Nitrification–denitrification dynamics and community structure of ammonia oxidizing bacteria in a high yield irrigated Philippine rice field

Abstract

Nitrogen is the single most limiting factor for rice production. Detailed knowledge on nitrogen dynamics in rice fields is therefore of major importance for developing sustainable rice production. A combination of state-of-the-art microsensor, stable isotope tracer, and molecular techniques was used to evaluate coupled nitrification–denitrification potentials and community structure of ammonia-oxidizing bacteria in a high yield irrigated rice cropping system in the Philippines, without the use of microcosm incubations. The multiple approaches showed a high degree of concordance among methods and thereby clarified the investigated processes. Numbers and potential activity of ammonia-oxidizing bacteria in the system reflected the availability of substrate in three defined soil factions with a ranking of: surface soil > rhizosphere > bulk soil. No nitrification activity was measured between spit applications of N fertilizer. However, nitrification was induced upon nitrogen amendment in intact soil cores. Despite induction by nitrogen amendment, the loss of nitrogen through coupled nitrification–denitrification was less than 10% of the plant nitrogen uptake. Denaturant gradient gel electrophoresis of amoA fragments revealed no differences in diversity profiles between the soil fractions, and phylogenetic analysis, based on amoA genes retrieved from the rice paddy soil, identified a set of mutually very similar sequences related to Nitrosomonas nitrosa.

Coupled nitrification–denitrification, Rice, amoA, Nitrogen, Microsensors, ¹⁵N-isotope technique

1 Introduction

Irrigated rice fields account for 75% of the world's rice production. In these intensively cultivated rice cropping systems, nitrogen is usually the limiting nutrient [1], and nitrogen amendment by fertilizers is of great importance to the crop yield. The application of inorganic ammonium-based N fertilizers has substantially raised the rice grain yield over the last decades, but at the same time plant assimilation efficiency of the applied nitrogen has shown a decreasing trend [2]. Microbial soil processes, e.g. mineralization, nitrification and denitrification, greatly affect nitrogen dynamics in soil and govern the supply of nitrogen to the plants. An in-depth understanding of these processes is therefore important.

Irrigated rice paddy soils have a specific zonation due to the constant flooding of the fields during rice growth [3]. Oxygen only penetrates a few millimeters into the soil surface, leaving the bulk soil anoxic. In planted soils, oxygen leakage from roots creates an oxic rhizosphere within the anoxic bulk soil [4,5]. Ammonium-based nitrogen fertilizers are usually applied as top dressings, making the surface soil a favorable niche for aerobic nitrification, the sequential oxidation of NH₄⁺ via NO₂⁻ to NO₃⁻ by nitrifiers [6]. In rhizosphere, NH₄⁺ is supplied from mineralization in the adjacent anoxic bulk soil [7,8], and another niche for nitrification is present around oxygen-excreting roots. Nitrate produced in the oxic niches readily diffuses into the surrounding anoxic bulk soil where it is prone to denitrification, leading to a tight coupling of nitrification and denitrification.

Coupled nitrification–denitrification has been suggested as a significant source of N loss in rice paddy soils, and earlier studies on rice fields have reported coupled nitrification–denitrification losses up to >40% of the applied nitrogen [9–12]. In a recent study, only 27–33% of the applied nitrogen was recovered in the plant material after growth [13] and the loss was mainly attributed to coupled nitrification–denitrification and NH₃ volatilization. In contrast, other recent studies have questioned whether coupled nitrification–denitrification is a significant cause of nitrogen loss in rice paddy soils [7,14]. These studies have however, mostly been conducted with planted microcosms under controlled conditions using soils, which have been oxidized and dried prior to experimental setup and re-flooding.

Ammonia oxidation is the rate-limiting step in nitrification, which in turn is the rate-limiting step for coupled nitrification–denitrification in rice paddy soils [15]. Furthermore, specific ammonia-oxidizing bacterial (AOB) populations in rhisozphere might facilitate plant nitrogen uptake in certain rice varieties by co-providing nitrate and ammonium as inorganic nitrogen source for the plant [16]. Detailed knowledge on the qualitative and quantitative importance of ammonia-oxidizing bacteria in rice paddy soils is thus essential for understanding the function of rice-cropping systems and predicting the impact of management strategies. Hence, the aim of the present study was to obtain detailed information on nitrification and denitrification potentials as well as community structure of the ammonia oxidizing bacteria in rice fields without reliance upon laboratory microcosm setups. We chose fields with plants at maximum tillering, preceding routine nitrogen amendment for panicle initiation, and divided the soil into three operationally defined fractions (surface soil, bulk soil, and rhizosphere) in order to evaluate the importance of each fraction. Abundance of ammonia-oxidizing bacteria was assessed by a competitive polymerase chain reaction (cPCR) assay using primers targeting the gene coding the active site polypeptide of ammonia monooxygenase (amoA). Furthermore, denaturing gradient gel electrophoresis (DGGE) and sequencing of partial amoA fragments retrieved from the rice paddies were used to assess the diversity of the ammonia-oxidizing community. Activity was assessed from NO_x⁻ pore water profiles measured with NO_x⁻ biosensors and potential nitrification capacity based on slurry incubations. In addition, stable isotope tracers were applied to determine coupled nitrification–denitrification rates and plant nitrogen uptake.

2 Materials and methods

2.1 Experimental plot

In situ measurements of microsensor profiles, coupled nitrification–denitrification and plant nitrogen uptake as well as all samplings for nucleic acid extraction and potential nitrification measurements were performed at the International Rice Research Institute (IRRI), Los Bãnos, Laguna Province, Philippines, October 1999, in the framework of a long-term continuous cropping experiment established in 1963.

Oriza sativa variety IR72 was grown in three successive annual crops (dry season, early wet season and late wet season). This investigation was carried out in the first half of the late wet season. Crop duration from transplanting to harvest was 105 days, with two weeks for land preparation between crops. The soil remained water-saturated or flooded from transplanting to physiological maturity and was drained about a week before harvest. The clay-rich soil (clay 52.7%; silt 36.7%; sand 10.4%; pH 6.0) has been re-classified as “Aquandic Epiaquoll”[17]. At a C/N ratio of 12.6 (3.54% C; 0.275% N) the bulk of soil N was organic nitrogen. Each crop received P and K fertilizer at constant rates of 26 kg P ha⁻¹ and 50 kg K ha⁻¹, respectively, whereas N addition rates were 135 kg ha⁻¹ during wet season crops. Top dressings of nitrogen fertilizer were added as urea in four split applications (10, 17, 38 and 69 days after transplanting) based on chlorophyll-meter readings of the uppermost leaf. The field received integrated pest management. Due to a fungal blast, the 1999 wet season spraying with diverse fungicides was far above average. Samples were obtained six weeks after transplanting, at maximum tillering of the plant.

2.2 Soil sampling

Three operational soil fractions were defined: surface soil, bulk soil and rhizosphere. Triplicate soil cores for surface and bulk soil samples were collected between the rice plants in Plexiglas tubes having an internal diameter of 5.5 cm. Surface samples were defined as the top 2–3 mm soil and was scraped off with a metal plate. Bulk soil was defined as soil without rice roots, and was sampled by sub-coring at 5–6 cm depth. Large metal cylinders (internal diameter 30 cm) were used to sample intact plants for rhizosphere samples. Rhizosphere samples were collected by extracting small roots using a fine pair of tweezers, and defined as roots with attached soil. All samples for molecular studies were immediately frozen at −80 °C while samples for potential nitrification measurements were processed within 4 h of sampling.

2.3 Nitrification potential

Potential nitrification activity was estimated from NO₃⁻ plus NO₂⁻ (NO_x⁻) production in NH₄⁺-enriched (approximately 500 μM NH₄⁺) slurries (n= 3 for each soil fraction) as described by Henriksen [18]. The samples were agitated on a rotating wheel and three samples for NO_x⁻ analysis were collected at fixed intervals during a 24 h incubation period. The samples were centrifuged and the supernatants were filter sterilized (Cellulose Acetate, 0.20 μm, Sartorius AG, Goettingen, Germany) and transferred to sterile glass vials, for later analysis. Nitrate and NO₂⁻ were determined on a HPLC system (Sycam, Gliching, Germany) equipped with a UV detector (220 nm, model Spectro monitor 3200, Thermoseparation Products, Riviera Beach, Florida) and an anion column (4 by 250 mm Aniontrenn LCA A14, Sycam, Gliching, Germany) held at 60 °C. The eluent was NaCl (40 mM), with a flow rate of 1.5 ml s⁻¹.

2.4 DNA extraction and PCR

Direct DNA extractions (n= 3 for each fraction) were performed using the FastDNA spin kit for soil (Bio 101, Vista, California) according to the manufacturer's protocol. Cell lysis was performed by bead beating (FastPrep DNA extractor, Bio 101, Vista, CA) followed by absorbance of DNA to a silica matrix in the presence of a high concentration of chaotrophic salts. DNA was washed in a salt/ethanol solution prior to elution in a low salt buffer.

Primers targeting amoA (amoA-1F [19] and amoA-2R-TC [20]) of ammonia-oxidizing bacteria from the β subdivision of the group Proteobacteria were used to obtain amplicons of partial amoA sequences. For DGGE, a semi-nested PCR approach was used to add a 33 base GC-clamp [21] to the amplicon, in order to improve melting behavior of the fragments, as described by Nicolaisen and Ramsing [20]. All PCR were run using the setup and cycling conditions described previously [20].

2.5 Competitive PCR

Competitive PCR was performed in a total volume of 15 μl (13 μl reaction mixture (9.1 μl dH₂O, 1.5 μl buffer 10× (100 mM Tris–HCl, pH 8.8, 750 mM KCl, and 15 mM MgCl₂)), 1.5 μl dNTP (1.25 mM each), 0.3 μl of each primer (amoA-1F/amoA-2R-TC (50 pmol/μl)), 0.3 μl competitive template (six concentrations for each sample) and 2 μl template/Taq mixture (12 μl template, 5 U Taq DNA polymerase (5 U/μl, Pharmacia, Uppsala, Sweden) and 2 μl 10× buffer)). The cPCR mixture was prepared in a larger volume (>200 μl) to reduce pipetting error, and subsequently divided into 13 μl aliquots. The template/Taq mixture was made fresh for every cPCR setup. PCR cycling conditions were: initial denaturation at 93 °C for 1 min, 35 cycles consisting of 30 s denaturation at 92 °C, 60 s annealing at 57 °C and 45 s extension at 72 °C (+1 s/cycle), 5 min final extension.

Gels were stained with SYBRGold, 100 ng ml⁻¹ (Molecular Probes, Leiden, Netherlands), visualized by UV-transillumination, and digitalized by the software Quantity One (Bio-Rad, Hercules, CA).

The number of cells in the environmental samples was calculated using a computer macro based on the calculations described by Bjerrum et al. [22]. The macro estimates the concentration of template copies in the PCR and includes correction for tailing of the competitive template band. The estimate was further corrected for dilution of sample, extraction efficiency and amount of soil used for the extraction.

2.6 DGGE

Polyacrylamid gels containing 30–70% denaturant (8%, 0.75 mm thick 16 × 10 cm; 0.5× TAE buffer, 37:1 acrylamid–bisacrylamid, 100% denaturing acrylamid defined as 7 M Urea and 40% formamide) were cast with a gradient maker (Biorad Laboratories, Hercules, CA) and a 3 ml top gel with no denaturant was added on top before polymerization started. Electrophoresis was performed using the D-Gene System (Biorad Laboratories, Hercules, CA), as described by Nicolaisen and Ramsing [20].

2.7 Cloning, sequencing and phylogenetic analysis

Direct cloning of amoA amplicons obtained from one surface sample was performed. Cloning and sequencing of amoA fragments of this site has been described in a recent paper [20]. The relevance with respect to rice growth was, however, not implemented in the previous description, and consequently they are included in this paper. Retrieved sequences were deposited in GenBank under accession numbers: AF489674–AF489680. Partial amoA gene sequences were manually aligned in Seqpup v. 0.6 (http://iubio.bio.indiana.edu/soft/molbio/seqpup/java/seqpup-doc.html) together with pure culture sequences available from the GenBank database.

Phylogenetic analysis was implemented on 453 unambiguously aligned nucleotides by distance matrix and parsimony algorithms in PAUP v. 4.0 (Sinauer Assosiates, Sunderland, Maine), using the default settings. Bootstrap (100 replicates) was performed using the distance matrix settings.

2.8 Microsensor profiles

Profiles of O₂ were measured in situ (irradiance: 1000 μmol phot m⁻² s⁻¹, temperature about 30 °C). In addition, intact soil cores collected in the field (n= 9) were brought to the laboratory, where profiles of O₂ and NO_x⁻ were measured at in situ temperature but at lower irradiance (200 μmol phot m⁻² s⁻¹). A total of 30 profiles were measured in the lab under illumination and at darkness. Subsequently, NH₄⁺ were added to the water column in an amount comparable to 1 kg N ha⁻¹, and NO₃⁻ in an amount comparable to 200 mg N h⁻¹, and the cores were incubated for 18 h before additional O₂ and NO_x⁻ profiles (n= 3) were measured under illumination and in darkness, respectively.

Oxygen profiles were measured by Clark-type oxygen microsensors equipped with a guard cathode [23]. The sensors had tip diameters of about 10 μm, 90% response times of < 1 s, and less than 2% signal difference between stagnant and vigorously stirred water. The oxygen sensors were 2-point calibrated in aerated water and anoxic sediment. Profiles of NO_x⁻ were measured with NO_x⁻ biosensors based on immobilized, N₂O-producing bacteria [24] having tip diameters of 30–70 μm and 90% response times of 30–60 s. The sensors were operated with a positively charged tip of +0.5 V versus a standard Calomel electrode [25] to increase the sensitivity to NO_x⁻. There was < 2% change in the response to NO_x⁻ when measuring in stagnant compared to vigorously stirred water. The NO_x⁻ sensors were tested for linearity before use in the aquarium with soil and plants and subsequently 2-point calibrated in nitrate-free deep soil layers and in the NO_x⁻-containing overlying water. The NO_x⁻ concentration in a sample from the overlying water was subsequently determined in a separate biosensor setup, as it was not possible to transfer the microscale biosensor being used to a separate setup for calibration. The signals from both oxygen and NO_x⁻ sensors were measured with a custom-built picoamperemeter and the currents were recorded on a strip chart recorder.

Depth profiles of O₂ and NO_x⁻ production were obtained using a computer-implemented diffusion-reaction model [26]. The efficient diffusivity (D_e=φD_s) in the soil was measured with a N₂O-based diffusivity sensor [27] in a soil-agar sandwich [28]. The diffusivities of O₂ and NO_x⁻ were then estimated assuming equal ratios between diffusivities for O₂, NO_x⁻ and N₂O in water and soil. Free solute diffusion coefficients for O₂ and NO_x⁻ were estimated using the equations given by Boudreau [29].

For the calculations of consumption rates we assumed molecular diffusion to be the only transport process, and we thus neglected the possible occurrence of water flow through the soil. Arth and Frenzel [7] observed percolation rates of 0.55–0.9 mm h⁻¹ in experimental soil microcosms with rice plants; however, in the present study, bromide-tracer experiments performed in the field suggested that convective water flow was not important as compared to diffusion (data not shown).

2.9 Coupled nitrification–denitrification and plant nitrogen uptake

Intact cores (n= 3) with intact four weeks old rice transplants were sampled with a stainless steel core sampler, and placed in a 12 cm high, 15 cm diameter PVC beaker. The beakers with the transplants were amended with NH₄⁺ (approx. 5 mmol) and left submerged in between rice plants for two weeks, to allow recovery of roots possibly damaged during sampling. After the pre-incubation period the beakers were placed within 14 cm high stainless steel beakers to which glass bell jars could be mounted, and 15 mmol ¹⁵NH₄⁺ was added to the rhizosphere by multiple vertical injections. Hereafter glass bell jars (height 30 cm) were sealed to the beakers. The bell jars were flushed with an 80:20% mixture of He and O₂ to reduce the concentration of N₂ and thereby increase the sensitivity of the ¹⁵N gas analysis.

During a 48 h incubation period in the field, five samples of 2 ml were collected periodically from the gas phase of each of the bell jars using a glass syringe inserted through a sampling port. The samples were transferred for storage to 5 ml glass vials (Exetainer, Labco High Wycombe, UK) prefilled with He purged demineralized water, that was replaced with the gas sample [30]. After the incubation, soil samples (5 ml) for analysis of ¹⁵N₂ accumulating in the soil pore water were collected with a syringe where the end was cut off. The samples were transferred to 50 ml Falcon tubes containing 10 ml demineralized water and 0.5 ml of 7 M ZnCl₂ to inhibit microbial activity. The soil and water were mixed and the resulting slurry was transferred to 5 ml glass vials (Exetainer, Labco, UK), to which an additional 250 μl of 7 M ZnCl₂ was added. Subsequently, the plants were removed from the beakers, rinsed with demineralized water, separated into leaves and roots, and freeze-dried to constant weight. The dried plant material was ground and stored in plastic cups in dry environment within an exicator for later analysis of ¹⁵N.

Concentrations of the isotopic ¹⁵N-containing gas species, ²⁹N₂ and ³⁰N₂, in the gas and slurry samples were determined on a mass spectrometer (Sira Series II, VG Isotech, Middlewich, UK) as described by Nielsen [31]. Production rates of ¹⁵N₂ were calculated from the slope of the regression line obtained from concentration vs. time plots plus the rate calculated from the amount of ¹⁵N₂ that accumulated in the soil.

The dried plant material was analyzed for both ¹⁵NO₃⁻ and total ¹⁵N. ¹⁵NO₃⁻ was extracted from the plant material by suspending a sub-sample of freeze-dried plant material from each of the plants in hot deionized water (80 °C) for 20 min. Hereafter the plant extract was filtered through a glass fiber filter (Whatmann GF/F), and then analyzed for ¹⁵NO₃⁻ as described by Risgaard-Petersen et al. [32]. Plant content of total ¹⁵N was determined on sub-samples processed in a CN elemental analyzer (NA1500 Nitrogen Analyser, Carlo Erba Strumentazione, Milano, Italy) coupled to a mass spectrometer system. Total content of ¹⁵N compounds in the plant material was calculated from the plant dry weight and the dry weight specific content of either ¹⁵NO₃⁻ or total ¹⁵N.

3 Results

3.1 Abundance and potential activity of ammonia-oxidizing bacteria

Potential nitrification values from the three operationally defined fractions of soil are shown in Fig. 1(a). The nitrification potentials in the different soil fractions (surface, rhizosphere and bulk) were significantly different (ANOVA p⩽0.0001, Scheffe' post hoc comparisons α= 0.5). Highest activity was observed in the surface soil and lowest in the bulk soil. Abundance of ammonia-oxidizing bacteria evaluated by cPCR was also different among the different soil fractions (Fig. 1(b), ANOVA p= 0.0024, when performed on log-transformed data) and followed the same order as observed in the nitrification potential. However, according to Scheffe' post hoc comparison procedure only the abundance of ammonia-oxidizing bacteria in the bulk soil was significantly lower than ammonia-oxidizing bacterial abundance in the surface and rhizosphere soil (α= 0.05).

1

(a) Potential nitrification rates in the three fractions of soil. SD based on three replicates for each fraction is shown. (b) Population sizes of ammonia-oxidizing bacteria based on competitive PCR for the three fractions of soil. Bars show 95% confidence intervals.

3.2 Diversity and phylogeny of ammonia-oxidizing bacteria

The denaturing gel presented in Fig. 2 of triplicate partial amoA amplicons from the various soil fractions shows a high reproducibility of the approach. A low diversity of ammonia-oxidizing bacteria was observed, with only three discernible bands detected. Two were found in all samples except one replicate, whereas one band was found in a single replicate and thus possibly represent a minor population with a heterogeneous distribution. The three soil fractions did not show significant differences in presence of particular bands, and hence no effect of fraction on the diversity of ammonia-oxidizing bacteria in soil was recognized. As no significant differences in population structure between the fractions was observed, only the surface sample corresponding to lane 2 in Fig. 2 was cloned and analyzed for phylogenetic affiliation of retrieved sequences.

2

DGGE profile showing partial amoA amplicons from the three fractions of soil. The band pattern reveals that only sequences affiliated with the Nitrosomonas genus are present in this paddy soil. Lanes 1–3: triplicate samples of surface soil; lanes 4–6: triplicate samples of bulk soil; lanes 7–9: triplicate samples of rhizosphere soil; M: marker based on amoA fragments from Nitrosomonas europaea and Nitrosospira multiformis as well as cloned and amplified amoA and 16S rDNA fragments from environmental clones.

All retrieved sequences were related to previously published amoA sequences from the β subdivision of Proteobacteria. Fig. 3 shows the phylogenetic affiliation of the rice field recovered amoA sequences as compared to a selection of reference sequences. The rice field sequences were mutually very similar and may originate from different gene copies in a set of highly related organisms. They grouped within the Nitrosomonas communis cluster (as defined by Purkhold et al. [33]) with the closest pure culture representative being Nitrosomonas nitrosa. The clustering of retrieved sequences together with N. nitrosa was supporter by a bootstrap value of 51.

3

Phylogenetic tree reflecting the relationships of partial amoA sequences recovered from the surface soil of the irrigated rice system at IRRI (Clone R_1–7) and partial amoA sequences from reference ammonia-oxidizers obtained from the GenBank database. The tree is a consensus distance matrix derived tree, rooted by inclusion of a sequence of the ammonium monooxygenase gene from the ammonia-oxidizing bacterium of the γ subdivision of Proteobacteria, Nitrosococcus oceani. Distance matrix bootstrap values (100 replicates) for branches are reported. Bootstrap values lower than 50 are not reported, as the branches in question were not supported in the majority of bootstrap replicates by the distance matrix method.

Though branch lengths varied, the same branching pattern was observed using both distance matrix and maximum parsimony algorithms to infer phylogeny. Furthermore, the branching pattern was congruent with previously published trees based on amoA fragments [33]. This observation gives us fidelity in the phylogenetic affiliation of the retrieved sequences.

Specific sequences cannot be directly ascribed to specific DGGE bands, as sequencing was done on directly cloned amplicons, and not on bands cut out from the DGGE gel. However, it has been shown previously [20] that the agreement between DGGE profiles of environmental samples and DGGE profiles of sequences obtained from direct cloning of the same PCR product is high. Cloning does, however, often lead to the detection of sequences not recognized by DGGE, due to higher detection limit for the DGGE approach. This is reflected in the present study, as more sequences are presented in Fig. 3 than bands detected by DGGE as presented in Fig. 2.

3.3 Activity and microdistribution of ammonia-oxidizing bacteria in the surface soil

Concentrations of NO_x⁻ in the floodwater and surface soil were found to be < 1 μM (data not shown). The NO_x⁻ profiles shown in Figs. 4(a) and (b) were thus obtained after amendment with NH₄⁺ and NO₃⁻, which did not change the oxygen profiles. Nitrate only penetrated about 1.4 mm when incubated in the dark as opposed to 2.5 mm in the light. When illuminated, a nitrate peak of 6–7 μM caused by nitrification was present at 1.0 mm depth.

4

Concentration profiles measured using microsensors. (a) Profiles obtained during illumination, (b) profiles obtained in the dark. ^: oxygen profiles; ●: NO_x⁻ profiles. The uppermost X-axis gives the oxygen concentration, whereas the lower X-axis gives the NO_x⁻ concentration. Calculated activity rates of oxygen production (photosynthesis) and consumption (respiration) as well as NO_x⁻ production (nitrification) and consumption (denitrification or assimilation) are shown for (c) illuminated and (d) dark conditions. Production is presented as positive values, whereas consumption is presented as negative values. NO_x⁻ production and consumption is presented as gray boxes, whereas oxygen production and consumption rates are represented by lines (white boxes). The uppermost X-axis gives the oxygen production, and the lower X-axis gives the NO_x⁻ production.

The oxygen penetration changed significantly from 1 mm in the dark to 1.7 mm when illuminated. The pronounced oxygen peak near the soil surface after illumination was due to net photosynthesis by benthic microphytes on the soil surface. Oxygen profiles measured in the field at light intensities of about 1000 μ moles photons m⁻² s⁻¹ at the soil surface often showed considerably larger oxygen peaks and oxygen penetrations of up to 3 mm (data not shown). The cover with benthic microphytes in the field was, however, extremely heterogeneous. One patch could have a pronounced oxygen peak in the light whereas the neighboring area exhibited only little effect of illumination.

In Figs. 4(c) and (d), the consumption and production rates estimated from the average oxygen and nitrate profiles measured after nitrogen addition are shown. During darkness (Fig. 4(d)), there was a gradually decreasing consumption of oxygen with depth down to 1 mm where oxygen was depleted. Oxygen production in the light (Fig. 4(c)) was associated with the upper 0.6 mm layer. Below this depth there was a net consumption until all the oxygen was consumed around 1.7 mm below the surface. Net nitrate consumption was observed in the dark incubated cores at all depths (Fig. 4(d)), whereas illumination induced nitrifying activity in the deeper parts of the oxic layer, 0.8–1.7 mm below the surface (Fig. 4(c)). The zone of net nitrate production by nitrification matches the zone of net oxygen consumption during illumination, whereas the nitrate consumption occurred below 1.7 mm where oxygen was depleted.

3.4 Coupled nitrification–denitrification and plant uptake

Rates of coupled nitrification–denitrification of ¹⁵NH₄⁺ and rates of plant ¹⁵N uptake were both highly variable (Fig. 5), reflecting an extremely heterogeneous environment. In general, however, plant uptake rate vastly exceeded the denitrification rate. The measured denitrification rates were only 1–9% of the measured uptake of ¹⁵N in the plants. No ¹⁵N-labeled NO₃⁻ was detected in the plant material.

5

Coupled nitrification–denitrification rates and plant uptake of ¹⁵N–NH₄⁺. The rates varied extensively between experiments (note the logarithmic Y-axis) and hence, the measured rates in each individual experiment are presented.

4 Discussion

Detailed knowledge on the influence of the soil microbial community on rice crop performance is crucial in order to improve field management to obtain higher yields and sustainable rice production. The present study contributes to this knowledge by presenting data obtained by direct measurements in the rice field without the use of microcosm setups.

Ammonia-oxidizing bacteria were enumerated using the molecular-based competitive PCR technique. This culture-independent technique has shown to be superior to the traditional most probable number (MPN) technique for enumerating ammonia-oxidizing bacteria in different soil systems [34], due to low cultivability of these bacteria. To evaluate the cPCR technique, we calculated the cell-specific activity of ammonia-oxidizing bacteria based on cPCR and potential nitrification of slurry incubations. For the surface sample, this activity was within the range of literature values of ammonia oxidation rates reported for Nitrosomonas species in pure culture (0,021 pmol cell⁻¹ h⁻¹, present study; 0.011–0.023 pmol cell⁻¹ h⁻¹[35]). Hence, the cPCR approach appeared to be highly effective in enumerating the ammonia-oxidizing population from the surface soil fraction. The agreement was not as pronounced for the bulk soil and rhizosphere samples (0.38 and 0.072 pmol cell⁻¹ h⁻¹, respectively), where the cell-specific activity was overestimated compared to literature values. This overestimation was probably due to a pronounced underestimation of the cell numbers by cPCR in the sediment matrix. As PCR is inhibited by even trace amounts of humic substances in the DNA extract, inhibition is often more pronounced in soil and sediment samples as compared to water and sludge samples. This in fact could explain the differences seen between the more biofilm-like surface soil and the bulk and rhizosphere soil. However, the overall distribution pattern of ammonia-oxidizing bacteria abundance and activity, measured by potential nitrification of slurry incubations and cPCR respectively, was similar, ranking surface soil > rhizosphere > bulk soil. This distribution pattern, with higher abundance of ammonia-oxidizing bacteria at the oxic niches than in the anoxic bulk soil, has previously been found for irrigated rice fields [36], and is in accordance with the presumed nutrient (oxygen and ammonia) availability in the fractions.

Most previous enumerations of ammonia-oxidizing bacteria in rice soil systems are based on MPN counts [37–39,36], and are 1–2 orders of magnitude lower than the ammonia-oxidizing bacteria abundance estimated based on the cPCR approach presented herein. This difference could be due to the lower cultivability of ammonia-oxidizing bacteria, and/or the sampling practice. Samples have often been taken as a 5–10 cm profile, ignoring the fractionation between oxidized surface soil and anoxic bulk soil, and thereby underestimating the ammonia-oxidizing population size of the surface soil, which might represent an important niche for nitrification. Nitrite oxidizing bacteria were not enumerated in the present study. However, abundance of nitrite oxidizing bacteria has previously been shown to correlate with ammonia-oxidizing population sizes in rice paddy soils [39]. Potential nitrification activities measured by microsensors and slurry incubations of the surface soil was in the same range as found for surface sediments of shallow waters [40], but considerably lower than rates found for other non-flooded agricultural soils [41], consistent with the lower oxygen penetration in the waterlogged paddy soil. The pronounced population size of ammonia-oxidizing bacteria found in the bulk soil fraction could be explained by the intensive cultivation with extensive soil preparation, which results in frequent mixing of the soil fractions. Ammonia-oxidizing bacteria can survive for prolonged periods in anoxic sediments without loosing the ability to respond when substrates are available [42–44]. Therefore, they are readily detectable by both potential nitrification and DNA-based methods despite presumed in situ dormancy.

Data were collected right before the third split application of urea to the field, and no net in situ nitrification activity was found at this time point based on microsensor profiles. After nitrogen amendment, potential nitrification activity in the soil was found to correspond to measurements obtained by the slurry incubations (48 vs. 54 nmol N g⁻¹ h⁻¹). This, together with the findings of a cell-specific activity close to cell-specific activities found for pure culture experiments, showed that the ammonia-oxidizing community was able to oxidize ammonia at a maximum rate in the surface soil upon amendment with nitrogen in the absence of plants, indicating nitrogen limitation of the system.

Induction of nitrification upon nitrogen amendment has previously been reported for microcosms systems [7], and our results confirm these findings in intact field cores. The lack of nitrification induction during darkness at the oxic/anoxic interphase is explained by oxygen limitation, as oxygen only penetrated to 1 mm in the dark, and thus barely reached the nitrification zone identified during illumination (Fig. 4). The position of nitrification and denitrification zones in close proximity around the oxic/anoxic interphase below the photosynthetic zone has previously been reported from waterlogged soils and sediments [7,45], and was mainly attributed to microphytobenthic induced nitrogen limitation [46]; in addition, high pH caused by photosynthetic activity [47] and photoinactivation of the amoA gene [48] also might force the nitrifying bacteria deeper in the sediment.

Despite induction of nitrification in plant-free setups, the ¹⁵N-isotope technique revealed that nitrogen loss through coupled nitrification–denitrification was only 1–9% of the total uptake of ¹⁵N by the plant. Urea is the nitrogen source normally used for fertilization in these plots, and the use of NH₄⁺ in the nitrification–denitrification studies might underestimate the plant nitrogen uptake, as NH₄⁺ is expected to be more easily used by nitrifying bacteria. These results show that even though a pronounced nitrification peak is induced upon nitrogen amendment in cores without plants, the rice plants are good competitors for nitrogen, as has also been found previously for other plant species [49,50]. Loss due to coupled nitrification–denitrification might thus be most pronounced after basal fertilizer amendments around transplanting of the rice, where the plant root web is not fully developed. Nitrate as well as NH₄⁺[51,52] is taken up by the plant, which means that even though some ammonium might be oxidized in the rhizosphere, it is still possible for the plant to assimilate this nitrogen source. No NO₃⁻ was measured in the plant material in the present study, indicating an efficient plant uptake of the available NH₄⁺ or a possible fast reduction of NO₃⁻ after uptake.

Recently, Briones et al. [16] suggested that the community structure of ammonia-oxidizing bacteria in the root zone might also have a positive influence on the plant nitrogen uptake by co-providing nitrate at a certain rate. The importance of the diversity of the nitrifying community on plant nitrogen uptake is not fully investigated, and the attempt of the present study was therefore to divide the soil into fractions reflecting different physicochemical niches, and look for differences in community structure of the ammonia-oxidizing populations and possibly the presence of specific rhizosphere associated ammonia-oxidizing bacteria. Despite the major differences with respect to physicochemical parameters as well as biological parameters between the surface soil, rhizosphere and bulk soil, no detectable difference in the diversity of retrieved amoA sequences was observed among the fractions (Fig. 2). Furthermore, the overall diversity of ammonia-oxidizing bacteria was low compared to other soil and sediment systems [20]. A negative effect on microorganism diversity due to heavy amendment with chemical fertilizers and pesticides could be responsible for the observed low diversity of ammonia-oxidizing bacteria, as a negative effect of chemical fertilizers on total microbial biomass has previously been observed [53]. Furthermore, a comparison of signature lipid biomarker profiles (PLFA) (Reichardt, unpublished) and DGGE profiles (Nicolaisen, unpublished) in soil from the long-term field experiment investigated here, with soil from a traditional rice cropping system of the Ifugao rice terraces, without application of chemical fertilizers and pesticides, supported this hypothesis. However, more data are needed to confirm a hypothesis of depression of microbial diversity due to chemical fertilizers and pesticides.

Nitrosomonas species belonging to the Nitrosomonas europaea and Nitrosomonas communis clusters have often been found in environments with high substrate concentrations [54]. These organisms often have higher K_s (lower substrate affinity) values and higher growth rates compared to organisms belonging to the Nitrosospira genus [55,56]. The high load of surface amended nitrogen fertilizer provides the system with pulses of nitrogen. However, nitrogen might only be available to the ammonia-oxidizing community for a limited time as plants are better competitors for NH₄⁺ than ammonia-oxidizing bacteria [49,50, this study]. An r-selected growth strategy with high V_max might thus be preferential in this kind of system in order to exploit the energy resource, when available, with maximum efficiency.

The data obtained here are in accordance with a more elaborate study of ammonia-oxidizing bacterial diversity in rice fields, where sequences affiliating with N. nitrosa are often recovered from irrigated fields with high nitrogen applications (Nicolaisen, unpublished data), and the findings that Nitrosomonas species are often affiliated with roots of modern high yield rice varieties [16]. N. nitrosa possess urease activity [55], whereas N. communis does not. Despite the low bootstrap value obtained for the clustering of the retrieved species with N. nitrosa, it could be hypothesized that the rice field ammonia-oxidizing community possessed urease activity, which would be a clear advantage as nitrogen fertilizer is added as urea in this system.

In conclusion, the distribution of cells between fractions followed the expected distribution based on substrate availability, with higher abundance and potential activity in the oxic niches. No in situ nitrification was observed, but was induced in non-planted soils after nitrogen amendment. However, the data showed coupled nitrification–denitrification to be less than 10% of plant nitrogen uptake, and a possible significant loss due to nitrification–denitrification would thus only be expected at times when the plant root web is not developed. In order to evaluate the interaction between the plant and ammonia-oxidizing bacteria under different management practices, a comparison of traditional cropping systems, with the intensively cultivated system described here, would be valuable.

Acknowledgements

This work was supported by RUF, Danish International Development Agency and Danish Natural Science Research Council. We thank Dorte Ganzhorn and Pernille Vester Thykier for excellent technical assistance. Furthermore, we express our gratitude to everybody at the Soil and Water Division, IRRI, Los Baños, Philippines for their hospitality during field research. In particular, we thank Nilo Driz for his indispensable help in the fields. We also thank Mette Mathiasen for measuring profiles for diffusivity calculations.

Author notes

1

Department of Marine Ecology, National Environmental Research Institute, DK-8600 Silkeborg, Denmark.

2

Marine Science Institute, University of the Philippines at Diliman, Diliman, 1101 Quezon City, Philippines.

3

Unisense A/S, Gustav Wieds Vej 10, DK-8000 Aarhus C, Denmark.