Spatial Variations of Soil N₂ and N₂O Emissions from a Temperate Forest: Quantified by the In Situ ¹⁵N Labeling Method

Abstract

Emissions of dinitrogen (N₂) and nitrous oxide (N₂O) from soil are important components of the global nitrogen cycle. Soil N₂O emissions from terrestrial ecosystems have been well studied. However, patterns and mechanisms of N₂ emissions remain unclear due to the technical difficulty in measuring N₂ production. In this study, an in situ ¹⁵N labeling method was employed to determine soil N₂ and N₂O emission rates from the lower, middle, and upper slopes, which correspond to different moisture conditions, in a temperate forest in Northeast China. We found that N₂ emissions varied from 85 to 3442 μg N m⁻² h⁻¹ across the slopes and were dominated by denitrification. The emissions of bulk N₂O (22 to 258 μg N m⁻² h⁻¹) and denitrification-derived N₂O (14 to 246 μg N m⁻² h⁻¹) were significantly lower than N₂ emissions from their corresponding slope positions. Both N₂ and N₂O emissions significantly increased when soils become wetter. The ratios of N₂O/(N₂O + N₂) were significantly higher at the upper and middle slopes (0.22 and 0.20, respectively) compared with those at the lower slope (0.08 ± 0.01). At the catchment scale, N₂ accounted for 85% of the total gaseous N losses (N₂O + N₂). Our study shows that soil moisture drives the patterns of N₂ and N₂O emissions and field quantification of N₂O/(N₂O + N₂) ratio should further consider the effect of slope position of forest ecosystems to estimate total soil gaseous N losses.

1. Introduction

Gaseous nitrogen (N) emissions (e.g., NO, N₂O, N₂) from soils play a crucial role in the global N cycle and climate change [1] and have been proposed as an important mechanism of terrestrial ecosystem N limitation [2]. Nitrous oxide (N₂O) and NO, as by-products of denitrification and nitrification, strongly impact global warming and atmospheric chemistry, respectively [3]. Dinitrogen (N₂), the end product of denitrification, is relatively inert in the atmosphere [4]. Compared to N₂O and NO, soil N₂ emissions have not been well-quantified due to the high background atmospheric N₂ concentration [5]. Presently, the acetylene (C₂H₂) inhibition method [6], ¹⁵N isotope trace technique [7,8,9,10], and gas–flow soil core method [11,12,13], have all been used to determine N₂ emissions from the soil, and these methods contribute to our understanding of soil N₂ dynamics. However, at a large ecosystem scale, the field quantification of N₂ fluxes remains a huge challenge due to high spatial and temporal variations in soil environments [5,14].

Due to the above-mentioned difficulty in detecting soil N₂, the N₂O/(N₂O + N₂) ratios commonly obtained from the laboratory are applied to estimate soil N₂ flux combined with field N₂O flux [15,16,17]. Recently, for instance, soil N₂ flux from a maize field was calculated based on laboratory-quantified N₂O/(N₂O + N₂) ratios, in situ measured N₂O emission rate, and soil factors [13]. However, some studies indicated that the physical transport of N₂O and N₂ from soil to the atmosphere in the laboratory incubation system did not realistically reflect in situ conditions [18,19,20]. This is because field soil environment is highly variable, where oxygen (O₂) concentration, available substrates, and other soil properties change both spatially and temporally. Additionally, the changes in soil properties have different effects on the emissions of N₂ and N₂O [21,22,23], consequently affecting N₂O/(N₂O + N₂) ratios. Two previous studies suggested a wide range of N₂O/(N₂O + N₂) ratios in the terrestrial ecosystem through a synthesis of the relevant literature [16,24]. To date, few field studies from temperate forests investigated soil N₂O/(N₂O + N₂) ratios using either the acetylene inhibition method [25,26] or the ¹⁵N isotope trace technique [27,28]. However, it is still unclear to researchers whether the pattern of in situ N₂O/(N₂O + N₂) ratios is consistent with those in laboratory assays (i.e., Ref 9, 10 and 13). The lack of data obtained at the field scale creates great uncertainties on the N₂ fluxes and global N cycle estimated by the existing models. Therefore, it is crucial to clarify the pattern of N₂O/(N₂O + N₂) ratios in the field, which can be used as a promising tool to accurately estimate N₂ fluxes and denitrification at the ecosystem scale.

Forests cover 31.7% of global land and play a vital role in regulating the N cycle and global climates [29]. Soil N₂ and N₂O emissions are mainly mediated by microorganisms. Environmental factors, such as N and carbon (C) availability and soil moisture, influence the populations and activities of nitrifier and denitrifier, consequently causing the changes in N₂ and N₂O emissions and N₂O/(N₂O + N₂) ratios [21]. Studies on factors controlling N₂O emissions from forest soils have attracted a great attention due to their global warming effect. In addition, previous studies indicated that N₂O and N₂ could be simultaneously emitted from the same soil aggregate because of the development of aerobic and anaerobic environments [13]. Thus, factors that control N₂O emissions also regulate N₂ emissions. For example, exogenous N addition in a tropical forest reduced soil N₂O emissions but had an enhancing effect on N₂ emissions under anaerobic conditions [30]; the snowmelt process promoted soil N₂ and N₂O emissions in a northern hardwood forest [9]. These studies suggest that the created anaerobic microsites in soil can enhance the proportion of denitrification to nitrification affecting the pattern of N₂ and N₂O emissions. Soil moisture may be a key driver in the development of the anaerobic conditions and the relative proportion of N₂ and N₂O emissions in the field since water controls prerequisite conditions (i.e., NO₃⁻ and O₂ availability) for the occurrence of nitrification and denitrification [7,31]. Moreover, topography can affect denitrification rates by changing the distribution of the substrate (NO₃⁻ or C) in soil water [32]. However, the effects of soil water on N₂ emissions and N₂O/(N₂O + N₂) ratios remain unclear in the field under complex topography.

The purpose of the work reported was to evaluate the influence of slope position and associated soil moisture conditions on rates and ratios of N₂O and N₂ emissions in a temperate forest in Northeast China by employing an in situ ¹⁵N labeling technique. The specific objectives of the study were to: (1) Use an in situ ¹⁵N labeling method to determine soil N₂ and N₂O emission rates across three slopes (analogous to three moisture conditions) from a temperate forest in Northeast China; (2) explore the controlling factors of N₂O/(N₂O + N₂) ratios; and (3) estimate the N₂ flux at the current ecosystem scale by measuring N₂O/(N₂O + N₂) ratios and N₂O fluxes. We hypothesized that lower slopes would have greater conversion rates of N₂O to N₂ and lower N₂O/(N₂O + N₂) ratios due to higher soil moisture condition.

2. Materials and Methods

2.1. Site Description

This study was conducted at the Qingyuan Forest Station (124°54′32.6″ E, 41°51′6.1″ N, 500–1100 m elevation) of Chinese Ecosystem Research Network (Qingyuan Forest CERN), located in the Liaoning Province, Northeast China (Figure 1). The climate of this region is a continental monsoon type. In this station, the annual average precipitation from 2014 to 2020 is 666 mm with approximately 80% falling during the growing season from April to October, and the annual average temperature is 4.7 °C [33,34]. Total inorganic N deposition (TIN) in precipitation at the study site is 15 to 21 kg ha⁻¹ yr⁻¹ (during 2014 to 2016) with NH₄⁺ contributing 65% of the N input [34,35].

The study site was originally occupied by a primary mixed broadleaved–Korean pine forest until the 1930s. Subsequently, the original forest was destroyed by a large fire in the early 1950s and replaced by a mixture of naturally regenerating broadleaved native tree species [36]. This natural secondary forest consists of Quercus mongolica, Juglans mandshurica, Fraxinus mandschurica, Phellodendron amurense, and Larix olgensis in the tree layer; Syringa amurensis, Acer tegmentosum, and Fraxinus rhynchophylla in the understory; and Equisetum hyemale, Arisaema amurense, and Polygonatum involucratum in the herbaceous layer [37]. In this forest, soil with the depth of 60 to 80 cm is developed from granite gneiss and categorized into Udalfs according to the definition of the second edition of U.S. Soil Taxonomy (1999) [36]. The soil texture in 0–10 cm mineral layer is clay loam (17.2% sand, 52.8% silt, and 30% clay) with a soil pH of 5.8, bulk density (BD) of 0.70 kg m⁻³, total organic carbon (TOC) of 2.1%, total nitrogen (TN) of 0.4%, and C/N ratio of 9.1 [34,38]. Due to the multiple effects of the topography and precipitation, the soil in lower places is usually in a water-saturated state compared with that in higher places, where soil properties can change with slope position in forests (

Table 1. Physical–chemical properties of the 0–20 cm mineral soil layer at different slope positions in the mixed forest of Qingyuan station in Northeast China.

Slope Position	Site	Year	BD (g cm−3)	pH	WFPS (%)	TOC (g kg−1)	TN (g kg−1)	C/N Ratio	NH4+ (mg N kg−1)	NO3− (mg N kg−1)	NH4+/NO3− Ratio
Upper	Site 1	2015	0.60 ± 0.06 a	5.4 ± 0.0 b	36 ± 4 c	4.7 ± 0.2 b	0.47 ± 0.02 b	10.2 ± 0.1 c	2.1 ± 0.1 b	2.9 ± 0.1 a	0.7 ± 0.0 b
	Site 1	2017	0.58 ± 0.01 a	5.6 ± 0.1 a	40 ± 1 c	7.7 ± 1.3 a	0.74 ± 0.09 a	10.2 ± 0.5 a	5.9 ± 0.7 a	1.5 ± 0.2 a	3.9 ± 0.1 a
	Site 2		0.56 ± 0.01 a	5.9 ± 0.1 a	28 ± 1 c	5.8 ± 0.2 b	0.58 ± 0.01 b	9.9 ± 0.1 b	8.0 ± 0.3 a	1.7 ± 0.1 a	4.7 ± 0.2 a
	Site 3		0.60 ± 0.05 a	5.9 ± 0.1 a	40 ± 5 c	6.0 ± 0.4 a	0.62 ± 0.03 a	9.4 ± 0.1 c	6.4 ± 0.3 a	1.4 ± 0.1 a	4.7 ± 0.4 a
	Total		0.59 ± 0.02 a	5. 7 ± 0.1 a	35 ± 2 c	6.0 ± 0.4 b	0.58 ± 0.03 b	9.9 ± 0.1 b	5.3 ± 0.7 b	2.0 ± 0.2 a	3.2 ± 0.5 b
Middle	Site 1	2015	0.48 ± 0.03 a	5.6 ± 0.1 a	67 ± 4 b	5.1 ± 0.3 b	0.44 ± 0.04 b	11.7 ± 0.4 b	2.6 ± 0.8 b	2.9 ± 0.3 a	0.8 ± 0.2 b
	Site 1	2017	0.53 ± 0.02 ab	5.6 ± 0.0 a	69 ± 2 a	8.5 ± 0.8 a	1.01 ± 0.16 a	10.2 ± 0.5 a	7.4 ± 0. 9 a	1.6 ± 0.2 a	4.8 ± 1.1 a
	Site 2		0.49 ± 0.06 a	6.0 ± 0.1 a	59 ± 6 b	7.5 ± 0.9 b	0.69 ± 0.11 ab	10.8 ± 0.7 ab	9.2 ± 1.5 a	2.9 ± 0.7 a	3.8 ± 0.9 a
	Site 3		0.56 ± 0.01 a	5.8 ± 0.1 a	65 ± 3 b	8.6 ± 1.6 a	0.79 ± 0.11 a	10.4 ± 0.5 ab	9.4 ± 1.5 a	1.1 ± 0.2 a	9.4 ± 2.5 a
	Total		0.52 ± 0.02 ab	5.7 ± 0.1 a	65 ± 2 b	7.4 ± 0.6 b	0.71 ± 0.07 ab	10.8 ± 0.3 b	7.1 ± 1.0 ab	2.2 ± 0.3 a	4.7 ± 1.1 b
Lower	Site 1	2015	0.44 ± 0.07 a	5.3 ± 0.1 b	128 ± 35 a	15.0 ± 3.2 a	1.07 ± 0.18 a	13.8 ± 0.6 a	10.4 ± 1.8 a	2.0 ± 0.5 a	5.6 ± 0.9 a
	Site 1	2017	0.49 ± 0.02 b	5.5 ± 0.0 a	100 ± 5 a	9.1 ± 1.2 a	0.80 ± 0.10 a	11.4 ± 0.5 a	7.3 ± 0. 8 a	1.0 ± 0.2 a	8.1 ± 1.8 a
	Site 2		0.46 ± 0.06 a	5.9 ± 0.1 a	83 ± 8 a	10.2 ± 0.7 a	0.89 ± 0.04 a	11.5 ± 0.3 a	10.5 ± 1.8 a	1.0 ± 0.3 a	11.8 ± 1.5 a
	Site 3		0.47 ± 0.10 a	6.0 ± 0.1 a	111 ± 24 a	8.5 ± 3.3 a	0.72 ± 0.25 a	11.4 ± 0.6 a	9.1 ± 3.6 a	1.7 ± 0.5 a	6.4 ± 1.9 a
	Total		0.47 ± 0.03 b	5.7 ± 0.1 a	105 ± 9 a	10.3 ± 1.1 a	0.85 ± 0.08 a	11.9 ± 0.3 a	9.1 ± 1.0 a	1.4 ± 0.2 b	8.1 ± 1.0 a

Data are presented as mean and standard errors. Values from different sites (1, 2, 3) and overall averages are shown. Different lowercase letters indicate significant differences among slopes (p < 0.05).

).

2.2. Field ¹⁵N Labeling Experiment

To determine in situ soil N₂ and N₂O emissions, static chamber and ¹⁵N gas flux methods were adopted for rate measurements. In June 2015, one transect zone (site 1) along a 20 m (width) × 35 m (length) slope was chosen and divided into lower (20 m × 13 m), middle (20 m × 12 m), and upper (20 m × 10 m) slopes with three distinctive soil moisture conditions: high, intermediate, and low, respectively (Figure 2 and Figure S1). At each slope zone, four 2 m × 3 m plots were randomly established, and a collar of stainless steel (basal area 0.09 m², 30 cm × 30 cm) was randomly anchored into the soil at 10 cm depth in each plot (Figure 2). All collars were finished one week before soil ¹⁵N labeling. The collar had a square groove of 3 cm × 3 cm depth for matching the stainless steel chambers, and the square groove could provide a gas-tight seal when filled with water. The chambers were equipped with a 3-way sampling port and a 3 mm diameter pressure equilibration tube (15 cm long) on the preinstalled frames. Solutions of labeled Na¹⁵NO₃ (99.26 atom% ¹⁵N, Shanghai Research Institute of Chemical Industry, Shanghai, China) were evenly injected into the soil within the collars at a rate of 2.5 g ¹⁵N m⁻² (diluted in 900 mL DI water, through 144 injections) by a syringe with a 10 cm long needle. The area inside the collar was then sprinkled with 300 mL DI water to wash the residual labeled solution on the litter into the soil. Subsequently, the collars were covered with the chambers, and a 50 mL gas sample was taken from the headspace with a gas-tight syringe at the time points of 4, 7, 24 and 30 h after the ¹⁵N tracer injection. The gas samples (50 mL) were transferred into the Tedlar gas bags (100 mL). At the end of incubation, five soil cores (0–20 cm) in each chamber were collected with a stainless steel sampler (2.5 cm diameter, 50 cm length) and composited to one sample. Field ambient air samples and composite soil samples (0–20 cm) were taken at the beginning of the incubations near each collar (within 15 cm distance).

In June 2017, an additional two transects (site 2 and site 3, respectively), where the aspect, vegetation composition and soil texture were similar to those in site 1, were established in the same forest and divided into lower, middle, and upper slopes as three different soil moisture zones. Using the same procedures as mentioned above, four stainless steel collars on each slope of site 1, site 2, and site 3 were randomly inserted into the soil one week before soil ¹⁵N labeling, and the same dose of ¹⁵N tracer solution was injected into the soil. The gas and soil samples were collected in the same way as described above.

2.3. Soil N₂O, N₂ Analysis and Flux Calculation

A 5 mL gas sample was injected into the gas chromatography for analyzing N₂O concentrations (C). The gas chromatography was fitted with a Porapak Q column (30 m length, 0.53 mm id) and equipped with an electron capture detector (ECD) (GC-2014, Shimadzu, Kyoto, Japan). Three standard N₂O samples with concentrations of 0.35, 5, 20 ppm (National Center for Standard Matters, Beijing, China) were used to calibrate the sample N₂O concentration. The bulk N₂O flux ($F_{{\mathrm{N}}_2\mathrm{O}}$, µg∙m⁻²∙h⁻¹) was calculated from the linear change in N₂O concentration over time and the following Equation (1):

$$F_{{\mathrm{N}}_2\mathrm{O}}=(d{C}_T/dT)\times {\mathsf{\rho }}_{{\mathrm{N}}_2\mathrm{O}}\times V/A$$

(1)

where bulk N₂O flux includes the fluxes from ¹⁵N-labeled and non-labeled sources; $C_T$ is the N₂O concentration in the mixed gases at the incubation time T (ppm); ρ_N2O is the N₂O density at an air temperature of 20 °C in the field; V and A are the volume (m³) and basal area (m²) of the chamber.

The enrichment of ¹⁵N in N₂O was measured by an IsoPrime trace gas analyzer (TG) coupled with an auto-sampler of 112 plots (Gilson GX-271, Dunstable, UK) and a continuous-flow isotope ratio mass spectrometer (IRMS, IsoPrime 100, Cheadle, UK). The peak areas for major (⁴⁴N₂O), minor 1 (⁴⁵N₂O), and minor 2 (⁴⁶N₂O) from IRMS, as well as the ratios ⁴⁵R (⁴⁵N₂O/⁴⁴N₂O) and ⁴⁶R (⁴⁶N₂O/⁴⁴N₂O), were reported in all the gas samples from enriched ($Tm$ = 4, 7, 24, 30 h) and ambient air samples ($T0$). Previous studies showed that the non-random ¹⁵N distribution in N₂O was observed due to the high ¹⁵N enrichment in the source pool [8]. Hence, the ¹⁵N enrichment of N₂O (${{}_{}{}^{15}X}_{N_{2}O}$) at each time point was calculated according to the ratios of ⁴⁵R and ⁴⁶R with the following Equation (2), assuming ¹⁷R (¹⁷O/¹⁶O) = 3.8861 × 10⁻⁴ and ¹⁸R (¹⁸O/¹⁶O) = 2.0947 × 10⁻³ [8,39]. The ¹⁵N₂O flux ($F{}_{}{}^{15}N{}_{2}O$) was estimated through bulk N₂O flux and the ¹⁵N difference of N₂O enrichment between enriched and air samples. Furthermore, the N₂O flux produced from denitrification ($F_{{\mathrm{N}}_2{\mathrm{O}}_{denitrification}}$) was calculated by dividing the ¹⁵N₂O flux by ¹⁵N enrichment of the soil-labeled NO₃⁻ pool (${{}_{}{}^{15}X}_{{{\mathrm{NO}}_3}^{-}}$) [8,40].

$${{}_{}{}^{15}X}_{{\mathrm{N}}_2\mathrm{O}}=100\times \frac{{}_{}{}^{45}R+2\times {}_{}{}^{46}R-{}_{}{}^{17}R-2\times {}_{}{}^{18}R)}{2+2\times {}_{}{}^{45}R+2\times {}_{}{}^{46}R}$$

(2)

$$F{}_{}{}^{15}\mathrm{N}{}_2\mathrm{O}=({}_{}{}^{15}X{}_{{\mathrm{N}}_2\mathrm{O}-Tm}-{}_{}{}^{15}X{}_{{\mathrm{N}}_2\mathrm{O}-T0})/100\times {\mathrm{F}}_{{\mathrm{N}}_2\mathrm{O}}$$

(3)

$$F_{{\mathrm{N}}_2{\mathrm{O}}_{denitrification}}=F{}_{}{}^{15}\mathrm{N}{}_2\mathrm{O}/{{}_{}{}^{15}X}_{{{\mathrm{NO}}_3}^{-}}$$

(4)

The content and rate of N₂ in samples were also determined using the TG-IRMS system. Gas sample (0.5 mL) was manually injected into the sample loop (50 μL) using a gas-tight syringe and the peak areas for major (²⁸N₂), minor 1 (²⁹N₂), and minor 2 (³⁰N₂) from IRMS as well as the ratios ${}_{}{}^{29}R$ (²⁹N₂/²⁸N₂) and ${}_{}{}^{30}R$ (³⁰N₂/²⁸N₂) were measured in both enriched ($Tm$ = 4, 7, 24, 30 h) and ambient air samples ($T0$). According to the difference in the ratios ${}_{}{}^{29}R$ and ${}_{}{}^{30}R$, respectively, between the sample and normal air, the ¹⁵N mole fraction (${}_{}{}^{15}X{}_{N_2}$) and ¹⁵N flux ($F^{15}{\mathrm{N}}_2$) of N₂ at each time point were calculated using the following Equations (5) and (6). Then, the N₂ flux produced from denitrification ($F_{N_{2}denitrification}$) at each time point was calculated using Equation (7) by dividing the ¹⁵N₂ flux by ¹⁵N enrichment of soil NO₃⁻ pool. The total N₂ flux ($F_{{\mathrm{N}}_2-total}$) was estimated by dividing ¹⁵N flux by the ¹⁵N enrichment of N₂O pool (${}_{}{}^{15}X{}_{{\mathrm{N}}_2\mathrm{O}}$), assuming that N₂ production is from both denitrification-derived N₂O and NH₄⁺ oxidation-derived N₂O [8].

$${}_{}{}^{15}X{}_{{\mathrm{N}}_2}=\frac{{}_{}{}^{29}R+2\times {}_{}{}^{30}R}{2+2\times {}_{}{}^{29}R+2\times {}_{}{}^{30}R}$$

(5)

$$F^{15}{\mathrm{N}}_2=\left({}_{}{}^{15}X{}_{{\mathrm{N}}_2-Tm}-{}_{}{}^{15}X{}_{{\mathrm{N}}_2-T0}\right)\times p_{{\mathrm{N}}_2}\times V/\left(A\times T\right)$$

(6)

$$F_{{\mathrm{N}}_{2}denitrification}=F^{15}{\mathrm{N}}_2/{}_{}{}^{15}X{}_{{{\mathrm{NO}}_3}^{-}}$$

(7)

$$F_{{\mathrm{N}}_2-total}=F^{15}{\mathrm{N}}_2/{}_{}{}^{15}X{}_{{\mathrm{N}}_2\mathrm{O}}$$

(8)

where p_N2 is the density of N₂ at air temperature of 20 °C in the field; V and A are the volume (m³) and basal area (m²) of the chamber.

We did not directly measure the ¹⁵N abundance of soil NO₃⁻ pool in this study. However, it can be calculated based on the ratios of ${}_{}{}^{45}R$ and ${}_{}{}^{46}R$ in N₂O, assuming that there is the same uniformly ¹⁵N-labeled pool of NO₃⁻ for ¹⁵N₂O and ¹⁵N₂ [41]. The detailed calculations can be found in Buchen et al. [40] and Spott et al. [42]. We further calculated N₂O/(N₂O + N₂) ratios from denitrification (R_N2O) according to N₂O_{denitrification} and N_{2denitrification} fluxes. The denitrification rates were present by the sum of N₂O_{denitrification} plus N_{2denitrification} fluxes. The total N₂O/(N₂O + N₂) ratios (R_N2O-total) were also calculated by total N₂ and bulk N₂O fluxes.

In this study, we defined three times the standard deviation as the minimum detectable change for N₂ and N₂O measurements [8]. Ultrahigh-purity N₂ gas (−1.63‰ δ¹⁵N) was manually injected (n = 60), and N₂O standard gas (0.37 ppm, n = 30) was automatically injected, in every batch as quality controls. The detection limit was 4.6 × 10⁻⁷ for ${}_{}{}^{29}R$, 3.9 × 10⁻⁷ for ${}_{}{}^{30}R$, 0.11‰ for δ¹⁵N or 4 × 10⁻⁵ atom% ¹⁵N in N₂, and 7.8 × 10⁻⁶ for ${}_{}{}^{45}R$, 3.8 × 10⁻⁵ for ${}_{}{}^{46}R$ in N₂O. These values were applied to determine if the sample at each time point was significantly different from reference samples (T = 0 h), and if not, they were defined as having no ¹⁵N-N₂ or N₂O production at this time point.

2.4. Soil Parameter Analysis

The soil samples were analyzed for bulk density (BD), soil moisture, pH, total nitrogen (TN), total organic carbon (TOC), and inorganic nitrogen (NH₄⁺, NO₃⁻) contents. Soil NH₄⁺ and NO₃⁻ were extracted with 2 M KCl and measured by an auto discrete analyzer (Smartchem 200, Rome, Italy). The soil pH was determined by a pH meter (PHS-3E, INESA Scientific Instrument Co., Ltd., Shanghai, China) in a 1:2.5 soil water suspension. The TN and TOC contents were determined using an elemental analyzer (Micro Isotope Cube, Langenselbold, Germany). Soil BD was measured using a known volume metal container. Soil gravity water content (SGWC) was quantified by drying in oven at 105 °C for 24 h to a constant weight. The corresponding soil water-filled pore space (WFPS) was calculated based on BD and SGWC ($\mathrm{WFPS}=\frac{\mathrm{SGWC}\times \mathrm{BD}}{1-\mathrm{BD}/2.65}\times 100\%$). The main soil characteristics were shown in Table 1.

2.5. Statistical Analyses

Prior to statistical analysis, data were checked for normality and homogeneity of variance with the Kolmogorov–Smirnov test and Levene’s test, respectively. All statistical analyses were performed with SPSS software (Version 16.0; SPSS Inc., Chicago, IL, USA). One-way ANOVA was used to check the differences in rates of N₂ and N₂O emissions, N₂O/(N₂O + N₂) ratios, and soil properties among slopes followed by multiple comparisons using the LSD method. Paired-samples t-test was applied to determine the differences in soil moisture, NH₄⁺, and NO₃⁻ contents between before and after ¹⁵N labeled incubation. The relationships between rates and ratios of gaseous N emissions and soil properties were examined using Pearson correlation analysis and regression analysis. To identify the dominant factors regulating soil N₂ and N₂O emissions as well as N₂O/(N₂O + N₂) ratios, a stepwise regression analysis was conducted. The statistically significant differences were set at a 0.05 level unless otherwise stated.

3. Results

3.1. Soil Physical-Chemical Parameters

The soil water-filled pore space (WFPS) varied from 28% to 128% and was distinctly different among slopes, with higher values at the lower slope (Table 1). In contrast, soil pH had no significant variation among slopes, except site 1 in 2015. The total average soil BD increased from the lower (0.47 ± 0.03 g cm⁻³) to the upper slope (0.59 ± 0.02 g cm⁻³). Higher average contents of TOC, TN, NH₄⁺ and ratios of C/N, NH₄⁺/NO₃⁻ were observed at the lower slope. The total average NO₃⁻ contents were significantly higher at the upper and middle slopes than at the lower slope, although there were no significant differences among slopes within sites (Table 1). After soil ¹⁵N was labeled for 30 h, no significant differences were found in soil moisture and NH₄⁺ content, except NO₃⁻ content (Figure S2).

3.2. Soil N₂O and N₂ Emissions

Soil bulk N₂O emissions (except site 1 in 2015) peaked at 7 h and then decreased with incubation time (Figure 3A). The average bulk N₂O emission rates from 7 to 30 h decreased more at the upper slope (62%) than the lower and middle slope (8% and 15%, respectively). However, the ¹⁵N₂O emissions at the upper slope did not follow the same pattern as bulk N₂O emissions (Figure 3B). Over the 30 h incubation, the mean bulk N₂O fluxes varied significantly across slopes, with higher values at the lower slope (Figure 4A,

Table 2. The area-weighed N₂ and N₂O fluxes and ratios in the mixed forest of Qingyuan station in Northeast China.

Slope Positions	Area (%)	N2Odenitrification (μg N m−2 h−1)	Bulk N2O (μg N m−2 h−1)	N2denitrification (μg N m−2 h−1)	Total N2 (μg N m−2 h−1)	Denitrification (μg N m−2 h−1)	RN2O	RN2O-total
Upper	80	14 ± 4 c	22 ± 5 c	39 ± 7 c	85 ± 15 c	54 ± 10 c	0.22 ± 0.03 a	0.20 ± 0.02 a
Middle	15	82 ± 21 b	145 ± 32 b	594 ± 205 b	896 ± 243 b	676 ± 220 b	0.20 ± 0.05 a	0.18 ± 0.04 a
Lower	5	246 ± 38 a	258 ± 37 a	3243 ± 330 a	3442 ± 342 a	3488 ± 340 a	0.08 ± 0.01 b	0.08 ± 0.01 b
Total	100	36 ± 8	52 ± 11	282 ± 53	375 ± 66	319 ± 58	0.16 ± 0.02	0.15 ± 0.02

Different lowercase letters indicate significant differences among slopes (p < 0.05). Denitrification: the sum of N₂O_{denitrification} plus N_{2denitrification}; R_N2O: ratio of N₂O_{denitrification} to N₂O_{denitrification} plus N_{2denitrification}; R_N2O-total: ratio of bulk N₂O to bulk N₂O plus total N₂.

). A similar pattern was also observed in the denitrification-derived N₂O fluxes (Figure 4B).

Similar to ¹⁵N₂O, the ¹⁵N₂ emissions with incubation time at each slope position also had different patterns among sites (Figure 3C). The average ¹⁵N₂ emissions at the lower and middle slopes significantly increased with incubation time, while a slight downward trend was observed at the upper slope (Figure 3C). Furthermore, the N₂ flux produced from denitrifying NO₃⁻ pool (N_{2denitrification}) varied from 4 to 4517 μg N m⁻² h⁻¹ among sites, being much higher than N₂O emissions (Figure 4). A large variation in denitrification rate (N₂ + N₂O) was also observed among sites (Figure S3A). Overall, the mean fluxes of N_{2denitrification} and total N₂, as well as denitrification rates, significantly decreased from the lower to the upper slope (Table 2, Figure 4C and Figure S3A).

3.3. Soil N₂O/(N₂O + N₂) Ratio

Significant variations in R_N2O (ratio of N₂O_{denitrification} to N₂O_{denitrification} plus N_{2denitrification}) among slopes were observed in all sites (Figure S3B). This ratio at the lower slope was significantly lower than those at the middle and upper slopes (Table 2). Based on the areas of different soil moisture zones occupied in the study forest, the weighted average R_N2O was 0.16 ± 0.02, being similar to R_N2O-total (ratio of bulk N₂O to bulk N₂O plus total N₂, 0.15 ± 0.02) (Table 2). Further, the area-weighted R_N2O and R_N2O-total were much lower than the value of soil N₂O/(N₂O + N₂) ratio (0.30) from natural forests recompiled by those previously reported literature (Table S1).

3.4. Relationships between Gaseous N Rates, Ratios, and Soil Properties

Soil N₂, N₂O_{denitrification}, bulk N₂O and denitrification rates were all positively correlated with soil WFPS, TOC, C/N and NH₄⁺/NO₃⁻ ratio, while they had a negative correlation with BD (

Table 3. Pearson correlations between soil properties and fluxes, ratios of N₂ and N₂O emissions in the mixed forest of Qingyuan station in Northeast China.

	N2denitrifcation	N2Odenitrifcation	Denitrification	RN2O	RN2O-total	Bulk N2O	Total N2
BD	−0.36 *	−0.34 *	−0.37 *	0.19	0.26	−0.38 *	−0.37 *
pH	−0.24	0.05	−0.22	0.20	0.27	0.10	−0.23
WFPS	0.74 **	0.51 **	0.73 **	−0.39 **	−0.41 **	0.47 **	0.75 **
TOC	0.45 **	0.38 *	0.46 **	−0.16	−0.18	0.42 **	0.45 **
TN	0.29	0.24	0.29	−0.12	−0.14	0.28	0.28
C/N ratio	0.62 **	0.54 **	0.63 **	−0.35 *	−0.39 *	0.52 **	0.63 **
NH4+	0.23	0.19	0.23	0.01	0.05	0.28	0.22
NO3−	−0.21	−0.18	−0.21	0.11	0.03	−0.17	−0.19
NH4+/NO3− ratio	0.31 *	0.43 **	0.33 *	−0.06	0.01	0.46 **	0.30 *

* and ** represent correlation significance at p < 0.05 and p < 0.01, respectively.

). The R_N2O was negatively correlated with soil WFPS and C/N ratio (Table 3). A stepwise regression analysis showed that soil WFPS was the key factor regulating N_{2denitrification}, denitrification rate and R_N2O, accounting for 63%, 63% and 13% of variation, respectively (

Table 4. Multiple linear regression models between fluxes, ratios of N₂ and N₂O from denitrification, bulk N₂O flux and selected soil properties in the mixed forest of Qingyuan station in Northeast China.

Species	WFPS	BD	C/N Ratio	NH4+/NO3− Ratio	R2
N2denitrification	0.72 **	−0.32 **			0.63
N2Odenitrification			0.48 **	0.35 **	0.38
Denitrification	0.71 **	−0.33 **			0.63
RN2O	−0.38 *				0.13
Bulk N2O			0.45 **	0.38 **	0.38

* and ** represent significance at p < 0.05 and p < 0.01, respectively. R square represents the determination coefficient.

). In contrast, the C/N and NH₄⁺/NO₃⁻ ratios mainly affected the emissions of bulk N₂O and N₂O_{denitrification} (Table 4).

4. Discussion

4.1. Variations of Soil N₂ and N₂O Emissions

The ¹⁵N isotope tracer technique allows us to determine in situ N₂ as well as N₂O emissions due to denitrification. In our study, soil N₂ emissions were mainly produced by denitrification (area-weighted: 282 μg N m⁻² h⁻¹) and denitrification accounted for average 75% of total N₂ emissions (area-weighted: 375 μg N m⁻² h⁻¹, Table 2), which was consistent with our previous laboratory findings at the same study site [37]. The mean N₂ fluxes compared well with the rates reported from a woodland forest in the UK [28], a forested wetland and a northern hardwood forest in the US [7,27], using the similar ¹⁵N tracer approach. Our result was also comparable to the rates reported in other temperate forests and maize soils using the gas–flow soil core method [11,13]. Furthermore, the fluxes of both bulk N₂O and N₂O_{denitrification} were significantly lower in comparison with the N₂ fluxes (Table 2, Figure 3 and Figure 4), which was consistent with the results reported from the ¹⁵N tracer studies in temperate forests [27], tropical forests [8,30], and other upland soils [43]. Therefore, our results, together with those of previous studies, suggest that N₂ emissions are likely the main gaseous N loss from terrestrial ecosystems. Moreover, the total denitrification rate (N₂ + N₂O) (54 to 3488 μg N m⁻² h⁻¹, Table 2) was in the range reported in the laboratory ¹⁵N tracer assay [7]. In contrast, this rate was significantly higher than those in other temperate forests from in situ ¹⁵N tracer studies [27,28]. This may be partly explained by the differences in NO₃⁻ and water input. In previous studies, the added NO₃⁻ and water only adjusted within 10% and 5% of the ambient soil NO₃⁻ pool and volumetric water, respectively. In our study, however, soil–extracted NO₃⁻ significantly increased by more than 5 times (Figure S2C). Moreover, about 11 mm precipitation was deposited into the soil, potentially leading to a 10% increase in soil volumetric, although no significant change was observed in soil gravity water content (Figure S2A). One laboratory ¹⁵N study from a tropical rainforest showed an increase in N₂O emission with increasing soil NO₃⁻ [8]. Additionally, high water input to soil may promote the formation of an anaerobic environment, potentially promoting the further reduction of the produced N₂O to N₂ during denitrification [30]. The significantly higher N₂ rather than N₂O flux from denitrification, was confirmed in our study (Table 2, Figure 3 and Figure 4).

In this study, we found the highest N₂ and N₂O emissions or denitrification rates occurring at the lower slope (Table 2, Figure 4 and Figure S3A). This was partly in line with our expectations. Topography plays an important role in the redistribution of water and substrates (available N or C) which can influence gaseous N emissions [17]. Substrates are prone to accumulate in the bottom position with the transportation of water [31,44]. In the present study, the lower slope had higher soil WFPS, TOC, TN, NH₄⁺ contents and lower soil BD compared to the upper slope, and the variation in soil WFPS across slopes was significantly larger than other soil parameters (Table 1). Previous studies indicated that soil water could affect the production of N gas from nitrification and denitrification by regulating O₂ and/or substrate availability [31,45]. In this study, we found that soil WFPS was significantly positively correlated with TOC and C/N ratio (Table S2). Furthermore, good positive relationships between soil WFPS and N_{2denitrification} flux, denitrification rate were observed, although soil BD had a negative contribution to denitrification rate and N_{2denitrification} flux (Table 4). It suggests that the change in soil water associated with bulk density is important factor driving the difference in denitrification across slopes. Additionally, it was noted that the underestimated N₂O and N₂ fluxes may occur in the current study due to the gas diffusion from soil surface to subsoil. Previous study demonstrated a large underestimation of denitrification rate from the soil surface using static chamber method and the underestimated extent could decrease with the increase in soil moisture [46]. In the present study, we speculated that the N₂ and N₂O fluxes measured at the lower slope were likely to reflect real values since soil water was oversaturated in this location (110% WFPS). However, these fluxes will be significantly underestimated at the middle and upper slopes, where soil WFPS were 40% and 60%, respectively, slightly lowering the values reported by Well et al. [46]. Nevertheless, our field results are consistent with previous laboratory findings, where high soil moisture corresponded to high N₂ and N₂O emissions [7,9]. These results confirm that the change in soil moisture could moderate the variations of N₂ and N₂O emissions.

In addition to soil moisture, the relative proportion of N₂ and N₂O was affected by N substrates [7,8]. The N availability, such as NO₃⁻ and NH₄⁺, is the primary requirement for denitrification and nitrification [17]. The high nitrification rates and NO₃⁻ contents were reported at lower slope sites in a temperate coniferous forest [44]. Previous assays indicated the simultaneous increase in N₂ and N₂O production with the increase in NO₃⁻ concentration [8,12]. However, our study showed a high NO₃⁻ content at the middle and upper slopes, which did not have the highest gaseous N emissions (Table 1, Figure 4). Alternatively, we found that soil N₂O emissions were affected by the NH₄⁺/NO₃⁻ ratio (Table 3 and Table 4), which was inconsistent with previous findings. It was likely that the relative proportion of substrate N species in comparison to their contents may be more important to N₂O rather than N₂ emissions. Moreover, the C/N ratio significantly affected soil bulk N₂O and N₂O_{denitrification} fluxes (Table 4), although soil TOC, C/N, or bulk density was individually observed to correlate with these fluxes (Table 3). These relationships indicate that the changes in soil parameters may have the different effects on N₂ and N₂O emissions. In addition, the temporal variation in N₂O and N₂ emissions was larger at the upper and middle slopes within the site (Figure 4). This suggests that the in situ denitrification rates may change with the seasons or years [28]. Therefore, the dynamics of in situ soil N₂ and N₂O fluxes should be addressed in the future.

4.2. Comparison of Soil N₂O/(N₂O + N₂) Ratios from Denitrification

Spatial variations of N₂ and N₂O emissions in this study were expected. Commonly measured N₂O fluxes are widely available, yet to accurately estimate denitrification rates requires a better understanding of the controls on N₂O/(N₂O + N₂) ratios from denitrification (R_N2O). In the current study, lower R_N2O were associated with high soil moisture (Table 2, Figure S3B). Similar findings were also observed from other temperate forests in laboratory assays using ¹⁵N labeling [7] or the direct gas flux method [11]. Increasing soil moisture will decrease O₂ diffusion, most likely forming anaerobic conditions for denitrification, and it is expected that the R_N2O will decrease [47]. Besides, NO₃⁻ availability, as the electron acceptor of denitrification, also affected its product ratio [13]. However, our results demonstrated that this ratio was mainly regulated by soil moisture (Table 4), further supporting our previous hypothesis. Moreover, the area-weighted average R_N2O from the entire forest (0.16, Table 2) was similar to the value from the ¹⁵N gas flux method reviewed by Scheer et al. [24]. In contrast, our result was significantly lower than previously reviewed global ratios in upland soils based on approaches of ¹⁵N trace and acetylene inhibition [16], while it was higher compared to previous field ¹⁵N trace studies on temperate forests (e.g., 0.008 and 0.01 in Refs. [27,28], respectively). We further calculated the R_N2O from forest soils, with an average value of 0.30 (Table S1), by compiling these data from previous research [16,24] and our study. This reflected the great spatial variability of R_N2O, and its dynamic change urgently requires to be clarified. Nevertheless, these results suggest that N₂ fluxes from forest ecosystems are likely to be underestimated by around one-third when using the N₂O/(N₂O + N₂) ratio in nature soils reported by Schlesinger [16].

Previous studies indicated the variations in the N₂O/(N₂O + N₂) ratio due to the difference in N₂ measurement among methods [41]. Based on the literature compiled with the current study, we observed a significant difference in N₂O/(N₂O + N₂) ratios among methods, with the highest values observed by the acetylene inhibition method and the lowest values by the ¹⁵N tracer method (Figure S4). As reported in previous studies, each method has a special application range and its advantage or disadvantage [11,48]. For instance, the ¹⁵N-gas flux method requires reasonable amounts of labeled NO₃⁻ introduced to soil systems, which may disturb the soil micro−environment and stimulate microbial N turnover [11], affecting soil N₂ emission. Therefore, the approach used in this study also had some bias for N₂ measurement. These bias may be linked to the amount of added NO₃⁻ and water, labeled soil depth, enclosure time and volume of the chamber, and even the precision of the instrument [8,41,46,48]. In our study, soil NO₃⁻ and moisture were higher than the common conditions (Figure S2A,C), although a high NO₃⁻ content (i.e., >20 mg N kg⁻¹) was often observed in some natural temperate forests in the same region as our study [49]. As discussed above, the potential interaction of NO₃⁻ and water regulated the proportion of N₂ and N₂O emissions, leading to a lower N₂O/(N₂O + N₂) ratio than the real ratios. As a result, it was likely to overestimate N₂ emission and underestimate R_N2O. In the preliminary experiment, we did not find a significant change in the δ¹⁵N-N₂ value between the labeled sample (24 h) and ambient air (0 h) after the soil was labeled with ¹⁵NO₃⁻ solution at a rate of 0.25 g ¹⁵N m⁻² (data not shown). Furthermore, an unknown proportion of downward diffusion of ¹⁵N−labeled gases to the non−labeled subsoil might cause the bias in N₂ and N₂O fluxes [41,46]. The extend closure time of chamber would further add the variations. Although the fluxes of ¹⁵N₂ and ¹⁵N₂O at all the slopes were found to increase during 30 h incubation (Figure 3B,C), the ¹⁵N gas diffusion to the subsoil (below 10 cm) could occur, especially on the middle and upper slopes with larger air-filled porosity (Table 1). Moreover, the large volume of the chamber (18 L) likely diluted the produced ¹⁵N₂ with air ¹⁴N. Consequently, it could lead to an underestimation of N₂ fluxes. These were thus responsible for the observation of low or undetectable N₂ flux explained in part due to these effects (Figure S5). In addition, quantifying the limit of detection for the ¹⁵N gas flux method is still a challenge in low N₂ emissions in nature forest ecosystems, although the detection sensitivity of our IRMS (0.79 μg ¹⁵N m⁻² h⁻¹) is better compared to previous studies [8,27,28]. As addressed above, undoubtedly, there are still some other bias regarding the accuracy of N₂ loss due to the interactions of multiple and unpredictable factors. Therefore, some improvements could be explored to reduce these bias for the in situ ¹⁵N tracer method. For example, one way to improve detection is to reduce the N₂ background concentration in the field by flushing chambers with a N₂–free air [20]. Additionally, based on this, shorting chamber height and sampling time can further improve the N₂ analytical accuracy. Overall, more field studies on N₂ emissions and the N₂O/(N₂O + N₂) ratios from forest ecosystems with ¹⁵N tracer method are required in the future.

4.3. Implications for Ecosystem N Loss

In this study, about 5.1 kg N ha⁻¹ yr⁻¹ of total N₂ flux was estimated based on the area–weighted total N₂O/(N₂O + N₂) ratio (0.15, Table 2) and the field average N₂O flux with 0.90 kg N ha⁻¹ yr⁻¹ monitored in 2019 to 2021 (Huang K. et al., unpublished). The actual N₂ fluxes were probably higher than this estimate because the chambers for monitoring field N₂O flux were mostly set at the upper and middle slope positions, where N₂O emissions were relatively lower than those at the lower slope position (Table 2). The total gaseous N emissions (N₂ + N₂O) were calculated at about 6.0 kg N ha⁻¹ yr⁻¹, accounting for 30% of the inorganic N (NH₄⁺ plus NO₃⁻, 20 kg N ha⁻¹ yr⁻¹) deposition in precipitation, as reported in this study forest by Huang et al. [35]. This result was comparable to those from Hawaiian rainforests [50,51] and temperate forests [52] estimated by the natural nitrogen and/or oxygen stable isotope approach. In all, these findings suggest that the gaseous N losses from the forest ecosystems may be larger than commonly thought, and the N cycle is likely to be more open.

The present study still have some uncertainties in estimating soil N₂ flux with measured N₂O/(N₂O + N₂) ratio, including the bias from the ¹⁵N–gas flux method as mentioned above. For instance, previous studies indicated the variations in N₂O/(N₂O + N₂) ratios during different growing periods [26,28]. In the current study, we carried out the ¹⁵N trace experiment in the middle period of the growing season (April to October) and monitored the N gas emission once. In this period, soil temperature was high which stimulated microbial activity to consume considerable substrates for nitrification and denitrification. These processes in turn promoted gaseous N losses from soil. In the early or later phase, however, plant growth and microbial activities may decrease due to temperature limitations. Low available substrates could affect microbial enzyme activity and community composition, causing an increase or decrease in N₂O/(N₂O + N₂) ratios [17]. As a result, the calculated N₂ loss was under– or over–estimated. Moreover, the change in soil moisture could affect the diffusion of N₂ and N₂O produced in soil pore space [41], leading to the differences in the N₂O/(N₂O + N₂) ratios among different periods. It was found that field soil WFPS changed from 24% to 73% throughout the growing season in our study site [34]. Previous studies showed a low N₂O/(N₂O + N₂) ratio with high soil moisture [7,15], being similar to our result. The effect extent of soil moisture on N₂O/(N₂O + N₂) ratio remains unclear and should be further clarified in the following studies. Additionally, the unquantified other microbial processes from N₂ and N₂O emissions are also contributed to the uncertainty in N₂O/(N₂O + N₂) ratios. More field research on differentiating production pathways of N₂ and N₂O emissions from forest ecosystems are urgently required for more robust estimates of gaseous N losses [53].

5. Conclusions

The results of this study showed that in situ soil N₂ and N₂O emissions significantly varied across slopes and were strongly affected by soil moisture. Soil N₂ emissions were significantly higher compared to N₂O emissions, accounting for 85% of gaseous N losses (N₂ + N₂O). The combined field N₂O flux and N₂O/(N₂O + N₂) ratio is likely to be a promising tool to quantify soil N₂ flux and even denitrification rate for a given forest ecosystem. In addition, we recognize that there are some bias for determining in situ soil N₂ emission and therefore the calculated N₂O/(N₂O + N₂) ratio in the current ¹⁵N trace study, such as the added ¹⁵NO₃⁻ amount, enclosure time of the chamber, and gas diffusion to the subsoil. Overall, to further elucidate the dynamic of soil N₂ and N₂O emissions associated with N₂O/(N₂O + N₂) ratios, in situ ¹⁵N–labeled experiments in different seasons from different forest ecosystems should be performed. Meanwhile, we also need to refine and combine established methodologies using models for better predicting ecosystem denitrification rate and gaseous N losses.