Introduction

Nitrogen (N) is one of the key elements required for crop growth. However, N fertilizer application to agricultural systems has had multiple negative effects on the environment (Sainju et al. 2008). In some natural ecosystems, reactive N is effectively conserved via the prevailing inherent soil N dynamics (Huygens et al. 2007; Rütting et al. 2008; Rütting and Müller 2007). In subtropical regions, which is characterized by the high precipitation, NO3− leaching losses occur readily via leaching or runoff, while NH4+ is less mobile (Huygens et al. 2007; Rütting and Müller 2007, 2008). Rates of autotrophic nitrification in native soils in the subtropical regions have been reported as low or absent due to low soil pH (Zhao et al. 2007), while rates of NO3− immobilization to organic N reported were relatively high (Zhang et al. 2013a), especially in low NH4+-N soils (Rice and Tiedje 1989). As a consequence, inorganic N in the native soils of these regions is dominated by NH4+, and available NO3− is thought to be conserved efficiently by NO3− consumption processes (Zhu et al. 2013b). However, when native land is converted to agricultural use, microbial immobilization of NO3− is substantially suppressed and nitrification is stimulated (Schimel and Bennett 2004), leading to NO3− dominance in inorganic N in the soil, with associated risks of N loss through runoff, leaching, and denitrification. Therefore, N retention in native subtropical soils is likely to be reduced when converted to agriculture (Han et al. 2012; Yang et al. 2010; Zhang et al. 2013a).

However, N transformations in these agricultural soils may differ under different water regimes. For example, Choi et al. (2003) showed that nitrification was stimulated following ammonium sulfate addition under unsaturated conditions, whereas under saturated conditions, immobilization and subsequent mineralization of organic N to NH4+ dominated. Nitrification in saturated soils is assumed to be very low, because of oxygen-limiting conditions. Additionally, dissimilatory NO3− reduction to NH4+ (DNRA) would become an important NO3− removal pathway, particularly under saturated conditions (Yin et al. 2002; Zhang et al. 2015a). During DNRA, N losses via leaching and gaseous emissions can be reduced by decreasing the size of the NO3− pool in soil and increasing the NH4+ pool, therefore providing additional NH4+ for immobilization or for uptake and assimilation by primary producers (Minick et al. 2016; Silver et al. 2005). Thus, comparing with unsaturated condition, low nitrification capacity with high DNRA rate in soils under saturated condition may lead to alleviate the potential risk of N losses by runoff or leaching.

The connection between moisture and nitrogen (N) transformation in soils is key to understanding N losses, particularly nitrate (NO3−) losses. The relationship between soil moisture content and net N transformation rates has been extensively studied by adjusting soil moisture content in the laboratory (Di et al. 2014; Guntinas et al. 2012; Wang et al. 2005; Yu and Ehrenfeld 2009). However, net N transformation rates are the outcome of several N cycling processes such as mineralization, immobilization, and nitrification. To quantify the response of inorganic N retention to soil moisture content, the gross rates of processes associated with NH4+ and NO3− cycling should be determined simultaneously. Additionally, few studies have been conducted with soils that have experienced a long-term difference in moisture status, as compared to an instantaneous adjustment, and this longer term legacy effect may influence soil microbial communities and in turn the N transformation dynamics.

The aim of this study, therefore, was to test the hypothesis that an inherent N retention mechanism exists in saturated, compared with unsaturated soil conditions, which reduces net NO3− production helping to keep the reactive N in the system. The study used the long-term field experiment in the subtropical region of China with a tobacco-rice cropping system incorporating an unsaturated tobacco-cropping season and a saturated rice-cropping season. The soil samples collected from tobacco season were incubated at 60% water holding capacity, while the samples collected from saturated rice-cropping season were incubated under saturated conditions, two 15N tracing treatments and a numerical modeling method was used to quantify gross N transformation rates. The 15N tracing model applied in this study has been tested extensively and used in studies on soils from orchard, grassland, paddy, vegetable, peanut, and maize fields, and for forest soils (Huygens et al. 2007; Müller et al. 2004; Rütting et al. 2008; Zhang et al. 2013b, 2015b; Zhu et al. 2011).

Materials and methods

Study site description

The long-term field experiment was established in 2008 in the Jiangle County, Fujian Province, China (26° 44′ 53″ N, 117° 26′ 48″ E). The site is characterized by a subtropical monsoon climate, with a mean annual temperature of 18.9 °C and mean annual precipitation of 1670 mm (over 30 years). The soil type is defined as an Anthrosol (WRB Soil Taxonomy), which has developed over granite bedrock. The annual cropping sequence was tobacco planted in February or March followed by rice planted in July. During the tobacco-cropping season, irrigation did not occur. However, during the rice-cropping season, the water level was maintained at 5 cm above the soil surface. The four treatments examined were (1) no fertilization applied (CK); (2) N, phosphorus (P), and potassium (K) fertilizers applied at the recommended rate (NPK); (3) N fertilizer applied at 50% above the recommended treatment and P and K at the recommended rate (NhPK); and (4) NPK with additional straw (NPKS), where rice straw was chopped and incorporated into soil by plowing at a rate of 3600 kg ha−1 (dry matter basis) for both crops (see below). Three replicate plots (7 × 4 m) of each treatment were established in a randomized block design, with plots separated by brick frames. Under the NPKS treatment, mean concentrations of N, P, and K in straw were 5.0, 1.6, and 20.5 g kg−1, respectively. Fertilizer rates and timings are given in Table 1. Compound fertilizers, urea, and potassium nitrate were applied as N sources for tobacco; ammonium bicarbonate and urea were applied as N sources for rice every year.

Table 1 Annual fertilizer application rates in the long-term field experiment over the period 2008–2013 (kg ha−1)
Full size table

Soil sampling

Soil samples were collected from the plow layer (0–20-cm depth) in July and October 2013, following the tobacco and rice harvest, respectively. Ten soil cores (5 cm in diameter) were taken from each plot and combined providing one composite sample per plot and three replicates for each treatment. Thus, 12 samples were collected following the tobacco harvest when the soil was unsaturated, and 12 additional samples were collected following the rice harvest when the soil was saturated. Soil samples were split into two subsamples, one of which was used to determine soil properties (Table 2) and the other stored at 4 °C in the dark prior to the laboratory incubation experiment.

Table 2 Properties of soils collected from the long-term field experiment after tobacco (unsaturated) and rice (saturated) harvesting in 2013
Full size table

Laboratory incubation

Soils from each field treatment collected at the end of the tobacco crop season were incubated under unsaturated conditions (at 60% water holding capacity (WHC)), while those collected following the saturated rice-cropping season were incubated under saturated conditions (soil/water = 1:1, w/w). Soils were incubated in 250-ml conical flasks at 25 °C in the dark, with 30 g of dry weight equivalent (DWE) soil in each flask. There were two 15N-labeled NH4NO3 treatments for every soil, one labeled with 15NH4NO3 and the other with NH415NO3, at 9.86 and 9.75 atom% 15N excess, respectively. Labeled ammonium nitrate (NH4NO3) fertilizer was added to each flask in solution at a rate of 60 μg N g−1 soil, and then further water added as necessary to achieve the required moisture content. The conical flasks were sealed with silicone rubber stoppers and incubated for a total of 144 h. During incubation, rubber stoppers were removed for 1 h every 2 days. Three conical flasks from each treatment were randomly selected at 0.5, 48, 96, and 144 h after 15N fertilizer solution application, and the soils were extracted by a soil/solution ratio of 1:5 with 2 M KCl, to determine exchangeable NH4+-N and NO3−-N concentration and their 15N abundance.

Soil analyses

Soil pH was measured in the ratio of soil to water of 1:5 (v/v) using a DMP-2 mV/pH detector (Quark Ltd., Nanjing, China; Thomas 1996; Zhang et al. 2013a, b). Soil organic C (SOC) was analyzed by wet digestion with H2SO4-K2Cr2O7, and total N was analyzed using a semi-micro Kjeldahl digestion with Se, CuSO4, and K2SO4 as catalysts (Nelson and Sommers 1996; Zhang et al. 2013a, b). Exchangeable NH4+-N and NO3−-N were extracted at a soil/solution ratio of 1:5 with 2 M KCl, shaking at 250 rpm for 60 min at 25 °C with a mechanical shaker. The extract was filtered, and concentrations of NH4+-N and NO3−-N determined using a segmented-continuous flow analyzer (Skalar, Breda, The Netherlands; Zhang et al. 2013a, b).

Ammonium and NO3− were separated by distillation with magnesium oxide and Devarda’s alloy for 15N measurements (Bremner and Keeney 1965). In brief, the extract was steam-distilled with MgO to separate NH4+; thereafter, NO3− was separated by Devarda’s alloy. The liberated NH3 was trapped in a conical flask with boric acid solution. The trapped N was acidified and converted to (NH4)2SO4 using 0.02 M H2SO4 solution. The H2SO4 solution containing NH4+ was evaporated to dryness at 65 °C in an oven. The 15N isotopic composition of NH4+ and NO3− was analyzed using a Sercon SL Elemental Analyzer coupled to a 20-20 isotope ratio mass spectrometer (IRMS, Sercon Ltd., Crewe, UK).

15N tracing model

The 15N tracing model was used to quantify the simultaneously occurring gross N transformation rates (Müller et al. 2007). The model includes the following processes: immobilization of NH4+ to recalcitrant organic N (I NH4 _ Nrec ) and labile organic N (I NH4 _ Nlab ), mineralization of labile organic N (M Nlab ) and recalcitrant organic N (M Nrec ) to NH4+, adsorption of NH4+ on cation exchange sites (A NH4 ) and release of adsorbed NH4+ (R NH4 ), oxidation of NH4+ to NO3− (O NH4 ) and oxidation of recalcitrant organic N to NO3− (O Nrec ), dissimilatory NO3− reduction to NH4+ (D NO3 ), and immobilization and denitrification of NO3− (I NO3 ).

The variables required by the model are the concentrations and 15N enrichments of NH4+ and NO3−. Further detail on this 15N transformation model is given by Müller et al. (2007). However, in brief, gross N transformation rates were calculated by zero-order or first-order kinetics by minimizing the difference between modeled and measured contents of exchangeable NH4+-N and NO3−-N and their 15N abundances. Several model modifications, kinetic settings, and variations in the number of considered N transformations and considered N pools were tested to identify the most appropriate model (in accordance with Rütting and Müller 2008). The final model used was selected according to Akaike’s information criterion (AIC). Optimization parameters were carried out by Markov Chain Monte Carlo Metropolis algorithm (MCMC-MA). The steps in the optimization algorithm and the model development are described in detail by Müller et al. (2007). The variance of the individual observations was accounted for by the misfit function f(m) between observations and simulation output. The MCMC-MA routine was carried out using MatLab (Version 7.2, The MathWorks Inc.).

Calculation and statistical analyses

The parameter averages and standard deviations of the model were calculated from the probability density function of each parameter, which was gained from the optimization procedure (Müller et al. 2007). To identify adequate iteration numbers, three parallel sequences were carried out in each analysis. Based on the kinetic settings and parameters, average N transformation rates were calculated and expressed in units of milligrams of N per kilogram soil per day (DWE). The differences in soil properties and N transformation rates between different treatments, within soil moisture groups, were examined by one-way ANOVA. When the difference was significant, Duncan’s test was used to compare the differences between treatments. The differences in soil properties and N transformation rates between different soil moisture conditions were examined by paired-samples t test for the grouped fertilizer treatments (omitting the CK treatment).

Competition between nitrification and immobilization of NH4+ is used as an index of NO3− loss (N/I), which is calculated using Eq. 1:

$$ N/I={O}_{NH4}/ TI $$
(1)

where TI is the total rate of NH4+ immobilization (I NH4 _ Nrec  + I NH4 _ Nlab ).

The nitrification capacity (NC) is calculated using Eq. 2:

$$ N\mathrm{C}={O}_{NH4}/ TM $$
(2)

where TM is the total rate of organic N mineralization (M Nlab  + M Nrec ).

Results

Properties of soils sampled from the field experiment

When compared to the CK treatment, the soil pH was significantly lower for the fertilized treatments, following both the unsaturated and saturated conditions (p < 0.01, Table 2), with lowest pH in the NhPK treatment. Total N (TN) content was greater in fertilizer treatments than in the CK treatment (p < 0.05, Table 2), while SOC content and C/N ratio were relatively stable (although SOC was numerically greater for the treatment where straw residue was added). Within the soil moisture grouping, there were no significant differences between fertilizer treatments in exchangeable NH4+-N content. However, for the unsaturated soil, NO3−-N content of in the NhPK treatment was 45.48 mg kg−1 DWE, which was significantly higher than the other treatments (Table 2). Thus, the nitrate ratio (defined as the ratio of NO3−-N to total inorganic N) in the NhPK treatment was significantly higher than for other treatments.

Compared to the unsaturated soils, soil pH increased significantly following the saturated conditions of the saturated rice-cropping season (p < 0.01). Soil exchangeable NH4+-N and NO3−-N contents were greater in the unsaturated soil (p < 0.01). For the fertilizer treatments, the soil NO3− ratios were greater in the unsaturated soil (p < 0.01).

Laboratory incubation experiment

Inorganic N pool concentration and 15N enrichment changes

Soil exchangeable NH4+-N concentration decreased, while soil NO3−-N concentration increased for all treatments over the incubation period (Fig. 1). The rate of increase of soil NO3−-N concentration (the slopes of the lines in Fig. 1) was greater under unsaturated than saturated soil conditions, indicating a greater net nitrification rate for the unsaturated soil. For the unsaturated soil, the rate of increase of soil NO3−-N concentration was greater for the NPK, NhPK, and NPKS treatments than for CK, demonstrating the stimulation of fertilizers on nitrification. A significant dilution was observed in all soils over the incubation period in the 15N enrichment of the NH4+-N pool following the addition of the 15NH4NO3 and in the 15N enrichment of the NO3−-N pool following addition of the NH415NO3 (Fig. 2). The 15NH4+-N enrichment where the NH4+-N pool was labeled was decreased greater under saturated conditions than under unsaturated conditions, indicating that the mineralization rate was greater under saturated conditions than under unsaturated conditions (Fig. 2). While, the 15N enrichment of NO3−-N following addition of the NH415NO3 was decreased greater under unsaturated conditions than under saturated conditions, suggesting that nitrification rate was greater under unsaturated conditions than under saturated conditions (Fig. 2).

Fig. 1
figure 1

Measured concentrations of exchangeable NH4+ and NO3− pools of CK (a), NPK (b), NhPK (c), and NPKS (d) after different incubation time. The concentrations of exchangeable NH4+-N or NO3−-N in soils under saturated or unsaturated conditions were the average of those for the 15NH4NO3 and NH415NO3 treatments (n = 18). Treatments were CK, no fertilization; NPK, mineral N, P, and K fertilizers applied at recommended rates; NhPK, mineral N fertilizers applied at rates 50% higher than NPK; and NPKS, mineral NPK fertilizer applied with straw

Full size image
Fig. 2
figure 2

Measured 15N enrichments of exchangeable NH4+ and NO3− pools of CK (a), NPK (b), NhPK (c), and NPKS (d) after different incubation times. The 15N enrichment of NH4+-15N or NO3−-15N in soils under saturated or unsaturated conditions was the average of those for the 15NH4NO3 or NH415NO3 treatments (n = 9). Treatments were CK, no fertilization; NPK, mineral N, P, and K fertilizers applied at recommended rates; NhPK, mineral N fertilizers applied at rates 50% higher than NPK; and NPKS, mineral NPK fertilizer applied with straw

Full size image

Effect of fertilizer treatment on N transformation rates

Gross N transformation rates under the two soil moisture conditions are presented in Fig. 3. For the soil samples collected following the rice harvest and incubated under saturated conditions, mineralization and immobilization rates were not significantly different between treatments (Fig. 3a, b). I NO3 increased in the order CK < NPK < NhPK < NPKS and was significantly higher for NPKS than for the other treatments (p < 0.05, Fig. 3c). The rates of O NH4 were significantly greater for the fertilization treatments than the CK treatment (p < 0.05), but no significant differences were observed between fertilization treatments (Fig. 3d). Similar trends were observed for the gross rate of D NO3 , although the magnitudes of D NO3 were much smaller than those of O NH4 (Fig. 3e).

Fig. 3
figure 3

Gross N transformation rates for different treatments under saturated and unsaturated conditions as estimated using the 15N tracing model (mg N kg−1 soil day−1). Different lowercase or uppercase letters indicate significant differences at p < 0.05 under saturated and unsaturated conditions, respectively. TM total mineralization of organic N to NH4+, TI total immobilization of NH4+ to organic N, ONH4 oxidation of NH4+ to NO3−, O Nrec oxidation of recalcitrant organic N to NO3−, I NO3 NO3− immobilization by the microbial biomass and other NO3− consumption rates, such as denitrification, D NO3 dissimilatory NO3− reduction to NH4+. Treatments were CK, no fertilization; NPK, mineral N, P, and K fertilizers applied at recommended rates; NhPK, mineral N fertilizers applied at rates 50% higher than NPK; and NPKS, mineral NPK fertilizer applied with straw

Full size image

For the soil samples collected following the tobacco harvest and incubated under unsaturated conditions, there was no clear trend in the gross rate of organic N mineralization (Fig. 3a); however, the gross rates of NH4+-N immobilization were significantly higher for CK and NPK than for NhPK and NPKS (Fig. 3b). The gross rates of I NO3 were very low (less than 0.1 mg N kg−1 soil day−1 on average) with no significant differences between the treatments (Fig. 3c). The variation of O NH4 among the treatments was the same as that observed under saturated condition, but the magnitudes were larger compared to the corresponding treatment (Fig. 3d). The gross rate of DNRA (D NO3 ) was low under saturated condition with a significantly greater rate observed for NPKS relative to the other fertilizer treatments (p < 0.05, Fig. 3e).

Effect of soil moisture condition on N transformation rates

The rates of organic N mineralization (TM) and immobilization of NH4+ (TI) and the ratio of TI to TM in saturated conditions were significantly higher (15-fold for TI) than in unsaturated conditions (p < 0.05) (Figs. 4 and 5). There were positive relationships between soil pH and the rates of organic N mineralization (R2 = 0.62; p < 0.01, Fig. 6) and immobilization of NH4+ (R2 = 0.59; p < 0.01, Fig. 6). The rates of I NO3 and D NO3 were also higher under saturated soil conditions, whereas the rate of autotrophic nitrification decreased under saturated conditions (Fig. 4).

Fig. 4
figure 4

Nitrogen transformation pathways in soils under saturated and unsaturated conditions (mg N kg−1 soil day−1). The red arrows signify transformation processes where significant differences were found between the unsaturated and saturated soils (paired-sample t tests). The yellow highlight signifies the dominant form of inorganic N in the soils. Data presented are the means of gross rates of N transformation for all the fertilizer treatments combined (NPK + NhPK + NPKS; n = 9)

Full size image
Fig. 5
figure 5

The ratio of NH4+ immobilization to organic N mineralization (I/M) for different soil moisture conditions

Full size image
Fig. 6
figure 6

Relationship between soil pH and a gross rate of mineralization and b NH4+ immobilization. The data from the NPK, NhPK, and NPKS treatments were analyzed together, n = 18

Full size image

The ratio of autotrophic nitrification to NH4+ immobilization (N/I) and nitrification capacity (NC, defined as the ratio of gross rate of nitrification to total gross rate of mineralization) was lower under the saturated soil conditions than unsaturated soil conditions (Fig. 7). Soil NO3−-N concentration increased linearly with NC (R2 = 0.84; p < 0.01).

Fig. 7
figure 7

The nitrification capacity (NC) (a) and ratio of nitrification to immobilization (N/I) (b) of soils with different moisture content. The data from the NPK, NhPK, and NPKS treatments were analyzed together, n = 18. NC was calculated as [O NH4 /TM]. Different lowercase letters indicate significant difference of N/I and NC at p < 0.05

Full size image

Discussion

The effects of fertilization on N transformation in soil

It has been reported that N fertilizer application stimulates both gross mineralization and NH4+ immobilization in soils (Zhang et al. 2012). The 15N tracing results in our study indicated that fertilization did not substantially affect the rates of organic N mineralization (TM), except for the NPKS treatment under tobacco-cropping conditions, which was significantly higher than the CK treatment under unsaturated conditions (Fig. 3a). There was a positive relationship between the rates of organic N mineralization and soil pH (Fig. 6a), suggesting that the stimulated organic N mineralization in the NPKS treatment was likely due to the soil pH increase during the tobacco-cropping season (Table 1).

Long-term fertilizer additions stimulated autotrophic nitrification both under saturated and unsaturated conditions (Fig. 3), which agrees with previous reports (Ding et al. 2010; Hao et al. 2003; Meng et al. 2005). The ability of N fertilizers to stimulate O NH4 may be due to a positive stimulation of the microbial biomass and activity with the increase in the abundance of ammonia-oxidizing bacteria (AOB) and archaeal nitrifiers (AOA; Chu et al. 2008; Shen et al. 2008; Zhang et al. 2015b). Increased applications of N fertilizers have been shown to result in a significant increase in population size of autotrophic AOB (Chu et al. 2008; Shen et al. 2008), which in turn are responsible for an increased rate of autotrophic nitrification (Zhang et al. 2013b).

The rate of autotrophic nitrification in acid soils increases with increasing soil pH (Jiang et al. 2015; Stempfhuber et al. 2015). We observed a pH increase in the NPKS treatment during the unsaturated tobacco-cropping season (Table 1), but this did not result in an increased rate of autotrophic nitrification (Fig. 3d). It is possible that this was because of the effect of allelochemicals (e.g., ferulic acid and benzoic acid), as products of the straw decomposition process in this treatment which can inhibit the numbers and activity of nitrifiers (e.g., Nitrosomonas and Nitrobacter) in the soil (Chung et al. 2001; Zhang et al. 2015c).

Generally, SOC content exerts a key control on denitrification and NO3−-N immobilization (Firestone and Davidson 1989; Greenan et al. 2006; Hayakawa et al. 2012; Recous et al. 1990). The application of organic material increases the content of SOC and thereby may increase the rates of denitrification and NO3−-N immobilization (Burger and Jackson 2003; Miller et al. 2008; Nishio et al. 2001). In our study, for the saturated soil, rice straw application (which numerically increased soil SOC; although not statistically significant) increased the I NO3 rate (Fig. 3c) and a positive relationship between the rate of I NO3 and soil SOC content was observed (R2 = 0.524, p < 0.05). However, no such relationship was observed for the unsaturated soil conditions (Table 2), indicating that C substrate was not limiting I NO3 under unsaturated conditions. With the 15N tracing model applied in this study (Müller et al. 2004, 2007), removal of NO3−-N through immobilization and denitrification is quantified together as a single rate and cannot be separated into the pathway-specific rates. To achieve this, details on nitrite (NO2−) dynamics in soils are also required (Rütting and Müller 2008) which were not determined in this study but would provide valuable additional information if measured in future studies.

Nitrogen conservation mechanisms in saturated soil conditions

Ammonium is the preferred N source for uptake by rice (Luo et al. 1993); organic N mineralization and NH4+ immobilization are important processes of NH4+ production and consumption in soils. The measured data of 15N enrichment of NH4+ pool under 15NH4NO3 treatment and 15N enrichment of NO3− under NH415NO3 treatment suggested greater mineralization rate and lower nitrification rate under saturated conditions than under unsaturated conditions (Fig. 2). However, comparing with unsaturated conditions, there was no increasing of exchangeable NH4+-N concentrations under saturated conditions (Fig. 1), indicating that immobilization rate of NH4+-N was greater under saturated conditions than under unsaturated conditions.

In line with the qualitative results observed in measured data, the numerical model analysis showed that the rates of organic N mineralization (TM) under saturated soil conditions were significantly higher than under aerobic conditions (Fig. 4), which agrees with Mathieu et al. (2006) who made similar observations in the Saône river plain, near Dijon (Eastern France) in an agricultural soil. There was a positive correlation between the rate of organic N mineralization and soil pH (Fig. 6a), which suggest that the increased rate of organic N mineralization in saturated soils may be attributable to soil pH increase (Table 2).

High NH4+-N availability in paddy soils may increase N loss through runoff or leaching too, even though NH4+-N is less mobile than NO3−-N, and through NH4+ oxidation may result in elevated NO3− concentrations (Huygens et al. 2007; Inagaki and Miura 2002; Zhu et al. 2013a). Moreover, NH4+ immobilization in soils, controlled by biotic and abiotic processes, may significantly influence N availability for crops as well as affect N losses (Said-Pullicino et al. 2014). Our results showed that the rates of NH4+ immobilization (TI) in the saturated soil were significantly higher than that under unsaturated conditions (Fig. 4), which corresponds to findings by Compton and Boone (2002) who identified a positive relationship between the rate of inorganic N immobilization and soil moisture content. This indicates that NH4+ will not easily accumulate in soil under saturated conditions, which is further supported by the correlation between soil pH and NH4+ immobilization rate (p < 0.01, Fig. 6b). The ratio of TI/TM under saturated conditions was significantly higher than that under unsaturated condition, and in more than 75% of all cases higher than 1.2 (Fig. 5). Those results explained the lower soil exchangeable NH4+-N concentrations under saturated conditions than that under unsaturated conditions (Table 2). Low soil exchangeable NH4+-N concentrations was expected to reduce N loss via anaerobic ammonium oxidation, which was considered as an important N removal pathway in anaerobic agricultural soil (Shen et al. 2016).

Generally, autotrophic nitrification is considered to be a microbial oxidation process under aerobic conditions. A previous study found that the optimum moisture conditions for nitrification under different cropping systems were about 60% water-filled pore space (WFPS) (Linn and Doran 1984). Kiese et al. (2008) found that the relationship between the rates of gross nitrification and soil moisture could be described best by the O’Neill functions, with a soil moisture optimum for nitrification at 65% WFPS. However, Yang et al. (2016) reported that nitrification activity in paddy soils was not suppressed by low oxygen concentrations. Soil nitrifying microorganisms are known to adapt to the local conditions (Mahendrappa et al. 1966; Myers 1975), such as low pH (Wang et al. 2014), low temperature (Wang et al. 2012), or low oxygen concentration (Yang et al. 2016), and O2 transport to the rhizosphere of rice plants via parenchyma gas transport (Arth et al. 1998; Inubushi et al. 2002) may also facilitate nitrification.

However, the rates of nitrification under saturated conditions were still significantly lower than that under unsaturated conditions (Fig. 4), contributing to a lower accumulation of NO3−-N and a reduced risk of N loss from the paddy soil. Nitrification capacity (NC) is an index of the ability of autotrophic nitrifiers to compete with immobilizing bacteria for NH4+ and is a key factor in controlling N losses (Hart et al. 1994). Our study showed that the NC of the saturated rice-cropping season soil incubated under saturated conditions was significantly lower than the unsaturated tobacco-cropping soil incubated under unsaturated conditions (Fig. 7a). The positive relationship between NO3− contents with NC (R2 = 0.84; p < 0.01) also suggested that low nitrification rate under water-saturated condition contributed to low NO3− content and a lower risk of N loss.

The ratio of autotrophic nitrification to NH4+ immobilization (N/I) has been suggested as an index to quantify the competitive ability of heterotrophic microbes and nitrobacteria for NH4+ and the likelihood for N losses from soils (Stockdale et al. 2002). For the soil samples collected following rice harvest and incubated under saturated condition, N/I was lower than 1 (Fig. 7b), suggesting that the substrate of autotrophic nitrification (NH4+) would be deficient because of high immobilization rates. This result implies that, if plenty of NH4+-N is made available (e.g., through fertilizer application), the rate of nitrification in paddy soil would increase due to increased availability of NH4+ to the nitrifying microorganisms. Thus, split applications of N fertilizer are an important management practice to decrease the risk of N loss in paddy soils.

The processes of DNRA reduce the soil NO3− content while retaining the N in an available form (Silver et al. 2005). In our study, the rate of DNRA was significantly higher under saturated soil conditions (Fig. 4), which can be explained by the requirement of anoxic soil conditions to promote the DNRA process, and also explains why DNRA plays only a minor or negligible role in aerobic soils (Tiedje 1988). Therefore, under saturated conditions, surplus NO3−-N can be transformed to NH4+-N through the DNRA process thereby potentially conserving N in the system. Accompanied with higher rates of organic N mineralization and NH4+ immobilization and lower rates of NH4+ autotrophic nitrification in the saturated soils than in unsaturated soils, N can be retained in soils during the saturated rice-cropping season. This observation was supported by the view of Houtermans et al. (2017) that paddy management promotes N sequestration in soils.

It was reported that NO3−-N immobilization occurs when NH4+-N is very low (Rice and Tiedje 1989). In our study, soil exchangeable NH4+-N contents were greater in the unsaturated soil than that in the saturated soil. Thereby, the rate of I NO3 under saturated conditions was significantly higher than under unsaturated conditions (Fig. 4). Although the processes of NO3− immobilization and denitrification could not be considered independently (see above), saturated conditions generally favor the denitrification process (Xing et al. 2002) and N losses via that pathway were therefore likely to have been much higher than N conserved through immobilization. However, because NO3− availability is a key factor for denitrification (Greenan et al. 2006), the high rates of NH4+ immobilization and DNRA and low nitrification rate resulted in low NO3− availability under the saturated soil condition (Table 2), thereby decreasing opportunities for gaseous N losses via denitrification.

Conclusion

Contrary to the general assumption that soils under saturated conditions are prone to N losses, there seems to be an inherent soil N conservation mechanism during the saturated rice-cropping season to effectively conserve N. This is predominantly via reduced NO3− production and improved DNRA process. Higher organic N mineralization rates in paddy soils provide sufficient NH4+-N for crop uptake, and surplus NH4+-N is immobilized rather than nitrified under the saturated soil conditions. Soil inorganic N concentrations under saturated conditions were significantly lower than under unsaturated conditions, decreasing the potential for N loss through runoff or leaching. However, the rates of NO3− immobilization and denitrification of the unsaturated rice-cropping soil were significantly higher than in the unsaturated tobacco-cropping soil, and the risk of N loss through denitrification maybe higher in the saturated soil than in the unsaturated soil. The relationship between nitrification, NO3− immobilization, and denitrification should be evaluated in more detail in future studies.