Background & Summary

Nitrogen is an important element for life and dominates Earth’s atmosphere, yet over 99% remains inaccessible to most organisms1. Reactive nitrogen (Nr or fixed N), readily available to life forms, profoundly influences ecosystem productivity2. An overabundance of reactive nitrogen, mainly in the forms of ammonia (NH3), nitrogen oxides (NOx), and nitrous oxide (N2O), can considerably deteriorate air quality and adversely impact global climate3,4. NH3 is a crucial precursor to PM2.5 (particulate matter with an aerodynamic diameter less than 2.5 μm) pollution, contributing to the formation of secondary inorganic aerosols5, secondary organic aerosols6, and brown carbon7,8. NOx acts as a precursor to both PM2.5 and O3, thus alterations in Nr emissions can profoundly impact the levels of PM2.5 and O3 pollution. The presence of fine particles in the atmosphere notably impacts climate dynamics by modifying radiative transport, as well as altering the optical properties and lifespan of clouds9. N2O is acknowledged as one of the most critical greenhouse gases. In the stratosphere, N2O is a potent ozone-depleting substance10. Excessive deposition of Nr in terrestrial and aquatic ecosystems leads to soil acidification, eutrophication of water bodies, and jeopardizes ecosystem biodiversity1,11. Enhancing Nr emissions monitoring can markedly help to achieve multiple Sustainable Development Goals (SDGs), including SDG2 Zero Hunger, SDG 3 Good Health and Well-Being, SDG 13 Climate Action, and SDG 15 Life on Land.

Livestock is a major contributor to both air pollution and climate change12. As the most abundant alkaline gas, over 85% of NH3 originated from agriculture13,14,15, with 33% from nitrogen fertilizer application and 54% from livestock husbandry systems13. Other sources include the chemical industry16, human waste17, transportation18,19, and rewetting processes in soils20. NOx emissions are predominantly emitted as a byproduct from combustion processes21. Over 5% of NOx is emitted from soil22 which predominantly depends on nitrogen injection from agriculture activities. In natural ecosystems, NOx and N2O are emitted through microbial processes like nitrification and denitrification in soils23. Thus, livestock excretion could further contribute to the release of NH3, NOx, and N2O, acting as an important fertilizer. Agricultural emissions, particularly from livestock, are the largest anthropogenic N2O sources24 and have risen by more than 45% since 198025, mainly due to the increased use of nitrogen fertilizer and manure24. In addition, industrial, specifically chemical processing, wastewater, the combustion of fossil fuels, and biomass burning contribute a minor percentage of N2O emissions25.

China stands out as a significant hotspot for NH3 and NOx concentrations as observed from space26,27,28,29. This is attributed to large population, rapid industrialization, urban expansion, and substantial agricultural demands30. Approximately 15% of China’s land experiences nitrogen deposition exceeding critical loads31. China contributes over 16%, 21%, and 16% of global NH3, NOx, and N2O emissions respectively (EDGAR: Emission Database for Global Atmospheric Research v8.1 (https://edgar.jrc.ec.europa.eu/dataset_ap81) for NH3 and NOx, EDGAR 2024 GHG (https://edgar.jrc.ec.europa.eu/dataset_ghg2024) for N2O). As chemical products from SO2, NH3, and NOx, sulfate-nitrate-ammonium (SNA) aerosols account for 40–57% of PM2.5 concentrations in China5,32,33. Soil NOx emissions significantly reduce the sensitivity of surface ozone to anthropogenic emissions in North China and will therefore reduce the effectiveness of ozone mitigation policies34.

Recently, there has been a notable change in Nr emissions in China, yet existing high-resolution estimates of livestock Nr emissions primarily focus on singular Nr species and are mostly outdated (Table 1). NH3 and NOx concentrations in China have changed markedly since 2000, as observed from satellites35,36,37. With a growing population and a shift toward more meat-intensive diets, China has experienced a notable increase in meat demand38. Aiming to further enhance the overall environmental conditions, Chinese authorities have begun promoting the reduction of NH3 emissions39,40. Therefore, it is crucial to assess the changes in Nr emissions from livestock husbandry in China over the past decade. While numerous research has investigated changes in NH3 emissions recent years41,42,43,44, most studies focused on emissions only up to 2018 in China45. It is essential to capture emissions data from more recent years to effectively shape environmental policies. NOx emissions from agriculture, particularly livestock, have received scant attention due to the perception of agriculture as a minor NOx contributor compared to other sources46,47. Manure-derived NOx emissions are critical in influencing local and regional ozone formation and air quality, particularly in agricultural areas with high livestock densities48. Unlike NOx, agriculture stands out as a substantial contributor to N2O emissions49. There is plenty of research aiming to estimate N2O emissions in China, but most of these studies took place in or before 200850,51,52,53,54. In addition, it is worth noting that there has been a relative dearth of research focused on estimating N2O emissions from livestock in China55. Furthermore, most previous studies have focused on individual Nr species, without considering the interactions between different Nr species. To address these issues, here we develop a high-resolution Nr emission inventory from livestock in China during 2005–2022, that simultaneously considers NH3, NOx, and N2O, based on our previous research on NH3 emission inventory15,56. The datasets can be downloaded freely through Figshare57. Our high-resolution, long-term, animal-specific Nr emissions inventory offers a robust basis for designing mitigation policies. These include strategies such as low-protein animal feed, updated livestock housing structures, the implementation of manure acidification, and deep manure application, all aimed at mitigating air pollution, climate change and promoting sustainable agriculture.

Table 1 Comparison with previous studies for Nr emissions from livestock in China.
Full size table

Methods

Livestock distribution in different raising systems

The spatial distributions of livestock (including dairy cattle, beef cattle, sheep, goats, pigs, donkeys, mules, camels, horses, rabbits, broilers, meat duck, meat geese, laying chicken, laying duck, and laying geese) originate from diverse sources. Buffaloes, cattle, chickens, ducks, horses, goats, pigs, and sheep were sourced from the Gridded Livestock of the World (GLW, http://www.fao.org/livestock-systems/global-distributions/) dataset58 (Fig. S1). We calculated three sets of annual livestock density from 2005 to 2022. Each set was based on a different reference year for livestock distribution data: 2010, 2015, and 2020. Due to the lack of updated spatial distribution data for horses and ducks in 2020, we applied the baseline distribution from 2015 for these two species in the 2020-based dataset. The original resolution of the livestock density data was 5 arc-minutes and was resampled to 0.1°. For rabbits, their distribution correlated with human population density, based on data from the Gridded Population of the World (GPW) v4, which was obtained from NASA (https://www.earthdata.nasa.gov/data/projects/gpw). Specifically, population data from 2005, 2010, 2015, and 2020 were used to represent the periods 2005–2007, 2008–2012, 2013–2017, and 2018–2022, respectively. The original resolution of the human population density data was 2.5 arc-minutes and was resampled to 0.1°. Donkeys, mules, and camels followed the distribution pattern of horses, while geese distribution mirrored that of ducks. We adjusted the Average Annual Population (AAP) (Eqs. 1, 2) for each animal category with provincial annual records from the Chinese Animal Husbandry and Veterinary Yearbook (2006–2023)59, the National Bureau of Statistics of China60 and website of National Bureau of Statistics(https://data.stats.gov.cn/easyquery.htm?cn=C01) (Eq. 3).

The AAP represents the average population of a specific animal category maintained throughout the year. For animals with lifespan over a year, we calculated AAP following Eq. 1.

$${AAP}={n}_{{places}}\times (1-{t}_{{empty}}/365)$$
(1)

where tempty is the average annual duration the place is unoccupied, and nplaces is the number of places. If the lifespan of an animal is under a year, multiple production cycles would occur annually. In this case, we calculated the AAP by Eq. 2, where nprod is the number of animals produced per year according to statistics, nround is the number of production cycles or rounds each year, xns is the proportion of animals that die without being sold. The definition and calculation equations for the AAP are derived from EEA (2019)16 (European Environment Agency).

$${AAP}={n}_{{prod}}/({n}_{{round}}\times \left(1-{x}_{{ns}}\right))$$
(2)
$${AAP}_{i,j,l,p}={AAP}_{i,j,l,p{\rm{\_}}org}\times \frac{{AAP}_{p,l}}{\sum _{i}\sum _{j}{AAP}_{i,j,l,p{\rm{\_}}org}}$$
(3)

Where gridded \({{AAP}}_{i,j,l,p}\) is AAP in grid i,j in province p for livestock l, \({{AAP}}_{i,j,l,{p\_org}}\) is obtained from GLW, and AAPp,l is calculated AAP in province p for livestock l based on statistical data from the yearbook (2006–2023).

$${{AAP}}_{i,j,{sub},p}={{AAP}}_{i,j,l,p}\times {r}_{{sub},p}$$
(4)

Where gridded \({{AAP}}_{i,j,{sub},p}\) is AAP in grid i,j in province p for subcategory sub, and \({r}_{{sub},p}\) is subcategory rate in province p.

The adjustment excludes Hong Kong, Macao, and Taiwan due to the absence of statistical data. It is crucial to categorize the AAP into different raising systems to align with different emissions factors. Livestock raised by grazing systems are situated in pastoral and semi-pastoral areas. We estimated the AAP in the intensive raising system, with a threshold value for each livestock category as provided by Kang et al.42, while other animals were assumed to be raised in the free-range system. Additionally, 23 subcategories were divided based on gender and age (Eq. 4) (Distributions in Figs. S2, S3). Pigs were subdivided into sows, weaners, nursing pigs, and fattening pigs based on statistical data from Fu et al.61 Similarly, dairy cattle, beef cattle, sheep, and goats were classified into older (>1 year) and younger (<1 year) groups, respectively, with ratios derived from Huang et al.13

The combined N-flow and TAN-flow method for estimating Nr emissions from livestock

Livestock excrement releases gaseous NH3, NOx, and N2O during housing, storage, and spreading onto land. Most earlier inventories estimate Nr emissions using annual emission factors per animal14,42,62,63,64,65. Recently, the mass-flow approach has been widely used in China13,42,43,66 and endorsed by the Ministry of Ecology and Environment of China67. In our previous research, we used the TAN (Total ammoniacal nitrogen) mass-flow approach to estimate the NH3 emissions from agriculture in China15,56. Our NH3 emissions inventory enables chemical transport models to effectively capture the spatial patterns of NH3, secondary inorganic aerosol concentrations, and nitrogen deposition in China56. TAN refers to ammonium-N and nitrogen compounds such as urea and uric acid that are readily broken down to ammonium68. Tracing the TAN through manure management systems enables more accurate estimations at each step, as NH3 and other gaseous nitrogen emissions originate from TAN. Given that TAN and N (nitrogen) respond differently to changes in livestock diets and manure management, it’s essential to track TAN and N separately69,70. Besides, estimating NOx and N2O emissions requires simultaneous calculation of N mass flow applied to the soil. Therefore, in this work, we updated the emission factors based on the European Environment Agency16 and added the N-mass flow pathway to calculate the NH3, NOx (presented as NO in the following text), and N2O emissions from livestock in the resolution of 0.1° × 0.1° from 2005–2022 in China (Fig. 1).

Fig. 1
figure 1

Combined TAN-flows and N-flows in livestock manure management system (EEA, 2019) Nr emissions from the pool of TAN are present in blue and Nr emissions based on N are in purple.

Full size image

The combined N-flow and TAN-flow methods are adopted to calculate the NH3, NO, and N2O emissions from each stage of livestock manure management. Following are descriptions and equations of detailed calculations derived from EEA16, IPCC71 (Intergovernmental Panel on Climate Change), and NDRC72 (National Development and Reform Commission).

Excreta N (Nex) from each subcategory are calculated through Eq. 5, Nrate is the total annual excretion of N by the animals, unit: kg AAP−1 a−1.

$${N}_{{ex}}={AAP}\times {N}_{{rate}}$$
(5)

The amount of Nex in buildings where livestock are housed (mN_indoor), on uncovered yards, or during grazing (mN_outdoor) is estimated by Eqs. 6, 7. The proportions depend on the fraction of time that livestock spends in buildings, yards, and grazing land throughout the year, presented as xindoor and xoutdoor.

$${m}_{N{\rm{\_}}{indoor}}={N}_{{ex}}\times {x}_{{indoor}}$$
(6)
$${m}_{N{\rm{\_}}{outdoor}}={N}_{{ex}}\times {x}_{{outdoor}}$$
(7)

Excreta TAN indoor and outdoor are calculated by mN and the proportion of the N excreted as TAN (xTAN).

$${m}_{TAN{\rm{\_}}indoor}={m}_{N{\rm{\_}}indoor}\times {x}_{TAN}$$
(8)
$${m}_{{TAN}{\rm{\_}}{outdoor}}={m}_{N{\rm{\_}}{outdoor}}\times {x}_{{TAN}}$$
(9)

For excreta left on grazing land or yard, NH3, NO, and N2O emissions are calculated through Eqs. 10–12.

$${E}_{{outdoor}{\rm{\_}}{{NH}}_{3}}={m}_{{TAN}{\rm{\_}}{outdoor}}\times {{EF}}_{{outdoor}{\rm{\_}}{{NH}}_{3}}$$
(10)
$${E}_{{outdoor}{\rm{\_}}{NO}}={m}_{{TAN}{\rm{\_}}{outdoor}}\times {{EF}}_{{outdoor}{\rm{\_}}{NO}}$$
(11)
$${E}_{{outdoor}{\rm{\_}}{N}_{2}O}={m}_{{TAN}{\rm{\_}}{outdoor}}\times {{EF}}_{{outdoor}{\rm{\_}}{N}_{2}O}$$
(12)

Excreta as liquid slurry and as solid have different emission parameters, so Eqs. 13–16 are used to calculate the amounts of TAN and total N deposited during the housing stage as slurry (\({m}_{{TAN\_housing\_slurry}},{m}_{{N\_housing\_slurry}}\)) and solid (\({m}_{{TAN\_housing\_solid}},{m}_{{N\_housing\_solid}}\)). xslurry is the proportion of livestock manure handled as slurry, and the remainder is manure handled as solid.

$${m}_{N{\rm{\_}}{housing}{\rm{\_}}{slurry}}={m}_{N{\rm{\_}}{indoor}}\times {x}_{{slurry}}$$
(13)
$${m}_{{TAN}{\rm{\_}}{housing}{\rm{\_}}{slurry}}={m}_{{TAN}{\rm{\_}}{indoor}}\times {x}_{{slurry}}$$
(14)
$${m}_{N{\rm{\_}}{housing}{\rm{\_}}{solid}}={m}_{N{\rm{\_}}{indoor}}\times (1-{x}_{{slurry}})$$
(15)
$${m}_{{TAN}{\rm{\_}}{housing}{\rm{\_}}{solid}}={m}_{{TAN}{\rm{\_}}{indoor}}\times {(1-x}_{{slurry}})$$
(16)

NH3 emissions during housing are calculated by multiplying the amount of TAN (mTAN_housing) and corresponding \({{EF}}_{{housing\_}{{NH}}_{3}}\) for both slurry and solid manure.

$${E}_{{housing}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}={m}_{{TAN}{\rm{\_}}{housing}{\rm{\_}}{slurry}}\times {{EF}}_{{housing}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}$$
(17)
$${E}_{{housing}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}={m}_{{TAN}{\rm{\_}}{housing}{\rm{\_}}{solid}}\times {{EF}}_{{housing}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}$$
(18)

Equations 19–21 are applied to only solid manure. A fraction of TAN is immobilized in organic matter (fimm) when manure is managed as a litter-based solid and the litter is straw. The immobilization can greatly reduce the potential NH3 emissions afterward, such as during storage and after application. The addition of N in animal bedding (mbeding) for each category of livestock is derived from table 3.7 in EEA16. The amounts of straw used and the N inputs (\({m}_{{N\_bedding}}\)) are estimated by Eq. 19.

$${m}_{N{\rm{\_}}{bedding}}={m}_{{bedding}}/100$$
(19)
$${m}_{{ex}-{TAN}{\rm{\_}}{housing}{\rm{\_}}{solid}}={m}_{{TAN}{\rm{\_}}{housing}{\rm{\_}}{solid}}-({E}_{{housing}{\rm{\_}}{solid}{\rm{\_}}{NH}3}+({m}_{{bedding}}\times {f}_{{imm}}))$$
(20)
$${m}_{{ex}-N{\rm{\_}}{housing}{\rm{\_}}{solid}}={m}_{N{\rm{\_}}{housing}{\rm{\_}}{solid}}+{m}_{N{\rm{\_}}{bedding}}-{E}_{{housing}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}$$
(21)

Manure on housing will be collected for storage. The amounts of TAN and total N to storage are calculated through Eqs. 22–25.

$${m}_{{TAN}{\rm{\_}}{storage}{\rm{\_}}{solid}}={m}_{{ex}-{TAN}{\rm{\_}}{housing}{\rm{\_}}{solid}}$$
(22)
$${m}_{N{\rm{\_}}{storage}{\rm{\_}}{solid}}={m}_{{ex}-N{\rm{\_}}{housing}{\rm{\_}}{solid}}$$
(23)
$${m}_{{TAN}{\rm{\_}}{storage}{\rm{\_}}{slurry}}={m}_{{TAN}{\rm{\_}}{housing}{\rm{\_}}{slurry}}-{E}_{{housing}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}$$
(24)
$${m}_{N{\rm{\_}}{storage}{\rm{\_}}{slurry}}={m}_{N{\rm{\_}}{housing}{\rm{\_}}{slurry}}-{E}_{{housing}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}$$
(25)

For slurries, a part of the organic N will be mineralized (fmin) to TAN before Nr gas emissions during storage, as calculated through Eq. (26).

$${{mm}}_{{TAN}{\rm{\_}}{storage}{\rm{\_}}{slurry}}={m}_{{TAN}{\rm{\_}}{storage}{\rm{\_}}{slurry}}+({{m}_{N{\rm{\_}}{storage}{\rm{\_}}{slurry}}-{m}_{{TAN}{\rm{\_}}{storage}{\rm{\_}}{slurry}})\times f}_{\min }$$
(26)

Equations 27–34 are used to calculate the gaseous N emissions (Estorage) during storage for both slurry and solid with corresponding quantities (mTAN) and emissions factors (EF).

$${E}_{{storage}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}={{mm}}_{{TAN}{\rm{\_}}{storage}{\rm{\_}}{slurry}}{\times {EF}}_{{storage}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}$$
(27)
$${E}_{storage{\rm{\_}}solid{\rm{\_}}{NH}_{3}}={m}_{TAN{\rm{\_}}storage{\rm{\_}}solid}{\times EF}_{storage{\rm{\_}}solid{\rm{\_}}{NH}_{3}}$$
(28)
$${E}_{storage{\rm{\_}}slurry{\rm{\_}}{N}_{2}O}={mm}_{TAN{\rm{\_}}storage{\rm{\_}}slurry}{\times EF}_{storage{\rm{\_}}solid{\rm{\_}}{N}_{2}O}$$
(29)
$${E}_{storage{\rm{\_}}solid{\rm{\_}}{N}_{2}O}={m}_{TAN{\rm{\_}}storage{\rm{\_}}solid}{\times EF}_{storage{\rm{\_}}solid{\rm{\_}}{N}_{2}O}$$
(30)
$${E}_{storage{\rm{\_}}slurry{\rm{\_}}NO}={mm}_{TAN{\rm{\_}}storage{\rm{\_}}slurry}{\times EF}_{storage{\rm{\_}}solid{\rm{\_}}NO}$$
(31)
$${E}_{storage{\rm{\_}}solid{\rm{\_}}NO}={m}_{TAN{\rm{\_}}storage{\rm{\_}}solid}{\times EF}_{storage{\rm{\_}}solid{\rm{\_}}NO}$$
(32)
$${E}_{storage{\rm{\_}}slurry{\rm{\_}}{N}_{2}}={mm}_{TAN{\rm{\_}}storage{\rm{\_}}slurry}{\times EF}_{storage{\rm{\_}}solid{\rm{\_}}{N}_{2}}$$
(33)
$${E}_{storage{\rm{\_}}solid{\rm{\_}}{N}_{2}}={m}_{TAN{\rm{\_}}storage{\rm{\_}}solid}{\times EF}_{storage{\rm{\_}}solid{\rm{\_}}{N}_{2}}$$
(34)

Equations 35, 36 sum the gaseous N loss during manure storage.

$${E}_{{storage}{\rm{\_}}{slurry}}={{E}_{{storage}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}+E}_{{storage}{\rm{\_}}{slurry}{\rm{\_}}{N}_{2}O}+{E}_{{storage}{\rm{\_}}{slurry}{\rm{\_}}{NO}}+{E}_{{storage}{\rm{\_}}{slurry}{\rm{\_}}{N}_{2}}$$
(35)
$${E}_{{storage}{\rm{\_}}{solid}}={{E}_{{storage}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}+E}_{{storage}{\rm{\_}}{solid}{\rm{\_}}{N}_{2}O}+{E}_{{storage}{\rm{\_}}{solid}{\rm{\_}}{NO}}+{E}_{{storage}{\rm{\_}}{solid}{\rm{\_}}{N}_{2}}$$
(36)

A part of the manure is used for ecofeed in the intensive system before being applied to cropland. For example, chicken manure is an excellent source of high-protein feed for fish73. Equations 37–41 are used to calculate the quantities of TAN and N in manure applied to cropland.

$${m}_{{TAN}{\rm{\_}}{applic}{\rm{\_}}{slurry}}=({{mm}}_{{TAN}{\rm{\_}}{storage}{\rm{\_}}{slurry}}-{E}_{{storage}{\rm{\_}}{slurry}})\times (1-{feed})$$
(37)
$${m}_{N{\rm{\_}}{applic}{\rm{\_}}{slurry}}=({m}_{N{\rm{\_}}{storage}{\rm{\_}}{slurry}}-{E}_{{storage}{\rm{\_}}{slurry}})\times \left(1-{feed}\right)$$
(38)
$${m}_{TAN{\rm{\_}}applic{\rm{\_}}solid}={(m}_{TAN{\rm{\_}}storage{\rm{\_}}solid}-{E}_{storage{\rm{\_}}solid})\times (1-feed)$$
(39)
$${m}_{N{\rm{\_}}{applic}{\rm{\_}}{solid}}={(m}_{N{\rm{\_}}{storage}{\rm{\_}}{solid}}-{E}_{{storage}{\rm{\_}}{solid}})\times (1-{feed})$$
(40)
$${m}_{N{\rm{\_}}{applic}}={m}_{N{\rm{\_}}{applic}{\rm{\_}}{slurry}}+{m}_{N{\rm{\_}}{applic}{\rm{\_}}{solid}}$$
(41)

NH3 emissions after field application are calculated using EFapplic combined with mTAN_applic through Eqs. 42, 43.

$${E}_{{applic}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}={m}_{{TAN}{\rm{\_}}{applic}{\rm{\_}}{slurry}}\times {{EF}}_{{applic}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}$$
(42)
$${E}_{{applic}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}={m}_{{TAN}{\rm{\_}}{applic}{\rm{\_}}{solid}}\times {{EF}}_{{applic}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}$$
(43)

NO and N2O emissions after field application are calculated based on mN and EFmanure through Eqs. 44, 45, respectively.

$${E}_{{applic}{\rm{\_}}{NO}}={m}_{N{\rm{\_}}{applic}}\times {{EF}}_{{manure}{\rm{\_}}{NO}}$$
(44)
$${E}_{{applic}{\rm{\_}}{N}_{2}O}={m}_{N{\rm{\_}}{applic}}\times {{EF}}_{{manure}{\rm{\_}}{N}_{2}O}$$
(45)

All the emissions of NH3 and NO from the manure management system that are to be reported are summed as Eqs. 46, 47, respectively.

$${E}_{{{NH}}_{3}}={E}_{{outdoor}{\rm{\_}}{{NH}}_{3}}+{E}_{{housing}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}+{E}_{{housing}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}+{E}_{{storage}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}+{E}_{{storage}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}+{E}_{{applic}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}+{E}_{{applic}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}$$
(46)
$${E}_{{NO}}={E}_{{outdoor}{\rm{\_}}{NO}}+{E}_{{storage}{\rm{\_}}{slurry}{\rm{\_}}{NO}}+{E}_{{storage}{\rm{\_}}{solid}{\rm{\_}}{NO}}+{E}_{{applic}{\rm{\_}}{NO}}$$
(47)

Indirect N2O emissions occur as a result of the volatilization and leaching of nitrogen (N) during manure management. The indirect N2O emissions from the volatilization of N in the forms of NH3 and NO are estimated through Eq. 48. \({{EF}}_{{indirect\_deposition\_}{N}_{2}O}\) is emission factor for N2O emissions from atmospheric deposition of N on soils and water surface, kg N2O-N (kg NH3-N + NO-N volatilized)−1, and the default value is 0.01 kg N2O-N (kg NH3-N + NO-N volatilized)−1 71.

$${E}_{{indirect}{\rm{\_}}{deposition}{\rm{\_}}{N}_{2}O}={(E}_{{{NH}}_{3}}+{E}_{{NO}})\times {{EF}}_{{indirect}{\rm{\_}}{deposition}{\rm{\_}}{N}_{2}O}$$
(48)

The amount of manure N that leached from the manure management system (mN_leaching) is calculated by the net amount of N applied to soil from manure (mN_applic) and fraction of managed manure N losses for livestock due to runoff and leaching during solid and liquid storage of manure (fleaching), the typical range is 1–20%, in this paper we use 10% as suggested from the Provincial Greenhouse Gas Inventory Preparation Guidelines72. Indirect N2O emissions from runoff and leaching are calculated through Eq. 50 with mN_leaching multiple with corresponding emission factor (\({{EF}}_{{leaching\_}{N}_{2}O}\)).

$${m}_{N{\rm{\_}}{leaching}}={m}_{N{\rm{\_}}{applic}}\times {f}_{{leaching}}$$
(49)
$${E}_{{indirect}{\rm{\_}}{leaching}{\rm{\_}}{N}_{2}O}={m}_{N{\rm{\_}}{leaching}}\times {{EF}}_{{leaching}{\rm{\_}}{N}_{2}O}$$
(50)

Indirect N2O emissions are the sum of indirect emissions from N deposition (\({E}_{{indirect\_deposition\_}{N}_{2}O}\)) and leaching (\({E}_{{indirect\_leaching\_}{N}_{2}O}\)).

$${E}_{{indirect}{\rm{\_}}{N}_{2}O}={E}_{{indirect}{\rm{\_}}{deposition}{\rm{\_}}{N}_{2}O}+{E}_{{indirect}{\rm{\_}}{leaching}{\rm{\_}}{N}_{2}O}$$
(51)

All the emissions of N2O from the manure management system that are to be reported are summed as Eq. 52.

$${E}_{{N}_{2}O}={E}_{{outdoor}{\rm{\_}}{N}_{2}O}+{{E}_{{storage}{\rm{\_}}{slurry}{\rm{\_}}{N}_{2}O}+{E}_{{storage}{\rm{\_}}{solid}{\rm{\_}}{N}_{2}O}+E}_{{applic}{\rm{\_}}{N}_{2}O}+{E}_{{indirect}{\rm{\_}}{N}_{2}O}$$
(52)

The net amount of TAN and N returned to the soil from manure (mTAN_returned and mN_returned) after loss of NH3, NO, and N2O are calculated through Eq. 53, 54.

$${m}_{{TAN}{\rm{\_}}{returned}}={m}_{{TAN}{\rm{\_}}{applic}{\rm{\_}}{slurry}}+{m}_{{TAN}{\rm{\_}}{applic}{\rm{\_}}{solid}}-{E}_{{applic}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}-{E}_{{applic}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}-{E}_{{applic}{\rm{\_}}{NO}}-{E}_{{applic}{\rm{\_}}{N}_{2}O}-{E}_{{indirect}{\rm{\_}}{N}_{2}O}$$
(53)
$${m}_{N{\rm{\_}}{returned}}={m}_{N{\rm{\_}}{applic}{\rm{\_}}{slurry}}{+m}_{{TAN}{\rm{\_}}{applic}{\rm{\_}}{solid}}-{E}_{{applic}{\rm{\_}}{slurry}{\rm{\_}}{{NH}}_{3}}-{E}_{{applic}{\rm{\_}}{solid}{\rm{\_}}{{NH}}_{3}}-{E}_{{applic}{\rm{\_}}{NO}}-{E}_{{applic}{\rm{\_}}{N}_{2}O}-{E}_{{indirect}{\rm{\_}}{N}_{2}O}$$
(54)

Meteorological conditions are considered to determine the seasonal fluctuations in livestock Nr emissions. For NH3 emissions, adjustments are made for outdoor factors considering air temperature and wind speed (Eq. 55), and indoor emissions are adjusted for air temperature only74 (Eq. 56). The seasonality of outdoor (\({\alpha }_{{meteo\_i}}\)) and indoor (\({\alpha }_{{temp\_i}}\)) emissions factors for month i are estimated by Eqs. 55, 56:

$${\alpha }_{{meteo}{\rm{\_}}i{\rm{\_}}{{NH}}_{3}}=\frac{12\times {e}^{0.0223\times {T}_{i}+0.0419\times {W}_{i}}}{\mathop{\sum }\limits_{j=0}^{12}{e}^{0.0223\times {T}_{j}+0.0419\times {W}_{j}}}$$
(55)
$${\alpha }_{{temp}{\rm{\_}}i{\rm{\_}}{{NH}}_{3}}=\frac{12\times {e}^{0.0223\times {T}_{i}}}{\mathop{\sum }\limits_{j=1}^{12}{e}^{0.0223\times {T}_{j}}}$$
(56)

Where Ti and Wi are air temperature (°C) in 2 meters and wind speed (m s−1) in 10 meters for month i, respectively75,76.

Additionally, an exponential temperature dependence is added to NO emission factors between 0 °C and 30 °C (constant at T > 30 °C)77. A Gaussian distribution for temperature is incorporated for N2O emissions, with 67% occurring between 15 °C and 25 °C53. We obtaine monthly 2 m air temperature and 10 m wind speed meteorological data from ERA5-land (https://cds.climate.copernicus.eu/datasets/reanalysis-era5-land-monthly-means?tab=overview), the fifth generation of the European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric reanalysis for global climate. The data have a spatial resolution of 0.1° by 0.1°.

Data Records

Our dataset is available at Figshare57. It comprises twelve NetCDF (.nc) files that provide gridded data on livestock AAP and monthly livestock emissions of NH3, NO and N2O from 2005 to 2022 in China. Each dataset includes 23 livestock categories and 3 raising systems, provided at a spatial resolution of 0.1° × 0.1°. To reflect different baseline in livestock distribution, three versions are provided for each dataset, based on GLW data from 2010, 2015, and 2020. The dataset is compatible with multiple tools like IDL, Python, MatLab, and others. For each dataset, the base-year versions are available (base2010, base2015, base2020):

  • Livestock AAP: ‘Livestock_distribution_China_2005–2022_base2010.nc’, ‘Livestock_distribution_China_2005–2022_base2015.nc’, ‘Livestock_distribution_China_2005–2022_base2020.nc’;

  • Monthly NH3 emissions: ‘NH3_emission_livestock_China_2005–2022_base2010.nc’, ‘NH3_emission_livestock_China_2005–2022_base2015.nc’, ‘NH3_emission_livestock_China_2005–2022_base2020.nc’;

  • Monthly NOx emissions: ‘NOx_emission_livestock_China_2005–2022_base2010.nc’, ‘NOx_emission_livestock_China_2005–2022_base2015.nc’, ‘NOx_emission_livestock_China_2005–2022_base2020.nc’;

  • Monthly N2O emissions: ‘N2O_emission_livestock_China_2005–2022_base2010.nc’, ‘N2O_emission_livestock_China_2005–2022_base2015.nc’, ‘N2O_emission_livestock_China_2005–2022_base2020.nc’.

  • Each NetCDF file includes the following dimensions and variables:

  • Latitude/Longitude: Gridded coordinates at 0.1° resolution.

  • Time: Annual intervals for AAP and monthly intervals for emissions from 2005 to 2022.

  • Category: 23 livestock types, including beef cattle less than one year, beef cattle greater than one year, dairy cattle less than one year, dairy cattle greater than one year, goats less than one year, goats greater than one year, sheep less than one year, sheep greater than one year, sows, weaners, nursing pigs, fattening pigs, donkeys, mules, camels, horses, rabbits, broilers, meat duck, meat geese, laying chicken, laying duck, and laying geese.

  • Raising systems: 3 raising systems, free-range, intensive and grazing.

The year to year evolution of the AAP of livestock groups from 2005 to 2022 can be captured using our dataset (Fig. 2). For example, cattle decreased by 2.1 million (−18.9%) from 2005 to 2018 due to the Livestock Farming Zoning policy78, then increased by 1.3 million in 2022. Sheep and goats increased by 1.4 million (4.6%) from 2005 to 2015, then decreased to 29.7 million in 2018 before rebounding to 32.6 million in 2022. Total pig numbers increased by 4.7 million (−10.9%) from 2005 to 2012, then slightly decreased by 5.2 million to 2018. The outbreak of the African swine fever virus and the implementation of strict livestock and poultry restricted zone policy79,80 led to a 27.5% (11.8 million) decrease in the AAP of pigs in 2019 compared to 2018, but returned to normal levels in 2022, reaching 45.3 million.

Fig. 2
figure 2

Average annual population (AAP) of each livestock group in China from 2005 to 2022, the y-axis for poultry is one order of magnitude larger than other livestock.

Full size image

Our datasets can capture the spatial distribution and temporal trends of Nr emissions from livestock in China between 2005 and 2022 (Fig. 3 for NH3, Fig. S4 for NO, Fig. S5 for N2O, following analyses derived from dataset calculated based on the 2010 livestock distribution). Annual emissions averaged 5.0 Tg NH3-N, peaking at 5.2 Tg NH3-N in 2005, declining to 4.9 Tg NH3-N in 2007, and stabilizing around 5.0 Tg NH3-N from 2008 to 2015. A significant drop to 4.8 Tg NH3-N in 2019 was linked to a severe disease outbreak among pigs, followed by an increase to 5.5 Tg NH3-N in 2022. Summer emissions contribute 28% of total NH3 due to higher temperatures. Emissions from free-range livestock decreased from 77% in 2005 to 40% in 2022, while emissions from intensive systems rose from 17% to 51%, reflecting shifts in management practices.

Fig. 3
figure 3

(a) Spatial distribution of average livestock NH3 emissions in China (2005–2022); (b) Monthly livestock NH3 emissions for 2005, 2010, 2010, 2015, and 2022; (c) Annual NH3 emissions by livestock type and raising system in China during 2005–2022.

Full size image

Technical Validation

Uncertainty

Uncertainties in Nr emissions from livestock arise from both livestock number and corresponding parameters. N and TAN are calculated based on AAP, excrement rate, nitrogen content, and TAN percentage, with subsequent Nr emissions estimated using corresponding parameters. For lack of local estimates, some parameters are sourced from other countries, such as the immobilized rate (fimm) in Eq. 20, and the mineralized rate (fmin) in Eq. 2616. Strategies such as low-protein feed, modernization of livestock housing structures, the implementation of manure acidification, and the deep application of manure43,81 were introduced to reduce Nr emissions from livestock. Uncertainties may occur due to unchanged parameters, especially in recent years47. Furthermore, NO and N2O emissions from soil can be influenced by environmental factors like temperature, climate condition, soil water content, soil structure, types of nitrogen input, and microorganism activity82,83,84. Except for the temperature, these potential changes are not considered in our study due to parameter limitations.

To address uncertainties, we drew from existing research and determined the probability density function (PDF)13,85,86 (Table S1). We allowed for variations in parameters representing rates or volumes (e.g., excretion rates) within a range of 75% to 125% of the default data86. Emission factors were assumed to span a range of 50% to 150% of the default values. For parameters that represent fractions, we established a broad variation range of −0.25 to +0.25 around the default value, such as immobilization and mineralization rates, ensuring values fall between 0 and 1. We employed the Monte Carlo simulation to evaluate uncertainties in emission inventories42. In this simulation, random numbers were generated from the PDF of input variables, such as activity data, emission factors, and related parameters (Table S1). We conducted 50,000 Monte Carlo simulations to estimate the range of Nr emissions from livestock, with a 95% confidence interval, for each year during 2005–2022. The estimated emissions ranges are as follows: 3.6–6.6 Tg N for NH3; 141.8–254.8 Gg N for NO; 164.5–286.88 Gg N for N2O during 2005–2022 in China (Fig. 4). We anticipate that future field measurements exploring these parameters will be conducted in China to enhance the accuracy of emissions inventory assessments.

Fig. 4
figure 4

Annual total (a) NH3, (b) NO, and (c) N2O emissions from livestock in China from different inventories, the red symbols are single-year estimates. The red triangle, cross, and multiplication symbols are estimates based on years before 2005. The light purple shading in each plot represents the uncertainty associated with livestock Nr emissions, providing a 95% confidence interval for the estimates.

Full size image

Comparisons with previous studies

We compared annual livestock NH3, NO, and N2O emissions with previous research, considering both annual trends and single-year estimates (Fig. 4). Nr emissions from livestock including those from manure management and its application to soil. Notably, emissions from fertilizer application and manure on cropland are typically categorized as “emissions from soils” in most global inventories like the Regional Emissions in ASia (REAS v3.2.1)87, Community Emissions Data System (CEDS)88,89, and EDGAR. To facilitate comparison, Nr emissions from manure applied to soil are extracted based on proportions provided by Yan et al.90. Regarding N2O, indirect emissions estimated from other inventories include those sourced from NH3 and NOx depositions from other sources (like industry). Thus, we solely compare emissions from manure management and direct emissions from manure applied to soil. Given that most studies focus on N2O emissions from soil, we assume N2O emissions from manure management are 20% of total livestock emissions (manure management and manure on soil)50. These unchanged proportions across different years could also contribute to the discrepancies in values and trends when comparing our results with existing research.

Our estimations of NH3 emissions from livestock fall within the range of values reported in existing inventories (Fig. 4). NH3 estimates closely match REAS v3.2.1 and Fu et al.91, but are larger than EDGAR v8.1, Liao et al.45, Liu et al.43, Kang et al.42, and Li et al.92, mostly based on Huang et al.13 (except EDGAR v8.1). EDGAR v8.193 estimates NH3 emissions using country-level activity data from FAO (Food and Agriculture Organization), combined with weighted average emission factors from the IPCC. Our inventory uses provincial-level activity data and distinguishes between free-range, intensive, and grazing systems, which may result in 1.0–2.0 Tg higher NH3-N emissions. Incorporating TAN mineralized from organic nitrogen during storage16, our estimates are approximately 1.0 Tg NH3-N higher than Liao et al.45, Liu et al.43, Kang et al.42, and Li et al.92 Conversely, our estimates are 0.2–0.9 Tg NH3-N lower than CEDS89, possibly due to reliance on emissions factors from other countries and FAO activity data. Our estimates are notably lower than those of Zhang et al.44, which were derived from a mass-balanced model, and were larger than most existing inventories.

Livestock NO and N2O emissions in this study closely match previous local research but show substantial discrepancies with global inventories. Estimations of livestock NO emissions in China vary widely, ranging from 72.490 to 410.894 Gg N. Our results align with Wang et al.95 for higher nitrogen inputs from local observation but lower than Tie et al.94, likely due to their exponential dependence on soil temperature. Those local estimates are higher than the global inventories like CEDS and EDGAR, indicating previous underestimation of agricultural emissions. Our estimation is consistent with Yan et al.96, both nearly doubling of their earlier version90. The N2O emissions from livestock in our study are found to be comparable with those reported in EDGAR v8.1, they are, however, 25.9–54.4% lower than CEDS, Hu and Wang55, and Zhou et al.54, and 30.4–37.3% higher than estimates from FAOSTAT and Ying et al.97. This difference can also be attributed to variations in emissions factors, the livestock population, the proportions of different raising systems, and the assumption of a constant rate of manure management and manure applied to soil. Some environmental factors, like soil structure and climate conditions, are not considered for missing necessary parameters. Due to the different base years, our estimates are higher than Wang et al.51, and Xing and Yan52.

The changes in livestock numbers and raising system have jointly determined the interannual variability of Nr emissions from livestock. From 2005 to 2022, most livestock numbers consistently increased, alongside a shift from free-range to intensive farming systems. Before 2019, the growth in Nr emissions from the rising livestock population was offset by the reductions achieved through more efficient raising systems, leading to stable or even declining Nr emissions. A significant drop of Nr emissions in 2019 was linked to a severe disease outbreak among pigs. After 2019, while livestock numbers continued to rise and the shift to intensive systems persisted, the increase in Nr emissions resulting from the growing livestock population has outweighed the reductions achieved through the adoption of more efficient raising systems. This trend has been particularly pronounced with the recovery of pig populations and the ongoing rise in the numbers of poultry and sheep.

The interannual variability of livestock Nr emissions in our study is consistent with most existing studies (Fig. 4) before 2019 but shows an increasing trend thereafter. In most inventories, livestock Nr emissions have been relatively stable or even decreased slightly from 2005 to 201842,43,45. This trend can be attributed to the transition to more efficient raising systems offsetting any increase in emissions due to population growth. The updated livestock numbers from the NBSC60 have helped to flatten the dramatic decrease in NH3 emissions from livestock that occurred during 2005–2007 in previous works42,43. NH3 emissions in CEDS increased from 2005 to 2012, remaining steady until 2019. This trend may reflect changes in agriculture, including both livestock and fertilizer applications. NH3 emission estimates from Zhang et al.44 showed an increased trend from 2005 to 2015, following the NH3 concentration trend. Previous studies suggest this increase could be associated with the declining emissions of SO2 and NO98,99.

The interannual trends of livestock NO and N2O in this study are aligned with existing research. Like EDGAR v8.1 and CEDS, livestock NO emissions in our estimates are almost unchanged over 2005–2015. After 2015, NO emissions slightly decreased until 2019 and then started to increase, likely influenced by fluctuations in the pig population. The N2O emissions estimated in our study match well with EDGAR 2024 GHG and FAOSTAT. All datasets show a decreasing trend in N2O emissions from 2005 to 2019. Except for 2020 and 2022, our results and N2O emissions from FAOSTAT present an increasing trend while EDGAR 2024 GHG shows no apparent change. The N2O emissions reported in CEDS remained stable over 2005–2019.

Limitations

In summary, our dataset provides livestock AAP and monthly emissions of NH3, NO and N2O from livestock in China from 2005 to 2022. This dataset can serve as a foundational resource assessing emission reduction potential and the environmental benefits of related policies. However, several limitations that should be acknowledged and may be addressed in the future. First, due to lack of monthly livestock population data, the seasonal variability of emissions may not be fully described. Once more detailed and higher-resolution monthly livestock data become available, the accuracy of our monthly livestock emission estimates can be further improved. Second, while our inventory includes most major livestock categories, some animals, such as deer and alpaca, are excluded due to lack of statistical data. Although these animals make up small part of total livestock, their absence may lead to underestimate of total emissions. Third, the emission estimates rely on multiple parameters used, such as immobilization and mineralization rates, are based on default values provided by the EEA. Emission factors for Nr emissions from manure applications vary with soil type, crop type and other environmental conditions. These factors are not considered in this study due to data limitation. These parameters could be better acquainted through local filed observations in the future. Additionally, due to limitations in statistical quality, uncertainty in the statistical data is inevitable. Improving the accuracy of local parameters and statistical data in future work will help reduce uncertainties in livestock emission estimates.