Concurrent Elevation of CO₂ and Temperature Stimulates N₂O Emissions from Rice Paddies in a Rice–Wheat Cropping System

Abstract

Paddy fields are a major agricultural hotspot for nitrous oxide (N₂O), contributing approximately 11% of global agricultural emissions. While elevated CO₂ and warming individually regulate N₂O emissions by modulating soil carbon, nitrogen (N) availability and microbial activity, the effects of concurrent elevated CO₂ and temperature (ECT) and the underlying microbial mechanisms under field conditions remain poorly understood. Here, we used a free-air CO₂ enrichment and temperature increase (T-FACE) system in a rice–wheat cropping system to investigate the impacts of ECT on N₂O emissions from rice paddies and identify the underlying biogeochemical and microbial mechanisms. Results showed that ECT increased area-scaled and yield-scaled N₂O emissions by 15.3% and 17.6%, respectively. Mechanistically, during the peak emission period, ECT significantly increased soil NH₄⁺–N content by 40.2% and the denitrification gene ratio [(nirK + nirS)/nosZ] by 27.5%. Furthermore, ECT increased the diversity of nitrifying communities but decreased that of denitrifying communities, while reshaping the composition of ammonia-oxidizing archaea and denitrifiers, thereby altering nitrification and denitrification. Overall, our field-based evidence suggests that ECT can stimulate N₂O emissions primarily by increasing soil N substrate availability and shifting denitrifier communities in ways that may favor N₂O accumulation. These findings offer mechanistic insights into climate-driven N₂O emissions.

1. Introduction

Rising atmospheric CO₂ concentrations and temperatures, driven by human activities since the Industrial Revolution, represent defining hallmarks of contemporary climate change [1,2,3]. As CO₂ acts as the primary substrate for plant photosynthesis and temperature fundamentally dictates the metabolic rates of plants and soil microorganisms, their concurrent elevation exerts profound challenges to the stability, biogeochemical functioning, and sustainability of global agroecosystems [4,5,6,7].

Rice (Oryza sativa L.) serves as the primary staple food for more than half of the global population, playing a central role in global food security [8,9]. However, flooded paddy soils are also a major agricultural hotspot for nitrous oxide (N₂O) emissions—a potent, long-lived greenhouse gas with a global warming potential nearly 300 times that of CO₂ and the primary agent of stratospheric ozone depletion [10,11]. Globally, rice paddies emit approximately 1.7 kg N ha⁻¹ yr⁻¹ of N₂O, accumulating to a total annual flux of ~130 Gg yr⁻¹ and accounting for ~11% of global agricultural N₂O emissions [12,13]. Driven by intensive reactive nitrogen (N) inputs, atmospheric N₂O concentrations are accelerating at a rate of 0.75–1.0 ppb yr⁻¹ [14,15,16]. Given that agricultural soils contribute roughly half of global N₂O emissions [15,17], deciphering how paddy N₂O emissions respond to future climate scenarios is critical for accurately predicting climate–carbon/nitrogen feedbacks and mitigating agricultural greenhouse gas forcing.

Soil N₂O emissions stem primarily from microbial nitrification and denitrification processes, both of which are highly sensitive to CO₂ enrichment and elevated temperatures [18,19]. While a substantial body of research has documented the individual effects of elevated CO₂ or warming on paddy N₂O emissions [20,21,22], studies examining concurrent elevated CO₂ and temperature (ECT) remain limited [23,24,25]. For instance, Pereira et al. (2013) [24] reported that ECT exerted a negligible effect on seasonal N₂O emissions, whereas Bhattacharyya et al. (2013) [23] and Wang et al. (2018) [25] observed substantial emission surges under ECT scenarios. A recent global synthesis further highlighted that ECT increased N₂O emissions by an average of 36% relative to ambient conditions [19]. Conceptually, these emission increases are likely attributed to enhanced photosynthate allocation to roots under elevated CO₂, which elevates labile soil carbon availability and fuels denitrification [26], coupled with warming-induced acceleration of soil organic N mineralization [27,28], which boosts substrate supply for both nitrification and denitrification [29]. Nevertheless, these studies have predominantly focused on N₂O emission dynamics, lacking direct empirical evidence that couples emission patterns with underlying biogeochemical and microbial pathways.

Existing literature indicates that ECT can reshape soil N transformations via dual biogeochemical pathways. On the one hand, ECT alters substrate availability by expanding total and labile N pools [30], reshaping microbial necromass accumulation and amino sugar dynamics [31], and stimulating root exudation through increased C allocation [23]. On the other hand, ECT restructures key functional microbial communities, selectively modulating root-associated C/N-metabolizing bacteria [32], shifting ammonia-oxidizing bacteria (AOB) community structures [33,34], and altering the balance between active AOB and nitrite-oxidizing bacteria (NOB) [35]. Notwithstanding these isolated biogeochemical and microbiological insights, most studies have evaluated soil chemical properties or microbial taxonomic shifts independently, without synchronously tracking field-scale N₂O flux trajectories. Crucially, how these ECT-induced microbial restructurings and substrate shifts interactively govern real-time paddy N₂O emissions under realistic field conditions remains a major knowledge gap.

To address this knowledge gap, we conducted a field experiment using a free-air CO₂ enrichment and temperature increase (T-FACE) system in a typical rice–wheat cropping system. The objectives of this study were to quantify the impacts of concurrent elevated CO₂ and warming on in situ N₂O emissions from rice paddies and to unravel the underlying substrate and microbial mechanisms driving these emission responses. We hypothesized that ECT would exert a synergistic positive effect on N₂O emissions, driven by enhanced root C exudation and accelerated N mineralization, which collectively stimulate the abundance and metabolic activity of nitrifiers and denitrifiers. Findings from this study will provide a mechanistic foundation for refining global biogeochemical models and developing high-yield, low-emission rice management strategies under future climate scenarios.

2. Materials and Methods

2.1. Experimental Design

The field experiment was initiated in 2021 at the Danyang Experimental Station of Nanjing Agricultural University (31.9° N, 119.5° E), Baolin Village, Jiangsu Province, China. The site is characterized by a typical subtropical monsoon climate with four distinct seasons and has been managed under a winter wheat–rice cropping system for many years. The mean annual temperature and precipitation for 2021–2024 were 17.0 °C and 1281 mm, respectively. Prior to establishment of the experimental platform in 2021, the topsoil (0–20 cm) had the following baseline properties: soil organic carbon, 21.1 g kg⁻¹; total nitrogen, 1.1 g kg⁻¹; total phosphorus, 0.5 g kg⁻¹; and total potassium, 12.6 g kg⁻¹.

The field experiment comprised two treatments, namely an ambient control (CK; ambient CO₂ concentration and temperature) and a concurrent elevated CO₂ and temperature (ECT) treatment, with three replicates arranged in a randomized complete block design. Prior to the experiment, the soil within each block was thoroughly tilled and mixed to ensure within-block homogeneity. The study was conducted using a free-air CO₂ enrichment and temperature increase (T-FACE) system, which integrates a Free-Air CO₂ Enrichment (FACE) system with a Free-Air Temperature Increase (FATI) array (Figure S1). The FACE system consisted of six octagonal rings (8 m in diameter), with a minimum spacing of 25 m between rings to avoid gas cross-contamination. In the ECT treatment, atmospheric CO₂ concentration was elevated by approximately 160 ppm above ambient, reaching about 550 ppm, consistent with the IPCC medium-emission scenario (SSP2-4.5) for the end of this century [36]. High-purity CO₂ from a 20-m³ liquid storage tank was delivered through a main pipeline and injected into each ring via micro-porous perimeter tubes. Real-time control was managed by an automated central system (Yisheng Taihe, China). Each ring was equipped with eight evenly distributed CO₂ sensors (VC2008T, SenseAir, Delsbo, Sweden), a temperature sensor, and an anemometer-wind vane. Gas release was dynamically regulated according to wind speed and direction, and injection was automatically suspended when wind speed exceeded 5 m s⁻¹ to improve gas use efficiency. Fumigation was conducted during daytime hours (05:00–19:00). Injection tubes and sensor arrays were suspended 60–75 cm above the rice canopy and adjusted weekly as the crop grew. Control rings were fitted with identical non-fumigating dummy pipelines to account for potential microclimatic effects. During the 2024 rice-growing season, mean daytime CO₂ concentrations were 542 ppm in ECT rings and 397 ppm in CK rings.

The FATI system consisted of an infrared heating array coupled with an automatic temperature-tracking platform. Each warming unit included an aluminum reflector plate (180 cm × 20 cm) fitted with an 800-W far-infrared electric heating element. The target warming level was set at 1.0–2.0 °C above ambient, which is within the range of projected warming under future climate scenarios [37]. Heating plates were suspended approximately 75 cm above the canopy to warm both the upper canopy and surface soil through downward infrared radiation. Unheated dummy plates were installed in CK plots at the same height to minimize shading and other microclimatic artifacts. Canopy temperature was monitored using infrared sensors directed at the uppermost fully expanded leaves, whereas soil temperature was continuously recorded at a 5-cm depth using vertically inserted probes connected to a multi-channel automatic logger at 10-min intervals. Plate height was adjusted weekly to maintain a constant distance from the canopy. During the 2024 rice-growing season, mean canopy temperature was 26.73 °C in CK and 27.63 °C in ECT; mean surface soil temperature was 25.27 °C in CK and 25.62 °C in ECT.

2.2. Crop Management

The local conventional japonica rice cultivar Wuyunjing 23 was used. Twenty-five-day-old seedlings were manually transplanted in late June and harvested in early November. Plants were spaced at 25 cm × 15 cm, with three seedlings per hill. Buffer rows with the same spacing were established around the experimental area to minimize edge effects. Nitrogen fertilizer was applied at 225 kg N ha⁻¹ as urea (46% N) in three splits, i.e., 40% as basal fertilizer 1 d before transplanting, 30% at tillering (7 d after transplanting), and 30% at panicle initiation. Phosphorus was applied as superphosphate (120 kg P₂O₅ ha⁻¹; 12% P₂O₅) entirely as a basal dressing. Potassium was applied as potassium chloride (160 kg K₂O ha⁻¹; 60% K₂O), split equally between basal application and panicle initiation. A conventional alternate wetting and drying (AWD) regime was adopted, including shallow flooding during the early growth stage to promote tillering, mid-season drainage to suppress unproductive tillers, and AWD cycles thereafter until final drainage 7 d before harvest.

The octagonal rings and permanent PVC bases remained in place throughout the rice and wheat growing seasons and were designed to minimize interference with normal field operations. All agronomic practices inside the rings, including tillage, planting, fertilization, irrigation, and harvest, were performed manually and followed the same schedule and methods as the surrounding field. The permanent PVC bases remained flush with the soil surface and therefore did not obstruct planting, manual weeding, fertilizer application, or harvesting; planting density and crop management within the bases were identical to those in the surrounding ring area. Specifically, during the wheat season, the field was maintained under rainfed conditions with supplemental irrigation as needed. Between the rice and wheat seasons, crop residues were returned to the field, and the land was prepared by shallow manual tillage within the rings to simulate field-scale operations. All other management practices, including pest, disease, and weed control, followed local high-yield recommendations.

2.3. Sampling and Measurement Methods

2.3.1. N₂O Emission Measurement

N₂O fluxes were measured during the entire three-year experimental period (2022–2024) using the static closed-chamber technique [38]. Since the interannual patterns were highly similar, the 2024 dataset is presented herein as a representative case.

Each sampling system consisted of a permanent PVC base (50 cm × 50 cm × 15 cm), inserted 2–3 d before transplanting, and a removable PVC chamber (50 cm × 50 cm; 50 or 100 cm high depending on plant height). The chamber was externally wrapped with reflective insulation and sponge to minimize solar heating during sampling. To standardize diurnal variation, gas sampling was consistently performed between 09:00 and 11:00 a.m. on each sampling day. At the time of measurement, thermally insulated chambers were fitted into the water-filled grooves of the permanent PVC bases to establish an airtight seal. Gas samples were withdrawn at 10-min intervals over a 30-min closure period (i.e., at 0, 10, 20, and 30 min after chamber deployment) and transferred into pre-evacuated 500-mL aluminum foil gas bags for offline analysis. Samples were transported to the laboratory and analyzed for N₂O concentrations within 24 h using a photoacoustic gas analyzer (INNOVA 1412, LumaSense, Ballerup, Denmark), which was calibrated with standard gases prior to each measurement to ensure analytical precision.

Gas sampling was performed weekly, with additional sampling after fertilization and 2–3 extra samplings during the mid-season drainage period. N₂O flux (F, mg m⁻² h⁻¹) was calculated as:

$$F=\rho \times {\displaystyle \frac{273}{(273+T)}}\times {\displaystyle \frac{V}{S}}\times {\displaystyle \frac{\mathrm{\Delta }c}{\mathrm{\Delta }t}}$$

(1)

where ρ denotes the standard-state density of N₂O (1.977 g L⁻¹); T is the mean chamber air temperature (°C); V and S refer to headspace volume (m³) and base area (m²), respectively; and Δc/Δt is the linear rate of N₂O concentration change (µL L⁻¹ h⁻¹). To ensure data reliability, only flux measurements with a coefficient of determination (R²) > 0.90 for the linear regression were retained for further analysis. To ensure representative gas sampling and data accuracy, two gas sampling points (i.e., chambers) were established within each octagonal ring as technical replicates (Figure S1), and the measurements from the two chambers were averaged to obtain a single value per biological replicate (ring). Seasonal cumulative area-scaled N₂O emissions were calculated by trapezoidal integration of all fluxes over the rice-growing season, and yield-scaled emissions were expressed as kg N₂O t⁻¹ grain yield.

2.3.2. Aboveground Biomass and Plant N Accumulation

At maturity, three representative hills were selected from each plot (octagonal ring) according to the average number of productive tillers. Aboveground biomass was determined after oven-drying the samples to constant weight at 70 °C. The dried plant material was ground, and N concentration was measured by Dumas combustion using an elemental analyzer (vario EL cube, Langenselbold, Germany). Plant N accumulation was calculated as biomass multiplied by N concentration. Grain yield was determined by harvesting, threshing, and air-drying representative hills from the center of each plot. Final yield was adjusted to 14.5% moisture content according to the national standard for japonica rice.

2.3.3. Soil Characteristics

During the 2024 N₂O emission peak, soil samples (0–15 cm depth) were collected by five-point sampling with a 5-cm-diameter corer and pooled per ring to obtain a representative composite sample. After homogenization and sieving, each sample was split into two portions. One was stored at 4 °C for analysis of dissolved organic carbon (DOC), NH₄⁺-N, NO₃⁻-N [39,40] and soil N mineralization rate [41], and the other was stored at −80 °C for microbial functional gene abundance and high-throughput sequencing. DOC was extracted with deionized water (1:5 w/v) and quantified on a total organic carbon analyzer (Multi N/C 3100, Analytik Jena AG, Jena, Germany). NH₄⁺-N and NO₃⁻-N were extracted with 2 M KCl (1:5 w/v) and determined using a flow autoanalyzer (Auto Analyzer 3, BRAN LUEBBE, Norderstedt, Germany). Soil N mineralization rate was measured using an incubation method. In brief, duplicate fresh soil samples, each equivalent to 5 g of dry soil, were prepared. One subsample was immediately extracted for NH₄⁺–N and NO₃⁻–N analysis, while the other was incubated in the dark at 25 °C for 28 days. During incubation, deionized water was added periodically to maintain constant soil moisture. At the end of the incubation, the second subsample was similarly extracted and analyzed. The N mineralization rate was calculated using the following equation [42]:

$$N_m=(N_I-N_S)/D$$

(2)

where N_m is the soil N mineralization rate (mg kg⁻¹ d⁻¹), N_I and N_S are the total soil NH₄⁺-N and NO₃⁻-N concentrations (mg kg⁻¹) after and before incubation, respectively, and D is the incubation duration (d).

2.3.4. Microbial DNA Extraction and Functional Gene Quantification

Frozen soil samples were thawed and DNA was extracted from 0.25 g of soil using the DNeasy PowerSoil Pro Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s protocol. DNA concentration and purity were assessed using a NanoDrop spectrophotometer (Thermo Scientific, Wilmington, DE, USA). Absolute quantitative PCR was performed on an ABI Q6 Flex Real-Time PCR system using ChamQ SYBR qPCR Master Mix (Vazyme, Nanjing, China). The primer pairs ArchamoAF-ArchamoAR [43], amoA1F-amoA2R [44], NirK1F-NirK5R [45], nirSF-NirSR [46], and nosZ1F-nosZ1R [47] were used to quantify the copy numbers of amoA-AOA, amoA-AOB, nirK, nirS, and nosZ genes, reflecting nitrifier and denitrifier abundances, respectively. Each 20-µL reaction consisted of 10 µL 2× ChamQ SYBR qPCR Master Mix (Vazyme, China), 0.4 µL of each primer (10 µM), 0.4 µL 50× ROX Reference Dye 2, 1.5 µL DNA template, and nuclease-free water to volume. Thermal cycling followed the manufacturer’s recommendations with slight adjustments to the annealing temperature. All runs were executed on a QuantStudio™ 6 Flex Real-Time PCR System (Applied Biosystems, Foster City, CA, USA).

2.3.5. High-Throughput Sequencing and Analysis

Soil microbial DNA was extracted using the same protocol and analyzed by functional gene amplicon sequencing. The sequencing pipeline comprised PCR amplification, product purification, quantification and normalization, PE300 library preparation, and high-throughput sequencing. Before library construction, DNA integrity was assessed by 1% agarose gel electrophoresis. Target genes were amplified with the same primers as in the qPCR assay, and the amplicons were visualized on 2% agarose, purified with the AxyPrep DNA Gel Recovery Kit (AXYGEN, Union City, CA, USA), and quantified on a Quantus™ Fluorometer (Promega, Madison, WI, USA). Purified products were then pooled proportionally to meet the sequencing depth required per sample. Libraries were constructed with the NEXTFLEX^® Rapid DNA-Seq Kit and sequenced on the Illumina MiSeq PE300 platform.

Raw reads were quality-filtered, and non-redundant sequences were retained after singleton removal. OTUs were delineated at 97% similarity using UPARSE [48], with chimeras discarded and representative sequences selected. An OTU table was generated by mapping quality-filtered reads to these representative sequences at ≥97% identity. Taxonomic classification was performed by aligning representative sequences against the SILVA 16S rRNA database (Release 138) with the RDP Classifier 2.2 at an 80% confidence threshold [49], and community composition was summarized at different taxonomic levels.

Subsequent analyses included alpha diversity, beta diversity and community composition. Alpha diversity was assessed via rarefaction analysis using Mothur v.1.21.1 [50], yielding observed species, Chao1, ACE, Shannon, Simpson, and Pielou_J indices; between-treatment differences for each index were tested using the Wilcoxon rank-sum test. Beta diversity was assessed by principal coordinate analysis (PCoA) of Bray–Curtis distances, computed using the vegan package, and the PCoA was performed with the ape package. Community composition was analyzed at the genus level based on relative abundances, and treatment effects were compared using the Wilcoxon rank-sum test.

2.4. Statistical Analysis

The independent-samples t-test was used to evaluate the effects of concurrent elevated CO₂ and temperature on N₂O emissions, rice biomass, plant N accumulation, soil properties, and the abundances of nitrifiers and denitrifiers. All statistical analyses were performed using IBM SPSS Statistics 20, and figures were prepared with GraphPad Prism 9. Statistical significance was accepted at p < 0.05.

3. Results

3.1. N₂O Emissions

The temporal pattern of N₂O fluxes was similar in CK and ECT, with fluxes remaining low under continuous flooding and increasing sharply after mid-season drainage and during the subsequent alternate wetting and drying cycles (Figure 1a). Relative to CK, ECT significantly increased seasonal cumulative area-scaled N₂O emissions by 15.3% (Figure 1b). Yield-scaled N₂O emissions were also higher under ECT, increasing by 17.6% compared with CK (Figure 1c).

3.2. Rice Biomass, Plant N Accumulation and Soil Properties

ECT did not significantly affect aboveground biomass or plant N accumulation at maturity (Figure 2a,b). At the peak stage of N₂O emission, ECT significantly increased the soil N mineralization rate by 55.9% (Figure 2c). Soil DOC showed an increasing trend under ECT, with an 8.3% increase relative to CK (Figure 2d). In contrast, ECT increased soil NH₄⁺–N content by 40.2% and decreased NO₃⁻–N content by 30.5% (Figure 2e,f).

3.3. Nitrifier and Denitrifier Abundances

ECT did not significantly affect the abundances of AOA, AOB, nirK-type denitrifiers, or nosZ-type denitrifiers (Figure 3a,b,e). The abundance of nirS-type denitrifiers showed an increasing trend under ECT, with a 22.1% increase relative to CK, although the difference was not significant (Figure 3d). In contrast, the (nirK + nirS)/nosZ ratio was significantly increased by 27.5% under ECT (Figure 3f).

3.4. Nitrifier and Denitrifier Communities

3.4.1. Community Alpha Diversity

Compared with CK, ECT significantly increased AOA richness, diversity, and evenness. Specifically, observed species, Chao1, and ACE increased by 20.1%, 13.7%, and 15.4%, respectively, while the Shannon index, Simpson index, and Pielou’s evenness increased by 20.9%, 8.5%, and 16.8%, respectively (

Table 1. Alpha diversity of soil nitrifier and denitrifier communities as affected by concurrent elevated CO₂ and temperature.

		Community Richness			Community Diversity		Pielou’s Evenness
		Observed Species	Chao 1	ACE	Shannon	Simpson	Pielou_J
AOA	CK	193 ± 7	230 ± 12	226 ± 9	2.41 ± 0.02	0.8260 ± 0.0045	0.4582 ± 0.0070
	ECT	232 ± 4 *	261 ± 6 1	261 ± 4 *	2.91 ± 0.03 *	0.8958 ± 0.0028 *	0.5349 ± 0.0033 *
AOB	CK	868 ± 9	1058 ± 24	1053 ± 17	3.96 ± 0.03	0.9430 ± 0.0019	0.5850 ± 0.0037
	ECT	886 ± 6	1088 ± 19	1080 ± 8	4.02 ± 0.02	0.9454 ± 0.0014	0.5929 ± 0.0036
nirK-type	CK	724 ± 29	877 ± 24	870 ± 22	4.94 ± 0.12	0.9762 ± 0.0036	0.7503 ± 0.0153
	ECT	718 ± 59	882 ± 61	860 ± 67	4.31 ± 0.10 *	0.9421 ± 0.0042 *	0.6570 ± 0.0090 *
nirS-type	CK	1333 ± 32	1567 ± 53	1596 ± 60	5.23 ± 0.07	0.9794 ± 0.0012	0.7272 ± 0.0104
	ECT	1279 ± 44	1515 ± 66	1544 ± 69	4.92 ± 0.06 *	0.9506 ± 0.0039 *	0.6885 ± 0.0066 *
nosZ-type	CK	1592 ± 30	1935 ± 37	1915 ± 33	5.88 ± 0.02	0.9917 ± 0.0004	0.7971 ± 0.0023
	ECT	1642 ± 25	1997 ± 29	1978 ± 28	5.81 ± 0.03	0.9907 ± 0.0006	0.7847 ± 0.0033 *

Notes. Values are presented as mean ± standard error. An asterisk (*) indicates a significant difference at p < 0.05, whereas the absence of an asterisk denotes no significant difference between the CK and ECT treatments. ¹ p = 0.05. CK, ambient CO₂ and temperature; ECT, concurrent elevated CO₂ and temperature.

). In contrast, AOB alpha diversity did not differ significantly between treatments (Table 1).

For the nirK-type denitrifier community, richness indices did not differ significantly between CK and ECT, but ECT significantly decreased the Shannon index by 12.7%, the Simpson index by 3.5%, and Pielou’s evenness by 12.4% (Table 1). Similarly, for the nirS-type community, richness was unaffected, whereas the Shannon index, Simpson index, and Pielou’s evenness were significantly reduced by 5.9%, 2.9%, and 5.3%, respectively (Table 1). For the nosZ-type community, ECT significantly decreased Pielou’s evenness by 1.6%, while the other alpha diversity indices remained unchanged (Table 1).

3.4.2. Community Beta Diversity

ECT markedly altered the beta diversity of soil nitrifier and denitrifier communities. The AOA and AOB communities under ECT were clearly separated from those under CK. The first two principal coordinates explained 84.1% and 6.4% of the variance for AOA, and 50.1% and 13.1% for AOB, respectively (Figure 4a,b).

Similarly, the nirK-, nirS- and nosZ-type denitrifier communities under ECT were clearly separated from CK. The first two principal coordinates explained 34.0% and 17.6% of the variance for nirK, 59.1% and 12.0% for nirS, and 31.2% and 15.7% for nosZ, respectively (Figure 4c–e).

3.4.3. Composition of Nitrifier and Denitrifier Communities

ECT changed the community composition of AOA and nirK-, nirS-, and nosZ-type denitrifiers at the genus level (Figure 5). Within the AOA community, ECT reduced the relative abundance of Nitrosopumilus by 29.5% and increased the relative abundance of Candidatus Nitrosotalea approximately sevenfold compared with CK (Figure 5a). The AOB community was overwhelmingly dominated by Nitrosospira in both CK and ECT (>99% relative abundance), and no genus showed a significant response to ECT (Figure 5b).

Among nirK-type denitrifiers, ECT increased the relative abundance of Rhodanobacter and Gemmatimonas by 38.6% and 127.0%, respectively (Figure 5c). For nirS-type denitrifiers, ECT significantly decreased Anaerolinea by 59.6% but increased Sideroxydans by 156.7% (Figure 5d). For the nosZ-type community, ECT did not significantly alter the relative abundance of the dominant genera (Figure 5e).

4. Discussion

Our field experiment showed that concurrent elevated CO₂ and temperature significantly increased N₂O emissions from rice paddies in a rice–wheat cropping system, consistent with earlier findings [23,25]. This stimulation may be attributed to several mechanisms. First, ECT may have increased labile carbon inputs to the paddy soil. Elevated CO₂ often enhances photosynthesis and root carbon allocation, thereby increasing root exudation and the supply of labile C to the rhizosphere, which in turn can stimulate nitrification and denitrification and promote N₂O production [26,51]. In our study, soil DOC content tended to be higher under ECT (Figure 2), implying a greater supply of labile C for nitrification and denitrification that likely enhanced N₂O emissions [23,25,51].

Second, ECT likely enhanced N turnover in the soil, thereby increasing the supply of substrates for N₂O production. Warming can accelerate soil organic N mineralization and increase the availability of mineral N, especially NH₄⁺-N, which can support both nitrification and denitrification processes [20,29]. Consistent with this mechanism, ECT significantly increased the soil N mineralization rate and NH₄⁺-N content in our study (Figure 2). The enrichment of available N under ECT likely provided more substrates for microbial N transformations, ultimately facilitating greater N₂O emissions [16]. By contrast, plant N accumulation was not significantly increased, consistent with Pereira et al. (2013) [24], indicating that the additional mineral N generated under ECT was not fully retained by crop uptake and may instead have been more readily available for microbial processes.

Third, ECT altered the abundance and structure of nitrifying and denitrifying communities, thereby shifting the balance between N₂O production and reduction. Nitrifiers and denitrifiers are the key microbial groups governing N₂O emissions in rice paddies [18,19]. In our study, ECT significantly increased the functional denitrification gene ratio (nirK + nirS)/nosZ (Figure 3), indicating that denitrification was shifted toward N₂O production rather than complete reduction to N₂. However, ECT did not significantly affect the abundances of AOA and AOB, suggesting that the stimulatory effect of ECT on N₂O emissions was more likely driven by changes in denitrifier function than by changes in nitrifier abundance. In addition, ECT increased nirS abundance more strongly than nirK, implying that nirS-type denitrifiers may have played a greater role in promoting incomplete denitrification and the resulting N₂O accumulation.

At the community level, ECT reshaped both nitrifier and denitrifier assemblages (Table 1; Figure 5). ECT increased the alpha diversity of the AOA community (Table 1), accompanied by a decrease in the relative abundance of Nitrosopumilus and an increase in the relative abundance of Ca. Nitrosotalea (Figure 5). This pattern likely reflected the altered soil carbon and nitrogen status under ECT. Specifically, the increased NH₄⁺-N concentration and the tendency toward higher DOC content (Figure 2) may have created more heterogeneous niches for different AOA groups, thereby promoting community diversification and compositional turnover [52]. This is consistent with a previous study showing that ECT can enhance nitrifier diversity [35].

However, ECT did not significantly affect denitrifier richness but reduced community diversity and evenness (Table 1), indicating a more uneven and selectively structured denitrifier community (Figure 5). These changes were likely driven by shifts in soil pH, C/N ratio, and O₂ availability under ECT, which may have acted as environmental filters [23,30,53,54]. Among the changed taxa, the enrichment of Rhodanobacter is notable because this genus is often associated with acidic and low-redox environments and has been linked to N₂O production [54,55,56,57]. In some Rhodanobacter strains, the absence of N₂O reductase prevents the final reduction of N₂O to N₂, which may favor N₂O accumulation [58]. The increase in the relative abundance of Gemmatimonas may have been supported by greater labile C input from elevated CO₂ and enhanced N mineralization under warming [26,27,30,59].

In contrast, the decline in the relative abundance of Anaerolinea may have been associated with increased O₂ release from roots under elevated CO₂, which would be unfavorable for this obligate anaerobe [53,60,61]. The increase in Sideroxydans may further indicate strengthened Fe and N cycling under ECT, as this nitrate-dependent Fe(II)-oxidizing bacterium can couple Fe²⁺ oxidation to nitrate reduction [62,63]. Under ECT, warming may have accelerated microbial O₂ consumption, lowered soil redox potential, and enhanced Fe³⁺ reduction, thereby increasing Fe²⁺ availability and creating more favorable conditions for Sideroxydans [23,64]. Taken together, these results suggest that ECT promoted N₂O emissions mainly by altering the abundance and community structure of key nitrifying and denitrifying microbes, which likely shifted microbial N transformation toward greater N₂O production.

Our results showed that ECT increased yield-scaled N₂O emissions, underscoring a dual challenge for rice–wheat systems, i.e., maintaining yield while mitigating N₂O release under future warmer and CO₂-enriched conditions. Addressing this challenge will require integrated strategies that simultaneously improve productivity and reduce N₂O emissions. For example, optimizing straw management by converting straw into biochar before soil incorporation may enhance carbon sequestration, improve soil fertility, and substantially reduce N₂O emissions [65,66,67]. The use of enhanced-efficiency fertilizers, such as nitrification inhibitors, urease inhibitors, and controlled-release fertilizers, can also improve nitrogen use efficiency and mitigate N₂O emissions in intensive paddy systems [65,68,69]. In addition, deep placement of nitrogen fertilizer has been shown to increase rice yield while reducing N₂O emissions [66]. Finally, breeding climate-resilient rice varieties, particularly through hybrid breeding approaches, may help combine heat tolerance with low-emission traits and thereby reduce both yield losses and N₂O release under future climate change scenarios [70,71]. Moving forward, increased efforts will be needed to adapt to and mitigate the negative effects of climate change, ensuring the sustainability of rice yields and minimizing N₂O emissions from rice agriculture.

This study has some limitations. First, soil NH₄⁺, NO₃⁻, and DOC were measured only once, during the peak N₂O emission period. This timing was chosen to capture soil N and C availability under conditions most relevant to the observed fluxes while minimizing disturbance to the long-term non-destructive gas flux measurements. However, a single sampling event cannot capture the seasonal dynamics of soil N and C pools, so relationships with N₂O emissions should be interpreted cautiously. Future studies should adopt a dynamic soil sampling scheme synchronized with N₂O flux measurements to better resolve the temporal coupling between soil N/C availability and N₂O emissions. Second, plant biomass was measured only at the ripening stage and thus represents cumulative growth over the entire season rather than stage-specific growth. The biomass–N₂O relationship should therefore be interpreted at the seasonal scale, reflecting overall photosynthetic carbon allocation belowground rather than a direct phase-specific link. Root-derived carbon inputs, such as rhizodeposition and fine-root turnover, vary among growth stages and may contribute to temporal fluctuations in N₂O emissions. However, phase-specific belowground carbon inputs were not quantified, which limits our ability to directly link biomass dynamics to N₂O fluxes at particular developmental stages. Future studies should measure root biomass, root exudates, or belowground carbon allocation at multiple growth stages to better resolve the temporal coupling between plant carbon supply and N₂O emissions.

5. Conclusions

Our results indicated that concurrent elevation of CO₂ and temperature increased both area-scaled and yield-scaled N₂O emissions by 15.3% and 17.6% from rice paddies in a rice–wheat cropping system. This stimulation was mainly associated with enhanced soil N availability, as ECT increased the soil N mineralization rate and NH₄⁺–N content, together with shifts in nitrifying and denitrifying microbial communities. Specifically, ECT increased the (nirK + nirS)/nosZ ratio and altered the alpha diversity and composition of AOA and nirK-/nirS-type denitrifier communities, likely suggesting a greater potential for N₂O production under future climate change. These findings emphasize the urgent need for agronomic practices that can sustain rice production and reduce N₂O emissions from flooded rice agroecosystems.