Ammonia (NH₃) emissions from livestock manure and mineral fertilizers constitute a large loss of reactive nitrogen (N) (Amann et al., 2013). Agriculture is the largest source of NH₃ in the atmosphere (Sailesh et al., 2013) and global agricultural NH₃ emissions increased by 90% between 1970 and 2005.

The NH₃ emitted is a threat to human health, because it reacts with acidic compounds in the atmosphere (Walker et al., 2006), forming fine particles (PM2.5) that cause lung diseases (Wang et al., 2017). Ammonia-N deposited to land or waters may exceed the critical N loads of the ecosystems causing eutrophication (e.g., Sutton et al., 2011).

Emissions of NH₃ from livestock manure account for a large but very variable proportion of the plant-available N in the manure, from a few percent to 100% of TAN (e.g., Hafner et al., 2018, 2019). For farmers, the loss of fertilizer N due to NH₃ emissions may significantly reduce N fertilizer efficiency, contributing to low overall use efficiency of applied N, i.e., <50% in the tropics and <70% in temperate areas (Malhi et al., 2001). Experiments on commercial farms in Denmark have shown that N use efficiency of the ammonium (NH4+) in manure is highly variable (Olesen et al., 2009). Consequently, farmers tend to underestimate the fertilizer value of manure, so that the total available N applied (manure + mineral fertilizer) may exceed crop requirements. This increases fertilizer costs, contributes to nitrate pollution of drinking water and aquatic ecosystems, and increases emissions of NH₃ (Sutton et al., 2011).

Concern over the harmful consequences of NH₃ emissions led to the adoption in 1999 of the Gothenburg Protocol of the UNECE Convention on Long-range Transboundary Air Pollution (CLRTAP; UNECE, 2019). The European Union has since enacted the National Emission Ceilings Directive (NECD; EP, 2016), which brings CLRTAP into the EU's legal framework, including the NH₃ emission ceilings applied to the Member States. Both CLRTAP and NECD require that the EU countries must submit an annual NH₃ emission inventory and this calls for models that can be used for calculating national emissions.

The EMEP/EEA air pollutant emission inventory guidebook (“the Guidebook,” European Environment Agency, 2016) provides guidance to European emission inventory compilers on methods for estimating emissions of compounds encompassed by CLRTAP and NECD, including NH₃. The methodologies are divided into Tiers of increasing complexity and detail and for key sources such as NH₃, the use of a Tier 2 or higher methodology is mandatory. The emission factors (EFs) were last systematically reviewed in 2008 and with the limited resources available at the time, the EFs were based on the expert judgement of the EAGER group of researchers (http://www.eager.ch/). The aim of this review was to provide state of art EFs for the Tier 2 calculation of NH₃ emission from the manure management chain. Emission factors have been derived for several livestock categories (e.g., dairy cows, sows, finishing pigs etc.) and for solid or liquid manures. This review was undertaken as part of the revision of the 2016 version of the Guidebook (European Environment Agency, 2016), and the EFs are primarily relevant for European livestock systems but may be applicable to equivalent systems in similar climates.

Data were collected from articles published in peer reviewed journals and from proceeding articles presented at international conferences or published in national reports.

The EFs are expressed as a function of TAN to enable a mass-flow approach to calculation NH₃ emissions so that calculations of emissions during manure storage and following manure application will take account of differences in emissions at early steps in manure management, e.g., housing and storage (Webb and Misselbrook, 2004).

Most of the studies of emission from buildings used in this study are published in peer reviewed papers (Table 4), while reports make up a large fraction of the data on emissions from stored and applied slurry. Emissions from applied solid manure contribute with most of the reported data. The few studies of emission from grazing have all been published in peer reviewed journals.

The number of studies decreased in the order; fattening finishing pigs in buildings with a slatted floor, dairy cows in buildings with a slatted floor, fatteners (solid floor), laying hen, and broilers (solid and slatted floor) (Table 5). There were data available from fewer than 10 studies for dairy cows (solid floor), tied dairy cows and non-dairy cows (solid floor), sows and piglets (solid floor), and laying hen and turkeys on solid floor (Table 5). The variation in measured EF is about 50% of the average in most studies with the exception of EF from pig buildings and turkey. The estimated EF for dairy cows (slurry), fatteners (slurry), fatteners (solid), and sows and piglets are similar to the factors in the current Guidebook (European Environment Agency, 2016). For the other categories the calculated EF deviate by more than 50% from the estimates given in the current Guidebook.

As for building EF, the final means reported here were weighted according to the number of stores from which measurements were taken. The means were not weighted according to the length of the measurement period. There are more studies of emissions from stored solid manure than from stored liquid manure, respectively, 50 and 21 studies (Table 6). The larger number of studies of emission from stored solid manure is not reflected in a low SD in percent of the estimated EF, which with exception of the EF for biogas slurry was higher than 50%. The estimated EF deviated with more than 33% from the guidebook data for dairy cows (solid), other cattle, horses, mules, asses, and laying hen.

The NH₃ EFs for pig and cattle slurry applied in the field were calculated for the seasonal temperatures shown in Table 3 and are shown in Table 7. The EF for each region were calculated as weighted averages, using the seasonal distribution shown in Table 2. Finally, the Tier 2 NH₃ EFs were calculated using an assumed distribution of slurry between cool, temperate and warm regions of 25, 50, and 25%, respectively. This led to NH₃ EFs of 19 and 25% of TAN for pig and cattle slurry, respectively.

There is a large number of data from studies measuring emission from solid manure applied in the field (Table 8). For dairy cows and non-dairy solid manure categories the SD is about 30% of the average EF, but for the other categories the SD is <53% of TAN irrespective that the number of studies are between 20 and 32. For the cattle categories the EF is near similar to the EF given in the Guidebook, but for the other categories it differs much more.

For grazing cattle the median NH₃ emission was 6% of total N excreted, average emission was 9% and SD of the average emission 7% (n = 12). For sows with piglets on grassland the NH₃ emission was 19% of N-total (n = 1). In the Guidebook the EF were 10% of TAN for dairy cows, 6% of TAN for non-dairy cattle and 25% of TAN for sows and piglets (European Environment Agency, 2016).

This study shows that there is a large variation in the results from studies measuring NH₃ emission from buildings, manure stores and following manure application to land. Consequently the EF given have SD that are around 50% of the estimated average for a category of livestock.

The SD of EF for housing may be due to the housing categories being only broadly defined and emission being affected by several factors which vary within each broad category. Those factors are: (1) the ratio of slatted floor to concrete floor area for pigs (Sommer et al., 2006), (2) floor opening area (Philippe et al., 2016), (3) distribution of excreta within the building, which is affected by positioning of feeders and drinkers, and behavior of pigs as related to age and temperature (Aarnink et al., 2006), (4) capacity of in-house storage, (5) age of animals, (6) climate, and (7) feeding practice.

In their review, Philippe et al. (2011) stated “The large range in published NH₃ emissions may be attributed to differences in house design (e.g., type of slatted floor, ventilation system, manure management) and management practices (e.g., age and weight at weaning, diet formulation).” This variation in emission caused by livestock housing systems and management was also pointed out by Groot Koerkamp et al. (1998) and the effect of climate by Saha et al. (2014). The large SD is therefore due to a range of factors, which are unaccounted for when calculating an average for the broadly defined categories used to give EF that can be used throughout Europe considering that in many regions little is known about either the proportions of livestock housed with different types of building and manure management practices.

Emission factors from pig buildings in % of TAN have a smaller SD than most other livestock buildings due to the more precise measurements of emissions from the pig buildings with forced ventilation and to some extend less variation in housing systems, feeding practice, and manure management.

The SD for the estimated EF (NH₃ in % of TAN) for stored manure was large for both stored slurry and stored solid manure. For stored slurry this may be due to different design of the store, i.e., the depth of stored slurry and climatic conditions and for solid manure it may be an effect of density and water content of the heap (Bernal et al., 2017). In most publications the annual emissions of NH₃ from liquid manure stores are given as g NH₃ m⁻² y⁻¹. This is a meaningful unit, because emission from stored slurry is related to the surface area, in addition to the effect of surface TAN concentration, emission is affected by surface pH, weather etc. The effect of surface to depth is illustrated by emission from a storage diameter 30 m and depth 1 m being similar to the emission from a storage diameter 30 m and depth 3 m, if the composition and climate are similar, but EF for the deep store should be 1/3 of the emission from the shallow one. For solid manure stores we present the EF in percent of TAN at initiation of the storage period, which is not an ideal solution, because transformations between organic N and TAN depend on a range of factors, including carbon to nitrogen (C:N), resilience of organic matter, oxygen content—porosity, size of heap, cover, turning etc. (Bernal et al., 2017). In future EFs should be related to the most important parameters which should include the transformation of organic N to TAN.

Emission from applied slurry is calculated using standard European conditions and is related to slurry DM and application rate, rain intensity and temperature (Hafner et al., 2018; Huijsmans et al., 2018). Soil characteristics will in practice have an effect on emissions (Chantigny et al., 2004; Seçer et al., 2018), but the statistical model does not include soil characteristics, because there was no clear effect of soil type in the analyses of emission data from the 12 countries that provided data for the model, which were from south to north of Europe (Hafner et al., 2018). Losses of NH₃ from solid livestock manure applied to soil are not as well-understood as the emissions from slurry application due to the greater variation in the composition of solid manure and there is widely differing practices with respect to litter use and storage, which contribute to the high SD of the average EF.

Ammonia emission from cattle on grassland is highly variable and most of the emission originates from urine patches (Laubach et al., 2012, 2013; Nichols et al., 2018). Increasing addition of N in fertilizer to the grassland will cause an increase in NH₃ emission due to increased grass N production, greater intake of N and increasing N excretion of which most ends up in the urine (Jarvis et al., 1989; Bussink, 1992; Voglmeier et al., 2018). Ammonia emissions increase with increasing soil moisture content (Bussink, 1992), and air temperature, wind speed, global radiation, and rainfall all influence emissions (Bell et al., 2017; Voglmeier et al., 2018). The same pattern is seen when studying NH₃ emission from sows on grassland, which was related to the amount of feed given to the sows, incident solar radiation, air temperature and rain (Sommer and Hutchings, 2001). These variations are not accounted for with the average EF given in the Guidebook. The few studies carried out do not support a distribution in EF as affected by soil and climate.

The EFs presented here differ in many instances from those in the current Guidebook (European Environment Agency, 2016). This only partly reflects the availability of new results from studies published in the 10 years since the previous systematic update. The resources available for the earlier update were limited, reflecting the lack of appreciation in Europe of the negative effects of NH₃ emissions in the past. As a result, the EFs were developed from the values used in the Tier 3 emission inventory methods of a small number of predominantly northern European countries. In the succeeding period, there has been an increase in awareness of the cost of atmospheric NH₃ pollution, especially with regards to human health (Giannadaki et al., 2018), and the cost-effectiveness of abatement measures (Backes et al., 2016). This has led to the establishment of more restrictive limits on national NH₃ emissions, within the scope of both CLRTP and NECD. This has prompted an awareness among regulators and policymakers of the need for more accurate and more detailed NH₃ emission inventories, and to an increase in resources for methodological development.

Changes in Tier 2 EFs varied widely among the different stages of manure management and livestock types (see Tables 5–8), hence the impact of the changes on national Inventories will vary according to the different proportions of livestock classes and manure management systems in each country. The impact of the changes in Tier 2 EFs on total annual emissions from each livestock type can be estimated from the default Tier 1 EFs, expressed as kg(NH₃) a⁻¹ AAP⁻¹ (Table 9). These Tier 1 defaults are produced by using the revised Tier 2 EFs from each stage of manure management to calculate NH₃ emissions based on standard annual N excretion values for each class of livestock. These revised Tier 1 default EFs can be compared with the previous default Tier 1 EFs to give an estimate of the change in annual NH₃ emissions from each class of livestock (Table 9).

Most of the changes to the Tier 2 EFs lead to only small (c. 5–10%) changes in emissions expressed as kg(NH₃) a⁻¹ AAP⁻¹ by the Tier 1 default EFs. Moreover, the changes include both increases, generally for livestock systems producing slurry, and decreases, generally for livestock systems producing FYM. There are some substantial reductions in Tier 1 EFs for poultry, but poultry are generally only a small item in national inventories.

Hence increases in total national emissions from countries with a large proportion of livestock systems producing slurry can be expected while total national emissions from countries in which livestock production is dominated by livestock systems producing FYM are expected to be small. For most countries NH₃ emissions from livestock production are dominated by emissions from cattle production. Hence total national emissions are expected to increase for countries where cattle are housed in slurry producing systems. It should be noted that the Guidebook EFs are guidelines only and many countries estimate national emissions using a Tier 3 approach in which EFs for each stage of manure management are based on measurements of NH₃ emissions from within that country. Hence changes to the Guidebook EFs will not impact on countries using such Tier 3 methodologies. Overall therefore the changes Tier 2 EFs reported here are not expected to make major changes to national emissions estimated using the Guidebook default methodology and none to national emissions estimated using national EFs.

The new NH₃ EFs from manure management are derived as weighted means of studies from across Europe, of which most are published in peer-reviewed papers and the application of a model of emission from field-applied slurry. The SD of the average of mean was high, due to the very broad definition of livestock and management systems, and to variations in climate and management practices. As a consequence, national NH₃ emission inventories developed using the Tier 2 methodology in the Guidebook will be associated with substantial uncertainty.

The datasets for this study can be found in the ALFAM website (http://www.alfam.dk).

SS and JW added data to the files used in the data analysis. Data analysis was carried out by SS, JW, and NH. SS wrote the first draft of the article which was then revised by the co-authors.