Effect of straw and inhibitors on the fate of nitrogen applied to paddy soil

A pot experiment was used to explore the distribution of fertilizer N and agronomic effects in a paddy soil-rice (Oryza sativa L.) system. Five treatments were set: without nitrogen, straw and inhibitor (C), urea (U), urea + straw (US), urea + urease + nitrification inhibitor (UI) and urea + urease + nitrification inhibitor + straw (UIS). Soil and urea-derived microbial biomass N increased significantly in US and UIS compared with straw-free treatments at seedling and tillering, indicating that biotic process play an important role in the retention of fertilizer N with straw addition. About 10% urea-N was recovered as fixed ammonium (FA) at seedling stage, subsequently released at tillering and maturation regardless of treatments, which emphasizes the importance of FA in the retention and supply of fertilizer N in paddy soils. Compared with U, rice yield and N uptake in US decreased by 7.8% and 25.2% respectively, while inhibitors (UIS) alleviated the reduction by 16.4% and 31.6%. The current research indicated UIS is recommended as the most appropriate management strategy in paddy soils of Northeast China based on N dynamics. But the economic effect as well as the field-scale validation need to be further evaluated.

Rice (Oryza sativa L.) cultivation in China covers about 3.02 × 10 7 ha, and overall yield reached 2.07 × 10 8 t in 2018, accounting for 50% of Chinese grain production 1 . As the most important nutrient element, an adequate supply of nitrogen (N) contributes to more than half of the rice yield 2 . Over-application of N, however, results in low use efficiency and high N losses, and causes environmental problems 3 . An input of fertilizer N can be absorbed by rice directly or temporarily immobilized by biotic and abiotic processes and subsequently released to meet crop demand. These processes are influenced by different management practices such as the addition of urease and nitrification inhibitors, and return of organic materials 4,5 .
Improving immobilization of fertilizer N is an effective way to retain N and reduce N losses with the addition of decomposable carbon (C) 6 . Previous studies have indicated that available C inputs temporarily enhance microbial immobilization of N, resulting in an increase in soil microbial nitrogen (MBN) 7 ; this process is closely correlated with biotic processes. Ma et al. 7 indicated that MBN consistently increased with inputs of the nitrification inhibitor DMPP (3,4-dimethylpyrazole phosphate), whereas the urease inhibitor NBPT (N-butyl thiophosphotriamine) increased MBN in the absence of glucose but decreased it in the presence of glucose. Rice paddy soils show a different response pattern due to flooded conditions: Wang et al. 8 found that flooded soil has a lower immobilization rate of N into microbial biomass because of the loss of ammonium N from volatilization and/or greater NO 3 − leaching after nitrification, but straw return generally stimulates immobilization [9][10][11] , and consequently changes the way fertilizer N is conserved and supplied [12][13][14] . With the addition of urease inhibitor and nitrification inhibitor, the forms of inorganic N are regulated in soil, and under flooded conditions, the persistence of fertilizer NH 4 + influences the retention and distribution pattern of N in soil, especially combined with the application of straw. Currently, however, little information is available regarding the scenarios mentioned above.
Fertilizer NH 4 + can also be rapidly fixed by soil clay minerals, especially 2:1 clay minerals. More than 70% of applied fertilizer N can be retained as fixed NH 4 + in some arable soils 15,16 . Fixed ammonium (FA) is an active Scientific Reports | (2020) 10:21582 | https://doi.org/10.1038/s41598-020-78648-w www.nature.com/scientificreports/ abiotic N pool and more than 80% of recently FA can be released in subsequent growing seasons 17 . Increasing fixed NH 4 + is another way to build a pool of N that will later become available to crops, improving fertilizer recovery and minimizing N losses 18 . In flooded paddy soils, how the FA pool responds to the addition of inhibitors and straw remains to be evaluated.
Although the response of crop yield and N uptake to inhibitors is variable, a meta-analysis concluded that the addition of inhibitors caused the yield and N use efficiency to increase by an average of 7.5% and 12.9%, respectively 19 . The greater persistence or continuous supply of NH 4 + with the addition of nitrification and urease inhibitors was reported and responsible for the improved agronomic effect. Other research indicated that declining yield was found in continuously cropped irrigated rice systems, where a significant portion of the rice residue is returned 3 . An inadequate N supply in the late season was suggested to be responsible for the declining yield when rice residue is returned 20 . Eagle (2000) also indicated that the decrease in N uptake does not result from a reduction in the capacity of the root system to acquire N from soil 20 . The strong retention of fertilizer N in different soil N pools with the application of straw is probably responsible for the inadequate supply of N to rice. It is therefore necessary to evaluate the fate of fertilizer N and its retention in different pools, with the concomitant use of inhibitors and straw application, and observe how this retention affects its uptake by rice.
Based on the analysis above, a pot experiment was set up to clarify the impacts of inhibitor and straw additions on the dynamics of fertilizer N in a paddy soil-rice system by using 15 N to trace the fate of applied N. We hypothesized that: (1) Straw decreases N uptake and crop yield, stimulates microbial growth and increases the ability of organic nitrogen pools to store fertilizer N; (2) Inhibitor application improves retention of urea-N through abiotic process in the absence of additional C; (3) Inhibitor application with straw alleviates the decrease of N uptake and crop yield caused by straw because of the persistence of NH 4 + under the combined application of inhibitors.

Results
Total soil N and retention of urea-N. Total soil N did not show any significant differences with either application of inhibitors or straw return in the current research (P > 0.05). Only the combined application of inhibitors and straw (UIS) increased soil total N at the rice maturation stage compared with the other three treatments (Table 1). In contrast, different treatments and sampling times showed significant effects on recovery of urea-N (Table 1). UI decreased N recovery rate by 18.3%, 28.4%, and 39.4% for urea-N at the maturation stage compared with U, US and UIS (P < 0.05). Straw return promoted the retention of urea-N in soil, especially at the tillering stage, which is the rice growth period during which most N is taken up. Combined application of inhibitors and straw (UIS) alleviated the retention of urea-N in soil and decreased urea-N recovery rate by 41.29% compared with US at the tillering stage.
Dynamics of soil N and fate of urea-N in NH 4 + -N, MBN, FA, DON pools. In general, urea-derived NH 4 + -N decreased with rice growth for all treatments except UIS, which increased initially and then decreased. At the seedling stage, after 7 days of fertilization, U and UI treatments showed a significant higher 15 NH 4 + content than US and UIS treatments (P < 0.05). Compared with urea only (U), urea-derived NH 4 + in US, UI and UIS significantly decreased by 57.8%, 26.5% and 64.5%, respectively, at the seedling stage (P < 0.05) (Fig. 1A). At tillering, just after the top dressing fertilization, the lowest amount of fertilizer-derived NH 4 + was found in the U treatment while the highest recovery rate was found in the UIS treatment, which was 20 times higher than the U treatment and 23.5% higher than the US treatment (Fig. 1A, Table 2). There were no significant differences in NH 4 + and 15 NH 4 + content at the end of plantation among treatments (Fig. 2, Table 2, P > 0.05). Soil microbial biomass N (SMBN) and fertilizer-derived N in microbial biomass are shown in Fig. 1B; uptake by microbial biomass is a biotic process by which soil can retain the added N. The treatment effect was stronger than the effect of sampling stage for this parameter (Fig. 1B). It is apparent that, compared with U, straw addition (US, UIS) significantly increased the recovery of urea-N in the microbial biomass N pool, as the urea-derived MBN increased 77.9% and 122.7% in the seedling stage and 191.9% and 269.6% at tillering (Fig. 1B, Table 2). At maturation, however, there was no significant difference among treatments for MBN and MB 15 N, except that the U treatment exhibited 20.35-27.92 mg N kg −1 higher MBN than other three treatments (Fig. 1B). It is emphasized Table 1. Soil total nitrogen, soil urea-derived nitrogen and the recovery of urea-N in soils at three sampling stages. U: 15 N labeled urea; US: 15 N labeled urea + straw; UI: 15 N labeled urea + inhibitor; UIS: 15 N labeled urea + inhibitor + straw. The values in the table represent the mean of three replicates. Different capital letters represent significant differences of the same treatment at different sampling times; different lowercase letters represent significant differences among treatments during the same sampling period (Duncan, P < 0.05).    Table 2). The fixed ammonium (FA) pool is correlated with abiotic process. Generally, both total soil FA and ureaderived FA exhibited the pattern of seedling > tillering > maturation (P < 0.01, Fig. 1C), with the highest FA recovery of 10% at the seedling stage ( Table 2), indicating that the soil had a strong ability to fix N at first and this pool was released thereafter. At seedling, UI showed the highest FA and the treatments without straw (U, UI) had higher FA contents 9.8% and 31.1% than that in the treatments receiving straw (US, UIS). Soil FA decreased sharply to 12.61-27.87 mg N kg −1 at tillering for all treatments and U exhibited the lowest value with 30.42 mg N kg −1 . FA decreased further at maturation stage except in the U treatment. The amount of urea-derived The numbers on the arrows are normalized path coefficients. The thickness of the arrow is proportional to the size of the path coefficient. Continuous arrows and dotted arrows indicate positive and negative correlations, respectively. Asterisks after data indicates a significant difference (*P < 0.05, **P < 0.01, *** P < 0.001). U: 15 N labeled urea; US: 15 N labeled urea + straw; UI: 15 N labeled urea + inhibitor; UIS: 15 N labeled urea + inhibitor + straw. FA stands for fixed ammonium derived from urea; MBN stands for microbial biomass nitrogen derived from urea. DON represents soluble organic nitrogen derived from urea; TN represents total nitrogen derived from urea sources; NH 4 + -N denotes ammonium nitrogen derived from urea. TDN represent total dissolvable nitrogen derived from urea. Some variables have no arrows because in the operation of maximum likelihood number method, the route with relatively small path coefficient is removed in order to make the model more fit and more in line with the fitting index described below in the figure.
Scientific Reports | (2020) 10:21582 | https://doi.org/10.1038/s41598-020-78648-w www.nature.com/scientificreports/ N conserved in FA did not show any significant difference at seedling and tillering in different treatments (P > 0.05), and at maturation, the addition of inhibitors (UI, UIS) led to greater release of urea-derived FA and showed lower values compared with U and US (Fig. 1C). Another N pool which is correlated with both abiotic and biotic process is dissolved organic N (DON). The recovery of urea-derived DON was not significantly different among treatments at all three sampling periods, except the US treatment at tillering, which increased 3.82 times relative to U, reaching 4.67 mg N kg −1 (P > 0.05, Fig. 1D, Table 2). For total DON, different treatments showed contrasting changes. US decreased at the seedling stage and increased at tillering compared with U, and UI increased significantly at both seedling and tillering. UIS changed only at tillering stage, which increased 2.38 times compared with U (Fig. 1D). Table 3, rice yield and biomass in the US treatment was significantly reduced compared with the other treatments, and the loss of urea-N was significantly higher than the other treatments. Inhibitor addition together with straw (UIS) alleviated the reductions in yield and rice biomass caused by straw only (US). These two treatments with straw application permitted greater retention of urea N in the MBN pool. Compared with C, the U, UI and UIS treatments significantly increased rice biomass by 20.0%, 24.7% and 19.8% respectively (P < 0.05), and rice yield increased by 8.2%, 21.9% and 16.3%, respectively. Furthermore, the UI treatment achieved the highest rice yield, biomass, panicle N content and total fertilizer N uptake with only 30% fertilizer N loss.

Discussion
Effect of straw return on the fate of urea-N. Hypothesis 1 was confirmed by the current research. The lowest N uptake and crop yields were found in the US treatment (Table 3), which is in accordance with previous research conducted by Pan et al. 17 and Ma et al. 21 . In irrigated rice, grain yield is closely associated with N uptake when the availability of other nutrients is adequate and pest damage does not limit crop growth. Straw decomposition and utilization by microorganisms led to a state of N limitation in the soil due to the high C/N ratio of straw, and microorganisms competed for N with plants, especially in the critical period of crop growth 21 . The immobilization of urea-N (MBN) was enhanced by 2.53, 7.78 and 2.41 times, respectively, at all three sampling stages compared with U (Fig. 1B), and there were significant correlations among urea-derived MBN and DON (P < 0.01, Fig. 2), in which straw return showed the main effect on soil MBN and DON at seedling and tillering (P < 0.05) and sampling time had a significant influence (P < 0.05, Table 4). This is likely one of the reasons for the decrease in crop yield and N uptake. Devevre and Horwath 22 indicated that in addition to microbially mediated biotic process, chemically mediated abiotic process are also responsible for the retention and supply of fertilizer N in paddy soil systems. They postulated that fertilizer-derived organic nitrogen accumulated in soil undergoes partial polymerization and enters the soil humus, resulting in the decrease in N availability. We observed that Table 3. Rice yield, biomass, nitrogen content assimilated from urea and the fate of urea-derived N at harvest (% of applied N). C: no nitrogen, inhibitor and straw; U: 15 N labeled urea; US: 15 N labeled urea + straw; UI: 15 N labeled urea + inhibitor; UIS: 15 N labeled urea + inhibitor + straw. The values in the table represent the mean of three replicates. Different lowercase letters represent significant differences among treatments during the same sampling period (Duncan, P < 0.05).   www.nature.com/scientificreports/ urea-derived DON was significantly higher at the earlier stages of sampling in the current study. Although we did not measure SOM fractions, we assumed the increased DON would help form stabilized humic substances with the help of C existing in straw. Olk et al. 23 also suggested that the accumulation of recalcitrant rice straw products, such as lignin-derived phenols, was responsible for reducing soil N availability. They theorized that stabilization of N into humic fractions might be the dominant process affecting the availability of fertilizer N under the combined application of straw. Despite the immediate negative effect of straw application for rice growth and N uptake, the long-term effects of N retention induced by plant residue addition appear to be positive. Bird et al. 24 examined the effects of residue management on N immobilization and uptake by rice plants. They observed that although N fertilizer efficiency fell after the return of rice straw in the short term, 4-6 years of straw addition produced a labile pool of available N, which led to a reduction in N fertilizer dependency by the rice plant in subsequent seasons. It should be noted that a large part of urea N was fixed in the form of FA, and FA decreased significantly over time, indicating that the release of FA can contribute to the uptake of N by rice (Fig. 1C, Table 2). This result was not in agreement with Gouveia and Eudoxie 25 and Akter 26 , who indicated that under anaerobic conditions, the surface of clay minerals was covered by iron oxides, which limited the diffusion of NH 4 + into the interlayer space, weakening NH 4 + fixation by clay minerals. The reason for the discrepancy can be attributed to the variety of parent materials in different soil types. Alfisols originate from residual deposits, slope deposits and some loessial materials with more montmorillonite and illite, abundant 2:1 clay minerals that favor the fixation of NH 4 + in soil 27 . In the present experiment, structural equation modeling also indicated that extractable NH 4 + was significantly correlated with FA (the pathway coefficient was 0.83, P < 0.01) (Fig. 2 US). Recently, several researchers have indicated that FA is an important N pool for the retention of urea-N in soils similar to that used here 7,27 , and the newly formed FA can be later released and afford available N for plant growth 28,29 . The current study confirmed this process. Other research has indicated that with addition of glucose, competition for NH 4 + between microbial immobilization and mineral fixation was intensified and a larger proportion of urea-N was found in the organic N pool, reducing the effect of FA on urea-N conservation 21,30 . In our experiment, however, the decrease in FA did not necessarily exhibit any relationship with MBN (Fig. 2); other research has also shown that the addition of wide C/N ratio organic material does not necessarily decrease FA 31 . In addition to the various types of soils tested and their mineral composition 32 , the discrepancies among the studies were induced by differences in rate and availability of the C addition 33 . The current research emphasized the importance of NH 4 + fixation in the early season and N release later in the season for the growth of rice. Reserves of N in FA can be a substantial source of N for plant uptake and other soil processes, and should be given greater attention in future work.
The N unaccounted for in the US treatment was significantly higher than in other treatments (Table 3), which is in disagreement with previous research 3,22 , but in accordance with the research conducted by Phongpan and Mosier 34 and Wang 35 . The contrasting results from different trials are probably attributable to differences in soils and different C/N of straw. Loss of N through surface runoff did not occur in this study, and the pot experiment also avoided losses through percolation 3 . Therefore, ammonia volatilization and gases produced during nitrification-denitrification are responsible for the N loss 36 . Zaman and Blennerhassett 37 indicated that decomposition of the incorporated straw triggered the release of base cations as well as increased floodwater pH, which promoted NH 3 volatilization. Moreover, total dissolved nitrogen (TDN) and urea-derived TDN was higher in the US treatment (data not shown), which increased the risk of N volatilization loss through the water, and is probably the reason why unaccounted-for N is higher in the US treatment. Zhong (2017) 38 indicated that the percentage of N loss by ammonia volatilization increased with the nitrogen application rate. 48.77% loss as ammonia volatilization was reported when 225 kg N ha −1 was applied while 58% loss was found when 375 kg N ha −1 was applied. We did not measure ammonia volatilization in the current research, but still can speculate that the volatilization loss of N was lower than the experiment which applied more N to soil.

Fate of urea-N under the application of inhibitors. Application of inhibitor with urea (UI) showed
the highest rice yield and N uptake as well as lowest unaccounted-for N among all the treatments (Table 3). Nitrification and urease inhibitors have been proposed as a method to improve N use efficiency and reduce N losses 39,40 . Although the response of crop yield to inhibitors is variable, a meta-analysis concluded that the addition of inhibitors caused yield and N use efficiency to increase by 7.5% and 12.9%, respectively 19,29,41 , which is in accordance with our research. Zhang et al. 42 and Tobias et al. 43 indicated that urea hydrolysis can be delayed 3-7 days with the application of urease inhibitor. At the same time, nitrification inhibitors regulate the existing forms of extractable N, allowing a better synchrony between N supply and rice demand 44 . In our experiment, we observed relatively high urea-derived NH 4 + at the seedling stage in the presence of inhibitor, and a decrease at tillering, but the decrease was not as strong as in the U treatment (Fig. 1A), which confirmed the regulatory function of inhibitors as well. This is beneficial for rice uptake of N to promote higher yield. MBN in UI was not as strong as that in the treatments with application of straw (US and UIS) at seedling and maturation (Fig. 1B). This phenomenon can be attributed to the widespread view that there is less available C supply with the application of straw 7 . FA showed the same pattern of change as the other treatments, showing the highest value at seedling which decreased sharply over time (Fig. 1C). Compared with other treatments, the proportion that decreased is higher, indicating that newly fixed ammonium was more easily released in this treatment. According to the SEM analysis shown in Fig. 2, the urea-derived NH 4 + had a significant direct pathway to FA (the pathway coefficient was 0.95, P < 0.01). This phenomenon verified hypothesis 2, but only at the seedling stage of rice growth.
The recovery rate of urea-N in soil was significantly lower in UI than in other treatments, and loss (unaccounted-for N) was lower as well, but uptake of urea N by plants was significantly higher (Tables 1, 3). Basically, urea-N can be partitioned into three fates after it is applied to soil: retained in soil, taken up by plants and lost Scientific Reports | (2020) 10:21582 | https://doi.org/10.1038/s41598-020-78648-w www.nature.com/scientificreports/ from the soil-plant system. As mentioned earlier, the possible loss pathways in this study research are ammonia volatilization and nitrification-denitrification. Although a lot of work has indicated increased ammonia volatilization under waterlogged conditions of paddy soil with the application of NI 45,46 ; our experiment still showed a decreased N loss. This can likely be attributed to decreased loss of gases during nitrification and denitrification. More importantly, application of inhibitor made urea-N transformation characteristics better match the demand of rice growth. The advantages of inhibitor amendment were demonstrated in the current research, i.e. increased rice yield and decreased N loss, which implies better protection of the environment. But the disadvantage of inhibitor amendments was also evident: the lower retention of fertilizer N in soil may cause a stronger dependence on fertilizer N in rice-soil systems for ensuring favorable long-term yields.

Fate of urea-N under application of inhibitors and straw.
The UIS treatment exhibited significantly higher urea-N recovery compared with UI or any other treatment at maturation, and was also higher than UI at tillering (Table 1), which indicated the good effect of combined application of straw on N retention and soil fertility improvement, where a positive interaction was evident between them on soil N retention at tillering (P < 0.05). Yu et al. 27 also indicated that in the presence of glucose, microbial immobilization was the principal mechanism of fertilizer N retention, and this phenomenon was mitigated by urease inhibitor. Our results were also consistent with Ma et al. 7 , who inferred that the application of straw together with inhibitor promotes fertilizer N retention in soil compared with inhibitor application alone. This is in contrast to the results of Wang et al. 47 , who observed that inhibitors decreased N immobilization when soil was amended with barley straw. The difference may be attributed to the contrasting C/N ratio of straw, different soil type and planting system, and the fertilizer regimes; according to Table 4, sampling times had a significant influence on recovery of urea-N in soil pools except for FA. When compared with US, UIS alleviated the decrease in crop yield and N uptake of urea-N (Table 3), which verifies the proposed hypothesis 3, and is in accordance with the results published by Ma et al. 21 . He proved that it is urease inhibitor, not nitrification inhibitor, which alleviates the decrease in crop yield induced by straw application. The delayed effect of urease inhibitor for appearance of NH 4 + (and presumably reduced NH 3 emission) in UIS gave rice roots more opportunity to catch and absorb N, which led to higher uptake of N and higher crop yield 21,48 . FA showed the same pattern as the other treatments ( Fig. 1, Table 2), which also emphasizes the importance of FA in the earlier growth stage and release of FA in the later growth stage for the retention and supply of N in soil 30,49 . Future experiments, particularly field experiments, will further test this effect.

Materials and methods
Soil collection. Soil samples were obtained from a paddy field to a depth of 20 cm at the University of Shenyang Agriculture Rice Institute Experiment Site, Northeast China (41°31′ N, 123°24′ E), which had received no fertilizer in the past 20 years. The soil type was an Alfisol before it was changed to rice cultivation and represents the main soil type for agricultural production in the region. The field-moist soil was passed through a 5 mm sieve to remove stones, crop residues and roots. The basic soil physical and chemical properties were as follows: the pH (H 2 O) was 6.18, soil organic matter content was 26.1 g kg −1 , total N content was 1.95 g N kg −1 , total P and K were 2.18 g kg −1 and 22.4 g kg −1 respectively. Soil CEC was 22.3 cmol kg −1 , and clay, silt and sand were 16.5%, 55.3% and 28.2% respectively. Set-up. An outdoor pot experiment was conducted with rice in a net house from May to October, 2018. Five treatments were set up in a completely randomized design: (1) C (no nitrogen); (2) U ( 15 N-labeled urea); (3) US ( 15 urea + rice straw); (4) UI ( 15 urea + urease inhibitors + nitrification inhibitor); (5) UIS ( 15 urea + urease inhibitors + nitrification inhibitor + rice straw). A total of 45 pots (5 treatments × 3 sampling times × 3 replicates) were used, with 3 kg air dried soil per pot. The 15 N-labelled urea (10.02% 15 N at%) was applied at a rate equivalent to 318 kg N ha −1 in three applications: 50% as a basal application, and 25% each as the first and second top dressing. Super-phosphate calcium and potassium chloride were applied at a rate of 212 kg P ha −1 and 318 kg K ha −1 respectively. Urease inhibitors (PPD-phenylphosphodiamine and NBPT) and nitrification inhibitor (DMPP) were applied at rates of 1%, 1% and 2% of applied urea-N (w/w). Rice straw with a C/N ratio of 63 (C and N contents were 37.8% and 0.59%, respectively) was ground into powder with a pulverizer. The application amount of straw was 10.6 t ha −1 (5 g kg −1 ). The basal fertilizer (with or without inhibitor), straw and soil were mixed thoroughly into the pots on May 25 and left overnight under flooded conditions. Rice was transplanted into pots the next day. Fertilizer (with or without inhibitor) was dissolved in water and injected into the soil with a syringe when topdressing on June 18 and August 20.
Rice (Meifeng 9) with five or six fully expanded leaves was grown in plastic pots (d = 18 cm, h = 25 cm) under no-leached conditions. Crop management was consistent with local management practices for conventional cultivation, in particular maintaining 3-5 cm floodwater until 1 week before harvest. Soil and rice plants were sampled with a five-point method after 7 days fertilization at the seedling stage (June 4), and then at the tillering (June 25) and maturation stage (September 21). Soil analysis. Soil pH was determined by a glass electrode pH meter using a soil:water ratio of 1:2.5 50 , Total organic C content of soils was determined using a C and N analyzer (Vario TOC Analyzer, Elementar, Germany). Soil total P and K was determined by sodium carbonate fusion and the molybdenum antimony-ascorbic acid colorimetric method 51 . The CEC was determined using a modified NH 4 -acetate compulsory displacement method 52 . Soil particle size distribution was determined by pipette method 41 . Ammonium was determined by extracting a 10 g soil subsample with 100 mL of 2 M potassium chloride (KCl). Samples were shaken for 1 h on a reciprocal shaker, filtered and analyzed on a continuous flow analyzer (AA3, Bran + Luebbe, Germany) 53 14 . Total soil N was analyzed with an elemental analyzer after the soil was air-dried and ball-milled to a fine powder. Three pots of crop were taken at maturation stage to calculate biomass and then separated to stem and panicle to get yield and panicle N uptake. Cleaned rice plant samples were oven-dried at 65 °C to constant weight and ground for analysis of total N analysis and 15 N content. The atom% 15 N of the NH 4 + pool was determined as described by Sebilo et al. 58 . Briefly, the extracted solutions were first transferred to glass vials with caps. Filter packages for the NH 4 + diffusion technique were prepared as follows: KHSO 4 (10 μL, 2.5 M) was pipetted onto a strip of glass fibre filter (APFE, 25 mm, Millipore), and the filter was then wrapped and enclosed with a hydrophobic filter ("Mitex", PTFE, 47 mm diameter, Millipore) to form the filter package. After the addition of MgO (0.25 g) and the filter package to the samples, the vials were closed immediately and the solution was stirred slowly for one week at room temperature. After one week, the filter package was removed from the vial, dried in a freeze dryer for 24 h and the glass fibre strip recovered for atom% 15 N analysis. For measurement of 15 MBN, 4 mL fumigated and unfumigated subsamples were digested to transform N to NO 3 −53 , then 15 NO 3 − was measured with 0.25 g devarda alloy reagent (Merck, German) and a diffusion package method as described in Sebilo et al. 58 . The digested unfumigated glass fiber samples were considered as 15   The amount of urea-derived N in a certain N pool was calculated as follows 27 : where Ndfu is the amount N derived from urea; N soil N pool refers to the N content from different soil N pools; 15 N AT% excess soil N pool is the 15 N enrichment of different soil N pools; 15 N AT% excess urea refers to the enrichment of urea added. Recovery of urea-N in a specific N pool was calculated as follows: where Ndfu is the amount N derived from urea; N urea applied is the content of urea applied. Uptake of urea-N by rice (N r ) was calculated as follow: where N total in rice is rice total N content; % atom excess 15 N in rice is the enrichment of rice; % atom excess 15 N in urea is the 15 N enrichment of urea.
Ndfu mg kg −1 = N soil N pool × 15 N AT% excess soil N pool 15 NAT% excess urea Recovery(%) = N dfu N urea applied × 100 N r = N total in rice × % atom excess 15N in rice %atom excess 15N in urea