Evidence for biological nitrification inhibition in Brachiaria pastures

Abstract

Nitrification, a key process in the global nitrogen cycle that generates nitrate through microbial activity, may enhance losses of fertilizer nitrogen by leaching and denitrification. Certain plants can suppress soil-nitrification by releasing inhibitors from roots, a phenomenon termed biological nitrification inhibition (BNI). Here, we report the discovery of an effective nitrification inhibitor in the root-exudates of the tropical forage grass Brachiaria humidicola (Rendle) Schweick. Named “brachialactone,” this inhibitor is a recently discovered cyclic diterpene with a unique 5-8-5-membered ring system and a γ-lactone ring. It contributed 60–90% of the inhibitory activity released from the roots of this tropical grass. Unlike nitrapyrin (a synthetic nitrification inhibitor), which affects only the ammonia monooxygenase (AMO) pathway, brachialactone appears to block both AMO and hydroxylamine oxidoreductase enzymatic pathways in Nitrosomonas. Release of this inhibitor is a regulated plant function, triggered and sustained by the availability of ammonium (NH₄⁺) in the root environment. Brachialactone release is restricted to those roots that are directly exposed to NH₄⁺. Within 3 years of establishment, Brachiaria pastures have suppressed soil nitrifier populations (determined as amoA genes; ammonia-oxidizing bacteria and ammonia-oxidizing archaea), along with nitrification and nitrous oxide emissions. These findings provide direct evidence for the existence and active regulation of a nitrification inhibitor (or inhibitors) release from tropical pasture root systems. Exploiting the BNI function could become a powerful strategy toward the development of low-nitrifying agronomic systems, benefiting both agriculture and the environment.

Most modern agricultural systems are based on large inputs of inorganic nitrogen (N), with ammonium (NH₄⁺) being the primary N source (1, 2). Also, current crop management practices result in the development of highly nitrifying soil environments (3, 4). Nitrification results in the transformation of the relatively immobile NH₄⁺ to highly mobile nitrate (NO₃⁻), making inorganic N susceptible to losses through leaching of NO₃⁻ and/or gaseous N emissions, potentially initiating a cascade of environmental and health problems (1, 2, 5, 6). Nitrous oxide (N₂O) is one of the three major biogenic greenhouse gases contributing to global warming, produced primarily from denitrification processes in agricultural systems (5, 7). Also, assimilation of NO₃⁻ by plants can result in further N₂O emissions directly from plant canopies (8). The low agronomic N-use efficiency (NUE) found in many agricultural systems is largely the result of N losses associated with nitrification (i.e., N losses from NO₃⁻ leaching and denitrification) (9–11). Most plants have the ability to assimilate both NH₄⁺ and NO₃⁻ (12); therefore, nitrification does not need to be a dominant process in the N cycle for efficient N use.

Nitrification is low in some forest and grassland soils (13–17). Since the early 1960s, some tropical grasses have been suspected of having the capacity to inhibit nitrification (18–21). However, this concept remained controversial due to the lack of direct evidence showing such inhibitory effects or the identification of specific inhibitors (22).

We adopted a very sensitive bioassay using a recombinant luminescent Nitrosomonas europaea to detect biological nitrification inhibition (BNI) in plant–soil systems with the inhibitory activity of roots expressed in allylthiourea units (ATU) (23). Using this methodology, we were able to show that certain plants release nitrification inhibitors from their roots (23–26). Such BNI capacity appears to be relatively widespread among tropical pasture plants, with Brachiaria spp. showing the highest capacity among the pasture grasses tested (24). The potential for high BNI capacity also exists in wild wheat (26). A pasture grass, Brachiaria humidicola (Rendle) Schweick, native to tropical Africa and grown extensively in tropical South American grasslands, releases substantial amounts of BNIs from its roots, ranging from 17 to 50 ATU per gram of root dry weight per day (23, 24). Here, we report the identity of the major nitrification inhibitor (which is a recently discovered cyclic diterpene), its most likely mode of inhibitory action, the regulatory nature of its release from roots, and the effectiveness of this BNI function in suppressing soil nitrification and N₂O emissions from Brachiaria pastures. The potential implications of the BNI function limiting N losses from agricultural systems and in reducing the ecological footprint of food production are indicated.

Results and Discussion

Isolation, Characterization, and Structure Determination of Brachialactone.

Bioassay-guided fractionation of the root exudates achieved the isolation of a cyclic diterpene, which we named “brachialactone” (Fig. 1 or its enantiomer). It has a unique dicyclopenta[a,d]cyclooctane skeleton (5-8-5 ring system) with a γ-lactone ring bridging one of the five-membered rings and the eight-membered ring. Similar 5-8-5 tricyclic terpenoids (ophiobolanes and fusicoccanes) are found in fungi and plants (27). However, to our knowledge, any derivative that has a lactone ring is novel. Fusicoccane-type cyclic diterpenes are biologically synthesized from geranylgeranyl diphosphate by two-step cyclization catalyzed by terpene cyclases (27, 28). Fusicoccins can act as activators of H⁺-ATPases in plants by modifying the function of 14-3-3 proteins and also exhibit unique biological activity (i.e., anticancer activity) in animal cells (28).

Fig. 1.

Chemical structure of brachialactone, the major nitrification inhibitor isolated from root exudates of B. humidicola.

Inhibition of nitrification in our in vitro assay system of N. europaea was linearly related to brachialactone concentration over the range of 1.3–13.3 μM (r² = 0.99, P < 0.01; Fig. 2A). Brachialactone, with an ED₈₀ of 10.6 μM, should be considered a potent nitrification inhibitor when compared with nitrapyrin or dicyandiamide, two of the most widely used synthetic nitrification inhibitors (ED₈₀ of 5.8 μM for nitrapyrin and 2,200 μM for dicyandiamide). Contribution of brachialactone to the total inhibitory activity in these root exudates ranged from 60% to 96% (r² = 0.91, P < 0.01; Fig. 2B and Table S1).

Fig. 2.

Inhibition of nitrification by brachialactone and contribution of brachialactone to BNI activity released from roots. (A) Inhibitory effect of brachialactone on N. europaea in an in vitro assay. (B) Contribution of brachialactone to the BNI activity released from roots (i.e., in root exudates) of B. humidicola. Root exudates were collected from intact plants using 1 L of aerated solution of 1 mM NH₄Cl with 200 μM CaCl₂ over 24 h. Each data point represents root exudates collected from hydroponically grown plants in a glasshouse during March to May of 2007 and 2008.

Mode of Inhibitory Action of Brachialactone on Nitrosomonas.

Brachialactone inhibited Nitrosomonas function, possibly by blocking both enzymatic pathways, ammonia monooxygenase (AMO) and hydroxylamino oxidoreductase (HAO), that are involved in ammonia oxidation; however, the inhibitory effect on the HAO pathway is less than its effect on the AMO pathway (Table 1; for details, see Table S2). Crude extract of root exudates containing BNI activity showed an inhibitory effect of similar strength on both enzymatic pathways (Table 1; Table S2), indicating that other BNIs released from roots have a mode of action different from that of brachialactone. Recently, linolenic acid, a major BNI compound present in the leaf tissue of B. humidicola, was shown to block both AMO and HAO enzymatic pathways in a manner similar to BNI activity of crude root exudates, indicating the possibility of a single inhibitor affecting both the enzymatic pathways in Nitrosomonas (29). When a fatty acid binding protein, BSA, was added (after the addition of linolenic acid) to the Nitrosomonas pure cultures, a major portion of the inhibitory effect was removed, indicating the reversible nature of the inhibitory effect from linolenic acid (29). The reducing power generated from the oxidation of hydroxylamine by HAO is thought to pass through cytochrome c-554 to both cytochrome aa₃ oxidase and ubiquinone (30), which is subsequently used for the reduction of NAD(P)⁺, as well as for the maintenance of the AMO reaction (31). We cannot as yet rule out the possibility of brachialactone disrupting the generation of reductive power, NAD(P)H₂, by interfering directly with the electron transfer pathways of the cytochrome chain in the inner membrane of Nitrosomonas; thus, loss of light emission, which is independent of the direct effect on enzymatic pathways (i.e., AMO and HAO). In contrast, the inhibitory effect of nitrapyrin, a synthetic nitrification inhibitor, was nearly eliminated when hydroxylamine was added to the assay, indicating that only the AMO enzymatic pathway was affected (Table 1), which is in agreement with its known mode of action (32). High concentrations of monoterpenes found in conifer forest systems are reported to suppress nitrifier activity by blocking the AMO pathway (33). Only a few compounds—phenyl, methyl, or hydroxyethyl hydrazine and hydrogen peroxide—are known to inhibit the HAO enzymatic pathway in Nitrosomonas (34). Most commercial nitrification inhibitors (such as dicyandiamide or nitrapyrin) suppress nitrifier activity by targeting primarily the AMO pathway; thus, they could be vulnerable to genetic changes in nitrifier populations or to natural genetic diversity in ammonia-oxidizers (AOs) (35, 36). Given the inherent genetic variability in nitrifier populations (35), it is likely that BNIs released from Brachiaria spp. will be less vulnerable to genetic changes due to their more diverse modes of action on Nitrosomonas.

Table 1.

Inhibitory strength of brachialactone on AMO or HAO enzymatic pathways of N. europaea

Compound	Concentration in the in vitro assay, μM	Inhibition, %
		AMO pathway	HAO pathway
BNI-root exudate (crude methanol extract)	—	63.4 ± 0.8	63.8 ± 0.8
Brachialactone	5.0	59.7 ± 0.9	37.7 ± 0.9
Nitrapyrin	3.0	82.3 ± 1.5	8.1 ± 1.2

Root exudate was collected from intact BH (CIAT 679) plants (root fresh weight of ≈20 g) using aerated solutions of 1 mM NH₄Cl, evaporated to dryness, extracted with methanol, and evaporated to dryness and dissolved in 200 μL of dimethyl sulfoxide; 1 μL of the crude extract was used for the determination of BNI activity. Values are means ± SE from four replications.

Influence of NH₄⁺ on the Release of Brachialactone.

Assuming that the BNI function evolved as a mechanism to conserve N by limiting nitrification and the associated N losses, we hypothesized that the inhibitory activity would respond to the presence of NH₄⁺ in soil, because its availability determines the extent of soil nitrifier activity (37). We showed earlier that BNIs are released from roots of B. humidicola plants when the sole N source was NH₄⁺, but not when they were grown with NO₃⁻ (38). Results from the present study demonstrate that N form (NH₄⁺ vs. NO₃⁻) in the root environment has a major influence on the release of brachialactone, which is accelerated only in the presence of NH₄⁺ (Table 2).

Table 2.

Influence of NH₄⁺ in the root environment on the release of BNI activity and brachialactone from roots of Brachiaria humidicola grown with NH₄⁺ as the N source

Exudate-collection solution	BNI activity released from the roots, ATU per g root dry wt per day	Brachialactone released from the roots, μg per g root dry wt per day
Distilled water	6.6 ± 1.2	1.3 ± 0.1
Acidified water (pH 3.0)	4.8 ± 0.4	1.6 ± 0.6
KNO3 (1 mM)	ND	ND
NH4Cl (1 mM)	13.9 ± 1.6	8.3 ± 0.2

All collection solutions contained 200 μM CaCl₂. Values are means ± SE from three replications. ND, not detected.

For inhibitors to be most effective, their release should be concentrated in the area where the root system is exposed to NH₄⁺. To test this hypothesis, we used a split-root system in which plants were initially grown with (NH₄)₂SO₄ as the sole N source. Then one half of the root system was exposed to NH₄⁺ and the other half to NO₃⁻ in separate tanks. Release of brachialactone was triggered only in the part of the root system exposed to NH₄⁺, and not in the entire root system (Table 3), indicating a localized release response.

Table 3.

Influence of nitrogen form (NH₄⁺ vs. NO₃⁻) in the exudate collection solutions on the release of BNI activity and brachialactone from the roots of B. humidicola in a split-root system

Split-root system treatment	BNI activity released from the roots, ATU per g root dry wt per day	Brachialactone released from the roots, μg per g root dry wt per day
The half of the root system exposed to NH4+	9.5 ± 1.4	9.4 ± 0.2
The half of the root system exposed to NO3−	ND	ND

Values are means ± SE from three replications.

Validation of Effectiveness of BNI Function in Suppressing Soil Nitrification and N₂O Emissions.

Based on conservative estimates of the live root biomass from a long-term grass pasture being ≈1.5 Mg·ha⁻¹ (39) with a BNI capacity of 17–50 ATU per g of root dry wt per day (24), we estimate that BNI activity of 2.6 × 10⁶ to 7.5 × 10⁶ ATU·ha⁻¹·day⁻¹ can potentially be released from B. humidicola roots. This estimate amounts to an inhibitory potential equivalent to the application of ≈6.2–18 kg of nitrapyrin per ha per year (based on 1 ATU being equal to 0.6 μg of nitrapyrin), which is large enough to have a significant influence on the function of soil nitrifier populations and nitrification rates.

Field studies at Centro Internacional de Agricultura Tropical (CIAT; Palmira, Colombia) indicated a 90% decline in soil ammonium oxidation rates (Fig. 3) due to extremely small populations of nitrifiers [AO bacteria (Fig. 4B) and AO archaea (Fig. 4A); determined as amoA genes] in B. humidicola plots within 3 years of establishment (Fig. 4 A and B). Ammonium availability was relatively high (≥10 mg of NH₄ N per kg of soil) in the field plots during the experimental period of 2005 to 2007 (Fig. S1). This observation suggests that the extremely low nitrifier populations observed in B. humidicola plots were not due to lack of soil ammonium nitrogen. However, there was little effect on the total soil bacterial population (expressed as gene copy number) (Fig. 4C), indicating the highly specific nature of the inhibitory effect toward the AO bacterial populations. Nevertheless, the archaea population in general is suppressed by Brachiaria sp. and Panicum sp. pastures (Fig. 4D), indicating that inhibitors produced by the root systems of these pastures may not be entirely specific to AO archaea. N₂O emissions were also suppressed >90% in field plots of B. humidicola (CIAT 16888) compared with plots of soybean, which lacks BNI capacity, or control plots (plant-free field-plots) (Fig. 5). Two other pasture grasses, Panicum maximum and Brachiaria spp. hybrid cv. Mulato that have a low to moderate level of BNI capacity (3 to 10 ATU g⁻¹ root dwt d⁻¹), showed only an intermediate level of inhibitory effect on soil ammonium oxidation rates (Fig. 3).

Fig. 3.

Soil ammonium oxidation rates (mg of NO₂⁻ N per kg of soil per day) in field plots planted with tropical pasture grasses (differing in BNI capacity) and soybean (lacking BNI capacity in roots) [over 3 years from establishment of pastures (September 2004 to November 2007); for soybean, two planting seasons every year and after six seasons of cultivation]. CON, control (plant-free) plots; SOY, soybean; PM, P. maximum; BHM, Brachiaria hybrid cv. Mulato; BH-679, B. humidicola CIAT 679 (standard cultivar); BH-16888, B. humidicola accession CIAT 16888 (a germ plasm accession). Values are means ± SE from three replications.

Fig. 4.

Influence of tropical pasture grass cultivation (in field plots, over 3 years: September 2004 to November 2007) on soil microorganism populations at 1 day after ammonium sulfate fertilization by estimating copy number of AOB amoA genes (A); AOA amoA genes (B); bacterial small-subunit (SSU) rRNA genes (C); and archaea SSU 16S rRNA genes (D). Plots are identified in Fig. 3 legend. Gene copy number was expressed as copy number per gram of dried soil and obtained through absolute quantification by using real-time PCR. Values are means ± SE from three replications.

Fig. 5.

Cumulative N₂O emissions (mg of N₂O N per m² per year) from field plots of tropical pasture grasses (monitored monthly over a 3-year period, from September 2004 to November 2007). Plots are identified in Fig. 3 legend. Values are means ± SE from three replications.

Conclusions and Perspectives

These results provide evidence for the existence of a plant-controlled mechanism by which nitrification inhibitors are produced and delivered by roots to soil-nitrifier sites. Here, a biological molecule providing a major portion of the BNI activity from Brachiaria root systems has been identified and characterized, and its release has been shown to be a tightly controlled physiological function. This report solves nearly 3 decades of mystery surrounding the low nitrification rates found in Brachiaria-dominated tropical pastures.

Fertilizer-N consumption is expected to reach 200 Tg·y⁻¹ by 2025 from the present 150 Tg·y⁻¹ (2, 40, 41). Environmental damage could result, given the pervasive inefficiencies in N use (<40% of applied N is recovered by most field crops) (2, 42), due largely to nitrification and its associated processes (1, 13). The economic implications of this “wasted N” can be enormous; they are currently estimated as US$ 17 billion from cereal production systems alone (13, 43). N₂O emissions from agricultural systems are expected to reach 25.7 Tg of N per year by 2025, contributing significantly to global warming (1, 7, 10, 44).

Our results suggest potential differences in N₂O emissions among plant species (Fig. 5) linked to their differential BNI capacities (Table S4) (24); however, such differences are not presently considered by the Intergovernmental Panel on Climate Change in their estimations of projected N₂O emissions from agricultural systems (45). For example, South American savannas occupy ≈250 million (M)ha, largely under native grass or pastures of introduced Brachiaria spp. (39). These pastures could be of low nitrifying (24) and low N₂O emitting systems. If these grasslands were converted to soybean and maize, crops that lack BNI capacity (24, 46), there could be major implications for N₂O emissions. Approximately 11 Mha of pastoral land in the Cerrados region of Brazil has already been converted to soybean and maize (37, 47), and an additional 35–40 Mha could suffer such conversion. Such land-use changes could have major consequences on N₂O emissions from this region.

Given the current environmental concerns, it is desirable to develop new technologies and approaches for combating the rampant and rapid nitrification in agricultural systems to reduce N pollution and improve NUE (1). Development of improved forage grasses for low nitrifying pasture-based production systems is possible given the significant genetic variability found for the BNI function within the Brachiaria spp. (24). Also, introducing high BNI capacity from wild wheat (Leymus racemosus) into cultivated wheat could be an option in the foreseeable future (26, 48). A fundamental shift toward NH₄⁺-dominated crop nutrition can be achieved by using crops and pastures that have high BNI capacity or integrating annual crop production with a high BNI-capacity forage component, resulting in low-nitrifying agronomic production systems, benefiting both agriculture and the environment.

Materials and Methods

Bioassay-Guided Fractionation and Structure Elucidation.

Isolation of brachialactone from root exudates.

B. humidicola (Rendle) Schweick (CIAT 679) plants were raised hydroponically in the greenhouse (23). Root exudates were collected by using aerated solutions of NH₄Cl (1 mM), the water was evaporated, and the residue was extracted with dichloromethane (23). The dichloromethane was evaporated, and the residue was dissolved in methanol and separated by using an HPLC system (Tosoh 8020 photodiode array system) equipped with a C₁₈ reverse-phase column (Tosoh TSKgel SuperODS). BNI activity of HPLC effluents was determined (23). The major BNI-active compound was eluted in the 20–50% acetonitrile gradient in water. The active fractions were pooled and further purified by repeated HPLC to yield 0.7 mg (purity, >95% on HPLC) as an amorphous powder.

Characterization and structure determination of brachialactone.

The UV and CD spectra of brachialactone in methanol at room temperature were recorded on a UV-1600 spectrophotometer (Shimadzu) and a J-820 spectropolarimeter (JASCO), respectively. As shown in Fig. S2, the UV absorption spectrum showed the absorption maximum at a wavelength of 230 nm (ε = 2.4 × 10³ M⁻¹·cm⁻¹). CD spectrum gave a positive band at 232 nm ([θ] = +6.5°·cm²·dmol⁻¹, 0.027 mmol/100 mL).

Brachialactone was dissolved in methanol and applied to EI mass spectrometry on a QP2010 (Shimadzu) at 70 keV by direct inlet. The peaks observed (m/z, %) were 334 (M⁺, 17), 291 (9), 245 (5), 227 (5), 199 (11), 149 (21), 137 (100), 136 (75), 121 (47), and 107 (22) (Fig. S3). For the accurate mass analysis, the sample was diluted using 50% aqueous methanol including 1% acetic acid and measured in the positive-ion mode on an ESI-FTICR mass spectrometer (ApexII 70e; Bruker Daltonics). The measurement gave the protonated molecular ion at m/z 335.2213 corresponding to C₂₀H₃₁O₄ (335.2217) with 1.2 × 10⁻⁶ error [M + H]⁺. These results contributed to determine the molecular formula of brachialactone to be C₂₀H₃₀O₄.

The ¹H, ¹³C and 2D NMR spectra were measured in CD₃OD on Avance 800 and Avance 500 spectrometers (Bruker Biospin). Assignment of NMR signals is listed in Table S3. Chemical bond connections in the molecule were determined based on DQF-COSY, TOCSY, HSQC, and HMBC correlations. The presence of a lactone ring was confirmed by HMBC from carbonyl to H4. Relative configuration was determined by using NOE information, and the absolute configuration in Fig. 1 and Fig. S4 was drawn according to that of ophiobolin D (27).

Determining the Mode of Brachialactone Inhibitory Action on Nitrosomonas.

Mode of inhibitory action on Nitrosomonas was determined by incubating the pure cultures of luminescent N. europaea with brachialactone in the presence of hydroxylamine (i.e., inhibition of the HAO enzymatic pathway) or in the absence of hydroxylamine (i.e., inhibition of the AMO enzymatic pathway) using a previously reported protocol (23, 29); 1 μL of purified brachialactone (dissolved in dimethyl sulfoxide) to give 5 μM in the assay medium was added to 250 μL of bacterial culture and 199 μL of distilled water; the contents were incubated for 10 min before the addition of 200 μL of 1 mM hydroxylamine (to give 307 μM) to give an assay volume of 650 μL. The mean of the five bioluminescence measurements made during the subsequent 10-min incubation at 15 ^οC was taken as the activity level. Every measurement was repeated four times, and they were considered to be replications for the calculation of the SE (Table S2).

Influence of NH₄⁺ on the Release of Brachialactone.

Plants were grown with NH₄⁺ or NO₃⁻ as sole N sources (38). Root exudates were collected using aerated treatment solutions and BNI activity was extracted (23); brachialactone levels were determined by using HPLC.

Split-Root System Studies.

Plants were raised hydroponically with (NH₄)₂SO₄ as the sole N source. After 6 months of growth, the root system of each plant was divided in half and each half was grown in a separate nutrient tank. After 2 weeks of separate growth, one half of the root system was used for collecting root exudates in a 1 mM N solution from NH₄Cl; and the other half was used for collecting root exudates using 1 mM KNO₃; root exudate solutions were evaporated, and BNI activity was determined (23); brachialactone levels were determined by using HPLC.

Field Validation of the Effectiveness of B. humidicola in Suppressing Soil Nitrification and N₂O Emissions.

A field experiment was established on August 30, 2004, and continued until November 2007 at CIAT headquarters (3°30′N, 76°21′W) on a Vertisol (Typic Pellustert), pH 7.4, with an annual mean rainfall of ≈1,000 mm, annual mean temperature of 26 °C, and an elevation of 965 m above sea level. The six treatments were: (i) B. humidicola (CIAT 679, a standard cultivar with a BNI capacity of 17.3 ATU per g of root dry wt per day) (Table S4); (ii) B. humidicola (CIAT 16888, a germ plasm accession with a BNI capacity of 53.8 ATU pr g of root dry wt per day) (Table S4); (iii) Brachiaria hybrid cv. Mulato (an improved pasture with a BNI capacity of 10.2 ATU g of root dry wt per day) (Table S4); (iv) P. maximum cv. Common (with a BNI capacity of 3.3 ATU g of root dry wt per day) (24); (v) soybean cv. ICAP 34 (lacks BNI capacity in roots) (24); and (vi) control (bare soil, no plants). The experimental unit was a 10 × 10 m plot. The treatments were replicated three times in randomized complete blocks. For each soybean crop cycle, fertilizer was applied twice: for the first application (4 weeks after planting), plots were fertilized (kg·ha⁻¹) with: 48 N, 48 K, 16 P, 0.4 Zn, and 0.4 B, and for the second application (8 weeks after planting) only N was applied at 48 kg·ha⁻¹. The pasture grass plots and the bare soil plots also received the same amount of fertilizers at the same time. Pastures were cut twice a year, coinciding with the harvesting of the soybean crop. Soil inorganic N levels were monitored (30 days after ammonium sulfate fertilization) twice a year.

Soil Sampling, Determination of Ammonium Oxidation Rates, and AOB and AOA Gene Quantification.

Two 1-m² quadrants within each experimental unit were marked as permanently treated sampling subplots. N fertilizer was applied as ammonium sulfate solution in water to the sampling subplots. Three years after establishing the pastures (November 2007), soil samples (five samples from the top 10-cm soil layer and pooled for each experimental unit) were collected from subplots 1 day after fertilizer application and ammonium oxidation rates in the soil were determined (49). Quantification of AOB and AOA functional genes was carried out on these soil samples by real-time PCR using the primer combinations amoA-1F/amoA-2R (50), amoA19F (6)/amoA643R (51), BACT1369F/PROK1541R (52), and Arch20F-/-Arch958R (53) for AOB amoA gene, AOA amoA gene, bacterial SSU rRNA gene, and archaeal SSU rRNA gene, respectively (for details, see SI Methods).

N₂O Emission Measurements in the Field.

N₂O emissions from the sampling subplots were measured monthly from September 2004 to November 2007. When fertilizer was applied, gas sampling was done 1 day after application of fertilizer to the subplots by using the static chamber technique and analyzed for N₂O on a gas chromatograph equipped with an electron capture detector (ECD) (54). N₂O flux was determined from concentration plotted against time. Results presented in Fig. 5 are the cumulative amounts of N₂O expressed in mg of N₂O N per m² per year from September 2004 to November, 2007 (for detailed methods, see SI Methods).