phillips and Hayman, 1970). Thirty pieces of roots from each sample were randomly selected, arranged on slides, and then mounted under a light microscope (Axioplan, Zeiss, Germany). The AMF root colonization rate was determined using the magnified intersection method (Mcgonigle et al., 1990).

The HLD of AMF in patches was determined by a modified aqueous extraction and membrane filter technique using air-dried samples of patches in triplicate (Jakobsen et al., 1992). Hyphal length density was calculated according to the modified Newman formula for calculating root length (Tennant, 1975).

The potential extracellular enzyme activities of the patch samples were measured using fluorogenically labeled substrates (Bell et al., 2013). Five fluorogenic enzyme substrates based on 4-methylumbelliferone (MUF) were used to assess enzyme activities: MUB-α-D-glucopyranoside for α-1,4-glucosidase, MUF-β-D-glucopyranoside for β-1,4-glucosidase, MUF-β-D-cellobioside for β-cellobiosidase, MUB-β-D-xylopyranoside for β-1,4-xylosidase, and MUB-N-acetyl-β-D-glucosaminide dehydrate for β-1,4-N-acetylglucosaminidase. L-Leucine-7-amido-4-methylcoumarin was used to analyze the activity of leucine amino peptidase. Briefly, fresh patch samples (equivalent to 1 g dry weight) were suspended in 100 mL of 50 mM sodium acetate buffer (pH 6.0) by shaking for 30 min, and then 200 μL of the suspensions and 50 μL of a 200-μM substrate (labeled with the MUB) were pipetted into a 96-well microplate and incubated in the dark at 25°C for 4 h. All solutions and suspensions were prepared with sterile water. A negative control was arranged with blanks (without soil suspension). After a 4-h incubation period, the microplates were measured fluorometrically with an automated luminescence spectrophotometer (FLx800 microplate, Bio-Tek Instruments, Winooski, VT, USA; emission at 450 nm, extinction at 345 nm). After correcting for the negative control and quenching effects, enzyme activities were expressed in units of nmol h⁻¹ g⁻¹ dry sample as MUB release.

The genomic DNA was extracted from fresh patch samples using the Fast DNA^® SPIN Kit for soil (MP Biomedicals, Cleveland, OH, USA) according to the manufacturer’s instructions. The quantity and quality of DNA were checked using a Nanodrop spectrophotometer (Thermo Fisher, Waltham, MA, USA). Real-time quantitative polymerase chain reactions (qPCRs) of the bacterial and archaeal amoA genes involved in nitrification and nirK, nirS, and nosZ genes involved in denitrification were conducted using a CFX96 Optical Real-Time detection system (Bio-Rad Laboratories, Inc., Hercules, CA, USA). Each reaction system was performed in a 20-µL solution containing 10 µL of SYBR Premix Ex Taq (Tli RNaseH Plus, 2×, Takara Bio, Shiga, Japan), 1 µL of DNA template, 0.5 µL of each primer (10 mM), and 8 µL of ddH₂O. The qPCR primer pairs and thermal conditions are shown in Supplementary Table 1 . Standard curves were obtained using a 10-fold dilution series (10²–10⁸ copies) of plasmid DNA containing target genes with known copy numbers. Each sample was performed in technical triplicate. The amplification efficiencies were 90%–105%, and the R ² value of the standard curves was 0.99–1.00 for all genes.

All data were first tested for normality and homogeneity of variances. To fulfill normality and homogeneity of variances, the copy numbers of N-cycling genes were log₁₀(x)-transformed, and AM fungal variables (AMF root colonization rate and HLD) were transformed by square roots. The effects of patch types, inoculation treatments and their interactions on plant and AMF variables, patch N₂O concentration, and gene copies were examined by two-way analysis of variance (ANOVA). The effects of patch types, inoculation treatments, and their interactions on soil physicochemical properties and activities of extracellular enzymes in patches which clearly did not fit the homogeneity of variances were examined by the Scheirer–Ray–Hare test. The effects of inoculation treatments on the variables within each patch type were determined by the two-tailed unpaired t-test. Multiple stepwise regression was employed to identify the most influential variables affecting N₂O concentration. Independent variables include NH₄ ⁺ and NO₃ ^– contents, as well as the copy numbers of N-cycling genes. To assess the overall extracellular enzyme activities, we calculate the standardized scores (Z scores) of six extracellular enzyme activities and sum them up to obtain a single index (Z value). Pearson correlation analysis was conducted to test the relationships between the copy numbers of N-cycling genes, HLD, physiochemical properties, and the Z value of extracellular enzyme activities. The above Scheirer–Ray–Hare test and Pearson correlation analysis were conducted using R version 4.0.4. The rest of the statistical analysis was performed using the SPSS statistical software package version 22 (SPSS, Inc., Chicago, IL, USA).

The roots of the maize plants were successfully colonized by AMF. The AMF root colonization rates were greatly increased in the +AMF treatment but not significantly affected by patch types ( Figure 1A and Supplementary Table 2 ).

The HLD in the +AMF treatment was 7.41 m g⁻¹ on average, which was significantly higher than that in the –AMF treatment, where a small number of extraradical hyphae were detected. The HLD values in the organic patches were higher than those in the control patches ( Figure 1B and Supplementary Table 2 ).

The water, NH₄ ⁺, DOC, TDN, TC, and TN contents in the organic patches were much higher than those in the control patches ( Figures 2 , 3 and Supplementary Table 3 ). AMF significantly reduced the NH₄ ⁺, NO₃ ⁻, DOC, TDN, and TN contents in the organic patches. In the control patches, only the NO₃ ⁻ and TDN contents were reduced by AMF. The patch water and TC contents were not affected by AMF, regardless of patch type ( Figures 2 , 3 and Supplementary Table 3 ).

The activities of the six extracellular enzymes in the organic patches (α-1,4-glucosidase, β-1,4-glucosidase, β-cellobiosidase, β-1,4-xylosidase, β-1,4-N-acetyl-glucosamidase, and leucine amino peptidase) were approximately 30–230 times higher than those in the control patches. AMF significantly reduced the activities of the six extracellular enzymes in the organic patches by 38% to 70%. In the control patches, the activities of α-1,4-glucosidase, β-1,4-glucosidase, and leucine amino peptidase were significantly reduced, while the activity of β-1,4-N-acetylglucosaminidase was increased by AMF. The activities of the other two extracellular enzymes (β-cellobiosidase and β-1,4-xylosidase) were not significantly affected by AMF in the control patches ( Figure 4 and Supplementary Table 3 ).

AMF significantly reduced the N₂O concentration from the organic patches and did not significantly affect the N₂O concentration from the control patches ( Figure 5A and Supplementary Table 2 ). The abundance of the bacterial amoA gene ranged from 8.60 × 10⁵ to 3.56 × 10⁶ copies g⁻¹ dry soil. The values were several times lower than those of the archaeal amoA gene (2.56 × 10⁶ to 1.35 × 10⁷ copies g⁻¹ dry soil). In the organic patches, AMF significantly reduced the abundance of the bacterial amoA gene and did not affect the abundance of the archaeal amoA gene, while in the control patches, AMF significantly increased the abundance of the bacterial and archaeal amoA genes ( Figures 5B, C and Supplementary Table 2 ).

The abundance of the nirK gene ranged from 2.82 × 10⁷ to 7.88 × 10⁹ copies g⁻¹ soil, which was approximately an order of magnitude higher than that of the nirS gene in patches (ranged from 7.58 × 10⁶ to 5.50 × 10⁸ copies g⁻¹ soil). The abundance of the nirK, nirS, and nosZ genes in the organic patches was higher than that in the control patches. In the organic patches, AMF significantly reduced the nirS gene abundance by 67% on average and did not significantly affect the abundance of the nirK and nosZ genes. In the control patches, AMF significantly increased the nirK gene abundance and did not significantly affect the abundance of the nirS and nosZ genes ( Figures 5D–F and Supplementary Table 2 ).

Multiple stepwise regression analysis showed that the abundance of the nirS gene alone was the main factor controlling the N₂O concentration in the organic patches, explaining 68.2% of the variation in N₂O concentration. In the control patches, no relevant independent variables were entered into the model ( Supplementary Table 4 ). The abundance of the bacterial amoA and nirS genes was negatively correlated with HLD in the organic patches. In the control patches, HLD was positively correlated with the abundance of the bacterial amoA gene and showed no significant linear correlation with the nirS gene abundance ( Figures 6A, C ). The abundance of the bacterial amoA gene was positively correlated with NH₄ ⁺ content (r = 0.72, P = 0.043), and the abundance of the nirS gene was positively correlated with DOC content (r = 0.76, P = 0.028) in the organic patches but not in the control patches ( Figures 6B, D ). The Z value of extracellular enzyme activities was positively correlated with DOC content in the organic patches (r = 0.75, P = 0.031; Figure 6E ).

The shoot and root biomass and nitrogen uptake of maize plants were significantly affected by the inoculation treatment but not by patch type ( Supplementary Table 5 ). AMF significantly increased the shoot and root biomass of host plants by 1.79 and 1.65 times, respectively. The shoot and root N contents of maize plants were not significantly affected by AMF except for a slight but significant decrease in shoot N content in the control patches. AMF also significantly enhanced the shoot and root N uptake of maize plants by 1.58 and 1.73 times, respectively ( Supplementary Table 6 ).

AMF hyphae can penetrate into crop residue patches to search for patchily distributed nutrients and alter carbon and nitrogen cycling during the decomposition of crop residues (Hodge et al., 2001). Growing evidence shows that AMF reduce soil N₂O emissions in the mycorrhizosphere or hyphosphere (Bender et al., 2014; Zhang et al., 2015; Storer et al., 2017; Gui et al., 2021; Zhao et al., 2021). However, the reasons for the decrease were variable and environmentally dependent. The successful colonization of extraradical hyphae in patches is a prerequisite for the effect of AMF on patch N₂O emissions. The hyphal growth rate of Funneliformis mosseae after initial infection with host plants is only about 3 mm d^–1 (Smith and Read, 2008), and time is needed for substantial hyphae to proliferate in patches. In the present study, the average HLD in the +AMF treatment after approximately 30 days of patch addition was 7.41 m g^–1, much higher compared to the –AMF treatment (Figure 1B), which is consistent with previous reports (Hodge and Fitter, 2010; Gui et al., 2021). In the present study, AMF reduced N₂O concentration from the organic patches after approximately 30 days of patch addition (Figure 5A), when crop residues in the organic patches had undergone a period of rapid degradation (Wu et al., 2019). This reduction in N₂O emissions from degraded residue patches by AMF could potentially be attributed to the decreased abundance of nirS-type denitrifiers and ammonia-oxidizing bacteria, as well as the decrease in N-substrate availability, which will be discussed in detail below.

AMF significantly reduced the abundance of the nirS gene in organic patches which negatively correlated with HLD ( Figures 5E , 6C ). Multiple stepwise regression analysis suggested that the abundance of the nirS gene best explained the decrease of N₂O concentration from organic patches ( Supplementary Table 4 ). Denitrification is carried out by microbes as an alternate respiration to conserve energy under O₂ constraints (Zumft, 1997), with either organic carbon (for heterotrophic denitrification) or reduced inorganic compounds (for autotrophic denitrification) as electron donors (Korom, 1992). In the present study, the DOC content in the organic patches was significantly decreased by AMF and positively correlated with the abundance of the nirS gene ( Figures 3A , 6D ). This result indicates that the growth of the nirS-type denitrifiers may be limited by the availability of organic carbon, which is a result of AMF interference in residue degradation. This finding is inconsistent with a previous study that found that AMF increased the content of DOC in patches, possibly due to hyphal exudates (Li et al., 2023). The possible explanation for this inconsistency is that the ratio of crop residues in the present study was 33%, which is much higher than in the previous study (1%), and the contribution of carbon released from degraded residues outweighs the carbon exuded by AMF hyphae in the patch DOC pool.

In the present study, inoculation with AMF significantly reduced all tested extracellular enzyme activities in the organic patches ( Figure 4 ). This demonstrates that the presence of AMF inhibits the activity of decomposers in degraded residue after approximately 30 days of incubation. This finding is consistent with previous studies in which AMF suppressed the decomposition of woody plant litters and wheat stubbles after incubation for 3 months in the field (Leifheit et al., 2015; Verbruggen et al., 2016). In contrast, AMF have also been reported to accelerate litter decomposition (Hodge et al., 2001). The discrepancy may be attributed to the quality, quantity, composition, and decomposing time of the residues. AMF could initially stimulate decomposers by providing hyphal exudates in the early stages of residue decomposition (Herman et al., 2012; Kaiser et al., 2015). Over time, AMF hyphae efficiently scavenge mineral nutrients, such as N released from the organic patches, and potentially suppress these decomposers by competing substrates (Leifheit et al., 2015). This is consistent with lower N availability including NH₄ ⁺, NO₃ ⁻, TDN, and TN in the organic patches due to AMF in the present study ( Figures 2 , 3 ). The low decomposition rate in the presence of AMF may mitigate the release of DOC in plant residues and offer great potential to occlude C within soil aggregates (Leifheit et al., 2015; Verbruggen et al., 2016). The decrease in DOC content by AMF in the organic patches, which is positively correlated with the Z value of extracellular enzyme activities, supports this speculation ( Figures 3A , 6E ). However, AMF only significantly decreased the abundance of the nirS gene, but not the nirK gene, in the organic patches (Figures 5D, E). Compared with nirK-type denitrifiers, nirS-type denitrifiers are more sensitive to DOC, and their populations increase with soil DOC content within a certain range. Meanwhile, nirK-type denitrifiers have a broader profile of carbon substrate utilization and do not significantly respond to alterations in soil DOC content (Hou et al., 2018; Langarica-Fuentes et al., 2018). In addition, one limitation of the study is that it did not adequately reflect the impact of AMF on the early residue degradation phase. Future studies should focus on improving the experimental design and the dynamic monitoring throughout the entire cultivation period to obtain a more comprehensive understanding of the impact of AMF on residue degradation.

In the present study, AMF significantly decreased NH₄ ⁺ and NO₃ ⁻ contents in the organic patches resulting in low N-substrate availability to nitrifying and denitrifying microbes ( Figure 2 ). AMF can detect, proliferate, and acquire N from decomposing organic matter such as clover, cellulose, albumin, DNA, and chitin (Bukovska et al., 2018) and, consequently, utilize this N primarily for their own growth and transfer some to the host plant. This is supported by the significantly improved N uptake of maize plants in the presence of AMF ( Supplementary Table 6 ). It is estimated that nearly a third of the total N detected in the AMF hyphae is derived from the organic patches (Hodge and Fitter, 2010).

In the present study, AMF hyphae significantly reduced the abundance of the bacterial amoA gene in the organic patches ( Figure 5B ), which is consistent with previous studies that showed AMF suppressed AOB and soil potential nitrification rates (Veresoglou et al., 2011; Bukovska et al., 2018). AMF can absorb NH₄ ⁺, NO₃ ⁻, and some simple organic N from the soil, with a preference for NH₄ ⁺ (Govindarajulu et al., 2005; Tanaka and Yano, 2005). AOB are a group of slow-growing microbes and generally thought to be weak competitors for NH₄ ⁺ compared with AMF (Bollmann et al., 2002; Storer et al., 2017). In the present study, AMF hyphae proliferated extensively in the organic patches and increased the N uptake of maize plants ( Figure 1B and Supplementary Table 6 ). The NH₄ ⁺ content in the organic patches decreased nearly by 50% in the presence of AMF, which was also positively correlated with the abundance of the bacterial amoA gene ( Figures 2B , 6B ). This demonstrates that AMF outclassed AOB in acquiring NH₄ ⁺ in degraded residue patches and subsequently suppressed the growth of AOB. In contrast to the organic patches, AOB and AOA abundance was promoted by AMF in the control patches ( Figures 5B, C ), which is consistent with increased AOB abundance after urea application by native AMF in a pot experiment (Teutscherova et al., 2019). A possible explanation is that AMF promoted soil aeration by improving soil aggregation (Rillig and Mummey, 2006; Morris et al., 2019).