The fate of fertilizer nitrogen in a high nitrate accumulated agricultural soil

Well-acclimatized nitrifiers in high-nitrate agricultural soils can quickly nitrify NH4+ into NO3− subject to leaching and denitrifying loss. A 120-day incubation experiment was conducted using a greenhouse soil to explore the fates of applied fertilizer N entering into seven soil N pools and to examine if green manure (as ryegrass) co-application can increase immobilization of the applied N into relatively stable N pools and thereby reduce NO3− accumulation and loss. We found that 87–92% of the applied 15N-labelled NH4+ was rapidly recovered as NO3− since day 3 and only 2–4% as microbial biomass and soil organic matter (SOM), while ryegrass co-application significantly decreased its recovery as NO3− but enhanced its recovery as SOM (17%) at the end of incubation. The trade-off relationship between 15N recoveries in microbial biomass and SOM indicated that ryegrass co-application stabilized newly immobilized N via initial microbial uptake and later breakdown. Nevertheless, ryegrass application didn’t decrease soil total NO3− accumulation due to its own decay. Our results suggest that green manure co-application can increase immobilization of applied N into stable organic N via microbial turnover, but the quantity and quality of green manure should be well considered to reduce N release from itself.

Human-induced nitrogen (N) input increased over tenfold with respect to one century ago 1,2 . Excessive N fertilization caused accumulation of significant amounts of N beyond crop absorption in soils, largely in the form of nitrate (NO 3 − ) 3,4 . Surplus NO 3 − in soil is problematic because it is susceptible to loss by leaching or denitrification 2,5 , which is both economically and environmentally undesirable 4,6 . High nitrification potential and low NO 3 − immobilization are responsible for NO 3 − accumulation in agricultural soils 7 . Long-term N fertilization and tillage greatly increase the population and alter the community structure of ammonia oxidizing bacteria, as well as the nitrification capacity 4,[8][9][10] . This is confirmed by the significant positively relationship between nitrification potential and soil NO 3 − content 7,11 . Co-application of N fertilizer with organic materials is especially common in high-input cropping systems, and it is considered to be a good agronomic practice as it is thought to elevate microbial bioavailability of the applied N and reduce N loss to the environment 12 . Much of immobilized N was speculated to be rapidly transformed to microbial residues or necromass as the average life cycle of N in microbial biomass was just several days 13,14 . While a recent study shows that soil newly synthesized amino acids are relatively easier to decompose in contrast with original soil amino acids 15 , the fate of newly immobilized N in various soil N pools is generally not well quantified.
In recent years, large areas of conventional cereal cultivation in China have been transferred to intensive greenhouse cultivation due to fast economic development and increased consumer demand 3 . Greenhouse soil cultivated with vegetables represents typical high NO 3 − soil in agricultural system 3,16 . Constant monoculture, excessive fertilization, and high-intensity anthropogenic interference during greenhouse cultivation change the process of soil N transformation and accelerate the accumulation of NO 3 − 10, 16 . Thus, special attention is required to pay on N transformation in the NO 3 − polluted soil (e.g., greenhouse soil) after the application of ammonium-based fertilizers.
Isotope 15 N tracing technology is widely used to study the fate of chemical fertilizers and other N amendments in ecosystems. Previous studies on the fate of soil added 15 N mainly focused on its transformation to Scientific RepoRts | 6:21539 | DOI: 10.1038/srep21539 extractable NH 4 + and NO 3 − pools, as well as N retention during a certain period (either short-term or long-term mostly) 7,9,17,18 . Due to the difficulties in experimental operation, dynamic and systematic studies on the fates of soil added 15 N entering into various N pools (e.g. microbial biomass N, extractable organic N, mineral fixed N, non-extractable organic N) are rare, while these forms of N may play significant roles as important intermediates in biogeochemical N transformation processes in soil 15,19 .
Altering the fate of chemical fertilizer N to lessen NO 3 − loss by manure co-application largely relies on the scale and the stability of manure-enhanced immobilization of fertilizer N. In this study, we conducted a 120-day incubation experiment to trace the fates of the inorganic fertilizer N alone or in combination with a green manure (as ryegrass in this study) to a greenhouse soil, using a 15 N trace technique. The purpose of this study was to examine the effect of co-application of green manure on the fates of applied N entering into seven different N pools. Specifically, we examined if co-application of green manure promoted the incorporation of soil newly immobilized N into relatively stable organic N pools instead of remineralization, and in a long run, reduced NO 3 − accumulation and loss risk in high-input greenhouse soil. We expected that green manure co-application would increase the stable immobilization of fertilizer N in the studied soil.

Dynamics of soil extractable N (NH 4
+ -N, NO 3 − -N, EON). In the control treatment, NH 4 + concentration was 0.6 to 4.9 mg N kg −1 , and did not change much over the 120-day incubation, while NO 3 − concentration increased somewhat from 152 mg N kg −1 at the beginning of incubation to 199 mg N kg −1 at the end of incubation (Fig. 1a,c). In the N addition treatment, NH 4 + decreased by 93-97% within 3 days from > 80 mg N kg −1 to background level of 2.7-4.5 mg N kg −1 . In contrast, NO 3 − concentration increased rapidly to 225-244 mg N kg −1 from 152 mg N kg −1 in the first 3 days (Fig. 1b,d). From the 7 to 120 days, NO 3 − concentration slightly increased, and 15 N recovery in NO 3 − pool, however, was relatively constant. Co-application of ryegrass significantly decreased the proportion of applied fertilizer NH 4 + conversion to NO 3 − . During the days from day 3 to day 120, 56-70% of the applied fertilizer NH 4 + was recovered as NO 3 − in the + 15 NH 4 + + Ryegrass treatment, being significantly lower than that (87-92%) in the + 15 NH 4 + treatment (Fig. 1d). Extractable organic N (EON) was 8.4 to 25.3 mg N kg −1 and accounted for 3-11% of total extractable nitrogen (TEN, NH 4 + -N + NO 3 − -N + EON) (Fig. 1e). Ryegrass addition slightly increased EON concentrations. Less than 2% of the applied fertilizer NH 4 + was recovered as EON, and there was no significant difference in the 15 N recovery in EON between the two 15 NH 4 + addition treatments (Fig. 1f).

Dynamics of soil microbial biomass N (MBN) and mineral fixed N (MFN).
In the control treatment, MBN concentrations was relatively constant, being 32.9 to 52.0 mg N kg −1 (Fig. 1g). NH 4 + addition increased MBN concentration slightly only in the first 3 day of incubation. However, co-application of ryegrass with NH 4 + increased MBN significantly in the first 30 days. After 60 days, there was no significant difference in MBN concentrations among the three treatments. In the + 15 NH 4 + treatment, 1.5-2.9% of the applied fertilizer NH 4 + was recovered to microbial biomass (Fig. 1h). In the + 15 NH 4 + + Ryegrass treatment, the proportion to microbial biomass changed over the incubation, from 5% in the 0.1 day increased to a maximum of 22% in the 1 day and decreased to 4% at the end of incubation.
MFN concentrations were comparable to NO 3 − concentration, but were not different among the three treatments (Fig. 1i). The 15 N recovery in MFN pool did not change and stabilize at 2-3% in the two N addition treatments during day 1-120 ( Fig. 1j).
Incorporation into soil organic matter (SOM) and N loss. The two 15 N addition treatments yielded different 15 N recoveries in TEN and non-extractable N (NEN) pools ( Fig. 2b,d). The 15 N recoveries in TEN pool were 88-99% for the + 15 NH 4 + treatment and 57-81% for the + 15 NH 4 + + Ryegrass treatment during the 120-day incubation, and showed a slight increase after an initial sharp drop. A reverse trend was found for the 15 N recovery in NEN pool (Fig. 2).
We defined the 15 N recovery in NEN pool minus in MFN and MBN pools (NEN-MFN-MBN) in the unfumigated post-extracted soil as incorporation into SOM, and defined the unrecovered 15 N (neither in TEN nor NEN pools) as loss (Fig. 3). During day 3-120 in the + 15 NH 4 + treatment, only 0.1-2% and 3-6% of the applied fertilizer N was incorporated into SOM and lost, respectively. When ryegrass was co-applied, 8-17% of the applied fertilizer N was incorporated into SOM and 6-15% of applied N was lost, respectively ( Fig. 3).

Discussion
Nitrification. Our results show that the studied greenhouse soil is characterized by high NO 3 − accumulation (> 150 mg N kg −1 ) and very fast nitrification. The net nitrification rates (net NO 3 − change) of the applied fertilizer NH 4 + were 28.7 and 27.7 mg N kg −1 d −1 for the + 15 NH 4 + and + 15 NH 4 + + Ryegrass treatments respectively during day 0.1-1, and 17.9 and 8.3 mg N kg −1 d −1 during day 1-3 ( Fig. 4), much higher than the reported values for many other ecosystems 4,11,20 . Nitrifiers were acclimatized with high quantity and improved function in the tested soils due to high-rate fertilization and long-term vegetable cultivation but limited C input 20,21 . In contrast, in N-limited forest soil, low N input and sufficient available C to fuel high denitrification and immobilization, prevents the NO 3 − accumulation 7,22,23 . The relative strength of N mineralization and immobilization mainly depends on the C/N of added organic material as well as the incubated soil [24][25][26] . The total rate of nitrification (including 14 N and 15 N) in the ryegrass co-applied soil was not reduced in this study (Figs 1c and 4), primarily due to the supply of non-labelled NH 4 + from relatively fast mineralization of the ryegrass with a C/N ratio only 12.4 (soil C/N ratios were 10.5, 10.0 and 10.2 respectively after the addition for Control, + 15 NH 4 + and + 15 NH 4 + + Ryegrass treatments). The total nitrification rate and N release would get reduced if organic material with a higher C/N than ryegrass was applied 27 .
But the use of organic material with a wider C/N ratio (e.g. crop straw) may not really favor transformation of NH 4 + into organic N pool instead of NO 3 − pool, due to the desynchrony between slow release of available C from crop straw and fast nitrification of applied NH 4 + in greenhouse soil. Therefore, reducing nitrification of applied fertilizer N by co-application of organic material in high-input agricultural soils should well consider the quality (e.g. C/N, carbon availability) of added organic material 22,26,27 . Incorporation into MBN and SOM. Our study dynamically evaluated the stability and further transformation of microbially assimilated N using the indicator "non-extractable organic 15 N". The results suggest that co-application of green manure led to effective NH 4 + immobilization and stabilization in soil (Fig. 3). During day 1-120, 23-28% of the applied fertilizer N was incorporated into non-extractable organic N, initially primarily as MBN, and eventually primarily as SOM. According to the low soil NH 4 + concentration after day 3, and the low microbial assimilation to NO 3 − in agricultural soils 7,28 , we propose that the slow decline of 15 N recovery in non-extractable organic N pool ( 15 N incorporated into MBN and SOM) after day 3 was dominated by  + +Ryegrass) to the tested soil. 15 N recovery in SOM pool was calculated as the difference of recoveries in "non-extractable N" and "microbial biomass N + mineral fixed N". Nitrogen loss is the unrecovered 15 N neither in extractable nor non-extractable pools. Numbers in the areas are the corresponding 15 N recoveries in relevant N pool at the related incubation time.
the remineralization of newly immobilized organic 15 N (Fig. 3). Our results also show that newly immobilized organic 15 N is a relatively stable N pool that can accumulate rapidly but is not readily accessible to microbial mineralization.
Studies of soil organic matter after 15 NH 4 + or 15 N-urea addition by nuclear magnetic resonance technique (NMR) confirm that the immobilized organic 15 N is concentrated as the form of peptide or amide 17,29,30 , which is regarded as a microbial source 19,31 . In this study, the ryegrass-induced rapid up-down change of 15 N recovery in MBN (Fig. 1h), as well as the trade-off changes of 15 N recoveries in MBN and "incorporated into SOM" pools ( Fig. 3), indicate that the initial immobilization of 15 NH 4 + and its subsequent transformation in the soil should be mainly microbially mediated [32][33][34] , and the disappeared MB 15 N was mainly further incorporated into SOM, a stable soil organic N pool that hard to decompose 35 . Appel et al. 36 confirmed that almost no extractable 15 N was extracted from soils when 15 N-labelled bacterial biomass was added prior to extraction. This goes against usual understanding that the newly immobilized N will release to bioavailable N after the death and breakdown of live microbes. It is expected that the soil N immobilized by microbes may contribute a lot to the low but long availability of residual fertilizer-N (residual effect) to subsequent crops 37,38 .
The biotic immobilization of added NH 4 + was mainly initiated and regulated by active microorganisms 39 . MB 15 N was too variable within a short period 26 as also evidenced by this study to reliably reflect the scale of 15 NH 4 + immobilization, especially in a relatively long period. In contrast, newly immobilized organic 15 N, composed of 15 N incorporation into both MBN and SOM, can be fast accumulated but slowly mineralized in the long-term process (Fig. 3), and may be a better indicator versus MB 15 N to more reliably quantify the microbial assimilation of added 15 NH 4 + in a long period. Immobilization of N in soils might also be a result of abiotic reaction 40,41 , but the proportion is generally low in agricultural soil 42 and in our study soil (Fig. 1j). Nevertheless, no matter how much the immobilization of applied 15 NH 4 + is biotically or abiotically regulated, the evidence provided by the present study suggested that applying green manures (e.g. ryegrass) to soil could significantly enhance the stable immobilization of applied NH 4 + . This is very significant for N management in intensive cultivation systems (e.g. greenhouse cultivation, orchard, vineyard, etc.), where large amounts of N fertilizer are applied. (Figs 1 and 3), which was consistent with previous studies 26,43,44 . Because most of EON in soils is resistant to decomposition, rapid turnover of easily degradable EON released from fresh organic matter will lead to the low contribution of added organic material to the total EON pool 44 . With the decline of the recovery of MB 15 N, the EO 15 N recovery kept continuously low (Fig. 3), reflecting indirectly a low contribution of the disappeared MBN to EON during the 120-day incubation. The content of clay-fixed ammonium depends on the degree of K saturation of the interlayers of 2:1 clay minerals 45,46 . Low 15 N recovery in MFN pool in this study may be due to the block of sufficient K + and NH 4 + resulting from high rate fertilization of K and N in greenhouse vegetable cultivation process. Available K in the greenhouse soil is up to 201.7 ± 2.8 mg kg −1 , being much higher than that in nearby cropland soil (74.4-127.0 mg kg −1 ) 35,47 Still, low moisture change during the first three days of the incubation may also contribute a part to the low fixation as soil dry-wet cycle is also thought to be an important factor controlling NH 4 + fixation 45 . Co-application of ryegrass might increase the loss of applied fertilizer NH 4 + (Fig. 3), in the forms of nitrogenous gases (NH 3 , N 2 , N 2 O, NO, HONO, etc.). The quantity of loss via ammonia volatilization in the soil was low (measured by absorption with dilute H 2 SO 4 , Kjeldahl distillation and titration) and the difference was not significant among treatments (< 0.5 mg N kg −1 during the 120-day incubation period, detailed data was not shown here). Compared to the treatment with only NH 4 + addition, we suggest that denitrification was fueled by available C from ryegrass and thereby increased the loss of gaseous N. Also, fast ryegrass mineralization rapidly stimulated aerobic microbial activity, likely reducing the permeability and availability of oxygen in soil pores to increase denitrifying loss of 15 NO 3 − in the very initial period, which was verified by the observation of lower water perviousness (need more time to completely infiltrate) in the soil with ryegrass addition than that without ryegrass addition when we added deionized water at regular intervals.

Soil and ryegrass samples preparation. In the present study, Fluvo-aquic soil (silty loam, Hapli-Udic
Cambisols in Chinese soil taxonomy) samples were collected from the surface layer (0-20 cm) of a greenhouse field planted with pepper (Capsicum annuum L.) in Damintun town, Xinmin county, Liaoning province (122°50´E, 41°59´N). The greenhouse soil had a 7-yr history of vegetable cropping (cucumber, tomato, pepper, etc.), and before that it was under maize cultivation for decades. The soil had a low C/N ratio of 10.5, and a high NO 3 − accumulation (152 mg N kg −1 , Table 1). In this study, ryegrass (Lolium multiflorum Lam.) samples was applied as a green manure in the greenhouse soil, which was obtained by culturing ryegrass in pots (21 cm in diameter and 20 cm in height filled with 3.5 kg soil) until it matured. Each pot was applied with 1 g of non-labelled urea.
Experimental design. Three treatments were set with 4 replicates for each treatment: (1) Control, soil without additions; (2) + 15 NH 4 + , adding as 15 N-labelled ammonium sulfate ((NH 4 ) 2 SO 4 ); (3) + 15 NH 4 + + Ryegrass, adding as 15 N-labelled (NH 4 ) 2 SO 4 and non-labelled ryegrass. 15 N abundance of (NH 4 ) 2 SO 4 was 50.17 atom%, and (NH 4 ) 2 SO 4 was added at a rate of 80 mg N kg −1 dry soil, being equivalent to the addition of 200 kg N ha −1 into the 0-20 cm soil layer in the field. The ryegrass was smashed to powder and sieved through 0.5 mm before use, and added at a rate of 4.21 g kg −1 dry soil. Selected properties of the soil and ryegrass were listed in Table 1.
For the laboratory incubation, soil was sieved through 2 mm and mixed homogeneously after gravimetric soil moisture was lowered to approximately 15%, and then pre-incubated at 25 °C for 7 days. Thereafter, 25 g soil (on an oven-dried basis) was weighed into polyethylene plastic centrifuge cups (60 mm in diameter and 81 mm high, n = 64 for each treatment). After mixing the soil thoroughly with ryegrass for the + 15 NH 4 + + Ryegrass treatment, aqueous solution of ( 15 NH 4 ) 2 SO 4 or deionized water was added over the soil surface with a pipette to bring the gravimetric soil moisture uniformly to 25.3% (corresponded to 60% water filled pore space). The application was made drop-wise to the soil in the cup and the soil then mixed to ensure the application was even. Finally, the soil was mechanically compressed to field bulk density (1.25 g cm −3 ) using a small disc with a handle. All cups were covered with parafilm punctured with a needle to maintain aerobic conditions and incubated in an automatically controlled incubator (25 °C). Distilled H 2 O was added at regular intervals (4 days) to maintain the soil moisture and prevent from decreasing the microbial activity due to water limitation during the 120-day incubation.
Sample processing and chemical analysis. Approximately 0.1 (2.5 h), 1, 3, 7, 15, 30, 60 and 120 days after adding ( 15 NH 4 ) 2 SO 4 , eight replicates of each treatment were destructively sampled, extracted directly (four cups) or after fumigation (four cups) with 100 ml 0.5 mol l −1 K 2 SO 4 for 30 minute on a shaker (220 rev min −1 ) (Fig. 5). The fumigation was carried out with alcohol-free chloroform in a desiccator lined with wet filter paper at 25 °C for 24 h in the dark 48 . After centrifugation (5000× g, 10 min), the supernatant was filtered through cellulose acetate membranes (0.45 μ m, Xinya, Shanghai) that had been prewashed with 0.5 mol l −1 K 2 SO 4 . The extracts were determined immediately or stored at − 20 °C until future analysis. The soil residue was extracted two more times with K 2 SO 4 solution (100 ml each, supernatant was discarded) to remove extractable N thoroughly from soil particles. All residual soil was carefully recovered, air-dried, ground, and sieved (< 0.15 mm).
Soil extracts (20 ml) were distilled after the addition of MgO and Devarda's alloy respectively for quantifying the content of NH 4 + -N and NO 3 − -N 49 . A modified micro-diffusion procedure using Teflon strips was performed to prepare samples for analysis of 15  Extractable organic N (mg kg −1 ) 8 7760 Extractable N/TN (%) 11.7 28.9 Available K (mg kg −1 ) 201.7 - Table 1. Chemical properties of the tested soil and ryegrass. ξ 1:2.5 (soil: deionized water). § Extractable C and N were extracted by 2 mol l −1 KCl (1:5 for soil and 1:20 for ryegrass, w/v).
Concentrations of non-extractable N (NEN) and mineral fixed NH 4 + -N (MFN), and their 15 N abundances in residual soil after extraction mentioned above were determined by the EA-IRMS. Organic N in residual soil was removed by excessive fresh prepared alkaline KOBr solution before determining MFN 52,53 . Soil and ryegrass total organic carbon (TOC) was measured using the solid module of TOC/TN analyzer (Multi N/C 3100, Analytikjena, Germany). Soil and ryegrass total N were determined by the Kjeldahl method. Soil available potassium (K) was determined by a flame photometer (6400A, Shanghai) after extraction with 1 mol l −1 ammonium acetate (v/w = 10:1).
Calculation. The soil microbial biomass nitrogen (MBN) was quantified using the following formula with a K EN = 0.45 54  where, TEN (total extractable N) = NH 4 + -N + NO 3 − -N + EON. Letters of "f " and "uf " respectively mean "fumigated" and "unfumigated"; N _x is the amount of soil N components (NH 4 + -N, NO 3 − -N, EON, MFN or NEN, mg kg −1 ); 15 N_ x is the amount of 15 N (mg kg −1 ) from labelled ammonium sulfate added in related N pool; _ APE 15 N x , APE bg and APE AS are the atom percent excesses (%) of N component in related N pool, background and ammonium sulfate added, respectively; Recovery x is the percentage of the 15 N tracer recovered in the labelled N pool; Input 15 N is the amount of 15 N in ammonium sulfate added (mg kg −1 ).
Net nitrification rates were calculated as the difference between final and initial NO 3 − -N concentrations for three time intervals, days 0.1-1, days 1-3 and days 3-7 .
Statistical analyses. Data are expressed on oven-dried soil basis, and subjected to analyses of variance using SPSS for Windows v16.0 software package (SPSS Inc., Chicago, IL, USA). All figures were obtained from SigmaPlot 12.5. Error bars in figures represented standard errors. The differences between means of variables for different treatments were statistically tested by ANOVA procedure with least significant difference (LSD) test.   N components (ammonium, nitrate, organic-N), microbial biomass N, non-extractable N and mineral fixed N in tested soils.