Comparing field and microcosm experiments: a case study on methano- and methylo-trophic bacteria in paddy soil

Authors


*Corresponding author. Tel.: +49 4522 763 244; fax: +49 4522 763 310, E-mail address: eller@mpil-ploen.mpg.de

Abstract

Methane-oxidising bacteria (MOB) play an important role in the reduction of methane emissions from rice agriculture. In rice fields, they are subjected to many environmental and field management parameters, which may have a significant impact on their community composition. To study this in greater detail, the community structure of methano- and methylo-trophic bacteria was investigated in a rice field in northern Italy during the summer 1999 and compared to a microcosm study described previously. We used PCR-based denaturing gradient gel electrophoresis applying 16S rDNA (9α and 10γ) and mxaF (methanol-dehydrogenase) primer sets. In parallel, population size and activity of MOB were determined. This study provides the first comprehensive investigation of different compartments (bulk soil, rhizosphere, rhizoplane, and homogenate) throughout an entire rice-growing season in the field. Lower cell numbers of MOB were detected in the field compared to the microcosms, possibly due to lower CH4 concentrations in the soil pore water. In both studies, growth of MOB occurred predominantly at the root surface (rhizoplane) and in the root (homogenate), whereas cell numbers in bulk soil showed only minor changes throughout the season. Molecular analysis detected only few changes in α-proteobacterial methylotrophs during the season, whereas a higher variability was detected in γ-proteobacteria. Nevertheless, the sequences of electrophoretic bands showed that the diversity in the field study and in the microcosms was comparable. Activity patterns of MOB and the population structure of methylotrophic bacteria agreed well between both studies, even though the detected quantities differed. Extrapolations of microcosm data to the field scale are thus possible, but should be used carefully when concerning quantitative changes.

1Introduction

Whole-ecosystem studies suffer from certain problems such as large scale and freely floating environmental factors resulting in low tractability and long duration of experiments. However, whole-ecosystem studies are by definition most realistic and often regarded superior to microcosm experiments. Microcosms, on the other hand, are tractable, can be produced in sufficient numbers, and maintained under defined conditions. However, microcosm approach has been opposed by other researchers[1], and scale-dependent problems have been identified as crucial[2]. Microbial processes occur on the smallest possible scale, and much work has been done using microcosms. Consequently, a direct comparison between microcosms and the field is essential and should be done at least for some of the processes studied. Here we focus on methane turnover in rice fields, and on how activity and population dynamics of methanotrophic bacteria may control methane emission.

Flooded rice fields are environments with strong variations of both environmental and management parameters thus influencing plant growth and pore water chemistry. Rice fields in general contribute substantially (50–80 Tg CH4a−1) to the global CH4 emission [3,4] and so are of great concern. Methane-oxidising bacteria (MOB) have the potential to substantially reduce CH4 emissions from rice agriculture. Earlier estimates were close to a reduction by 90%[5], but had to be downscaled to about 30% for an entire rice-growing season[6]. Recent field studies from northern Italy verified this and suggested that nitrogen limitation of MOB might be the reason for the seasonality, resulting in high oxidation rates early in the season and low rates from late tillering onwards [6,7]. Additionally, it was found that rice cultivars may affect the rate of CH4 oxidation [8,9].

There is an ongoing discussion about the microsites where CH4 oxidation takes place. While in one of the earlier articles, the CH4 oxidation in the rhizosphere was assumed to be of minor importance[10], a series of papers dealing with model systems based on rice field soil from Vercelli (northern Italy) suggested a close association of MOB and the rice roots [11–14]. Furthermore, MOB also grow at the oxic surface of the soil where about 80% of the potentially emitted CH4 may become oxidised[15].

In this study, we investigate the possibility of extrapolating the data from microcosm experiments to the field scale. We present molecular and cultivation-based analyses of the size, diversity and dynamics of the methano- and methylo-trophic communities in a rice field throughout an entire rice-growing season. All compartments, including different soil layers, root rhizoplane and homogenate, were sampled in season 1999 in an experimental field in Vercelli, Italy. In parallel, CH4 fluxes and CH4 oxidation rates were measured in the field. These were published earlier[6], but are used here as background data for explanation of detected patterns. The results are discussed in context with important biogeochemical data from this field site [6,16] and compared to a previous microcosm study[12] on MOB.

2Materials and methods

2.1Field site

The experimental field was located at the Italian Rice Research Institute (Istituto sperimentale per la cerealicoltura) in Vercelli, northern Italy. The main characteristics of this site and the field management during the sampling in 1999 have been described in detail earlier [16–18]. Prior to flooding, the field was fertilized with 100 kg N ha−1 in the form of urea. On May 5, the field was flooded and a few days later rice (Oryza sativa, type Japonica, variety KORAL) was sowed onto the flooded field. The age of the rice plants was used as characterization of the sampling time points: 7 days after sowing was set as 0 dap (days after planting) to allow a comparison with the greenhouse study by Eller and Frenzel[12]. At 60 dap the flooded field was fertilized again with 40 kg N ha−1 as urea.

The biomass of rice plants was determined each third week using five plants. Roots were washed carefully with tap water to remove adhering soil particles. The number of leaves and the height of the plants above soil surface were determined before the plants were dried to constant weight at 65 °C.

2.2Field sampling and separation of compartments

Field samples were divided into four compartments: bulk soil (non-rooted soil), rhizosphere (rooted soil), rhizoplane (surface of the roots) and homogenate (root without rhizoplane). In Fig. 1 these compartments are illustrated and compared to the microcosms of the greenhouse study.

Figure 1.

Illustration of the different compartments examined in the field and microcosm study.

Three parallel soil cores were taken using a stainless steel cylinder (inner diameter 7 cm, length 26 cm). As the rice roots grew mainly in the upper few centimeters of the soil, the soil cores were divided into two layers: 0–3 and 3–10 cm, which were referred to as rhizosphere and bulk soil, respectively. Soil slurries were prepared from two parts of wet soil (by weight) mixed with one part of autoclaved tap water. Remaining root and straw particles were removed from the slurries before use in the experiments.

For the preparation of the roots, whole plants were dug out by hand and roots were washed carefully to remove adhering soil particles. At 20 dap, 21 plants were pooled into one sample without replicate. At all other dates 6–10 plants were pooled for each parallel in three replicates and analysed. The rhizoplane was detached from the roots by adding sterile glass beads and tap water to the roots in a proportion of 1:10:10 (roots:glass beads:water (by weight)), and shaking at 180 rpm for 30 min at room temperature [11,19]. The resulting suspension was decanted and the supernatant is referred to as rhizoplane. The remaining roots were thoroughly washed with sterile tap water and cut into small pieces. Root pieces and sterile tap water (ratio 1:10, w/w) were homogenised for 2 min (Standard Blender 260, Buddeberg, Germany), yielding the homogenate.

2.3Microcosm experiments

The microcosm system used in the comparative study by Eller and Frenzel[12] and referred to in this manuscrript was kindly provided by Paul Bodelier (Netherlands Institute for Ecology, Institute for Limnology) and has been described in detail before[20]. In brief, microcosms were built of stainless steel cylinders (height 12 cm, ∅ 9 cm) with the root compartment separated by a perforated stainless steel cylinder (∅ 4 cm, holes 1 mm), covered with two layers of nylon gauze (mesh size 0.45 μm, Fig. 1). At the indicated plant ages, the soil and root compartments from four parallel microcosms were prepared for CH4 oxidation measurements, DNA extraction and most-probable number (MPN) counts[12] as described here for the field samples. The soil used in the microcosm study was sampled in spring 1998 from the same site at which the field study was carried out.

2.4CH4 oxidation rates

To determine in situ CH4 oxidation rates, a specific inhibitor for CH4 oxidation, difluoromethane (CH2F2,[21]), was used in CH4 flux measurements at a concentration of 1% (v/v) in the headspace of flux chambers in the field[6] and microcosm study[12]. The increase of CH4 concentrations was followed for 2 h in treated and untreated microcosm or field sites, see Eller and Frenzel[12] and Krüger et al.[6] for a detailed description. In situ CH4 oxidation activity was calculated from the difference in CH4 emission with and without inhibitor and is given as % of net emission.

Potential CH4 oxidation rates in soil slurries were determined for four parallel microcosms and three replicate field samples of each plant age. Each rhizosphere and bulk soil sample was incubated in three replicates of 20 ml soil slurries in sterile glass bottles, supplemented with 10,000 ppmv CH4 each. Methane depletion was monitored by sampling the headspace and subsequent GC/FID analysis. From the mean CH4 depletion curves of the three parallel samples, two linear regressions were calculated. The first regression describes the initial CH4 oxidation, i.e., the CH4 depletion at the beginning of the measurements until the onset of fast (induced) CH4 oxidation. The second regression describes the fast (induced) CH4 oxidation period. The initial and induced methane oxidation rates were calculated from the slope of these regression lines. The time lag until the onset of induced CH4 oxidation was determined by the point of intersection of the two regression lines.

2.5Cell counts by most probable number technique

All four compartments (bulk soil, rhizosphere, rhizoplane and homogenate) were used for MPN determinations as described by Eller and Frenzel[12]. The soil slurries were further diluted 1:100 prior to the dilution in the microtiter plates. For the MPN determination, each sample (3 replicates per date, except roots at 20 dap, see above) was diluted in 2-fold steps in 2 × 8 replicates in microtiter plates. Cell numbers and standard deviations for the 8 replicates of one sample were calculated from positive dilutions using the code number system tabulated by Rowe et al.[22]. When discussed in the text, the statistical significance of the differences in MOB cell numbers was confirmed using the student's t-test. To illustrate the importance of each compartment for the whole ecosystem, cell numbers were calculated per m2.

2.6DNA extraction

Soil slurries (2 ml) or 5–10 ml of either rhizoplane or homogenate suspensions were centrifuged (13,000g, 5 min) and the supernatant discarded. All pellets were stored at −20 °C until DNA extraction. The DNA was extracted by cell lysis with 10% SDS in a cell disrupter (Fast Prep FP120, Savant Instruments Inc., USA), followed by purification using NH4 acetate and isopropanol precipitation as described previously [12,23]. For PCR amplification, DNA concentrations were adjusted to 2 ng DNA μl−1.

2.7PCR amplification

The DNA was amplified using 3 primer sets targeting 16S rRNA genes: the universal eubacterial primer set 533F/907R-GC[24], the 10γ primer set 197F/533R-GC targeting methylotrophs using the ribulose monophosphate (RuMP) pathway for carbon assimilation (including Methylococcaceae, type I MOB), and the 9α primer set 142F/533R-GC, targeting methylotrophs using the serine pathway for carbon assimilation (including Methylocystaceae, type II MOB) [23,25]. We used the PCR conditions as described by Henckel et al.[23] and Eller and Frenzel[12]. In addition to the 16S rRNA gene, the α-subunit of the methanol dehydrogenase (MDH) gene mxaF was targeted with the primer set 1003F/1562R-GC [23,26]. This gene is present in all gram-negative (proteobacterial) methylotrophs and a functional marker for the methylotrophic metabolism. All primer pairs had a GC-clamp attached to the 5′-end of one primer for subsequent DGGE analysis [23,27].

2.8Denaturing gradient gel electrophoresis

DGGE analysis was carried out at 60 °C and 150 V for 5 h (Dcode System, BioRad, Germany) as described previously[12]. Denaturing gradients from 40% to 70% were used for 9α amplification products, whereas for both other primer sets gradients from 35% to 70% were used, with 80% corresponding to 6.5% acrylamide, 5.6 M urea and 32% deionised formamide. To assign changes in banding patterns to methano- or methylo-trophic bacteria, as many DGGE bands as possible were excised and sequenced. Especially with the 10γ primer set, non-methylotrophic 16S rRNA genes are amplified in parallel to the methylotrophs and not all species of the different type I genera are covered. This might result in a more complex DGGE banding pattern than expected for a MOB specific PCR product, although some MOB genera might be underestimated. Nevertheless, the same primer sets and PCR conditions as described above have been used for the molecular analysis in the microcosm study, allowing a direct comparison of the targeted bacterial populations between microcosm and field.

2.9Phylogenetic analysis

Sequences were assembled using the program SeqMan II included in the DNAStar software package (SeqMan Vers. 4.05, 1989–2000 DNAStar, Lasergene) and the consensus sequence was used for phylogenetic analysis. The 16S rDNA sequences were aligned and phylogenetically analysed with the ARB software package[28] (database including about 13,000 sequences), using Maximum Parsimony equations and Jukes–Cantor correction[29] to account for possible multiple substitutions. All sequences were additionally compared to those from the recent EMBL Data Library by BLAST search to identify close relatives. The sequences of the close relatives found in EMBL were aligned and added to the ARB database before calculating the phylogenetic trees. The tree for type I MOB was constructed with the neighbour-joining (NJ) tool of the ARB program[30] using only full-length sequences. The partial sequences derived from DGGE bands were added to this tree using the Maximum Parsimony algorithm and only the positions present in the partial sequences, resulting in 258 valid columns. The tree topology was kept constant during addition of partial sequences[31]. The tree for type II MOB was calculated using type II 16S rRNA sequences as recommended by Heyer et al.[32], resulting in 95 type II MOB sequences. The Maximum Likelihood algorithm of the ARB software package (fastDNAml version 1.2 [33,34]) was used for the comparable clustering of Methylocystis and Methylosinus sequences as presented previously[32]. Partial sequences from DGGE bands were added to this tree by keeping the topology constant, using the Maximum Parsimony algorithm and only the positions present in the partial sequences, resulting in 356 valid columns. For a better graphical resolution in Fig. 3, some sequence names were deleted. MxaF sequences were manually aligned with sequences retrieved from the GenBank database based on amino acids. Trees were calculated using NJ and the protein correction algorithm “PAM” (point accepted mutations per 100 residues per 108 modelled evolutionary years[35]) included in the ARB-Software package.

Figure 3.

DGGE banding patterns retrieved after 16S rRNA gene amplification of DNA from soil and root compartments of rice field samples using the 10γ primer set. Field samples are compared to a bulk soil sample from the microcosm experiment, mixed with ampilification product of a Methylococcus capsulatus (Mcc) culture. Excised and sequenced bands are marked and numbered. Dry soil, mixed soil sample before flooding of the field; B, bulk soil; Rh, rhizosphere, RP, Rhizoplane; H, Homogenate; dap, days after planting.

2.10Nucleotide accession numbers

The sequences derived from 9α amplifications have the GenBank accession numbers AJ429165–AJ429174, sequences of 10γ amplifications AJ429175–AJ429186. The sequences for the α-subunit of the methanol dehydrogenase (mxaF) have the accession numbers AJ496525–AJ496533. All accession numbers are given in the respective figures.

3Results and discussion

In a heterogenous environment like rice fields, MOB are subjected to a large number of parameters, which may have a significant impact on their community. The study of microbial physiology and ecology of such a complex system is based on the use of simplified model systems, like compartmented microcosms planted with rice. However, in such greenhouse studies, not all parameters can be simulated thus questioning the extrapolation of the data to the field scale. To evaluate the justification of such an extrapolation for the methano- and methylo-trophic bacteria, we compared in this study a greenhouse[12] and a field experiment using the same methods for community analysis for this group of microorganisms.

3.1Cell numbers and activity of methane oxidising bacteria

To follow the population size of MOB during the rice-growing season, we carried out MPN counts with samples from all different compartments of the field. Cell numbers in the field increased only slightly in the soil compartments from 1 × 106 to 2 × 106 g dw−1 at 20 and 80 days after planting (dap), respectively (Table 1). MOB numbers in bulk soil and rhizosphere were comparable, except in the very beginning and at the end of the season, when they were higher in the rhizosphere than in the bulk soil. The strongest increase in cell numbers was detected in the root homogenate, from 3.26 × 106 to 2.9 × 107 g dw−1 at 20 and 60 dap, respectively. In the rhizoplane the cell numbers slightly increased to 1.3 × 107 g dw−1. To illustrate the importance of the different field compartments for the growth of MOB in the whole system, cell numbers per area of field plot were also calculated (Table 1).

Table 1.  Methanotrophic cell numbers in the different compartments of the rice field during the growing season of the rice plant in 1999
Plant agea (dap)bBulk soilRhizosphereRhizoplaneHomogenate
 ×105 (g dw−1)×109 (m−2)×105 (g dw−1)×109 (m−2)×105 (g dw−1)×109 (m−2)×105 (g dw−1)×109 (m−2)
  1. Cell numbers calculated per gram dry weight of soil or root and per square meter of field plot (this study).

  2. aSamples were taken in three replicates if not indicated differently. The values are the mean of three replicates (with 8 MPN replicates each) ± standard error.

  3. bDays after planting.

  4. cThe sample “dry soil” was a mixed soil sample of the top 10 cm, taken before the field was flooded.

  5. dOnly one sample taken for the indicated time-point. Values are the mean and standard deviation of the eight parallel MPN dilution series.

Dry soil c,d34 ± 10353 ± 109      
0d5.5 ± 2.758.1 ± 28.511 ± 1.449.5 ± 6.08    
2512.6 ± 2.5132 ± 25.912.8 ± 3.657.6 ± 16.373 ± 12d0.008 ± 0.001d32.6 ± 5.9d0.0039 ± 0.0007d
6123.7 ± 5.8249 ± 60.622.4 ± 9.4101 ± 42.3129 ± 660.119 ± 0.061289 ± 1200.266 ± 0.107
8122.7 ± 7.8238 ± 82.421.8 ± 6.098.1 ± 26.895 ± 390.126 ± 0.036105 ± 270.139 ± 0.025
1018.3 ± 0.287.4 ± 2.316.9 ± 4.476.1 ± 20.045.4 ± 2.00.101 ± 0.018118 ± 950.263 ± 0.088

On average, the total root biomass per plant was 0.05, 0.37, 0.53 and 0.89 g dry weight (dw) of roots for the plant ages of 20, 60, 80 and 100 dap, respectively. The average number of rice plants was 25 m−2[36]. This resulted in a total root biomass of 1.25–22.25 g dw m−2 depending on season. The amounts of soil for the bulk soil and rhizosphere compartment were calculated from the height of the respective soil layers (rhizosphere 3 cm, bulk soil 7 cm) and the dry weight of soil per volume (1.5 g cm−3). This resulted in a total mass of 105 and 45 kg dw m−2 for the bulk soil and the rhizosphere compartment, respectively. For the root compartments, 3.9 × 106–2.6 × 108 MOB cells were detected per m2. The bulk soil showed the highest cell numbers with 5.8 × 1010–3.5 × 1011 MOB m−2 compared to 5 × 1010–1 × 1011 MOB m−2 in the rhizosphere. The bulk soil was thus the most important reservoir for MOB during the whole season. After the end of the vegetative plant growth around 40 dap (data not shown), cell numbers of MOB did not increase any further (Table 1).

The seasonal changes in population size of MOB were similar in the field and in the microcosms. Cell numbers in the bulk soil were remarkably constant with time in both studies (Tables 1 and 2,[12]). Nevertheless, MOB cell numbers in the rhizosphere and root compartments were about one order of magnitude higher in the microcosm than in the field study. The higher cell numbers might be due to a stronger influence of the rice roots on the surrounding soil in the microcosms and higher CH4 concentrations in the microcosm pore water (1 mM compared to 0.4 mM in the field[36]). This might be explained by the more efficient separation of rhizosphere and bulk soil in the microcosms, resulting in a higher density of roots in the rhizosphere compartment than that in the field (Fig. 1). Nevertheless, a strong influence of the roots should also be expected in the field, since the Italian rice variety KORAL forms a dense root layer in the top 3 cm of the soil, containing up to 75% of the total root biomass[37].

Table 2.  Methanotrophic cell numbers in the different compartments of rice microcosms as presented in[12] as a figure, calculated per gram dry weight of soil or roota
Plant age (dap)Bulk soil × 105 (g dw−1)Rhizosphere × 105 (g dw−1)Rhizoplane times 105 (g dw−1)Homogenate × 105 (g dw−1)
  1. aThe values are the mean of two to four replicate microcosms per plant age with eight MPN replicates each ± standard error[12].

017.1 ± 12.817.1 ± 12.8n.d.n.d.
206.2 ± 0.83.6 ± 0.4116 ± 143.32 ± 0.3
2814.7 ± 8.963.9 ± 49.8391 ± 18815.3 ± 12.2
5714.8 ± 8.1150 ± 91.2472 ± 39641.2 ± 18.7
7024.4 ± 1.7238 ± 28.1837 ± 150141 ± 11.3
8528.1 ± 2.2262 ± 28.31210 ± 30.4129 ± 3.3
928.5 ± 6.265.9 ± 22.5361 ± 19619.4 ± 7

A tight association of MOB with the root was also found by other researchers [9,38,39], since in an otherwise anoxic soil the rice roots provide microsites where oxygen is available [39–42]. Additionally, plant exsudates provide important substrates for root-associated methanogens [43,44]. However, over a larger area the rice roots made up only a small portion of the studied soil volume. The total population per m2 was larger in the non-rooted and mostly anoxic bulk soil than in the thin oxic rhizosphere in both studies (Tables 1 and 2). The MOB cell numbers at the roots resulted in only 1% of the total population at the end of the rice-growing season (Table 1). Thus, the net growth of the root-associated MOB is small compared to the overall population size and the persistence of a permanent population depends not only on the growth at the oxic/anoxic interfaces but also on the ability to survive adverse conditions. MOB cells in the bulk soil seem to be the reservoir for the overall population, surviving under anoxic conditions possibly by assimilation of other carbon compounds than CH4[45]. They could thus serve as a “seed bank” for the MOB population in the root-influenced compartments during the next season. Plowing in spring would lead to a homogenous distribution of bacteria throughout the soil compartment explaining the uniform community structure in the different compartments that we have investigated. It is important to investigate in future studies how the MOB population changes its size and/or activity during the non-flooded winter season, with only atmospheric CH4 concentrations at the soil surface and no oxygen supply via rice roots to deeper soil layers.

The seasonal changes of the population size of MOB correlated well with the in situ CH4 oxidation measured in the field and in a microcosm study using the specific inhibitor CH2F2 (Tables 3 and 4, [6,12]). In both studies a high percentage of in situ CH4 oxidation was detected only during the first weeks of plant growth, with 40% in the field and 76% in the greenhouse study (Tables 3 and 4 and [6,12] for detailed discussion). In parallel to the in situ activity, the initial rates of the potential CH4 oxidation measured in vitro decreased (Tables 3 and 4). We found that after the first weeks of plant growth in both microcosms and field, CH4 oxidation became nitrogen-limited, despite a weekly fertilisation of the microcosms[12]. High ammonium concentrations in the porewater after fertilisation with urea rapidly decreased below the detection limit because of efficient uptake by the rice plants. An additional fertilisation led to a direct stimulation of methanotrophic activity. The effect of nitrogen for the field site has been discussed in detail earlier[7]. MOB cell numbers and the potential CH4 oxidation rates in the different compartments were comparable between rhizosphere and bulk soil in the field (Table 3,[6]), whereas in the microcosm experiment both the rates and MOB cell numbers were higher in the rhizosphere, as discussed above (Table 4). In both studies potential CH4 oxidation activity showed higher variation over the season in the rhizosphere than in the bulk soil compartment. In summary, the seasonal and compartmental pattern of MOB cell numbers, activity and controlling environmental factors were comparable between microcosm and field experiment, indicating a good comparability of the two systems, even though the MOB cell numbers were higher in the microcosm study.

Table 3.  Methane oxidation rates in situ and in vitro in the field study (all data as presented graphically in[6])
  RhizosphereBulk soilRhizosphereBulk soil
Plant age (dap)aMOb in situ (%)Initial MORc (nmol g dw−1 h−1)Induced MOR (nmol g dw−1 h−1)Initial MOR (nmol g dw−1 h−1)Induced MOR (nmol g dw−1 h−1)Lag-phase (h)Lag-phase (h)
  1. aDays after planting.

  2. bMethane oxidation.

  3. cMethane oxidation rates.

  4. dn.d., not determined.

Dry soiln.d.d95 ± 9316 ± 12132 ± 7263 ± 522.517.5
−7n.d.408 ± 5408 ± 5287 ± 3287 ± 52.56
2541.41 ± 15.96269 ± 18650 ± 5479 ± 10461 ± 4765
4822.60 ± 27.93146 ± 11315 ± 2886 ± 28257 ± 22.752.75
6116.19 ± 17.1926 ± 37324 ± 3073 ± 30266 ± 63.23.2
65n.d.95 ± 21148 ± 425 ± 3145 ± 21.756.25
81−4.14 ± 7.39105 ± 5267 ± 1632 ± 16259 ± 257.65.75
1053.98 ± 16150 ± 99310 ± 1151 ± 11204 ± 855
Table 4.  Methane oxidation rates in situ and in vitro in the microcosm study (all data presented graphically in[12])
  RhizosphereBulk soilRhizosphereBulk soil
Plant age (dap)aMOb in situ (%)Initial MORc (nmol g dw−1 h−1)Induced MOR (nmol g dw−1 h−1)Initial MOR (nmol g dw−1 h−1)Induced MOR (nmol g dw−1 h−1)Lag-phase (h)Lag-phase (h)
  1. aDays after planting.

  2. bMethane oxidation.

  3. cMethane oxidation rates.

  4. dn.d., not determined.

1676.29 ± 6.61n.d.dn.d.n.d.n.d.n.d.n.d.
2861.85 ± 6.9161 ± 70438 ± 1627 ± 45406 ± 587.7 ± 0.821.1 ± 0.4
4212.01 ± 10.9n.d.n.d.n.d.n.d.n.d.n.d.
57−28.43 ± 9.18148 ± 61479 ± 5852 ± 0.8435 ± 677.6 ± 0.826.4 ± 2.3
708.19 ± 9.46n.d.n.d.n.d.n.d.n.d.n.d.
83−10.33 ± 16.79n.d.n.d.n.d.n.d.n.d.n.d.
92n.d.17 ± 27396 ± 9834 ± 34373 ± 377 ± 0.522.9 ± 0.4

3.2Community structure and development

For the analysis of the population structure of methano- and methylo-trophic bacteria in the rice field, DNA extractions and PCR amplifications were carried out in all compartments of the field, i.e., bulk soil, rhizosphere, rhizoplane and homogenate (Fig. 1). With a primer set targeting the 16S rRNA gene of all eubacteria, amplicons of the predicted size were received for all DNA samples (data not shown). The 9α and 10γ primer sets targeting all methylotrophic bacteria were chosen [23,25] and have been previously applied in the microcosm study[12], thus allowing a direct comparison of both systems.

DGGE banding patterns of 9α amplifications were similar for the different compartments at the same time point and also over the entire season in the field (Fig. 2, insert). The phylogenetic analysis of major DGGE bands of the field samples resulted in only four sequences clustering with the known genera of Methylocystaceae, Methylosinus and Methylocystis. Another five bands clustered with sequences derived from different kinds of soil samples, including rice field soil (Fig. 2). Due to the lack of isolates with high sequence similarities to these fragments and since the similarity is only based on 16S rRNA gene, it is not possible to link these sequences to a function like methanotrophy. Both clusters of sequences from the field samples also contained sequences from the comparative microcosm experiment ([12], Fig. 2), indicating a comparable diversity detected with the 9α primer set in both systems.

Figure 2.

Phylogenetic relation of sequences retrieved from PCR-based DGGE of rice field (bold, this study) and microcosm samples (italic, from[12]) with the 9α primer set for 16S rDNA of serin pathway methylotrophs. The insert shows a part of the field DGGE with the excised bands marked and numbered compared to a bulk soil sample from the microcosm experiment, mixed with ampilification product of a Methylosinus sporium (Mss) culture: Dry soil, Mixed soil sample before flooding of the field; B, Bulk soil; Rh, Rhizosphere; RP, Rhizoplane; H, Homogenate; dap, days after planting. The tree was calculated using Maximum Likelihood analysis for full-length sequences. Sequences are marked as follows: The letters F or M refer to field and microcosm samples, respectively, followed by the number (#) of the DGGE band sequenced (inserted picture for field samples and Eller and Frenzel[12] for microcosm samples), followed by the compartment sampled (B, Bulk soil; Rh, Rhizosphere; RP, Rhizoplane; H, Homogenate) and the age of the rice plants (dap, days after planting).

Amplifications with the 10γ primer set and subsequent DGGE revealed differences in band numbers and intensity during the season in the field (Fig. 3). Changes occurred for one compartment between different time points (Fig. 3, bands marked with an arrow) and between different compartments at the same time point (Fig. 3, bands marked with a dot). The same observations were made during the microcosm study. Since the 10γ primer set also amplifies some non-methylotrophic bacteria, as many bands as possible were exised and sequenced. Unfortunately, we could not retrieve sequences of all visible bands. The sequences clustering within the Methylococcaceae were closely related to the genus Methylobacter (Fig. 4). However, the sequences from two bands with the same electrophoretic mobility (#9 and #12 in Fig. 3) as the Methylobacter-related bands clustered with Acinetobacter (Fig. 4). Sequences of fragments with higher mobility (#2, #6, #11 and #13 in Figs. 3 and 4) could not be affiliated to MOB but showed highest similarities to sequences of Achromatium and Azoarcus in a BLAST nucleotide sequence similarity search.

Figure 4.

Phylogenetic relations of sequences retrieved from PCR-based DGGE of rice field (bold, this study) and microcosm samples (italic,[12]) using the 10γ primer set for 16S rRNA gene. The tree was calculated with full-length sequences using Neighbour-Joining and Maximum Parsimony algorithm and Jukes–Cantor correction. Sequences are marked as follows: The letters F or M refer to field and microcosm samples, respectively, followed by the number (#) of the DGGE band sequenced (Figure 4 for field samples and Eller and Frenzel[12] for microcosm samples), followed by the compartment sampled (Dry soil, mixed soil sample before flooding of the field; B, Bulk soil; Rh, Rhizosphere; RP, Rhizoplane; H, Homogenate) and the age of the rice plants (dap, days after planting).

The 9α and 10γ primer sets were originally designed to amplify the 16S rRNA genes of all methylotrophic bacteria, including methanotrophs. A recent probe match test in ARB including the sequences of the new type II isolates described by Heyer et al.[32] showed that type II methanotrophs are quite well covered with the 9α primer, except the acidophilic genera Methylocella and Methylocapsa[46,47], which should not be important in rice fields. For the 10γ primer set the diversity of detected non-methylotrophic organisms is higher than for the 9α primer, and include Pseudomonas, Azoarcus, Acidithiobacillus and other genera. However, most genera of type I MOB (Methylobacter, Methylomicrobium, Methylomonas, Methylocaldum, Methylococcus) are detected as well, even though not for all genera all species are covered. This might lead to an underestimation of the diversity of type I MOB in our study. Nevertheless, the sequences detected in the field were similar to those of the microcosms, covering the amplicon diversity expected with the 10γ primer set. The predominance of Methylobacter related sequences for type I MOB might be due to the incomplete amplification of the type I MOB diversity with the primers used, as discussed above. Nevertheless, Kolb et al.[48] found a predominance of the Methylobacter/Methylosarcina group in a bulk soil sample from the same rice field in Vercelli. These authors used a quantitative PCR assay with a primer set targeting the α-subunit of the particulate methane monooxygenase, thus excluding a potential bias of the primers used in the present study. On the other hand, in a microcosm experiment with the rice variety ROMA, in which cloning and TRFLP-analysis were applied to examine the MOB population on rice roots in greater detail, several other Methylococcaceae were detected[49]. Unfortunately, no investigation of the rice field soil community was carried out in that study, which makes a comparison of the results difficult.

In addition to the 16S rRNA genes, the α subunit of the methanol dehydrogenase gene (mxaF) was amplified and analysed. This enzyme is found in all methylotrophic organisms, including the methanotrophic bacteria. DGGE bands were received from all four compartments in the field study. The sequences of these bands clustered with Methylomicrobium within the Methylococcaceae and with Methylocystis within the Methylocystaceae, as well as with the methylotrophic organisms Hyphomicrobium and Methylosulfonomonas (Fig. 5). However, the detected groups could not be assigned to a particular compartment or a certain period during the season but were ubiquitous throughout season and samples. The predominance of Methylobacter- related sequences detected with the 10γ primer set could not be verified with these data. For the greenhouse experiment, only four bands from the rhizosphere and rhizoplane were successfully sequenced and clustered with Methylomicrobium within the Methylococcaceae (Fig. 5,[12]).

Figure 5.

Unrooted phylogenetic tree for mxa F-sequences from rice field (bold, this study) and microcosm samples (italic,[12]) compared to published sequences. The letters F or M refer to field and microcosm samples, respectively, followed by the number (#) of the DGGE band sequenced (gel not shown for field samples, the reader is referred to Eller and Frenzel[12] for microcosm samples), followed by the compartment sampled (B, Bulk soil; Rh, Rhizosphere; RP, Rhizoplane, H, Homogenate) and the age of the rice plants (dap, days after planting).

The methylo- and methanotrophic communities in both experiments were remarkably stable in composition and size, in spite of the numerous environmental changes that occurred after the initial mixing and homogenisation of soil by plowing.

4Conclusions

Previous field studies in Vercelli focused on the biogeochemistry of CH4 emissions and the activity of the microorganisms involved in the CH4 cycle (e.g., [5,16,17]). The diversity of involved microorganisms was hitherto investigated mainly in greenhouse experiments using different model systems such as dry soil, unplanted flooded soil, rice roots and compartmented microcosms (eg. [9,23,49]). In this comprehensive study we compared a microcosm system and a field site during an entire rice-growing season including activity measurements and microbiological and molecular population analyses in all compartments. The comparison showed that in both systems the main factors controlling the population size and activity of MOB were plant growth and the availability of nitrogen. The persistent MOB population in the bulk soil seemed to serve as a “seed bank” for the growth of MOB during the next rice-growing season. The population structure detected in both studies was also comparable. This confirms that the results obtained with the compartmented microcosms in the greenhouse study can be extrapolated to the field environment, thus providing a suitable model system for future studies on this complex environment. Nevertheless, this might only be true for the field site and ecological conditions studied here. In California rice paddies the preference of MOB for the rice roots so typical of the Italian site was not observed[50]. Instead, in this environment MOB seemed to depend largely on the soil–water interface. Hence, care must be taken when combining greenhouse and field data from different environments and extrapolations should be made only from carefully adapted model systems using, for example, the same soil, fertilisation treatment and plant variety.

Acknowledgements

We thank Salvatore Russo for the possibility to work in the rice fields of the Istituto sperimentale per la cerealicoltura in Vercelli, Italy. This study was supported by a grant of the Deutsche Forschungsgemeinschaft (DFG) within the SFB 395 “Interactions, adaptations, and catalytic capabilities of soil microorganisms”. Further support came from the Max Planck Society, Germany.

Ancillary