Asymmetric Response of Grassland Greenhouse Gases to Nitrogen Addition: A Global Meta-Analysis

Abstract

Grassland ecosystems, a major component of the global carbon (C) and nitrogen (N) cycles, are increasingly impacted by anthropogenic N addition. However, a comprehensive, integrated assessment of all three major greenhouse gas (GHG) responses in grasslands is lacking. Here, we present the first global meta-analysis to evaluate the effects of N addition on all three major GHGs (i.e., nitrous oxide (N₂O), methane (CH₄), and carbon dioxide (CO₂) fluxes) in grasslands. Our results show that N addition significantly and consistently stimulates N₂O emissions, a response primarily modulated by key drivers such as grassland type, management, N addition rate and forms, humidity index (HI), and soil pH, clay, and total nitrogen (TN) content. In contrast, N addition has a minimal and non-significant overall effect on soil CO₂ fluxes. For CH₄, N addition causes a context-dependent reduction in uptake, an effect that is exacerbated by high mean annual precipitation (MAP) and soil bulk density (BD) but alleviated by high soil organic carbon (SOC) content. Notably, both CO₂ and N₂O showed a dose-dependent effect, while soil CO₂ fluxes were unexpectedly suppressed by nitrate nitrogen (NO₃⁻) addition. Our findings indicate that the pronounced and consistent increase in N₂O emissions is the dominant factor in GHG-related impacts in grasslands, implying a net positive climate forcing in grasslands from N enrichment, even if there is insufficient data to calculate net climate forcing directly. Our study highlights the heterogeneous nature of grassland GHG responses and provides critical insights for developing sustainable N management strategies to mitigate climate change.

1. Introduction

Grasslands, which make up more than 40% of the Earth’s land surface, represent a critically important component of the global carbon (C) and nitrogen (N) cycles [1,2,3,4]. These enormous ecosystems play a crucial role in regulating global greenhouse gas (GHG) budgets, serving as large stores of soil carbon and as active sinks or sources of atmospheric GHGs, such as nitrous oxide (N₂O), methane (CH₄), and carbon dioxide (CO₂) [5,6,7,8]. However, anthropogenic activities, especially the widespread application of synthetic fertilizers in agriculture and the combustion of fossil fuels, have dramatically increased the amount of reactive N that has been deposited into terrestrial ecosystems [9,10,11]. This increased N input significantly changes the biogeochemical processes within grassland ecosystems, potentially affecting their natural ability to slow down climate change or, on the other hand, aggravating it by changing GHG fluxes [5,8,12,13,14]. Comprehending these effects is crucial for forecasting future climate scenarios and developing effective mitigation strategies.

The impacts of N addition on individual GHGs (e.g., N₂O emissions, CH₄ uptake/emission, or CO₂ fluxes) in grasslands have been the subject of numerous individual studies or meta-analyses [15,16,17,18,19,20,21,22,23] but there is an absence of thorough, integrated evaluations of their concurrent responses to N enrichment. Relying solely on isolated investigations of individual GHGs often obscures potential trade-offs or synergistic effects among them, thereby hindering a holistic understanding of ecosystem-scale climate feedback [17,18,21,22,24,25,26]. For instance, an increase in the emission of one GHG might be partially or entirely offset by a decrease in another, or vice versa, leading to a complex net effect on radiative forcing that single-gas studies cannot fully capture [17,18,21,22,24]. An enhanced integrated perspective is of the utmost necessity for accurately assessing the genuine climate footprint of N deposition.

Existing syntheses addressing N deposition and GHG fluxes typically suffer from critical limitations. Many either focus exclusively on the response of a single GHG [27,28,29,30,31,32] or aggregate data across a wide diversity of terrestrial ecosystems, such as forests, wetlands, and croplands, despite each having inherently varying ecological characteristics [26]. Although useful for broad trends, this wide aggregate data frequently obscures distinctive, context-dependent reactions that are specific to grasslands. Because grasslands differ from other biomes in terms of soil–plant–microbe interactions, hydrological regimes, and C/N cycling pathways, a thorough and targeted investigation is required to precisely identify the precise processes behind their GHG responses to N enrichment [5,6,7,8]. In the absence of such a targeted approach, the precise climate forcing caused by anthropogenic N deposition in grasslands, a globally important ecosystem, remains ambiguous and inadequately quantified.

To address the aforementioned knowledge gaps, we conduct the first global meta-analysis that is specifically designed to assess the responses of grassland ecosystems to N addition in relation to all three major GHGs: CO₂, CH₄, and N₂O fluxes. Based on the current mechanistic understanding and our preliminary findings, we hypothesize that N enrichment will (i) consistently stimulate N₂O emissions by enhancing microbial nitrification and denitrification processes [33,34,35,36]; (ii) exert minimal net effects on soil CO₂ fluxes at the ecosystem level, primarily due to potentially offsetting changes between stimulated microbial activity and N-induced negative effects on root and microbial biomass [18,37]; and (iii) reduce CH₄ uptake by suppressing methanotrophic activity in non-saturated soils [15,18,19,28,32,35]. To test these hypotheses and provide a comprehensive assessment, we systematically synthesized dispersed evidence across a wide range of climatic gradients, soil properties, and management regimes. This approach allows us to quantify the net climate forcing of anthropogenic N enrichment on grassland ecosystems and to identify key environmental (i.e., edaphic properties and climatic variables) and experimental drivers (e.g., N addition rate, form of N addition) that mediate these complex GHG responses. The insights gained from this focused and integrated synthesis will provide critical information for future research directions, improve the accuracy of global biogeochemical models, and contribute directly to the development of sustainable N management strategies tailored for grassland ecosystems, ultimately supporting efforts to mitigate climate change.

2. Materials and Methods

2.1. Data Compilation

We conducted a comprehensive literature search to identify peer-reviewed studies investigating the effects of experimental N addition on N₂O, CH₄, and CO₂ fluxes in grassland ecosystems worldwide. Our search was performed using Web of Science, Google Scholar, and the China Knowledge Resource Integrated Database (http://www.cnki.net/, accessed on 30 June 2022) with combinations of keywords such as “nitrogen addition”, “nitrogen deposition”, “grassland”, “meadow”, “pasture”, “steppe”, “savanna”, “N₂O”, “nitrous oxide”, “nitrification”, “denitrification”, “CH₄”, “methane”, “methanogenesis”, “methane oxidation”, “CO₂”, “carbon dioxide”, “respiration”, “greenhouse gas”, “emission”, “flux”, and “uptake”. The search was limited to studies published before the year 2022. To avoid selection bias, studies included in the meta-analysis had to meet the following criteria: (1) Experiments were conducted in situ rather than being laboratory incubation or greenhouse-based experiments. (2) Studies must have been conducted with N addition only, with both a N addition treatment and a corresponding control plot. Any N addition experiments conducted along with phosphorus addition, warming, elevated CO₂, water, or drought were excluded. (3) Data on N₂O, CH₄, or CO₂ fluxes could be extracted directly from text, tables, or digitized graphs as means, standard deviations (or standard errors), and sample sizes for both treatment and control groups. (4) Fluxes were measured over a growing season or an annual cycle, or sufficiently long to capture meaningful responses. (5) Sufficient metadata were provided for each site, including geographical location, climate zone, soil properties (e.g., soil pH, organic carbon, and total nitrogen content), N addition rate and forms, and duration of the experiment.

Finally, a total of 810 paired observations from 172 peer-reviewed articles met the selection criteria. The distribution of the sampling sites was shown in Figure 1. All the data were systematically extracted from the text, tables, figures, and supplemental materials of the selected publications. When data were presented graphically, we used GetData Graph Digitizer (http://www.getdata-graph-digitizer.com/, accessed on 1 June 2025) to digitize and extract numerical values. For each study, we recorded mean values, sample sizes (n), and measures of variability (standard deviation, SD, or standard error, SE) for GHG fluxes from both control and N addition treatments. When SE was reported, it was converted to SD using the equation SD = SE × $\sqrt{n}$. Unidentified error bars in figures were conservatively assumed to represent SE. Where studies reported multiple N addition rates or multi-year datasets, each unique N-rate-by-year combination was treated as an independent observation to maximize data resolution while maintaining statistical independence.

In addition, we further extracted three categories of auxiliary variables to support moderator analysis. Experimental parameters including site location (e.g., latitude and longitude), grassland type (natural or artificial), grassland management (no grazing or mowing, enclosure, ungrazing, grazing, mowing, or grazing and mowing), experimental duration, N addition rate (kg N ha⁻¹ yr⁻¹), and form (e.g., ammonium (NH₄⁺), ammonium nitrate (NH₄NO₃), nitrate (NO₃⁻), urea, and organic nitrogen (ON)). Climatic variables (including mean annual temperature (MAT, units: °C)/precipitation (MAP, units: mm) and humidity index (HI)) were sourced primarily from original studies, supplemented by the Climatic Research Unit dataset (CRU TS4.07; https://www.worldclim.org/, accessed on 1 June 2025) using site coordinates and experimental years when unavailable. Baseline soil properties (including soil bulk density (BD, units: g cm⁻³), clay content (Clay, units: %), pH, organic carbon (SOC, units: %), total nitrogen (TN, units: %), and soil temperature (ST units: °C)) were obtained from study texts, tables, and figures or derived via geospatial matching and experimental years with the Harmonized World Soil Database (HWSD v2.0; https://www.fao.org/soils-portal/data-hub/soil-maps-and-databases/harmonized-world-soil-database-v12/en/, accessed on 1 June 2025) and ECMWF Reanalysis v5 (ERA5; https://cds.climate.copernicus.eu/datasets/reanalysis-era5-land?tab=overview, accessed on 1 June 2025).

2.2. Meta-Analysis Statistical Procedures

The effects of N addition on grassland soil GHG (N₂O, CH₄, and CO₂) fluxes were evaluated using the natural logarithmic form of the response ratio [38]. The response ratio (RR) was calculated as the ratio of the mean GHG flux in the N-added treatment group ($X_T$) to that in the control group ($X_C$):

$$RR={\displaystyle \frac{X_T}{X_C}}.$$

The lnRR was then calculated as

$$\mathrm{l}\mathrm{n}\left(RR\right)={\displaystyle \frac{\overline{X_T}}{\overline{X_C}}}.$$

The variance (v) and weighted amount (W) for each lnRR [38] were calculated using the following formula:

$$v={\displaystyle \frac{{SD}_T^2}{n_T{X}_T^2}}+{\displaystyle \frac{{SD}_C^2}{n_C{X}_C^2}}$$

$$w={\displaystyle \frac{1}{v}}$$

where SD_T and SD_C are the standard deviations, and n_T and n_C are the sample sizes for the treatment and control groups, respectively.

The overall mean effect size and 95% confidence intervals (CIs) for each GHG was estimated using a random-effect model, which was fitted using the rma.mv function in the “metafor” package in R software 4.3.2 (R Development Core Team). Due to the varying units and magnitudes across studies, the effect size metric was the natural logarithm of the response ratio (lnRR) weighted by its inverse variance (wi = 1/vi) [39]. Egger’s regression test and funnel plots were used to evaluate the publication bias. Cochran’s Q statistic and the I² index were used to evaluate the heterogeneity among studies. If an effect’s 95% confidence interval (CI) did not overlap with zero (for lnRR), it was deemed statistically significant.

Subgroup analyses were conducted to investigate the influence of various moderator variables, including grassland type (natural or artificial), grassland management (no grazing or mowing, enclosure, ungrazing, grazing, mowing, or grazing and mowing), N addition rate (e.g., low: <50 kg N ha⁻¹yr⁻¹, medium: 50−100 kg N ha⁻¹yr⁻¹, high: >100 kg N ha⁻¹yr⁻¹), N addition forms (e.g., NH₄⁺, NH₄NO₃, NO₃⁻, urea, ON), MAP (e.g., low: <400 mm, high: >400 mm), MAT (e.g., low: <0 °C, high: >0 °C), HI (e.g., arid: <0.2, semi-arid: 0.20–0.50, sub-humid: 0.50–0.65, humid: >0.65), soil pH (e.g., acid soil: <7, alkaline soil: >7), SOC (e.g., low: <10 g kg⁻¹, medium: 10−30 g kg⁻¹, high: >30 g kg⁻¹), and clay content (e.g., low: <15%, high: >15%).

To identify key environmental and experimental factors influencing the response of grassland GHGs to N addition, we employed a multi-model inference approach based on a multivariate meta-regressive framework. Prior to analysis, potential continuous moderators (including N application rate, soil BD, clay content, SOC, TN, soil pH, ST, MAP, MAT, and HI) were standardized (z-scored). A two-step was used to address the multicollinearity. Firstly, variables with absolute weighted correlations over 0.8 were iteratively eliminated by removing the variable with the highest sum of absolute correlations among highly correlated pairings. After that, a weighted linear model was then used to remove variables with variance inflation factor (VIF) values higher than 10. This process began with the variables with the highest VIF and continued until all variables were below the threshold. These procedures ensured that the independent variables chosen for subsequent models were not excessively collinear. As for categorical moderators, grassland type, management, and N addition forms were considered.

A multi-model inference approach was then used to identify the best-fitting models and quantify the relative importance of each moderator [40]. All possible combinations of the abovementioned continuous moderators and the categorical variables were used to construct candidate meta-regression models using the rma.mv function in the “metafor” package. Reference as random effect was included in all models to account for non-independence of multiple observations from the same study. The general model structure was as follows:

$$\ln {RR}_{ij}={\alpha }_0+{\beta }_0{+\beta }_1{X}_{1ij}+{\beta }_2{X}_{2ij}+\cdots +{\beta }_k{X}_{kij}+{\epsilon }_{ij}$$

where $\ln {RR}_{ij}$ is the j effect size from the i reference, ${\beta }_0$ is the overall intercept, ${\beta }_1\cdots {\beta }_k$ are the coefficients for the moderators with $X_1\cdots X_k$, ${\alpha }_0$ is the random effect associated with the i reference, and ${\epsilon }_{ij}$ is the sampling error, with ${\epsilon }_{ij}~N\left(0,{\tau }^2\right)\mathrm{a}\mathrm{n}\mathrm{d}{\alpha }_0~N\left({\mu }_0,{\sigma }_0^2\right)$.

Models were estimated using the REML method. Model performance was compared using the Akaike information criterion corrected for small sample sizes (AICc), and the proportion of heterogeneity explained (R²) was calculated for each. Model weights, based on ΔAICc, were used to determine the relative importance of each moderator variable by summing the weights of all models in which that variable appeared.

For individual continuous moderator variables exhibiting a relative importance value greater than 0.9, we conducted separate univariate meta-regression models to visually depict their associations with effect sizes. Using point sizes scaled by the inverse of their variance, the estimated regression line and its 95% CI were shown against the observed effect sizes. The QM test was used to determine the significance of the moderator effect [39].

3. Results

3.1. Response of Grassland Soil GHG Fluxes to N Addition

Nitrogen enrichment elicited divergent responses of soil GHG emissions to N addition in global grasslands (Figure 2, Figure 3 and Figure 4). Overall, N addition significantly increased N₂O emissions, with an effect size (lnRR) of 0.77 (95% CI: 0.55, 1.00; n = 340; p < 0.001) (Figure 2). In contrast, N addition exerted no significant overall effect on CO₂ fluxes (mean: 0.06; 95% CI: −0.11, 0.22; n = 175; p = 0.80) (Figure 3), while it showed a non-significant reduction on CH₄ uptake (mean: −0.07; 95% CI: −0.14, 0.01; n = 295; p = 0.12) (Figure 4).

Since there was significant residual heterogeneity in the global grassland N₂O emissions (Qb = 86,446.83; p < 0.001; n = 340; Figure 2), we attempted to explain the heterogeneity by considering different factors. Among the three categories of moderators (e.g., experimental, climatic, and edaphic), grassland type, management, N addition rate, N form, MAP, MAT, HI, pH, SOC, and clay had significant effects on the change in soil N₂O emissions induced by N enrichment (p < 0.01;

Table 1. The results of meta-regressions for grassland soil GHG fluxes under different grassland type, management, N addition, and climatic and edaphic conditions.

GHGs	Variable	Meta-Regression Results
		Qm	p Value
N2O	Type	247.2	<0.001 ***
	Management	394.44	<0.001 ***
	N addition rate	2393.51	<0.001 ***
	N addition forms	1436.3	<0.001 ***
	MAP	47.46	<0.001 ***
	MAT	49.66	<0.001 ***
	pH	88.08	<0.001 ***
	SOC	104.13	<0.001 ***
	Clay	74.6	<0.001 ***
	HI	65.97	<0.001 ***
CO2	Type	2.24	0.33
	Management	23.66	<0.001 ***
	N addition rate	2,571,277.15	<0.001 ***
	N addition forms	13.67	0.01 **
	MAP	0.30	0.86
	MAT	1.99	0.37
	pH	3.34	0.19
	SOC	1.48	0.69
	Clay	0.50	0.78
	HI	24.89	<0.001 ***
CH4	Type	14.24	<0.001 ***
	Management	92.29	<0.001 ***
	N addition rate	6.32	0.1
	N addition forms	19.03	<0.001 ***
	MAP	2.92	0.23
	MAT	28.35	<0.001 ***
	pH	3.03	0.22
	SOC	12.48	0.01 **
	Clay	3.09	0.21
	HI	13.20	0.01 *

Note: MAT, MAP, and HI refer to mean annual temperature (units: °C), mean annual precipitation (units: mm), and humidity index, respectively. Edaphic properties include soil pH, organic carbon (SOC, units: %), and clay content (Clay, units: %). ***, **, and * indicate at 0.001, 0.01 and 0.05 significance, respectively.

). Subsequent subgroup analysis revealed significant positive responses (p < 0.05) with the exception of artificial grasslands (−0.61 ± 0.18, p < 0.001), grazing management (−1.51 ± 0.28, p < 0.001), subzero temperatures (MAP < 0 °C: 0.46 ± 0.24, p = 0.05) and arid regions (0.79 ± 0.64, p = 0.22). The strongest stimulation was found under mowing management (1.59 ± 0.25), high N addition rate (>100 kg N ha⁻¹ yr⁻¹: 1.50 ± 0.11), NH₄⁺-based and organic N addition forms (NH₄⁺: 1.41 ± 0.16; ON: 2.12 ± 0.10;), acidic (pH < 7: 0.90 ± 0.11), moderate SOC (1.01 ± 0.13), and high-clay-content (0.92 ± 0.12) soils.

Although the overall effect of N addition on CO₂ flux was neutral, significant residual heterogeneity was detected (Qb = 22,035,843.06; p < 0.0001; n = 175; Figure 3). Meta-regression showed that the management of grasslands, N addition amount and forms, and HI had significant effects on the change in soil CO₂ fluxes (p < 0.01; Table 1). The majority of subgroups showed no significant change in CO2 fluxes when N was added, but five conditions induced significant changes: enclosure management (lnRR = 0.43, 95% CI: 0.20, 0.67; n = 17, p = 0.02) and N addition in sub-humid (HI: 0.50–0.65; lnRR = 0.28, 95% CI: 0.06, 0.49; n = 19, p = 0.01) and humid regions (HI: >0.65; lnRR = 0.33, 95% CI: 0.09, 0.56; n = 40, p = 0.006) caused increases, while grazing management (lnRR = −0.90, 95% CI: −1.49, −0.31; n = 17, p = 0.02) and the NO₃⁻ addition form caused significant decreases in CO₂ fluxes (lnRR = −0.56, 95% CI: −1.03, −0.08; n = 4, p = 0.02).

For CH₄, despite the non-significant overall effect, significant residual heterogeneity was detected (Qb = 1598.21; p < 0.001; n = 295; Figure 4). Meta-regression showed that type of grassland and management, MAP, SOC, N addition forms, and HI had significant effects on the change in soil CH₄ fluxes (p < 0.01; Table 1). Except for the significant increased effect of grazing and mowing management, subgroup analysis further elucidated the significantly inhibited influence on CH₄ uptake, in particular low N addition rates (<50 kg N/ha; lnRR = −0.09 ± 0.04, n = 152), N addition forms such as NH₄NO₃ (lnRR = −0.11 ± 0.05, n = 62), specific environmental conditions like both hotter-temperature (MAT > 0 °C; lnRR = −0.17 ± 0.05, n = 181) and cooler-temperature (MAT < 0 °C; lnRR = 0.14 ± 0.06, n = 114), dry (HI < 0.20; lnRR = −0.22 ± 0.06, n = 22) climates, and soils with high SOC content (>30 g/kg; lnRR = −0.16 ± 0.05, n = 119).

3.2. Factors Driving the Response of Grassland Soil GHG Fluxes to N Addition

To identify the key environmental and experimental factors influencing grassland GHG responses to N addition, a multi-model inference approach was employed to evaluate the relative importance of potential moderator variables (Figure 5). The response of N addition on N₂O fluxes were predominantly driven by N addition forms, HI, N addition rate, management and type of grassland, soil pH, soil TN, and clay content, with importance values greater than 0.9. This indicates that the magnitude of N₂O response to N addition is primarily shaped by a combination of climatic conditions, N application specifics, management and type of grassland, and fundamental soil properties (Figure 5a). However, the most influential factors (importance ≥ 0.8) in CO₂ flux responses to N addition were identified as the rate and forms of N addition, as well as the management of grasslands. This suggests that for CO₂, variables related to N addition and management of grasslands are more dominant drivers compared to climatic or inherent soil characteristics (Figure 5b). For CH₄ fluxes, the dominant variables (importance ≥ 0.7) were the management and type of grassland, N addition forms, soil BD, and MAP. This highlights the critical role that soil physical properties, moisture availability, N addition forms, and grassland type and management play in modulating CH₄ responses to N addition (Figure 5c).

The relationships between these critically important moderator variables and the effect sizes of respective GHGs are further illustrated in scatter plots (Figure 6). Alongside the plots, the QM test results, slopes, and p-values are also presented. For N₂O emissions, the effect size showed significant positive slopes with increasing HI, N addition rate, and soil TN and clay content (p < 0.01), indicating that higher values of these factors generally amplify N-induced N₂O emissions. A significant and non-linear relationship was also observed for pH (p < 0.001), indicating that N₂O emissions decline as pH diverges further from a specific optimum value. For CO₂ emissions, the N addition rate showed a positive slope, but this relationship was not statistically significant (p = 0.069). These specific relationships highlight the complex interplay between N addition and various environmental and experimental factors in regulating grassland CO₂ flux dynamics. For CH₄ uptake, the effect size exhibited significant negative slopes with soil BD (p = 0.02), MAP (p < 0.001), and clay content (p < 0.001), implying that higher soil BD, MAP, and clay content tend to reduce CH₄ uptake in response to N addition. Conversely, a significant positive slope was found for SOC (p = 0.005), indicating that higher SOC content tends to promote CH₄ uptake or alleviate the N-induced suppression of CH₄ uptake.

4. Discussion

Through a global meta-analysis, we provide a comprehensive and quantitative assessment of how N addition affects fluxes in the three major GHGs in grassland ecosystems. The results reveal heterogeneous but mechanistically explainable patterns across gases, aligning with theoretical expectations and previous studies [26,27,28,30,31,32,33,37,41,42]. Specifically, it was found that N enrichment significantly enhanced N₂O emissions, had a minimal effect on CO₂ fluxes, and decreased CH₄ uptake in a way that depended on context. This observed variation in GHG responses was predominantly mediated by key experimental and environmental factors, as demonstrated by multi-model inference and meta-regression analyses.

The broad stimulation of N₂O emissions (+116.86 ± 12.03%) in response to N addition is a significant and reliable finding that supports previous earlier plot-level/localized and regional syntheses [18,21,22,23,26,27,30,33,41,42]. The primary underlying mechanism involves enhanced substrate availability for nitrification and denitrification, which were the main microbial pathways for N₂O production in grassland soils [34,36]. In addition, the coupled influence of soil moisture conditions and aeration was also detected in shaping the response ratio of N₂O emissions to N addition, as indicated by the relative importance results [43,44]. A significant quadratic relationship with soil pH indicated a reverse-U-shaped response, which likely reflected niche partitioning between acidophilic nitrifiers and alkaliphilic denitrifiers [45]. Furthermore, under low-temperature conditions (especially in subzero climates), N₂O responses were substantially weakened or nullified, indicating temperature-mediated decoupling of substrate availability and microbial activity [46]. In particular, N addition inhibited the response of N₂O under artificial grassland and grazing management, which may be explained by a shift from N to carbon in the primary limiting factor for microbial processes. Specifically, grazing management practices can induce carbon limitation by reducing the quantity and/or quality of labile carbon inputs [47]. Additionally, grazing-induced soil compaction may increase N loss via leaching [48], which would thereby decrease the substrate in the soil profile that is accessible for nitrification and denitrification. Lastly, these grassland systems may reach N saturation as a result of prolonged exposure to N inputs from fertilization or animal feces [49], which would reduce the responsiveness of N cycling to additional N addition.

In contrast, the relatively minor net effect of N addition on soil CO₂ fluxes (soil respiration) probably results from a balance between several underlying processes [18,26,37,50]. While N inputs can stimulate microbial activity and decomposition by alleviating N limitation [18], N addition can also induce soil acidification, especially when ammonium-based addition is used, which inhibits sensitive microorganisms and lowers root biomass [37,51]. Subgroup analysis also revealed greater benefits at high N addition rates compared to low ones, indicating that microbial respiration is more impacted by excess N above plant need [52]. In contrast to the typical assumption of stimulation, nitrate-based fertilization dramatically reduced soil CO₂ fluxes due to nitrate toxicity or changes in the microbial population [22]. Furthermore, only under higher humidity index were significant responses elicited, underscoring soil moisture in moderating the effects of N on CO₂ emissions [53]. Notably, the impact of nitrogen addition on grassland CO₂ emissions yielded opposite results under enclosure and grazing management. This discrepancy implies that grazing causes carbon constraint [47,54], which changes the main effect of N addition from promoting microbial and plant activity in enclosures to inhibiting microbial activity in grazed systems because of nutritional imbalance or acidity.

For CH₄ uptake, our synthesis validates the general inhibitory impact of N enrichment [26,28,31,32], which is consistent with the documented mechanism of NH₄⁺ inhibition of methane monooxygenase, a major enzyme in methanotrophic bacteria [55,56,57]. However, this suppression was rather context-dependent. Due to decreased soil aeration and limited CH₄ diffusion in wetter and more compacted conditions, which restricts the activity of the aerobic methanotrophic bacteria that are sensitive to anoxic conditions, CH₄ uptake responses were significantly inversely correlated with MAP and soil BD [58,59]. Interestingly, we found a strong positive correlation between CH₄ uptake and soil organic carbon (SOC), indicating that high SOC may alleviate the suppressive effects of N addition, perhaps by fostering a more varied or resilient methanotrophic community [60]. These findings reveal a grassland-specific mechanism that has been previously missed in wetland-dominated syntheses.

In general, these findings show that N enrichment alters GHG fluxes in grasslands through gas-specific and context-dependent mechanisms. The kind of grassland, climate conditions (e.g., HI), management, N addition rate and forms, and soil properties (e.g., pH, TN, Clay, and SOC) all affect N₂O responses. CH₄ uptake is primarily controlled by hydrological and carbon-related factors, including MAP, soil BD, clay, and SOC content. The restricted response of CO₂ fluxes from grassland soils is probably caused by the counterbalancing effects of multiple interrelated processes. While N enrichment can boost plant production and microbial activity, resulting in greater autotrophic and heterotrophic respiration, it can also suppress microbial decomposition through soil acidification, nutritional imbalances, and changes in plant–microbe interactions. The total effect of N addition on soil CO₂ emissions is neutral due to these offsetting mechanisms and significant environmental variability [22]. Our results emphasize the importance of taking environmental context into consideration when forecasting biogeochemical feedback in terrestrial ecosystems and contribute to mechanistic understanding of how grassland GHG fluxes react to human N inputs. They also highlight chances for climate-focused management. Reducing N inputs and increasing aeration may help reduce emissions in arid or compacted soils with high N₂O sensitivity [30,43,44,46,61]. Reduction in CH₄ uptake in humid, carbon-rich settings may be lessened by substituting urea or organic additions for ammonium-based fertilizers [17]. Changing the type of fertilizer and adding more organic matter could support soil carbon balance in systems where nitrate additions lower CO₂ emissions [17,22]. All in all, these findings support the development of precise N strategies tailored to specific grassland conditions to minimize GHG fluxes [22].

A key limitation of this meta-analysis is the lack of concurrent measurements of CO₂, CH₄, and N₂O fluxes within the same experimental settings, which precludes a direct estimation of the net climate forcing of N addition in terms of CO₂-equivalents. Nevertheless, given the consistent and substantial stimulation of N₂O emissions—alongside its high global warming potential (298 times that of CO₂ over a 100-year horizon) [62]—which outweighs the comparatively minor or variable responses of CO₂ and CH₄, it is likely that N addition results in a net warming effect in grassland ecosystems [25]. Future studies should focus on integrated field experiments that simultaneously monitor all three primary GHGs using standardized techniques in order to reduce uncertainty, especially in underrepresented environments like tropical and semi-arid grasslands. Furthermore, in order to improve predictive models and guide efficient N management in the face of rapidly accelerating global change, it will be crucial to advance our understanding of microbes and biogeochemistry using techniques like metagenomics, stable isotope tracing, and high-frequency flux monitoring [63,64,65].

5. Conclusions

Our global meta-analysis reveals a complex yet discernible pattern in the responses of GHG fluxes from grassland soils to nitrogen (N) addition. Even though the effects of different GHGs vary greatly, when important environmental and experimental factors are considered, the general response patterns become predictable. In particular, N enrichment causes a pronounced and consistent increase in N₂O emissions, which are mostly caused by improved microbial activity and are further amplified under conditions of high N addition rates and high humidity. However, the total impact on soil CO₂ fluxes remains negligible and statistically insignificant, suggesting a delicately regulated interaction between stimulatory drivers of respiration and antagonistic factors, including soil acidification. For CH₄, N addition results in a context-dependent reduction in uptake, with the magnitude of this effect strongly mediated by soil properties—including BD and SOC content, as well as precipitation amount. Our findings underscore that a single-gas perspective is insufficient for understanding ecosystem-scale climate feedbacks Although direct quantification of net climate forcing was precluded by data limitations, the consistent and substantial increase in N₂O emissions indicates that this gas is likely the predominant contributor to the climate impact of GHGs in grasslands, thereby suggesting a net warming effect of nitrogen enrichment in grasslands. The work provides a critical, data-driven foundation for future research to adopt an integrated, multi-gas approach. In order to reduce GHG fluxes and lessen the wider climatic effects of anthropogenic N deposition, it also offers valuable insights for developing targeted and sustainable N management strategies that take local environmental conditions into consideration.