Effects of Fe oxide on N transformations in subtropical acid soils

Subtropical ecosystems are often characterized by high N cycling rates, but net nitrification rates are often low in subtropical acid soils. NO3−-N immobilization into organic N may be a contributing factor to understand the observed low net nitrification rates in these acid soils. The effects of Fe oxide and organic matter on soil N transformations were evaluated using a 15N tracing study. Soil net nitrification was low for highly acidic yellow soil (Ferralsols), but gross ammonia oxidation was 7 times higher than net nitrification. In weakly acidic purple soil (Cambisols), net nitrification was 8 times higher than in Ferralsols. The addition of 5% Fe oxide to Cambisols, reduced the net nitrification rate to a negative rate, while NO3−-N immobilization rate increased 8 fold. NO3−-N immobilization was also observed in Ferralsols which contained high Fe oxides levels. A possible mechanism for these reactions could be stimulation of NO3−-N immobilization by Fe oxide which promoted the abiotic formation of nitrogenous polymers, suggesting that the absence of net nitrification in some highly acid soils may be due to high rates of NO3−-N immobilization caused by high Fe oxide content rather than a low pH.

Subtropical ecosystems are often characterized by high N cycling rates, but net nitrification rates are often low in subtropical acid soils. NO 3 2 -N immobilization into organic N may be a contributing factor to understand the observed low net nitrification rates in these acid soils. The effects of Fe oxide and organic matter on soil N transformations were evaluated using a 15 N tracing study. Soil net nitrification was low for highly acidic yellow soil (Ferralsols), but gross ammonia oxidation was 7 times higher than net nitrification. In weakly acidic purple soil (Cambisols), net nitrification was 8 times higher than in Ferralsols. The addition of 5% Fe oxide to Cambisols, reduced the net nitrification rate to a negative rate, while NO 3 2 -N immobilization rate increased 8 fold. NO 3 2 -N immobilization was also observed in Ferralsols which contained high Fe oxides levels. A possible mechanism for these reactions could be stimulation of NO 3 2 -N immobilization by Fe oxide which promoted the abiotic formation of nitrogenous polymers, suggesting that the absence of net nitrification in some highly acid soils may be due to high rates of NO 3 2 -N immobilization caused by high Fe oxide content rather than a low pH. S oils in subtropical and tropical forests are generally characterized by rapid N cycling rates and high N availability 1-3 . However, a few studies indicated that nitrification seems to be absent or a minor process in subtropical acid soils [4][5][6][7] . This interpretation was supported by two observations: one is that NO 3 2 concentrations are often consistently low in these soils, and that net nitrification rates are low or even negative during incubation assays of soil samples [6][7][8][9] . However, this does not necessarily mean that the nitrification does not occur in these subtropical acid soils. Alternatively, it is possible that NO 3 2 -N produced in these soils during nitrification (ammonia plus organic N oxidation) is immobilized through both abiotic and biotic processes as for observed in coniferous forest soils 10 . For example, high rates of NO 3 2 -N immobilization into organic N can be responsible for N enrichment in subtropical acid forest soils 3 which could protect the soil N from leaching. Furthermore, NO 3 2 -N can be converted readily to other inorganic forms through denitrification or dissimilatory NO 3 2 reduction to NH 4 1 (DRNA). Previous studies also suggested that N immobilization can be controlled by the concentration of available C 11 and available inorganic N 12,13 . Nugroho et al. 14 observed that net nitrification rates in acid soils can vary about three fold at similar soil pH but with different organic matter content. Of course, both biotic and abiotic processes may be significant or one process may dominate, depending on the soil conditions. However, the causes of variation of N immobilization between soil types are not well known 15 .
Subtropical and tropical acid soils are often characterized by high levels of Fe oxides. It has been hypothesized that Fe may play a key role in regulating NO 3 2 immobilization in acid forest soils 16 . Likewise, the inhibition of nitrification by high amounts of iron from pyrite (FeS 2 ) can also occur 17 . Shindo and Huang 18 found that Mn oxide promoted the abiotic formation of nitrogenous polymers. Since Fe and Mn oxides have similar chemical properties, it is possible that Fe oxide can also promote organic N formation from NO 3 2 Therefore, Fe oxide may be a primary player in N transformations in subtropical acid soils and may help to explain the low NO 3 content. The effects of Fe oxide or organic substances addition on N transformations in these soils were estimated using a 15 N tracing method and N transformation model 19 . -N concentrations did not change significantly for the Ferralsols control and its two treatments ( Figure 1A). However, NH 4 1 -N dynamics for the Cambisols differed from Ferralsols, as Fe oxide and organic substances addition caused higher significant effects for Cambisols than Ferralsols. At half an hour, after 50 mg kg 21 15 NH 4 NO 3 was introduced, NH 4 1 -N concentrations decreased rapidly for the Cambisols control, and further declined during the 7-day incubation (p , 0.05). A similar trend was observed when organic substances were added to Cambisols. However, Fe oxide addition did not change NH 4 1 -N concentrations throughout the 7day incubation ( Figure 1B).

Effects of
Half an hour after the addition of 50 mg kg 21 NH 4 15 NO 3 , NO 3 2 -N concentrations for the Ferralsols control averaged 9.89 6 0.36 mg NO 3 2 -N kg 21 . Addition of organic substances or Fe oxide did not affect NO 3 2 -N concentrations. During the 7 days incubation period, NO 3 2 -N concentrations did not change significantly, and did not respond to Fe oxide or organic substances addition in the Ferralsols control ( Figure 1C). Conversely, NO 3 2 -N dynamics in the Cambisols differed were significantly from the one in the Ferralsols. Half an hour, after 50 mg kg 21 NH 4 15 NO 3 application, NO 3 2 -N concentrations for the Cambisols control averaged 10.5 6 0.20 mg N kg 21 . NO 3 2 -N concentrations increased significantly for the Cambisols control and organic substances treatment through time, but decreased significantly in the Fe oxide treatment during the 7 days incubation ( Figure 1D).
Effects of Fe oxide or organic substances addition on N transformation rates. Our results clearly showed that soil net nitrification rates, gross autotrophic nitrification (ammonia oxidation) rates, NH 4 1 immobilization rates and NO 3 2 immobilization rates were significantly affected by Fe oxide addition.
The net nitrification rate for the Ferralsols averaged 0.05 6 0.01 mg N kg 21 soil day 21 , which was significantly lower than for Cambisols (0.41 6 0.04 mg N kg 21 soil day 21 ), and the organic  substances treatment showed a non-significant decreasing trend in nitrification rates (p . 0.05). However, Fe oxide addition decreased the net nitrification rate significantly from 0.41 6 0.04 to 0.28 6 0.03 mg N kg 21 soil day 21 for Cambisols but not for Ferralsols.
The gross ammonia oxidation rates were 0.21 6 0.03 mg N kg 21 day 21 and 0.74 6 0.05 mg N kg 21 day 21 for the Ferralsols control and Cambisols respectively (p , 0.05) ( Figure 2B). Furthermore, Fe oxide addition significantly decreased the gross NH 4 oxidation rate in Cambisols, whereas the organic substances amendment did not change gross ammonia oxidation (p . 0.05). Neither Fe oxide nor organic substances amendments changed gross NH 3 oxidation rates in the Ferralsol (Figure 2A).
Fe oxide or organic substances amendments did not change heterotrophic nitrification rates for the Ferralsols ( Figure 2C). However, the Fe oxide treatment increased the heterotrophic nitrification rate for Cambisols ( Figure 2D). NO 3 2 immobilization rates for Cambisols were not significantly affected by organic substances addition ( Figure 3D) but they significantly increased by Fe oxide addition. NO 3 2 immobilization was also observed for the Ferralsol treatments, including the control and both Fe oxide and organic substances treatments. However, Fe and organic substances amendments did not significantly change NO 3 2 immobilization rates for Ferralsols compared to the unamended control ( Figure 3C).
Gross NH 4 1 immobilization was four times higher for Cambisols than Ferralsols ( Figure 3A, B). Hematite addition significantly decreased the NH 4 1 immobilization rate for Cambisols, and the organic substances treatment also showed a decreasing trend but the difference was not significant (p . 0.05). NH 4 1 immobilization rates for Ferralsols did not change in response to Fe oxide and organic substances addition.
Gross N mineralization rates did not differ significantly between Ferralsols and Cambisols in the Fe oxide and organic substance treatment (p . 0.05), although Fe oxide addition tended to decrease gross N mineralization for both soils (Figure 4).
Ammonia-oxidizing bacterial (AOB) and ammonia-oxidizing archaeal (AOA) amoA gene copies. Abundance of AOB and AOA was estimated by quantifying their respective amoA gene copy numbers after pre-incubation ( Figure 5). The organic substances treatment tended to decrease AOB abundance for both Ferralsols and Cambisols but results were not significant (p . 0.05). However, Fe oxide addition significantly decreased AOB amoA gene copy numbers by about 50% for Cambisols. For Ferralsols, Fe oxide addition also tended to decrease AOB abundance, but not significantly. The abundance of AOB was significantly higher for Cambisols than Ferralsols, but the difference was not significant after 5% Fe oxide addition.     AOA amoA gene copy numbers were significantly higher for Cambisols than Ferralsols ( Figure 5). Fe oxide addition decreased AOA amoA gene copy numbers for Cambisols to a similar level in Ferralsols. For Ferralsols, Fe oxide addition also did not change AOA abundance significantly. Similar to AOB, AOA abundance showed a non-significant decreasing trend after the organic substances addition for both Ferralsols and Cambisols (p . 0.05).

Discussion
While net nitrification was low for the highly acidic Ferralsols, gross autotrophic nitrification rates were about three times higher than net nitrification rates. The net nitrification rate was 8 times higher for weakly acidic Cambisols than highly acidic Ferralsols. However, the net nitrification rate decreased to a negative value for the weakly acidic Cambisols after Fe oxide addition. Results clearly showed that soil net nitrification rate was significantly inhibited by Fe oxide addition. However, only the evaluation of the associated individual gross N rates sheds light whether the oxidation to NO 3 2 or immobilization of NO 3 2 was responsible for this result. Nitrification could be retarded by high amounts of Fe from pyrite (FeS 2 ). Blaise 17 reported that nitrification was inhibited by 40.3% with the addition of 10000 mg pyrite kg 21 soil at the end of a 30-day incubation. However, they were unable to identify the exact mechanism by which pyrite inhibits nitrification/immobilization. Furthermore, postulated that the effect of pyrite could be due to the toxic action of any one or a combination of the following: (1) sulphides 20 ; (2) the oxidized forms of the sulphides, or (3) presence of Fe 21 ions 21 . Compared to their results, a stronger inhibition of net nitrification by 50000 mg hematite kg 21 soil was observed in the present study. However, the mechanisms they postulated for pyrite inhibition of nitrification did not seem to be applicable for the present study since S or sulphides were not involved. Therefore, other mechanisms must be responsible for ferric iron inhibition of nitrification.
The toxicity of Fe to nitrifying microorganisms may partly contribute to the depression of nitrification. A pioneer study by Meiklejohn 22 indicated that small amounts of Fe (0.1-0.6 mg L 21 ) stimulated growth of nitrifying bacteria and increased the oxidation of NH 3 to NO 2 2 , whereas high concentrations of Fe (.112 mg L 21 ) was toxic to nitrifying bacteria. The present results clearly demonstrated that both amoA genes for the ammonia monooxygenase (AMO) of AOB and AOA decreased two or three times after Fe oxide addition to Cambisols, and the soil gross ammonia nitrification rate also decreased. Higher amoA gene copy numbers of AOB and AOA were observed for Cambisols than for Ferralsols. Since Fe oxide content of Ferralsols (50.1 mg kg 21 soil) was significantly higher than for Cambisols (16.6 mg kg 21 soil), AOB and AOA were likely inhibited by high Fe oxide content. However, Fe toxicity to AOB and AOA cannot explain the decrease of gross ammonia nitrification completely, because the average activity was estimated around 40 fmol NH 3 cell 21 day 21 for AOA and AOB 23,24 . The gene copy numbers of AOB and AOA after Fe oxide addition still have potential nitrification up to 20 mg N kg 21 soil day 21 .
The gross NO 3 2 -N immobilization rate for Cambisols was 8 times higher in the 5% Fe oxide compared to the control treatment. In general immobilization of NO 3 2 is assumed to be assimilated into microbial biomass. However, higher NO 3 2 -N immobilization induced by Fe oxide in the present soils can not be explained by the assimilation of NO 3 2 -N into microbial biomass. This was supported by the fact that NH 4 , rather than NO 3 , is the preferred form of N for assimilation by soil microorganisms [25][26][27] . Further evidence was that the gene abundance of nitrifying microorganisms decreased after Fe oxide addition ( Figure 5). Results from upland tropical forest soils in America also showed that microbial biomass 15 N did not increase after NO 3 2 addition 28 . Positive charges on the soil surface may also increase NO 3 2 -N immobilization under field conditions [29][30][31] . Anion exchange capacity (AEC) increased significantly after Fe oxide addition (Table 1) A potential mechanism of abiotic immobilization of NO 3 2 has been postulated, and the authors suggested that Fe plays a key role in NO 3 2 -N immobilization by promoting organic N formation from inorganic N 16 . Shindo and Huang found that Mn oxide significantly promoted the abiotic formation of nitrogenous polymers in the pH range 4 to 8 within 24 hours 18 . Since Fe and Mn oxides have similar chemical properties, it is possible that Fe oxide also functions similarly and can thus promote organic N formation. The mechanism for this organic N formation may explain the N dynamics for the two soils evaluated in our study. In a recent study, Zhang et al. reported that the gross autotrophic nitrification ranged from 0 to 0.5 mg N g 21 d 21 and the NO 3 2 -N immobilization rate into organic N ranged from 0.11 to 1.71 mg N g 21 d 21 in a subtropical acid forest soil 3 . Their results indicated high rates of NO 3 2 -N immobilization into organic N were responsible for the N enrichment in humid subtropical forest soils which contained large amounts of Fe and Mn oxides. A higher net nitrification rate occurred for weakly acidic Cambisols than for highly acidic Ferralsols in our study, and the NO 3 2 -N immobilization was higher for Ferralsols than for Cambisols. The Fe oxide content of Ferralsols (50.1 g kg 21 soil) was higher than for Cambisols (16.6 g kg 21 soil), which implied that the absence of nitrification in some highly acid soils may due to high amounts of NO 3 2 -N immobilization stimulated by high Fe oxide concentrations.
Soil net nitrification, gross NH 4 1 oxidation, total gross NH 4 1 mineralization and NO 3 2 -N immobilization did not change significantly by organic substances addition for both soils. The insignificant response of N transformations by C/N ratio or organic substances addition may be due to the short incubation period which was not enough to highlight possible effects.
Cambisols and Ferralsols formed from the same parent material in the same climate, but differing in pH. While net nitrification occurred for the slightly acidic Cambisols, no obvious net nitrification was observed for highly acidic Ferralsols. The soil net nitrification rate strongly decreased to a negative value after Fe oxide addition to the slightly acidic Cambisols. However, significant autotrophic nitrification was observed using 15 N tracers for both soils. Fe oxide greatly increased NO 3 2 -N immobilization, probably by promoting the abiotic formation of nitrogenous polymers. This suggests that the absence of net nitrification in some highly acid soils may due to high amounts of NO 3 2 -N immobilization caused by the high Fe oxide content rather than the low pH.

Methods
Site description and soil sampling. The yellow soil samples (Ferralsols, Xanthic) were collected from a pine forest soil derived from a sandstone parent material in Jingyun Mountain, Chongqing, China (29u839N,106u399E). In this region, the annual mean temperature is 19.6uC and annual rainfall is 1611 mm. Middle subtropical evergreen broad-leaved forests characterize the vegetation of this area. Purple soils (Cambisols, Chromic) were collected at a nearby hill. The Ferralsols samples were highly weathered and experienced a long developing process. To the contrary, Cambisols were slightly weathered and have not experienced a long developing process because of soil erosion. Selected soil properties for two soils and each treatment were listed in Table 1.
Four field replicates were taken at each site to a 0-20 cm depth in October, 2012, with each replicate being composed of seven individual soil cores (diameter was 5.5 cm), which were pooled and homogenized to reduce heterogeneity. Samples for each of the four replicates per site were were air-dried and separated into two parts. One was ground to pass a 2-mm sieve and used to make subsamples for incubation, another was ground to pass a 1-mm sieve and used for chemical analyses.
Preparation of Fe oxide and organic substances treatments. The precipitates of hematite were prepared from ferric chlorite of at least 98% purity, and solutions containing 100 g of FeCl 3 ?6H 2 O/L were prepared. Ferric oxides were precipitated by the addition of NH 3 to the FeCl 3 solution until pH 7.0. The resultant suspension was thoroughly dialyzed for 24 hours at which time no further Cl 2 and Fe 31 was detected in the external solution, followed by aging for 48 hours in a 160uC oven 32 . Obtained   hematite was checked by X-ray powder diffractometer (XRD) (see Supplementary figure 1). The organic substances were collected from the organic layer of the Ferralsols, and sieved to pass a sieve ,2 mm to remove coarse roots, mixed thoroughly and then air dried (20uC) for 3 days before use. Subsamples were obtained by adding 5% hematite or 5% organic substances into Ferralsols or Cambisols, respectively. Six subsamples were obtained as follows: Ferralsols 1 5% hematite, Ferralsols 1 5% humic substances, Cambisols 1 5% hematite and Cambisols 1 5% humic substances. Subsamples were air-dried, and separated into two parts. One was ground to pass a 2-mm sieve and stored at 4uC for 2 months before use, another was ground to pass a 1-mm sieve and used for chemical analyses.
Physical and chemical analyses. Soil pH was measured in a soil to water ratio of 152.5 (v/v) by a DMP-2 mV/pH detector (Quark Ltd, Nanjing, China). Total N (TN) and soil organic matter (SOM) contents were determined by a Macro Elemental Analyzer (Elementar Analysensysteme GmbH, Hanau, Germany). Total soil Fe was extracted with HNO 3 -HF-HClO 4 , and free Fe oxides were extracted with Na 2 S 2 O 4 -Na 3 C 6 H 5 O 7 -NaHCO 3 . Fe content was measured by atomic absorption spectrophotometry with a graphite furnace (GFAAS) using a model Z-8200 spectrophotometer. Soil cation exchange capacity (CEC) and anion exchange capacity (AEC) were determined after extraction with 0.0215 mol (NH 4 NO 3 ) L 21 solution using the method described by Gillman & Sumpter 33 .
Experimental design and 15 N addition. The soil subsamples were adjusted to 60% water holding capacity (WHC) and pre-incubated for 7 days at 25uC. We employed a combination of 15 N tracing experiments and full process-based N cycle models to quantify process-specific and pool-specific N transformation rates, which is the standard method for the quantification of N dynamics in soils 19,[34][35][36][37] . For all samples, 15 N tracing studies were carried out under controlled conditions. For the 15  NO 3 solution was added to each conical flask at a rate of 7.14 mmol N g 21 soil (50 mg NH 4 1 -N g 21 soil and 50 mg NO 3 2 -N g 21 soil). The soil was adjusted to 60% waterholding capacity (WHC) and incubated for 7 days at 25uC. All bottles were covered with polyethylene film punctured with needle holes to maintain aerobic conditions. The soils (three replications for each treatment) were extracted at 0.5, 24, 72, 120 and 168 hours after fertilizer application to determine the concentrations and isotopic composition of NH 4 1 and NO 3 2 . For isotopic analysis, NH 4 1 and NO 3 2 were separated by distillation with Mg oxide and Devarda's alloy 38,39 (see Supplementary information). The isotopic composition of NH 4 1 and NO 3 2 were measured using an automated C/N analyzer coupled to an isotope ratio mass spectrometer (Europa Scientific Integra, UK). Simultaneous gross N transformations in soil were quantified using a process-based 15 N tracing model 19 (see Supplementary information).
DNA extraction & quantitative PCR assay. Right after pre-incubation, 4 replicates of each treatment were randomly selected to extract DNA and amoA genes were analyzed by quantitative PCR (qPCR). The DNA was extracted for three sub-samples from 0.50 g of soil with the FastDNA Spin Kit for soil (MP Biomedicals, United States), according to the protocol of the manufacturer. The quality and quantity of the DNA extracts were determined with a spectrophotometer (Nanodrop, PeqLab, Germany), and were pooled and stored at 220uC until use. Quantitative PCR of amoA genes was performed to estimate the abundance of the ammonia-oxidizing bacterial and archaeal communities, respectively (see Supplementary information).
Statistical analyses. Data (measured or calculated) were subjected to one-way ANOVA and mean values were separated using Duncan's New Multiple Range Test at p , 0.05. All statistical analyses were performed by SPSS statistical package.