Contrasting effects of different pH‐raising materials on N2O emissions in acidic upland soils

Acidic soils, occupying ca. 40% of the world's arable soils, often need to be managed (e.g., to raise their pH and to improve crop productivity); however, the environmental impact of raising soil pH is often difficult to assess. Increasing soil pH stimulates the reduction of N2O to N2, thus lowering N2O emissions associated with denitrification, but can also increase autotrophic nitrification rates and related N2O emission. Using a 15N tracing technique, we provide process‐based insights into the effects of two acid‐neutralizing materials (quicklime [CaO] vs. pig manure) on N2O emissions in an acidified upland soil that had experienced excessive N application. Without pH adjustments we found that N2O emissions, stimulated by supply of reactive N, were related to denitrification‐ and heterotrophic nitrification‐derived N2O emissions, whereas autotrophic nitrification‐derived N2O emissions declined with decreasing soil pH. These effects were reversed by increasing soil pH via liming. However, increasing the soil pH via application of pig manure significantly increased soil N2O emissions from both nitrification and denitrification. Our study highlights that pH‐amelioration practices may enhance N2O emissions depending on the type of material applied to the soil. Therefore, both pH remediation and greenhouse gas mitigation options need to be considered together to avoid adverse environmental effects. The effect of different acid‐neutralizing materials on soil N2O emissions should be incorporated into ecosystem models to better estimate global N2O emissions when pH amelioration is practised.

soil. Therefore, both pH remediation and greenhouse gas mitigation options need to be considered together to avoid adverse environmental effects. The effect of different acid-neutralizing materials on soil N 2 O emissions should be incorporated into ecosystem models to better estimate global N 2 O emissions when pH amelioration is practised.

Highlights
• Enhanced N 2 O emission by N input was from denitrification and heterotrophic nitrification.
• Chemical N input and liming have reversible effects on N 2 O emission.
• Soil N 2 O emission was decreased by liming but increased by animal manure input.
• Careful consideration of pH raising substrates is needed to avoid adverse effects. Nitrous oxide (N 2 O) is an important greenhouse gas with a global warming potential of 298 times that of CO 2 over a 100-year time period, contributing to positive radiative forcing and ozone destruction in the stratosphere (Kanter et al., 2013;Stocker et al., 2013). Globally, agricultural soils have been found to contribute approximately 50% of anthropogenic N 2 O emissions, mainly due to the input of chemical nitrogen (N) fertilizers (Bouwman, Boumans, & Batjes, 2002;Tian et al., 2019). A global meta-analysis of 1,104 field measurements further indicated that N 2 O emissions in acidic soils are highly sensitive to changing N fertilization practices compared to neutral-alkaline soils .
Approximately 40% of the world's arable soils are acidic (i.e., pH < 5.5), with a tendency to become even more severe in recent years (Bian, Zhou, Sun, & Li, 2013;Von Uexkuell & Mutert, 1995). There are global concerns regarding acidified arable land, particularly in China (Guo et al., 2010;Hajkowicz & Young, 2005), because soil acidification has adverse effects on plants and soil microorganisms due to aluminium toxicity and deficiencies of some nutrients (Kunhikrishnan et al., 2016). Liming is recommended to alleviate soil acidification (Goulding, 2016;Holland et al., 2018). The application of animal manure may also have a liming effect due to its ash alkalinity , with the added advantage that the animal manure supplies some of the N required by crops. Increasing soil pH by liming and/or animal manure can affect N transformation rates such as denitrification and autotrophic and heterotrophic nitrifications, with all of them involved in the production of N 2 O (Zhang, Müller, & Cai, 2015). It has been suggested that increasing soil pH may be a viable abatement technique to mitigate N 2 O emission from acid soils (Wang, Guo, et al., 2018). However, the effects of different acid-neutralizing materials (e.g., quicklime vs. animal manure) on N 2 O emissions from acidic soils are unclear. A better process-based understanding is pivotal to accurately estimate global N 2 O emissions from acid soils undergoing pH-raising amelioration treatments.
The effect of liming on soil N 2 O emissions is complex (McMillan et al., 2016;Senbayram et al., 2019). An important mitigation effect appears to be the shifting of the N 2 O to N 2 ratio during denitrification towards N 2 under less acidic conditions (Liu, Mørkved, Frostegård, & Bakken, 2010;Šimek & Cooper, 2002). Although increased soil pH often stimulates the reduction of N 2 O (lower N 2 O/N 2 ratio), reduction may be partially or completely counterbalanced by increased total denitrification rates under higher soil pH (Saggar et al., 2013). It has also been shown that liming decreases the relative proportion of N 2 O emissions via autotrophic nitrification, but enhances autotrophic nitrification, which increases the production of NO 3 − , and therefore the N 2 O emission resulting from the coupling of nitrification-denitrification processes (Barton, Gleeson, Maccarone, Zúñiga, & Murphy, 2013;Senbayram et al., 2019). Thus, there is considerable debate over whether liming would be a viable mitigation strategy to reduce N 2 O emissions from N-fertilized acidic arable soils (Higgins, Laughlin, & Watson, 2013;Qu, Wang, Almoy, & Bakken, 2014;Senbayram et al., 2019). The question is how liming affects denitrification-, autotrophic nitrification-and heterotrophic nitrification-derived N 2 O emissions. In addition to the changed pH, animal manure application can stimulate N 2 O emissions through the supply of labile carbon (C) and N, thus potentially increasing denitrification activity, which may outweigh the N 2 O emission savings due to a decreased N 2 O/N 2 ratio as a result of increasing pH (Saggar et al., 2013;Wang, Chadwick, Cheng, & Yan, 2018). Thus, understanding how liming and animal manure affect N 2 O production through denitrification and autotrophic and heterotrophic nitrification is critical for developing more reliable process-based N 2 O emission estimation models and specific N 2 O emission mitigation strategies.
Here, we investigated how long-term application of quicklime (CaO) and pig manure influences gross N transformation rates and associated N 2 O emissions under aerobic conditions in an acidified upland soil caused by excessive N application. We hypothesized that (a) the stimulation of N 2 O emission by chemical N fertilizer applicationinduced soil acidification would be completely reversed by quicklime application, given that both applications mainly affect soil pH, and (b) different acid-neutralizing materials (quicklime vs. pig manure) have different effects on soil N 2 O emissions due to the interactive effects of pH, and C and N availability following animal manure application.

| Site description and experimental design
The long-term field experiment was initiated in 1990 at Qiyang Experimental Station (26 45 0 12 00 N, 111 52 0 32 00 E) at the Chinese Academy of Agricultural Sciences, Hunan Province, China, and is still running now. The site has been managed according to a typical agriculture practice in the area, with a summer maize (Zea mays L.)-winter wheat (Triticum aestivum L.) rotation system. In each year, winter wheat was sown in strips in early November and harvested in early May in the following year. Summer maize was sown between the wheat strips in early April and harvested in July. The site is characterized by a subtropical monsoon climate with an annual average temperature of about 18 C and an annual average precipitation of approximately 1,250 mm . The soil is classified as Ferric Acrisols (WRB) originated from the Quaternary red clay parent material . In the region, soil acidification is a natural process as a result of intensive leaching, whereas increased input of chemical N fertilizers for achieving high yields has been accelerating the process . Prior to establishing the long-term field experiment, the field was under a wheat-maize rotation for 3 years without fertilization to achieve low and uniform nutrient levels. The soil (0-20 cm) at the beginning of the long-term experiment had a pH of 5.7, organic C of 6.1 g kg −1 , total N of 1.1 g kg −1 , total P of 0.5 g kg −1 , total K of 13.7 g kg −1 , available N of 79.0 mg kg −1 , available P of 13.9 mg kg −1 and available K of 104 mg kg −1 .
The long-term experiment included 11 treatments: UC (unfertilized control); NK, NP and NPK (mineral fertilizer, where N, P and K stand for nitrogen, potassium and phosphorus, respectively); NKL, NPL and NPKL (mineral fertilizer + L = quicklime); NPKS (mineral fertilizer + S = straw); NPKSL (mineral fertilizer + straw + quicklime); NPKM (mineral fertilizer, 30% of applied N,+ M = pig manure [70% of applied N]); and M (pig manure only). For the treatments receiving N fertilizer, the total amount of N (manure and urea) application was the same each year. For the NPKM treatment, 30% of the total N was applied as urea and the remaining 70% was applied as pig manure. The pig manure with an average water content of around 70% was collected from local farms. The pig manure contained 82.7 cmol kg −1 ash alkalinity, 368 g kg −1 organic C, 18.2 g kg −1 total N, 13.7 g kg −1 total P and 13.3 g kg −1 total K, with a pH of 8.8. For the NPKS and NPKSL treatments, half of the crop residues were left on the soil surface each year, considering that returning all straw with wide C/N ratios to the soil would result in high microbial N demand and subsequently severe N limitation for crop production (Chen, Liu, Tian, & Zhang, 2014). The chemical fertilizers were applied in the form of urea (300 kg N ha −1 year −1 ), superphosphate (53 kg P ha −1 year −1 ) and potassium chloride (100 kg K ha −1 year −1 ). Annually, 30% of the fertilizers were applied in the winter wheat season (November-May) and 70% in the summer maize season (April-July). Both chemical fertilizers and manure were applied by band placement at a depth of 10 cm after sowing of each crop. Because of rapid soil acidification in plots subjected to annual inputs of chemical N fertilizers since 1990, in 2010, half of the plots in the NK, NP, NPK and NPKS treatments received 2,250 kg ha −1 quicklime according to the same N fertilization amount to alleviate soil acidification, followed by the addition of 1,500 kg ha −1 quicklime in 2014 . Quicklime powder was broadcasted in the fallow season (August-October) and mixed with the soil by subsequent manual ploughing (10 cm).
Fresh soil samples (0-20 cm) were collected in October 2015 after the maize harvest using a 5-cm-diameter soil auger. A total of nine soil cores were randomly taken and homogenized to produce one composite soil sample from each plot. The soil samples were immediately sieved through a 2-mm mesh, homogenized thoroughly and then divided into two subsamples. One subsample was stored at 4 C for less than a week prior to conducting the 15 N tracing incubation experiments; another was air-dried for the determination of soil pH, total C and N concentrations ( Figure 1).

| 15 N tracing experiment
A 15 N tracing experiment was conducted in triplicate with 15 N-labelled NH 4 + and 15 N-labelled NO 3 − . Fresh soil samples (20 g, oven-dried basis) were placed in 250-mL flasks and sealed. The flasks were then preincubated in the dark at 25 C in the laboratory for 1 day to allow equilibration. After preincubation, 2 mL of either 15 NH 4 NO 3 or NH 4 15 NO 3 solution (10 atom% 15 N excess) were added to each soil sample by pipetting the solutions homogenously over the soil surface, reaching final concentrations of 50 mg of NH 4 + -N and 50 mg of NO 3 − -N kg −1 soil. Subsequently, the moisture content was adjusted to 60% water-holding capacity by adding deionized water. The flasks were then sealed with rubber stoppers and incubated at 25 C in the dark for an additional 6 days. During the incubation, the samples were aerated for 30 min each day to maintain aerobic conditions inside the flasks, and any water loss was replaced every 3 days with deionized water as required. Incubation conditions mimicked the actual conditions of upland soils.
Gas samples were taken from the headspace of the flasks on days 2, 4 and 6. Before each gas sampling event, the flasks were opened for 30 min to renew the atmosphere inside and immediately sealed for 6 hours using a silicone sealant. Before sampling, the headspace gas was mixed by withdrawing and back injecting headspace gas five times using a 20-mL gas-tight syringe with a stopcock. There were two groups of flasks with the same soil samples used for N 2 O and NO sampling, respectively. For one group of the flasks, 40-mL gas samples were collected using a 50-mL gas-tight syringe with a stopcock from the headspace of each flask at the end of 6-hr incubations and immediately injected into two pre-evacuated vials (18.5 mL), to determine the concentrations of N 2 O and CO 2 and isotopic composition of the N 2 O. For another group of the flasks, 40-mL gas samples were also collected and transferred to a multi-layer foil sampling bag (2 L) for NO analysis. After headspace samples were taken, three flasks were randomly selected from each labelling treatment, and the soils were extracted by 2 M KCl (at a soil:solution ratio of 1:5) for determination of the concentration and isotopic composition of NH 4 + and NO 3 − . The extraction procedure was also performed 0.5 hr following 15 N addition to avoid disturbance of abiotic N immobilization. After KCl extraction, residual soils were washed with deionized water three times until any residual inorganic N was removed, oven-dried at 60 C to a constant weight, and ground to pass through a 0.15-mm sieve for 15 N analysis of insoluble organic N.

| Analyses
Prior to NO determination, gas samples drawn from flasks were diluted to 1 L with high-purity N 2 . The NO concentrations were measured using a NO x analyser (ThermoFisher 42i, Chemiluminescence Detector, Ther-moFisher, Franklin, MA, USA). The N 2 O and CO 2 concentrations were determined using an Agilent 7,890 gas chromatograph (Santa Clara, CA, USA). The isotopic compositions of the N 2 O samples were determined via mass spectrometry (Thermo Finnigan, MAT 253, Waltham, MA, USA). The concentrations of NH 4 + and NO 3 − in the KCl extracts were determined using a continuous- flow analyser (SA1000, Skalar, Breda, the Netherlands). NH 4 + and NO 3 − were separated for 15 N measurements by distillation with magnesium oxide and Devarda's alloy, respectively (Bremner, 1996). The isotopic compositions of NH 4 + , NO 3 − and insoluble organic N were determined by an automated C/N analyser isotope ratio mass spectrometer (Europa Scientific Integra, Sercon 20-22, Crewe, UK). Soil total C and N concentrations were determined by dry combustion using an elemental analyser (Vario MAX CN Elemental Analyzer, Elementar, Hanau, Germany). Soil pH was determined in deionized water (soil:water, 1:2.5) using a pH metre (Mettler Toledo FE20, Shanghai, China).

| 15 N tracing model and statistical analyses
Soil gross N transformation rates were quantified using the 15 N tracing analysis model developed by Müller, Rütting, Kattge, Laughlin, and Stevens (2007) Figure S2 and S3) (Müller, Stevens, & Laughlin, 2004). According to the kinetic settings and the final parameters, average gross N transformation rates were calculated over the 6-day period and expressed in units of mg N kg −1 soil day −1 . Cumulative emissions of NO, N 2 O and CO 2 were estimated by successive linear interpolations across sampling intervals. The relative contribution of different processes to the N 2 O production was calculated using the equations developed by Stevens, Laughlin, Burns, Arah, and Hood (1997), and modified and extended by Rütting, Clough, Müller, Lieffering, and Newton (2010). Three pools were considered as possible substrates for N 2 O, which differed in their 15 N atom fraction, α d (NO 3 − pool, assumed to be the e-acceptor denitrification), α a (NH 4 + pool, assumed to be driving autotrophic nitrification) and α h (N pool at natural abundance, assumed to be the substrate for oxidation of organic N). The 15 N atom fraction of N 2 O (αN 2 O) is determined by:  (1) allows the calculation of d, n a and n h (via the solver routine available in Microsoft-Excel). Paired sample t-tests were used to compare significant differences in soil properties, cumulative CO 2 , N 2 O and NO emissions, NO/N 2 O ratios and gross N transformation rates between NKL, NPL, NPKL and NPKSL treatments compared to their respective quicklime-free treatments (NK, NP, NPK and NPKS), and between NPK and NPKM treatments. One-way ANOVA with a least significant difference (LSD) test was used to assess differences in various variables among all treatments. All statistical analyses were performed with SPSS 16.0. All results are reported as means ± standard deviations on a soil dry weight basis.

| Soil properties
In comparison with UC, long-term application of NK, NP, NPK and NPKS (N-containing treatments) significantly decreased soil pH from 5.49 to approximately 4.10, whereas soil pH was maintained close to or above the level of UC after application of either quicklime (NKL, NPL, NPKL and NPKSL, NL-containing treatments) or animal manure (NPKM and M, M-containing treatments) (Figure 1). Compared with UC, the longterm application of NK, NP, NPK and NPKS generally increased soil total C and N concentrations (Figure 1). However, the application of quicklime (NKL, NPL, NPKL and NPKSL) in most cases did not further increase soil total C and N concentrations compared with their respective quicklime-free treatments. In contrast, soil total C and N concentrations were significantly higher in NPKM and M treatments than in NPK and UC treatments.

| Gross N transformation rates
The observed and modelled concentrations and 15 N enrichment values for NH 4 + and NO 3 − were in good agreement for all treatments ( Figures S2 and S3), indicating the modelled gross N transformation rates were reliable. Gross N mineralization rates were significantly greater in M treatments compared to the other treatments but not different among the various M treatments (Table 1). Gross NH 4 + immobilization rates were consistently greater in NKL, NPL, NPKL and NPKSL treatments than their respective quicklime-free treatments. Animal manure application (NPKM and M) also resulted in a significant increase in gross NH 4 + immobilization rates compared with NPK and UC treatments. Gross autotrophic nitrification rates were almost completely suppressed by the application of NK, NP, NPK and NPKS (N-containing treatments), but were maintained at a similar level to UC treatment due to the application of NKL, NPL, NPKL and NPKSL (NL-containing treatments) (Table 1). Animal manure application (NPKM and M) led to an approximately 3-6-fold increase in gross autotrophic nitrification rates in comparison with UC. Both gross heterotrophic nitrification and NO 3 − immobilization rates were generally negligible for all treatments. The DNRA rates were negligible in UC, NK, NP, NPK and NPKS treatments, but were enhanced by quicklime application except for NPKS. Manure application resulted in a more pronounced increase in DNRA rates than the quicklime application. Gross N mineralization rates showed a quadratic response function with increasing total C and N concentrations ( Figure S4a). The response of gross autotrophic nitrification rates to soil pH was different with the two acid-neutralizing substances ( Figure S4b). There was a linear and positive relationship between gross autotrophic nitrification rates and soil pH when quicklime was used to alleviate soil acidity, whereas an exponential and positive relationship between gross autotrophic nitrification rates and soil pH was observed when manure was used as ameliorant. Such results indicated that gross autotrophic nitrification rates could be stimulated to a greater degree by manure than by quicklime when soil pH was increased to a similar value. Gross NH 4 + immobilization rates were positively correlated with soil pH (in fact logarithmic if the H + ion concentrations are taken into account) and positively correlated with total C concentrations, plateauing at C concentrations of 14 g C kg −1 (i.e., in the form of a cubic response function, Figure S4c,d). The DNRA rates were positively correlated with soil pH and total C concentrations ( Figure S4e,f).

| Soil CO 2 , N 2 O and NO emissions
Cumulative CO 2 emissions over the whole incubation period were highest in the M treatment, followed by NPKM and NPKSL treatments (Figure 2a). The cumulative CO 2 emissions were generally greater in NKL, NPL, NPKL and NPKSL treatments than in their respective quicklime-free counterparts, although not significantly. Compared with their respective quicklime-free treatments (NK, NP, NPK and NPKS), the cumulative NO and N 2 O emissions were reduced by quicklime application (NKL, NPL, NPKL and NPKSL treatments) by 20-58% and 28-38%, respectively, although not significantly in the latter (Figure 2b,c). In contrast, manure applications (NPKM and M) led to around 28% decrease in NO emissions and 87-828% increase in N 2 O emissions compared with NPK and UC treatments. The NO/N 2 O ratio was 1.2 and 0.3 in the NPKM and M treatments, respectively, and both were significantly lower compared to other treatments (2.8-5.6) (Figure 2d). The response of cumulative N 2 O emissions to soil parameters varied with the pH amelioration substance ( Figure S5). Cumulative N 2 O emissions exhibited an either exponential or linear decline with soil pH, total C concentration, cumulative CO 2 emissions and gross autotrophic nitrification rates when quicklime was used to alleviate soil acidification. However, the trends were reversed when manure was used as ameliorant. Both cumulative NO emissions and the NO/N 2 O ratios were negatively correlated with soil pH (Figure S6a,b). In addition, cumulative NO emissions showed a steep decline with increasing gross autotrophic nitrification rate and cumulative CO 2 emissions and then levelled out ( Figure S6c,d).  (Figure S3). In the UC treatment, the average contribution of denitrification and autotrophic and heterotrophic nitrification to N 2 O production was 46, 32 and 21%, respectively, whereas denitrification and autotrophic and heterotrophic nitrification were responsible for 41-53%, 6-18% and 37-47% of N 2 O production in NK, NP, NPK and NPKS treatments, respectively (Figure 3a). In NKL, NPL, NPKL and NPKSL treatments, denitrification and autotrophic and heterotrophic nitrification contributed 20-63%, 19-59% and 8-54% of N 2 O production, respectively. In NPKM and M treatments, denitrification and autotrophic and heterotrophic nitrification contributed 38-42%, 41-48% and 10-22% of N 2 O production, respectively (Figure 3a). The contribution of denitrification and heterotrophic nitrification to N 2 O production decreased linearly and exponentially with increasing soil pH, respectively (Figure 4a,b), whereas the contribution of autotrophic nitrification to N 2 O production was linearly and positively correlated with soil pH (Figure 4c). In addition, the contribution of autotrophic nitrification to N 2 O production increased linearly and exponentially with the increase of autotrophic nitrification rates after quicklime and manure application, respectively (Figure 4d).

| Sources of N 2 O emissions
In the UC treatment, the cumulative N 2 O production from processes of denitrification and autotrophic and heterotrophic nitrification were 1.1, 0.8 and 0.5 μg N kg −1 , respectively (Figure 3b). In NK, NP, NPK and NPKS treatments, the cumulative N 2 O production from denitrification and autotrophic and heterotrophic nitrification was 1.2-1.7, 0.2-0.5 and 1.2-1.7 μg N kg −1 , respectively.
In NKL, NPL, NPKL and NPKSL treatments, the cumulative N 2 O production from denitrification and autotrophic and heterotrophic nitrification showed tendencies of decrease (0.4-1.6 μg N kg −1 ), increase (0.4-1.1 μg N kg −1 ) and decrease (0.2-1.1 μg N kg −1 ) compared with their respective quicklime-free treatments (NK, NP, NPK and NPKS), respectively (Figure 3b). In contrast, in NPKM and M treatments, the cumulative N 2 O production from denitrification and autotrophic and heterotrophic nitrification was as high as 2.2-12, 2.4-14 and 1.3-2.9 μg N kg −1 , respectively. Cumulative N 2 O production from denitrification decreased linearly with increasing soil pH when quicklime was used to alleviate acidification, but increased exponentially with increasing soil pH and total C concentration when manure was used as ameliorant (Figure 5a, b). Cumulative N 2 O production from heterotrophic nitrification was negatively and linearly correlated with soil pH when quicklime was used to alleviate acidification, but was positively and exponentially correlated with total C concentration when manure was used as ameliorant (Figure 5c,d). With increasing pH and gross autotrophic nitrification rates, cumulative N 2 O production from autotrophic nitrification increased linearly and exponentially when quicklime and manure were used to alleviate acidification, respectively (Figure 5e,f). F I G U R E 5 Relationships between soil pH and total C concentration and cumulative N 2 O flux from denitrification (EDN, a, b) and heterotrophic nitrification (EHN, c, d) and between soil pH and gross autotrophic nitrification rate (O NH4 ) and cumulative N 2 O flux from autotrophic nitrification (EAN, e, f). Insert (small figures) shows that NPKM and M treatments were excluded from the analysis, and big figures show that NKL, NPL, NPKL and NPKSL treatments were excluded from the analysis. NPKM, mineral NPK fertilizer (30% of applied N) plus swine manure (70% of applied N) each crop season; M, manure only; NKL, NPL, NPKL, mineral NK, NP, NPK fertilizer plus quicklime, respectively; NPKSL, mineral NPK fertilizer plus straw and quicklime [Color figure can be viewed at wileyonlinelibrary.com]

| DISCUSSION
It is well known that excessive N use contributes substantially to soil acidification (Guo et al., 2010). Our results showed that 25 years of chemical N fertilizer application (NK, NP, NPK and NPKS, N-containing treatments) resulted in a significant decline in soil pH from 5.49 to 4.10 ( Figure 1). Meanwhile, soil pH was maintained close to or above the original level after application of either quicklime (NKL, NPL, NPKL and NPKSL, NL-containing treatments) or animal manure (NPKM and M, M-containing treatments). Such results demonstrated that animal manure has the same potential to alleviate N-induced soil acidification with quicklime. Changes in soil pH have been found to affect N 2 O emissions through changing microbial-mediated N transformation processes (Li, Sørensen, Olesen, & Petersen, 2016;Zhang, Zhao, Cai, Müller, & Zhang, 2018). However, the effects of the two contrasting acid-neutralizing materials (quicklime vs. animal manure) on soil N 2 O emissions and their underlying mechanisms have not been addressed explicitly. Our study clearly shows that quicklime and animal manure applications have contrasting effects on soil N 2 O emissions in an acidified upland red soil ( Figure 6). Our findings of quicklime-induced decrease and animal manure-induced increase in N 2 O emissions are in contrast to the prevailing view that increasing soil pH can be a viable strategy for lowering N 2 O emissions in acidic soils (Qu et al., 2014;Russenes, Korsaeth, Bakken, & Dorsch, 2016;Wang, Guo, et al., 2018), thus highlighting extreme caution in applying pH-raising substances to mitigate N 2 O emissions. Our study also provides a compelling processbased understanding of how N 2 O emissions change in response to application of quicklime and animal manure ( Figure 6).

| Potential mechanisms to explain the effects of quicklime and animal manure application on soil N 2 O emissions
We found that increasing soil N 2 O emissions following long-term application of chemical NK, NP, NPK and NPKS F I G U R E 6 Schematic showing how the quicklime and animal manure application affects N 2 O emission differently. The data in the UC, N-containing, NL-containing and M-containing treatments represent the value from original UC treatment, the average value from the original NK, NP, NPK and NPKS treatments, NKL, NPL, NPKL and NPKSL treatments, and NPKM and M treatments, respectively. SON, soil organic N pool; Den, denitrification. CDN, CHN and CAN are the contributions of denitrification, heterotrophic nitrification and autotrophic nitrification to N 2 O production, respectively. EDN, EHN and EAN are the cumulative N 2 O fluxes from denitrification, heterotrophic nitrification and autotrophic nitrification, respectively. UC, unfertilized control; NK, NP, NPK, mineral NK, NP, NPK fertilizer alone, respectively; NKL, NPL, NPKL, mineral NK, NP, NPK fertilizer plus quicklime, respectively; NPKS, mineral NPK fertilizer plus straw; NPKSL, mineral NPK fertilizer plus straw and quicklime; NPKM, mineral NPK fertilizer (30% of applied N) plus swine manure (70% of applied N) each crop season; M, manure only [Color figure can be viewed at wileyonlinelibrary.com] fertilizers was attributed to enhanced denitrification-and heterotrophic nitrification-derived N 2 O emissions, which offset reduced autotrophic nitrification-derived N 2 O emissions ( Figure 6). The opposite effect was observed in all lime treatments (i.e. NKL, NPL, NPKL and NPKSL) compared with NK, NP, NPK and NPKS treatments. We found that both denitrification-and heterotrophic nitrification-derived N 2 O emissions were negatively correlated with soil pH, whereas autotrophic nitrification-derived N 2 O emissions were positively correlated with soil pH when quicklime was used to alleviate acidification ( Figure 5). This is in line with other studies, showing a pH-associated stimulation of N 2 O reduction (lower N 2 O/N 2 ratio) and in turn reduced N 2 O emissions via denitrification (Mukumbuta, Uchida, & Hatano, 2018;Qu et al., 2014), as well as lower N 2 O emissions via heterotrophic nitrification (Li, Chapmanc, Nicold, & Yao, 2018;Zhang et al., 2015;Zhang et al., 2018), but increased N 2 O emissions via autotrophic nitrification (Zhang et al., 2013). Recent studies on the same site indicated that ammonia-oxidizing bacteria (AOB) rather than ammonia-oxidizing archaea (AOA) were mainly responsible for increased autotrophic nitrification after long-term liming despite a general dominance of AOA in these acidic soils .
Another important factor to explain the reduced N 2 O emissions due to liming is the aerobic state of the soil. Under anaerobic conditions, increasing soil pH by liming promotes the reduction of N 2 O (lower N 2 O/N 2 ratio) through denitrification when most of the N 2 O originates from denitrification (Mukumbuta et al., 2018;Qu et al., 2014;Russenes et al., 2016). This was due to the fact that the activity of the N 2 O reductase enzyme generally increases with increasing pH values (Saggar et al., 2013;Šimek & Cooper, 2002). Further study showed that the low N 2 O reductase activity in acid soils could not be ascribed to the low number of nosZ transcripts . However, under aerobic conditions and increasing soil pH, N 2 O reduction is restricted by the availability of NO 2 − (Feng, Yan, Hütsch, & Schubert, 2003) and only noticeable when N 2 O via autotrophic nitrification is the dominant N 2 O source (Barton et al., 2013;Mørkved, Dorsch, & Bakken, 2007). Overall, these explanations were prevailing but split. Instead, our results integrated the above explanations and provided a full picture of how increased soil pH following quicklime application changed N 2 O emission in an acidified upland soil caused by chemical N application ( Figure 6). In contrast to liming, animal manure application significantly increased denitrification-, autotrophic-and heterotrophic nitrification-derived N 2 O emissions ( Figure 6). Our results are in contrast with the common viewpoint that organic material input stimulates N 2 O emissions via denitrification under aerobic conditions (Charles et al., 2017;Chen, Li, Hu, & Shi, 2013;Li et al., 2016). Autotrophic nitrification-derived N 2 O emissions were positively correlated with soil pH and gross autotrophic nitrification rates (Figure 5e,f), showing a stimulating effect of manure on autotrophic nitrification rates and associated N 2 O emissions, probably by increased soil pH ( Figure S4b). Other possible explanations for manureinduced stimulation of autotrophic nitrification and N 2 O emission could be the enhanced substrate  availability from what the pig manure itself contains, the mineralization of pig manure and increased soil N mineralization due to a priming effect (Table 1). This is in line with previous findings at the same site that the NPKM treatment had the highest copy numbers of AOB and AOA amoA genes among the treatments that received mineral fertilizers (He et al., 2007). The positive relationship between denitrification-derived N 2 O emissions, soil pH and total C concentrations (Figure 5a,b) showed that the stimulation of denitrification to soil N 2 O emissions with animal manure application outweighed any possible decline in the N 2 O/N 2 ratio with pH increase, possibly due to the C-induced activity and associated increase of the anaerobic microenvironment (Saggar et al., 2013). Also the pH-stimulated effect on autotrophic nitrification rates and NO 3 − build-up (Table 1) in conjunction with increased C availability further enhanced denitrification activity (Senbayram et al., 2019). Heterotrophic nitrification-derived N 2 O emissions were positively correlated with soil total C concentrations, but not with increased soil pH (Figure 5c,d), which is in line with other studies Zhang et al., 2018). Next to the observed C-induced increase of heterotrophic nitrification-derived N 2 O emissions it is, however, possible that soil pH decreased the relative proportion of heterotrophic nitrification-related N 2 O emissions but this was masked by the overall C-stimulated activity increase. Thus, complex interactions among soil pH, available C input and relative N 2 O emissions seem to control N 2 O emissions from this acidified upland acidic soil under animal manure application.
4.2 | Potential mechanisms for individual N 2 O production processes in response to quicklime and animal manure application and implications for N 2 O mitigation Studies found that nitrification-related N 2 O emissions could be dominant under aerobic conditions (Mathieu et al., 2006;Stevens et al., 1997). Our study showed that nitrification (autotrophic + heterotrophic) and denitrification contributed almost equally to N 2 O emissions in the UC and N-containing treatments in an upland red soil. Meanwhile, long-term quicklime and animal manure application decreased the contribution of denitrification to N 2 O emissions. There was a negative relationship between the contribution of denitrification to N 2 O emissions and soil pH (Figure 4a). A shift in microbial N 2 O source in response to a short-term change in pH has been reported by Baggs, Smales, and Bateman (2010), who observed that liming of a pH 4.5 soil to pH 7 shifted the predominance away from denitrification to autotrophic nitrification in an arable soil. Previously we showed via a regression analysis that the contribution of denitrification to N 2 O production decreased with increasing soil pH, with a threshold of approximately pH 4.4, below which denitrification dominated N 2 O production (Cheng, Zhang, Wang, Cai, & Wang, 2015). Moreover, a global meta-analysis from agriculture revealed that pH was a critical factor regulating N 2 O production but the authors failed to present the underlying mechanisms (Wang, Guo, et al., 2018). Results from this long-term study demonstrated that soil pH is a key predictor of the main N 2 O production pathways. Understanding and accurately identifying the dominating N 2 O production source is of particular importance to developing targeted N 2 O mitigation strategies. For instance, in manure-containing treatments, autotrophic nitrification accounted for 45% of N 2 O emissions, whereas denitrification contributed to 40% of N 2 O emissions. Although a nitrification inhibitor could not directly affect denitrification, the application of a nitrification inhibitor, which depresses autotrophic nitrification, would also limit denitrification via reduced nitrification-denitrification transformations, that is, decreasing NO 3 − concentrations. A recent study even found that the effectiveness of nitrification inhibitors used for mitigating N 2 O emissions seems to be more pronounced under conditions favouring denitrification . Nguyen et al. (2017) further demonstrated that the effectiveness of nitrification inhibitors in reducing denitrification-derived N 2 O emissions depended on whether denitrification is controlled by NO 3 − produced during nitrification.

| CONCLUSIONS
We showed that long-term N applications, causing soil acidification and an associated increase in N 2 O emissions, could be completely reversed by increasing soil pH via liming. However, application of animal manure, also used for pH amelioration, had the opposite effects on N 2 O emissions. This shows that we need to understand the full complexity of the effects of a substance used to ameliorate soil conditions (such as pH) before this can be recommended. Thus, careful consideration of pH-raising substrates is needed to avoid adverse environmental effects. Our study provides a full account of the effect of two acidneutralizing materials (quicklime vs. pig manure) on N transformation processes in the soil and associated N 2 O emissions. Co-application of nitrification inhibitors in combination with various acid-neutralizing materials should be tested to avoid negative effects on N 2 O emissions. Thus, acid-neutralizing materials need to be carefully chosen and the effect of these substances on the overall elemental cycles in soils should be taken into account in models that aim to predict global N 2 O emissions. However, it should be noticed that the soil used in this study was sampled only once from the field and the soil microbial community was not analysed. Further research is therefore required to link microbial community composition or potential function to the N 2 O emission processes in response to the input of acid-neutralizing materials with high-frequency soil sampling from the field.