Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2024 Dec 30;31(1):e70015. doi: 10.1111/gcb.70015

Global Greenhouse Gas Emissions From Agriculture: Pathways to Sustainable Reductions

Lidong Li 1,✉, Tala Awada 1, Yeyin Shi 1, Virginia L Jin 2, Michael Kaiser 1
PMCID: PMC11683860  PMID: 39740017

ABSTRACT

Agriculture serves as both a source and a sink of global greenhouse gases (GHGs), with agricultural intensification continuing to contribute to GHG emissions. Climate‐smart agriculture, encompassing both nature‐ and technology‐based actions, offers promising solutions to mitigate GHG emissions. We synthesized global data, between 1990 and 2021, from the Food and Agriculture Organization (FAO) of the United Nations to analyze the impacts of agricultural activities on global GHG emissions from agricultural land, using structural equation modeling. We then obtained predictive estimates of agricultural GHG emissions for the future period of 2022–2050 using deep‐learning models. The FAO data show that, from 1990 to 2021, global livestock numbers, inorganic nitrogen (N) fertilizer use, crop residue, and irrigation area increased by 27%, 47%, 49%, and 37%, respectively. The increased livestock numbers contributed to the increases in CH4 and N2O emissions, while inorganic N fertilizer, crop residue, and irrigation mainly contributed to the increases in N2O emissions. Emissions of CO2 decreased because of a 29% reduction in net forest loss. As a result of the reduced deforestation emissions, the overall agricultural GHG emissions declined from 11.50 to 10.89 GtCO2eq from 1990 to 2021 despite the increases in livestock numbers, inorganic N fertilizer, crop residue, and irrigation. Looking ahead, our model predicts that if current agricultural trends persist, GHG emissions will rise to 11.82 ± 0.07 GtCO2eq in 2050. However, maintaining agricultural GHG emissions at the 2021 level through 2050 is possible if the rate of reduction in net forest loss is doubled. Furthermore, if the rate is tripled, agricultural GHG emissions can be limited to 9.85 ± 0.07 GtCO2eq in 2050. Our findings suggest that reductions in agricultural GHG emissions, alongside sustainable agricultural intensification and climate‐smart agricultural practices, can be achieved through parallel efforts emphasizing accelerated forest conservation.

Keywords: carbon dioxide, cropland, fertilizer, forest loss, livestock, methane, nitrous oxide, pasture

Agriculture contributes to greenhouse gas (GHG) emissions through activities such as deforestation, livestock production, nitrogen fertilizer use, and irrigation. From 1990 to 2021, emissions declined slightly due to reduced deforestation, despite increased agricultural activity. Using global data and predictive models, we found that if current trends continue, GHG emissions will rise by 2050. However, doubling or tripling the reduction rate of deforestation could stabilize or even lower emissions. Our study highlights the importance of sustainable practices and accelerated forest conservation to mitigate climate change while maintaining agricultural productivity.

graphic file with name GCB-31-e70015-g005.jpg

1. Introduction

Greenhouse gases (GHGs) trap heat radiation in the Earth's atmosphere, preventing it from escaping into space. This trapped heat leads to global warming, which drives extreme weather patterns such as intense heat waves, prolonged droughts, more frequent wildfires, as well as heavier precipitation and more severe storms (IPCC 2022). The three most relevant trace gases in GHGs are carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O). In 2024, global mean atmospheric concentrations of CO2, CH4, and N2O had reached record highs of 423 ppm, 1932 ppb, and 338 ppb, respectively, representing increases of 19%, 13%, and 9% from their 1990 levels (Lan, Tans, and Thoning 2024; Lan, Thoning, and Dlugokencky 2024). Although CH4 and N2O are present in lower concentrations compared to CO2, their global warming potentials over a 100‐year time scale are 28 and 265 times greater, respectively, making them significant contributors to global warming (IPCC 2014). The global rise in these GHG emissions is primarily driven by anthropogenic activities. Agricultural practices—including forest conversion for agriculture use, livestock production, manure management, inorganic N fertilizer use, crop residue left on field, and irrigation—are significant sources of agricultural GHG emissions. Interestingly, despite increases in agricultural activity and productivity, global agricultural GHG emissions have declined from 11.4 to 10.9 GtCO2eq between 1992 and 2020 (FAO 2024; Li, Awada, et al. 2024). This apparent paradox, where we have an increase in agricultural activities but a decline in agricultural GHG emissions, warrants further investigation to understand the underlying causes and mechanisms.

Agricultural activities influence GHG emissions by altering both above‐ and belowground carbon (C) and nitrogen (N) processes. Deforestation for agricultural use, often in conjunction with drainage of land and burning of biomass, reduces aboveground biomass and belowground soil organic carbon (SOC) (Baldassini and Paruelo 2020; Hosseini et al. 2024). In contrast, afforestation or reforestation efforts—whether creating new forests or restoring degraded ones—offer substantial potential to enhance terrestrial C sinks (Xiang et al. 2022; Ménard et al. 2023). Therefore, net forest loss leads to C loss and thus GHG emissions. Furthermore, agricultural practices such as application of livestock manure and inorganic N fertilizers typically lead to GHG emissions. Climate‐smart agriculture has been proposed as a practice aimed at reducing or removing GHG emissions, enhancing C and N storage, achieving food security, and meeting relevant sustainable development goals (Lou et al. 2024).

Alterations in C and N processes in ecosystems impact GHG emissions at multiple scales. Despite substantial regional studies, there is a lack of global assessments of agricultural GHG emissions that leverage long‐term FAO historical data and integrate advanced modeling techniques, such as deep learning, to project future emissions under varying scenarios. To address these gaps, we synthesized 32 years of global data (1990–2021) from the Food and Agriculture Organization (FAO) of the United Nations (FAO 2024) to evaluate the impacts of agricultural activities on GHG emissions. Using structural equation modeling, we provide a nuanced understanding of causal relationships and relative contributions of agricultural activities to emissions. Furthermore, we applied a deep‐learning approach to predict future GHG emissions (2022–2050) from agricultural land under various hypothetical scenarios. Our analyses offer a novel global perspective on how agricultural practices shape GHG emissions and provide actionable insights for mitigating their environmental impacts.

2. Materials and Methods

2.1. Data Syntheses

The historical data for the period of 1990–2021 were obtained from FAO (2024). Data were accessed in September 2024. Data are available by year and country, with global coverage. We compiled agricultural GHG emissions and the related agricultural activities (net forest loss to agricultural use, livestock production, inorganic N fertilizer application, crop residue left on agricultural land, and irrigation) for up to 214 countries and territories. Following the FAO definitions, in this paper, agricultural GHG emissions (reported as CO2 equivalents) are the sum of CO2, CH4, and N2O emissions from agricultural land, including emissions from deforestation for agricultural use (FAO 2024). Most global reports do not typically include deforestation emissions within the scope of agriculture. However, the FAO approach incorporates deforestation emissions to provide a more comprehensive assessment of the broader impacts of agricultural activities on the planet. Forest loss is the net conversion of forestland, which is assumed to be due to agricultural use (FAO 2024). Livestock includes buffalo, camels, cattle, chickens, donkeys, goats, horses, mules, hinnies, sheep, and swine/pigs (FAO 2024). Inorganic N fertilizers are the sum of the N content from urea, ammonium sulphate, ammonium nitrate, calcium ammonium nitrate and other mixtures with calcium carbonate, sodium nitrate, urea and ammonium nitrate solutions, anhydrous ammonia, and other not elsewhere classified inorganic N fertilizers (FAO 2024). Crop residues are the sum of N content in crop residues left on agricultural land (FAO 2024). Irrigation is the agricultural land area equipped for irrigation (FAO 2024).

2.2. Statistical Analyses

We used structural equation modeling (SEM) to quantify the effects of agricultural activities on GHG emissions using the historical data from 1990 to 2021. We used SEM as a confirmatory approach and followed the standard procedures of conducting SEM (Li et al. 2019). The selection of variables—forest loss, livestock numbers, inorganic N fertilizer, crop residues, and irrigation—was based on their established roles as significant drivers of agricultural GHG emissions. The expected interactions, based on the basic equations used by FAO and the guidelines of the International Panel on Climate Change (IPCC 2006), are that increases in forest loss, livestock numbers, inorganic N fertilizer, crop residues, and irrigation will lead to increased GHG emissions. To validate the expected interactions, we conducted a literature review to ensure the alignment of these relationships with known causal pathways (Shcherbak, Millar, and Robertson 2014; Duan et al. 2022; Min et al. 2022; Li et al. 2023; Li, Hong, et al. 2024; Pan et al. 2024) and then tested their significance using SEM. The SEM was conducted using the Structural Equation Models Optimization in Python (semopy; Python 3.12). Path coefficients were tested by the maximum likelihood estimation at a p < 0.05 level and reported in the standard deviation unit. Model fit was evaluated by the comparative fit index (CFI) and the adjusted goodness of fit (AGFI). Model sensitivity was informed by bootstrap resampling with 1000 iterations to assess robustness of SEM and quantify uncertainty.

2.3. Model Development

We developed a deep‐learning approach to project future agricultural GHG emissions. Model simulations were performed using cloud clusters at the Holland Computing Center of the University of Nebraska. We used the Long Short‐Term Memory (LSTM)‐based Recurrent Neural Network (RNN) as the algorithm for modeling using Keras in TensorFlow (Python 3.12). The LSTM‐based RNN was chosen due to their ability to capture long‐term dependencies and temporal patterns in sequential data (Li, Awada, et al. 2024). Its ability to retain information over long sequences allows the consideration of the cumulative effects of past agricultural activities on GHG emissions, improving the reliability of projections. Alternative machine learning techniques, such as random forest or gradient boosting machine, while powerful, are less adept at handling temporal sequence data directly without feature engineering. The LSTM‐based RNN, in contrast, does not require additional preprocessing steps to incorporate temporal relationships, reducing the potential to introduce biases. Unlike simpler regression models, LSTM‐based RNN can model complex, nonlinear interactions and account for dynamic temporal dependencies. Our model architecture was tailored for sequential data. Specifically, the input layer corresponds to the number of features and time steps in the dataset, forming the entry point for sequential data into the model. A series of LSTM layers are implemented to capture the complex patterns in the data. The number of neurons in the LSTM layers decreases progressively (256, 128, 64, 32) to refine the patterns as the data moves deeper into the network. Dropout layers are incorporated after each LSTM layer to reduce the risk of overfitting by randomly disabling a fraction (30%) of the neurons during training. The output from the final LSTM layer is further refined through two dense layers. The second dense layer outputs the final value. Correlations among input features and their temporal dependencies were implicitly handled. Deep architectures like this are capable of modeling interactions and correlations among the features. By maintaining a memory of past inputs, LSTM can learn patterns that reflect how features evolve and interact across time steps. The historical data from 1990 to 2021 were used for model training and testing. We manually tuned the hyperparameters with the 10‐fold cross‐validation for robust evaluation of model performance. Model fit was evaluated by coefficient of determination (R 2), root mean squared error (RMSE), and loss function of mean squared error. Model uncertainty was evaluated by the standard deviation of the folds. Model sensitivity was assessed by the first order sensitivity index using the Sobol sensitivity analysis. Based on the established data‐driven relationships obtained from model training and testing, we predicted agricultural GHG emissions from 2022 to 2050 in three hypothesized scenarios using future agricultural activities as predictors. The future agricultural activities were extrapolated by fitting a trend line to each predictor with specified changing rates. In Scenario 1, net forest loss, livestock numbers, inorganic N fertilizer use, crop residue left on agricultural land, and irrigation area continue changing at the historical rates. In Scenarios 2 and 3, the rate of reduction in net forest loss is respectively doubled and tripled compared with the historical rates, while the other predictors continue increasing at the historical rates.

3. Results

3.1. Historical Trends Informed by FAO Data

We analyzed the FAO historical data (FAO 2024) and reported the FAO data and our statistical results in this section. The FAO historical data (FAO 2024) showed that global agricultural GHG emissions significantly decreased from 11.50 to 10.89 GtCO2eq between 1990 and 2021 (p = 0.01, R 2 = 0.19, Figure 1a). Our analyses showed statistical significance but relatively weak explanatory power. The low R 2 value may be attributed to model limitations. The regression model assumes linear relationships between variables, which may not fully account for the complex, nonlinear interactions driving GHG emissions. The declining trend of global net forest loss to agricultural use was more significant (p < 0.01, R 2 = 0.92, Figure 1b). The other agricultural activities, that is, livestock production, inorganic N fertilizer application, crop residue left on field, and irrigation, all significantly increased (Figure 1c–f). Indices of fit showed optimal model fit of the SEM (CFI = 1.00, AGFI = 0.99; Figure 2). Bootstrap resampling indicated narrow uncertainties of SEM (Table S1). Net forest loss had the largest effect on GHG emissions. The magnitude of impact on GHG emissions ranked as net forest loss >livestock >fertilizer >crop residue >irrigation. The FAO historical data (FAO 2024) showed that while CO2 emissions decreased, CH4 and N2O emissions increased from 1990 to 2021 (Figure 3).

FIGURE 1.

FIGURE 1

Open in a new tab

Global agricultural greenhouse gas (GHG) emissions and agricultural activities from 1990 to 2021. Data were obtained from FAO. GtCO2eq, gigaton CO2 equivalent; LSU, livestock unit; Mha, million hectare; Mt, million ton. The shade represents a 95% confidence interval. The dashed lines indicate a 95% prediction interval. (a) global GHG emissions from agricultural land, (b) global net forest loss due to agricultural use, (c) global livestock numbers, (d) global inorganic N fertilizer use in agriculture, (e) global crop residue left on agricultural land, (f) global agricultural land area equipped with irrigation.

FIGURE 2.

FIGURE 2

Open in a new tab

Effects of agricultural activities on global agricultural greenhouse gas emissions from 1990 to 2021 tested by structural equation modeling. Boxes represent variables. Arrows indicate causal relationships. The arrow width indicates the magnitude of effect size. Values beside arrows are standardized path coefficients, that is, effect sizes. Large values indicate larger magnitudes of impacts. Data were obtained from FAO, totaling 37,824 data points from 214 countries and territories.

FIGURE 3.

FIGURE 3

Open in a new tab

Global agricultural emissions of CO2, N2O, and CH4 from 1990 to 2021. Data were obtained from FAO. GtCO2eq, gigaton CO2 equivalent. The shade represents a 95% confidence interval. The dashed lines indicate a 95% prediction interval. (a) global CO2 emissions from agricultural land, (b) global CH4 emissions from agricultural land, (c) global N2O emissions from agricultural land.

3.2. Future Projections Informed by Deep‐Learning Model

The performance metrics of model training and testing showed optimal model fit (Figures S1 and S2; Table S2). Our deep‐learning model explained 94% of the variance in agricultural GHG emissions (Table S2). Standard deviations of the folds indicated narrow uncertainty (Table S2). The sensitivity index showed different importance of each input feature (Table S3). In Scenario 1, agricultural GHG emissions increased from 10.91 ± 0.05 to 11.82 ± 0.07 GtCO2eq between 2022 and 2050 (Figure 4). Reduction in net forest loss resulted in decreases in GHG emissions (Figures 5 and 6). Specifically, in Scenario 2, agricultural GHG emissions remained relatively unchanged, with a slight decrease from 10.88 ± 0.05 to 10.79 ± 0.07 GtCO2eq between 2022 and 2050 (Figure 5). In Scenario 3, agricultural GHG emissions decreased from 10.84 ± 0.05 to 9.85 ± 0.07 GtCO2eq between 2022 and 2050 (Figure 6). These low uncertainties (< 1%) only statistically reflect model performance. Real‐life uncertainty must also account for uncertainties in the training data, particularly those arising from the Tier 1 approach.

FIGURE 4.

FIGURE 4

Open in a new tab

Global agricultural greenhouse gas (GHG) emissions in Scenario 1. The changing rates of forest loss, livestock production, inorganic N fertilizer application, crop residue left on agricultural land, and irrigation are maintained the same as the historical period. GtCO2eq, gigaton CO2 equivalent; LSU, livestock unit; Mha, million hectare; Mt, million ton. The shade represents the prediction uncertainty. (a) global GHG emissions from agricultural land, (b) global net forest loss due to agricultural use, (c) global livestock numbers, (d) global inorganic N fertilizer use in agriculture, (e) global crop residue left on agricultural land, (f) global agricultural land area equipped with irrigation.

FIGURE 5.

FIGURE 5

Open in a new tab

Global agricultural greenhouse gas (GHG) emissions in Scenario 2. The rate of reduction in forest loss is doubled compared with that in the historical period; the changing rates of the other predictors are maintained the same as the historical period. GtCO2eq, gigaton CO2 equivalent; LSU, livestock unit; Mha, million hectare; Mt, million ton. The shade represents the prediction uncertainty. (a) global GHG emissions from agricultural land, (b) global net forest loss due to agricultural use, (c) global livestock numbers, (d) global inorganic N fertilizer use in agriculture, (e) global crop residue left on agricultural land, (f) global agricultural land area equipped with irrigation.

FIGURE 6.

FIGURE 6

Open in a new tab

Global agricultural greenhouse gas (GHG) emissions in Scenario 3. The rate of reduction in forest loss is tripled compared with that in the historical period; the changing rates of the other predictors are maintained the same as the historical period. GtCO2eq, gigaton CO2 equivalent; LSU, livestock unit; Mha, million hectare; Mt, million ton. The shade represents the prediction uncertainty. (a) global GHG emissions from agricultural land, (b) global net forest loss due to agricultural use, (c) global livestock numbers, (d) global inorganic N fertilizer use in agriculture, (e) global crop residue left on agricultural land, (f) global agricultural land area equipped with irrigation.

4. Discussion

In the context of global agricultural GHG emissions, forest loss, livestock production, inorganic N fertilizer application, crop residue left on agricultural land, and irrigation each contribute to CO2, CH4, and N2O emissions through distinct mechanisms, with different magnitudes of impact (Figure 2).

4.1. Forest Loss

The decline of agricultural GHG emissions from 1990 to 2021 was driven by the decrease in CO2 emissions (Figures 1a and 3a). This decrease is attributed to the reduced net forest loss (Figures 1b and 2). Forest ecosystems function as critical C sinks by capturing CO2 through photosynthesis (Roebroek et al. 2023). Globally, forest C sink is estimated to offset nearly half of the CO2 emissions from fossil fuel combustion (Pan et al. 2024). However, the forest C sink is highly vulnerable to deforestation. Deforestation drives GHG emissions through the direct removal of forest biomass and subsequent changes in soil dynamics (Figure 2). More specifically, global forest C stocks are distributed as 43% in live biomass and 45% in soils, respectively (Pan et al. 2024). When forests are converted to agricultural land, for example, croplands, pastures, or farm infrastructure, the live biomass is often burned or left to decompose. Complete combustion of forest biomass releases nearly all the stored C as CO2. Burning of forest biomass account for 17%–31% of global agricultural GHG emissions (Figure S3) (FAO 2024). The loss of forest cover and the disturbance of soil accelerate SOC decomposition, resulting in CO2 emissions. Particularly, deforestation on organic soils such as peatlands leads to substantial GHG emissions. Global CO2 emissions from organic soils that are drained for agriculture account for 6%–8% of agricultural GHG emissions (Figure S4) (FAO 2024). The drainage of organic soils allows oxygen to enter the soil, accelerating SOC decomposition and enhancing CO2 emissions (Kasimir‐Klemedtsson et al. 1997; Prananto et al. 2020).

Burning of forest biomass and drainage of organic soils have continuously contributed to CO2 emissions (Figures S3 and S4). However, the contribution of CO2 emissions from net forest loss in global agricultural GHG emissions decreased from 37% to 27% between 1990 and 2021 (Figure S5) (FAO 2024). This contradiction, where we have continuous deforestation but a decline of CO2 emissions from net forest loss, could be attributed to the increasing global efforts in afforestation and reforestation, partially offsetting the CO2 emissions from deforestation. One of the largest national afforestation programs, China's Three‐North Shelterbelt program, initiated in 1978, has afforested 30 million hectares by 2018 (Cao, Suo, and Xia 2020; Li, Xu, et al. 2021). A global effort, the Bonn Challenge, launched in 2011, aims to restore 350 million hectares of deforested and degraded lands by 2030, using the principles of forest landscape restoration (De Pinto et al. 2020). As of 2024, 210 million hectares are being restored under this initiative, with more than 60 countries committing substantial land areas for reforestation.

4.2. Livestock Production

The livestock sector is a major source of the non‐CO2 GHGs, primarily contributing to CH4 emissions from enteric fermentation in livestock digestion, as well as both CH4 and N2O from manure management (Gerber et al. 2013; FAO 2023). Emissions of CH4 from livestock enteric fermentation account for 20%–26% of global agricultural GHG emissions (Figure S6) (FAO 2024). These estimates usually carry an uncertainty of ±30%–50% because of the Tier 1 approach (IPCC 2019; Zhang et al. 2022). A recent estimation, using the 2019 Refinement to the 2006 IPCC Guidelines, showed ~20% higher values of CH4 emissions from livestock than these based on the 2006 IPCC Guidelines (Zhang et al. 2022). This difference arises from the different global warning potential values used in the guidelines (IPCC 2006, 2019). By all accounts, CH4 is the most substantial GHG within the livestock sector. Ruminant species, including cattle, sheep, and goats, produce CH4 via enteric fermentation as part of their digestive physiology. In the rumen, anaerobic fermentation breaks down structural carbohydrates into volatile fatty acids, producing hydrogen gas (H2) and CO2 (Buccioni, Cappucci, and Mele 2015; Min et al. 2022). Hydrogenotrophic methanogenic archaea utilize the H2 as an electron donor to reduce CO2 into CH4 (Buccioni, Cappucci, and Mele 2015; Min et al. 2022). As a metabolic byproduct of enteric fermentation, CH4 is expelled by the animal primarily through eructation. Diets with high fiber content result in increased CH4 emissions per unit of energy intake due to the greater reliance on fermentation for digestion (Min et al. 2022). Monogastric species, such as swine and chickens, also produce CH4, though in significantly lower quantities, as their digestive systems rely less on enteric fermentation (Gerber et al. 2013). Meanwhile, CH4 can be produced from livestock manure, especially when manure is stored or treated in anaerobic conditions such as in lagoons or anaerobic digesters. Emissions of CH4 from manure management systems account for 2%–3% of global agricultural GHG emissions (Figure S7) (FAO 2024).

Livestock manure also contributes to N2O emissions. The N in manure can be lost to the atmosphere as N2O primarily through microbial processes of aerobic nitrification and nitrifier denitrification, and anaerobic denitrification (Smith 2017; Wang et al. 2021). The manure left on pasture, applied to soil, and treated in management system together contribute to 7%–10% of global agricultural GHG emissions (Figure S8) (FAO 2024). These estimates are based on the 2006 IPCC Guidelines. Estimations using the 2019 Refinement showed 20%–50% lower values than those based on the 2006 IPCC Guidelines (FAO 2023; Zhang et al. 2024). This gap primarily stems from the differences in emission factors for manure left on pasture, which was 2% in the 2006 IPCC Guidelines but was reduced to 0.4% in the 2019 Refinement (IPCC 2006, 2019). Emission factors represent the proportion of N that is emitted as N2O. A relatively higher emission factor can arise from the localized deposition of manure and compaction of soil from livestock on pasture (Oenema et al. 2005; Wecking et al. 2020). This higher emission factor might overestimate N2O emissions under certain conditions, particularly in areas with improved grazing management. Well‐managed pastures typically have improved soil aeration and structure due to healthier plant rooting systems, which can reduce anaerobic conditions and thus the potential for denitrification (Li, Jin, et al. 2021; Van Der Weerden et al. 2023). The uncertainty in the estimates also depends on climate. Compared with estimates from the 2006 IPCC Guidelines, estimates from the 2019 Refinement were 15%–46% lower in dry climates but 7%–37% higher in wet climates (Hergoualc'h et al. 2021). To further refine those estimates, more empirical studies are necessary for the continuous development of more realistic emission factors and to reduce uncertainties in global GHG inventories.

4.3. Inorganic N Fertilizers, Crop Residues, and Irrigation

Inorganic N fertilizer application is the other significant source of N2O emissions, contributing to 4%–6% of global agricultural GHG emissions (Figure S9) (FAO 2024). Inorganic N fertilizers provide readily available N to plants and soil microbes, but a portion of the applied N is lost to the atmosphere as N2O via nitrification and denitrification. These emissions are especially high when N is applied in excess of plant requirements. Excessive use of inorganic N fertilizers can increase the N2O/(N2O + N2) production ratio because high nitrate levels in the soil inhibit the reduction of N2O to N2 during denitrification (Blackmer and Bremner 1978; Senbayram et al. 2012; Li, Hong, et al. 2024). Moreover, the acidifying effect of inorganic N fertilizers can increase the N2O/(N2O + N2) production ratio in that the reduction of N2O to N2 is more sensitive to lower pH than the reduction of nitrate to N2O (Knowles 1982; Li, Hong, et al. 2024). Consequently, soil N2O emissions are exponentially correlated with inorganic N fertilizer application rates (Shcherbak, Millar, and Robertson 2014; Li, Hong, et al. 2024). In comparison, crop residues have a smaller impact on N2O emissions due to the slower release of N into soil (Figure 2). Crop residues contribute only 1.3%–1.5% of global agricultural GHG emissions as their N input to soil is lower than that of inorganic N fertilizers (Figure S10) (FAO 2024). Particularly, when crop residues on the field and inorganic N fertilizers are combined with irrigation, irrigation can exacerbate both CO2 and N2O emissions by promoting optimal soil moisture levels for microbial mineralization and nitrification (Li et al. 2023; Li, Hong, et al. 2024; Wang et al. 2024). Excessive irrigation can create anaerobic microenvironments in the soil conducive to CH4 and N2O production (Li et al. 2023; Li, Hong, et al. 2024; Wang et al. 2024). Additionally, the energy required for the installation and operation of irrigation systems contributes to GHG emissions.

4.4. Future Projections and Climate‐Smart Agriculture

Feeding a growing global population requires increases in food production in the face of climate change and declining natural resources in the coming decades, likely driving further intensification of the agricultural activities. If these agricultural activities continue changing at the historical rates, our projections indicate that agricultural GHG emissions will rise by 8.6% ± 0.6% by 2050 compared with the 2021 levels (Figure 4). This presents a significant environmental challenge, but it is important that the concerns about GHG emissions do not compromise global food security. Currently, the most effective and practical strategy for mitigating agricultural GHG emissions is to limit CO2 emissions from forest loss (Figure 2). To maintain agricultural GHG emissions from 2022 to 2050 at the 2021 level, the rate of reduction in forest loss should be doubled compared with the historical trends (Figure 5). Furthermore, if the rate of reduction in forest loss is tripled, agricultural GHG emissions can decrease by 9.5% ± 0.6% in 2050 compared with the 2021 levels (Figure 6). While the benefits of this scenario are clear, its real‐world feasibility faces potential challenges. A tripling in the rate of forest loss reduction demands robust policy frameworks at both global and national levels. Many countries experience economic pressures to expand agricultural land to meet growing food demands or boost export revenue. Some local communities, particularly in tropical regions, depend on forested areas for their livelihoods. Resolving such land use competition necessitates integrating the needs of local communities with conservation objectives. Financial incentives can encourage reduced deforestation. Programs such as payments for ecosystem services and carbon markets can make conservation economically viable. The successful realization of this scenario depends on embedding conservation efforts within broader development objectives, fostering international cooperation, and ensuring that local stakeholders are empowered and incentivized to participate. Such a holistic approach can help bridge the gap between ambition and action.

From a global perspective, improving productivity on existing agricultural land to meet the increasing demands for food without expanding into forested areas (e.g., more efficient use of nutrients and water, implementing conservation practices, increasing resistance to abiotic and biotic stresses, and improving crop varieties and livestock breeds) is key to climate‐smart agriculture. This approach reduces the need for converting forest to cropland or pasture, while maintaining or improving productivity. In addition to reducing CO2 emissions, it is equally important to mitigate CH4 and N2O emissions. Our projected CH4 and N2O emissions keep increasing in the future scenarios (Figures S11–S16). Multiple strategies can be used to effectively mitigate CH4 emissions from livestock, including the use of methanogenesis inhibitors, genetic selections, gut microbial community selection, and dietary interventions (De Haas et al. 2011; Ungerfeld 2018; Berhanu et al. 2019). Similarly, improving N use efficiency is essential for reducing N2O emissions from inorganic N fertilizers and manure. This can be accomplished by optimization of the timing, amount, placement, and/or delivery of fertilizers, along with the use of controlled‐release fertilizers or nitrification inhibitors (Lyu et al. 2021; Wu et al. 2021; Wang et al. 2023; Li, Hong, et al. 2024). To effectively accelerate agricultural GHG mitigation, stakeholders can adopt the innovative technologies that are locally feasible. Examples of successful climate‐smart agriculture applications include the system of rice intensification (SRI) implemented in various parts of India. The SRI method has led to increases in rice yields while significantly reducing fertilizer and water usage and CH4 emissions (Jain et al. 2014; Varma 2019). Implementing improved feeding strategies for cattle in Brazil, such as dietary supplements and rotational grazing, has reduced CH4 emissions per kilogram of meat produced (D'aurea et al. 2021). Optimizing irrigation amounts has reduced N2O emissions while maintaining corn yields in the United States Corn Belt (Li et al. 2023). Climate‐smart agriculture presents a diverse set of strategies, including nature‐ and technology‐based, to mitigate GHG emissions while maintaining or enhancing agricultural productivity. Addressing the full scope of GHG challenges requires an integrated, multisectoral approach. Integrating climate‐smart agriculture with technological, economic, and policy innovations across sectors offers a more sustainable pathway for agricultural development, contributing to long‐term food security while supporting global efforts to mitigate climate change.

4.5. Limitations and Future Work

The strength of our model is limited to the training data. The country‐level dataset may contain potential biases or gaps. Different countries might use varied reporting standards and methods, leading to discrepancies in the reported data. Certain regions, particularly low‐income or conflict‐affected areas, may be underrepresented due to challenges in data collection. Also, aggregating data at the country level can obscure localized events or trends, such as region‐specific droughts, floods, or deforestation rates. Practices like fertilizer use or livestock management, which vary by region, may not be fully captured due to the aggregation of data at the country level. To enhance accuracy, higher spatial resolution is necessary to train our model. Once more data become available, they can be incorporated into our model to update its output.

The other limitation arises from the reliance on extrapolation in future scenarios. Our model assumes that historical trends in agricultural activities persist into the future. However, this assumption may not hold under the influence of abrupt policy changes, economic crises, or rapid technological advancements. Simulating diverse future scenarios in future research by incorporating varying policy frameworks and technological advancements could provide a more comprehensive reflection of potential pathways. Integrating real‐time data streams into future models could enhance their capacity to adapt to rapid shifts in agricultural practices and environmental conditions, offering more dynamic and accurate insights. Moreover, field experiments are critical for validating model predictions and generating high‐quality data for model training. Field experiments should incorporate input from local stakeholders and policymakers to ensure that model‐driven recommendations are practical and socially acceptable. Overall, combining advancements in modeling with rigorous field experimentation will enable more accurate, transparent, and actionable insights for reducing agricultural GHG emissions.

Author Contributions

Lidong Li: conceptualization, data curation, formal analysis, investigation, methodology, software, validation, visualization, writing – original draft, writing – review and editing. Tala Awada: conceptualization, funding acquisition, project administration, resources, supervision, validation, writing – review and editing. Yeyin Shi: resources. Virginia L. Jin: funding acquisition, project administration. Michael Kaiser: resources, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1.

GCB-31-e70015-s001.docx (4MB, docx)

Acknowledgments

We thank Christopher Misar for providing insights on agricultural management. We thank Francesco Tubiello for providing comments and edits. We thank FAO for the publicly available database (FAO 2024). This work was completed utilizing the Holland Computing Center of the University of Nebraska, which receives support from the UNL Office of Research and Economic Development, the Nebraska Research Initiative and the Agricultural Research Division. This study was supported by the Long‐Term Agroecosystem Research (LTAR) network of the United States Department of Agriculture (USDA), and Partnership on Data Innovation (PDI), USDA.

Funding: This work was supported by the Partnership on Data Innovation of the United States Department of Agriculture and Long‐Term Agroecosystem Research (LTAR) network of the United States Department of Agriculture.

Data Availability Statement

The data and code that support the findings of this study are openly available in Dryad at https://doi.org/10.5061/dryad.73n5tb36k. GHG emissions and agricultural activities data were obtained from FAOSTAT at https://www.fao.org/faostat/en/#data.

References

  1. Baldassini, P. , and Paruelo J. M.. 2020. “Deforestation and Current Management Practices Reduce Soil Organic Carbon in the Semi‐Arid Chaco, Argentina.” Agricultural Systems 178: 102749. [Google Scholar]
  2. Berhanu, Y. , Olav L., Nurfeta A., Angassa A., and Aune J. B.. 2019. “Methane Emissions From Ruminant Livestock in Ethiopia: Promising Forage Species to Reduce CH4 Emissions.” Agriculture 9: 130. [Google Scholar]
  3. Blackmer, A. , and Bremner J.. 1978. “Inhibitory Effect of Nitrate on Reduction of N2O to N2 by Soil Microorganisms.” Soil Biology and Biochemistry 10: 187–191. [Google Scholar]
  4. Buccioni, A. , Cappucci A., and Mele M.. 2015. “Methane Emission From Enteric Fermentation: Methanogenesis and Fermentation.” Climate Change Impact on Livestock: Adaptation and Mitigation 186. [Google Scholar]
  5. Cao, S. , Suo X., and Xia C.. 2020. “Payoff From Afforestation Under the Three‐North Shelter Forest Program.” Journal of Cleaner Production 256: 120461. [Google Scholar]
  6. D'aurea, A. P. , da Silva Cardoso A., Guimarães Y. S. R., Fernandes L. B., Ferreira L. E., and Reis R. A.. 2021. “Mitigating Greenhouse Gas Emissions From Beef Cattle Production in Brazil Through Animal Management.” Sustainability 13: 7207. [Google Scholar]
  7. De Haas, Y. , Windig J., Calus M., et al. 2011. “Genetic Parameters for Predicted Methane Production and Potential for Reducing Enteric Emissions Through Genomic Selection.” Journal of Dairy Science 94: 6122–6134. [DOI] [PubMed] [Google Scholar]
  8. De Pinto, A. , Cenacchi N., Robertson R., et al. 2020. “The Role of Crop Production in the Forest Landscape Restoration Approach—Assessing the Potential Benefits of Meeting the Bonn Challenge.” Frontiers in Sustainable Food Systems 4: 61. [Google Scholar]
  9. Duan, N. , Li L., Liang X., et al. 2022. “Variation in Bacterial Community Structure Under Long‐Term Fertilization, Tillage, and Cover Cropping in Continuous Cotton Production.” Frontiers in Microbiology 13: 847005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. FAO . 2023. Pathways Towards Lower Emissions—A Global Assessment of the Greenhouse Gas Emissions and Mitigation Options From Livestock Agrifood Systems. Rome: Food and Agriculture Organization of the United Nations. [Google Scholar]
  11. FAO . 2024. “Name of Database: FAOSTAT.” https://www.fao.org/faostat/en/#data/. Licence: CC‐BY‐4.0.
  12. Gerber, P. J. , Steinfeld H., Henderson B., et al. 2013. “Tackling Climate Change Through Livestock: A Global Assessment of Emissions and Mitigation Opportunities”.
  13. Hergoualc'h, K. , Mueller N., Bernoux M., Kasimir Ä., van der Weerden T. J., and Ogle S. M.. 2021. “Improved Accuracy and Reduced Uncertainty in Greenhouse Gas Inventories by Refining the IPCC Emission Factor for Direct N2O Emissions From Nitrogen Inputs to Managed Soils.” Global Change Biology 27: 6536–6550. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hosseini, M. , Bahrami H., Khormali F., Khavazi K., Mokhtassi‐Bidgoli A., and Rezaei M.. 2024. “Effect of Deforestation on Soil Organic Carbon Fate and Pool, a Case Study in Mazandaran, Iran.” Catena 237: 107804. [Google Scholar]
  15. IPCC . 2006. 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Prepared by the National Greenhouse Gas Inventories Programme, edited by Eggleston H. S., Buendia L., Miwa K., Ngara T., and Tanabe K., 4. Japan: IGES. [Google Scholar]
  16. IPCC . 2014. Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, edited by Pachauri R. K. and Meyer L. A., 87. Geneva, Switzerland: IPCC Core Writing Team. [Google Scholar]
  17. IPCC . 2019. 2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse gas Inventories, edited by Calvo Buendia E., Tanabe K., Kranjc A., et al., 4. Switzerland: IPCC. [Google Scholar]
  18. IPCC . 2022. Climate Change 2022: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, edited by Pörtner H.‐O., Roberts D. C., Tignor M., et al., 3056. Cambridge, UK and New York: Cambridge University Press. 10.1017/9781009325844. [DOI] [Google Scholar]
  19. Jain, N. , Dubey R., Dubey D., et al. 2014. “Mitigation of Greenhouse Gas Emission With System of Rice Intensification in the Indo‐Gangetic Plains.” Paddy and Water Environment 12: 355–363. [Google Scholar]
  20. Kasimir‐Klemedtsson, Å. , Klemedtsson L., Berglund K., Martikainen P., Silvola J., and Oenema O.. 1997. “Greenhouse Gas Emissions From Farmed Organic Soils: A Review.” Soil Use and Management 13: 245–250. [Google Scholar]
  21. Knowles, R. 1982. “Denitrification.” Microbiological Reviews 46: 43–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Lan, X. , Tans P., and Thoning K. W.. 2024. “Trends in Globally‐Averaged CO2 Determined From NOAA Global Monitoring Laboratory Measurements.” Version 2024‐08. 10.15138/9N0H-ZH07. [DOI]
  23. Lan, X. , Thoning K. W., and Dlugokencky E. J.. 2024. “Trends in Globally‐Averaged CH4, N2O, and SF6 Determined From NOAA Global Monitoring Laboratory Measurements.” Version 2024‐08. 10.15138/P8XG-AA10. [DOI]
  24. Li, H. , Xu F., Li Z., et al. 2021. “Forest Changes by Precipitation Zones in Northern China After the Three‐North Shelterbelt Forest Program in China.” Remote Sensing 13: 543. [Google Scholar]
  25. Li, L. , Awada T., Zhang Y., and Paustian K.. 2024. “Global Land Use Change and Its Impact on Greenhouse Gas Emissions.” Global Change Biology 30: e17604. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Li, L. , Hong M., Zhang Y., and Paustian K.. 2024. “Soil N2O Emissions From Specialty Crop Systems: A Global Estimation and Meta‐Analysis.” Global Change Biology 30: e17233. [DOI] [PubMed] [Google Scholar]
  27. Li, L. , Jin V. L., Kettler T., et al. 2021. “Decreased Land Use Intensity Improves Surface Soil Quality on Marginal Lands.” Agrosystems, Geosciences & Environment 4: e20226. [Google Scholar]
  28. Li, L. , Ma L., Qi Z., et al. 2023. “Measured and Simulated Effects of Residue Removal and Amelioration Practices in No‐Till Irrigated Corn ( Zea mays L.).” European Journal of Agronomy 146: 126807. [Google Scholar]
  29. Li, L. , Wilson C. B., He H., Zhang X., Zhou F., and Schaeffer S. M.. 2019. “Physical, Biochemical, and Microbial Controls on Amino Sugar Accumulation in Soils Under Long‐Term Cover Cropping and No‐Tillage Farming.” Soil Biology and Biochemistry 135: 369–378. [Google Scholar]
  30. Lou, Y. , Feng L., Xing W., et al. 2024. “Climate‐Smart Agriculture: Insights and Challenges.” Elsevier 1: 100003. [Google Scholar]
  31. Lyu, X. , Wang T., Song X., et al. 2021. “Reducing N2O Emissions With Enhanced Efficiency Nitrogen Fertilizers (EENFs) in a High‐Yielding Spring Maize System.” Environmental Pollution 273: 116422. [DOI] [PubMed] [Google Scholar]
  32. Ménard, I. , Thiffault E., Kurz W. A., and Boucher J.‐F.. 2023. “Carbon Sequestration and Emission Mitigation Potential of Afforestation and Reforestation of Unproductive Territories.” New Forests 54: 1013–1035. [Google Scholar]
  33. Min, B.‐R. , Lee S., Jung H., Miller D. N., and Chen R.. 2022. “Enteric Methane Emissions and Animal Performance in Dairy and Beef Cattle Production: Strategies, Opportunities, and Impact of Reducing Emissions.” Animals 12: 948. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Oenema, O. , Wrage N., Velthof G. L., van Groenigen J. W., Dolfing J., and Kuikman P. J.. 2005. “Trends in Global Nitrous Oxide Emissions From Animal Production Systems.” Nutrient Cycling in Agroecosystems 72: 51–65. [Google Scholar]
  35. Pan, Y. , Birdsey R. A., Phillips O. L., et al. 2024. “The Enduring World Forest Carbon Sink.” Nature 631: 563–569. [DOI] [PubMed] [Google Scholar]
  36. Prananto, J. A. , Minasny B., Comeau L. P., Rudiyanto R., and Grace P.. 2020. “Drainage Increases CO2 and N2O Emissions From Tropical Peat Soils.” Global Change Biology 26: 4583–4600. [DOI] [PubMed] [Google Scholar]
  37. Roebroek, C. T. , Duveiller G., Seneviratne S. I., Davin E. L., and Cescatti A.. 2023. “Releasing Global Forests From Human Management: How Much More Carbon Could Be Stored?” Science 380: 749–753. [DOI] [PubMed] [Google Scholar]
  38. Senbayram, M. , Chen R., Budai A., Bakken L., and Dittert K.. 2012. “N2O Emission and the N2O/(N2O+ N2) Product Ratio of Denitrification as Controlled by Available Carbon Substrates and Nitrate Concentrations.” Agriculture, Ecosystems & Environment 147: 4–12. [Google Scholar]
  39. Shcherbak, I. , Millar N., and Robertson G. P.. 2014. “Global Metaanalysis of the Nonlinear Response of Soil Nitrous Oxide (N2O) Emissions to Fertilizer Nitrogen.” Proceedings of the National Academy of Sciences 111: 9199–9204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Smith, K. 2017. “Changing Views of Nitrous Oxide Emissions From Agricultural Soil: Key Controlling Processes and Assessment at Different Spatial Scales.” European Journal of Soil Science 68: 137–155. [Google Scholar]
  41. Ungerfeld, E. M. 2018. “Inhibition of Rumen Methanogenesis and Ruminant Productivity: A Meta‐Analysis.” Frontiers in Veterinary Science 5: 113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Van Der Weerden, T. , Noble A., Beltrán I., et al. 2023. “Influence of Key Factors on Ammonia and Nitrous Oxide Emission Factors for Excreta Deposited by Livestock and Land‐Applied Manure.” Science of the Total Environment 889: 164066. [DOI] [PubMed] [Google Scholar]
  43. Varma, P. 2019. “Adoption and the Impact of System of Rice Intensification on Rice Yields and Household Income: An Analysis for India.” Applied Economics 51: 4956–4972. [Google Scholar]
  44. Wang, C. , Amon B., Schulz K., and Mehdi B.. 2021. “Factors That Influence Nitrous Oxide Emissions From Agricultural Soils as Well as Their Representation in Simulation Models: A Review.” Agronomy 11: 770. [Google Scholar]
  45. Wang, C. , Qi Z., Zhao J., et al. 2023. “Sustainable Water and Nitrogen Optimization to Adapt to Different Temperature Variations and Rainfall Patterns for a Trade‐Off Between Winter Wheat Yield and N2O Emissions.” Science of the Total Environment 854: 158822. [DOI] [PubMed] [Google Scholar]
  46. Wang, W. , Ye Z., Li J., et al. 2024. “Intermediate Irrigation With Low Fertilization Promotes Soil Nutrient Cycling and Reduces CO2 and CH4 Emissions via Regulating Fungal Communities in Arid Agroecosystems.” Journal of Environmental Management 351: 119688. [DOI] [PubMed] [Google Scholar]
  47. Wecking, A. R. , Wall A. M., Liáng L. L., et al. 2020. “Reconciling Annual Nitrous Oxide Emissions of an Intensively Grazed Dairy Pasture Determined by Eddy Covariance and Emission Factors.” Agriculture, Ecosystems & Environment 287: 106646. [Google Scholar]
  48. Wu, P. , Chen G., Liu F., Cai T., Zhang P., and Jia Z.. 2021. “How Does Deep‐Band Fertilizer Placement Reduce N2O Emissions and Increase Maize Yields?” Agriculture Ecosystems & Environment 322: 107672. [Google Scholar]
  49. Xiang, Y. , Li Y., Luo X., et al. 2022. “Mixed Plantations Enhance More Soil Organic Carbon Stocks Than Monocultures Across China: Implication for Optimizing Afforestation/Reforestation Strategies.” Science of the Total Environment 821: 153449. [DOI] [PubMed] [Google Scholar]
  50. Zhang, L. , Pan S., Ouyang Z., et al. 2024. “Global Nitrous Oxide Emissions From Livestock Manure During 1890–2020: An IPCC Tier 2 Inventory.” Global Change Biology 30: e17303. [DOI] [PubMed] [Google Scholar]
  51. Zhang, L. , Tian H., Shi H., et al. 2022. “A 130‐Year Global Inventory of Methane Emissions From Livestock: Trends, Patterns, and Drivers.” Global Change Biology 28: 5142–5158. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data S1.

GCB-31-e70015-s001.docx (4MB, docx)

Data Availability Statement

The data and code that support the findings of this study are openly available in Dryad at https://doi.org/10.5061/dryad.73n5tb36k. GHG emissions and agricultural activities data were obtained from FAOSTAT at https://www.fao.org/faostat/en/#data.

Articles from Global Change Biology are provided here courtesy of Wiley

RESOURCES