Contrasting Land-Use Legacies Reorganize Soil Resource–Function Balance and Alter Potential N₂O Emissions Following Tropical Forest Conversion

Abstract

Loss of soil organic resources after forest conversion is often assumed to constrain microbial functioning; however, resource status and functional potential may respond differently to land-use legacies. We compared soils under natural secondary forest, six-year poultry-integrated agroforestry, and 60-year paddy cultivation derived from the same regional forest type. Soil properties, microbial biomass, hydrolyzable organic N fractions, extracellular enzyme activities, nitrogen (N)-cycling genes, and potential N₂O emissions were measured using three independent field replicates per land use. To quantify the alignment between resources and microbial functioning, we separately standardized indicators of retained organic resources (soil organic carbon (SOC), total N, microbial biomass C, and four hydrolyzable organic N fractions) and microbial functional potential (four extracellular enzymes and five N-cycling genes). We then calculated their difference as a study-specific organic resource–microbial activation imbalance index (ORMAI), for which positive values indicate high functional potential relative to retained resources. Relative to the forest, SOC declined by 29.5% under agroforestry and 61.1% under paddy cultivation, while total N declined by 16.6% and 59.7%, respectively. Paddy soil also contained the lowest concentrations of all hydrolyzable organic N fractions, with reductions of 28.3–81.5% relative to the forest. Despite this depleted resource status, paddy soil exhibited the highest activities of all measured C-, N-, and phosphorus-acquiring enzymes and the greatest abundances of AOA-amoA, AOB-amoA, nirK, nirS, and nosZ. It also had a higher (nirK + nirS)/nosZ ratio than the other soils. Paddy soil produced 3.91 and 4.69 times as much cumulative N₂O as forest and agroforestry soils, respectively. The resulting ORMAI was strongly positive in paddy soil (2.34) but negative in the forest and agroforestry soils (approximately −1.17). The three systems therefore exhibited distinct configurations: high resource status with intermediate functional potential in the forest, partial resource retention with restrained functional activation under agroforestry, and depleted resource status coupled with high functional and emission potentials in paddy soil. These findings identify resource–function imbalance as an informative dimension of land-use legacy, revealing that soil organic-resource depletion can coincide with enhanced microbial functional and N₂O emission potentials rather than constraining them. These results highlight the importance of considering microbial functional potential alongside soil organic-resource status when assessing potential N₂O emission risks and developing land-management strategies following tropical forest conversion.

1. Introduction

Soil carbon (C) and nitrogen (N) concentrations are standing pools rather than direct measures of biogeochemical activity. Their sizes reflect the cumulative balance among organic inputs, microbial transformation, physicochemical stabilization, plant uptake, and hydrological or gaseous losses [1,2]. A large soil organic C (SOC) or total N pool may therefore coexist with slow short-term turnover when a substantial proportion of the retained material is physically protected, mineral-associated, chemically resistant, or otherwise inaccessible to microorganisms [3]. Conversely, relatively small organic pools may support substantial microbial processing when environmental conditions increase substrate accessibility or favor rapid nutrient transformation [4]. Distinguishing the amount of resource retained from the microbial capacity to process that resource is consequently essential for understanding how land-use change affects soil functioning. Conversion of tropical forests to managed agricultural systems can alter both components of this relationship. Removal of forest vegetation modifies litter inputs, root turnover, and rhizosphere processes, whereas subsequent cultivation changes soil disturbance, aeration, hydrology, acidity, nutrient inputs, and the physical protection of organic matter [5,6]. Forest conversion frequently reduces SOC and total N, but its effects on nutrient availability, microbial biomass, extracellular enzyme activity, and N-transformation pathways vary among management systems and with time since conversion [6,7]. The functional consequences of a given change in SOC or total N therefore depend on the environmental filters established by the subsequent land use and on how these filters regulate microbial access to the resources that remain [8]. Forest conversion should thus be examined not only in terms of changes in individual soil pools, but also in terms of how the resulting land-use legacy reorganizes the relationship between resource status and microbial functional potential.

This distinction is particularly relevant when a common forest type is converted to agricultural systems with contrasting vegetation, organic inputs, hydrological regimes, and management histories [9,10]. Agroforestry can maintain perennial vegetation and pathways of litter, root, and manure input, potentially preserving some features of forest nutrient cycling [11,12]. However, these effects are not inherent to all agroforestry systems and depend on system design, vegetation composition, animal integration, management intensity, and time since establishment [13,14]. Paddy cultivation establishes a markedly different soil environment through puddling, fertilization, residue management, and repeated transitions between flooded and drained conditions [9]. These practices modify redox dynamics, soil structure, nutrient availability, microbial habitats, and the relative importance of aerobic and anaerobic transformations [15]. Agroforestry and paddy cultivation may therefore produce distinct soil biogeochemical configurations even when both originate from the same forest type. Their comparison can reveal contrasting land-use legacies; however, land-use type, management regime, and time since conversion vary simultaneously, and their individual effects therefore cannot be separated.

Total N provides limited information about the chemical organization and potential accessibility of soil organic N [16]. Soil organic N comprises compounds that differ in their biological origin, molecular structure, association with minerals, and susceptibility to microbial depolymerization [17]. Acid hydrolysis operationally separates part of this pool into hydrolyzable ammonium N (HAN), amino sugar N (ASN), amino acid N (AAN), and hydrolyzable unidentified N (HUN) [18,19]. These fractions are not direct measurements of immediately available N; instead, they characterize differences in the composition of the hydrolyzable organic N pool and its potential contribution to longer-term N supply [20]. Their measurement can therefore indicate whether contrasting land-use legacies are associated with broad changes in organic N status or with disproportionate responses among fractions linked to amino compounds, microbial residues, and unidentified hydrolyzable materials [21]. The functional significance of these organic pools depends partly on microbial access to them. Decomposition of complex organic substrates requires extracellular depolymerization before their products can be assimilated by microbial cells [17]. β-1,4-glucosidase contributes to microbial C acquisition, β-1,4-N-acetylglucosaminidase and leucine aminopeptidase contribute to organic N acquisition, and acid phosphatase contributes to organic phosphorus (P) acquisition [22,23]. Their potential activities reflect microbial investment in resource acquisition, but they are also influenced by microbial biomass and physiological state, nutrient demand, enzyme stabilization, substrate accessibility, and assay conditions [24]. They should therefore be interpreted as indicators of potential resource-acquisition capacity rather than direct measurements of in situ decomposition. Importantly, enzyme activity need not vary proportionally with the size of the corresponding organic-resource pool. High activity may accompany abundant accessible substrates, but it may also reflect intensified microbial investment in acquiring resources that have become scarce or difficult to access [25]. Conversely, soils containing large organic pools may exhibit comparatively low potential activity when those resources are physically protected, chemically resistant, or insufficient to induce enzyme production [26]. Land-use conversion may therefore alter not only organic-resource status and enzyme activities individually but also the degree of correspondence between them. Evaluating both dimensions is necessary to test whether contrasting land-use legacies preserve resource–function alignment or generate a mismatch between retained organic resources and microbial resource-acquisition potential.

The balance between retained organic resources and microbial functional potential may be particularly consequential for soil N₂O emissions. Forest conversion can reduce SOC and organic N pools while simultaneously altering pH, aeration, redox dynamics, mineral N availability, and nutrient inputs—all of which regulate the microbial transformation of the resources that remain [27]. Consequently, soils with depleted organic-resource pools may not exhibit proportionally lower N-cycling activity. Instead, management legacies may increase substrate turnover, microbial resource-acquisition investment, or the abundance of N-transforming functional groups, thereby sustaining or enhancing N₂O production potential despite lower SOC and organic N concentrations [28]. This possibility is especially relevant to paddy soils, where long-term fertilization and repeated flooding–drainage cycles can create alternating conditions favorable to nitrification, denitrification, and the coupling of their products [9]. Net N₂O release then emerges from the interaction between available substrates, microbial acquisition capacity, the size and relative organization of nitrifying and denitrifying guilds, and the environmental regulation of N₂O production and reduction [29,30]. Measurements of AOA-amoA, AOB-amoA, nirK, nirS, and nosZ, together with their abundance ratios, can characterize relevant components of this functional organization [31,32], while potential enzyme activities indicate microbial investment in accessing C, N, and P resources. CO₂ release provides a complementary expression of overall microbial C processing under the same conditions [33]. Integrating these measurements therefore makes it possible to test whether differences in potential N₂O emissions primarily coincide with the amount of organic resources retained in soil or with a broader functional configuration that remains highly active relative to those resources.

Despite substantial progress in documenting land-use effects on soil organic matter, microbial biomass, extracellular enzymes, N-cycling communities, and greenhouse gas emissions [7,27], it remains unclear whether these responses remain functionally aligned after forest conversion. These measurements occupy different positions along a resource pool functional capacity–process response continuum: organic C and N fractions characterize retained-resource status, extracellular enzymes represent potential resource-acquisition capacity, functional-gene abundances describe selected components of microbial metabolic potential, and gas emissions integrate the net expression of production and consumption processes under the measurement conditions. Changes at one level cannot therefore be assumed to propagate proportionally to the next [34]. Nevertheless, these dimensions are commonly interpreted separately or related through pairwise correlations and ordination, approaches that reveal differences and covariation but do not directly quantify whether microbial functional potential is high or low relative to the organic resources retained in soil [35]. The central unresolved issue is therefore whether contrasting land-use legacies preserve correspondence between organic-resource status and microbial functional potential or shift these dimensions disproportionately, and whether such shifts are reflected in potential gaseous emissions.

To examine this issue, we used a domain-based framework that represented organic-resource status and microbial functional potential independently before comparing their relative configuration. SOC, total N, microbial biomass C (MBC), and hydrolyzable organic N fractions were integrated into a study-specific organic resource retention index (ORRI), whereas extracellular enzyme activities and N-cycling functional-gene abundances were integrated into a microbial functional activation index (MFAI). Their difference was expressed as an organic resource–microbial activation imbalance index (ORMAI), with positive values indicating greater functional potential relative to organic-resource status and negative values indicating the converse. These indices were designed as complementary summaries of the measured domains, not as universal measures of soil quality or substitutes for their constituent variables. Importantly, N₂O and CO₂ emissions were excluded from index construction, allowing their associations with the resulting resource–function configuration to be evaluated without mathematical circularity.

We applied this framework to soils under reference natural secondary forest, six-year poultry-integrated agroforestry, and approximately 60-year paddy cultivation. Both managed systems originated from the same regional forest type but differed in vegetation, management, hydrological regime, and conversion history. This comparison is particularly relevant to Hainan Province, where warm and humid tropical conditions favor rapid organic-matter turnover and nutrient transformation, potentially increasing the sensitivity of soil C and N cycling to land-use change [36,37]. In the present study, forest conversion in Wenchang was represented by two contrasting land-use systems: poultry-integrated agroforestry and long-established paddy cultivation, which differed markedly in vegetation, organic inputs, disturbance, fertilization, and hydrological regimes. Establishing whether these pathways preserve soil organic-resource status or generate microbial functional and N₂O-production potentials that are disproportionate to the remaining resource base is important for evaluating the regional biogeochemical consequences of forest conversion. Such evidence can also help distinguish land-management trajectories that better reconcile agricultural use with the conservation of soil resources and the limitation of potential gaseous N losses in tropical agricultural landscapes. We characterized soil physicochemical properties, microbial biomass, hydrolyzable organic N fractions, potential extracellular enzyme activities, and N-cycling functional-gene abundances and evaluated potential N₂O and CO₂ production under common incubation conditions. AOA-amoA/AOB-amoA and (nirK + nirS)/nosZ abundance balances were additionally calculated to characterize the relative organization of ammonia-oxidizing and denitrification-related groups. This design enabled us to determine whether contrasting land-use legacies were associated only with changes in individual properties or with a coordinated reorganization of organic-resource status, microbial functional potential, and potential gaseous responses.

We addressed three questions: (1) how do the three land-use systems differ in organic-resource status and microbial functional potential; (2) do the relative abundance balances of ammonia-oxidizing and denitrification-related groups differ among the systems; and (3) are potential N₂O and CO₂ emissions associated more closely with individual resource pools or with the relative balance between organic-resource status and microbial functional potential? We hypothesized that (H1) both managed systems would have lower organic-resource status than the reference forest, with the largest reduction under long-term paddy cultivation; (H2) the management and hydrological legacy of paddy cultivation would be associated with greater potential enzyme activities, larger N-cycling functional groups, an increased N₂O production-to-reduction gene-abundance balance, and greater potential N₂O and CO₂ production; and (H3) potential N₂O production would be associated more strongly with microbial functional potential and a positive resource–function imbalance than with SOC, total N, or individual hydrolyzable organic N fractions alone. By testing these hypotheses, we evaluated whether forest conversion changes not only the quantities of retained resources and microbial functional attributes but also the proportional relationship between them.

2. Materials and Methods

2.1. Study Area and Land-Use History

The study was conducted in Wenchang City, Hainan Province, southern China. Based on the available field records, the sampling sites were located within an approximate geographical range of 19°39′–19°45′ N, 110°40′–110°44′ E. The region has a tropical monsoon climate, with a mean annual temperature of approximately 25.0 °C, mean annual precipitation of approximately 1500 mm, and mean annual evaporation of approximately 500 mm. The study sites were selected within the same regional climatic and soil-forming environment to minimize variation unrelated to land use. The three land-use systems shared a comparable regional soil-forming background. All sampling sites occurred within the same soil mapping unit (SMU 11774), classified as Ferralic Cambisol with a loam surface texture according to the Harmonized World Soil Database v2.0. The study area was also associated with Quaternary unconsolidated sedimentary parent material according to the regional soil environmental background classification (DB46/T 687–2025) [38]. Thus, the selected sites were comparable in mapped soil type, texture, and parent-material setting.

Three land-use systems were selected in September 2025: natural secondary forest, poultry-integrated agroforestry, and paddy field. The natural secondary forest represented the prevailing reference forest condition in the study region and was used as the reference land-use system. It was dominated by native tropical broad-leaved tree species and had experienced relatively limited recent management disturbance. Both the agroforestry and paddy systems were established through conversion of the same regional natural secondary forest type represented by the reference sites. The poultry-integrated agroforestry system had been established approximately 6 years before sampling through conversion of the same natural forest type. It consisted primarily of perennial cash crops, including areca nut, citrus, and coconut, integrated with free-ranging poultry such as chickens, ducks, and geese. The system received organic inputs from litter deposition, root turnover, crop residues, and poultry excreta. The paddy fields had also been established through conversion of the same natural forest type and had been cultivated for 60 years. They represented a long-term, intensively managed agricultural system characterized by rice cultivation, fertilization, and periodic flooding and drainage. Soil sampling was conducted approximately two weeks after rice harvest. The paddy fields had remained without managed surface flooding for approximately four to five weeks before sampling, including the pre-harvest drainage period. At sampling, the fields contained neither a standing crop nor surface water, although the soil remained moist under the prevailing humid tropical conditions.

Within each land-use system, the three replicate sites had comparable land-use histories, vegetation or cropping characteristics, and management durations. The selected systems therefore represented two contrasting forest-conversion pathways: relatively recent conversion to poultry-integrated agroforestry and long-term conversion to paddy cultivation. Because land-use type, management regime, and time since conversion varied concurrently, the comparison was designed to evaluate the integrated soil legacies associated with the two conversion pathways rather than to isolate the effect of conversion duration alone.

2.2. Soil Sampling and Sample Preparation

Each land-use system was represented by three spatially independent replicate sites, giving a total of nine independent field replicates. Replicate sites were separated by more than 3 km to ensure spatial independence. All samples were collected on the same day and using an identical sampling procedure. At each replicate site, the surface litter layer and loose aboveground residues were carefully removed before sampling. Five soil cores, each 5 cm in diameter, were collected from the 0–20 cm mineral soil layer using a five-point sampling method across an approximately 20 m × 20 m area, with one point near the center and four points toward the surrounding corners. The five cores collected within each site were thoroughly homogenized to form one composite sample representing that independent field replicate. The exact minimum distance between individual soil cores was not recorded, but the five-point layout ensured that cores were spatially distributed across the replicate site to reduce microsite bias. Composite samples from different replicate sites were retained and processed separately throughout the study. Thus, each land-use system comprised three independent field-level composite samples (n = 3). The nine composite samples were placed individually in sterile polyethylene bags and transported to the laboratory under cooled conditions. Visible roots, stones, and recognizable plant and animal residues were removed, after which the samples were gently homogenized and passed through a 2 mm sieve.

Each field replicate was then divided into three portions. One portion was air-dried for the determination of soil particle-size distribution, basic physicochemical properties, and organic N fractions. A second portion was maintained fresh at 4 °C for the determination of inorganic N, microbial biomass, extracellular enzyme activities, and laboratory incubation. The third portion was stored at −80 °C until DNA extraction and quantitative PCR analysis. Samples from the nine independent field replicates were kept separate throughout sample preparation, laboratory measurements, incubation, and statistical analysis.

2.3. Determination of Soil Physicochemical Properties

The soil pH was determined by the potentiometric method with a soil-to-water ratio of 1:2.5. Soil NH₄⁺-N and NO₃⁻-N were extracted with 2 M KCl solution at a ratio of 1:5 (soil:water) and determined by spectrophotometer (Shanghai Metash Instruments Co., Ltd., Shanghai, China). Soil organic carbon (SOC) was determined by the potassium dichromate-sulfuric acid oxidation method, and total N was determined using the semi-micro Kjeldahl method. Total P was measured by the molybdenum-antimony colorimetric method after digestion. Soil MBC, microbial biomass N (MBN), and microbial biomass P (MBP) were measured using the chloroform fumigation-extraction method and calculated using the corresponding conversion factors. Soil acid-hydrolyzable N fractions, including AHN, HAN, ASN, and AAN, were determined following established organic N fractionation procedures [20]. HUN was calculated as the difference between AHN and the sum of HAN, ASN, and AAN.

2.4. Determination of N₂O and CO₂ Emissions

A 32-day laboratory incubation experiment was conducted to evaluate potential N₂O and CO₂ emissions from soils under different land-use types. For each land-use type, three incubation replicates were prepared from the corresponding homogenized composite soil sample, and blank controls were established to exclude system errors. Three treatment groups were established: natural secondary forest soil, poultry-integrated agroforestry soil, and paddy soil. For each incubation replicate, a fresh-soil subsample equivalent to 20 g of oven-dry soil was placed into each bottle, and distilled water was added to adjust soil moisture to 40% water-holding capacity. All serum bottles were covered with perforated plastic wrap to maintain aerobic conditions and pre-incubated in the dark at 25 °C for 7 days to restore and stabilize soil microbial activity. After pre-incubation, urea solution was added to each bottle to achieve an exogenous N addition rate of 150 mg N kg⁻¹ dry soil, and soil moisture was adjusted again to 60% water-holding capacity. Urea was added at 150 mg N kg⁻¹ dry soil to provide a common N-substrate challenge across the three land-use systems, following [39]. This standardized addition was intended to reduce variation arising from differences in initial mineral-N supply and to permit comparison of the potential N₂O response associated with the properties retained under each land-use legacy. The dose was not intended to reproduce a specific field fertilizer application rate; therefore, the resulting emissions were interpreted as comparative potentials under the imposed incubation conditions rather than as estimates of field-scale fertilizer-induced emissions. The serum bottles were then incubated at 25 °C. During incubation, the bottles were weighed regularly, and deionized water was added to maintain relatively stable soil moisture.

Gas samples were collected on incubation days 1, 2, 4, 7, 11, 15, 23, and 32. Cumulative N₂O and CO₂ emissions over the 32-day incubation period were calculated separately for each replicate using trapezoidal integration, with the Day 1 flux used to represent the Day 0–1 interval; the detailed equation is provided in the note to Table S3. Before each gas collection, the bottles were sealed for a fixed enclosure period to allow headspace gas accumulation, after which sterilized 25 mL glass syringes were used for gas sampling. Before sampling, residual gas in the syringe was expelled, the serum bottle septum was disinfected, and the needle was inserted vertically through the septum. A 15 mL gas sample was withdrawn from the bottle, and the syringe plunger was pushed and pulled 5–6 times to ensure thorough mixing of the headspace gas. After gas collection, the bottle cap was opened, and the bottle was again covered with perforated plastic wrap to maintain gas exchange during incubation. N₂O concentrations were measured using a Shimadzu GC-2014 gas chromatograph. The detection conditions were as follows: a mixed gas of 95% argon and 5% methane was used as the carrier gas, high-purity N₂ was used as the backflush gas, the gas flow rate was 25 mL min⁻¹, the column temperature was 50 °C, and the detector temperature was 300 °C. CO₂ concentrations were measured simultaneously from the same gas samples and used to calculate CO₂ emission dynamics during incubation. Gas emission rates were calculated from the increase in headspace gas concentration during the enclosure period and standardized by dry soil mass. Cumulative emissions were estimated by linear interpolation between adjacent sampling dates and summing emissions across the 32-day incubation. Thus, the reported gas data represent incubation-derived potential emissions rather than direct field fluxes.

2.5. Measurement of Soil Extracellular Enzyme Activities

Soil extracellular enzyme activities were measured using a microplate fluorometric method following established fluorometric enzyme assay procedures [40], including β-glucosidase, N-acetyl-β-D-glucosaminidase, leucine aminopeptidase, and acid phosphatase. Fresh soil (1.0000 g) was placed into a 250 mL conical flask, mixed with 125 mL acetate buffer (pH 5.0), and magnetically stirred for 5 min to prepare a soil suspension. Then, 200 μL of the soil suspension was added to a 96-well microplate, followed by 50 μL of the corresponding fluorogenic substrate solution. The substrates for acid phosphatase, β-glucosidase, N-acetyl-β-D-glucosaminidase, and leucine aminopeptidase were 4-MUB-phosphate, 4-MUB-β-D-glucopyranoside, 4-MUB-N-acetyl-β-D-glucosaminide, and L-leucine-7-amido-4-methylcoumarin, respectively. After incubation in the dark at 25 °C for 4 h, 10 μL of 1 mol L⁻¹ NaOH was added to terminate the reaction, and fluorescence intensity was measured at an excitation wavelength of 365 nm and an emission wavelength of 450 nm. A standard curve was constructed using 4-MUB standard solution to calculate acid phosphatase, β-glucosidase, and N-acetyl-β-D-glucosaminidase activities; leucine aminopeptidase activity was converted based on the corresponding AMC fluorescent product. Enzyme activities were expressed as nmol g⁻¹ h⁻¹.

2.6. DNA Extraction and Quantitative PCR Analysis

Total DNA was extracted from 0.5 g soil samples using the Soil Rapid DNA^® Spin Kit (MP Biomedicals, Solon, OH, USA) according to the manufacturer’s instructions. The extracted DNA was stored at −20 °C until qPCR analysis. DNA concentration and purity were determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA). Absolute quantitative PCR was performed using the SYBR Green method on an ABI 7500 real-time PCR system (Applied Biosystems, Foster City, CA, USA) to quantify the abundances of the archaeal ammonia monooxygenase gene (AOA-amoA), bacterial ammonia monooxygenase gene (AOB-amoA), nitrite reductase genes (nirK and nirS), and nitrous oxide reductase gene (nosZ). These genes were used as functional markers for ammonia oxidation, nitrite reduction, and N₂O reduction [41]. Primer sequences and amplicon sizes for each target gene are provided in Table S1.

The qPCR reaction mixture had a total volume of 25 μL, containing 12.5 μL Bestar SybrGreen qPCR MasterMix (DBI Bioscience, Shanghai, China), 0.25 μL forward primer (10 μM), 0.25 μL reverse primer (10 μM), 1 μL DNA template, and 11 μL sterile ddH₂O. The qPCR amplification program was as follows: initial denaturation at 95 °C for 2 min, followed by 45 cycles of 95 °C for 10 s and 60 °C for 1 min, followed by melting curve analysis. Plasmid DNA containing representative cloned fragments of the target genes was used as the qPCR standard, and standard templates were serially diluted by 10-fold gradients across seven orders of magnitude to generate standard curves. Product specificity was verified by melting curve analysis, and no-template controls were included by replacing DNA templates with sterile water. Gene copy numbers in the qPCR template were converted to copies g⁻¹ dry soil using the template copy concentration, dilution factor, DNA elution volume, and dry soil mass used for DNA extraction. Functional gene abundances were log₁₀-transformed before statistical analysis.

2.7. Derived Functional-Gene Balance and Resource–Activation Indices

To characterize relative functional balances within the soil N-cycling community, two gene-abundance indices were calculated separately for each field replicate using the original, untransformed gene-copy abundances. The ammonia-oxidizer balance was expressed as [42]:

$${\mathrm{l}\mathrm{o}\mathrm{g}}_{10}(\mathrm{A}\mathrm{O}\mathrm{A}/\mathrm{A}\mathrm{O}\mathrm{B})={\mathrm{l}\mathrm{o}\mathrm{g}}_{10}(\mathrm{A}\mathrm{O}\mathrm{A}-amoA)-{\mathrm{l}\mathrm{o}\mathrm{g}}_{10}(\mathrm{A}\mathrm{O}\mathrm{B}-amoA)$$

Higher values indicate greater relative dominance of archaeal over bacterial ammonia oxidizers. A genetic N₂O production-to-reduction balance was calculated as [43]:

$${\mathrm{l}\mathrm{o}\mathrm{g}}_{10}\left[{\displaystyle \frac{nirK+nirS}{nosZ}}\right]$$

where the original nirK and nirS copy abundances were summed before division by the original nosZ abundance and subsequent log₁₀ transformation. Higher values indicate a greater relative abundance of genes associated with nitrite reduction, a potential N₂O-producing step, relative to the gene associated with N₂O reduction. These indices represent relative gene-abundance balances and not direct measurements of nitrification, denitrification, or net N₂O production rates.

To quantify the balance between soil organic-resource retention and microbial functional activation, three study-specific composite indices were subsequently constructed using replicate-level data [44,45]. Cumulative N₂O and CO₂ emissions were excluded from index construction and retained as independent response variables. Before index calculation, each constituent variable was standardized across the nine samples as [45]:

$${\mathrm{Z}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{{\mathrm{x}}_{\mathrm{i}\mathrm{j}}-{\overline{\mathrm{x}}}_{\mathrm{j}}}{{\mathrm{s}}_{\mathrm{j}}}}$$

where ${\mathrm{x}}_{\mathrm{i}\mathrm{j}}$ is the value of variable $\mathrm{j}$ in sample $\mathrm{i}$, and ${\overline{\mathrm{x}}}_{\mathrm{j}}$ and ${\mathrm{s}}_{\mathrm{j}}$ are the overall mean and sample standard deviation of variable $\mathrm{j}$, respectively. Functional gene abundances were log₁₀-transformed before standardization. The organic resource retention index (ORRI) incorporated two equally weighted domains [45]. The bulk organic-resource domain was calculated as:

$${\mathrm{B}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{Z}}_{\mathrm{S}\mathrm{O}\mathrm{C},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{T}\mathrm{N},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{M}\mathrm{B}\mathrm{C},\mathrm{i}}}{3}}$$

where TN is total N.

The organic N-resource domain was calculated as:

$$\mathrm{O}{\mathrm{N}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{Z}}_{\mathrm{H}\mathrm{A}\mathrm{N},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{A}\mathrm{S}\mathrm{N},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{A}\mathrm{A}\mathrm{N},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{H}\mathrm{U}\mathrm{N},\mathrm{i}}}{4}}$$

ORRI was then calculated as:

$$\mathrm{O}\mathrm{R}\mathrm{R}{\mathrm{I}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{B}}_{\mathrm{i}}+\mathrm{O}{\mathrm{N}}_{\mathrm{i}}}{2}}$$

Higher ORRI values therefore indicate greater retention of bulk soil organic resources and hydrolyzable organic N fractions. The microbial functional activation index (MFAI) was also constructed from two equally weighted domains [45]. The extracellular-enzyme domain was calculated as:

$${\mathrm{E}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{Z}}_{\mathrm{A}\mathrm{P},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{B}\mathrm{G},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{N}\mathrm{A}\mathrm{G},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{L}\mathrm{A}\mathrm{P},\mathrm{i}}}{4}}$$

AP, BG, NAG, and LAP are acid phosphatase, β-glucosidase, N-acetyl-β-D-glucosaminidase, and leucine aminopeptidase, respectively. The N-cycling functional-gene domain was calculated as:

$${\mathrm{G}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{Z}}_{\mathrm{A}\mathrm{O}\mathrm{A},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{A}\mathrm{O}\mathrm{B},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{n}\mathrm{i}\mathrm{r}\mathrm{K},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{n}\mathrm{i}\mathrm{r}\mathrm{S},\mathrm{i}}+{\mathrm{Z}}_{\mathrm{n}\mathrm{o}\mathrm{s}\mathrm{Z},\mathrm{i}}}{5}}$$

MFAI was calculated as:

$$\mathrm{M}\mathrm{F}\mathrm{A}{\mathrm{I}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{E}}_{\mathrm{i}}+{\mathrm{G}}_{\mathrm{i}}}{2}}$$

Higher MFAI values indicate greater combined potential extracellular enzyme activity and abundance of N-cycling functional groups. Finally, the organic resource–microbial activation imbalance index (ORMAI) was calculated as:

$$\mathrm{O}\mathrm{R}\mathrm{M}\mathrm{A}{\mathrm{I}}_{\mathrm{i}}=\mathrm{M}\mathrm{F}\mathrm{A}{\mathrm{I}}_{\mathrm{i}}-\mathrm{O}\mathrm{R}\mathrm{R}{\mathrm{I}}_{\mathrm{i}}$$

Positive ORMAI values indicate that microbial functional activation was high relative to organic-resource retention, whereas negative values indicate comparatively stronger organic-resource retention than functional activation. Equal weighting of the constituent domains prevented groups represented by a larger number of variables from disproportionately influencing the composite indices [45].

2.8. Statistical Analysis

Experimental data were organized using Excel 2019. Analysis of variance (ANOVA) combined with Tukey’s HSD test was used to analyze intergroup differences in soil physicochemical properties, mineral N, organic N fractions, microbial biomass, extracellular enzyme activities, N-cycling functional gene abundances, and cumulative N₂O and CO₂ emissions among land-use types (p < 0.05). The temporal dynamics of N₂O and CO₂ emissions were analyzed using repeated-measures ANOVA. Pearson’s correlation analysis was employed to assess the associations between greenhouse gas emissions and soil biochemical properties, extracellular enzyme activities, and N-cycling functional genes (p < 0.05). Differences in the gene-abundance balance indices, ORRI, MFAI, and ORMAI among land-use systems were assessed using one-way ANOVA followed by Tukey’s HSD test. Pearson correlations were used to evaluate their associations with cumulative N₂O and CO₂ emissions. The robustness of ORMAI was evaluated by sequentially excluding each constituent variable and recalculating the index. An alternative calculation in which all individual variables received equal weight was also performed to assess sensitivity to the domain-weighting procedure. PERMANOVA based on Euclidean distances was used to evaluate differences in the z-score-standardized soil-resource and microbial functional profiles among land-use systems, and PERMDISP was used to test homogeneity of multivariate dispersion. Except for PERMANOVA and PERMDISP, statistical analyses and correlation analyses were performed using IBM SPSS Version 24, and corresponding charts were drawn using Origin 2021. PERMANOVA and PERMDISP were performed using R 4.4.1 with the vegan package (version 2.7-5).

3. Results

3.1. Land-Use Conversion Reorganized Soil C–N Status and Microbial Nutrient Pools

Soil properties, inorganic N availability, and microbial nutrient pools differed substantially among the three land-use systems (

Table 1. Soil physicochemical properties, inorganic nitrogen concentrations, and microbial biomass under reference forest, poultry-integrated agroforestry, and paddy land-use systems.

Parameter	Forest	Poultry-Integrated Agroforestry	Paddy	p-Value
pH	4.74 ± 0.04 b	4.84 ± 0.04 b	6.14 ± 0.06 a	<0.001
NO3−-N (mg kg−1)	13.99 ± 0.42 b	17.87 ± 0.16 a	9.09 ± 0.20 c	<0.001
NH4+-N (mg kg−1)	6.63 ± 0.25 b	7.36 ± 0.19 a	7.93 ± 0.49 a	0.009
NO3−-N/NH4+-N	2.11 ± 0.14 b	2.43 ± 0.08 a	1.15 ± 0.08 c	<0.001
SOC (g C kg−1)	16.5 ± 0.04 a	11.7 ± 0.21 b	6.43 ± 0.20 c	<0.001
Total N (g N kg−1)	1.45 ± 0.128 a	1.21 ± 0.038 b	0.58 ± 0.025 c	<0.001
C/N	11.5 ± 0.96 a	9.67 ± 0.48 a	11.0 ± 0.65 a	0.0504
Total P (g kg−1)	0.25 ± 0.005 c	0.46 ± 0.012 b	0.69 ± 0.008 a	<0.001
MBC (mg kg−1)	337 ± 4.80 a	328 ± 3.81 a	312 ± 6.78 b	0.003
MBN (mg kg−1)	4.32 ± 0.27 c	11.8 ± 0.91 a	10.2 ± 0.17 b	<0.001
MBP (mg kg−1)	6.46 ± 0.35 c	26.1 ± 0.17 b	41.8 ± 0.77 a	<0.001

Values are means ± SD of three independent field replicates (n = 3). Within each row, different lowercase letters indicate significant differences among land-use systems at p < 0.05. The p-value column reports the omnibus one-way ANOVA result for each parameter. SOC, soil organic carbon; N, nitrogen; P, phosphorus; MBC, microbial biomass carbon; MBN, microbial biomass nitrogen; MBP, microbial biomass phosphorus.

). Complete one-way ANOVA and Tukey’s HSD results for all measured variables are provided in Table S4. Land use significantly affected soil pH, SOC, total N, and total P (all p < 0.001). Soil pH was significantly higher under paddy cultivation than under forest and agroforestry systems, increasing from 4.74 and 4.84 in forest and agroforestry soils, respectively, to 6.14 in paddy soil. Forest and agroforestry soils were both strongly acidic and did not differ significantly in pH. In contrast, SOC and total N concentrations declined significantly and progressively following forest conversion. Relative to the reference forest, SOC declined by 29.5% under agroforestry and 61.1% under paddy cultivation, while total N declined by 16.6% and 59.7%, respectively. Soil total P showed the opposite pattern, increasing significantly from 0.251 g kg⁻¹ in the forest to 0.46 g kg⁻¹ under agroforestry and 0.69 g kg⁻¹ under paddy cultivation.

Land use also significantly affected NO₃⁻-N (p < 0.001), NH₄⁺-N (p = 0.009), and the NO₃⁻-N/NH₄⁺-N ratio (p < 0.001). Agroforestry soil had the highest NO₃^--N concentration (17.9 mg kg⁻¹), exceeding the forest and paddy soils by 27.7% and 96.5%, respectively, with significant differences among all three systems. NH₄⁺-N concentrations were significantly higher in the agroforestry and paddy soils than in the forest soil, but did not differ significantly between the two managed systems. Consequently, the NO₃⁻-N/NH₄⁺-N ratio was significantly highest under agroforestry (2.43), intermediate in the forest (2.11), and lowest in the paddy soil (1.15). Microbial biomass components showed contrasting element-specific responses. Land use significantly affected MBC (p = 0.003), MBN (p < 0.001), and MBP (p < 0.001). MBC did not differ significantly between the forest and agroforestry soils but was significantly lower in paddy soil, representing a 7.4% reduction relative to the forest. MBN was significantly highest under agroforestry (11.8 mg kg⁻¹), followed by paddy soil (10.2 mg kg⁻¹) and forest soil (4.32 mg kg⁻¹). MBP increased significantly from 6.46 mg kg⁻¹ in the forest to 26.14 mg kg⁻¹ under agroforestry and 41.8 mg kg⁻¹ under paddy cultivation, with significant differences among all three systems. The soil C/N ratio did not differ significantly among the three land-use systems (p = 0.0504).

3.2. Land-Use Differences in Organic N Fractions

Land use significantly affected all measured hydrolyzable organic N fractions (all p < 0.001; Figure 1). HAN decreased significantly from 329 mg N kg⁻¹ in the reference forest to 210 mg N kg⁻¹ under agroforestry and 149 mg N kg⁻¹ in paddy soil, corresponding to reductions of 36.3% and 54.7%, respectively, relative to the forest (Figure 1a). HUN showed an even stronger decline, decreasing from 536 mg N kg⁻¹ in the forest to 386 mg N kg⁻¹ under agroforestry and 98.9 mg N kg⁻¹ in paddy soil (Figure 1d). Thus, relative to the reference forest, HUN was reduced by 27.9% under agroforestry and by 81.5% under paddy cultivation. Both HAN and HUN differed significantly among all three systems.

ASN and AAN showed a different response pattern. Their concentrations did not differ significantly between the reference forest and agroforestry soils, but were significantly lower in the paddy soil. ASN declined from 208 and 197 mg N kg⁻¹ in the forest and agroforestry soils, respectively, to 64.4 mg N kg⁻¹ in the paddy soil, representing a 69.1% reduction relative to the forest (Figure 1b). Similarly, AAN declined from 280 mg N kg⁻¹ in the forest and 289 mg N kg⁻¹ under agroforestry to 201 mg N kg⁻¹ in paddy soil, equivalent to a 28.3% reduction relative to the forest (Figure 1c). Overall, paddy soil consistently contained the lowest concentrations of all measured hydrolyzable organic N fractions, with the strongest proportional reductions observed for HUN and ASN.

3.3. Temporal Dynamics and Cumulative N₂O and CO₂ Emissions

Land use, incubation time, and their interaction significantly affected the temporal dynamics of both N₂O and CO₂ emissions (all p < 0.001; Figure 2a,b; Table S2). N₂O emission rates from paddy soil increased during the first half of the incubation, reached a maximum of approximately 18.0 μg kg⁻¹ h⁻¹ on day 15, and subsequently declined to 2.17 μg kg⁻¹ h⁻¹ by day 32. In contrast, N₂O emission rates from the reference forest and agroforestry soils remained comparatively low throughout the incubation and did not exceed 4.20 μg kg⁻¹ h⁻¹. Emission rates of CO₂ peaked during the initial incubation period, reaching 6.48 mg kg⁻¹ h⁻¹ in paddy soil and approximately 5.0 mg kg⁻¹ h⁻¹ in both forest and agroforestry soils on day 2. Following this initial peak, CO₂ emission rates declined sharply, showed a smaller transient increase around day 7, and then gradually approached relatively stable levels. Paddy soil generally maintained the highest CO₂ emission rates during the later incubation period, whereas forest soil showed the lowest rates.

Cumulative N₂O emissions differed significantly among all three land-use systems (p < 0.001) and followed the order paddy > forest > agroforestry (Figure 2c). Paddy soil emitted 3.91 and 4.69 times as much cumulative N₂O as forest and agroforestry soils, respectively. Forest soil also emitted 20.1% more cumulative N₂O than agroforestry soil. Cumulative CO₂ emissions also differed significantly among the three systems (p < 0.001) and followed the order paddy > agroforestry > forest (Figure 2d). Relative to the forest, cumulative CO₂ emissions were 16.1% higher under agroforestry and 37.9% higher under paddy cultivation. Therefore, under the standardized temperature, moisture, and urea-addition conditions, paddy soil showed the greatest potential for both N₂O and CO₂ emissions.

3.4. Potential Extracellular Enzyme Activities Diverged Between Agroforestry and Paddy Soils

Land use significantly affected the potential activities of acid phosphatase, β-glucosidase, N-acetyl-β-D-glucosaminidase, and leucine aminopeptidase (all p < 0.001; Figure 3). All four enzymes showed a consistent response pattern, with the highest activities in paddy soil, intermediate activities in the reference forest, and the lowest activities in agroforestry soil. Acid phosphatase activity increased from 71.5 nmol g⁻¹ h⁻¹ in agroforestry soil to 85.3 nmol g⁻¹ h⁻¹ in forest soil and 92.1 nmol g⁻¹ h⁻¹ in paddy soil (Figure 3a). Similarly, β-glucosidase activity was 65.9, 82.7, and 89.1 nmol g⁻¹ h⁻¹ in agroforestry, forest, and paddy soils, respectively (Figure 3b). Significant pairwise differences were detected among all three land-use systems for both enzymes.

The two N-acquiring enzymes showed comparable responses. N-acetyl-β-D-glucosaminidase activity increased from 69.2 nmol g⁻¹ h⁻¹ under agroforestry to 76.6 nmol g⁻¹ h⁻¹ in forest soil and 88.7 nmol g⁻¹ h⁻¹ in paddy soil (Figure 3c). Leucine aminopeptidase activity similarly increased from 64.3 nmol g⁻¹ h⁻¹ under agroforestry to 77.1 nmol g⁻¹ h⁻¹ in the forest and 88.0 nmol g⁻¹ h⁻¹ in paddy soil (Figure 3d). Relative to agroforestry, paddy soil showed 28.7%, 35.3%, 28.1%, and 36.8% higher acid phosphatase, β-glucosidase, N-acetyl-β-D-glucosaminidase, and leucine aminopeptidase activities, respectively. Compared with the reference forest, the corresponding increases in paddy soil ranged from 7.80% for β-glucosidase to 15.8% for N-acetyl-β-D-glucosaminidase. Thus, the two managed systems diverged consistently in their potential extracellular enzyme activities, with suppression under agroforestry and enhancement under paddy cultivation relative to the reference forest.

3.5. Paddy Soil Supported Greater Abundances of N-Cycling Functional Genes

The abundances of all measured N-cycling functional genes differed significantly among the three land-use systems (p < 0.001; Figure 4). Paddy soil had the highest abundance of each functional gene. The abundance of AOA-amoA followed the order paddy (8.40 ± 0.05) > agroforestry (8.05 ± 0.02) > reference forest (7.89 ± 0.05 log₁₀ copies g⁻¹ dry soil), with significant differences among all three systems. AOB-amoA showed a similar pattern, increasing from 6.00 ± 0.12 in the reference forest to 6.58 ± 0.16 in agroforestry and 7.45 ± 0.06 log₁₀ copies g⁻¹ dry soil in the paddy soil, with all pairwise differences being significant.

The denitrification-related genes showed gene-specific responses. The abundance of nirK followed the order paddy (5.94 ± 0.11) > reference forest (5.41 ± 0.06) > agroforestry (4.99 ± 0.11 log₁₀ copies g⁻¹ dry soil), with significant differences among all systems. Paddy soil also had a significantly higher nirS abundance (5.85 ± 0.06) than both the reference forest (5.48 ± 0.02) and agroforestry soils (5.37 ± 0.06 log₁₀ copies g⁻¹ dry soil), which did not differ significantly from each other. The abundance of nosZ followed the order paddy (7.43 ± 0.03) > reference forest (7.15 ± 0.05) > agroforestry (6.98 ± 0.04 log₁₀ copies g⁻¹ dry soil), with significant differences among all three systems. On the original abundance scale, paddy soil contained approximately 3.2-, 28.5-, 3.4-, 2.4-, and 1.9-fold greater abundances of AOA-amoA, AOB-amoA, nirK, nirS, and nosZ, respectively, than the reference forest. Thus, paddy soil supported larger nitrifier and denitrifier functional groups under the conditions examined; however, gene abundance indicates functional population size rather than the realized rates of the corresponding N-transformation processes.

The relative balances among N-cycling functional genes also differed significantly among the land-use systems (Figure 5; Table S6). The log₁₀-transformed AOA-amoA/AOB-amoA abundance ratio decreased progressively from the reference forest (1.90 ± 0.16) to agroforestry (1.46 ± 0.14) and paddy soil (0.95 ± 0.11), with significant differences among all three systems (overall ANOVA, F_2,6 = 36.0, p < 0.001; Figure 5a). The N₂O production-to-reduction gene-abundance balance, expressed as log₁₀((nirK + nirS)/nosZ), also differed among land-use systems (F_2,6 = 13.1, p = 0.007; Figure 5b). This index was significantly higher in paddy soil than in the reference forest and agroforestry soils, whereas the latter two did not differ significantly (p = 0.591). Across the nine replicates, the production-to-reduction gene-abundance balance was positively correlated with cumulative N₂O emissions (r = 0.89, p = 0.001; Figure 5c). Thus, despite the greater nosZ abundance in paddy soil, its potential N₂O production-related gene abundance increased more strongly relative to its N₂O reduction-related gene abundance.

3.6. Soil–Microbial Attributes Associated with Potential N₂O and CO₂ Emissions

PERMANOVA indicated significant differentiation among the land-use systems for both the soil physicochemical and resource-variable profile (pseudo-F = 45.0, R² = 0.94, p = 0.004) and the microbial enzyme and N-cycling functional-gene profile (pseudo-F = 70.5, R² = 0.96, p = 0.004; Table S7). PERMDISP detected no significant differences in multivariate dispersion among the land-use systems for either profile (p = 0.76 and 0.79, respectively), indicating that the observed PERMANOVA differentiation was not explained by unequal within-group dispersion.

Pearson correlation analysis revealed distinct soil–microbial associations with cumulative N₂O and CO₂ emissions (Figure 6). Cumulative N₂O emissions were positively correlated with cumulative CO₂ emissions (r = 0.88, p < 0.01), soil pH (r = 0.99, p < 0.001), NH₄⁺-N (r = 0.72, p < 0.05), total P (r = 0.86, p < 0.01), and MBP (r = 0.80, p < 0.01). They were also positively associated with the potential activities of all four extracellular enzymes, with correlation coefficients ranging from 0.73 for β-glucosidase to 0.91 for N-acetyl-β-D-glucosaminidase (p < 0.05), and with the abundances of all measured N-cycling functional genes (r = 0.89–0.97, all p < 0.01). Conversely, cumulative N₂O emissions were negatively correlated with NO₃^--N (r = −0.92), the NO₃⁻-N/NH₄⁺-N ratio (r = −0.97), SOC (r = −0.85), total N (r = −0.93), and MBC (r = −0.83; all p < 0.01). Negative correlations were also observed with all four organic N fractions, ranging from r = −0.72 for HAN to r = −0.99 for ASN (p < 0.05). In contrast, microbial biomass N was not significantly correlated with cumulative N₂O emissions. Cumulative CO₂ emissions displayed a partly similar but less uniform pattern. They were positively correlated with NH₄⁺-N (r = 0.84), total P (r = 0.99), pH (r = 0.92), MBP (r = 0.98), AOA-amoA (r = 0.979), AOB-amoA (r = 0.98), and nirS (r = 0.78; p < 0.05). Cumulative CO₂ emissions were negatively correlated with the NO₃⁻-N/NH₄⁺-N ratio, SOC, total N, MBC, and all four organic N fractions (r = −0.76 to −0.99, p < 0.05). However, their relationships with NO₃^--N, extracellular enzyme activities, nirK, and nosZ were not significant. The four organic N fractions were positively intercorrelated and generally positively associated with SOC and total N, but negatively associated with pH, extracellular enzyme activities, and N-cycling functional gene abundances.

3.7. Resource Retention and Microbial Functional Activation Became Imbalanced in Paddy Soil

The composite indices revealed contrasting resource-retention and microbial-activation configurations among the three land-use systems (Figure 7; Table S6). ORRI decreased significantly from the reference forest (0.95 ± 0.15) to agroforestry (0.25 ± 0.05) and paddy soil (−1.20 ± 0.08), with significant differences among all three systems (p < 0.001; Figure 7a). Conversely, MFAI was highest in paddy soil (1.14 ± 0.06), intermediate in the reference forest (−0.21 ± 0.06), and lowest in agroforestry soil (−0.93 ± 0.09), with all pairwise differences being significant (p < 0.001; Figure 7b). Consequently, ORMAI was markedly higher in paddy soil (2.34 ± 0.12) than in either the reference forest (−1.16 ± 0.19) or agroforestry soil (−1.18 ± 0.04), whereas the latter two systems did not differ significantly (p = 0.983; Figure 7c). Across the nine replicates, ORMAI was positively correlated with cumulative N₂O emissions (r = 0.996, p < 0.001; Figure 7d) and cumulative CO₂ emissions (r = 0.90, p < 0.001). The paddy-associated elevation of ORMAI remained unchanged when each constituent variable was sequentially excluded and when an alternative equal-variable-weighting procedure was applied (Table S6, Part D). These results identify a robust paddy-associated imbalance in which high microbial functional activation coincided with diminished organic-resource retention under the standardized incubation conditions. Because the association largely reflects separation among land-use systems, it should not be interpreted as evidence of a direct causal relationship between ORMAI and gaseous emissions.

4. Discussion

Contrasting land-use legacies reorganized the relationship between soil organic-resource status and microbial functional potential rather than shifting these dimensions proportionally. The clearest expression of this reorganization occurred in the paddy soil, where substantial depletion of SOC, total N, and hydrolyzable organic N fractions coincided with high potential enzyme activities, greater abundances of N-cycling functional groups, and the largest potential N₂O and CO₂ emissions. Agroforestry represented a different configuration: it retained more organic resources than the paddy soil but exhibited the lowest microbial functional activation and potential N₂O production. These contrasting configurations indicate that neither the size of organic-resource pools nor conversion to managed land alone was sufficient to predict microbial functional or emission potential. Rather, the potential N₂O response was most clearly expressed when resource status and microbial functional attributes were considered together. The observed patterns were therefore consistent with our hypotheses that managed soils, particularly long-term paddy soil, would have lower organic-resource status than the reference forest, and that high potential N₂O production would coincide with greater microbial functional potential relative to retained resources. These conclusions describe associations among contrasting land-use legacies under standardized incubation conditions; they neither isolate the effects of conversion age nor quantify process rates or annual field emissions.

4.1. Contrasting Land-Use Legacies Reorganized Organic-Resource Status and Microbial Nutrient Allocation

The progressive decline in ORRI from the reference forest to agroforestry and paddy soil observed in our study indicates that the managed systems differed markedly in their relative organic-resource status. This pattern supports the first hypothesis, which predicted that both managed systems would have lower organic-resource status than the reference forest, with the largest reduction under long-term paddy cultivation. Here, ORRI represents an integrated index of measured resource concentrations rather than a direct estimate of areal C or N stocks or in situ retention rates. The greater SOC, total N, MBC, and hydrolyzable organic N concentrations in the reference forest are consistent with sustained litter and root inputs, limited soil disturbance, and the physical and chemical stabilization of organic matter under forest vegetation [7,10,46]. Reductions in SOC and total N following conversion of tropical forests to agricultural land have been reported widely; however, their magnitude varies with climate, soil properties, management, and conversion duration [6,47]. In the present study, the approximately 30% decline in SOC under agroforestry and 61% decline under paddy cultivation were accompanied by corresponding reductions in total N. Their parallel responses suggest that the change was not restricted to a particular N fraction but occurred within a broader reorganization of the soil organic-matter pool.

The organic N fractions refined this interpretation. HAN and HUN decreased progressively across the forest–agroforestry–paddy comparison, whereas ASN and AAN were maintained at similar concentrations in the forest and agroforestry soils but declined sharply in the paddy soil. This contrast indicates that the agroforestry soil did not simply retain a fixed proportion of every forest-associated N fraction. Instead, some components of the hydrolyzable N pool were more strongly preserved than others [20]. Amino sugars are commonly associated with microbial residues, whereas amino acid N includes proteinaceous and peptide-derived compounds with diverse stabilization and turnover characteristics [17,48]. Their relative maintenance under agroforestry may be compatible with continued perennial-root inputs, organic returns, and microbial residue formation; however, the present measurements cannot distinguish among these sources or determine turnover rates [48,49]. Conversely, the consistently low HAN, ASN, AAN, and HUN concentrations in paddy soil indicate broad depletion or redistribution of the measured hydrolyzable organic N pool rather than loss of only one chemical category.

Several mechanisms could contribute to this paddy-associated pattern. Long-term cultivation can alter organic inputs, aggregate protection, mineral–organic associations, and the balance between C stabilization and mineralization [50]. Periodic drainage may expose previously reduced soil microsites to oxygen, potentially stimulating decomposition of susceptible organic matter, whereas puddling and repeated cultivation can modify soil structure and substrate accessibility [51,52]. However, these mechanisms were not measured directly, and long-term paddy management does not invariably reduce SOC or organic N; its effects depend strongly on residue return, fertilization, water regime, and initial soil conditions [53,54]. The present results therefore establish a low organic-resource configuration in the sampled paddy soils but do not identify the individual process responsible for its development.

The inorganic and microbial nutrient pools showed that lower organic-resource status did not correspond uniformly to lower nutrient availability. Agroforestry soil contained the highest NO₃⁻-N concentration and NO₃⁻-N/NH₄⁺-N ratio, together with the highest MBN, whereas paddy soil combined a low NO₃⁻-N/NH₄⁺-N ratio with high NH₄⁺-N, total P, MBP, and MBN. These pools are net outcomes of simultaneous production, immobilization, uptake, and loss and cannot be interpreted as direct measures of the underlying transformation rates [1]. Nevertheless, they demonstrate contrasting nutrient-allocation patterns: agroforestry was associated with relatively oxidized mineral N and strong microbial N accumulation, whereas paddy soil was associated with a greater relative contribution of NH₄⁺-N and strong microbial P accumulation. The fourfold increase in MBN and MBP under agroforestry relative to the forest, despite lower SOC and total N, further indicates that microbial nutrient pools responded differently from bulk organic-resource concentrations. Thus, the managed systems did not follow a single continuum of generalized soil degradation; they developed distinct distributions of organic, mineral, and microbial nutrients.

4.2. Microbial Functional Potential Did Not Vary in Parallel with Organic-Resource Status

The opposing responses of ORRI and MFAI provide the clearest evidence that organic-resource status and microbial functional potential did not vary in parallel. Paddy soil had the lowest ORRI but the highest MFAI, whereas agroforestry retained an intermediate ORRI while exhibiting the lowest MFAI. This divergence is important because it shows why SOC, total N, or individual organic N fractions cannot be used alone to infer potential microbial functioning.

All four extracellular enzymes showed the same land-use ordering—paddy > forest > agroforestry—despite representing microbial acquisition of different resources. The coordinated increase in acid phosphatase, β-glucosidase, N-acetyl-β-D-glucosaminidase, and leucine aminopeptidase under paddy cultivation is therefore more consistent with a broad difference in potential resource-acquisition capacity than with a response confined to a single nutrient cycle. Extracellular enzyme production can be influenced by microbial biomass, community composition, substrate accessibility, nutrient demand, and stabilization of enzymes on soil minerals [55]. Higher activity does not necessarily indicate that the corresponding resource was more abundant or that its in situ decomposition rate was greater. It indicates that, under the assay conditions, the paddy soil possessed a greater potential capacity to catalyze the measured reactions.

The paddy-associated increase in potential enzyme activity despite lower SOC and organic N concentrations can be interpreted in at least two non-exclusive ways. First, reduced resource availability may increase microbial allocation toward extracellular acquisition when organisms must invest more strongly to obtain limiting substrates [56]. Second, the long-term paddy environment may have selected a microbial community with greater enzyme-production capacity or may have altered enzyme stabilization in the soil matrix [57,58]. The higher pH of the paddy soil may also have reduced the strong acidity constraint experienced by the forest and agroforestry communities and contributed to differences in microbial community composition and functional potential [59]. Because the enzymes were measured under standardized assay conditions, however, their activities cannot be attributed directly to the field pH at the moment of measurement. Soil pH is better interpreted as a long-term environmental filter associated with community structure and enzyme production than as the immediate cause of the measured activity [60].

The simultaneous elevation of acid phosphatase, total P, and MBP in paddy soil also requires cautious interpretation. Acid phosphatase activity is often discussed as an indicator of microbial or plant P demand, but its activity can persist when total P is high because total P does not represent immediately available phosphate and because extracellular enzymes can remain stabilized outside living cells [61,62]. Consequently, the high acid phosphatase activity does not by itself demonstrate P limitation. Together with the high MBP, it instead indicates that P acquisition and microbial P immobilization were prominent components of the paddy soil’s functional configuration.

The agroforestry soil presents an equally important result. Its lower enzyme activities cannot be explained simply by depletion of all measured resources because SOC, total N, ASN, AAN, and ORRI remained substantially higher than in paddy soil. The high MBN and NO₃⁻-N concentrations also show that the agroforestry soil was not uniformly nutrient poor. One possibility is that recent organic inputs and microbial immobilization supported nutrient retention without requiring a correspondingly large investment in the measured extracellular enzymes [63]. Alternatively, the acidic conditions, vegetation composition, or management history may have favored a community with lower potential enzyme production [64]. These mechanisms were not measured directly. The defensible conclusion is therefore that six-year poultry-integrated agroforestry established an intermediate-resource but comparatively low-activation configuration, rather than functioning simply as an intermediate stage between forest and paddy soil.

The functional-gene data independently supported this separation between resource status and microbial functional potential. Paddy soil contained the greatest abundances of all measured nitrification- and denitrification-related genes despite having the smallest SOC, total N, and hydrolyzable organic N concentrations. Gene abundances indicate the sizes of populations possessing the targeted functions, not their gene expression or realized transformation rates [34,65]. Nevertheless, the consistent enrichment of AOA-amoA, AOB-amoA, nirK, nirS, and nosZ indicates that the paddy-associated shift was not limited to one functional group. The ammonia-oxidizer balance provides more information than the absolute gene abundances alone [66]. AOA-amoA remained more abundant than AOB-amoA in all three systems, but log₁₀(AOA/AOB) declined progressively from the forest to agroforestry and paddy soils. Thus, the land-use pattern reflected a disproportionately larger increase in AOB rather than a general replacement of AOA by AOB [66]. AOA and AOB occupy overlapping but differentiated niches, with AOA often favored under acidic or low-N conditions and AOB responding more strongly to higher NH₄⁺ supply and less acidic agricultural soils [67,68]. The higher pH, greater NH₄⁺-N availability, and long management history of the paddy soil are consistent with the observed relative expansion of AOB [69]; however, nitrification rates were not measured and cannot be inferred quantitatively from the ratio.

The denitrification-related genes require similar restraint. The concurrent enrichment of nirK, nirS, and nosZ indicates expansion of populations possessing genetic capacities for both nitrite reduction and N₂O reduction [70,71]. Therefore, the higher nosZ abundance in paddy soil occurred within a broader expansion of the denitrifier functional community and provided no evidence of lower net N₂O emissions. The higher log₁₀((nirK + nirS)/nosZ) value in paddy soil indicates that nitrite-reduction genes increased more strongly relative to nosZ than in the other systems. Because this index remained negative, it does not mean that the summed abundance of nirK and nirS exceeded nosZ. It represents a relative shift in gene-abundance balance, not a direct production-to-consumption rate ratio. Its positive association with cumulative N₂O is therefore consistent with, but does not prove, a shift toward greater net N₂O production potential. Overall, these enzyme and gene patterns support the second hypothesis, which suggested that the management and hydrological legacy of paddy cultivation would be associated with greater potential enzyme activities, larger N-cycling functional groups, an increased N₂O production-to-reduction gene-abundance balance, and greater potential N₂O and CO₂ production.

4.3. Potential Gas Emissions Reflected an Integrated Resource–Function Configuration

The standardized incubation provided a common test of how the three soils responded to the same temperature, moisture regime, and urea addition. The resulting fluxes represent potential responses under the specified experimental conditions rather than in situ or annual field-scale emissions. Within this scope, the substantially greater cumulative N₂O and CO₂ emissions from paddy soil show that its low organic-resource status did not translate into low short-term gaseous-emission potential. Quantitative data from comparable tropical forest–paddy land-use systems remain scarce. In a tropical peatland land-use comparison in South Kalimantan, Indonesia, mean N₂O fluxes were approximately 46.0 μg N m⁻² h⁻¹ in secondary forest and 154 μg N m⁻² h⁻¹ in paddy fields, representing a 3.3-fold numerical difference; however, the two land uses did not differ significantly [72]. This contrast is similar in direction to the 3.91-fold higher cumulative potential N₂O emission from paddy than forest soil observed here. However, the absolute values are not directly comparable because the earlier study measured area-based field fluxes in tropical peat soil, whereas our study measured mass-based cumulative emissions during a standardized urea-amended incubation.

The temporal response of N₂O provides additional context. Forest and agroforestry soils maintained relatively low emission rates, whereas paddy soil developed a pronounced peak around day 15 before declining. This delayed peak is compatible with progressive changes in urea hydrolysis, mineral N availability, microbial growth, and the balance between N₂O production and consumption during incubation [73]. However, the experiment did not measure urea hydrolysis, gross nitrification, denitrification, oxygen distributions, or N₂O reduction rates over time. The peak should therefore be interpreted as evidence of a stronger dynamic response by the paddy soil, not as evidence for a particular N₂O-producing pathway.

Several measured features could have contributed to this response. Paddy soil combined higher NH₄⁺-N, less acidic conditions, larger ammonia-oxidizer and denitrifier populations, and greater potential N-acquiring enzyme activities. These attributes provide a plausible functional context for rapid transformation of the added urea N and subsequent N₂O formation [74]. Nevertheless, none of them individually establishes process control. Accordingly, the negative relationships of N₂O to NO₃⁻-N and NO₃⁻-N/NH₄⁺-N ratio are insufficient to infer that low NO₃⁻ availability caused greater N₂O production. A small measured NO₃⁻ pool can result from limited production, rapid microbial or plant immobilization, denitrification, or other consumption pathways [75]. Pool size alone cannot resolve these alternatives.

The CO₂ response indicates that the paddy-associated emission pattern was embedded within a broader change in microbial processing. Cumulative CO₂ was 38% greater in paddy than forest soil, while agroforestry showed an intermediate response. This pattern is compatible with greater utilization of available C during incubation, but it does not demonstrate greater decomposition of the total SOC pool. The amount of SOC was smallest in paddy soil, and cumulative CO₂ reflects the mineralization of accessible substrates rather than the total quantity of C stored in the soil [76]. Moreover, CO₂ did not reproduce the complete correlation pattern observed for N₂O: its associations with the measured extracellular enzymes, nirK, and nosZ were not significant. Thus, the positive relationship between cumulative CO₂ and N₂O identifies a shared land-use-associated response but does not imply identical microbial controls over the two gases.

Together, the univariate results, correlation matrix, and PERMANOVA indicated that resource pools and functional attributes were organized differently among the land uses. ORMAI then expressed the relative distance between these two dimensions. Its markedly positive value in paddy soil indicates that standardized functional attributes were high relative to the measured organic-resource status. By contrast, both forest and agroforestry had negative ORMAI values, indicating that their standardized resource status exceeded their functional activation potential. The similarity of ORMAI between the forest and agroforestry systems does not mean that the two soils were functionally equivalent. Forest soil had both higher ORRI and higher MFAI than agroforestry soil, but the difference between these dimensions was similar. The two systems therefore reached comparable ORMAI values through different component configurations: high resource status with intermediate activation in the forest and intermediate resource status with low activation under agroforestry. Meaningful interpretation of ORMAI therefore requires consideration of ORRI, MFAI, and their constituent variables, rather than treating it as a stand-alone soil-quality score.

The paddy-associated elevation of ORMAI persisted when each constituent variable was removed sequentially and when equal weighting of individual variables replaced domain weighting. This sensitivity supports the conclusion that the pattern was not generated by one unusually responsive enzyme, gene, or organic N fraction. Because N₂O and CO₂ were excluded from index construction, their correlations with ORMAI were not produced by direct mathematical inclusion of the response variables. Nevertheless, the very strong ORMAI–N₂O relationship largely reflects the separation of paddy soil from the other systems and does not establish that ORMAI is a causal predictor of N₂O. Within these boundaries, the results support the third hypothesis, which predicted that potential N₂O emission would be aligned more closely with the integrated functional configuration and its imbalance relative to retained resources than with large SOC, total N, or hydrolyzable organic N concentrations.

4.4. Implications of the Resource–Function Imbalance Framework

Our results show that contrasting land-use legacies can reorganize the correspondence between retained organic resources and microbial functional potential in qualitatively different ways. Long-term paddy cultivation produced a resource-depleted but functionally activated configuration, whereas the agroforestry soil retained an intermediate resource base but showed comparatively restrained functional activation. These configurations remained evident when the soils were exposed to common incubation conditions, indicating that their potential gaseous responses reflected legacy-associated differences in soil resource–function organization rather than differences in immediate temperature, moisture, or N addition. Thus, evaluating the two dimensions independently—and quantifying their relative imbalance—provides information that cannot be obtained by ranking individual resource or functional variables alone.

The strong contrast in ORMAI between paddy soil and the other systems illustrates the potential diagnostic value of this framework. A positive imbalance does not indicate desirable soil functioning, nor does it imply that resource depletion stimulates N₂O production. Rather, it identifies a configuration in which measured microbial functional potential is high relative to the organic-resource base available to support it. Such a configuration may signal greater potential for rapid resource processing and N₂O emissions following favorable environmental conditions or external N supply [68]. Conversely, a negative imbalance identifies systems in which retained-resource status exceeds the measured level of functional activation, without implying that those resources are permanently stabilized or unavailable. ORMAI should therefore be regarded as a study-specific descriptor of resource–function configuration, not as a universal soil-quality index or a causal predictor of emissions.

The agroforestry soil further demonstrates why land-use systems should not be arranged along a single gradient from forest to intensive agriculture. Its combination of partial organic-resource preservation, high MBN and measured inorganic N concentrations, low enzyme and functional-gene indices, and the lowest cumulative potential N₂O emissions represented a distinct configuration rather than an intermediate stage between forest and paddy soil. This pattern suggests that the vegetation and management characteristics of the six-year poultry-integrated system had not generated the functionally activated configuration observed under long-term paddy cultivation. Nevertheless, because only one agroforestry design and sampling period were examined, and field N inputs, plant uptake, seasonal dynamics, and annual gaseous losses were not quantified, these findings do not establish agroforestry generally as an N₂O-mitigation strategy. They instead identify a specific land-use configuration that merits testing across management intensities, seasons, and field conditions.

4.5. Limitations

Several design limitations define the strength of inference. First, the study used a spatial comparison rather than repeated measurements of the same sites before and after conversion. Comparable texture, parent material, weathering status, regional climate, and original forest type strengthen the comparison, but they cannot eliminate all unmeasured site differences. Second, land-use type and conversion duration were confounded: agroforestry had been established for approximately six years, whereas paddy cultivation had continued for about 60 years. The results therefore represent the integrated legacies of contrasting conversion pathways rather than independent effects of management type or elapsed time. Third, each land-use system was represented by three independent field replicates. This design was sufficient to detect large and internally consistent differences, but it limits the stability and generalizability of replicate-level correlations. In particular, the high correlations between ORMAI and gas emissions were dominated by between-system separation and require validation across more sites and within broader gradients of management intensity. Fourth, ORRI summarizes relative concentrations of selected organic-resource indicators; it is not a measurement of C or N stocks, stabilization rates, or retention efficiency. MFAI similarly combines potential enzyme activities and DNA-level gene abundances and therefore represents potential functional capacity rather than realized microbial process rates [77]. ORMAI is therefore a study-specific descriptor of relative resource–function configuration, not a universal index or mechanistic rate model.

4.6. Future Perspectives

Future work could test this framework using seasonal field measurements, bulk-density-corrected C and N stocks, gross N-transformation rates determined with 15N tracers, direct measurements of N₂O production and reduction, and expression of the targeted functional genes [78,79]. Repeated sampling across paddy drainage cycles and agroforestry development stages would help separate management effects from conversion duration. Such measurements would also determine whether the resource–function configurations identified under standardized incubation persist under field conditions.

5. Conclusions

Within these boundaries, the study shows that conversion of a common natural secondary forest type was associated with contrasting soil biogeochemical legacies. The reference forest retained the largest organic-resource concentrations, poultry-integrated agroforestry maintained an intermediate resource status but comparatively low functional activation, and long-term paddy soil exhibited the strongest imbalance between diminished organic-resource status and elevated microbial functional and emission potentials. The value of the proposed framework lies not in replacing individual soil indicators, but in showing that retained resources, microbial functional capacity, and gaseous responses represent different dimensions of land-use change. Their alignment—or lack of alignment—provides a more informative basis for interpreting soil functional responses than changes in SOC or total N alone. From a practical perspective, the results indicate that organic-resource concentrations alone may not adequately identify soils with high potential N₂O responses. The resource-depleted but functionally activated configuration of the paddy soil suggests that assessments of potential N₂O susceptibility should consider microbial functional attributes alongside SOC and organic N status, particularly when evaluating N-input and drainage–rewetting management. By contrast, the agroforestry soil represented a distinct configuration with partial resource preservation and comparatively low functional activation; however, the present evidence is insufficient to generalize this system as an N₂O-mitigation strategy. ORMAI may therefore serve as a study-specific screening descriptor for identifying resource–function configurations that warrant closer field evaluation, rather than as a universal soil-quality or emission-risk index. Future studies should test this framework across additional sites, seasons, management intensities, and conversion stages using year-round field N₂O measurements, bulk-density-corrected C and N stocks, drainage and fertilization records, ¹⁵N-based gross N-transformation rates, direct N₂O production and reduction measurements, and functional-gene expression. These measurements are needed to determine whether the configurations observed under standardized incubation persist under field conditions and to separate the effects of management regime from those of conversion duration.