Introduction

In China, nitrate (NO3 −) deposition is increasing rapidly, although NH4 + is currently the dominant form of N in bulk deposition (Liu et al. 2013). Enhanced deposition of NO3 − may fundamentally change microbial cycling of N in forest soils. In vegetated soils, additions of NO3 − can stimulate more root-derived respiration (Gavrichkova and Kuzyakov 2008) and soil CO2 flux (Wang et al. 2015) than NH4 +, partly because the reduction of NO3 – to NH4 + prior to assimilation requires oxidation of carbohydrates (Aslam 1982; Ninomiya and Sato 1984). More recently, NO3 – has been shown as an inhibitor of several fungal lignolytic enzyme production and activity, e.g., fungal phenol oxidases and peroxidases of white-rot basidiomycetes (Ekberg et al. 2007; Kaiser et al. 2010; Waldrop and Zak 2006).

The soil internal N cycling exerts a strong control over N availability, ultimately affecting net primary productivity, plant growth, microbial function, and N loss of an ecosystem (Schimel and Bennett 2004). Many factors influence the rates of soil N cycling, including soil moisture, soil pH, soil organic matter (SOM), and land use and management (Cheng et al. 2014; Gao et al. 2015; Lang et al. 2016). Soil organic matter is the largest soil N pool (Huygens et al. 2008), and its composition and concentration influence the production and fate of mineral N (Booth et al. 2005; Laungani and Knops 2012). Microbial communities play key roles in decomposition and utilization of organic substrates (Nannipieri et al. 2003). Fungi, compared to bacteria, are more efficient decomposers of recalcitrant SOM (Hodge et al. 2000), and gram-negative bacteria are typically linked to the degradation and utilization of protein and chitin (liable, N-contained compounds) (Koranda et al. 2014).

Global change drivers such as N deposition are likely to feed back on forest soil N cycling through the effects on SOM composition and concentration, and microbial activity, biomass, and composition. In the short term, low additions of N to forest soils generally stimulated fungal activity and growth, decomposition of low N-wood substrates, and microbial respiration (Allison et al. 2009; Koranda et al. 2014), while it reduced protein- and chitin-degrading enzyme activities and the abundance of gram-negative bacteria (Allison et al. 2008; Koranda et al. 2014), and resulted in a depletion of labile SOM pool (Cusack et al. 2011). These observations indicate that at low N deposition condition, a shift of N mineralization from fast to slow turnover pools might occur. However, a shift towards fast turnover organic N pools may occur at excessive N deposition, due to either enhanced formation of recalcitrant compounds that are resistant to microbial decay, shifts in microbial enzyme synthesis, and activity towards preferential decomposition of labile and energy-rich compounds, or both (Janssens et al. 2010; Maaroufi et al. 2015). Detailed investigations of these responses are central to understanding how forest ecosystems may maintain their functional stability under anthropogenic N inputs, but are currently lacking.

Despite presence of widespread evidence for N addition-induced nonlinear increase in net production and loss of nitrate (NO3 –) from soils, it remains uncertain to what extent this increase results from stimulation of either autotrophic or heterotrophic nitrifiers, or from the saturation of N uptake by plants, heterotrophic microbes, mycorrhizae, and abiotic reactions (Perakis et al. 2005). In acid soils of subtropical/tropical forests, heterotrophic nitrification dominates over autotrophic nitrification in NO3 – production, and NO3 – immobilization is an effective N retention mechanism (Zhang et al. 2013; Zhu et al. 2013). In these soils, microbial assimilation of NO3 – depends likely on heterotrophic nitrification for substrate generation, and both processes are functionally linked (Huygens et al. 2008; Zhang et al. 2013; Zhu et al. 2013). Acidic forest soils with high soil C/N ratio and high fungal biomass generally show a high rate of heterotrophic nitrification (Zhang et al. 2015; Zhu et al. 2013). There is increasing evidence of stimulated effects of low N additions on fungal growth or activity (Koranda et al. 2014; Rousk and Baath 2007), suggesting that there would be an increase in heterotrophic nitrification rates (Zhu et al. 2013). However, soil C/N ratio (Corre et al. 2010; Templer et al. 2012) and fungal biomass (Demoling et al. 2008; Frey et al. 2004, 2014) decreased at forest sites receiving long-term or high N deposition, which might inhibit heterotrophic nitrification and also NO3 − immobilization.

In this paper, we report results from a 15N tracing study with soil fertilized for more than 2.5 years at a rate of 0, 40, and 120 kg NO3 −–N ha−1 year−1, respectively, in a slash pine (Pinus elliottii) forest of subtropical China. Our objective was to determine if and how enhanced deposition of NO3 − would affect microbial cycling of N in the soil. We hypothesized that (1) mineralization of slow turnover organic N would increase, and mineralization of fast turnover organic N would decrease under low NO3 − additions, and vise versa under high NO3 − additions; (2) heterotrophic nitrification, would first increase, then decrease with increasing NO3 − additions; and (3) NO3 − immobilization would show a similar change as heterotrophic nitrification in response to increasing NO3 − additions. To test our hypotheses, process- and pool-specific N transformation rates were quantified via a process-based 15N model (Müller et al. 2007; Rütting and Müller 2007).

Materials and methods

Site description, experimental design, and sampling

The study site is a 28-year-old subtropical slash pine (P. elliottii) forest at Qianyanzhou (QYZ) Red Soil Hilly Station, Jiangxi, Southern China (115° 03′ 29.2″ E, 26° 44′ 29.1″ N, 102 m a.s.l.). The mean annual temperature is 17.9 °C, and the mean annual precipitation is 1475 mm (Wen et al. 2010). The understory is dominated by Woodwardia japonica, Loropetalum chinense, and Dicranopteris dichotoma (Wang et al. 2012). The soils are classified as Typic Dystrudepts (USDA Soil Taxonomy). Annual precipitation deposition of dissolved inorganic N (DIN) equals 12.6 ± 1.5 kg N hm−2 a−1, with a NH4 +–N/NO3 −–N ratio of 1.9:1 (Zhan et al. 2014).

To investigate the influence of atmospheric NO3 − deposition on microbial N cycling in the soil, we established a simulated NO3 − deposition experiment. Starting in May 2012, we fertilized 20 m × 20 m plots with 0, 40, and 120 kg N ha−1 year−1 in a randomized block design (three replicates per treatment, totaling 9 plots). Nitrogen was added as sodium nitrate (NaNO3). The three blocks were laid out across 40 ha, and there were at least 10-m buffer zones between plots within each block. N solutions were sprayed monthly to the forest floors with a backpack sprayer. Each month, fertilizer was weighed and mixed with 30-L water for each plot, and each control plot received 30 L of water without fertilizer.

In late November 2014 (roughly 2.5 years after N additions), 8 subsamples were taken from the 0–10 cm in each plot using a PVC tube (inner diameter of 7 cm; length of 15 cm). Subsamples were pooled together by plot and were not mixed for the same treatment (totaling to nine composited samples). Each composited sample was sieved (2-mm), homogenized, and subsequently divided into two subsamples: one for analysis of soil properties, and another for the laboratory 15N tracing experiments. The tracing experiments were carried out within 2 weeks, and soil samples were stored at 4 °C before analyses.

15N tracing experiment

There were two 15N treatments (each with three replications), of which either ammonium (15NH4NO3) or nitrate (NH4 15NO3) were labeled with 15N at 10 atom percent excess. For each plot, the sieved soil was placed in four sets of conical flasks (six conical flasks per sets, three of the six conical flasks for 15NH4NO3 labeling, and the remaining three for NH4 15NO3 labeling; each conical flask containing fresh soil with the equivalent of 20 g of dry soil). These conical flasks were preincubated in the dark for 24 h at 25 °C prior to 15N labeling after sealing with parafilm with five pin holes for gas exchange. 15NH4NO3 or NH4 15NO3 was added at a rate of 2.86 μmol N g−1 dry soil (20 μg NH4 +–N g−1 dry soil and 20 μg NO3 −–N g−1 dry soil) in 3-ml per conical flask. The conical flask was incubated in the dark for 144 h at 25 °C after adjusting the soil to 60 % water holding capacity (WHC) and sealing with parafilm (with five pin holes for air exchange). Soil extractions were carried out at 0.5, 48, 96, and 144 h after label addition to determine the concentrations and isotopic compositions of NH4 + and NO3 −. A detailed description of laboratory 15N tracing study on each soil sample can be found in Fig. S1 (see supporting information).

Analyses

Soil moisture, soil pH, and contents of total C, total N, exchangeable NH4 +, and NO3 − were determined. Soil moisture was measured by drying at 105 °C for 48 h. Total C and N contents were analyzed by an elemental analyzer (Europa Scientific Integra, UK) using air-dried, finely ground soil. Soil pH was determined in a soil (air-dried)/water ratio of 1:2.5 by a DMP-2 mV/pH detector (Quark Ltd., Nanjing, China). Soil exchangeable NH4 + and NO3 − were extracted with 2 M KCl at a soil/extractant ratio of 1:5 after shaking for 60 min at 250 rpm and 25 °C. The concentrations of exchangeable NH4 + and NO3 − were assayed with a continuous-flow analyzer (Skalar, Breda, the Netherlands). The soil properties are presented in Table 1.

Table 1 Soil properties at 0–10-cm depth after 2.5 years of NO3 − additions
Full size table

The isotopic compositions of exchangeable NH4 + and NO3 − were determined using an automated C–N analyzer coupled to an isotope ratio mass spectrometer (Europa Scientific Integra, UK). For isotopic analysis, NH4 + and NO3 − were separated by distillation with magnesium oxide (MgO) and Devarda’s alloy. A detailed description of this method can be found in previous studies (Zhang et al. 2012, 2013).

15N tracing model

Gross rates of N transformations were quantified with a process-based 15N tracing model (Fig. 1) (Müller et al. 2007). The model considered 10 simultaneously occurring gross N transformations: M Nrec mineralization of recalcitrant organic N to NH4 +, M Nlab mineralization of labile organic N to NH4 +, I NH4_Nrec immobilization of exchangeable NH4 + to recalcitrant organic N, I NH4_Nlab immobilization of exchangeable NH4 + to labile organic N, R NH4ads release of adsorbed NH4 +, A NH4 adsorption of NH4 + on cation exchange sites, O NH4, oxidation of NH4 + to NO3 − (autotrophic nitrification), O Nrec oxidation of recalcitrant organic N to NO3 − (heterotrophic nitrification), INO3-Nrec immobilization of NO3 − to recalcitrant organic N, and D NO3 dissimilatory NO3 − reduction to NH4 +. The transformation rates were calculated either by zero (M Nrec and O Nrec) or first-order (M Nlab, I NH4_Nrec, I NH4_Nlab, O NH4, I NO3-Nrec, and D NO3) kinetics. Based on available N cycling parameters, gross N mineralization (M tot = M Nlab + M Nrec), gross nitrification/gross NO3 − production (O NH4 + O Nrec), NH4 + immobilization (I NH4 = I NH4_Nrec + I NH4_Nlab), and gross NO3 − immobilization (I NO3 = I NO3_Nrec) were calculated. The data supplied to the model were the concentrations (μmol N g−1 dry soil) and enrichments (at.% excess 15N) of exchangeable NH4 + and NO3 − in the form of mean ± standard deviations (SD) (Fig. S1). For each plot, the simulated concentrations and isotopic enrichments were generally within the range of the observed averages ± SD (see supporting information, Figs. S2–10).

Fig. 1
figure 1

Conceptual 15N tracing model (Müller et al., 2007). N rec recalcitrant organic N; N lab labile organic N; NH 4 + ammonium; NO 3 − nitrate; M Nrec mineralization of recalcitrant organic N to NH4 +; M Nlab mineralization of labile organic N to NH4 +; I NH4_Nrec immobilization of exchangeable NH4 + to recalcitrant organic N; I NH4_Nlab immobilization of exchangeable NH4 + to labile organic N; R NH4ads release of adsorbed NH4 +; A NH4 adsorption of NH4 + on cation exchange sites; O NH4 oxidation of exchangeable NH4 + to NO3 − (autotrophic nitrification); O Nrec oxidation of recalcitrant organic N to NO3 − (heterotrophic nitrification); I NO3_Nrec immobilization of exchangeable NO3 − to recalcitrant organic N; and D NO3 dissimilatory NO3 − reduction to NH4 +

Full size image

The model calculated gross N transformation rates by simultaneously optimizing the kinetic parameters for each individual process via minimizing the misfit between modeled and observed concentrations of exchangeable NH4 + and NO3 − and their respective 15N enrichments. To identify the most suitable model that could best simulate the observed data and the measured soil N cycling, several model modifications, which varied in the number of N transformations, kinetic settings of individual processes (zero/first-order kinetics/Michaelise-Menten kinetics) and considered N pools, were tested. The final model (Fig. 1) was identified based on the lowest Aikaike’s information criterion (AIC) (Cox et al. 2006). Initially, all parameters (N pools and N transformations) from the conceptual model (Fig. 1) were included in the optimization run and the kinetic settings adjusted to reach an AIC as low as possible. Parameters, which did not significantly improve the model fit, were excluded in the following step when they approached zero in the analysis. In general, zero-order kinetics are more appropriate to describe N transformations (e.g., M Nrec and O Nrec) originating from large pool sizes, while N transformations (M Nlab, I NH4_Nrec, I NH4_Nlab, I NO3-Nrec, and D NO3) originating from small pool sizes are likely to follow first-order kinetics (Myrold and Tiedje 1986). During the incubation, the activity of nitrifying microorganisms may undergo a rapid change from non-NH4 + limiting conditions (zero-order kinetics) to NH4 + limiting conditions (first-order kinetics) (Müller et al. 2007). In this case, Michaelise-Menten kinetics, rather than zero- or first-order kinetics, could be more appropriate to describe NH4 + oxidation. A detailed description of stepwise modification of parameters and their kinetic settings to find the lowest resulting AIC have been reported (Inselsbacher et al. 2013; Müller et al. 2004, 2007; Rütting and Müller 2007). The model parameters were optimized with Markov Chain Monte Carlo Metropolis algorithm (MCMC-MA), which has been described in detail by Müller et al. (2007). To get a better resolution of soil processes, and in line with previous studies, soil organic N pool was conceptually divided into two fractions, a liable (active) and a recalcitrant (slow) pool (Inselsbacher et al. 2013; Müller et al. 2007; Zhang et al. 2013). The initial (i.e., t = 0) pool sizes of mineral N (14N and 15N) were estimated based on Müller et al. (2004)). The optimization procedure produced a probability density function (PDF) for each process, from which the average and standard deviation of each process was calculated. For N transformations following first-order kinetics, average gross rates were calculated by integrating the gross rates over the entire experimental period, divided by the incubation time (Inselsbacher et al. 2013; Müller et al. 2007). The gross N transformation rates were expressed in milligrams N per kilogram dry soil per day. The MCMC-MA routine was processed by the software Matlab (Version 7.2, The Math Works Inc.), which considers models separately.

Data analysis

All the data were tested for normality (Shapiro–Wilk test) and homogeneity (Levene test) of variance before analysis. If preconditions of ANOVA were not met, we employed the Kruskal-Wallis H test with paired comparisons for testing differences among treatments. Data that showed normal distribution and homogeneity were tested by using the one-way ANOVA with least significant difference (LSD) test to compare the differences between treatments. Nitrogen dose was set as main effects. All analyses were conducted using SPSS version 20.0 (IBM Co., Armonk, NY, USA), and α < 0.10 was considered as the threshold value for significance.

Results

Dynamics of N pool sizes and 15N enrichments

The stimulated and observed concentrations and 15N enrichments of exchangeable NH4 + and NO3 − matched well. For each plot, the simulated concentrations and isotopic enrichments were generally within the range of the observed averages ± SD (Figs. S2–10). Concentrations of exchangeable NH4 + and NO3 − increased with incubation time (Figs. S2–10a, b) The dilution of 15N in the exchangeable NH4 + (or NO3 −) pool when the N pool was labeled indicated an inflow of unlabeled exchangeable NH4 + (or NO3 −) into the labeled N pool. 15N of the exchangeable NH4 + pool under NH4 15NO3 labeling showed a very small change, suggesting that dissimilatory nitrate reduction to ammonium (DNRA) rates were negligible (Figs. S2–10c). The 15N of the NO3 − pool showed a slow increase under 15NH4NO3 labeling, indicating a low potential to oxidize exchangeable NH4 + to NO3 − (autotrophic nitrification) (Figs. S2–10d).

Production and consumption of exchangeable NH4 +

There were treatment effects of NO3 − additions on gross N mineralization (M tot = M Nlab + M Nrec, p = 0.10; Fig. 2). M tot in the high N addition plots were 23.7 % lower than in the control plots (p = 0.04), while M tot in the low N addition plots did not differ significantly from other plots at p < 0.10 (Fig. 2). Exchangeable NH4 + production via mineralization of recalcitrant organic N (M Nrec) increased from 1.94 ± 0.12 mg N kg−1 soil day−1 in the control plots to 2.78 ± 0.24 mg N kg−1 soil day−1 in the low N addition plots (p = 0.011), and then decreased by 58.3 % to 1.16 ± 0.09 mg N kg−1 soil day−1 in the high N addition plots (p < 0.001) (Fig. 3). Exchangeable NH4 + production via mineralization of labile organic N (M Nlab) showed the opposite in response to increasing NO3 − additions (Fig. 3).

Fig. 2
figure 2

Gross rates of N cycling in the surface soil (0–10 cm) in response to experimental NO3 − additions. Rates are means with standard errors (mg N kg−1 dry soil day−1; n = 3 plots), and see Fig. 1 for definition of abbreviations. There was no difference detected among treatments for NH4 + immobilization, heterotrophic nitrification, and DNRA. For N mineralization, autotrophic nitrification and NO3 − immobilization, means followed by the same letter (black, blue, and red for N mineralization, autotrophic nitrification and NO3 − immobilization, respectively) indicated no significance among treatments (one-way ANOVA, least significant difference test at p < 0.10)

Full size image
Fig. 3
figure 3

Mineralization–immobilization turnover in the surface soil (0–10 cm) and its response to experimental NO3 − additions. Rates are means with standard errors (mg N kg−1 dry soil day−1; n = 3 plots), and see Fig. 1 for definition of abbreviations. There was no difference detected among treatments for exchangeable NH4 + immobilization into labile and recalcitrant organic N. For mineralization of labile and recalcitrant organic N to NH4 +, means followed by the different letter (upper case for mineralization of labile organic N to NH4 +, and lower case for mineralization of recalcitrant organic N to NH4 +) indicated significance among treatments (one-way ANOVA, least significant difference test at p < 0.10)

Full size image

The main consumption pathway for the produced exchangeable NH4 + was NH4 + immobilization rather than autotrophic nitrification (Fig. 2). Gross NH4 + immobilization rates (I NH4) did not differ significantly among treatments (p = 0.382). Most of the produced exchangeable NH4 + was immobilized into labile organic N (59 to 71 %) rather than recalcitrant organic N (29 to 41 %) (Fig. 3), and the respective rate of NH4 + immobilization into labile (I NH4_Nlab) and recalcitrant (I NH4_Nrec) organic N did not differ significantly across treatments (Fig. 3).

Production and consumption of NO3 −

Forty-one to 85 % of the NO3 − production was derived from oxidation of N rec (O Nrec), and the rest was the result of NH4 + oxidation (O NH4) (Fig. 2). Heterotrophic nitrification rates (O Nrec) did not vary significantly among treatments (Fig. 2). Autotrophic nitrification rates (O NH4) increased significantly with increasing NO3 − additions (Fig. 2). However, gross rates of NO3 − production (O Nrec + O NH4) were similar among treatments (p = 0.313). Nitrate immobilization (I NO3) dominated over DNRA in NO3 − consumption, and I NO3 rates differed significantly among treatments (p = 0.016; Fig. 2). I NO3 rates decreased from 0.44 ± 0.08 mg N kg−1 soil day−1 in the control plots to 0.09 ± 0.02 mg N kg−1 soil day−1 in the low N addition plots (p = 0.006), and then increased by 200 % to 0.27 ± 0.06 mg N kg−1 soil day−1 in the high N addition plots (p = 0.074) which were still lower than the controls (p = 0.088) (Fig. 2). DNRA retained 6.2 to 6.8 % of the NO3 − produced (O Nrec + O NH4), and DNRA rates did not change significantly with treatments (Fig. 2).

Discussion

The effects of NO3 − addition on microbial NH4 + cycling

Previous studies have indicated that soil organic matter fractions, characterized by different size, quality, and turnover time, and linked to  physiologically and functionally different soil microbes, respond differently to N deposition over time (Cusack et al. 2011; Frey et al. 2014; Janssens et al. 2010; Koranda et al. 2014; Maaroufi et al. 2015; Swanston et al. 2004). Thus, in the current study, we quantified two specific gross N mineralization rates, related to either a rapid (N lab) or a slower (N rec) turnover of organic N pool (Fig. 1). Obviously, the reduced gross N mineralization in the high N addition plots was primarily a result of decreased mineralization of N rec (M Nrec). The reduction in M Nrec rates under high NO3 − additions is not surprising. Unlike mineralization of N lab, mineralization of N rec requires a depolymerization step, which is generally carried out by exoenzymes that are often produced by fungi (Schimel and Bennett 2004). There is growing evidence of N enrichment-induced reductions in fungal biomass and activity in forest soils (Demoling et al. 2008; Frey et al. 2014; Maaroufi et al. 2015). Moreover, the observed inhibition of several fungal lignolytic enzyme production and activity, e.g., fungal phenol oxidases and peroxidases of white-rot basidiomycetes, by excessive additions of NO3 − further supports our results (Ekberg et al. 2007; Kaiser et al. 2010; Waldrop and Zak 2006). It has been suggested that fungal decline under excessive N additions might be a result of enhanced stabilization of organic matter into recalcitrant compounds, as which are resistant to microbial decay or impairs fungal metabolism (Janssens et al. 2010; Maaroufi et al. 2015).

Low NO3 − addition treatment had a contrasting effect on mineralization of N rec relative to high NO3 − addition treatment. Low NO3 − additions stimulated mineralization from N rec. This finding is in line with studies of inorganic N addition to a beech forest soil (Koranda et al. 2014) or an arable soil (Zhang et al. 2012). These occur partly because under low NO3 − addition treatment, (1) microbial N limitation (especially fungal N limitation) was alleviated, resulting in increased respiratorily responsive biomass and growth-active biomass (especially fungi) to stimulate decomposition of complex C/N substrates (Koranda et al. 2014); (2) decreased C/N ratio of N rec (Aber et al. 1998); and (3) a combination of (1) and (2).

In support of our first hypothesis, the directions of changes in M Nrec and M Nlab rates were opposite under increased NO3 − additions (Fig. 3). Our results suggested that microbial mineralization of Nrec (M Nrec) would be upregulated when microbial mineralization of N lab (M Nlab) was downregulated under low NO3 − additions, and the possible mechanisms have been presented above. Two, mutually non-exclusive, mechanisms proposed by Janssens et al. (2010) might contribute to the contrasting responses of M Nrec and M Nlab under high NO3 − addition treatment: enhanced stabilization of organic matter into recalcitrant compounds; shifts in synthesis and activity of microbial extracellular enzymes towards preferential decomposition of liable, energy-rich compounds.

The change in N mineralization (gross N mineralization, mineralization of N rec, and mineralization of N lab) did not induce a corresponding change in exchangeable NH4 + immobilization (gross NH4 + immobilization, exchangeable NH4 + immobilization into labile organic N, and exchangeable NH4 + immobilization into recalcitrant organic N). Such a phenomenon might be explained by the high variance of the observed exchangeable NH4 + immobilization rates (Fig. 3). Thus, our results contrast with other findings. A number of researchers have reported decreased NH4 + immobilization with either increased or decrease N mineralization in forests receiving high levels of N deposition (Corre et al. 2003, 2007, 2010; Venterea et al. 2004). However, NH4 + immobilization might not necessarily change with the change in N mineralization if there is ample NH4 + availability, and microbes are lack of N limitation in the soil, as we observed in this study site. Neither an increase nor a decrease in the rates of exchangeable NH4 + immobilization into recalcitrant organic N occurred, suggesting that microbial N demand did not change under increased NO3 − additions (Rütting et al. 2010). High microbial demand for exchangeable NH4 + (including autotrophic nitrifies), together with reduced exchangeable NH4 + production from recalcitrant organic N, indicates that plant–microbe competition for available NH4 + might increase in the high NO3 − addition plots.

The effects of NO3 − addition on microbial NO3 − cycling

Nitrate additions did not stimulate gross nitrification in the subtropical forest soil. This result does not agree with previous findings of increased gross nitrification in tropical forests under enhanced N deposition (Baldos et al. 2015; Corre et al. 2010; Silver et al. 2005). Contrary to our second hypothesis, heterotrophic nitrification did not change significantly with increasing NO3 − additions. Previous studies have suggested that in acidic subtropical/tropical forests, soil C/N, fungal biomass, and soil pH were the key controls of heterotrophic nitrification, and their reductions could with a decline in the heterotrophic nitrification rates under enhanced N deposition (De Boer and Kowalchuk 2001; Zhang et al. 2015; Zhu et al. 2013). Thus, the lack of significant response from heterotrophic nitrification might be related to the lack of changes in soil C/N and soil pH (Table 1). Zhu et al. (2013) found that oxidation of recalcitrant organic N (heterotrophic nitrification) in acid forest soils of subtropical China decreased as gross N mineralization increased (or when mineralization of organic N was relatively high). This suggests that microbial mineralization of organic N to NO3 − and NH4 + might be mutually exclusive or competitive. However, this is not confirmed by our data as heterotrophic nitrification in the high N addition plots showed a decreasing trend with decreased N mineralization (Fig. 2).

However, autotrophic nitrification rates were observed to increase with increasing NO3 − additions, possibly as a result of NO3 − addition-induced reductions in microbial assimilation of NH4 + (although not statistically significant; Fig. 3) and the associated increase in NH4 + availability (Booth et al. 2005; Corre et al. 2010). Specifically, autotrophic nitrification dominated over heterotrophic nitrification in NO3 − production under high NO3 − addition treatment, indicating an increase in the risk of N losses.

Again, inconsistent with our third hypothesis, NO3 − immobilization first decreased, then increased with increasing NO3 − additions. However, NO3 − immobilization in the high N addition plots was still lower than the control plots (Fig. 3). Previous studies have suggested that NO3 − immobilization depends directly on heterotrophic nitrification for substrate generation, and both pathways are functionally linked in subtropical/tropical acidic forest soils (Gao et al. 2016). At our site, NO3 − immobilization rates were higher in the control plots with higher heterotrophic nitrification rates, and were lower in the N addition plots with lower heterotrophic nitrification rates (Fig. 2). In other subtropical/tropical acidic forest soils with a high rate of heterotrophic nitrification, NO3 − immobilization rates were also higher, but this was not the case for forest soils with high autotrophic nitrification rates (Huygens et al. 2007, 2008; Zhang et al. 2013; Zhu et al. 2013). Also, autotrophic nitrification rates in grassland (Müller et al. 2011; Rütting et al. 2010) and arable (Zhang et al. 2012) soils were generally high; however, NO3 − immobilization rates in these soils were extremely low, possibly due to the effectiveness of low NH4 + concentrations in inhibiting NO3 − assimilation (Rice and Tiedje 1989). Thus, the reduced NO3 − immobilization under NO3 − additions could be related to decreased heterotrophic nitrification and increased autotrophic nitrification.

Low NO3 − addition treatment had a greater inhibition on NO3 − immobilization than high NO3 − addition treatment (Fig. 2). This may be due to NH4 + being energetically favorable for microbes, and the low microbial NO3 − demand under high NH4 + immobilization (Jansson et al. 1955; Puri and Ashman 1999). Indeed, rates of NH4 + immobilization into recalcitrant organic N in the low NO3 − addition plots were obviously greater than in the high NO3 − addition plots (Fig. 3). Thus, changes in rates of NH4 + immobilization into recalcitrant organic N, heterotrophic nitrification, and autotrophic nitrification could explain the nonlinear changes in NO3 − immobilization rates under increased NO3 − additions.

The importance of DNRA in N retention for tropical/subtropical forests and its response to N deposition vary with site conditions, and DNRA rates in situ may be underestimated by laboratory incubation (Baldos et al. 2015; Huygens et al. 2008; Silver et al. 2005; Templer et al. 2008; Zhang et al. 2013). Although DNRA rates (0.05 to 0.06 mg N kg−1 soil day−1), quantified by our laboratory incubation, were low, they were still equal to or higher than NO3 − leaching (0.04 ± 0.02 mg N kg−1 soil day−1) and denitrification activity (<0.015 mg N kg−1 soil day−1) measured in other tropical forests (Silver et al. 2005; Templer et al. 2008), highlighting the importance of DNRA in N retention in subtropical/tropical forest ecosystems. DNRA rates were similar among treatments, suggesting that increased NO3 − additions did not affect the rate of N retention via this pathway, even though NH4 + production declined significantly in the high NO3 − addition plots.

The effects of laboratory incubations of mixed, cold-stored soil on microbial N cycling

In the current study, gross N transformation rates were estimated by laboratory incubations of disturbed/mixed, cold-stored soil. To ensure uniform labeling, soil samples were sieved and homogenized. However, soil sieving could result in reductions in fungi-mediated transformations, e.g., heterotrophic nitrification and NO3 − immobilization because of its disruption of established mycelium networks (Huygens et al. 2007; Johnson et al. 2005; Zhu et al. 2013). However, mixing the soil might have increased exchangeable NH4 + production (Schimel et al. 1989). As suggested by Arnold et al. (2008), the cold storage and laboratory incubation of soil samples might have led to a reduction in gross N mineralization and NH4 + consumption, and an increase in gross nitrification and NO3 − immobilization. This occurred because microorganisms in subtropical/tropical soils are typically accustomed to relatively high temperatures with small fluctuations, and easily mineralizable organic N decreases during storage (Arnold et al. 2008). Zhang et al. (2013) found that DNRA rates in situ might be also underestimated by laboratory measurements. In brief, although laboratory measurement of gross rates of N transformations might allow identifying changes in N transformations between treatments (Paterson 2003), it did not reflect in situ N cycling rates (Arnold et al. 2008). Thus, 15N injection into intact soil cores, in situ incubation, and mineral N extraction in the field to estimate gross N transformations of forest soils are recommended.

Conclusions

Our work revealed that microbial mineralization of fast and slow turnover organic N pool respond differently to NO3 − additions, and high NO3 − addition treatment had a contrasting effect relative to low NO3 − addition treatment. After 2.5 years of treatments, mineralization of labile organic N first decreased, then increased, while mineralization of recalcitrant organic N changed oppositely in response to increasing NO3 − additions, showing support for our first hypothesis. Our results also identified the suggested mechanisms mediating microbial cycling of N in forest soils under enhanced N deposition: when NH4 + production via mineralization of fast turnover organic N was blocked, microbial mineralization of slow turnover organic N be upregulated, and vice versa. Exchangeable NH4 + immobilization (exchangeable NH4 + immobilization into labile and recalcitrant organic N, and gross NH4 + immobilization), however, were not changed significantly by increasing additions of NO3 −. These results suggests that in N-rich subtropical forests, enhanced NO3 − deposition might have more of an effect on NH4 + production than NH4 + consumption in the soil. Contrary to our second hypothesis, heterotrophic and gross nitrification did not change with increasing additions of NO3 −. However, it remains to be seen whether enhanced NO3 − deposition will affect heterotrophic nitrification in this acidic subtropical soil by decreasing soil C/N ratio and/or fungal biomass in the long term. Again, inconsistent with our third hypothesis, immobilization of NO3 − into recalcitrant organic N (NO3 − immobilization) decreased under increased NO3 − additions, with a greater reduction under low NO3 − addition treatment than high NO3 − addition treatment. Unexpectedly, autotrophic nitrification increased with increasing additions of NO3 −. Thus, we conclude that enhanced NO3 − deposition will increase the risk of N loss from this subtropical forest ecosystem by stimulating autotrophic nitrification, inhibiting NO3 − immobilization, or both. Microbial mechanisms controlling the changes in NO3 − production and soil retention of NO3 − under enhanced NO3 − deposition need to be explored.