Carbon footprint of synthetic nitrogen under staple crops: A first cradle‐to‐grave analysis

More than half of the world's population is nourished by crops fertilized with synthetic nitrogen (N) fertilizers. However, N fertilization is a major source of anthropogenic emissions, augmenting the carbon footprint (CF). To date, no global quantification of the CF induced by N fertilization of the main grain crops has been performed, and quantifications at the national scale have neglected the CO2 assimilated by plants. A first cradle‐to‐grave life cycle assessment was performed to quantify the CF of the N fertilizers' production, transportation, and application to the field and the uses of the produced biomass in livestock feed and human food, as well as biofuel production. We quantified the direct and indirect inventories emitted or sequestered by N fertilization of main grain crops: wheat, maize, and rice. Grain food produced with N fertilization had a net CF of 7.4 Gt CO2eq. in 2019 after excluding the assimilated C in plant biomass, which accounted for a quarter of the total CF. The cradle (fertilizer production and transportation), gate (fertilizer application, and soil and plant systems), and grave (feed, food, biofuel, and losses) stages contributed to the CF by 2%, 11%, and 87%, respectively. Although Asia was the top grain producer, North America contributed 38% of the CF due to the greatest CF of the grave stage (2.5 Gt CO2eq.). The CF of grain crops will increase to 21.2 Gt CO2eq. in 2100, driven by the rise in N fertilization to meet the growing food demand without actions to stop the decline in N use efficiency. To meet the targets of climate change, we introduced an ambitious mitigation strategy, including the improvement of N agronomic efficiency (6% average target for the three crops) and manufacturing technology, reducing food losses, and global conversion to healthy diets, whereby the CF can be reduced to 5.6 Gt CO2eq. in 2100.


| INTRODUC TI ON
The three main staple crops, maize, rice, and wheat, comprise twothirds of global food consumption (Day, 2013), representing 47% of agricultural land in 2019 (calculated from FAOSTAT).These three crops accounted for more than half (52.5 Tg N) (Tg = 10 12 g) of the global nitrogen (N) usage in agriculture (IFA, 2017), and this amount will increase substantially by 2050 due to the doubled food demand (Tilman et al., 2011).Globally, three-quarters of the applied N is surplus (Zhang et al., 2015), indicating that 40 Tg of N used to fertilize staple crops is lost to the environment, which results in acidification, contamination, eutrophication, stratospheric ozone depletion, and global warming (Galloway et al., 2003(Galloway et al., , 2008;;Reay et al., 2012;Steffen et al., 2015;Zamanian et al., 2018).
On average, the carbon footprint (CF) of energy production and supply for N fertilizer production are 116 and 29 kg CO 2 eq.GJ −1 respectively (Hoxha & Christensen, 2019).Fertilizer production, transportation, and spreading in the field resulted in CF values of 4.2, 0.10, and 0.26 kg CO 2 eq.kg −1 N, respectively (Amenumey & Capel, 2013;Hoxha & Christensen, 2019;Zhang et al., 2013).We updated and recalculated these CF values per unit of produced grain at the country level globally (Zhang et al., 2013).Few studies have stated the CF of N fertilization as fixed values (5.2 and 5.6 kg CO 2 eq.kg −1 N) based on field observations (Chojnacka et al., 2019;Zhang et al., 2013).These CF values are the aggregations of direct N 2 O emissions and indirect emissions via ammonia volatilization (NH 3 ) and nitrate leaching (NO − 3 ), while the carbon dioxide (CO 2 ) emitted directly after N application and the effects of N fertilization on CO 2 efflux from soil, C sequestration, and methane (CH 4 ) uptake or release were neglected.The IPCC stated that 20% of the applied urea is emitted directly as CO 2 after N application (2006a), in addition to a 0.124 kg C kg −1 N increase in soil respiration due to N fertilization (Zhou et al., 2014).Furthermore, N fertilization stimulates soil organic and microbial C sequestration (Jian et al., 2016), CH 4 uptake in uplands (Chen et al., 2021), CH 4 emissions in rice paddies (Banger et al., 2012), NH 3 emission (Zhan et al., 2021), nitrogen monoxide (NO) emission (Kayatz et al., 2020), and nitrate leaching (Wang et al., 2019).Minimum uncertainties in the global CF quantification of N fertilizers can be achieved by including all these inventories in the life cycle assessment (LCA).
Recently, the Paris Agreement established a requirement for biannual reports of greenhouse gas (GHG) inventories from over 190 national signatories with sufficient accuracy and details to follow progress toward nationally determined contributions.Unfortunately, no inventories consider all these paths, and even the available inventories of some parameters did not investigate N fertilizers as the main study factor or exclude the soil productivity index without N applications (control).Most of these studies have not included the C assimilated in crop biomass due to N fertilization and the gas emissions of the produced biomass until the end of life.Therefore, a framework that reconciles all the possible inventories, including data-driven and process-based models and bottom-up methods (FAO, 2019;IPCC, 2006a;Kayatz et al., 2020;Zhan et al., 2021), was developed based on field observations, including global metaanalyses (Cui et al., 2021;Shcherbak et al., 2014;Zhan et al., 2021).
Nitrogen fertilization doubled the number of humans fed by a hectare of arable land between 1908 and 2008, mainly after developing the Haber-Bosch process (Erisman et al., 2008).Cereal production is the main source of anthropogenic GHG emissions from food production systems (Crippa et al., 2021), whereas only 20% of these emissions come from N fertilization (Zhang et al., 2018).As a global average of increased grain production induced by N fertilization, nitrogen agronomic efficiency (NAE, a parameter for the quantity of grain produced by a kilogram of N) was assessed at 28 kg grain kg −1 N (van Grinsven et al., 2022).Therefore, we included the sequestered C by plant biomass in our LCA (Yara, 2010) to stand against the thought that crop CO 2 assimilation does not impact the global CF long enough due to the further processing of plant biomass for food or feed (Chojnacka et al., 2019).Grain represents 46% of the total cereal biomass.In comparison, the remaining parts (40% straw and 14% roots) are being returned to the soil, burned, or removed for animal feeding and other uses as byproducts (Dubey & Lal, 2009;Liu et al., 2018).The crop residues returned to the soil, and the part burnt of the three crops resulted in 162 and 33 Tg CO 2 eq., respectively, as calculated from FAOSTAT (https:// www.fao.org/ faost at/ en/# data/ GA) based on IPCC, tier 1 (assuming that only 10% is burnt and the remaining is returned).
Moreover, the biomass produced by N fertilization contributes substantially to the CF through livestock feed, human food, and the production and consumption of biofuels.Here, the feed and food supply chain encompasses production, packaging, transportation, processing, retail, consumption, and end of life, and resulted in CF by 4 × 10 3 and 18 × 10 3 Tg CO 2 eq. of the feed system in 2019 (FAOSTAT, 2023) and food system in 2015 (Crippa strategy, including the improvement of N agronomic efficiency (6% average target for the three crops) and manufacturing technology, reducing food losses, and global conversion to healthy diets, whereby the CF can be reduced to 5.6 Gt CO 2 eq. in 2100.

K E Y W O R D S
food supply chain, GHG emissions, global warming, nitrogen agronomic efficiency, soil and plant systems, synthetic nitrogen | 3 of 16 ABDO et al. et al., 2021) respectively.Globally, we estimated 374 and 171 Tg of grains used as biofuel or lost during consumption of the three crops (FAOSTAT, 2023), which can affect the CF substantially.
Considering the abovementioned findings, the alleged environmental impacts of N fertilizer usage are inequitably distributed, which may not be as strongly negative as has been reported (Gao & Cabrera Serrenho, 2023).The methodology divergence and overlooking some direct and indirect sources of GHG emissions and sequestrations trigger large differences.Thus, the cradle-to-grave LCA approach was established.Our study aimed to assess the CF induced by N fertilizers applied to staple crops.Here, we implemented a comprehensive and consistent LCA across all the system boundaries (N fertilizer in grain production), from energy mining to grain production and end of life.We applied the LCA from the cradle stage to the grave stage as recommended by the International Organization for Standardization's (ISO, 2000) international standard to cover the net CF of all boundaries, including energy production and supply for fertilizer plants, transportation, spreading, and soil and plant sequestrations and emissions.Additionally, the LCA tracked the consumption of the crop biomass produced by N fertilizers until the end of life.In this respect, the grave stage covered the boundaries of all possible systems of biomass use, including feed, food, biofuel, and losses.
Our assessment deepens our understanding of the CF of N fertilizer application to grain crops to test the role of N management (including changed manufacturing technology and improved N use efficiency) in mitigating global warming according to the standards of the Paris Agreement.Therefore, we suggested mitigation options for the CF across our LCA, such as improving the NAE, decreasing fertilizer production emissions, transportation, and application by developing the current technologies, and reducing losses and conversion to a healthy diet.The role of these mitigation options in controlling the CF was evaluated until 2100.

| Terminology
This study estimated the net CF of staple crop production induced by N fertilization, considering all the possible direct and indirect emissions and sequestrations of GHGs.A detailed LCA encompassing the cradle, gate, and grave stages to cover all the system (N production and usage) boundaries was assigned here.These stages comprised six main groups (Figure 1 The LCA functional unit is kg CO 2 eq.Mg −1 grain encompassing CO 2 , CH 4 , and nitrous oxide (N 2 O) as direct inventories and NO throughout the system.We added ammonia emissions (NH 3 ) and nitrate leaching (NO − 3 ) as indirect CF sources along the soil boundaries.We used a bottom-up approach (inventory, process-based modeling, and statistical extrapolations such as meta-analysis studies) to calculate the inventories of each system boundary.A list of abbreviations is presented in the Supporting Information.

| Data analysis
The calculation methodology is presented in detail in Supporting Information Section S1.The functional unit of our LCA impact is kg CO 2 eq.Mg −1 , the amount of the grain CF produced with N fertilization only after removing the effect of soil productivity when no N fertilization is added.Therefore, we used the NAE (Mg grain kg −1 N; Data S1) as a parameter of the produced grain by N fertilizer only to remove the effects of other agricultural inputs and calculate the F CF .
The following general equation (Equation 1) was used to calculate the total F CF of the main system boundaries: where FP CF , FT CF , FA CF , FS CF , FC CF , FO CF , FE CF , FB CF , and FL CF are the CF factors (kg CO 2 eq.Mg −1 grain) of fertilizer production, transportation and application (spreading), soil system, C sequestration by plant system, emissions of plant residues, food, feed, biofuel, and losses, respectively.

Fertilizer production (FP CF )
We collected the emission factors (kg CO 2 eq.GJ −1 ) of energy production and supply (Table S1) and the standards of the amount of energy used (GJ) for the production of ammonia and nitric acid (Mg) and the formation of the investigated fertilizer types using previous studies and calculator tools (Tables S1 and S2).We included emitted and sequestered gases during fertilizer production.We calculated the total CF of each N fertilizer type (FP CF ; kg CO 2 eq.kg −1 product) using the following equation (Equation 2) (Gunnar, 1998;Zhang et al., 2013) where FP Amm and FP Nit are the CF factors of ammonia and nitric acid production respectively.W Amm and W Nit are the amount of ammonia and nitric acid used to produce a fertilizer type (Table S2). (2) and ME CH 4 are the emitted gasses, while MS CO 2 , MS N 2 O , and MS CH 4 are the gasses sequestered (kg kg −1 ) during the manufacture of the final N fertilizer type.FP EI is the energy input during the formation of each fertilizer type.K refers to all inputs except ammonia and nitric acid, which produce the final N fertilizer.We calculated the weighted arithmetic mean of the CF for all the N fertilizer types utilizing the ratio of each fertilizer type to the total consumption.Then, we divided these values by the NAE to calculate the FP CF (kg CO 2 eq.Mg −1 grain; Data S2 and S3).

Fertilizer transportation (FT CF )
We calculated that 30% of the N applied to the staple crops is imported, crossing a distance of 27.4 × 10 6 km internationally and 0.14 × 10 6 km nationally.Therefore, we calculated the FT CF (kg CO 2 eq.kg −1 N) of fertilizer transportation using the following equation (Equation 3) based on the IPCC TIER 2 method (IPCC, 2006b; Data S5).
where FTS CF is the value of the reference country (kg CO 2 eq.kg −1 N km −1 ; Table S3).The EF C and EF S are the emission factors (kg CO 2 eq.GJ −1 ) of the transportation sector in each country and the reference country respectively (Data S5).IN CF , the emission factor of the international shipping, which was set as 7.81 × 10 −6 kg CO 2 eq.kg −1 N km −1 in 2015 according to IEA (2021) (https:// www.iea.org/ repor ts/ inter natio nal-shipping).D IN is the international distance between N fertilizer importers and exporters (Data S4).

Fertilizer application (FA CF )
Spreading fertilizers on soil affects the CF through GHG emissions related to fuel consumption by machinery.Due to data shortages and inconsistent observations of the CF of fertilizer application (Lai, 2004), we performed a case study (Supporting Information Section S2) to measure the FA CF (kg CO 2 eq.kg −1 N) of fertilizer application using Equation (4) (Amenumey & Capel, 2013; IPCC, 2006a; Data S6): (3) Life cycle assessment of the carbon footprint (CF) induced by N fertilizers that were applied to the grain crops (kg CO 2 eq.Mg −1 ) in 2019.Values are global averages of the CF values with the boundaries of the life cycle assessment (functional unit: kg CO 2 eq.Mg −1 grains).Positive values indicate greenhouse gas emissions and increase in the CF, while negative values refer to declines in the CF.Black, red, and blue colors refer to maize, rice and wheat, respectively.AN, ammonium nitrate (33.5% N); AS, ammonium sulfate (21%); CAN, calcium ammonium nitrate (27%); UAN, urea ammonium nitrate (30%).The FP CF is the CF of fertilizer production including mining, supply and combustion of the energy, in addition to gas emission or sink during manufacture process.The FT CF is the CF of fertilizer transportation including emissions through the national and international journeys.The FA CF is the CF of fertilizer application to field (mechanical spreading).The FS CF is the CF of the soil system accounting for the emissions and sinks of gases due to N fertilization.E CO 2 is the direct emissions of carbon dioxide (urea fertilizer hydrolysis and soil respiration), while S CO 2 is the carbon sequestration in soil induced by N fertilization.E N 2 O and E CH 4 are the CF equivalents of emitted nitrous oxide and methane after conversion using IPCC values (298 and 25 kg CO 2 eq.kg −1 of these gases) respectively.E NH 3 , E NO , and are the indirect CF induced by N fertilizers through the emissions of ammonia, nitrogen monoxide, and nitrate leaching using IPCC factors (0.01, 0.01 and 0.0075) for conversion into N 2 O equivalents respectively.P CO 2 is the carbon assimilated by plant biomass through photosynthesis, of which part was burnt (EPB), other was returned to the soil (EPR) and the remaining was included in the grave stage to accumulate the net CF after different uses of the crop biomass.The total CF represents the CF from the cradle stage to the end of life (grave stage).Data of this figure are presented in Supplementary Figure Data (Data S1).
where E is the energy required (MJ) per kg of fertilizer, N is the percentage of nitrogen content in each fertilizer type, F is the fuel consumption (L MJ −1 ), and EF is the emission factor of the consumed fuel (2.327 kg CO 2 L −1 , IPCC).Here, we assumed that fertilizer was applied only by machines due to the difficulties associated with calculating emissions from manual application by people.

CF of the fertilized soil (FS CF )
Agricultural soil is a primary source of anthropogenic GHGs.Soil is a complex dynamic system encompassing several boundaries and inventories affected by N fertilizers as the primary input.In addition to containing CO 2 during manufacture, N fertilizers stimulate soil respiration; N 2 O, NO, NH 3 , and CH 4 emissions; NO − 3 leaching; and C sequestration via soil organic C (SOC).Therefore, a bottomup approach was implemented using the following general equation (Equation 5) to calculate the FS CF (kg CO 2 eq.kg −1 N) induced by N fertilization (IPCC, 2006a;Kayatz et al., 2020) (Supporting where E CO 2 refers to the direct emissions of CO 2 (kg CO 2 eq.kg −1 N),

The carbon sequestration in the crop biomass ( )
The C sequestration in the plant biomass induced by N fertilization (P CO 2 ) (kg CO 2 kg −1 N) was calculated using the following equation (Equation 6) according to Dubey and Lal (2009) where C G , C S , and C R are the C ratios in grain (42%, 40%, and 41%), straw (44%, 42%, and 43%), and root (38%, 40%, and 38%) for maize, rice, and wheat respectively (Dubey & Lal, 2009).GB, SB, and RB are the grain, straw, and root amounts induced by kilogram of N fertilizer (kg kg −1 N).The converting coefficient of C to CO 2 is 44/12.

The CF of the produced biomass (FC CF )
This stage included the direct and indirect N 2 O emissions of returned plant residues and N 2 O and CH 4 emissions of burned residues.
The FC CF (kg CO 2 eq.kg −1 N) was given by the following equation (Equation 7), based on the IPCC Guidelines for National greenhouse gas Inventories (Tier 2) and FAOSTAT databases (Supporting Information Section S1; Data S8).
where  , 2006a, 2007).Finally, we used the NAE and the total grain production to calculate the FC CF as kg CO 2 eq.Mg −1 grain and at country level (Gg CO 2 eq.) (Data S8).

| Grave stage
The grave stage included the CF of the grain and straw produced by The N feeding ratios (%) of straw and cereal grain were calculated from FAO databases, following Mottet et al. (2017), using the total feed consumption by the livestock of all sources and the feeding amount of maize, rice, and wheat.The FCR is the feed conversion ratio calculated from FAO databases (FAOSTAT, 2023) using the amount of livestock products and feed consumption.NAE (kg crop kg −1 N) represents the feed production amount by a unit of N fertilizers.The EF factors were calculated as an average of all livestock types and products (meat, milk, eggs, etc.) (Data S9). (4) The CF of grain-food consumption (FO CF ) (kg CO 2 eq.kg −1 N) was calculated using the following equation (Equation 9), based on a comprehensive dietary emission, and data from different sources (Crippa et al., 2022;Ivanovich et al., 2023;Poore & Nemecek, 2018).
The EF Transportation (kg CO 2 eq.kg −1 food) included both national and international transportation of the grain to the processing units and was generated from (Forum, 2017;Li et al., 2022) based on the multi-region input-output (MRIO) model (Lenzen et al., 2017) (Data S8, Sheet 2).The EF values (kg CO 2 eq.kg −1 food) for processing, packaging, retail, consumption, and end of life were aggregated for various grain products using the data derived from Crippa The CF of biofuel production and combustion (FB CF ) (kg CO 2 eq. kg −1 N) was given by the following equation (Equation 10) based on the IPCC methodology (Jeevan Kumar et al., 2020;Roberts et al., 2022).
where Energy balance (MJ L −1 ) is the difference between the energy output and input during the biofuel production of cereals (Mekonnen et al., 2018).The energy input included energy consumed in the ethanol plant for transportation and processing.The Biofuel combustion CF is the emissions of the biofuel combustion (kg CO 2 equation MJ −1 ) (Roberts et al., 2022), and Biofuel yield (L kg −1 ) refers to the volume of the biofuel produced by 1 kg of crop biomass (0.39, 0.53, and 0.38 L kg −1 of maize, rice, and wheat respectively) (Jeevan Kumar et al., 2020) (Data S11).
Given that the C assimilated in the lost grain is released to the atmosphere, the CF of losses (FL CF ) (kg CO 2 eq.kg −1 N) was calculated using the following equation (Equation 11) (Data S12).
where P CO 2 is the assimilated C in crop biomass (kg CO 2 eq.kg −1 N), The EF Production is the emissions through the production of lost biomass and the loss fraction refers to the fraction of the biomass losses to the total biomass production by N fertilization (FAOSTAT, 2023).
The CF of all stages was calculated for grain, hectare, and country level, as presented in Data S13.

| Future projections and mitigation options
Based on the future projections of food demand until 2100 using the IPCC Special Report on Emissions Scenarios (Bodirsky et al., 2015) and the historical production of grain globally from 1961 to 2019 (FAOSTAT), we interpreted the changes in the grain production needed to meet that future demand by the following equation (Equation 12; Tilman et al., 2011).
where GY and D refer to the grain yield production and demand (Mg), and i and b refer to the future and base (2019) years respectively.
Similarly, we used the historical trend   accompanied by the improvements in NAE (S4).All these scenarios assumed decadal stability in all other criteria (crop genotypes, soil productivity index, etc.).

| Statistical analysis and validations
An independent sample t-test was performed to compare our results with those reported by previous studies at a significant level of p ≤ .05.Pearson correlation and linear regression models were implemented in the R package.Maps were drawn using the "tmap" package (Tennekes, 2018), and other figures were drawn using Origin 2021 (OriginLab, USA).

| Uncertainties and validations
Current analyses of the CF induced by N fertilization of grain crops are based on various calculation methods (bottom-up approaches and data-driven and process-based models) using various data sources (IPCC, IFASTAT, FAOSTAT, and meta-analysis studies).
These approaches are subject to uncertainties from various sources throughout the LCA stages at the global scale, including energy type and amount in fertilizer production, transportation, and application and the type and amount of fertilizers consumed, produced, imported, and exported (Supporting Information Section S3).
Additionally, the area harvested, grain yield, application rate, NAE, and N-type for each crop may induce high uncertainties if they are not addressed.Soil and plant systems exhibit continuous dynamic reactions influenced by ecology, climatic conditions, and agronomic practices.Therefore, gas emission and sequestration inventories of ( 9) these systems are hindered by various uncertainties addressed to the greatest extent possible, as recommended by Tian et al. (2019) (Supporting Information Section S3).The independent t-test showed insignificant differences in the calculated N rate (kg ha −1 ) at the country level between our study and that of Zhang et al. (2015) (Table S9); however, our values were higher owing to the increased N rate with time (FAO, 2019).
Additionally, no significant differences were found between our CF values of fertilizer production and those calculated from http:// www.calcf ert.com and our NAE values calculated using the regression models developed by Guo et al. (2022) and those based on N the budget (FAO, 2019) (Table S9).Due to a data shortage about the specific coefficients of CO 2 emission as a result of the dissolution of calcium carbonate at a country level and for each crop type, we neglected that source of the CF.Here, the CO 2 efflux of soil calcium carbonate is usually neglected (Rey, 2015;Zamanian et al., 2016), although soil inorganic carbon encompasses the majority of the soil C pool in semiarid and arid areas (Lal, 2000;Wang et al., 2015); this is because the calcium carbonates are concentrated in the subsoil (Wang et al., 2010), which reduces their exchange with the atmospheric CO 2 (Zamanian et al., 2018).
Additionally, the distribution of calcium carbonate within the soil profile is very low, driven by water deficiency in semiarid and arid regions (Zamanian et al., 2016).
The CF of biomass produced with N fertilizers through feed, food, biofuel, and losses has not been introduced before; therefore, the comparison with other data sources was impossible.Due to data limitations at the country level of feed and food supply chain, the aggregations made in this study may have uncertainties ranging between 20% and 30% of the 95% confidence interval for livestock feed (Gerber et al., 2013) and less than 30%-70% for food supply chain (Crippa et al., 2021).

| Boundaries of the CF of N fertilization
Our LCA showed that N fertilization of grain crops resulted in CF values of 2878, 1715, and 853 kg CO 2 eq.Mg −1 grain in 2019 as a global average of maize, rice, and wheat respectively (Figure 1).However, rice had the highest soil CF (1470 kg CO 2 eq.Mg −1 grain) as compared with maize (1010 kg CO 2 eq.Mg −1 grain) and wheat (660 kg CO 2 eq. Mg −1 grain).In the LCA, the cradle-stage boundaries (fertilizer production and transportation) had the lowest contribution to the total CF of maize, rice, and wheat by 1.3%, 2.6%, and 2.6% respectively.
In contrast, the gate stage (fertilizer application and soil and crop systems) contributed substantially to the CF by 23%, 35%, and 24% of the total CF respectively (Figure 1).The N 2 O emission due to N fertilization was the main contributor to the CF, accounting for 85% of the soil CF.However, the emitted N 2 O accounted for only 1% of the total soil emissions.In contrast, the N fertilization contributed to the mitigation of the CF by increasing the soil C sequestration by 9%, 8%, and 12% of the soil CF in maize, rice, and wheat respectively.
The N fertilization increased the assimilated C in the plant biomass by an average of 3090 kg CO 2 eq.Mg −1 grain, of which 0.3% and 8.8% were emitted again to the atmosphere through burning and mulching respectively.The CF declined by 49%, 66%, and 79% due to the assimilated atmospheric C in the crop biomass induced by N fertilization of maize, wheat, and rice respectively.
The remaining part of the crop biomass produced by N fertilization was included in the grave stage (food, feed, biofuel, and losses), which had the greatest contribution to the total CF of maize, wheat, and rice by 76%, 63%, and 74% respectively (Figure 1).On average, for the three crops, 2200 kg CO 2 eq.Mg −1 grain produced with N fertilizers was emitted from the food supply chain until the end of life.About half of the global maize production was used in livestock feed, while only 13% of rice and wheat was used in the feed system (Data S14).Therefore, the CF of maize production with N fertilization was the highest of the other two crops, corroborating with the greater CF of the livestock feed (1425 kg CO 2 eq. Mg −1 grain) than rice (160 kg CO 2 eq.Mg −1 grain) and wheat (113 kg CO 2 eq.Mg −1 grain).Biofuel production and losses of crop biomass contributed to the CF by 11% and 3.3% on average of the three crops respectively.

| Emissions per unit of N fertilizer at the country level
The main producers of maize (the United States, China, Brazil, and Argentina) had CF values greater than the global average by twofold (Figure 2a) and accounted for an average of 195 kg CO 2 eq.kg −1 N.
On the other hand, Ukraine (one of the main producers of maize) had a CF value of 77% lower than the global average, corroborating a lower feed ratio (60%).The CF value of the cradle stage was higher in Ukraine (5.7 kg CO 2 eq.kg −1 N) than in other main producers of maize, which accounted for an average of 3.3 kg CO 2 eq.kg −1 N. The CF value of the gate stage in Ukraine was lower (28 kg CO 2 eq.kg −1 N) than that of the other main producers of maize (32 kg CO 2 eq.kg −1 N) (Data S3-S8).
The CF values in China, India, Indonesia, and Vietnam (the main rice producers) were down from the global average by 65%-99% (Figure 2b).The CF values of the cradle and gate stages in these countries (3.2 and 33 kg CO 2 eq.kg −1 N respectively) were lower than the global average of other countries (3.6 and 34 kg CO 2 eq.kg −1 N respectively) (Data S3-S8).The CF values of China and India (the top wheat producers) were lower by 36% and 70% than the global average and accounted for 19 and 40 kg CO 2 eq.kg −1 N respectively (Figure 2c).On the other hand, Russia (one of the main wheat producers) represents the highest increase over the global average of the CF, with a threefold increment (Figure 2c).However, Russia had higher C sequestration in crop biomass than China and India by 29% and 59% respectively (Data S8).

| Total CF of N fertilization
Globally, the maize, rice, and wheat produced by N fertilizers resulted in CF values of 4.5, 0.9, and 2.0 Gt CO 2 eq., accumulating for 7.4 Gt CO 2 eq.(Figure 3).North America had the greatest CF through the grave stage, including feed, food, and biofuel for grain crops; therefore, North America was the top contributor to the CF by 38%, followed by Asia (Top producer of grain crops) and Latin America by 27% and 17% respectively.As the main region of rice and wheat production, Asia had the highest CF induced by N fertilization of those two crops, accounting for approximately 89% and 38% of the global level respectively.The USA and China were the top contributors to the CF of the three crops, accounting for 2.7 and 1.3 Gt CO 2 eq.respectively.The greatest CF value (2.5 Gt CO 2 eq.) of maize was reported in the United States (Figure 4a), while China had the highest CF values of rice (0.6 Gt CO 2 eq.) and wheat (0.4 Gt CO 2 eq.) (Figure 4b,c).
The relative importance of each emission source to the total CF differs among crops and regions.As a global average, the CF of the grave stage contributed 84%, 66%, and 79% to the total CF of maize, rice, and wheat, respectively, where North America, East Asia, and Southeast Asia were the top contributors to the three crops.The gate stage was also an important source of the CF by 14%, 31%, and 19% maize, rice, and wheat, respectively, of which 58%, 76%, and 76% were sourced soil emissions, indicating the great CH 4 emissions from rice fields.Fertilizer production and transportation, which present the cradle stage, had a substantial contribution to the global CF, especially in the main fertilizer producers (Asia, North America, and CIS) (Figure 3).Interestingly, we estimated that 3, 2, and 0.9 Gt of the atmospheric C were assimilated in the biomass of maize,

| Sources of the CF of the N-fertilized grains
The rice grain produced with N fertilizers had higher soil CF values than that of maize and wheat, agreeing well with the findings of Liu et al. (2018), who reported a higher C footprint in rice production than that in maize and wheat by twofold owing to the emitted CH 4 from paddies and that uptake in uplands.However, the total CF of maize was greater than that of rice and wheat, driven by the higher CF of the grave stage, which contributed to the total CF by more than twofold compared to the gate stage.The estimated soil CF was almost identical to the values reported by Hoxha and Christensen (2019) with strong correlation coefficients that range between 0.76 and 0.91 (Table S9).The transportation of N fertilizers had the lowest contribution to the CF (<2.5 kg CO 2 eq.Mg −1 ), agreeing well with the results reported by Walling and Vaneeckhaute (2020).The energy consumption by N fertilizer transportation accounted for only 6% of the total energy consumed by fertilizer production (89%), packaging, transportation, and application (Amenumey & Capel, 2013).The CF of transportation, however, was higher for N importers such as Brazil (the top N importer) by twofold than China (the top N producer) (Data S5).
Despite the smaller contribution, the CF of fertilizer production, transportation, and application correlated strongly with the CF and was higher than that of soil CF (Figure S3).
Agricultural soils are the main source of anthropogenic GHGs such as N 2 O, especially under N fertilization (Tian et al., 2020;Zhang et al., 2013), as the soil CF accounted for one-fifth of the total CF.
The N rate is the main driver of the soil CF because it stimulates the emissions of all N forms, including N 2 O by 1% (IPCC, 2006a; Shcherbak et al., 2014), NO by 0.4% (Kayatz et al., 2020), NH 3 by 12.6% (Zhan et al., 2021), NO 3 by 19% (Zhou & Butterbach-Bahl, 2014), CH 4 in rice by 15% (Banger et al., 2012), and CO 2 by 20% (IPCC, 2006a), while also mitigating the CF due to stimulating the C stored in soil organic matter (Jian et al., 2016).Despite the minor contribution of N 2 O to the total gas emissions of the soil, it was the greatest contributor to the soil CF because it has the highest global warming potential, which is 300-fold that of CO 2 over a 100-year period (IPCC, 2007).The declining CH 4 emissions due to N fertilization contributed weakly to the soil CF mitigation (less than 0.3%) in maize and wheat, while CH 4 emissions raised the soil CF of rice by one-third (Zhang et al., 2018).| 11 of 16 productivity, and in turn, the residues returned to the soil or burned (Tian et al., 2012;Zhang et al., 2016), which caused a decline in the net CF.
The food supply chain was the main source of the CF of grain crops, accounting for half of the total CF.In this respect, Crippa et al. ( 2021) estimated a 32% contribution of the food supply chain to the total GHG emissions of land use change and food production and supply.Our study focused on using N fertilizers only, including all emissions and sequestration sources and excluding land use change.The estimated CF of the grave stage through our analysis is comparable with that reported by Poore and Nemecek (2018), who calculated a total CF of 2700 kg CO 2 eq.Mg −1 of the grain supply chain.
Given that the feed crops accounted for a quarter of the emissions of 150 kg CO 2 eq.kg −1 protein from livestock production, the CF of feed crops was 1250 kg CO 2 eq.kg −1 feed (Gerber et al., 2013), comparable with our estimation (Data S9).

| Difference in the CF at country level
The dispersion coefficients ranged between 79% and 105% among the CF values of N consumption at the country level due to specifics of the fertilizer production technology, crop management practices, and the uses of the produced biomass.The CF of maize in the top producers (the USA, China, Brazil, and Argentina) was higher than the global average because the produced maize was used mainly in livestock feed by an average of 76% (Data S14).Additionally, the N application rate was higher in these countries than the global average by one to five-thirds (Figure S1).Our CF values through the cradle to gate stages are comparable to those estimated for China (Chen et al., 2014) and the United States (Grassini & Cassman, 2012), despite differences in the methods used, such as including soil C sequestration and removing the control effect.
Food is the main path of rice in the grave stage, which comprised lower CF in China, India, Indonesia, and Vietnam by an average of 9% than the global average due to direct consumption of the produced rice with less transportation, processing, and retail activities (Crippa et al., 2021).Losses were higher in these countries by 3.1 kg CO 2 eq.kg −1 N than the global average, but the C assimilated in the rice biomass was greater by 17 kg CO 2 eq.kg −1 N, indicating lower CF in these countries than the global average.The CF values of rice were lower than those of maize due to the greater emissions of maize feed than other crops and the increased effect of the N rate (r = .26) and NAE (r = −.52)(Figure S3), which is in line with previous observations (Chen et al., 2014).This finding explains why the CF values of Bangladesh (one of the top rice producers) were higher than the global average; indeed, the N rate was 12% higher and the NAE was 52% lower than those estimated for China (Figure S1).The utilization of rice biomass in livestock feed in Bangladesh was higher than that in China and India by one to two times approximately (Data S14), indicating greater CF of the grave stage in Bangladesh than in China and India.
The CF values of the cradle and gate stages were two-to fourfold higher in Russia than in India and China, indicating a higher total CF of wheat in Russia than in India and China.The CF of the grave stage in Russia was greater than that in China and India by three to eight times based on higher feed consumption (34%) and losses (17%) (Data S9-S12).These findings are comparable with those reported by Ivanovich et al. (2023), who estimated 1400 kg CO 2 eq.
kg −1 of wheat food across global regions without differences.
The role of the N rate and NAE became more significant in wheat systems (Figure S3) and reflected the magnitudes of the CF values in India across the cradle and gate stages, where the consumption of N fertilizers was 18% higher, and the NAE and C assimilation were 45% and 57% lower than that estimated for China, agreeing with Zhang et al. (2015).Brazil represents a case in which increased CF was observed in the maize system through the cradle stage due to the import of N fertilizers, despite the low N rate and high NAE compared with those in China (Figure 2).

| Global and regional variations in the total CF
Nitrogen fertilization resulted in a total of 7.4 Gt CO 2 eq.emitted from grain production and consumption, accounting for 41% of the total emissions of the food system (Crippa et al., 2021).Here, the underestimation of the CF by N fertilizers was noticed in the previous studies [1.2 Gt CO 2 eq.(Menegat et al., 2022) and 1.1 Gt CO 2 eq.(Gao & Cabrera Serrenho, 2023)], because these studies estimated the emissions of nitrous oxides only at the level of cradle-to-gate LCA.The previous studies at the cradle-to-gate LCA level dealt with the CF of rice from one perspective, which is CH 4 emissions, while neglecting the emissions across the consumption of the crops in feed and biofuel and the sequestered C by crop biomass (Crippa et al., 2021;Ivanovich et al., 2023).Therefore, these studies overestimated the C footprint of rice as compared with other crops, which had the highest contribution to the CF compared with other crops.The rice consumption in animal feeding was lower than that of maize by fourfold and was not used in biofuel production (Data S9-S12); this is consistent with the greatest CF of maize in the United States than other countries, as biofuel production is the main use of maize in the United States (Austin et al., 2022).China is the main producer and consumer of rice and wheat (FAOSTAT, 2023), and China recorded the highest CF values for these crops.
The differences in the CF among crops and regions can also be attributed mainly to the total yield, harvested area, and N rate.
Here, wheat had higher CF values than rice, reflecting the higher grain production by 27% and N usage by 18% in wheat than in rice (FAO, 2019).These results (especially the N rate) also explain the differences in the contributions of the CF of fertilizer production and that of the soil system among the three crops.The occurrence of the highest CF values in Asia, North America, and Europe is associated with the large-scale application of N fertilizers, and this finding is consistent with these areas having higher agricultural sources of anthropogenic N 2 O than other regions (Tian et al., 2020).
The differences in the regional contributions of the LCA boundaries are comparable with the findings of Crippa et al. (2021)  to a healthy diet by reducing the consumption of meat (Ivanovich et al., 2023).Consequently, we found that the CF will rise by only 64% by 2100 if the NAE is held constant with no future decline (Figure 5), which indicates the importance of reducing the emissions of other sources across the cradle and grave stages.This result is consistent with a 40% increase in N 2 O (Tian et al., 2020), the dominant contributor to the soil CF.
Unprecedented global challenges are affecting agriculture, including the need to increase crop production by 60%-100% from 2007 to 2050 to meet the growth in food demand (Alexandratos & Bruinsma, 2012;Searchinger et al., 2014;Tilman et al., 2011), while N inputs have already surpassed the natural planetary boundary (de Vries et al., 2013;Steffen et al., 2015).The key to overcoming this dilemma is to achieve ambitious NAE targets of more than 0.75 in Europe and North America, 0.6 in Asia, and 0.7 in other regions (Zhang et al., 2015).When achieving these targets, we estimated increases in the CF by only 10%, 16%, and 4% in maize, rice, and wheat in response to 6%, 5%, and 7% growth in the NAE, respectively, over a decade.Numerous GHG mitigation options are available for immediate deployment to improve NAE; these options include precise fertilizer delivery, application rate, and timing to meet the crop demand effectively in addition to the split addition method, prevention of waterlogging, conservation tillage, and inhibitor usage (Winiwarter et al., 2018;Zhang et al., 2015).Europe and the United States represent success stories for controlling GHG emissions while maintaining or increasing crop yields (Mueller et al., 2017).
Interestingly, the combined application of the three suggested mitigation options (S4) across the cradle-to-grave LCA will feed the world with maize, rice, and wheat while reducing the future New technologies in ammonia production (the highest energyconsuming process) could also reduce the energy used from 51 to 33 GJ Mg −1 NH 3 (IFA, 2009).The new technologies could decrease the energy needed to form urea, ammonium nitrate, nitric acid, and other N forms (Chai et al., 2019), reducing the soil CF by 43% (Zhang et al., 2013).We suggest that establishing new N-producing factories with advanced technological capabilities in the main importers, such as Brazil, could substantially decrease the cradle stage emissions and fertilizer application emissions.The losses of grain crops across transportation, processing, packaging, and retail represented 5% of the total grain consumption (Data S14), which resulted in 5.4 kg CO 2 eq.kg −1 N. Therefore, controlling the food losses will ultimately contribute to mitigating the CF (Porter et al., 2016), corroborating the findings by Ivanovich et al. (2023), who estimated a 5% reduction in global warming associated with the reductions in food losses.More than 50% of the maize production was used in animal feeding, substantially increasing the CF of N fertilizers.In this respect, a global conversion to a healthy diet by reducing the consumption of livestock products (Green et al., 2015) would considerably mitigate the CF of grain crops produced by N fertilizers (Ivanovich et al., 2023).

| CON CLUS IONS
Nitrogen fertilization should be considered when addressing the triple challenges of food security, climate change, and environmental degradation.We present the first comprehensive CF of the N fertilizers applied to global grain production and consumption.
Nitrogen fertilizers contributed to the global CF by 4.5, 0.90, and 2.0 Gt CO 2 eq.from maize, rice, and wheat production, respectively, in 2019, of which more than three-quarters were sourced from the grave stage (feed, food, biofuel, and losses).One-fifth of the global CF was sourced from the gate stage (fertilizer application and soil and plant emissions in the field).The CF of N fertilization is expected to be doubled by 2100 compared with 2019, responding to the increment in food demand and N fertilization and the decline in nitrogen agronomic efficiency (NAE).
Conserving or improving the current NAE would alleviate this increment in the CF to 10% or 64%, respectively.The future increment in food demand can be met with more N fertilization, while reducing the CF by one-fourth through improving the manufacturing technology of N fertilizers, increasing the NAE, reducing losses, and converting to healthy diets.This will help to curb global warming and meet the standards of the long-term goal of the Paris Agreement.
): (1) fertilizer production, including energy production and supply (cradle stage); (2) transportation of N fertilizers to the field (cradle stage); (3) application of fertilizer (gate stage); (4) the gas emissions and sequestrations of soil as stimulated by N fertilizer (gate stage); (5) the assimilated and emitted gases of the produced biomass as induced by N fertilization (gate stage); and (6) the consumption of the produced biomass until the end of life (grave stage).Our LCA included five types of N fertilizers (urea, ammonium nitrate, urea ammonium nitrate, calcium ammonium nitrate, and ammonium sulfate), representing 99.9% of the total N production and consumption (FAOSTAT; IFASTAT), with the three staple crops consuming 51.5% of global N usage (IFA, 2017).

D
N C is the national distance (Data S5).The fractions of nationally produced N and imported N in the FT CF values were calculated based on the N consumption of nationally produced N (W National ), internationally imported N (W International ), and total N consumption (W Total ) by each crop (Data S5).
emissions of the applied N (kg kg −1 N), and E NH 3 and E NO (kg kg −1 N) are the indirect emissions via ammonia and NO losses after conversion to N 2 O using a factor of 0.01 (IPCC, 2006a, 2006b), E NO 3 (kg kg −1 N) refers to the emitted N 2 O equivalent to nitrate leaching, 0.0075 is the N 2 O emission factor of N leaching (IPCC, 2006a), E CH 4 is the emitted methane (kg kg −1 N), and 298 and 25 are the CO 2 equivalency factors for the emitted N 2 O and CH 4 respectively (IPCC, 2006a).Here, the data used in the bottom-up approach (Data S7) were collected from recent meta-analysis studies [E CO 2 : Zhou et al. (2014), S CO 2 : Geisseler and Scow (2014), Jian et al. (2016), E N 2 O : Shcherbak et al. (2014), rice E CH 4 : Banger et al. (2012), maize and wheat E CH 4 : Chen et al. (2021), E NH 3 : Zhan et al. (2021), E NO : Kayatz et al. (2020), and E NO 3 : Wang et al. (2019)].
et al. (2021) and Poore and Nemecek (2018) based on the food utilization databases by FAO (Data S10).
in the N surplus calculated byZhang et al. (2015) to forecast the changes in the NAE due to N losses when no actions to improve N use efficiency (NUE) shall be done using a linear regression model (R 2 = .83).We then divided the grain yield (Mg) by the NAE (Mg kg −1 N) to calculate the required N (kg), which was used to calculate the future CF of N fertilization.Here, we investigated four scenarios: a decrease in NAE (S1), no changes in NAE or manufacture technology (S2), and decade improvements in NAE by 6%, 5%, and 7%, which was assumed based on the NUE targeted byZhang et al. (2015) (S3).Additionally, we suggested an integrated mitigation strategy with various options, including a decadal reduction in the CF of fertilizer production, transportation, and application by 5%(Zhang et al., 2013) and a 5% decadal reduction in the losses and food emissions by conversion to healthy diets(Ivanovich et al., 2023)

E 2
The carbon footprint (CF) induced by N fertilizers (kg CO 2 eq.kg −1 N) of maize (a), rice (b), and wheat (c) in 2019 at country level.The column graph shows the differences in the CF among the top producers of maize (d), rice (e), and wheat (f) across the stages of the LCA (cradle, gate, and grave).Data about the production amount of grain crops at country level are presented in Data S1 based on FAO databases.The CF is the net CF after deducting the assimilated atmospheric carbon in the crop biomass.The detailed values of emissions and sinks of greenhouse gases within each boundary are available in the Supporting Information.The data of this figure are presented in Supplementary Figure Data (Data S2).Map lines delineate study areas and do not necessarily depict accepted national boundaries.LCA, life cycle assessment.F I G U R E 3 The total carbon footprint (CF) induced by N fertilizers through production and consumption of maize (a), rice (b), and wheat (c) in 2019.The left part of the Sankey diagram represents the regional CF, the middle part shows the main stages of the LCA, and the right part shows the boundaries of the LCA.FP CF and FT CF are the CFs of fertilizer production and transportation, respectively (cradle stage).FA CF , FS CF , EPB, and EPR are the CFs of fertilizer application (spreading), soil system, biomass burning, and biomass mulching, respectively (gate stage).Food, feed, biofuel, and losses represent the CFs of different uses of the biomass produced by N fertilizers (grave stage).The data of this figure are presented in Supplementary Figure Data (Data S3).LCA, life cycle assessment.wheat, and rice, respectively, due to N fertilization, indicating higher net CF in wheat than in rice.
Nitrogen fertilization stimulates soil C sequestration (Wu et al., 2019; Zang et al., 2016), crop growth, and F I G U R E 4 The total carbon footprint (Gt CO 2 eq.) of maize (a), rice (b), and wheat (c) at country level in 2019.The data of this figure are presented in Supplementary Figure Data (Data S4).Map lines delineate study areas and do not necessarily depict accepted national boundaries.

4. 4 |
Historical trend and future scenariosMitigation strategies for GHGs are vital to obviate the gravest consequences of the climatic changes under the expected growth in the global population of 42% by 2100(UN, 2019).In this respect, we calculated a 9-to 16-fold increment in the CF under N fertilization of the three crops from 1961 to 2019 due to the fivefold growth in grain production and N consumption (FAO, 2019), accompanied by a 22% decline in the NAE(Zhang et al., 2015) (Figure5).The CF of grain crops produced by N fertilizers will increase from 7.4 Gt CO 2 eq. in 2019 to 21.2 Gt CO 2 eq.until 2100, when no actions (S1) are implemented to stop that decline in the NAE.Here, we suggested and evaluated three mitigation options across the cradle-tograve LCA to reduce the CF induced by N fertilizers until 2100.The first option was at the cradle stage to improve fertilizer production technology and decrease transportation emissions(Gao & Cabrera Serrenho, 2023).The second option at the gate stage encompasses the decline in the emissions of fertilizer application and improving the NAE by decreasing soil emissions through various cultivation practices(Abdo et al., 2021;Gao & Cabrera Serrenho, 2023).The third option at the grave stage is reducing the losses and converting CF of N fertilizers by 19%, 9%, and 26%, respectively, until 2100.That means only 6 Gt CO 2 eq.will be released into the atmosphere from grain production by N fertilization in 2100.The improvements in the manufacturing technologies of N fertilizers include but are not limited to, increasing the recovery of emitted CH 4 during energy mining (EPA, 2009) and improving the efficiency of energy combustion for electricity production(Chai et al., 2019).F I G U R E 5 Historical trend and future scenarios of carbon footprint (CF) induced by N fertilization from 1960 to 2100.The S1 scenario represents a decrease in the NAE.The S2 scenario indicates no changes in the NAE or production technology.The S3 scenario refers to decade improvements in the NAE by 6%, 5%, and 7% in maize (a), rice (b), and wheat (c).The S4 scenario indicates a decade reduction in the CF of fertilizer production, transportation, and application by 5% accompanied by the improvements in NAE, in addition to a 5% decadal reductions in the CF through decreasing losses and global conversion to a healthy dietary.Detailed information of this figure is presented in Supplementary Figure Data (Data S5).NAE, nitrogen agronomic efficiency.
EPR N 2 O is the nitrous oxide emissions of plant residues returned to soil (kg N 2 O kg −1 N).EPB CO 2 , EPB N 2 O , and EPB CH 4 are the emitted GHGs of burnt plant residues (kg kg −1 N).Values 298 and 25 are the CO 2 equivalency factors for the emitted N 2 O and CH 4 respec- , who found that the CF of agriculture contributed to the food supply chain by 10% in industrialized countries, while this contribution increased to 15% in developing countries.The soil CF contributed more to the CF in Africa, the Middle East, and Latin America than in Europe, reflecting the greater NUE in Europe than in the former regions(Zhang