Soil compaction raises nitrous oxide emissions in managed agroecosystems. A review

Nitrous oxide (N2O) is the contributor to agricultural greenhouse gas emissions with the highest warming global potential. It is widely recognised that traffic and animal-induced compaction can lead to an increased potential for N2O emissions by decreasing soil oxygen supply. The extent to which the spatial and temporal variability of N2O emissions can be explained by soil compaction is unclear. This review aims to comprehensively discuss soil compaction effects on N2O emissions, and to understand how compaction may promote N2O emission hotspots and hot moments. An impact factor of N2O emissions due to compaction was calculated for each selected study; compaction effects were evaluated separately for croplands, grasslands and forest lands. Topsoil compaction was found to increase N2O emissions by 1.3 to 42 times across sites and land uses. Large impact factors were especially reported for cropland and grassland soils when topsoil compaction—induced by field traffic and/or grazing—is combined with nitrogen input from fertiliser or urine. Little is known about the contribution of subsoil compaction to N2O emissions. Water-filled pore space is the most common water metric used to explain N2O emission variability, but gas diffusivity is a parameter with higher prediction potential. Microbial community composition may be less critical than the soil environment for N2O emissions, and there is a need for comprehensive studies on association between environmental drivers and soil compaction. Lack of knowledge about the interacting factors causing N2O accumulation in compacted soils, at different degrees of compactness and across different spatial scales, limits the identification of high-risk areas and development of efficient mitigation strategies. Soil compaction mitigation strategies that aim to loosen the soil and recover pore system functionality, in combination with other agricultural management practices to regulate N2O emission, should be evaluated for their effectiveness across different agro-climatic conditions and scales.


Introduction
Soil compaction is a component of land degradation, which has mainly been associated with agricultural traffic, forest harvesting, animal trampling and industrial activities (Batey, 2009). This degradation process is defined by the European Soil Data Centre (ESDAC) as 'a form of physical degradation resulting in densification and distortion of the soil where biological activity, porosity and permeability are reduced, strength is increased and soil structure partly destroyed'.
Soil compaction is exacerbated under wet conditions and at low soil organic matter contents (Hamza and Anderson, 2005). Under such conditions, the intrinsic soil factors (e.g. texture, aggregate stability) interact with the external pressure forces (e.g. wheel load, inflation pressure, traction, number of passes, stocking rate, trampling frequency) to determine the extent (i.e. topsoil only, below the plough layer, or to greater depth) and degree of compactness.
One of the concerns about soil compaction is its potential contribution to emission of nitrous oxide (N 2 O) by promoting oxygen (O 2 ) limited conditions (Fig. 1). Nitrous oxide is a byproduct of nitrification and a free intermediate of denitrification at the interface (in space or time) between aerobic and anaerobic conditions (Butterbach-Bahl et al., 2013), and changes in soil structural quality caused by compaction may influence both production, consumption and transport of N 2 O (Ball et al., 1999a). Depending on the climate scenario and management practices adopted, emissions of N 2 O from agricultural soils are expected to increase with increasing compaction (Flynn et al., 2005), which represents a potentially high off-site cost for environmental damage , as N 2 O is assigned a global warming potential of 265 (IPCC, 2014).
The fact that soil compaction can potentially increase N 2 O emissions from agricultural soils has been recognised in the literature, and for example Hu et al. (2021) published a review of soil compaction effects on productivity and environment with New Zealand as a case study, which concluded that the extent to which the variability in N 2 O emissions can be explained by soil compaction is unclear. Alleviating C (Ambus and Christensen, 1994) or N limitation (Ball et al., 2000) has been found to reduce the spatial variability of N 2 O emissions, and there were positive interactions with, respectively, wet depressions and soil compaction, presumably because production of N 2 O was sustained in a larger fraction of the soil volume. Furthermore, the response of N 2 O emission to soil physical changes caused by soil compaction are not yet well understood, and knowledge gaps related to soil physical parameters, organic matter decomposition, microbiology and compaction drivers were identified. The present review aims to comprehensively discuss the mechanisms behind soil compaction effects on N 2 O emissions, with emphasis on understanding the promotion of N 2 O emission hotspots and hot moments by soil compaction.
We first introduce the factors determining the effect of soil compaction on N 2 O emissions. Then follows a section reviewing, based on calculated impact factors, observations of soil compaction effects separately for cropland, grassland and forest land. This is followed by sections discussing knowledge gaps and strategies for mitigating compaction effects on N 2 O emissions, respectively.

Soil compaction as a driver of N O emissions
At the level of the soil profile, soil N 2 O emission responses to soil compaction are regulated by factors such as microbial abundance and activity, mineral N, labile C, soil pH, soil temperature, water content or water-filled pore space (WFPS), soil texture, soil structure (aggregate sizes, pore space characteristics, gas diffusion rate), surface sealing and soil drainage (Laudone et al., 2011;Ball, 2013;Garcia-Marco et al., 2014). The extent of soil compaction (Garcia-Marco et al., 2014), and the spatiotemporal distribution of the abovementioned factors in the soil profile (Groffman et al., 2009), are additional factors in the regulation of N 2 O emissions.
The release of N 2 O to the atmosphere is determined by the balance between production, consumption and transport (Soane and Vanouwerkerk, 1995), and therefore the pore system of a given soil, as determined by texture and structure, is critical in regulating the gas exchange between soil and atmosphere (Laudone et al., 2011). The soil pore system is studied in many research papers, and found to changes with depth, being best described as having a sponge-like system in the topsoil, whereas a tube-like system dominates the subsoil . When compaction occurs, pores are not completely destroyed, but instead, a closing of branching pores and diameter reduction of vertical (tube-like) pores can be seen (Schäffer et al., 2008;Schjønning et al., 2013). In general, compaction, therefore, promotes the development of a direction-dependent behaviour of the pore system Horn, 2006, 2009) that negatively affects the size, tortuosity and connectivity of pores, and directly affects fluid transport in soil (e.g. Kim et al., 2010;Berisso et al., 2012;Berisso et al., 2013;Kuncoro et al., 2014;Zhai and Horn, 2019). This in turn may promote anaerobic conditions and change the direction of soil processes (e.g. Ruser et al., 2006;Chamen et al., 2015;Müller et al., 2019;Rohe et al., 2021). As summarised by Ball (2013), limited pore continuity and gas transport capacity within (leading to anaerobic centres) and between aggregates (blocked or reduced inter-aggregate porosity) influence N 2 O production, consumption and transport to the soil surface.
As a product of microbial nitrogen transformations in soil, N 2 O emissions depend on the co-occurrence of suitable Fig. 1 Example of nitrous oxide flux measurement in a long-term tillage experiment (no-till and ploughed soil with and without a winter cover crop) (left), and a compacted soil structure with high potential to increase nitrous oxide emissions (right). Photographs by the authors. electron donors and acceptors. Denitrification including nitrifier denitrification has been identified as the main source of N 2 O emissions from soil (Skiba et al., 1993;Saggar et al., 2009;Kool et al., 2011;Harris et al., 2021), and this process requires degradable organic matter (energy source and O 2 sink) and nitrogen oxyanions (NO 3 − or NO 2 − ). For a given soil, as defined by texture, pH etc., the balance between N 2 O and N 2 production further depends on the degree of anaerobiosis, since even traces of O 2 inhibit the expression of N 2 O reductase, the enzyme responsible for N 2 O reduction to N 2 (Spiro, 2012). Therefore, soil compaction effects on N 2 O emissions will depend on management factors such as residue recycling, fertilisation with manure or synthetic N and traffic, factors which all influence soil O 2 status.
Soil volumes and episodes supporting N 2 O emissions are referred to as hotspots and hot moments, respectively (Wagner-Riddle et al., 2020). Figure 2 presents a conceptual framework to illustrate how differing soil structural states and interacting factors may trigger net N 2 O production and transport. The occurrence of N 2 O emitting hotspots varies with the scale of measurement (Luo et al., 2017;Wagner-Riddle et al., 2020), i.e. ranging from a few millimetres (Laudone et al., 2011;Rohe et al., 2021) to metres at field or landscape level (Ambus and Christensen, 1994;Jacinthe and Lal, 2006), or even to regional level (Groffman et al., 2009). The variability of N 2 O emission hotspot may depend on spatial scale, with gas diffusion as an important environmental factor (van den Heuvel et al., 2009).
External factors affecting the occurrence of N 2 O hotspots and hot moments include weather conditions and farming operations . Hotspots may become activated under wet conditions (Grant et al., 2006;Ruser et al., 2006). In the presence of substrates, precipitation (or irrigation) can induce N 2 O emissions by increasing soil water content and thereby reduce the supply of O 2 to sites of microbial activity (e.g. Ruser et al., 2006;Beare et al., 2009). Air temperature also controls N 2 O emissions from soils by affecting soil temperature and, consequently, rates of enzymatic processes (Schindlbacher et al., 2004;Flynn et al., 2005). Soil structural conditions have been found to affect soil thermal properties. Schjønning (2021), for example, found that thermal conductivity increases with bulk density, and Zhen et al. (2019) showed that thermal conductivity of undisturbed samples is larger than on remolded samples when measured at the same degree of saturation and dry bulk density.
Depending on the farming system, management strategies such as the quantity of N applied in animal excreta and fertilisers (e.g. Hu et al., 2020), the intensity of animal Fig. 2 Conceptual framework illustrating how differing soil structural statuses, and interacting factors, may trigger hotspots and hot moments for N 2 O production and transport. Soil quality score refers to the structural soil quality based on the Visual Evaluation of Soil Structure test (Guimarães et al., 2011). trampling (e.g. de Klein and Eckard, 2008) or traffic (e.g. Pradel et al., 2013) in combination with local weather and soil conditions (soil type, structural status and landscape position), all contribute to determine the occurrence of N 2 O hotspots and hot moments .
Spatiotemporal distribution of N 2 O emission hotspots is also influenced by management practices (Ball, 2013). Under good structural conditions, the soil is mostly well-aerated, and denitrification and N 2 O production is restricted to patches within the soil that are dominated by fresh organic matter undergoing decomposition, such as plant debris (Parkin, 1987;Li et al., 2016) or manure (Markfoged et al., 2011). Less intense decomposer activity may also, with adequate N supply, support N 2 O emission when soil volumes are saturated following rainfall or irrigation (hot moments) (Kostyanovsky et al., 2019).
Within compacted soil layers, the dominant tube-like pores in the system are critical upward and downward conductive paths for O 2 and gases produced (Laudone et al., 2011). Soil structure with preferential pathways allows applied N fertilisers to be transport with infiltrating water, which may then become a source of N 2 O at depth below the plough layer (Ball, 2013). Through the creation of bio-pores, earthworm activity has been reported to increase N 2 O production, yet its direct effect on emissions is negligible when compared with the overall soil fluxes (e.g. Bertora et al., 2007). Rather, in the massive structure of compacted soil, particularly the subsoil, the transport functionality of burrows could be a factor of importance for the release of N 2 O from the sites of production to the atmosphere. A similar contribution is expected from deep cracks. Blocked or disconnected structural pores may hold N 2 O which could be released when these soil structural pores are disrupted (Ball, 2013). Importantly, the distribution and connectivity between hotspots zones and the upward transport pathways controls the rate of N 2 O emissions to atmosphere, as a longer path delays and allows for reduction of N 2 O during transport (Laudone et al., 2011). In summary, N 2 O emissions from soil depend on biophysical interactions, structural stratification in the soil and management practices, but the net effect of the many potential interactions on N 2 O emissions is complex and requires further investigation. The work of Rohe et al. (2021) is an example of potential protocols for the assessment of the relationships between soil structural changes, climatic conditions and the denitrification process through the use of advanced imaging techniques in combination with transport parameter measurements.

Impact of soil compaction on N 2 O emissions
We used Web of Science to review the literature published before 19 April 2021 using the search term 'soil-compaction AND (nitrous-oxide-emission* OR N 2 O)'. To determine the effects of soil compaction on N 2 O emissions, papers including one or several compaction treatments (traffic-, animal-or repacking-induced compaction) were selected and then organised in tables by land use (cropland, grassland and forest land) and measurement method (in situ or ex-situ). The impact factor for N 2 O emissions shown in the tables is calculated as the ratio of the emissions from the compaction treatment to the non-compacted/control, using either mean or cumulative values depending on what was reported in the papers. In cases where the papers provided data from different sites and reported combined effects, a mean value across factors was calculated instead. Additionally, the information provided in the papers with respect to the type of compactness indicators reported was used to assess the level of detail evaluated and the type of association with N 2 O emissions provided (theoretical or mathematical associations).

Impact of topsoil compaction on N 2 O emissions
For this review, topsoil is defined as the soil depth of the plough layer, which is around 0.25 m. Traffic-induced compaction occurs from the top few centimetres up to 0.9 m depth (e.g. Håkansson and Reeder, 1994;Berisso et al., 2012) and can become a long-lasting problem (Berisso et al., 2012;Etana et al., 2013), whereas trampling-induced compaction is reported to occur only in the topsoil, with the greatest impact caused at depths of < 0.10 m (Hamza and Anderson, 2005).
In Tables 1, 2 and 3, it can be seen that topsoil compaction generally increases N 2 O emissions, with the highest reported rate being 42 times higher than in uncompacted soil.

Croplands
N 2 O emissions from croplands are characterised by a large temporal variation, with seasonal peaks as a response to fertiliser application, precipitation/irrigation and/or freezethaw events (Bessou et al., 2010;Gregorich et al., 2014;Liu et al., 2017). As mentioned above, soil compaction may exacerbate the production and emission of N 2 O associated with fertilisation and other management practices.
Fertilisation Table 1 shows for studies conducted in croplands that topsoil compaction is reported to increase N 2 O emissions between 1.4 (Hansen et al., 1993) and 9.9 times (Ruser et al., 1998), and up to 42 times when compaction + NO 3 − based Nfertiliser + glucose was tested (Bao et al., 2012). Across a number of published studies, the differences in N 2 O emissions between compacted and non-compacted areas were especially large after N fertilisation (Ball et al., 1999b;Sitaula et al., 2000). The potential residual effect of soil compaction and N fertiliser on cumulative N 2 O emissions was found to be significant 1 year after compaction and where fertiliser treatments were applied to a clay loam soil cropped to maize (Gregorich et al.,     2014). N 2 O emissions were found to correlate better with soil compaction than with N fertiliser rates on a silt loam soil under sugar beet with preceding winter wheat and winter barley straw incorporation (Bessou et al., 2010), which indicates that O 2 limitation was the most important driver. However, Ball et al. (2000) found that N 2 O fluxes were better correlated with soil physical properties after N fertilisation, and they stressed the importance of interacting effects. The type of fertiliser is another key factor controlling N 2 O emissions. For example, in sandy loam soil under rotation (green fodder/barley/peas/vetch and rye-grass), Hansen et al. (1993) reported that in compacted soils the soil air concentration of N 2 O was seven times higher in NPK-fertilised plots than in plots fertilised with cattle slurry, whereas no difference was found in the uncompacted soil. This was probably because NPK fertilisers contain NO 3 − , which can directly serve as an electron acceptor for denitrification, whereas ammoniacal N in manure must await nitrification, which may be delayed by reduced O 2 availability in compacted soil (Jensen et al., 1996).
Other examples of combined compaction effects include a study by Ball et al. (1999a) for loam and sandy loam soils under winter wheat in the UK. Their study found that compaction significantly increased N 2 O emissions after fertiliser application or residue incorporation, with marked emissions in the periods of the year when the soil was wet (volumetric soil water content > ca. 38%). Similarly, Sitaula et al. (2000) reported that prevalent high volumetric water contents of > 45% on measurement days favoured greater N 2 O production in a traffic-compacted sandy loam soil compared to uncompacted soil.
For maize on a silt loam soil, a combined compaction (+NO 3 +glucose) treatment was found to explain nearly 70% of the variation in N 2 O emissions compared to 24% for a control treatment, and 60% WFPS was found to be a threshold for increasing N 2 O emissions in all the treatments both with and without compaction (Bao et al., 2012).
In incubation experiments, an increase in N 2 O emissions by compaction, in combination with fertiliser and/or water content, was reported to range from 1.3 to 20 times that in the uncompacted soils (e.g. Ruser et al., 2006;Ball and Crawford, 2009;Beare et al., 2009).
Drying/rewetting Irrigation resulting in drying/rewetting cycles is another factor of importance for N 2 O production in compacted agricultural soils (Ruser et al., 2006;Beare et al., 2009). In an incubation experiment on a clay loam soil, the total N 2 O production from compacted soil was 70, 3 and 20 times higher than that from uncompacted soil during, respectively, the preincubation phase at field capacity, the drying phase and the rewetting phase (Beare et al., 2009). In the same experiment, the production of N 2 O was increased by drying and rewetting cycles compared to the continuously wet treatment, though in both cases compaction resulted in a larger increase in N 2 O production relative to the uncompacted soil.
When comparing compacted interrow soil to uncompacted soil, Ruser et al. (2006) found that drying/rewetting cycles in the soils could induce high N 2 O emission peaks that exceed by about a factor of two the maximum flux rates measured after nitrate addition at constant soil water contents. Moreover, during the drying/rewetting cycles, larger emissions are measured after rewetting (Beare et al., 2009), which also proportionally increase with increasing water content (Ruser et al., 2006).

Row crop systems
In row crop systems, the tractor-compacted interrow areas are characterised by a poor structure with higher bulk density, a large reduction in the air-filled pore space and higher soil moisture content than the cropped rows (Ruser et al., 1998;Ball and Crawford, 2009). Anaerobic conditions in the compacted interrows induced a higher N 2 O production by an impact factor ranging between 2.0 and 9.9 times compared to the rows growing potato (Ruser et al., 1998;Flessa et al., 2002;Ruser et al., 2006) and corresponding impact factors of 1.3-4.6 in carrot fields (Ball and Crawford, 2009).
In an incubation experiment, the maximum N 2 O flux rates occurred from compacted interrow soil sampled from a potato field, but the cumulative N 2 O emission at 90% WFPS was higher for the ridge soil compared to the compacted interrow (Ruser et al., 2006). In row crop systems, N 2 O emissions are also found to increase after surface application of residues in combination with compaction (Flessa et al., 2002), and after heavy precipitation (Ruser et al., 1998;Flessa et al., 2002). However, N 2 O release can be relatively low in the interrows if measured immediately after precipitation (under waterlogged conditions) (Ruser et al., 1998). Importantly, N 2 O losses from the interrows might vary with the degree of compactness, the amount of WFPS and the rate of N inputs (Flessa et al., 2002), similar to other cropping systems.
In apple and cherry orchards, interrows were reported to increase daily N 2 O emissions approximately two-fold compared to within the tree lines during summer in Australia (Swarts et al., 2016). However, in both the interrows and tree lines, the emission rates during summer were low, which was attributed to the judicious management of irrigation and N fertiliser application through tree line drippers, as the volumetric water content rarely exceeded field capacity.
In relation to controlled-traffic farming, Tullberg et al. (2018) presented work wherein the N 2 O emission reduction ratio was calculated for 15 sites based on the traffic impact factors of permanent traffic lanes, random-trafficked soil and the non-wheeled area. This work showed that trafficked lanes increased N 2 O release 1.1 to 5.0 times compared to the untrafficked areas across sites.

Grasslands
Topsoil compaction in grasslands has been found to increase N 2 O emissions between 1.2 and 7.4 times when measured directly under field conditions, and between 1.0 and 8.3 times in incubation experiments (Table 2). In tractor compacted grasslands, Ball et al. (1997) found that trafficked areas with, respectively, 1.3-, 16-and 2-fold lower air-filled porosity, air permeability and relative gas diffusivity, produced about two times more N 2 O than the zero-traffic areas. Some studies, however, reported no significant compaction effect on N 2 O emissions from grasslands and pasture soil (Simek et al., 2006;van der Weerden and Styles, 2012;Harrison-Kirk et al., 2015); all studies concluded that production of N 2 was probably favoured.
Fertilisation Yamulki and Jarvis (2002), in a clay loam soil under a mixture of perennial grasses, found that trafficcompaction significantly increased the total cumulative flux of N 2 O regardless of fertiliser application, though the variation in N 2 O fluxes was large within and between the treatments. In contrast, work conducted by Schmeer et al. (2014) on traffic-compaction of a sandy loam soil under permanent perennial species only caused an increase in N 2 O emissions on N-fertilised plots (mean of three years). In a combined compaction-fertiliser experiment, Bhandral et al. (2007) showed that traffic-compaction of a sandy loam soil increased N 2 O emissions from grassland irrespective of the N source, yet the effect of nitrate application was more pronounced in the compacted soil compared to other N sources. These studies indicate that on light-textured soil with perennial vegetation, compaction alone will not greatly influence N 2 O emissions, possibly because plants ensure a low mineral N availability except for a period after fertilisation. With finertextured soil, compaction can result in more wide-spread O 2 limitation and hence potential for denitrification, but the potential for N 2 O reduction to N 2 will also increase.
Grazing The effect of trampling-induced soil compaction on N 2 O emissions has been widely investigated. In a cattle overwintering area in the Czech Republic, Simek et al. (2006) found higher N 2 O emissions in trampled areas compared to those of areas with less or no disturbance by trampling, but the difference was not statistically significant, which was attributed to a high spatial variability. In another study from Scotland, simulation of trampling in a wet dairy pasture soil showed a three-fold increase in N 2 O emissions (Ball et al., 2012).
In New Zealand, van der Weerden and Styles (2012) and van der Weerden et al. (2017) found that N 2 O fluxes from pasture on silt loam soil were greatest from compacted treatments after urine application and remained elevated for two to four weeks; thereafter fluxes declined and remained stable for about four months. In silt clay loam and sandy loam soils under perennial ryegrass, compaction by both traffic and trampling induced larger cumulative N 2 O emissions compared to uncompacted control soil, although the difference was only statistically significant for the traffic compaction treatment (Hargreaves et al., 2021). This study, conducted in the UK, confirmed that the largest mean daily N 2 O fluxes are generally observed after external input of N. In contrast to these various reports, a study by Piva et al. (2019) on a clay oxisol in Brazil found that the N 2 O emission peak after N application was more intense in ungrazed compared to grazed pasture, especially at low WFPS. Hence, grazing affects N 2 O emissions through interacting effects of N input and compaction that are modified by site-specific conditions. This has been confirmed in manipulative incubation experiments with urine, dung and simulated compaction (van Groenigen et al., 2005;Cardoso et al., 2017).
Mechanisms In an ex-situ experiment, conducted with a silty loam soil under ryegrass-white clover, soil cores were packed by applying pressures of 0, 220 or 400 kPa and treated with synthetic urine, and then subjected to successive saturationdrainage cycles on tension tables (Harrison-Kirk et al., 2015). At 0 and 220 kPa compaction levels, N 2 O fluxes dropped as soil cores were drained to 6 kPa matric potential, whereas N 2 O fluxes in the most compacted treatment persisted longer and persisted until the soil was drained to 8 kPa tension. This indicates a relationship between matric potential and soil structure with respect to N 2 O emissions, but Balaine et al. (2013) found that relative gas diffusivity was a better predictor compared to matric potential. In their study of a silt loam soil packed to five dry bulk densities and one of seven matric potentials, Balaine et al. (2013) found a consistent maximum of N 2 O-N fluxes when the relative gas diffusivity ranged between 0.0060 and 0.0067, regardless of bulk density. It should be stressed that this conclusion applied to bulk soil and did not consider variations in organic matter or nitrate availability.
Soil compaction will change soil structure, as shown in a study where repacked soil cores were prepared with aggregates of different sizes (Uchida et al., 2008). The highest N 2 O fluxes occurred at moderate to severe compaction and in the smallest aggregates (0-1.0 mm), which also had the lowest porosity after compaction. After a drying/rewetting cycle, N 2 O fluxes increased in all treatments but with the highest fluxes in the moderately to severely compacted soils. The smaller the aggregate size, the longer was the period in which N 2 O fluxes continued to increase (Uchida et al., 2008).

Forest land
Under forest land, compaction by mechanical disturbance has been reported to be an impact factor causing N 2 O emissions of 1.7 to 40 times (Table 3).
In oak forests, Goutal et al. (2013) found that trafficked plots had a higher N 2 O production in comparison to the control treatment, but only below 0.3 m depth where the soil air-filled porosity was significantly reduced. The residual effect of compaction on N 2 O emissions was evident after 2 years of applying compaction, although with seasonal variation. Compaction with heavy machinery also significantly increased N 2 O emissions in two forests in Switzerland, and the difference in N 2 O emission between the compacted and natural area remained largely consistent up to around 5 years post-disturbance (Hartmann et al., 2014). The N 2 O emission in forests on clay Oxisol showed high fluxes during the wet season and low fluxes during the dry season for both compacted and uncompacted areas, whereas in the logging decks an inverted pattern was observed (Keller et al., 2005). At another site under beech, cumulative annual N 2 O emissions were 3.3 times higher in the wheel track than in the undisturbed stand (Warlo et al., 2019). In that study, N 2 O emissions across trafficked and non/trafficked areas were larger under alder than under beech, but no compaction effect was observed in the site under alder stands. Presumably the role of traffic and climatic conditions are the same in managed forests as in the other land use categories.

Impact of subsoil compaction on N 2 O emissions
Although most studies focus on N 2 O emissions from the soil surface, the production of N 2 O may occur in the entire soil profile depending on soil conditions.
In compacted subsoil, gas and water transport mainly occurs through vertical biopores that remain functional, though with reduced volume, after compaction Schjønning et al., 2019). Anaerobic conditions in the soil matrix between vertical macropores in compacted subsoil may turn hardened layers into emission sources. Additionally, subsoil compaction may reduce water flow in saturated, or nearsaturated state, thus impeding drainage and resulting in a wetter topsoil in early spring, as shown by, for example, Pulido-Moncada et al. (2021). This is expected to increase the risk of N 2 O emissions associated with fertilisation, manure application and crop residue turnover. Despite this, there is a paucity of knowledge about the contribution of subsoil compaction to the emission of the greenhouse gas N 2 O .
Recently, Petersen and Abrahamsen (2021) simulated the expected long-term effects of traffic with heavy machinery (resulting in subsoil compaction) on nitrogen balances and the environment by using the model Daisy with input data from a 10-year field trial in Denmark. The study showed that the simulated extra nitrogen loss (as N 2 or N 2 O) associated with subsoil compaction can increase losses by up to 50%. This simulation result highlights the need to determine to what extent subsoil compaction contributes to losses of gaseous nitrogen (N 2 O in particular).

Responses of N 2 O emission to changes in soil physical properties: knowledge gaps
Soil compaction promotes N 2 O emission due to changes in soil physical and biological properties. The following section reviews literature on how to best describe changes in soil structure in relation to risk of creating N 2 O emission hotspots and hot moments and knowledge gaps are identified.

Soil physical parameters
Studies summarised in Tables 1, 2 and 3 across the three landuse categories, i.e. cropland, grassland and forest land, recognised that changes in soil structure strongly affect N 2 O emissions, but often limit their assessments of traffic/ trampling-compaction impact on N 2 O emissions to either a brief description of the compaction status (Ball et al., 1999a;van Groenigen et al., 2005;Schmeer et al., 2014;Cardoso et al., 2017;Tullberg et al., 2018;De Rosa et al., 2020), or to a theoretical or indirect association between N 2 O emissions and bulk density, water content and/or WFPS (Yamulki and Jarvis, 2002;Keller et al., 2005;Ruser et al., 2006;Uchida et al., 2008;van der Weerden and Styles, 2012;Gregorich et al., 2014;van der Weerden et al., 2017;Piva et al., 2019;Hargreaves et al., 2021); or with bulk density, total porosity and water holding capacity (Liu et al., 2017); with air-filled porosity (Hansen et al., 1993;Goutal et al., 2013); pore size distribution, bulk density and WFPS (Ruser et al., 1998;Teepe et al., 2004); or with soil strength measurements such a penetration resistance and vane shear (e.g. Ball et al., 1997;Bhandral et al., 2007;Ball and Crawford, 2009;Vermeulen and Mosquera, 2009). One notable exception is the work by Ball et al. (1997), who presented a complete description of soil physical status including the majority of the parameters mentioned above. This study established an association of increases in N 2 O emissions from trafficked grassland areas with poor structure and limited fluid transport-yet no direct relationships were established.
The response of N 2 O fluxes to soil compaction has often been quantified with a particular focus on the association with soil water content, most often represented by WFPS. An example of this is the study by Swarts et al. (2016) in tree cropping systems, where several soil physical parameters were measured, but direct associations with N 2 O emissions were only investigated with water content (gravimetric and volumetric) and WFPS. These associations were statistically weak (r < 0.40) and had no consistent pattern, neither in the tree line nor interrow, between seasons or sites. In contrast, in a study of grazed soils under winter forage crops, van der Weerden et al. (2017) showed a strong relationship between N 2 O emissions and WFPS (R 2 = 0.83, p = 0.005) across urine and compaction treatments. In a study from China, WFPS was directly proportional to N 2 O flux rates (R 2 = 0.57-0.70) across compaction and N source combined treatments, but only when WFPS reached 56-63% (Bao et al., 2012). A significant linear relationship between WFPS and log-transformed N 2 O emissions was also found by Flessa et al. (2002) for compacted and non-compacted inter-rows of a potato field with on average 61 and 49% WFPS, and by Simek et al. (2006) in a cattle overwintering area with WFPS at 65-82%. However, Flessa et al. (2002) also observed high N 2 O emissions from the ridge position at only 30% WFPS. Beare et al. (2009) found a significant exponential relationship (r 2 = 0.67, p < 0.001) between WFPS and N 2 O production during pre-incubation and drying phases of an incubation experiment, but no clear difference in this relationship between compacted and uncompacted soil was observed when WFPS was < 60%. To summarise, WFPS is not a general predictor of compaction effects on N 2 O emissions, indicating that the effect of WFPS interacts with other soil properties.
In early works by Stepniewski (1980), O 2 diffusion was found to be a potential parameter to determine critical ranges of soil compaction and moisture tension for plant growth. Later studies have shown relationship between N 2 O emissions and relative gas diffusivity. Balaine et al. (2013) showed that relative gas diffusivity was a better predictor of N 2 O emissions than WFPS across several combinations of soil bulk density and water potential. In accordance with this, Petersen et al. (2008), comparing N 2 O emissions from intact soil cores under no-till or moldboard ploughing at seven matric potentials, found that relative gas diffusivity was a stronger predictor than either WFPS or volumetric water content. Harrison-Kirk et al. (2015) also compared WFPS, volumetric water content and relative gas diffusivity in an experiment with compaction of urine-treated soil cores; they always found statistically significant relationships with N 2 O flux (R 2 = 0.46-0.62, p < 0.001), but again pointed to relative gas diffusivity as the best predictor across variable soil conditions. Furthermore, Harrison-Kirk et al. (2015) reported that soil compaction led to reduced macro-porosity and more complete denitrification to N 2 in the most compacted soil.
The observations from controlled laboratory incubations are confirmed by field observations. In a study of beech and alder forest, Warlo et al. (2019) evaluated the relationship between N 2 O flux and soil structure by continuously monitoring several soil physical parameters. Although the authors gave more attention to the N 2 O fluxes for the tree species than for the effect of traffic, the best tree-species specific models (R 2 = 0.26-0.64) showed that gas diffusivity was the main variable controlling N 2 O flux. This is in agreement with a compaction study conducted by Sitaula et al. (2000) where an increase in N 2 O emissions was related to a decrease in gas diffusivity, and with results from Mutegi et al. (2010) where gas diffusivity was found to be a better explanatory factor for N 2 O emissions compared to WFPS in tillage experiments. Furthermore, Rousset et al. (2020) showed that, across four soils of different texture, each packed to three bulk densities, gas diffusivity predicted the onset of N 2 O emissions under conditions of C and NO 3 availability supporting denitrification. Hence, gas diffusivity appears to be is an important characteristic, together with water saturation, in determining how soil structure contributes to the production and transport of denitrification products (Rohe et al., 2021). At this time, the morphology of the soil pore system and its contribution to N 2 O fluxes are poorly understood and documented, and studies characterising soil-gas phase relationships may help fill this knowledge gap. An example is the study of Chamindu Deepagoda et al. (2013), which presented a comprehensive analysis of pore tortuosity-discontinuity in variably saturated soils and showed strong relationships between pore tortuosity (air permeability-based index) and clay content, particle size distribution and water retention parameters.

Organic matter decomposition
The relationship between relative gas diffusivity and N 2 O emissions may be confounded by organic matter degradation (Petersen et al., 2013;Balaine et al., 2016). Fresh organic matter associated with plant residues or animal manure represent a local O 2 demand that may sustain denitrification activity and N 2 O emissions across a wide range of soil conditions, as demonstrated in laboratory studies (e.g. Parkin, 1987;Li et al., 2016), but also under field conditions (Flessa et al., 2002). In these situations, organic matter decomposition rather than bulk soil conditions is the main driver of N 2 O emissions (Wagner-Riddle et al., 2020). Baral et al. (2016) concluded, based on a factorial incubation experiment with three soil moisture levels and three manure types, that relative gas diffusivity controls the proportions of aerobic and anaerobic degradation through the O 2 supply to putative N 2 O emission hotspots, although also soil NO 3 availability affects the extent of denitrification and N 2 O emissions (Taghizadeh-Toosi et al., 2021). An effect of soil particle size distribution and N 2 O emissions from crop residues was also reported by Kravchenko et al. (2017). If the role of relative gas diffusivity for oxic vs anoxic decomposition is confirmed, this may link bulk soil conditions with C and N turnover in organic hotspots.

Microbiology
Soil compaction will reduce the volume of macropores and increase that of smaller pores. While this is mostly discussed in the context of water availability (Lipiec et al., 2012) and hydraulic properties (Tarawally et al., 2004), the change in pore size distribution could also alter conditions for microbial survival and activity. Postma and van Veen (1990) investigated microbial numbers at different bulk densities in two soil types and found little effect of increasing bulk density on microbial numbers. They estimated that less than 1% of the habitable pore space was occupied, which may explain the lack of response. This was in contrast to the effect of increasing soil moisture, which resulted in declining cell numbers, an effect that was explained by increasing oxygen limitation (Postma and van Veen, 1990). Frey et al. (2009) observed changes in bacterial community structure in severely compacted forest soils (32% higher bulk density) and related this to reduced air and water conductivities. In a study by Liu et al. (2017), compaction negatively affected soil physical properties, but the latter had little effect on N 2 O-related microbial community size as it was correlated only to a few microbial gene abundances. A study by Bao et al. (2012) showed that compaction combined with NO 3 + glucose enhanced the activity and abundance of denitrifiers in alignment with an increase in N 2 O emission, but did not significantly affect the overall community composition. In their study, however, the isolated effect of compaction on the microbial community was not investigated. A study by Hartmann et al. (2014), however, provided a comprehensive evaluation of compaction-associated alterations of N 2 O-related microbial community characteristics; this was an integrated approach (soil physical, microbial and functional characteristics) to measuring resistance and resilience of the soil system to compaction, e.g. by determining compaction thresholds of detrimental impact on ecosystem functioning.
Although the above-mentioned studies assessed the effect of compaction on the N 2 O-related microbial community, there is still a need for more comprehensive studies on how compaction-induced changes in soil physical properties (e.g. pore characteristics, thermal conductivity) influence the microbiome under different scenarios, including different degrees of compaction.

Compaction drivers
Based on the literature review conducted here, only a few studies quantifying traffic-compaction effects on N 2 O emissions characterised the traffic treatment applied (e.g. Hansen et al., 1993;Sitaula et al., 2000;Teepe et al., 2004;Ball and Crawford, 2009;Goutal et al., 2013;Gregorich et al., 2014;Hartmann et al., 2014). However, the degree of compactness depends on the soil-machinery interaction, in turn depending on the characteristics of the machinery used in the field, and key to understanding the traffic-induced soil stress (Lamandé and Schjønning, 2011;Keller et al., 2013;ten Damme et al., 2019). This calls for a better understanding of the specific aspects of machinery-soil interactions leading to N 2 O emissions in different soils and climates, including those associated with hotspots and hot moments, in order to identify risk conditions and critical thresholds.

Mitigation and avoidance of soil compaction: impact on N 2 O emissions
Wagner-Riddle et al. (2020) listed possible agroecosystem N 2 O mitigation strategies such as fertiliser management-source, rate, time and place; management of organic input (crop residues and manure); and soil-N source (O 2 supply interactions), with a particular focus on hotspots and hot moments. Here, we summarise possible mitigation approaches specifically related to compacted soils as a high-risk environment for N 2 O emissions.
In general, ploughing and subsoiling (biological or mechanical) are options to mitigate soil compaction, whereas compaction-intelligent traffic and controlled-traffic farming are soil compaction avoidance strategies . The adoption of controlled-traffic farming may give an overall improvement of soil conditions that can reduce greenhouse gas emissions (Antille et al., 2015). Mouazen and Palmqvist (2015) developed a framework for the evaluation of environmental benefits of controlled traffic farming based on a European Commission Soil Framework Directive and scientific literature review, where soil compaction and greenhouse gas emissions were identified as the main and secondary environmental parameters, respectively.
Reduction in traffic intensity through controlled-traffic farming translates into 10-20% trafficked area compared to > 80% for conventional management (Gasso et al., 2013;Tullberg et al., 2018), which in itself minimises the area at risk of increased WFPS due to compaction (Antille et al., 2015), thus potentially leading to lower risk of N 2 O emissions. Indeed, the use of seasonal or permanent controlled-traffic farming has been found to reduce N 2 O emissions by up to 50% when compared to random traffic (Vermeulen and Mosquera, 2009). Estimations based on Australian soils indicate that in non-controlled traffic systems with 50%, 75% or 100% randomly wheeled area, when replaced by controlledtraffic farming with 15% designated traffic lane area, the N 2 O emissions would be 69%, 58% or 50%, respectively, of their previous values (Tullberg et al., 2018). In the Netherlands, Vermeulen and Mosquera (2009) also found that the application of seasonal controlled-traffic farming decreased N 2 O emissions on average by 20-50% in four vegetable crops. Bluett et al. (2019) suggested that traffic-induced soil compaction could probably be avoided through the use of lightweight machinery, but that with the current available technology the 'solution' would be the adoption of controlled-traffic farming. However, while research supports that the number of passes is significant for the impact of wheel load (e.g. Chamen et al., 2015;Pulido-Moncada et al., 2019), reduction of traffic intensity is not the only factor of importance when attempting to reduce or avoid soil compaction. The driving factors for trafficked-soil compaction determine the magnitude of the stress imposed on soil, which is susceptible to deformation-typically wet soil . Hence, key elements in the reduction of soil compaction risks are tyre type, tyre inflation pressure and wheel load (Lamandé and Schjønning, 2011;ten Damme et al., 2019), the combination of wheel load with number of passes (Schjønning et al., 2016), and traction and repeated wheeling (ten Damme et al., 2021). This suggests that there are other specific field traffic practices, besides controlled-traffic farming, with a potential to reduce compaction and consequently the potential for N 2 O emissions. There is, however, a need for comprehensive studies on the causal mechanisms linking compaction to N 2 O emission in order to establish the least emissions-prone agricultural systems.
More than just minimising compaction, it is also necessary to consider how compaction may interact with agricultural management practices. The choice of fertiliser, for example, is a key factor in mitigating N 2 O emissions in compacted soils, as 10 times less N 2 O was emitted when reduced N sources such as urine, ammonium and urea were used compared to nitrate in a grassland soil affected by compaction (Bhandral et al., 2007). Presumably compaction increased the volume of soil with oxygen limitation supporting N 2 O production via denitrification, but not N 2 O production via ammonia oxidation. Fertiliser application management (rate, timing and placement) is also recognised as an important practice to minimise N 2 O emissions from soil (Snyder et al., 2009). In systems with compacted interrows, the selection of banded fertiliser placement could be a mitigation option, by separating N sources from high-risk areas, compared to broadcast placement (Nash et al., 2012).
In grassland soils, the regulation of grazing periods based on the soil water content (Bhandral et al., 2007), and the reduction of stocking rates and/or length of grazing periods (de Klein and Ledgard, 2005), is regarded as important in limiting the trampling-compaction impact on N 2 O emission. It was indicated above that delaying nitrification of reduced N sources can mitigate N 2 O emissions. In accordance with this, the use of nitrification inhibitors has been found to reduce N 2 O emissions from both trampled and non-trampled soils with prolonged effectiveness, examples of inhibitors are dicyandiamide, nitrapyrin and 3, 4-dimethyl pyrazole phosphate (DMPP) (Subbarao et al., 2006).
Another management strategy may be the use of biochar. In China, biochar application in compacted soils was found to mitigate N 2 O emissions by 18%, with significance for the time/magnitude of peak emissions after N fertilisation and precipitation/irrigation (Liu et al., 2017). However, in that same study, the biochar effect on N 2 O emission was mainly associated with a chemically-mediated (change in pH) rise in the abundance of both nitrifiers and denitrifiers, and biochar may not have directly contributed to the reduction of soil anaerobic microsites. Other studies have shown an effect of biochar on soil bulk density or hydraulic properties (e.g. Burrell et al., 2016;Verheijen et al., 2019;Toková et al., 2020), indicating that biochar has a N 2 O mitigation potential.

Future research directions
This literature review shows that significant efforts have already been made to elucidate how soil compaction contributes to N 2 O emissions in managed agroecosystems. Nevertheless, most studies have focused on specific N 2 O-soil interactions, and there is a lack of comprehensive studies which can relate the spatiotemporal distribution of N 2 O emissions, in an integrated way, to soil biophysical interactions as modified by structural stratification, management practices and climate variation. The complex nature of the interactions among these factors is poorly understood and this has revealed a number of more specific knowledge gaps. The present literature review identified a need for future research focusing on (i) understanding fluid transport-pore network behaviour in relation to denitrification as the main source of N 2 O; (ii) N 2 O flux thresholds as constrained by selected model explanatory variables (site-specific conditions) such as soil texture, structure, plant cover (mineral N availability) and fertilisation; (iii) understanding the relationship between N 2 O fluxes from organic hotspots and the interactive effects of gas diffusivity, labile organic matter and nitrate availability, as well as the influence of soil compaction on the development of N 2 O hotspots and hot moments; (iv) seasonal variations in the effects of topsoil and subsoil compaction on N 2 O emissions and the impact of soil recovery after compaction; (v) understanding the contribution of subsoil compaction to N 2 O emissions to the atmosphere; (vi) assessment of links and interactions among soil compaction, pore morphology, thermal conductivity and microbiome; and (vii) influence of soil compaction drivers (vehicular traffic, livestock trampling) on spatiotemporal changes in the degree of soil compaction and the development of N 2 O hotspots and hot moments. Future studies are hence called upon to contribute to developing least emissions-prone agricultural systems.

Conclusions
The international literature reviewed here recognises the significant risk for higher N 2 O emissions caused by soil compaction. Fertilisation, moisture content, drying/rewetting cycles and agricultural systems are the main observation criteria when evaluating compaction effects for cropland, grassland and forest land. The main focus has been given to topsoil compaction, since the contribution of subsoil compaction to N 2 O emission is poorly known. Most often soil water metrics interacting with soil compaction has been evaluated for regulation of N 2 O emissions, but gas diffusivity has been found to better explain N 2 O fluxes. Gas diffusivity in soil is determined by both air-filled porosity and tortuosity. Yet, a common issue in studies focusing on soil compaction-induced N 2 O emission is the poor characterisation of the structural state of the soil, meaning the degree of compactness and pore system functionality, and the drivers causing the structural damage. This leads to uncertainties in the selection and evaluation of critical management factors. A large proportion of N 2 O emissions are associated with hotspots and hot moments, and there is a need for comprehensive studies to understand how conditions for C and N turnover in these situations are modified by different degrees of compaction. Understanding the direct and indirect effects of soil physical conditions on microbial activities is key to the selection and implementation of effective mitigation strategies.