Hot moment of N₂O emissions in seasonally frozen peatlands

Abstract

Since the start of the Anthropocene, northern seasonally frozen peatlands have been warming at a rate of 0.6 °C per decade, twice that of the Earth’s average rate, thereby triggering increased nitrogen mineralization with subsequent potentially large losses of nitrous oxide (N₂O) to the atmosphere. Here we provide evidence that seasonally frozen peatlands are important N₂O emission sources in the Northern Hemisphere and the thawing periods are the hot moment of annual N₂O emissions. The flux during the hot moment of thawing in spring was 1.20 ± 0.82 mg N₂O m⁻² d⁻¹, significantly higher than that during the other periods (freezing, −0.12 ± 0.02 mg N₂O m⁻² d⁻¹; frozen, 0.04 ± 0.04 mg N₂O m⁻² d⁻¹; thawed, 0.09 ± 0.01 mg N₂O m⁻² d⁻¹) or observed for other ecosystems at the same latitude in previous studies. The observed emission flux is even higher than those of tropical forests, the World’s largest natural terrestrial N₂O source. Furthermore, based on soil incubation with ¹⁵N and ¹⁸O isotope tracing and differential inhibitors, heterotrophic bacterial and fungal denitrification was revealed as the main source of N₂O in peatland profiles (0–200 cm). Metagenomic, metatranscriptomic, and qPCR assays further revealed that seasonally frozen peatlands have high N₂O emission potential, but thawing significantly stimulates expression of genes encoding N₂O-producing protein complexes (hydroxylamine dehydrogenase (hao) and nitric oxide reductase (nor)), resulting in high N₂O emissions during spring. This hot moment converts seasonally frozen peatlands into an important N₂O emission source when it is otherwise a sink. Extrapolation of our data to all northern peatland areas reveals that the hot moment emissions could amount to approximately 0.17 Tg of N₂O yr⁻¹. However, these N₂O emissions are still not routinely included in Earth system models and global IPCC assessments.

Introduction

Northern peatlands contain about 9–16% and 30% of the global soil-associated total organic nitrogen (N) [1, 2] and carbon (C) [3, 4], respectively. Peatland disturbances can trigger the loss of peat and the release of greenhouse gases and thus exert a significant influence on global climate change and the atmospheric C and N budget [5]. More than half of the peatlands in the Northern Hemisphere are subjected to annual freeze-thaw cycles, including perennially, seasonally, and intermittently frozen regions [6]. These regions are characterized by their soil freezing time, i.e., two years or more, more than 15 days per year, and between one and 15 days a year [6]. At least 38% of the northern peatland area is seasonally frozen [2, 6]. Seasonally frozen peatlands are characterized by unidirectional freezing and bidirectional thawing, which are vulnerable to warming and precipitation [7]. Under the influence of global warming, the active-layer depths of permafrost have continuously increased during recent decades [8, 9]. Degraded permafrost peatlands can be transformed into seasonally frozen peatlands [10], which could have a great impact on the global GHG budget [3, 5].

Nitrous oxide (N₂O) emissions from soils in the Northern Hemisphere have traditionally been considered negligible as these soils were considered to be N limited [5, 11, 12], and plants competed favorably for N with microorganisms [13–15]. However, increasing evidence shows that thawing permafrost can produce and release substantial amounts of carbon dioxide, methane (CH₄), and N₂O [12, 16–19]. Compared with areas that experience thawing only in the active layer of the permafrost, seasonally frozen peatlands experience a more substantial freeze-thaw cycle over the whole soil profile and have a greater redox potential gradient; [3, 20] hence, they are potentially important drivers of N₂O emissions. Yet, both the contribution of this effect to global change and the biogeochemical mechanism of the N cycle in seasonally frozen peatlands have not been well explored.

Seasonally frozen peatlands have warmed at a rate of 0.6 °C per decade over the past 30 years, twice the global average surface temperature [21]. This warming has increased bioavailable N by enhancing soil organic matter decomposition and N mineralization [22, 23]. Soil thawing promotes frozen N release and influences microbial N transformation rates, which could alter soil N availability and regulate ecosystem functions [24–26]. Waterlogged conditions during the thawing process can produce oxygen-limited conditions throughout the soil profile and increase the redox conditions of the soil [27], further promoting microbial N-reduction pathways, such as the heterotrophic denitrification process that generates N₂O, especially at a lower pH [28–30]. Moreover, during the thawing season, there is relatively week competition for bioavailable N between plants and microorganisms [18, 31]. These processes could cause increased N₂O emissions during the thawing of seasonally frozen peatlands. Based on the studies described above [18, 24–27, 31], as well as our previous microbial-N cycle studies in wetland and permafrost ecosystems [12, 26, 32–36], we put forward the hypothesis that seasonally frozen peatlands can be an important N₂O emission source and that thawing periods in spring can be a hot moment of annual N₂O emissions.

Microbial denitrification and nitrification are considered the most important N₂O sources in soils [37–39]. There are many microbial N₂O production pathways in fungi and bacteria that are involved in denitrification and nitrification, including nitrifier nitrification, nitrifier denitrification, nitrification-coupled denitrification, and heterotrophic denitrification [40–44]. These pathways have different mechanisms and contributions to N₂O production in different habitats [40–44]. Previous studies showed that nitrifier denitrification is the dominate process of N₂O production of soils with low oxygen availability using ¹⁵N and ¹⁸O isotope tracing techniques [40–42]. Fungal denitrification can be a significant source of N₂O across diverse ecosystems, including forest, drylands, and coastal sediments [43, 44]. Although all these processes can produce N₂O, little is known about their relative contribution to N₂O production in seasonally frozen peatlands.

The Antu peatland, located in the Northeast Plain of China, is one of three black soil regions in the World. As it experiences a strong degree of seasonal freezing and thawing through a frozen layer depth of 100 cm, it was selected as the study site [45–47]. In this study, we first monitored the flux of N₂O emissions of a seasonally frozen peatland using a closed chamber technique over a whole year, including peatland thawing, thawed, freezing, and frozen periods. Then, we conducted a series of measurements along the Gudong River (river length, 156.6 km; drainage basin area, 4303 km²) at the regional scale with a focus on the thawing season. Next, we tested our hypothesis along a soil/peat in-situ successional transect (0–200 cm). The soil profile samples were used to investigate the processes responsible for microbial N₂O production during laboratory incubation using isotopic ¹⁵N and ¹⁸O tracing, inhibitor methods, metagenomic sequencing, and metatranscriptomic sequencing.

Materials and methods

Site-scale sampling

Xinhe (sampling site E (Fig. 1C), 597.0 m above sea level; N 42°56′50″, E 128°28′49″) is located in the Yanbian Korean Autonomous Prefecture in Jilin Province, in northeastern China. It is located in one of the three biggest peatland regions in the world [45–47]. Gas (to measure N₂O flux) and soil samples (to determine the mechanism of N₂O production) were collected from three separate profiles in March, May, July, September, and November 2019, as well as in January 2020. The sampling depth was from 0 to 200 cm, with samples taken at 20-cm intervals (0–20, 20–40, 40–60, 60–80, 80–100, 100–120, 120–140, 140–160, 160–180, and 180–200 cm) (n = 180). Within the whole peat/soil profile, the first 0–120 cm was peat, the 120–160 cm layer was mainly clay, and the 160–200 cm layer was gravel. The freezing depth was 100 cm in winter, and peat or soil below 100 cm remained frost-free. The depth of the frozen layer varied with the seasons: 0–40 cm in January, 0–100 cm in March, 20–60 cm in May, 40–60 cm in July, and 0–20 cm in November. It should be noted that the entire peat/soil profile remained unfrozen in September.

Fig. 1. Hot moment of N₂O emissions in seasonally frozen peatland.

A Photos showing the sampling site in different periods and the 200 cm-soil profile of sampling site E. “FL.” and “WT.” indicate the frozen layer and water table, respectively. B Site-scale in-situ N₂O fluxes in the seasonally frozen peatland. The blue curve is the Gaussian fit of the N₂O fluxes. Gas samples were collected in March, May, July, September, and November 2019, as well as in January 2020. C Regional scale in-situ N₂O fluxes in April, May, and June in 2020 at sites along the Gudong River. D Box plots showing the N₂O fluxes of the regional-scale sites (above), and box plots showing N₂O fluxes from this study compared with other ecosystems at the same latitude (below) (the horizontal line indicates the median, while the box shows the 25^th and 75^th percentiles). The abbreviations “wetl.,” “peatl.,” and “upl.” indicate wetland, peatland, and upland, respectively [13, 96]. Two and three asterisks indicate p < 0.01 and p < 0.001, respectively.

Three parallel quadrats (1.5 m length × 1 m width × 2 m depth) with an equal spacing of 10 m were dug each time. A new quadrat was dug next to the old quadrat each sampling time. For each soil quadrat, the soil profile was sampled at four side pits and sliced at specific intervals (20-cm intervals) along the depth from 0 to 200 cm. Soil samples collected from the same depth in all four side pits were thoroughly mixed to generate a composite sample, generating 10 samples per soil pit (n = 30) and sampling occasion. Soil samples for RNA isolation were immediately mixed with RNAlater (Thermo Fisher Scientific, Waltham, MA, USA) (6 mL of solution with 2 g of soil) according to the manufacturer’s instructions. The remaining soil samples were sealed in separate sterile plastic bags and placed on ice. Samples were transported to the laboratory immediately. Prior to experimentation, all visible plant residues were removed. All replicate samples were analyzed separately, and their assay results were averaged to represent the site conditions. One subsample was incubated to determine microbial activity, and another subsample was prepared for chemical analyses. A sub-fraction was stored at −80 °C for DNA extraction and molecular analysis.

Regional-scale sampling

We first selected site-scale sampling site (E) for the annual N₂O flux measurements throughout a whole year in order to determine the hot moment of N₂O emissions for the study site (n = 18), namely, the spring thawing period. Then, regional-scale N₂O flux measurements were conducted in eight peatlands along the Gudong River with a focus on April, May, and June in 2020 (n = 72). The aim of this regional-scale study was to confirm the hot moment identified in the site-scale study; thus, only gas samples were collected in these regional-scale sampling sites.

The Gudong River, the largest tributary of the Di’er Songhua Jiang River, originates from the East Valley of Laoling, which is also located in the Northeast Plain of China, one of three black soil regions worldwide [45–47]. The river is 156.6 km long, and its drainage basin area is 4303 km². All peatland sites were located in the Changbai Mountain region, where the total area of peatland was estimated at 463.31 km², and the estimated soil organic C storage was 47.64 Tg [48]. The plants of these peatlands are similar, generally dominated by hummock-forming Carex species [49]. The height of the hummocks is generally 30–40 cm, and the vegetation coverage is 60–70% [50]. Eight peatland sampling points were selected along the Gudong River; Liangjiang (A, 640.3 m above sea level; N 42°39′15″, E 128°2′43″), Lixin (B, 625.6 m above sea level; N 42°42′13″, E 128°7′40″), Dongqing (C, 636.3 m above sea level; N 42°47′26″, E 128°13′52″), Wanbao (D, 652.2 m above sea level; N 42°50′32″, E 128°18′55″), Xinhe (E), Daba (F, 710.4 m above sea level; N 42°55′27″, E 128°33′25″), Jifang Zi (G, 612.1 m above sea level; N 42°52′21″, E 128°38′55″), and Gudong (H, 656.2 m above sea level; N 42°52′57″, E 128°36′55″) (Fig. 1C).

N₂O flux measurements

The in-situ N₂O fluxes were determined by the closed-chamber technique (volume of 4.8 × 10⁴mL, area of 0.08 m²) [32]. At each sampling point and each time, three parallel static chambers were established. Approximately 300-mL gas samples were extracted from the top of the chamber at 0, 5, 10, 15, 20, and 30 min. The N₂O concentrations were measured upon arrival at the laboratory using gas chromatography (7890A; Agilent Technologies, Basel, Switzerland) with micro-electron capture detection (μECD) [32]. The carrier gas was N₂ with a flow rate of 25 mL min⁻¹. The temperatures of the oven, injection port, and μECD were 70 °C, room temperature, and 330 °C, respectively. The relative error for N₂O measurements by this instrument was <1.5%. N₂O flux was calculated based on the linear variation of its concentration in the chamber headspace as a function of time, base area, chamber volume, and molar volume of N₂O at the chamber headspace air temperature. For all data collected, the coefficient of determination (R²) of the linear regression of concentration change over time was greater than 0.85.

¹⁵N and ¹⁸O tracer assay to determine N₂O production source

In this study, ¹⁵N and ¹⁸O tracing was adopted to distinguish the sources of microbial N₂O in the peatland of sampling site E [40–42]. The surface peat of each sampling month as well as three soil layers (40–60, 100–120, and 160–180 cm) in May (n = 27) were chosen. Fresh and pre-processed peat/soil were pre-incubated for approximately 2 d to increase the homogeneity of the samples and to avoid respiration pulses from the soil samples. Five-gram pre-incubated samples were evenly distributed on the bottom of 60-mL glass serum vials (Ochs Laborbedarf, Bovenden, Germany), which were then sealed with plugs (Ochs Laborbedarf) and aluminum crimp caps (Agilent, Santa Clara, CA, USA). Four treatments enriched with ¹⁸O and ¹⁵N were applied:[40–42] (i) H₂¹⁸O (¹⁸O at 97.2%) + NH₄⁺ + NO₃⁻; (ii) N¹⁸O₃⁻ (¹⁸O at 96.3%) + NH₄⁺ + NO₃⁻; (iii) ¹⁵NO₃⁻ (¹⁵N at 99.19%) + NH₄⁺ + NO₃⁻; and (iv) ¹⁵NH₄⁺ (¹⁵N at 99.16%) + NH₄⁺ + NO₃⁻. All treatments were conducted in triplicate. For all treatments, final enrichments of ¹⁸O-H₂O/NO₃⁻ and ¹⁵N-NH₄⁺/NO₃⁻ were added to 1.0 atom% ¹⁸O and 30 atom% ¹⁵N excesses, respectively. The final N concentrations of both NH₄⁺ and NO₃⁻ were maintained at 50 mg/kg NH₄⁺-N/NO₃⁻-N [40–42]. Subsequently, the gas in the vials was evacuated and then balanced using high-purity Ar (99.99%; Beijing Huayuan Gas, Beijing, China) at standard atmospheric pressure. Treatments were incubated at 60 rpm at different in-situ air temperatures (January, 0–4 °C; March, 4 °C; May, 10 °C; July, 15 °C; September, 20 °C; November, 5 °C). The incubation temperatures of all the following experiments were the same as the temperatures used in this ¹⁵N-¹⁸O tracer experiment.

After one week, the incubation was terminated using 7 M ZnCl₂ solution (500 μL each), and then, the gas samples were transferred to 12-mL vacuum exetainers (Labco, Lampeter, UK). N₂O gas was quantified using gas chromatography. The ¹⁵N and ¹⁸O signatures of N₂O in the gas samples were measured using an isotope-ratio mass spectrometer (IRMS and Precon, Delta V Advantage, Thermo Fisher Scientific, Bremen, Germany; with precisions of <0.04 ‰ δ¹⁵N and <0.07‰ δ¹⁸O, respectively). The ¹⁵N enrichments of mineral N in the soil samples from different treatments were determined using the chemical transformation method [40–42].

The relative contributions of microbial N₂O production pathways (nitrifier nitrification, NN; nitrifier denitrification, ND; nitrification-coupled denitrification, NCD; heterotrophic denitrification, HD) (Supplementary Fig. 1) were calculated according to the method of Kool et al. [40–42].

¹⁵N-tracer assay for N-conversion rate measurement

Potential rates of anammox, denitrification, and dissimilatory nitrate reduction to ammonium (DNRA) were measured with the ¹⁵N-tracer technique, as described by our previous publications [32, 51]. The surface peat of each sampling month (n = 18) was used to perform this experiment. The yield rates of the related products were determined as the potential rates of the three processes, briefly, treatments (1) ¹⁵NH₄⁺ [¹⁵N at 99.16%]; (2) ¹⁵NH₄⁺ + ¹⁴NO₃^–; (3) ¹⁵NO₃^– [¹⁵N at 99.19%] + ¹⁴NH₄⁺. In the case of the soil amended solely with ¹⁵NH₄⁺ (treatment 1), no significant accumulation of ¹⁵N₂-labeled gas (²⁹N₂ and/or ³⁰N₂) was observed in any sample, indicating that all ambient ¹⁴NO_x⁻ had been consumed. When both ¹⁵NH₄⁺ and ¹⁴NO₃⁻ were added (treatment 2), ²⁹N₂ accumulated in every sample without the accumulation of ³⁰N₂. This pattern was reproducible. The results showed that anammox was detectable in the soil. Samples amended solely with ¹⁵NO₃⁻ (treatment 3) showed significant anammox and denitrification rates. The rates and potential contributions to N₂ formation from either anammox or denitrification were calculated from the produced ²⁹N₂ and ³⁰N₂ in treatment 3. Two hundred microliters of ZnCl₂ solution (7 M) was added to terminate the incubation at each time point (0, 6, 12, 24, and 36 h). For DNRA, 100 μL of diluted ¹⁵NO₂^– (¹⁵N at 98.17%) isotopic stock solution was added to the samples. At defined intervals (0, 6, 12, 24, and 36 h), the reaction was stopped, and hypobromite was added followed by incubation for more than 12 h to convert produced ¹⁵NH₄⁺ into ³⁰N₂ completely.

Acetylene inhibition experiment

An acetylene inhibition experiment was used to preliminarily judge the N₂O source (biotic nitrification and denitrification or abiotic processes). The surface peat of each sampling month (n = 18) was used to perform this experiment. Three treatments were performed. One treatment without acetylene (C₂H₂) was used to measure the potential rate of total N₂O production. The second treatment with 0.01% C₂H₂ (v/v) was used to measure the potential rate of the heterotrophic denitrification pathway and abiotic processes [52]. The last treatment used sterilized samples to determine the potential rate of abiotic processes. All treatments were conducted in triplicate. Gas samples were withdrawn from the headspace of the vials at different intervals (0, 3, 6, 12, 24, and 36 h) and transferred to evacuated exetainers (Labco) using a syringe with a three-port valve. The N₂O concentrations of the gas samples were also measured using gas chromatography. The potential rate was calculated from the linear change of N₂O concentrations over the sampling time, and a linear regression with a coefficient (R²) above 0.9 was selected.

Substrate-induced respiration inhibition

The substrate-induced respiration-inhibition (SIRIN) method was used to estimate the contribution from fungi and bacteria [44]. The inhibitor additivity ratio (IAR) was used to evaluate whether antibiotics exerted non-target effects [44]. It was estimated using Eq. (1):

$$\frac{\left(A_{C{O}_2}-B_{C{O}_2}\right)+\left(A_{C{O}_2}-C_{C{O}_2}\right)}{A_{C{O}_2}-D_{C{O}_2}}$$ 1

Here, $A_{C{O}_2}$, $B_{C{O}_2}$, $C_{C{O}_2}$, and ._. represent CO₂ fluxes in antibiotic-free control soil, cycloheximide (CYH)-amended soil, streptomycin (STP)-amended soil, and soil amended with both antibiotics, respectively. A preliminary experiment was designed to determine the optimal concentrations of antibiotics at 0, 4, 6, 8, 12, and 24 h following substrate amendments. Four treatments were performed: (A) antibiotic-free control, soil with no antibiotic addition; (B) CYH, soil with the addition of CYH at 3.0, 6.0, 10.0, and 15.0 mg g⁻¹ soil; (C) STP, soil with the addition of STP at 3.0, 6.0, and 10.0 mg g⁻¹ soil; and (D) both antibiotics, soil with the addition of both CYH (3.0, 6.0, 10.0, and 15.0 mg g⁻¹ soil) and STP (3.0, 6.0, and 10.0 mg g⁻¹ soil). Each sample received an addition of glucose (5.0 mg g⁻¹ soil) as a C source. N₂O and CO₂ concentrations were determined by gas chromatography. Finally, we found that peatland bacterial and fungal activities were largely terminated when STP and CYH concentrations were 10 mg g⁻¹ and 3 mg g⁻¹, respectively.

Fresh, pre-processed, and homogenized peat/soil (0–20, 30–40, 40–60, 100–120, and 160–180 cm of each sampling month, n = 90) were distributed on the bottom of 60-mL glass serum vials, which were then sealed with plugs and aluminum crimp caps. Six treatments in triplicate were applied to random vials to assess the potential rate and contribution to soil N₂O production by bacterial denitrification, fungal denitrification, nitrification, and abiotic processes: (A) antibiotic-free control, soil with no antibiotic addition; (B) CYH, soil with the addition of CYH at 3 mg g⁻¹; (C) STP, soil with the addition of STP at 10 mg g⁻¹; (D) both antibiotics, soil with the addition of both CYH and STP; (E) C₂H₂ at 0.01% (v/v; ammonia oxidation inhibitor), and (F) ZnCl₂ (600 μL, 7 M; biotic process inhibitor). The headspace gas was sampled with a syringe equipped with a Luer lock valve (25.0 mL; Agilent) and injected into a 12.0-mL vacuumed glass serum vial after 0, 4, 6, 8, 12, and 24 h. The gas samples were also measured using gas chromatography. The potential rate and contribution to soil N₂O production by bacterial and fungal denitrification were estimated by Eqs. (2) and (3):

$$\frac{100\times \left(A_{N_{2}O}-B_{N_{2}O}\right)}{A_{N_{2}O}-D_{N_{2}O}}$$ 2

$$\frac{100\times \left(A_{N_{2}O}-C_{N_{2}O}\right)}{A_{N_{2}O}-D_{N_{2}O}}$$ 3

Here, $A_{N_{2}O}$ is the soil N₂O production rate in the antibiotic-free control, $B_{N_{2}O}$, $C_{N_{2}O}$, and $D_{N_{2}O}$ represent the soil N₂O production rate in soils treated with cycloheximide (3 mg g⁻¹), streptomycin (10 mg g⁻¹), and both streptomycin (10 mg g⁻¹) and cycloheximide (3 mg g⁻¹), respectively. The potential rate and contribution to soil N₂O production by nitrification and abiotic process were assessed based on treatments (A), (E), and (F) according to previous studies [32, 44].

Extraction of soil total DNA and RNA

DNA and RNA extraction was conducted immediately after transport back to the laboratory. Total DNA and RNA were extracted from 2.0 g of soil using the DNeasy PowerSoil DNA Isolation Kit (QIAGEN, Hilden, Germany) and the RNeasy PowerMicrobiome RNA Isolation Kit (QIAGEN), respectively, according to the manufacturer’s instructions [53]. All extracted DNA and RNA samples were stored at −80 °C.

Molecular analysis

Quantitative PCR (qPCR) targeting the ammonia monooxygenase sub-unit A gene (archaeal amoA and bacterial amoA) [54] and the nitrite reductase gene (nirK and nirS) [52] were performed on a sequence detection system (ABI 7500; Applied Biosystems, Foster City, CA, USA) with SYBR-Green fluorescent dye (TaKaRa, Dalian, China) (Supplementary Table 1). Standard curves were obtained using 10-fold dilutions of standard plasmids containing archaeal-amoA, bacterial-amoA, nirK, and nirS genes, respectively. Negative controls without DNA template were included in each amplification reaction. The concentration of the primer was 10 μM. The amplification mixture (25 μL) contained the following: 10 μL of SYBR Premix Ex Taq (TaKaRa Biotechnology, Dalian, China), 0.4 μL of ROX reference dye, 0.5 μL of each primer (10 μM), 0.4 μL of bovine serum albumin (BSA), 2 μL of DNA, and 6.2 μL of ddH₂O. Reactions with amplification efficiencies ranging from 90% to 110% and correlation coefficients (R²) above 0.99 were selected for further analysis.

Metatranscriptomic library sequencing and analysis

Ribosomal RNA (rRNA) sequences in the total RNA extract were removed using the Ribo-zero rRNA Removal Kit (Epicentre, Madison, WI, USA). The RNA fragments were reverse-transcribed to create cDNA. Subsequently, the cDNA library was constructed with an average insert size of 300 bp (±50 bp). Finally, 150-bp paired-end sequencing was performed on the Illumina HiSeq 4000 (Illumina, San Diego, CA, USA) at LC-BIO TECHNOLOGIES (Hangzhou, China) (Supplementary Table 2). RNA concentrations for samples from January, obtained from soil depths of 40–60 cm, 60–80 cm, and 80–100 cm, as well as all samples below 1 m, were below the detection limit.

Clean data were generated from raw metatranscriptomic reads after filtering low-quality nucleotides and reads with any ambiguous base calls using Cutadapt (v1.9) [55] and Fqtrim (v0.94). Subsequently, clean data were quality-checked using FastQC [56]. The clean data from each sample were de novo assembled and integrated using Trinity [57]. A non-redundant UniGene catalog was constructed after removing redundant sequences using CD-HIT [58].

Metagenomic library sequencing and analysis

Approximately 1.5 μg of extracted DNA (per sample) was used for metagenomic library preparation and subsequent sequencing on the Illumina NovaSeq 6000 (150-bp paired-end) at LC-BIO TECHNOLOGIES (Hangzhou, China) (Supplementary Table 3). Clean data were generated from raw metagenomic reads after filtering out low-quality nucleotides and reads with any ambiguous base calls using Cutadapt (v1.9) [55] and Fqtrim (v0.94). Valid data from each sample were assembled using IDBA-UD (v1.1.1) [59]. The gene prediction of the assembled contigs was performed by CDS MetaGeneMark (v3.26); [60] then, predicted genes were clustered, and redundant predicted genes were removed using CD-HIT [58] (at 95% identity and 90% coverage thresholds). The non-redundant gene catalog (UniGenes) was obtained after removing the redundant genes. The relative abundance (transcripts per million, TPM) of UniGenes was calculated by comparing the reads of each sample with the CDS library using Bowtie2 (v2.2.0) [61]. Gene taxonomic and functional annotations were obtained using BLASTp [56] against the taxonomic NCBI-NR (v2016-07-12) database and the functional databases eggnog (v2016-06), KEGG (v2016-05), and NCyc (v2019-03) [62] at an e-value <10⁻⁵ threshold.

Genome binning, taxonomic classification, and functional annotation

The clean data obtained from metagenomic sequencing were used to retrieve metagenome-assembled genomes (MAGs) using MetaWRAP [63]. After quality control, clean data were assembled into contigs using MEGAHIT [64]. The assembled contigs were clustered to recover MAGs using MetaBAT2 [65], MaxBin2 [66], and CONCOCT [67]. The resulting three bins were consolidated and filtered with the Bin_refinement module (completeness >50%, contamination <10%) (Supplementary Table 4). CheckM [68] was used to evaluate the completeness and contamination of the MAGs. The taxa corresponding to each of the genome bins were annotated using NCBI_nt, NCBI_tax, and GTDB-Tk (v2.1.15) [69], and their corresponding abundances were estimated using Salmon [70]. Protein-coding genes of each MAG were assigned using the eggNOG, KEGG, and NCyc [62] databases for functional annotation with PROKKA.

Analytical procedures for environmental variables

The soil variables, namely moisture, pH, total organic carbon (TOC), NH₄⁺, NO₃⁻, total organic matter (TOM), total carbon (TC), total nitrogen (TN), and total sulfur (TS), were investigated as described previously [32, 51, 54] (Supplementary Fig. 2). The dissolved oxygen (DO) concentration was measured using a Pocket Oxygen Meter (PyroScience GmbH, Aachen, Germany) equipped with an optical oxygen sensor accordingly.

Statistical analysis

Spearman correlation analysis and one-way ANOVA were conducted using SPSS 10.0 (SPSS Inc., Chicago, IL, USA). Redundancy analysis (RDA) and non-metric multidimensional scaling analysis (nMDS) were performed with CANOCO 5.0. Network analysis controlling for the false discovery rate (FDR) was conducted using the psych package in R and visualized in Gephi [71]. Linear regression was conducted using Origin 9.0 (OriginLab, Northampton, MA, USA). A p value of less than 0.05 indicated statistical significance. All bar charts, scatter plots, and pie charts were generated using Origin.

Results

Hot moment of N₂O flux in peatland

Based on the site-scale investigation of sampling site E (Fig. 1A), the peak N₂O emissions occurred in April (3.08 ± 0.01 mg N₂O m⁻² d⁻¹) and May (0.30 ± 0.00 mg N₂O m⁻² d⁻¹) and were significantly higher than N₂O emissions during the other months (−0.002 ± 0.04 mg m⁻² d⁻¹) (one-way ANOVA, p < 0.001) (Fig. 1B). Moreover, in January and November, the N₂O fluxes were negative, indicating that the frozen peatland during this period was a sink for N₂O. The cumulative amount of N₂O emissions during the thawing period accounted for 103% of the annual net emissions. These results indicate that the seasonally frozen peatland had a high rate of N₂O emission during a hot moment in spring.

The N₂O emissions of the seasonally frozen peatland were further investigated based on regional investigation, including eight peatland sampling sites separated by an interval of 15 km (Fig. 1C). N₂O flux measurements in April, May, and June showed substantial temporal and spatial heterogeneity of N₂O emissions (one-way ANOVA, p < 0.05), with an average flux of 0.55 ± 0.29 mg N₂O m⁻² d⁻¹. However, the N₂O emission fluxes of all sampling sites peaked in April (1.20 ± 0.88 mg N₂O m⁻² d⁻¹) and May (0.38 ± 0.15 mg N₂O m⁻² d⁻¹), and were significantly higher than that in June (0.05 ± 0.05 mg N₂O m⁻² d⁻¹) (Fig. 1D, one-way ANOVA, p < 0.05) (Supplementary Fig. 3). This is in accordance with the site-scale investigation. These results indicate that the thawing period is the hot moment of N₂O emissions in the seasonally frozen peatland. The occurrence of this N₂O emission hot moment may convert the seasonally frozen peatland into an important N₂O emission source (Fig. 1D). If the contribution of the hot moment of these peatlands is not considered, the N₂O emission flux from northern peatlands will be underestimated.

Microbial source of N₂O production

Sources of N₂O production were further investigated in soil cores (0–200 cm cores from site E split at 20-cm intervals) using inhibitor methods (Fig. 2). We found that microbial processes contributed 99.1 ± 0.2% to the total N₂O production, while the remainder was attributed to abiotic processes (Fig. 2A). Among the microbial N₂O production rates, bacterial denitrification was the greatest source (0.27 ± 0.07 ug N₂O g⁻¹ d⁻¹, one-way ANOVA p < 0.01) and accounted for 52.9% of the peatland N₂O production (Fig. 2A), followed by fungal denitrification (0.15 ± 0.05 ug N₂O g⁻¹ d⁻¹, 30.2%) and microbial nitrification (0.07 ± 0.03 ug N₂O g⁻¹ d⁻¹, 15.8%). During the interval from the frozen period to the thawed period of the peatland, the contribution to N₂O production from fungal denitrification and nitrification increased, while that from bacterial denitrification decreased (Fig. 2A). For each sampling time, bacterial and fungal denitrification together contributed 83.1 ± 8.4% to N₂O production. Moreover, their production was highly associated with the N₂O emission flux (R² = 0.63 and R² = 0.80, for bacterial and fungal denitrification, respectively (Supplementary Fig. 4)). Hence, denitrification was the main source of N₂O production in the seasonally frozen peatland.

Fig. 2. Rates and contributions of bacterial, fungal denitrification and nitrification to N₂O production in the seasonally frozen peatland.

A Potential rates of bacterial denitrification, fungal denitrification, and nitrification pathways to the total N₂O production over a whole year in the seasonally frozen peatland, as determined using inhibitor methods. For each sampling month, the pie chart represents the average proportions of the three processes for the 2-m soil core. “BD.,” “FD.,” “MN.,” “AP.,” “FL.,” and “WT.” indicate bacterial denitrification, fungal denitrification, microbial nitrification, abiotic processes, frozen layer, and water table, respectively. The stacked area plot shows the contributions of bacterial denitrification, fungal denitrification, nitrification, and abiotic processes to N₂O production. B Contributions of nitrifier nitrification (NN), nitrifier denitrification (ND), nitrification-coupled denitrification (NCD), and heterotrophic denitrification (HD) pathways in the surface peat to N₂O production, as determined using isotopic ¹⁵N⁻¹⁸O techniques.

There were significant differences in N₂O production in the soil cores through the 200-cm profiles. The total N₂O production rates in different months decreased with soil depth (Fig. 2A). Surface peat (0–20 cm) was the main source of N₂O, contributing 49.1% to N₂O production (Fig. 2A, one-way ANOVA, p < 0.05). The contribution from deep soil (100–120 and 160–180 cm) was lower, but not negligible (17.1%). In the surface and sub-surface peat (20–40 cm), the production of N₂O was mainly attributed to bacterial denitrification. As the depth increased (40–60 cm peat), N₂O was mainly produced via fungal denitrification, accounting for 48.3 ± 8.2% of the N₂O (Fig. 2A). In all the layers, the contribution from nitrification (17.1 ± 5.0%) was lower than that from denitrification. TOC and TOM were the key influencing factors for N₂O production at all soil depths (Supplementary Fig. 5). Overall, the above results demonstrated that the hot moment of N₂O emissions of the seasonally frozen peatland was mainly driven by denitrification.

As surface peat is the main N₂O source of peatlands, peat samples were chosen to furtherly verify the above results using ¹⁵N and ¹⁸O tracing technology. In this analysis, denitrification was still the dominant process contributing to N₂O production (88.6 ± 4.7%). Furthermore, this process was significantly higher than the contribution of nitrification, which includes NN, ND, and NCD (one-way ANOVA, p < 0.001, Fig. 2B). Although N₂O production from ammonia oxidation pathways accounted for a small part, the distributions of NN, ND, and NCD differed among the months (Fig. 2B). ND and NCD pathways were more active in July and September, accounting for over 95% of the ammonia oxidation-derived N₂O (96.3% and 99.8% for July and September, respectively) (Fig. 2B). In the soil profile, the N₂O production from ammonia oxidation was still significantly lower than that of denitrification (Supplementary Fig. 6). The ¹⁵N and ¹⁸O isotope tracing further confirmed that N₂O production in the seasonally frozen peatland was mainly produced by heterotrophic NO₃^- reduction.

The inhibitor assay and isotopic ¹⁵N-¹⁸O tracing results demonstrated that NO₃⁻ reduction was the main driver of the hot moment of N₂O emissions of the seasonally frozen peatland. The NO₃⁻ reduction process includes heterotrophic denitrification, DNRA, and anammox, and their individual contributions were further quantified using ¹⁵N tracing. The contribution of heterotrophic denitrification to NO₃⁻ reduction (31.6 ± 2.0 nmol N g⁻¹ h⁻¹) was significantly higher than that of both anammox (1.14 ± 0.2 nmol N g⁻¹ h⁻¹) and DNRA (1.52 ± 0.4 nmol N g⁻¹ h⁻¹), demonstrating the dominant role of denitrification in the NO₃^- reduction process. However, in contrast with the N₂O emission flux, the denitrification rate had no obvious hot moment (Supplementary Fig. 7). The denitrification rate increased with temperature and peaked during the fully thawed period. This result conflicts with the N₂O emission hot moment observed during the thawing period. Therefore, it was necessary to further explore the microbial mechanism of N₂O production.

Thawing stimulated the expression of genes associated with N₂O production

Metatranscriptomic and metagenomic sequencing were used to analyze functional genes associated with N₂O production (norB, p450nor, and hao) and reduction (nosZ) in bacterial denitrification, fungal denitrification, and nitrification (Fig. 3). In the soil profile, both the DNA- and RNA-based relative abundances of norB showed a slight decrease, while that of hao increased with soil depth. For p450nor, there was no significant difference in abundance (Supplementary Fig. 8). At different sampling depths, both the DNA- and RNA-based relative abundances of nor in the bacterial and fungal denitrification pathway were significantly higher than that of hao in the nitrification pathway, which was further confirmed by DNA-based qPCR results (one-way ANOVA, p < 0.05, Fig. 3A, Supplementary Fig. 8, Supplementary Fig. 9). These multi-omics results indicate that denitrification was the dominant N₂O production route in the seasonally frozen peatland and are consistent with the inhibitor and isotopic tracing results.

Fig. 3. N₂O production- and reduction-related functional gene transcripts identified in seasonally frozen peatland.

A Histogram showing the transcripts of N₂O production genes in the seasonally frozen peatland across a whole year as derived from metatranscriptomic data. “NA” indicates RNA concentrations that were below the detection limit for metatranscriptomic sequencing. TPM means transcripts per million. B The box plots show the relative abundance of N₂O production- and reduction-related functional gene based on RNA data (the horizontal line indicates the median, while the box shows the 25^th and 75^th percentiles). “Tot. pro.” and “Tot. red.” indicate total production and total reduction, respectively. One, two, and three asterisks indicate p < 0.05, p < 0.01, and p < 0.001, respectively. The months of peak abundance were highlighted for N₂O production (light purple) and reduction (light yellow) genes. C Co-occurrence network analysis showing the interaction among active N₂O production- and reduction-related microorganisms during different periods of the seasonally frozen peatland. Node colors indicate the types of genes. Edges are colored according to the interactions among different genes. The abbreviations “pro.” and “red.” indicate production and reduction, respectively.

Metatranscriptomic and metagenomic analysis showed that the relative abundances of genes encoding N₂O production protein complexes (i.e., nor and hao genes) were significantly higher than those of genes encoding N₂O reduction proteins (i.e., nosZ). However, the abundances of these genes during the thawing period were 300–500 times higher than those during the thawed and freezing periods as RNA transcripts, but there was no significant difference across all the periods at the DNA level (Fig. 3B, Supplementary Fig. 8, one-way ANOVA, p > 0.05). The large difference between N₂O production and reduction (16–32 times) implies that the seasonally frozen peatland had high N₂O emission potential during all the periods, while only the soil properties and physical state during the thawing period stimulated the observed N₂O burst.

The high expression of genes associated with N₂O production in March indicates that a large amount of N₂O may also be produced in March. Because the peatland is completely frozen at this time (frozen-depth, 100 cm), the N₂O produced during the frozen period may be trapped by ice in the soil. This N₂O can then be released with soil thawing [19, 27] and thus may also contribute to the hot moment of N₂O emissions in spring (April and May). However, such a physical release of N₂O is unlikely to be the direct reason for the occurrence of the observed hot moment, because the potential N₂O production rate was low during the frozen period (Fig. 2A). Moreover, both the N₂O production rate and gene expression during the thawing period were significantly higher than those during the frozen period (Figs. 2A and 3A). Hence, microbial activity during thawing periods is expected to be the main cause of the hot moment of N₂O emissions.

As substrates for microbial growth, TOC (p = 0.004, F = 11) and TC (p = 0.004, F = 8) were the dominant factors influencing the abundance of N₂O production- and reduction-related functional genes according to RDA (Supplementary Fig. 10). TOC and TC explained 30.6% and 24.2% of the variation in gene abundances, respectively, as supported by correlation analysis (Supplementary Table 5). Repeated freezing and thawing of the peatland can lead to the breakdown of soil aggregates and render decomposable organic C more easily accessible to microorganisms, thereby activating N₂O producers [27, 72, 73]. The acidic peatland environment (pH 3.3–4.5) could inhibit N₂O reductase and thus cause N₂O emissions [28, 30]. These results show that the freeze-thaw cycle during the thawing period stimulated microbial activities related to N₂O production and reduction and further promoted peatland N₂O emissions.

The microbial community structure was then further analyzed. Metatranscriptomic analysis revealed that the samples from the thawing period not only showed a clear separation from those of other periods, but also had significantly higher microbial diversity based on RNA, while there was no significant difference between periods in terms of microbial community and diversity based on DNA (Supplementary Figs. 11 and 12), which indicates that thawing altered the active microbial community structure. Co-occurrence networks from the different seasonal periods revealed that thawing increased the complexity of the N₂O-related microbial networks (Fig. 3C, Supplementary Tables 6 and 7), thereby reflecting the strengthened interactions between N₂O-related communities. These results indicated that thawing promoted synergy among active microorganisms.

Metabolic potential related to N₂O production

The potential denitrification metabolic pathway related to the N₂O hot moment was further revealed by metagenome analysis (Fig. 4). The assembly and binning processes generated 77 metagenome-assembled genomes (MAGs), of which 41 genomes encode genes involved in N₂O production and reduction (Supplementary Fig. 13). A total of 22 genomes possessed denitrifying nirK or nirS genes (encoding proteins involved in NO₂⁻ reduction to NO), while no MAGs contained genes involved in ammonia oxidation (amoA/B/C) (Fig. 4, Supplementary Fig. 13), further indicating a dominant role of denitrification in N₂O production. No MAGs encoded capabilities for all denitrification genes, i.e., genes that encode proteins involved in NO₃⁻ reduction to NO₂⁻, NO₂⁻ reduction to NO, NO reduction to N₂O, and finally N₂O reduction to N₂ (Fig. 4). This suggested that denitrification was a sequential process catalyzed by different enzymes and microorganisms as was observed in other studies [74, 75], and during the thawing period, the interaction among N₂O producing microorganisms was significantly stronger than that between N₂O producing and reducing microorganisms (Fig. 3C, Supplementary Table 6). Additionally, the abundance and expression of N₂O production genes were also significantly higher than those of N₂O reduction genes (Fig. 3B, Supplementary Fig. 8). These dual effects resulted in the occurrence of an N₂O emission hot moment during the thawing period.

Fig. 4. Metabolic reconstruction and features of N₂O production- and reduction-related microorganisms identified in seasonally frozen peatland.

Metabolic features across selected metagenome assembled genome (MAG) phylogenetic clusters. MAGs with the nirK gene were selected for phylogeny construction. Having any one or more functional genes in a MAG that performs the same process is indicated by “Present.” Bootstrap support values for 1000 replicates are indicated at each node.

The N-cycle metabolic pathway potentially related to the N₂O hot moment was further examined. Among the MAGs, 76.9% contained genes (narG/H/I, napA/B, narB, nasA/B, and NR) involved in NO₃⁻ reduction to NO₂⁻ (Supplementary Fig. 13), indicating that under limited oxygen conditions, NO₃⁻ reduction may be an appropriate alternative pathway for energy conservation, which supports our previous results [51]. Additionally, 62.3% of the MAGs contained genes that encode proteins with capabilities for NO₂⁻ reduction to NH₄⁺ (nirB/D, nirA, and nrfA/H) through assimilatory/dissimilatory nitrate reduction (Supplementary Fig. 13). The gene abundances for NO₂⁻ reduction to NH₄⁺ (99 ± 6 TPM) were significantly higher than that for NH₄⁺ oxidation (3 ± 1 TPM). In addition, the gene abundances for organic nitrogen mineralization (138 ± 55 TPM) were significantly higher than that for nitrogen assimilation (78 ± 26 TPM). This resulted in the accumulation of NH₄⁺ in the peatland (median 15.0 mg kg⁻¹, IQR (3.6, 21.5) mg kg⁻¹), in accordance with our hypothesis of an increased microbial mineralization rate.

However, the driver of the N₂O emission hot moment was heterotrophic denitrification rather than ammonia oxidation, despite the high NH₄⁺ content of the peatland (Supplementary Fig. 2). While acidophilic nitrification occurs in soils [76], oxygen-limitation (DO 0.05 ± 0.04 mg L⁻¹) in the peatland environment is unlikely to be conducive to ammonia oxidation [77]. Denitrifying community members in the peatland may exhibit acid tolerance under a low pH value. For example, denitrifying genome bin 79, contains genes encoding urease, which can move cytoplasmic NH₃ through to the periplasm and buffer the periplasm through the reaction converting H⁺ and NH₃ to NH₄⁺, resulting in a relatively neutral periplasmic pH (Supplementary Fig. 14). All these results indicated that heterotrophic denitrification driven by the thaw-freeze cycle was the predominant factor causing the hot moment of N₂O emissions occurring in the seasonally frozen peatland.

Discussion

Northern peatlands are sensitive to climate change, while their vulnerability to climate warming is uncertain [2–4]. Our results reported that seasonally frozen peatlands can be important N₂O emission sources in the Northern Hemisphere and that thawing periods are a hot moment of N₂O emissions driven by freezing-thawing cycles. These emission rates matched the highest N₂O emissions recorded in natural terrestrial ecosystems [78], corresponding to about 6% of the global warming potential of natural northern peatlands CH₄ emissions [79]. The hot moment of N₂O emissions that occurred in the seasonally frozen peatlands (0.79 ± 0.29 mg N₂O m⁻² d⁻¹) was higher than those from drained peatland forests (spring, 0.58 mg N₂O m⁻² d⁻¹) [13], permafrost (growing season, 0.047 mg N₂O m⁻² d⁻¹) [13], and non-managed northern peatland (0.22–0.65 mg N₂O m⁻² d⁻¹) [80]. It was comparable to N₂O emissions from managed northern peatlands influenced by freeze-thaw cycles (0.13–10.2 mg N₂O m⁻² d⁻¹) [81], with equivalent N₂O production rates. Neglecting these hot moment emissions may lead to underestimation of N₂O emissions representing about 1.1% of N₂O emissions from natural sources [82].

The hot moment release characteristics of N₂O emissions from seasonally frozen peatlands should be mainly determined by the freeze-thaw state of the soil. During the freezing (fall), frozen (winter), and thawed (summer) periods of the seasonally frozen peatland, N₂O emissions were relatively low, and during these periods, the study site even acted as a sink rather than an emission source. However, the flux during the hot moment of thawing (spring) was significantly higher than that during the other periods. The ice-covered peat surface in the freezing and frozen periods prevents N₂O from escaping until peat thaw commences [27]. These N₂O emissions can be physically released with peat thawing, but are unlikely to be the direct reason for the occurrence of a hot moment, because the potential N₂O production rate during winter was relatively low. In contrast with previous studies, in which temperature was determined to be the most critical factor affecting N₂O emission flux [13, 17], in our study, the N₂O fluxes in both high (summer, 0.04 ± 0.01 mg N₂O m⁻² d⁻¹) and low temperature seasons (winter, -0.15 ± 0.03 mg N₂O m⁻² d⁻¹) were significantly lower than that during the thawing seasons (spring, 1.69 ± 0.98 mg N₂O m⁻² d⁻¹), indicating the key role of thawing in N₂O emission. The freezing-thawing cycles in the thawing periods increased both the quantities and availabilities of soil carbon and nitrogen compounds [20, 24], which supply substrates necessary for microbial N₂O production, especially via bacterial and fungal denitrification, underlying the hot moment of N₂O emission.

Heterotrophic denitrification driven by the thaw-freeze cycle was the predominant cause of the hot moment of N₂O emissions from seasonally frozen peatlands. Consistent with this study, denitrification can generate large amounts of N₂O in northern peat soils [11, 38], especially within deeper peat layers [83]. However, nitrification can be a key driver of high N₂O emissions from arctic peat soils [84, 85], as it regulates the supply of NO₂⁻ and NO₃⁻ for N₂O production through denitrification, especially in bare peat during dry years [83]. In this study, corresponding with the N₂O production rate, the abundances of nitrification-related genes were significantly lower than those of denitrification-related genes, indicating that nitrification may not provide enough NO₃⁻ for denitrification. The dominant vegetation type is often used to judge the nutritional status of peatlands [86, 87]. In this study, Carex is the dominant plant in the peatlands, which indicates a mesotrophic peat environment. The nutrient supply here, including NO₃⁻, comes from rainfall, surface water, and groundwater [86, 87], not mainly from nitrification.

N₂O can be produced at different depths in the soil profile. In this study, surface peat was the main source of N₂O attributed to denitrification. The relatively high oxygen concentration in the surface peat may impair the reduction of N₂O [88], leading to N₂O accumulation and emission. The subsurface peat is an anoxic environment with a high concentration of organic matter, which may promote complete denitrification [89]. The final product is mainly N₂, not N₂O. Low nutrient contents in deeper soil layers result in a decrease in denitrifying microbial abundance and a limitation of N₂O production by denitrification, which is consistent with a previous study [90].

By using combined metagenomic and metatranscriptomic sequence data designed to capture the genes responsible for key functions of the N cycle, we observed significant increases in the relative abundance of N₂O production genes related to bacterial denitrification, fungal denitrification, and nitrification during the thawing period, which was in line with previous findings [11, 91, 92]. The RNA transcripts of N₂O production genes in the thawed period were significantly lower than those in the thawing period, but not the potential rate of N₂O production and denitrification. On the one hand, this could be the result of intense competition for available N sources between plants and microorganisms [11, 17]. DNA-based qPCR and metagenomic data indicated high N₂O production potential across all periods. However, plants meet their N requirements by absorbing small compounds such as oligopeptides, amino acids, and NO₃⁻. Insufficient NO₃⁻ and TOC may lead to low activities of denitrification in the in-situ environment, which is manifested as low RNA transcript abundance and N₂O emission flux. On the other hand, laboratory microcosm incubation methods may disrupt mass transfer and increase substrate utilization, which just characterizes the potential rate of N₂O production and denitrification [93]. This highlights that short-term laboratory experiments used to examine the microbial limitations of the N cycles of seasonally frozen peatlands [11, 94] do not well represent the real changes in the microbial community and their functioning over time. These results further indicated that thawing provided ideal conditions for the development of active N₂O-producing microbial communities, resulting in high N₂O release in seasonally frozen peatlands.

Conclusion

With the intensification of global warming [21], the rate of carbon degradation and nitrogen mineralization in peatlands will be greatly increased [22, 23]. It can be predicted from our study that the warming-effect caused by hot moment N₂O emissions in seasonally frozen peatlands may make peatlands a net source of warming; however, both anthropogenic and natural disturbances may accelerate this occurrence [95]. Thus, it is important to protect peatlands for the sake of ecological security and climate change at a global scale. Overall, patterns and drivers of seasonally frozen peatland N₂O emission and production rates observed in this study offer new insights into potential nitrogen release upon peatland thawing and provide important information for use in Earth system models to better predict northern peatland biogeochemical cycles under a warming climate.

Author contributions

GZ designed the project. XW, GZ, and SW contributed to sample analysis. XW and GZ wrote the manuscript with contributions from SW, YY, HT, MSMJ, and CS. All authors discussed and interpreted the results and contributed to the manuscript. All authors discussed and commented on the manuscript. Correspondence and requests for materials should be addressed to GZ. (gbzhu@rcees.ac.cn).

Data availability

The metagenomic data and the metatranscriptomic data were deposited in the NCBI Sequence Read Archive under accession numbers SAMN26424243–SAMN26424272 and SRR14251110–SRR14251136, respectively.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41396-023-01389-x.