Exploring Sulfate as an Alternative Electron Acceptor: A Potential Strategy to Mitigate N₂O Emissions in Upland Arable Soils

ABSTRACT

Agricultural activities are a significant source of nitrous oxide (N₂O), accounting for approximately 60% of global emissions, highlighting the urgent need for innovative strategies to mitigate N₂O emissions. Microbes conserve nearly as much energy with nitrate (NO₃ ⁻) as oxygen (O₂) respiration under limited O₂ availability. Thus, microorganisms prioritize NO₃ ⁻, limiting exploration of alternative electron acceptors (EAs) to inhibit N₂O emissions through NO₃ ⁻ respiration in upland arable soils. Current approaches remain insufficient, and the interactions between alternative EA reduction and pathways for N₂O emissions remain poorly understood. This study evaluated oxidized iron, manganese, and sulfate as alternative EAs to reduce N₂O emissions, along with the effects of zero‐valent metals (ZVMs). Metal sulfates (MSs) significantly minimized N₂O emissions by inhibiting denitrification rather than altering nitrification in microcosms, as supported by isotope mapping and inorganic nitrogen concentrations. Among others, putative complete denitrifiers, N₂O reducers, and sulfate reducers were stimulated, whereas ZVMs stimulated N₂O emissions and 16S rRNA gene abundance. Moreover, the abundance of denitrifier‐related genes (nirK, nirS, norB, and nosZ) consistently decreased under MS treatments, while dsrA mRNA abundance significantly increased. Sulfate (SO₄ ²⁻) addition reshaped the soil microbial community by enriching sulfur‐cycling taxa—including sulfate‐reducing and sulfur‐oxidizing bacteria—while suppressing nitrifiers such as Nitrospira, potentially disrupting nitrification–denitrification coupling. Ureibacillus thermosphaerius, harboring genes for denitrification and SO₄ ²⁻ reduction, increased under MS treatment. These shifts likely redirected electron flow toward SO₄ ²⁻ respiration, reducing NO₃ ⁻ utilization and contributing to N₂O mitigation. Field‐based manipulation experiments over 2 years demonstrated the feasibility of MSs in upland arable soils, reducing yield‐scaled N₂O emissions by 21.5% without compromising crop yields. A systematic literature review and meta‐analysis revealed that SO₄ ²⁻ application mitigated N₂O emissions by an average of 9%, with over 70% of observations showing a decreasing trend, underscoring its potential as an effective soil amendment for sustainable agriculture.

This study explored the use of sulfate‐based soil amendments as a new method to mitigate N₂O—a powerful greenhouse gas—from agricultural soils. Laboratory and field experiments showed sulfate treatments significantly lowered N₂O emissions without negatively affecting crop yields, mainly by altering soil microbial processes. These findings suggest sulfate amendments can be a practical and sustainable approach to reducing agriculture's climate impact.

1. Introduction

Agricultural activities are a major source of nitrous oxide (N₂O), contributing approximately 60% of global emissions (FAO ²⁰²¹). As of 2025, agriculture emits an estimated 7.26 Tg N₂O‐N y⁻¹ globally, with 71.5% originating from arable soils treated with synthetic nitrogen fertilizers (NFs) (Reay et al. 2012). The nitrogen (N) cycle in agroecosystems is largely microbially driven and substantially influenced by the application of synthetic NFs, which stimulate microbial pathways responsible for N₂O production (Davidson 2009; Fudjoe et al. ²⁰²³). Fertilized soils can convert up to 1.8% of applied N into N₂O (IPCC ²⁰¹⁹), generated directly through soil‐applied N and indirectly via pathways including nitrate (NO₃ ⁻) leaching, ammonia (NH₃) volatilization, and dissimilatory NO₃ ⁻ reduction to NH₃ (Ferland et al. 2024). Therefore, developing strategies that reduce N₂O emissions while maintaining sufficient N availability for crops is critical to meeting the growing global food demand (Galloway et al. 2021; Mosier and Kroeze ²⁰⁰⁰). Technological advances such as slow‐release fertilizers (urea formaldehyde resin, polypeptide‐bonded urea), controlled‐release fertilizers, and the use of urease and nitrification inhibitors have shown promise in enhancing NF efficiency (Govil et al. 2024; Klimczyk et al. ²⁰²¹; Zhang et al. ²⁰²⁴). In addition, advanced management practices, including split fertilization and deep soil injection, also offer benefits on N efficiency (Dong et al. 2022; Millar et al. ²⁰¹⁰). However, there remains a need for innovative, scalable approaches that are not only agronomically effective but also economically feasible, labor‐efficient, and compatible with existing agricultural machinery.

Microbial‐mediated N processes in upland arable soils, primarily nitrification and denitrification, contribute significantly to global N₂O emissions, accounting for approximately 70% of the total anthropogenic flux (Braker and Conrad 2011; Syakila and Kroeze ²⁰¹¹). Under typical aerobic conditions prevalent in well‐drained upland arable fields, nitrification, driven primarily by ammonia‐oxidizing archaea and bacteria as well as nitrite‐oxidizing bacteria, is generally the dominant N₂O‐producing process during crop cultivation (Hu et al. 2022; Inatomi et al. ²⁰¹⁹). Additionally, Wrage et al. (2004) highlighted nitrifier denitrification as an important contributor to N₂O emissions under these aerobic conditions, further underscoring the complexity of the microbial pathways involved. However, denitrification, which typically requires anaerobic or substantially reduced conditions, can become a dominant pathway during episodic events such as heavy rainfall, irrigation, or temporary soil saturation. Under these transient anaerobic conditions, increased water‐filled pore space (WFPS) significantly promotes denitrification activity, leading to a short‐lived yet substantial peak in N₂O emissions (Ruser et al. 2006; Wang et al. ²⁰²¹). Consequently, despite the overall dominance of nitrification and nitrifier denitrification under common upland conditions, cumulative annual emissions can often be dominated by these episodic denitrification events (Ibraim et al. 2020; Scheer et al. ²⁰²⁰; Zhang et al. ²⁰²²). This duality in microbial contributions to N₂O emissions underscores the need for tailored mitigation strategies, emphasizing interventions that specifically address denitrification dynamics under episodic and field‐relevant anaerobic conditions.

Soil microorganisms utilize terminal electron acceptors (EAs) in a thermodynamically determined sequence: oxygen (O₂), then NO₃ ⁻, followed by ferric iron (Fe³⁺), manganese (Mn⁴⁺), sulfate (SO₄ ²⁻), and carbon dioxide (CO₂) (Meulepas et al. 2010; Sivan et al. ²⁰¹⁶). Among these, Fe³⁺, Mn⁴⁺, and SO₄ ²⁻—which serve as alternative EAs—are commonly included in fertilizers to supply secondary or micronutrients to arable soils (Epstein and Bloom 2005; Hafeez et al. ²⁰²¹). Thus, prior studies have demonstrated that these alternative EAs can effectively reduce methane (CH₄) emissions in rice paddies by competitively suppressing methanogens and generating inhibitory intermediates (Bethke et al. 2011; Klüber and Conrad ¹⁹⁹⁸; Sahrawat ²⁰⁰⁸; Yin et al. ²⁰²⁴). However, because NO₃ ⁻ respiration yields the highest energy and is generally prioritized by microorganisms (Strohm et al. 2007), it has limited the exploration of alternative EAs for mitigating N₂O emissions in upland arable soils. Contrary to traditional expectations, several studies have provided evidence for the simultaneous reduction of multiple terminal EAs, such as SO₄ ²⁻ and Fe³⁺, under varying soil redox conditions (Jakobsen and Postma 1999; Wunder et al. ²⁰²¹), and even concurrent Fe³⁺ reduction and methanogenesis (Schreiber et al. 2004). Field‐based observations further support these findings, demonstrating the concurrent use of multiple terminal EAs in natural soil systems (Canfield et al. 1993; Jørgensen et al. ²⁰¹⁹; Vandieken et al. ²⁰⁰⁶).

Climate change, characterized by increased flooding and elevated irrigation demands due to intensified drought and rising temperatures, is expected to enhance conditions favorable for anaerobic microsites in upland soils. These changes may alter soil redox dynamics and promote the concurrent reductions of multiple terminal EAs (Freeman 2020; Xin et al. ²⁰¹⁶). Under such fluctuating redox conditions, microorganisms may simultaneously activate terminal oxidase and N‐reductase pathways (Bourceau et al. 2023). Nevertheless, the interactions between alternative EA reduction and NO₃ ⁻/NO₂ ⁻ respiration pathways in upland arable soils remain poorly understood and warrant further investigation. Hence, in the heterogeneous environment of upland arable soils, we hypothesize that alternative EAs can function concurrently with NO₃ ⁻ under transient anaerobic conditions. This simultaneous utilization of multiple EAs may alter the dynamics of EA competition and microbial prioritization, potentially reducing microbial N₂O emissions.

To test this hypothesis, this study aimed to evaluate the effectiveness of alternative EAs in mitigating N₂O emissions and to elucidate the underlying microbial and biogeochemical mechanisms. To assess the role of metal redox states (Fe³⁺, Mn⁴⁺, and SO₄ ²⁻), we compared the effects of metal sulfates (MSs) with those of the corresponding zero‐valent metals (ZVMs). We conducted time‐batch microcosm experiments, quantified the expression of functional genes, characterized microbial communities, and measured concentrations of applied alternative EAs, along with isotope mapping. Additionally, 2‐year field experiments were conducted to monitor N₂O fluxes and crop yield, thereby evaluating the practical feasibility of selected alternative EAs. Finally, to broadly assess the potential of viable EA, we performed a global‐scale meta‐analysis based on a systematic literature review.

2. Materials and Methods

2.1. Soil Incubation Experiment

2.1.1. Soil Incubation Preparation

Soil samples were collected from an upland arable field at the experimental farm of Pusan National University in Miryang, Korea (35°30′08″ N 128°43′15″ E). The soil is classified as part of the Bongsan series (Fine loamy, mixed, mesic Typic Hapludults family), with its specific physical and chemical properties detailed in Table S1. The sampling site has been cultivated with maize for the past 6 years, including this study. Soil was collected from the top 15 cm, air‐dried at room temperature, sieved through a 2‐mm mesh to remove debris and stones, and homogenized. Subsequently, 400 g of the prepared soil was placed into each 860 mL glass microcosm jar.

To evaluate the role of metal redox states in N₂O mitigation, we selected MSs (FeSO₄, MnSO₄, and ZnSO₄) as candidate EAs and their corresponding ZVMs (Fe⁰, Mn⁰, and Zn⁰), alongside a no‐treatment control. The inclusion of ZVMs enabled us to isolate the effects of the metal components from those of SO₄ ²⁻ and to examine their potential as alternative EAs. Metal sulfates and ZVM were applied at 0.01% (w/w), and urea [(NH₂)₂CO] was applied at 0.02% (w/w) on a dry‐weight basis to each microcosm jar. Soil moisture was adjusted and maintained at 65% WFPS, and jars were incubated in darkness at 25°C (±1.5) for 21 days. Prior to treatment, soils underwent a 10‐day pre‐incubation for microbial and physicochemical stabilization. For each incubation period (7, 14, and 21 days), 21 jars (3 replicates per treatment) were prepared, resulting in 63 microcosms. Jars remained open except during gas sampling.

The N application rate was established based on the recommended fertilization rate for maize cultivation (186 kg N ha⁻¹) in the study region (National Institute of Agricultural Sciences ²⁰²²). This rate corresponds to approximately 0.015% N on a dry‐weight basis, assuming a bulk density of 1.20 g cm⁻³ and a 10 cm plow layer. To better distinguish treatment effects, we applied a slightly elevated N rate of 0.02%, while maintaining field‐relevant conditions. Unlike well‐defined macronutrients, guidelines for secondary and micronutrient application are less established. Previous studies suggest safe application ranges for secondary nutrients, including SO₄ ²⁻ (50–2500 mg kg⁻¹) and micronutrients, including iron, manganese, and zinc (20–600 mg kg⁻¹) (Becker and Asch 2005; Dhaliwal et al. ²⁰¹⁹; Epstein and Bloom ²⁰⁰⁵). Based on these recommendations, micronutrients were applied at 100 mg kg⁻¹ (0.01%, w/w). Thus, our incubation experiment was conducted with a well‐defined application rate of 0.02% N and 0.01% selected EAs materials, ensuring a thorough and precise observation of their effects.

2.1.2. Gas Sampling and N₂O Measurements

Each jar was sealed with airtight rubber stoppers equipped with gas sampling ports. During incubation, gas samples (15 mL) were collected from jar headspaces at 0, 10, and 20 min daily. Nitrous oxide concentration was measured using a gas chromatograph‐mass spectrometer (GC–MS, QP2020, Shimadzu) equipped with an HP‐PLOT Q column, using helium as carrier gas (flow rate = 4.25 mL min⁻¹). Calibration curves generated using certified N₂O standards were used to determine sample concentrations. N₂O flux (ng N₂O g⁻¹ h⁻¹) was calculated as follows:

$${\mathrm{N}}_2\mathrm{O}\text{flux}\left(\mathrm{ng}{\mathrm{N}}_2\mathrm{O}{\mathrm{g}}^{-1}{\mathrm{h}}^{-1}\right)=\left(\mathrm{\Delta }g/\mathrm{\Delta }t\right)\times d\times \left(273/T\right)\times \left(V/W\right)$$ (1)

where, Δg/Δt represents the rate of change in N₂O concentration inside the jar (g cm⁻³), d is the gas density (g m⁻³) at standard conditions (273 K and 0.101 Mpa), T is incubation temperature (K), V is the jar headspace volume (cm³), and W is the soil weight (400 g). Cumulative N₂O emissions were calculated as follows:

$$\text{Cumulative}{\mathrm{N}}_2\mathrm{O}\text{emissions}\left(\mathrm{\mu g}{\mathrm{g}}^{-1}{\mathrm{day}}^{-1}\right)={\sum }_i^n\left(R_i\times D_i\right)$$ (2)

Cumulative N₂O emissions for the experimental period were calculated by multiplying the mean value of N₂O emissions (N₂O μg g⁻¹ day⁻¹) for each period (R _i ) by its duration (D _i ), then adding these values to the previous cumulative total.

2.1.3. Nucleic Acid Extraction and Quantitative PCR

DNA and RNA were co‐extracted from time‐batch incubated soils, following the method described by Griffiths et al. (2000). To eliminate DNA, the extracted RNA was treated with TURBO DNase (Invitrogen, USA). The absence of DNA was confirmed through PCR (Bio‐Rad, Germany), which targeted the bacterial 16S rRNA gene. The purified RNA was then reverse transcribed to cDNA using LunaScript RT SuperMix (New England BioLabs GmbH, Germany), which includes random hexamer and poly‐dT primers to ensure coverage across the length of the RNA targets and an RNase inhibitor to prevent RNA template degradation. The quantity and quality of DNA and cDNA were measured using a NanoDrop (Allsheng, China) and stored at −20°C.

The abundance of various functional genes in the soil samples was assessed using qPCR on both DNA and cDNA. The targets included bacterial 16S rRNA; amoA and hao genes for nitrifiers; nirK, nirS, norB, and nosZ genes for denitrifiers; and the dsrA gene for sulfate‐reducing bacteria (SRB), serving as a proxy for transcript abundance in the soil. For each soil sample (n = 3), nested quantitative PCR (qPCR) was conducted using a CFX Connect Real‐Time PCR System (Bio‐Rad, Germany), resulting in six replicates per treatment and time point. Each qPCR assay (total volume = 20 μL) comprised 10 μL of 2 × AMPIGENE qPCR Green Mix Lo‐ROX (Enzo Life Sciences, USA), 2.0 μL each of forward and reverse primers (10 μM), 6 μL of DNase‐ and RNase‐free water (Thermo Fisher Scientific, Brunswick, Germany), and 2 μL of diluted template DNA or cDNA. The primers and qPCR conditions for targeting functional genes, along with amplification efficiencies (ranging from 89.4 to 111.0%), are provided in Table S2. Amplicon specificity was confirmed through melt curve analysis and 1.2% agarose gel electrophoresis. For absolute quantification of bacterial 16S rRNA and functional genes, plasmids containing the respective target gene sequences were constructed via PCR amplification and TA cloning using each primer set. Standard curves were generated from 10‐fold serial dilutions, and copy numbers were calculated as follows (Whelan et al. 2003).

2.1.4. Soil Bacterial Sequencing and Bioinformatics Analysis

PCR amplification was conducted to construct sequencing libraries targeting the V3 to V4 regions of the 16S rRNA gene. The target region was amplified using the fusion primers 341F (5′‐AATGATACGGCGACCACCGAGATCTACAC‐XXXXXXXX‐TCGTCGGCAGCGTC‐AGATGTGTATAAGAGACAG‐CCTACGGGNGGCWGCAG‐3′) and 805R (5′‐CAAGCAGAAGACGGCATACGAGAT‐XXXXXXXX‐GTCTCGTGGGCTCGG‐AGATGTGTATAAGAGACAG‐GACTACHVGGGTATCTAATCC‐3′). The order of primer components is P5 (P7) graft binding, i5 (i7) index, Nextera consensus, adaptor, and target region (underlined sequences). The PCR conditions were as follows: initial denaturation at 95°C for 3 min, 25 cycles at 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s, with a final elongation at 72°C for 5 min. The PCR products were confirmed using 1% agarose gel electrophoresis and visualized using a Gel Doc system (Bio‐Rad, USA). Subsequently, they were cleaned using CleanPCR (CleanNA, Waddinxveen, Netherlands), pooled in equal concentrations, and assessed for quality and size on a Bioanalyzer 2100 (Agilent, USA) using a DNA 7500 chip. Paired‐end sequencing (2 × 250 bp) was conducted by CJ Bioscience Inc. (CJ, Korea) on the Illumina MiSeq (Illumina, USA), following the manufacturer's protocols. The raw sequencing data were processed using Quantitative Insights into Microbial Ecology 2 (QIIME2, Version 2022.08) (Bolyen et al. 2019). Primer sequences were trimmed using the FASTX Toolkit (Version 0.0.14) (Liu et al. 2019), and demultiplexed sequences were merged using FLASH2 (Magoč and Salzberg 2011). The DADA2 plugin was used to denoise low‐quality reads (Q score < 25) and remove chimeric sequences. Overall, 545,414 reads were obtained from 16S rRNA gene amplicon sequencing, averaging 60,602 ± 6743 reads per soil sample (Table S3). After filtering low‐quality and chimeric reads, 372,414 reads were retained, averaging 41,379 ± 5,000 reads per soil sample. Good's coverage was above 99.5% in the soil samples (data not shown). Amplicon sequence variants (ASVs) were classified taxonomically using the SILVA 16S rRNA gene database (Version SSU138.1) (Quast et al. 2012), excluding unassigned ASVs, mitochondria, and chloroplast sequences. Alpha diversity metrics were calculated from rarefied tables with 31,118 ASVs per sample. Principal coordinate analysis (PCoA) was conducted based on unweighted and weighted UniFrac distances to assess dissimilarities in microbial communities. Shared and unique microbial taxa were visualized using the Venn diagram package in R (Version 4.2.1). The 16S rRNA gene sequencing data are available in the NCBI Sequence Read Archive (SRA) under BioProject No. PRJNA1107391 and SRA accession number SRX24441768‐76.

2.1.5. Edaphic Characteristics

Extractable inorganic N ions (NH₄ ⁺ and NO₃ ⁻) and SO₄ ²⁻ concentrations were quantified at incubation intervals (7, 14, and 21 days). For each time point, 5 g of wet soil was collected from each treatment, with a total of six replicates per treatment, including two technical replicates. Soil samples were extracted using 30 mL of 1 M KCl, a standard method for extracting plant‐ and microbial‐available forms of NO₃ ⁻ and SO₄ ²⁻, and then filtered through a 0.45 μm filter (Bloem et al. 2002; Kodithuwakku et al. ²⁰²⁴). Concentrations for NH₄ ⁺ and NO₃ ⁻ were analyzed using an autoanalyzer (Seal Analytical, Germany); SO₄ ²⁻ were determined by ion chromatograph (Model ICS‐2000, Dionex Corporation, USA), and Fe²⁺ and Mn²⁺ were quantified via inductively coupled plasma mass spectrometer (Thermo Scientific iCAP Q, Germany). Consumption of extractable NO₃ ⁻ and SO₄ ²⁻ was calculated as the concentration difference between incubation for 7 days and 21 days.

2.1.6. Isotope Mapping

To measure stable N₂O isotopes (δ ¹⁵N^bulk, δ ¹⁸O, and site preference [SP]), gas samples were collected from the headspace of the incubation jars (approximately 550 mL volume) on day 21. For each treatment, 15 mL of gas was extracted from each of the three replicates and pooled into 30 mL glass flasks (Pyrex, with stopcock). The pooled gas, after the 20‐min closure, was transferred using a vacuum line maintained with a turbo pump (~0.1 mTorr). Isotopic data were later corrected for ambient mixing effects based on mass balance equations. The isotopomer ratios of N₂O were analyzed using a continuous‐flow isotope ratio monitoring mass spectrometer system (MAT252, Thermo Scientific, Burladingen, Germany) at the Tokyo Institute of Technology, as detailed by Toyoda et al. (2021). The measurement precision was ~0.2‰ for δ ¹⁵N^bulk, ~0.3‰ for δ ¹⁸O, and ~0.5‰ for δ ¹⁵ N ^α and δ ¹⁵ N ^β . The equations for expressing the isotopomer ratios are described below (Toyoda et al. 2008).

(3)

(4)

(5)

The N₂O isotopic composition measured from the mixture of soil‐emitted air (δ_obs) and its concentration (C_obs) was corrected for atmospheric background mixing using ambient air data (δ_ambient and C_ambient), which were estimated based on the 2009 average isotopomer ratios of N₂O in the Northern Hemisphere, as reported in Toyoda et al. (2013). The correction was performed using the following mass balance equation (Toyoda et al. 2024):

$${\mathrm{\delta }}_{\mathrm{obs}}\times {\mathrm{C}}_{\mathrm{obs}}={\mathrm{\delta }}_{\text{ambient}}\times {\mathrm{C}}_{\text{ambient}}+{\mathrm{\delta }}_{\text{soil}}\times {\mathrm{C}}_{\text{soil}}$$

From this equation, δ_soil was derived to isolate the isotopic signature of N₂O emitted solely from the soil. As a result of this correction, δ¹⁵N^bulk decreased by approximately ~3‰; δ¹⁸O values decreased by ~0.5‰, and SP decreased by ~1‰ on average, depending on the treatment.

2.2. Field‐Based Manipulation Experiment

2.2.1. Field Experiment Design

This experiment evaluated the field‐scale feasibility of applying MSs to mitigate N₂O emissions. The experiment was established in the same field, where soils for the incubation experiment were collected (Table S1) in a randomized complete block design with four replicates per treatment. Each plot measured 5 m × 5 m. Maize ( Zea mays L.) was transplanted on July 25, 2018, and May 11, 2019, with a planting density of approximately 67,000 plants per hectare, based on an in‐row spacing of 25 cm and an inter‐row spacing of 60 cm. An annual fertilization regimen of 186–35–74 kg ha⁻¹ (N‐P₂O₅‐K₂O) was followed according to national recommendations (National Institute of Agricultural Sciences ²⁰²²). Fertilizer applications were split into two doses: half applied as basal fertilizer 10 days after transplanting and the other half at the seven‐ or eight‐leaf stage. Urea (46% N), potassium chloride (50% K₂O), and superphosphate (12% P₂O₅) were used as fertilizer sources. Maize was harvested on October 4, 2018, and July 23, 2019.

Climate parameters, including air temperature, precipitation, irrigation events, and WFPS, were monitored throughout the 2‐year cultivation period (Figure S1). Weather data were obtained from a weather station located 1 km from the study site (Korea Meteorological Administration ²⁰²⁵). Supplemental irrigation was applied when rainfall was insufficient for maize establishment and growth. Daily volumetric soil moisture (θ, m³ m⁻³) at a 5 cm depth was measured using a soil moisture sensor (WT1000B, RF sensor, Korea) installed in each plot. Daily WFPS (%) was calculated as follows:

$$\text{WFPS}=\left(\theta /\text{soil porosity}\right)\times 100$$ (6)

Soil porosity (m³ m⁻³) was estimated from monthly bulk density measurements, assuming a particle density of 2.65 g cm⁻³.

The MSs were applied at 0 and 20 kg ha⁻¹ rates, with the applications evenly broadcast near maize plants on May 16, 2018, and May 5, 2019. Although standard application methods for secondary and micronutrients are not well defined, the American Plant Food Control Officials (AAPFCO) recommends a minimum micronutrient rate of 5–10 kg ha⁻¹, including Fe, Mn, and Zn, with higher rates often necessary for agronomic effectiveness (AAPFCO ²⁰¹²). Similarly, sulfate application rates of 20–40 kg ha⁻¹ or greater are advised where deficiencies exist (Amissah et al. 2024). No phytotoxic effects have been reported at these levels (Gupta and Gupta 1998), indicating that the applied rate was both effective and environmentally safe. Thus, a 20 kg ha⁻¹ rate was selected as a balanced, agronomically appropriate, and environmentally sustainable application rate for this experiment.

2.2.2. N₂O Sampling, Measurements, and Calculation

N₂O emissions were measured using the static closed‐chamber method, which was modified and validated by our research group and officially adopted into national guidelines (Choi et al. 2023). PVC anchors equipped with removable lids were permanently installed in the field. Each chamber lid was placed over the chamber during gas sampling events. Gas samples (30 mL each) were collected from the chamber headspace at 0, 20, and 40 min after lid placement, typically between 10:00 a.m. and 12:00 p.m. Sampling was conducted once per week throughout the 2‐year period. To better capture N₂O flux dynamics during periods of high emission potential, such as after fertilizer and treatment applications, irrigation, rainfall, and harvest, the sampling frequency was increased to twice per week. Samples were immediately transferred into 12 mL evacuated glass vials (Vial‐evacuated 838 W, Labco, UK). N₂O concentrations were analyzed using the same procedure as described in Section 2.1.2. According to Equations (1) and (2), N₂O flux and cumulative N₂O emissions were calculated and adjusted using the closed‐chamber method for field‐based experiments. The equations were modified to reflect the specific dimensions of the field chambers, including a headspace volume of 1500 cm³ and a surface of 0.032 m², based on PVC lids with a diameter of 20.2 cm and a height of 17 cm. The area convection coefficient was set at 10.2 cm⁻². To determine yield‐scaled N₂O emissions (YSNE), the cumulative N₂O emission was divided by the dried maize ear yield (MEY) harvested from each plot as follows:

$$\text{YSNE}\left(\mathrm{kg}{\mathrm{N}}_2\mathrm{O}{\mathrm{Mg}}^{-1}\text{yield}\right)=\frac{\text{Cumulative}{\mathrm{N}}_2\mathrm{O}\text{emission}\left(\mathrm{kg}{\mathrm{N}}_2\mathrm{O}{\mathrm{ha}}^{-1}{\mathrm{yr}}^{-1}\right)}{\text{Dried maize}\mathrm{ear}\text{yield}\left(\mathrm{Mg}{\mathrm{ha}}^{-1}{\mathrm{yr}}^{-1}\right)}$$ (7)

2.3. Literature Review and Data Collection

We undertook a thorough systematic literature review, adhering strictly to PRISMA guidelines to maintain high methodological rigor, transparency, and reproducibility standards. By utilizing the keywords ‘(Sulfate) AND (Nitrous Oxide) AND (Soils)’, we explored relevant publications in the Web of Science database. A total of 327 records were initially identified (search conducted on March 1, 2025). Our selection criteria were rigorous: (1) exclusion of reviews, meta‐analyses, or model‐based studies; (2) inclusion of only field or controlled soil incubation studies that reported cumulative N₂O emissions; and (3) exclusion of experiments involving non‐upland or non‐agricultural soils, such as rice paddies and wetlands. We also excluded SO₄ ²⁻ typically applied with N fertilizers, such as ammonium sulfate [(NH₄)₂SO₄], since their co‐application with N can increase N₂O emissions, making it challenging to isolate sulfate's role as an alternative EA. Exceptionally, we included studies using (NH₄)₂SO₄ as a SO₄ ²⁻ source only when they offered a control treatment with an equivalent N amount without SO₄ ²⁻. This approach enabled us to assess the effects of SO₄ ²⁻ on N₂O emissions distinctly. We gathered 74 data points from 19 published studies and our field experiment. Key information extracted from the selected studies included the type of SO₄ ²⁻ material, the SO₄ ²⁻ application rate, the crop type, and the cumulative N₂O emissions for both treatment and control groups. Figure data were digitized using WebPlotDigitizer (Version 4.6, Pacifica, USA). We compared the response ratio of N₂O emissions from sulfate‐treated soils and control soils without SO₄ ²⁻ application, under otherwise comparable conditions. In cases where multiple controls were available (e.g., unfertilized vs. conventionally fertilized), we selected the control group that received the same nitrogen fertilization as the sulfate‐treated group.

2.4. Statistical Analysis

Statistical analyses were conducted using R Studio (Version 3.4.4, R Core Team, Austria) with the “Agricolae” package. Treatment effects on N₂O flux, cumulative N₂O emissions, MEY, YSNE, exchangeable NH₄ ⁺, NO₃ ⁻, and SO₄ ²⁻ concentrations, and the copy numbers of functional genes (amoA, hao, nirK, nirS, norB, nosZ, and dsrA) were evaluated using one‐way ANOVA. Pairwise treatment comparisons were performed using the least significant difference (LSD) test, applied only when the overall F‐test indicated significant differences (p < 0.05). In the global‐scale meta‐analysis based on a systematic literature review, the 95% confidence interval (CI) of the response ratio for cumulative N₂O emissions was calculated using a one‐sample t‐test, which assessed whether the mean response ratio significantly differed from zero.

Data from 16S rRNA sequencing did not conform to the normal distribution, even after several transformations, including log, square root, and arcsine. Therefore, nonparametric analyses were employed. For alpha diversity measurements, the Kruskal–Wallis rank sum test was used to identify significant differences among treatments (control, ZVM, and MS) using the kruskal.test function in the stats package in R. If a significant difference was observed in the Kruskal–Wallis rank sum test, Dunn's post hoc test with the Benjamini–Hochberg adjustment was conducted using the dunnTest function in the FSA package (Ogle et al. 2023). Permutational multivariate analysis of variance (PERMANOVA) and permutational multivariate analysis of dispersion (PERMDISP) with 9999 random permutations were conducted in QIIME2 to compare differences in soil microbiota among treatments and within‐group variances. Analysis of Compositions of Microbiomes with Bias Correction 2 (ANCOM‐bc2) was used to identify differentially predominant phyla and genera (Lin and Peddada 2020). Spearman correlation coefficients between soil properties and differentially abundant microbial taxa identified from ANCOM‐bc2 were computed using the microbiome package (Lahti et al. 2014) and visualized in R Studio. Spearman correlation analysis was conducted to identify potential links between differentially abundant genera and soil properties using the R package microbiome and visualized using the R package “ggplot2”. Statistical significance was defined as p < 0.05, while values between 0.05 and 0.10 were speculated as statistical trends.

3. Results

3.1. N₂O Fluxes

MS and ZVM treatments significantly influenced N₂O emissions (Figure 1a). ZVM application notably increased cumulative N₂O emissions, whereas MS application effectively minimized them. Specifically, MnSO₄ and ZnSO₄ treatments resulted in the most substantial decrease in N₂O fluxes, closely followed by FeSO₄. After 15 days of incubation, MS treatments achieved a 46%–78% decrease in N₂O flux compared to the control (0.61 μg g⁻¹ h⁻¹, Figure 1c). Conversely, ZVM treatments significantly elevated N₂O fluxes, ranging from 1.35 to 2.44 μg g⁻¹ h⁻¹ (Figure 1d). This pattern aligned with the cumulative N₂O emissions observed over time in the microcosm experiment, which showed consistent trends across all treatments without significant temporal variation (Figure S2). This study sought to uncover the mechanisms responsible for reducing N₂O emissions. To achieve this, we focused our further analyses—such as inorganic N concentrations, qPCR‐based functional gene abundances, and differential abundance testing—only on the MS treatments, which consistently displayed notable N₂O mitigation effects. In stark contrast, the ZVM treatments caused a significant spike in N₂O emissions, leading to their exclusion from subsequent analyses to identify effective mitigation pathways.

FIGURE 1.

Comparative effect of metal sulfates and zero‐valent metals (ZVMs) in soil microcosm experiments. (a) The mean value of cumulative N₂O emissions, (b) bacterial 16S rRNA gene abundance across treatments including control, metal sulfates, and ZVMs, (c) N₂O flux comparison for metal sulfates (control, FeSO₄, MnSO₄, and ZnSO₄), and (d) N₂O flux for ZVMs (control, Fe, Mn, and Zn) after 21 days of incubation. Statistically significant differences among the treatment groups based on one‐way ANOVA by the least significant difference test (p < 0.05) are indicated by differing letters.

3.2. Isotope Mapping

To investigate the pathways of N₂O production in the studied soils, we analyzed the stable isotopic composition of N₂O collected from soil air samples and employed dual isotope mapping, referencing established isotopic signatures from Yu et al. (2020) (Figure 2). The experiment consisted of three treatments: control, MS, and ZVM. Each isotopic data point represents the average value derived from three mixture combinations within each treatment group. The control treatment exhibited δ¹⁵N^bulk, δ¹⁸O, and SP values of 3.8‰ ± 0.4‰, 43.8‰ ± 0.3‰, and 12.8‰ ± 2.8‰, respectively. In comparison, the MS treatment showed δ¹⁵N^bulk, δ¹⁸O, and SP values of −2.7‰ ± 3.1‰, 42.8‰ ± 0.6‰, and 20.3‰ ± 2.8‰, respectively. The ZVM treatment exhibited δ¹⁵N^bulk, δ¹⁸O, and SP values of 1.1‰ ± 1.8‰, 42.8‰ ± 0.5‰, and 16.8‰ ± 0.4‰, respectively. These isotopic patterns suggest that heterotrophic denitrification was more prominent in the control, consistent with its lower SP values overlapping the bacterial denitrification (bD) domain (Figure 2a,b). Both MS and ZVM treatments exhibited decreased δ¹⁵N^bulk relative to the control, but the extent of depletion was more pronounced in the MS. Additionally, the MS treatment displayed distinctly elevated SP values compared to both control and ZVM treatments, suggesting a shift towards nitrification or suppression of N₂O reduction processes. In contrast, ZVM samples exhibited only minor differences from the control in SP and δ¹⁵N^bulk, suggesting limited changes in microbial N₂O production or reduction pathways. Across all treatments, δ¹⁸O values remained relatively stable, indicating that oxygen isotope composition was less sensitive to the treatments applied.

FIGURE 2.

Dual isotope mapping of N₂O to identify dominant microbial pathways of production across treatments. (a) Isotope map using site preference (SP) and δ¹⁵N^bulk and (b) isotope map using SP and δ¹⁸O. Each point represents the average isotopic value of three mixture combinations per treatment. Domains for nitrification (Ni), bacterial denitrification (bD), nitrifier denitrification (nD), and fungal denitrification (fD) are delineated based on end‐member values from Yu et al. (2020).

3.3. Ammonium, Nitrate, and Sulfate Concentrations

Ammonium and NO₃ ⁻, transformed from urea, are crucial substrates for N₂O production via nitrification and denitrification in soil. After 21 days of incubation, exchangeable NH₄ ⁺ concentrations did not differ significantly between the MS‐treated soils and the control (Figure 3a). In contrast, exchangeable NO₃ ⁻ concentrations were significantly higher in soils treated with MSs compared to the control, reaching 41.5, 46.4, and 45.0 mg kg⁻¹ for FeSO₄, MnSO₄, and ZnSO₄, respectively, versus 32.3 mg kg⁻¹ in control (p < 0.05, Figure 3b). Correspondingly, NO₃ ⁻ consumption was significantly minimized in the MS‐treated soils, with consumption rates of 35.6, 32.5, and 33.8 mg kg⁻¹ for FeSO₄, MnSO₄, and ZnSO₄, respectively, compared to 43.1 mg kg⁻¹ in control (p < 0.05, Figure 3c). Sulfate consumption was also significantly greater in MS‐treated soils (Figure 3d). The control exhibited the lowest SO₄ ²⁻ consumption at 8.03 mg kg⁻¹, whereas FeSO₄, MnSO₄, and ZnSO₄ application showed substantially higher SO₄ ²⁻ consumption levels of 63.5, 67.8, and 69.1 mg kg⁻¹, respectively (p < 0.05). These results indicate that MS application effectively enhanced SO₄ ²⁻ reduction. Conversely, decreases in exchangeable Fe and Mn concentrations in FeSO₄ and MnSO₄ applications were relatively minor (13.5 and 14.3 mg kg⁻¹, respectively) and statistically insignificant (p = 0.32 and 0.41, respectively; data not shown).

FIGURE 3.

Changes in soil ion concentration affected by metal sulfates. Extractable (a) NH₄ ⁺ and (b) NO₃ ⁻ concentrations in soil with metal sulfates (FeSO₄, MnSO₄, ZnSO₄, and control) after a 21‐day incubation period. Net consumption of (c) NO₃ ⁻ and (d) SO₄ ²⁻ in soils with metal sulfate treatments by determining the difference in each concentration between the 21‐ and 7‐day incubation periods. Statistically significant differences among the treatment groups based on one‐way ANOVA by the least significant difference test (p < 0.05) are indicated by differing letters.

3.4. Functional Gene Expressions

The absolute abundance of bacterial 16S rRNA genes across treatments reflected the observed trends in cumulative N₂O emissions (Figure 1a,b), which aligns with previous studies that identified bacterial activity as a key determinant of N₂O emissions (Butterbach‐Bahl et al. ²⁰¹³; Decock and Six ²⁰¹³; Philippot et al. ²⁰¹¹). Specifically, the ZVM treatments significantly increased bacterial 16S rRNA gene abundance compared to the control, whereas MS treatments exhibited no notable change (p < 0.01, Figure 1b). The abundance of amoA, hao, nirK, nirS, and norB mRNAs in MS treatments remained largely unchanged (not significant, Figure 4a–e). Nevertheless, a consistent decreasing trend in nirK, nirS, and norB gene expression was noted, indicating moderate inhibition of denitrification activity under MS treatments. Additionally, the abundance of nosZ mRNA under MS treatments (ranging between 1.07 × 10⁴ and 1.58 × 10⁴ copies per gram of dry soil) was significantly lower compared to the control (5.75 × 10⁴) (p < 0.05, Figure 4f), suggesting a potent inhibition of complete denitrification.

FIGURE 4.

Impact of metal sulfates on functional gene expression in the nitrogen cycle. These figures illustrate the transcript abundance of (a) bacterial amoA, (b) hao, (c) nirK, (d) nirS, (e) norB, and (f) nosZ genes per gram of dry soil treated with metal sulfates after a 21‐day incubation period. Each box represents the mean copy numbers for treatments with control, FeSO₄, MnSO₄, and ZnSO₄. The thick central line represents the median value, the boxed areas represent the interquartile range, and the spots show the maximum and minimum values (N = 6). Statistically significant differences among the treatment groups, determined by one‐way ANOVA with the least significant difference test (p < 0.05), are indicated by differing letters.

3.5. Microbial Communities

Across all alpha diversity measurements, the alpha rarefaction curves plateaued as sequencing depth increased, indicating sufficient sequencing coverage for all samples (Figure S3). No significant differences were observed in all alpha diversity measurements (observed ASVs, Chao1, Evenness, Faith's phylogenetic diversity, Shannon index, and Simpson index) among treatment groups (Figure 5, p > 0.10). Although overall differences in microbial community structures were detected among treatment groups (PERMANOVA: p = 0.053 for unweighted UniFrac; p = 0.044 for weighted UniFrac) and within‐group dispersion showed significance in unweighted distances (PERMDISP: p = 0.025), post hoc pairwise comparisons did not reveal statistically significant differences between any two treatment groups (Q > 0.05) (Figure S4). Venn diagram analysis revealed that 69.1% of phyla and 50.1% of genera were shared among the treatment groups (Figure 6a,b). At the phylum level, all microbial taxa detected in the ZVM group overlapped with those found in the control and/or MS groups (Figure 6a), whereas the ZVM group exhibited the highest number of unique genera (Figure 6b). The microbial taxa detected in at least 30% of occurrences and with a relative abundance of at least 0.1% among the soil samples were defined as core microbial taxa. The main phyla were primarily Proteobacteria, Actinobacteriota, and Firmicutes, accounting for an average of 60% of total sequences (Figure 6c). At the genus level, UCG_Gemmatimonadaceae, Sphingomonas, and Bacillus were the most abundant genera in amplicon libraries, constituting 11.8% of the total soil microbiota sequences across the samples (Figure 6d). Additionally, 18 genera at an average of 1% in relative abundance were assigned to seven different phyla: Gemmatimonadota, Proteobacteria, Firmicutes, Bacteroidota, Actinobacteriota, Chloroflexi, and Acidobacteriota, which indicates substantial variations at the genus level despite the similarities observed at the phylum level.

FIGURE 5.

Comparison of microbial community structures influenced by treatment groups. Treatment abbreviations: CON (control), MS (metal sulfate), and ZVM (zero‐valent metal). Principal coordinates analysis based on (a) unweighted and (b) weighted UniFrac distances of soil microbiota. Permutational multivariate analysis of variance (PERMANOVA) was conducted to compare differences in soil microbiota among treatments. Boxplot of homogeneity of group dispersions based on (c) unweighted and (d) weighted UniFrac distances. Permutational multivariate analysis of dispersion (PERMDISP) was used to compare within‐group variances.

FIGURE 6.

Distribution of core and distinct microbial taxa across treatment groups at phylum and genus levels. Treatment abbreviations: CON (control), MS (metal sulfate), and ZVM (zero‐valent metal). Venn diagram depicting shared and unique microbiota at (a) phylum and (b) genus levels. Predominant soil microbiota at the (c) phylum and (d) genus levels. Only taxa showing ≥ 30% prevalence and 1% relative abundance were visualized.

Through ANCOM‐bc2 analysis, we identified 24 differentially abundant genera between the control and MS treatments (Figure 7a and Table S4). In the control group, 9 genera were enriched, including unclassified genus (UG) Micrococcaeceae, Gitt‐GS‐136, Skermanella, Nocardioides, JG30‐KF‐CM66, AKAU4049, Nitrospira, uncultured genus (UCG) Elsterales, and Anaeromyxobacter. Interestingly, 7 out of 15 differentially abundant genera in the MS group belonged to the phylum Proteobacteria, i.e., Luteimonas, UG_Sphingomonadaceae, Afipia, B1‐7BS, Herminiimonas, Magnetospirillaceae, and UCG_Rhodospirillales. To further understand the ecological implications of these shifts, Spearman correlation analysis was conducted between the differentially abundant genera and soil physicochemical properties and N‐cycle functional genes (Figure 7b). Dominant community members in MS‐treated soils, such as Gemmatimonadaceae, Ureibacillus, and Pricia, were negatively correlated with N₂O. Herminiimonas, Cyclobacteriaceae, and Rhodospirillales, which exhibited increased relative abundance in MS treatments, also showed a negative correlation with nosZ.

FIGURE 7.

Differences in major soil microbiota by metal sulfate addition. Treatment abbreviations: CON (control) and MS (metal sulfate). (a) Differentially abundant genera identified from the Analysis of Compositions of Microbiomes with Bias Correction 2 (ANCOM‐bc2). Only genera detected with at least ≥ 30% prevalence and 0.1% relative abundance across the treatments were evaluated. Data are represented by log fold change ±95% confidence interval. Red or blue symbols represent significant differences or statistical tendencies, respectively. (b) Spearman correlation analysis between differentially abundant genera, functional genes, and soil properties. Only those with significant and strong coefficients ($\left|\rho \right|$ > 0.7, p < 0.05) were visualized using asterisks: *p ≤ 0.05; **p ≤ 0.01, ***p ≤ 0.001. UG_, unclassified genus; UCG_, uncultured genus.

3.6. N₂O Fluxes and Crops Yield Under Field Conditions

In field‐based manipulation experiments, daily N₂O fluxes with all MS treatments exhibited consistent patterns aligned with agricultural practices across 2 years (Figure S5). MS treatments significantly influenced cumulative N₂O emissions (p < 0.05, Table 1). Average cumulative N₂O emissions varied between the two experimental years, with 4.81 kg ha⁻¹ in 2018 (Y1) and 3.76 kg ha⁻¹ in 2019 (Y2). Over 2 years, the control exhibited the highest cumulative N₂O emissions, followed by MnSO₄, FeSO₄, and ZnSO₄ treatments. MS application consistently decreased N₂O emissions relative to the control, achieving reductions ranging from 9.6% to 24.3%. Notably, decreases in N₂O emissions associated with FeSO₄ and ZnSO₄ were statistically significant (p < 0.05), whereas MnSO₄ did not achieve statistical significance. The 95% CIs for cumulative N₂O emissions were 4.64–5.00 kg ha⁻¹ for the control and 3.95–4.79, 3.85–4.31, and 3.39–4.30 kg ha⁻¹ for MnSO₄, FeSO₄, and ZnSO₄, respectively.

TABLE 1.

Feasibility of metal sulfates as a soil amendment in field‐based manipulation experiments.

Metal sulfates (MSs)	Cumulative N2O emissions (kg ha−1)			Maize ear yield (Mg ha−1)			Yield‐scaled N2O emissions (kg Mg−1)
	Y1	Y2	MS mean	Y1	Y2	MS mean	Y1	Y2	MS mean
Control	5.52a	4.12a	4.82a	6.81b	5.32a	6.07a	0.81a	0.77a	0.79a
Iron sulfate	4.55b	3.55b	4.05b	6.74b	5.53a	6.14a	0.67b	0.64b	0.66b
Manganese sulfate	4.99b	3.75b	4.37 ab	7.28a	4.93a	6.11a	0.68b	0.75ab	0.72b
Zinc sulfate	4.18b	3.60b	3.89 b	7.19ab	5.53a	6.36a	0.58b	0.65b	0.62b
Year mean	4.81A	3.76B		7.16A	5.33B		0.69A	0.71A

Note: This table presents cumulative N₂O emissions, maize ear yield, and yield‐scaled N₂O emissions for metal sulfate treatments across two field experiment years, 2018 (Y1) and 2019 (Y2). The “MS mean” column represents the average values for each metal sulfate treatment over the 2 years, while the “Year mean” denotes the average across all treatments for each year. Statistical differences between treatments within a year are indicated by lowercase letters, and differences between the 2 years by uppercase letters, based on one‐way ANOVA by the least significant difference test (p < 0.05).

The mean MEY over the 2 years did not differ significantly between the control and MS treatments (Table 1). The average MEY was 7.16 Mg ha⁻¹ in Y1 and 5.33 Mg ha⁻¹ in Y2, ranging from 4.93 to 7.28 Mg ha⁻¹. These values fall within the midrange of the MEYs reported by researchers in similar regions using the same NF rates and cropping systems in upland arable soils, which have an average of 5.82 ± 2.08 Mg ha⁻¹ (Huang et al. 2017; Park et al. ²⁰²²). Over the 2 years, YSNE in MS treatments was significantly lower than in the control (p < 0.05, Table 1). While the control had an average YSNE of 0.79 kg Mg⁻¹, MS treatments resulted in significantly lower YSNE values, ranging from 0.62 to 0.72 kg Mg⁻¹, indicating a minimum of 7.6%–21.5%. This decrease in N₂O emissions and a slight increase in MEY contributed to the significantly lower YSNE in the MS treatments than the control. Nitrogen uptake by crops (grain, stem, and leaf) was also measured over the 2‐year experimental period; however, no statistically significant differences were observed among the treatments (Table S5).

4. Discussion

4.1. Contrasting Effects of Electron Acceptors and Donors on N₂O Emissions

We monitored N₂O emissions in a time‐batch incubation experiment to assess how MS and ZVM applications influenced N₂O fluxes in upland arable soils. Our findings revealed clear contrasts between treatments. Specifically, ZVM applications significantly increased N₂O emissions (Figure 1a). The typical oxidation reactions of ZVM in soils can be summarized as follows:

$${\mathrm{Fe}}^0\to {\mathrm{Fe}}^{2+}+{\text{2e}}^{-}\to {\mathrm{Fe}}^{3+}+{\mathrm{e}}^{-}$$ (9)

$${\mathrm{Mn}}^0\to {\mathrm{Mn}}^{2+}+{\text{2e}}^{-}\to {\mathrm{Mn}}^{4+}+{\text{2e}}^{-}$$ (10)

$${\mathrm{Zn}}^0\to {\mathrm{Zn}}^{2+}+{\text{2e}}^{-}$$ (11)

All ZVMs acted as electron donors and stimulated N₂O emissions according to the number of electrons released during their oxidation. Mn⁰ exhibited the highest N₂O flux, followed by Fe⁰ and Zn⁰, which correlated with the number of electrons each metal can release during oxidation into the soil (Figure S1). These findings suggest that the electrochemical properties of these metals significantly influenced their effects on N₂O emissions. Additionally, Fe, Mn, and Zn in soil may stimulate N₂O production due to their role as cofactors in microbial enzymes associated with the N cycle. Iron plays a crucial role in electron transport chains essential for microbial transformations, including reducing NO₃ ⁻ in denitrification processes (Xin et al. 2016). Manganese is essential for superoxide dismutase in oxidative stress response and Zn²⁺ for metalloproteases (Su et al. 2016). However, such metal ions rarely limit microbial activity in soils, and N‐cycle processes are not more demanding than other microbial processes. Thus, data suggest that ZVM stimulated N₂O emissions due to the alleviation of the electron donor limitation of denitrification.

Applying MSs emerged as an effective strategy to reduce N₂O emissions in upland arable soils, with ZnSO₄ exhibiting the greatest mitigation effect (Figure 1c). Zinc sulfate was deliberate, considering that Zn²⁺ is not typically an EA and its molecular weight is comparable to Fe and Mn, to isolate and examine the effects of SO₄ ²⁻ alone. We initially anticipated FeSO₄ and MnSO₄, with their synergistic combination of alternative EAs from cation and anion, to be more effective in reducing N₂O emissions. This expectation is based on the ability of divalent metal ions (Fe²⁺ and Mn²⁺) to oxidize to Fe³⁺ and Mn⁴⁺ in upland arable soils, offering higher energy yields than SO₄ ²⁻ under oxygen‐limited conditions (Bethke et al. 2011). However, our results indicate that SO₄ ²⁻ was preferentially utilized, reducing N₂O emissions more effectively than treatments including Fe or Mn.

The limited effectiveness of Fe and Mn as EAs in our study is likely attributed to their low bioavailability in upland arable soils. Even if Fe³⁺ and Mn⁴⁺ are present, these oxidized forms quickly become insoluble due to their strong adsorption to soil particles or precipitation as compounds such as ferric hydroxide, goethite, hematite, and manganese dioxide (Hem 1963; Lindsay ^1979b; Schwertmann and Murad ¹⁹⁸³). In contrast, sulfate‐based materials exhibit superior solubility and availability for microbial utilization. Sulfates are highly water‐soluble, allowing them to readily diffuse in soil solution and become accessible to microorganisms, unlike Fe and Mn oxides, which often remain in insoluble forms (Charlet et al. 1993). Another advantage of SO₄ ²⁻ as an EA in upland arable soils is its regenerative nature. While SO₄ ²⁻ reduction results in the formation of S²⁻, periodic aeration can re‐oxidize S²⁻ back to SO₄ ²⁻ (Eimers et al. 2003), maintaining its availability in the soil over time. Moreover, S²⁻ may bind to humic substances and organic matter, potentially influencing redox dynamics in ways that further modulate N₂O emissions (Heitmann and Blodau 2006). Additionally, SO₄ ²⁻ materials are inexpensive and widely available, making them a practical choice for large‐scale field applications (Worthington et al. 2017). These properties make SO₄ ²⁻ a more viable and sustainable candidate for mitigating N₂O emissions in upland arable soils than Fe³⁺ and Mn⁴⁺.

4.2. Mechanism of Reducing N₂O Emissions by MSs

In the heterogeneous environment of upland arable soils, we hypothesize that SO₄ ²⁻ can function as an EA alongside NO₃ ⁻ during anaerobic respiration. This simultaneous utilization of multiple EAs has the potential to minimize N₂O emissions by altering the dynamics of EA competition and prioritization within microbial communities.

4.2.1. Alterations in N‐Cycle Influenced by MS Application

The dual isotope mapping clearly demonstrated distinct shifts in SP values across treatments when plotted against δ¹⁵N^bulk − N₂O and δ¹⁸O‐N₂O (Figure 2). Specifically, the MS treatment exhibited significantly elevated SP values accompanied by pronounced depletion of δ¹⁵N^bulk relative to both the control and ZVM treatments, indicating a clear trend toward nitrification‐driven N₂O production. This isotopic signature strongly suggests a fundamental redirection of electron flow within the soil N‐cycle under MS treatments. Such changes likely occurred due to increased competition for substrate electrons between SO₄ ²⁻ reducers and denitrifying microbes. The electron demand of SO₄ ²⁻ reduction presumably suppressed denitrification activity, thereby reducing the associated N₂O emissions and shifting the overall balance of N₂O production toward nitrification processes.

In contrast, the ZVM treatment showed isotopic signatures (both SP and δ¹⁵N^bulk) that closely overlapped with those of the control soils, despite the observed increase in N₂O emissions. Our initial interpretation, attributing higher N₂O fluxes under ZVM solely to enhanced denitrification based on the result of 16S rRNA gene abundance (Figure 1b), was inconsistent with this isotopic evidence. To reconcile this discrepancy, we revised our interpretation, proposing that the electron‐donating properties of ZVM may have broadly stimulated microbial activities, enhancing both nitrification and denitrification pathways simultaneously. While denitrification is traditionally viewed as electron‐demanding, recent literature highlights that the early stages of nitrification also involve electron‐dependent activation of molecular oxygen. This initial electron requirement leads to the formation of hydroxylamine (Siegbahn 2024; Voss et al. ²⁰²⁴), an important intermediate and direct precursor of N₂O. Thus, in ZVM‐treated soils, electron availability might have stimulated both nitrification and denitrification without strongly biasing the isotopic signatures toward either pathway, resulting in increased total N₂O emissions but minimal changes in SP and δ¹⁵N^bulk values.

Collectively, our results underscore the complexity of interpreting isotopic data when multiple microbial processes coexist and are simultaneously influenced by treatment‐specific changes in electron availability. While MS clearly redirected electron flow toward sulfate reduction, thereby decreasing denitrification and emphasizing nitrification‐related N₂O emissions, ZVM treatments enhanced microbial activity broadly without substantially shifting isotopic signatures. These nuanced observations emphasize that isotopic data interpretation must carefully consider simultaneous microbial processes and electron dynamics to accurately discern mechanisms underlying N₂O production (Amundson et al. 2023).

The results for exchangeable ions supported the isotope mapping findings. MS application inhibited NO₃ ⁻ reduction rather than NH₄ ⁺ oxidation (Figure 3a,b). Notably, NO₃ ⁻ and SO₄ ²⁻ consumption patterns showed inverse relationships (Figure 3c,d). Compared to the control, NO₃ ⁻ consumption was significantly reduced by approximately 10 mg kg⁻¹ with MS application. Conversely, SO₄ ²⁻ reduction in MS treatments averaged 66 mg kg⁻¹, accounting for around 30% of the applied SO₄ ²⁻, indicating SO₄ ²⁻ likely outcompeted NO₃ ⁻ as an EA under anaerobic conditions. Luo et al. (2022) reported decreased NO₃ ⁻ reduction under elevated SO₄ ²⁻ concentrations, with residual NO₃ ⁻ ratios increasing by up to 60%. However, despite the large N input, the observed NO₃ ⁻ consumption was unexpectedly about six times lower than that of SO₄ ²⁻. This discrepancy is likely because NO₃ ⁻ consumption was estimated from the difference between incubation days 21 and 7; potentially missing ongoing nitrification or intermediate N transformations occurring beyond the initial incubation period. Given the inherent complexity and continuous transformations within the soil N cycle, accurately quantifying exact NO₃ ⁻ consumption presents a methodological challenge. Temporal changes in extractable NO₃ ⁻ and NH₄ ⁺ concentrations during incubation are provided in Figure S6.

Our study highlights the importance of further investigating the interactions between NO₃ ⁻ and SO₄ ²⁻ adsorption onto soils, which could significantly influence their bioavailability and microbial utilization, especially depending on soil pH (Essington and Stewart 2018; Zhou et al. ²⁰²⁵). Understanding how differing adsorption capacities of NO₃ ⁻ and SO₄ ²⁻ shape microbial competition and N cycling dynamics can yield valuable insights. Future research should incorporate detailed mineralogical and physicochemical analyses to enhance our knowledge, ultimately helping to optimize SO₄ ²⁻ amendments for effectively mitigating N₂O emissions across diverse soil conditions.

4.2.2. Soil Microbial Quantity and Functional Gene Expression

Results from the functional gene expression analyses related to the N‐cycle (Figure 4) suggest concomitant activity of nitrification and denitrification and a significant decrease in the expression of the nosZ gene, which encodes the enzyme responsible for reducing N₂O to N₂. Such a decrease could potentially elevate N₂O emissions by minimizing the efficiency of N₂O reduction. However, a comprehensive evaluation indicated that NH₃ oxidation remained unaffected. Other denitrification‐related genes (nirK, nirS, and norB) showed decreasing trends following MS application, but these changes were not statistically significant. Therefore, the substantial downregulation of nosZ, coupled with stable nitrification and minor decreases in other denitrification‐related genes, suggests an overall suppression of denitrification activity, resulting in reduced total N₂O production. Consequently, despite the impaired potential for complete N₂O reduction, total N₂O emissions from MS‐treated soils remained lower compared to the control, predominantly due to the reduced overall N₂O production through denitrification.

Furthermore, the abundance of dsrA mRNA was significantly increased under MS treatments, ranging from 8.14 × 10⁴ to 2.64 × 10⁵ copies per gram of dry soil (Figure S7). This increase in dsrA mRNA suggests that SO₄ ²⁻, when added as an EA in MS treatments, stimulated the expression of genes encoding for the bisulfite reductase of SO₄ ²⁻ reducers, thus indicating stimulation of active SO₄ ²⁻ reduction as an anaerobic respiration process, potentially leading to decreased N₂O emissions by outcompeting denitrification. Such observations align with findings from sulfate‐rich environments (e.g., coastal wetlands and riverbed soils) where NO₃–⁻ consumption processes, such as denitrification and dissimilatory ⁻⁻NO3–/NO2 –reduction to NH₄ ⁺+, are inhibited, decreasing N₂O production (Bourceau et al. 2023; Cui et al. ²⁰²⁰). In the metabolic pathways of SOB, SO₄ ²⁻ serves as a terminal EA, being reduced to sulfides (H₂S, HS⁻, or S²⁻). Sulfides can inhibit N₂O reduction, likely due to their toxicity and the formation of copper sulfides that interfere with key microbial enzymes (Pan et al. 2013). However, in well‐aerated arable soils, the large‐scale accumulation of toxic sulfides is unlikely. Under aerobic or fluctuating redox conditions, sulfur‐oxidizing microorganisms can re‐oxidize sulfides back to SO₄ ²⁻. Furthermore, HS–⁻ and S²⁻ can precipitate with divalent metals such as Fe²⁺, forming insoluble FeS (Berkowitz et al. 2019). Although the overall risk of sustained sulfide toxicity is low, transient sulfide production under temporarily oxygen‐limited conditions could still disrupt denitrification by interfering with microbial activity or altering electron flow dynamics.

4.2.3. Soil Microbial Community

Our study aims to identify materials that effectively reduce N₂O emissions from soil. Thus, our microbial community analysis focused on MS treatments, which were effective in decreasing N₂O emissions (Figure 1). Soils are known to harbor diverse microbial communities that play multifunctional roles in nutrient cycling, plant functional stability, and overall plant performance (Lou et al. 2022). In this study, although overall microbial diversity did not differ significantly among the treatments, MS supplementation led to an increased number of unique microbial genera (approximately 10% per treatment), suggesting treatment‐specific shifts in microbial community composition. Notably, a substantial proportion (~60%) of the total detected genera were minor taxa (relative abundances below 1%), highlighting the high complexity of the soil microbiota, in agreement with previous findings (Jiao et al. 2022; Jiao et al. ²⁰¹⁹). Using ANCOM‐bc2 analysis, we identified 24 differentially abundant genera between the control and MS treatments, representing 9 abundant genera in the control and 15 abundant genera in the MS groups (Figure 7a and Table S4). In addition, Spearman correlation analysis revealed significant associations between key differentially abundant genera and various soil parameters and N cycling genes (Figure 7b).

Nitrospira, traditionally a known nitrifier, especially a nitrite‐oxidizing bacterium, was more abundant in the control than in MS (Figure 7a). Findings have identified comammox‐capable species that perform complete ammonia oxidation, expanding their recognized role in the N‐cycle (Daims et al. 2015). While Nitrospira is not directly involved in denitrification or N₂O reduction, its contribution to upstream nitrification may indirectly influence N₂O emissions, particularly in C‐rich microzones where redox fluctuations facilitate tight nitrification–denitrification coupling (Kits et al. 2019). In our study, suppressing Nitrospira under MS treatments could reflect altered redox or substrate conditions unfavorable to nitrification. These shifts may indirectly influence N₂O emissions by disrupting the continuity of nitrification–denitrification coupling. UG_Gemmatimonadaceae, Herminiimonas, and Ureibacillus were enriched in sulfate‐treated soils and showed negative correlations with nosZ (Figure 7b), suggesting potential shifts in redox dynamics and microbial functional activity. Members of the family Gemmatimonadaceae, belonging to the phylum Gemmatimonadota, are widely distributed across diverse environments such as soils, sediments, and marine (Fudjoe et al. 2023; Gong et al. ²⁰²⁴). Recent metagenomic studies have revealed that Gemmatimonadaceae possesses genes related to SO₄ ²⁻ reduction but lacks genetic machinery for denitrification (Mujakić et al. 2023). Similarly, Koh et al. (2017) reported that Herminiimonas arsenitoxidans AS8 harbors genes associated with SO₄ ²⁻ reduction yet does not contain denitrification‐related genes. In contrast, Ureibacillus thermosphaerius strain thermos‐BF, as reported by Juibari et al. (2015), contains both genes of denitrification and SO₄ ²⁻ reduction, highlighting its dual metabolic capabilities.

Metal sulfate treatments specifically enriched various sulfur‐cycling bacteria taxa, including Magnetospirillaceae, UCG_Rhodospirillales, and Afipia (Figure 7a). Previous studies have reported that certain members of Magnetospirillaceae and UCG_Rhodospirillales (both belonging to the order Rhodospirillales) participate in sulfide oxidation and thus function as sulfur‐oxidizing bacteria (SOB) (Koziaeva et al. 2023; Yang et al. ²⁰²³). Afipia is also known for its capacity to oxidize reduced sulfur compounds (Huddy et al. 2021). The observed increase in SRB under MS treatments likely enhanced sulfide production, subsequently stimulating SOB. This increase in SOB activity could potentially suppress denitrification efficiency, leading to a reduction in overall N₂O emissions (Lu et al. 2018). Supporting this, we identified significant positive correlations between specific microbial genera and both SO₄ ²⁻ concentrations and dsrA expression, indicating enhanced SO₄ ²⁻ utilization and dsrA expression following MS supplementation. For instance, AKYG1722, which was notably higher in relative abundance under MS treatments, is known for utilizing SO₄ ²⁻ in N‐depleted environments (González‐Cortés et al. ²⁰²²). Introducing SO₄ ²⁻ into the soil markedly decreased the relative abundance of denitrification‐related microorganisms, such as Elsterales and Anaeromyxobacter (Mhuantong et al. 2015; Zecchin et al. ²⁰²³). This change in the microbial community, particularly in soils with high SO₄ ²⁻ concentrations, likely contributed to decreased NO₃ ⁻ consumption. The relationship between SO₄ ²⁻ and NO₃ ⁻ respiration may be more direct, as many organisms traditionally identified as SRB can also reduce NO₂ ⁻ through denitrification or DNRA (Pan et al. 2013). Some organisms transition to NO₃ ⁻/NO₂ ⁻ respiration when N‐oxides are accessible (Krekeler and Cypionka 1995; Seitz and Cypionka ¹⁹⁸⁶), while others consistently prioritize SO₄ ²⁻ reduction even when oxidized N compounds are present (Dalsgaard and Bak 1994; Marietou ²⁰¹⁶; Marietou et al. ²⁰⁰⁹).

4.3. Feasibility of MSs for Mitigating N₂O Emissions in Upland Arable Soils

The mean MEY over the 2 years did not differ significantly between the control and MS treatments (Table 1). Mahal et al. (2022) noted no increase in maize production at half of the 12 sites studied despite using four S fertilizers, indicating a weak positive relationship between corn productivity and S application. Although MnSO₄ showed an increase in MEY compared to the control in Y1, this does not conclusively indicate that MS can enhance crop productivity. A critical aspect of our field experiment was to evaluate the feasibility of SO₄ ²⁻ materials as soil amendments. Our results demonstrated that applying SO₄ ²⁻ at 20 kg ha⁻¹ of MS did not adversely affect crops. Other studies have shown that a S input of 15–50 kg ha⁻¹ can enhance pest resistance, tissue strength, and crop quality. In maize cultivation fields, YSNE values have been reported to be above 1 kg Mg⁻¹ (Halvorson et al. 2010), but recent global meta‐analyses have indicated a range of 0.18–0.36 kg Mg⁻¹ (Aliyu et al. 2021). YSNE is influenced by factors such as irrigation, temperature, and initial total N concentration (Wang et al. 2021). There is a strong correlation between NF input and YSNE (Kim et al. 2023). Given the importance of food security, unconditionally reducing NF input is challenging. Additionally, crop N uptake is closely linked to sulfur metabolism; our measurements showed no significant differences in total N uptake between the control and MS treatments (Table S5). Thus, enhanced N uptake does not appear to explain the observed mitigation in N₂O emissions in our study. Therefore, SO₄ ²⁻ should be considered a potential fertilizer or soil amendment for its ability to mitigate N₂O emissions in arable soils, balancing environmental concerns with agricultural productivity.

Changes in soil physicochemical properties following MS application can be indirectly assessed through soil pH variations. The applied MS treatments led to a minor and statistically non‐significant decline in soil pH throughout the cultivation period (Figure S8). Although metal ions such as Fe²⁺, Mn²⁺, and Zn²⁺ can release protons (H⁺) during oxidation and contribute to soil acidification (Grundl and Delwiche 1993; Lindsay ^1979a; Menard and Demopoulos ²⁰⁰⁷; Permyakov ²⁰²¹; Schwertmann and Murad ¹⁹⁸³), the relatively low application rates used here resulted in minimal and insignificant changes. Furthermore, intermittent SO₄ ²⁻ reduction and S²⁻ oxidation processes, typical of partially anaerobic microsites in upland arable soils, likely buffered these acidification effects. Despite this limited pH response, even subtle pH changes could influence microbial communities involved in nitrification and denitrification processes, potentially affecting ammonia oxidizer and denitrifier activities (Shaaban et al. 2018; Zhou et al. ²⁰²⁴). Thus, minor soil pH alterations induced by MS applications might partially contribute to the observed N₂O mitigation effects, although this is likely a secondary mechanism.

Furthermore, SO₄ ²⁻ applications have implications beyond N cycling, significantly influencing additional biogeochemical processes in agroecosystems such as the C cycle. Microbial SO₄ ²⁻ reduction may initially stimulate CO₂ emissions due to enhanced organic matter mineralization. However, the subsequent formation of sulfide compounds introduces additional complexity. For example, Pan et al. (2013) demonstrated that sulfide compounds formed through SO₄ ²⁻ reduction could inhibit microbial respiration, potentially decreasing CO₂ emissions from soils. Additionally, Wu et al. (2021) suggested that these sulfide compounds may chemically interact with soil organic matter, influencing its long‐term stability and overall carbon sequestration potential. These complex interactions underscore the multifaceted influence of microbial S cycling on other elemental cycles and highlight the challenge of accurately predicting long‐term impacts on soil C dynamics and greenhouse gas emissions.

The potential for SO₄ ²⁻ reduction in upland arable soils fundamentally differs from conditions typically encountered in marine sediments, wetlands, or rice paddy soils, where prolonged anoxic conditions enable continuous SO₄ ²⁻ reduction. In contrast, upland agricultural soils develop transient anaerobic microsites due to events such as heavy rainfall, intensive irrigation, or specific management practices, creating fluctuating soil redox conditions that alternate frequently between oxidized SO₄ ²⁻ and reduced S²⁻ forms. Nevertheless, our experimental results provide evidence that SO₄ ²⁻ reduction can still transiently occur under specific conditions in upland arable soils. This observation aligns with earlier studies, particularly those by Canfield and Des Marais (1991), who reported dissimilatory SO₄ ²⁻ reduction even in partially oxygenated environments. Such findings challenge the conventional assumption that SO₄ ²⁻ reduction is exclusively an anaerobic process and highlight the necessity for further research to better characterize and understand the specific environmental conditions that permit SO₄ ²⁻ reduction under partially aerobic circumstances.

4.4. Literature Review on the Role of Sulfate in Mitigating N₂O Emissions

Although the findings from this study support our hypothesis, they challenge the conventional understanding of alternative EA usage, which may raise critical questions about generalizing these results. To broaden the scope of this hypothesis, we conducted a systematic literature review (Table S6) and a global meta‐analysis to explore the effects of SO₄ ²⁻ additions on N₂O emissions in cultivated aerobic condition soils, particularly upland arable soils. The response ratio was calculated by analyzing the effects of SO₄ ²⁻ addition relative to control treatments without SO₄ ²⁻ addition. The dataset covered diverse SO₄ ²⁻ application rates, crop types, and soil conditions, strengthening the findings' relevance and applicability across different agricultural scenarios. Our results demonstrate that SO₄ ²⁻ amendment significantly decreases N₂O emissions, with a mean response ratio of −9.08% (SD = 28.4%, N = 76) and a 95% CI ranging from −14.10% to −3.09% (Figure 8). Importantly, as the 95% CI does not include zero, this minimization is statistically significant. Among the 76 individual observations analyzed, 70.3% exhibited negative response ratios, further supporting the effectiveness of SO₄ ²⁻ amendments in mitigating N₂O emissions from upland arable soils. These findings align well with our field and microcosm experiments, reinforcing the potential of SO₄ ²⁻ as a viable alternative EA. Additionally, SO₄ ²⁻ can function as an EA even under micro‐oxic conditions, not requiring fully anoxic environments like Fe³⁺ and CO₂ reduction (Brodersen et al. 2019). The results suggest that SO₄ ²⁻ amendment is an effective and broadly applicable strategy for reducing N₂O emissions from upland arable soils.

FIGURE 8.

Response ratio of N₂O emissions to sulfate addition across globally selected sites. This plot presents data compiled through a PRISMA‐based systematic literature review. Each blue dot (N = 76) represents an individual observation comparing sulfate‐treated soils to conventional management practices. The red boxplot summarizes the distribution of response ratios, with the solid black line indicating the mean (−9.08%) and the red dashed lines showing the 95% confidence interval (CI), which ranges from −14.10% to −3.09%. Since the CI does not include zero, this confirms a statistically significant reduction in N₂O emissions. The accompanying violin plot illustrates the distribution density of the effect sizes.

Our study used MSs, specifically FeSO₄, MnSO₄, and ZnSO₄, to evaluate their potential as alternative EAs. However, the existing literature shows that other sulfate‐based materials, such as CaSO₄ and MgSO₄, are frequently used in similar studies. These SO₄ ²⁻ forms, including by‐products such as phosphogypsum, desulfurization gypsum, and blast furnace slag, are readily available and cost‐effective EAs and provide essential macronutrients (Lee et al. 2023). Additionally, our study demonstrates that applying MS at a moderate rate of 20 kg ha⁻¹ can significantly reduce N₂O emissions. However, a review of existing literature reveals substantial variability in application rates and outcomes reported across different studies, highlighting an urgent need for further exploration. For instance, Gao et al. (2024) found effective N₂O mitigation with a high gypsum application rate of approximately 1.44 Mg ha⁻¹ (converted from 0.72 g kg⁻¹ of soil); whereas Watts et al. (2023) reported no significant effects even with a higher application rate of 2.20 Mg ha⁻¹. On the other hand, much lower SO₄ ²⁻ application rates of 10–27.5 kg ha⁻¹ have also been reported to effectively reduce N₂O emissions in maize cultivation fields (Majumdar et al. 2002; Montoya et al. ²⁰²¹). The inconsistencies in these findings likely result from variations in SO₄ ²⁻ forms, along with differing soil properties, crop management practices, microbial community dynamics, and local climate conditions. Thus, it is imperative that future research comprehensively examines these factors. By doing so, we can gain deeper insights into the complex interactions that determine the effectiveness of SO₄ ²⁻ in mitigating N₂O emissions from arable soils, paving the way for more effective agricultural practices.

5. Summary

Validating SO₄ ²⁻ material as an effective soil amendment for reducing N₂O emissions in upland arable soils hinges on two critical aspects emphasized in this study. First, it is essential to determine whether alternative EAs can be utilized under partially anaerobic conditions created by rainfall and irrigation in upland arable soils. Second, it is crucial to evaluate whether SO₄ ²⁻, despite being a lower‐priority EA compared to NO₃ ⁻, can be used concurrently with NO₃ ⁻.

The results of this study, particularly from isotope mapping, NO₃ ⁻ and SO₄ ²⁻ measurements, and gene expression, confirmed that SO₄ ²⁻ treatments significantly suppressed N₂O production from denitrification and stimulated SO₄ ²⁻ reduction. The results of metataxonomic analysis supported these findings, showing an increase in microbial taxa capable of using both NO₃ ⁻ and SO₄ ²⁻ as EAs in MS‐treated soils. Furthermore, MS treatments enriched sulfur‐cycling microbial taxa and concurrently suppressed nitrifiers. This microbial community shift suggests that the concurrent utilization of NO₃ ⁻ and SO₄ ²⁻ as EAs and the suppression of nitrifiers may have disrupted the coupling between nitrification and denitrification, thereby contributing to the reduction in N₂O emissions. Field‐manipulation experiments further demonstrated that adding practical amounts of MSs to soils effectively reduced YSNE. Additionally, due to greater bioavailability under less stringent redox conditions, SO₄ ²⁻ offers a more practical alternative EA than Fe³⁺ and Mn⁴⁺ in upland arable soils. This unique characteristic of SO₄ ²⁻ reduction opens up a new avenue for research and application, as it can be utilized alongside NO₃ ⁻ under partially anoxic conditions, highlighting the potential of SO₄ ²⁻ materials as a viable and intriguing soil amendment that promotes sustainable agriculture practices while mitigating GHG emissions.

Author Contributions

Hyun Ho Lee: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, supervision, validation, visualization, writing – original draft, writing – review and editing. Hanbeen Kim: data curation, formal analysis, visualization, writing – original draft, writing – review and editing. Ye Lim Park: formal analysis, investigation. Marcus A. Horn: supervision. Jeongeun Kim: formal analysis, investigation, methodology, visualization. Jaehyun Lee: investigation, visualization. Sakae Toyoda: investigation, methodology, visualization. Jeongeun Yun: formal analysis, investigation. Hojeong Kang: methodology, resources. Sang Yoon Kim: formal analysis. Jinho Ahn: formal analysis. Chang Oh Hong: conceptualization, funding acquisition, project administration, supervision.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: gcb70428‐sup‐0001‐supinfo.zip.

Lee, H. H. , Kim H., Park Y. L., et al. 2025. “Exploring Sulfate as an Alternative Electron Acceptor: A Potential Strategy to Mitigate N₂O Emissions in Upland Arable Soils.” Global Change Biology 31, no. 8: e70428. 10.1111/gcb.70428.

Data Availability Statement

The data and code that support the findings of this study are openly available in Zenodo at https://doi.org/10.5281/zenodo.16389568, and in the NCBI Sequence Read Archive (SRA) under BioProject number PRJNA1107391, with SRA accession numbers SRX24441768‐SRX24441776.