Partitioning of CH4 and CO2 Production Originating from Rice Straw, Soil and Root Organic Carbon in Rice Microcosms

Flooded rice fields are an important source of the greenhouse gas CH4. Possible carbon sources for CH4 and CO2 production in rice fields are soil organic matter (SOM), root organic carbon (ROC) and rice straw (RS), but partitioning of the flux between the different carbon sources is difficult. We conducted greenhouse experiments using soil microcosms planted with rice. The soil was amended with and without 13C-labeled RS, using two 13C-labeled RS treatments with equal RS (5 g kg−1 soil) but different δ13C of RS. This procedure allowed to determine the carbon flux from each of the three sources (SOM, ROC, RS) by determining the δ13C of CH4 and CO2 in the different incubations and from the δ13C of RS. Partitioning of carbon flux indicated that the contribution of ROC to CH4 production was 41% at tillering stage, increased with rice growth and was about 60% from the booting stage onwards. The contribution of ROC to CO2 was 43% at tillering stage, increased to around 70% at booting stage and stayed relatively constant afterwards. The contribution of RS was determined to be in a range of 12–24% for CH4 production and 11–31% for CO2 production; while the contribution of SOM was calculated to be 23–35% for CH4 production and 13–26% for CO2 production. The results indicate that ROC was the major source of CH4 though RS application greatly enhanced production and emission of CH4 in rice field soil. Our results also suggest that data of CH4 dissolved in rice field could be used as a proxy for the produced CH4 after tillering stage.


Introduction
Flooded rice fields are an important source of the greenhouse gas CH 4 [1,2]. Methane and CO 2 are end products of anoxic degradation of organic matter in rice field soil [3]. The organic matter is mainly derived from three sources [4]: (1) soil organic matter (SOM), (2) root organic carbon (ROC) including root exudates and sloughed-off dead root, and (3) dead plant organic matter, such as rice straw (RS), which is often applied in large amounts (up to 12 t ha 21 annually) to maintain soil fertility [5][6][7]. Methane production is partitioned mainly between these three types of organic matter. Knowledge of partitioning is important for improving process-based modeling of CH 4 emission from rice fields [8,9], which is the basis for predicting methane flux and assessing the impact of agricultural management and global change.
Quantification of carbon partitioning can in principle be achieved by pulse-labeling of rice plant with 13 CO 2 or 14 CO 4 [10][11][12]. Recently, free-air CO 2 enrichment (FACE) using 13 Cdepleted CO 2 was used for determining the contribution of ROC to production of CO 2 and CH 4 in rice field soil [13]. However, pulse-labeling only assesses the immediate contribution of root exudates, while the contribution of sloughed-off dead root cells cannot be fully accounted for [13][14][15][16]. Since FACE experiments apply elevated CO 2 concentrations, photoassimilation of CO 2 may be enhanced and thus increase the contribution of plants and soil organic matter to carbon flux [17][18][19]. Furthermore, most studies of carbon flux partitioning in rice fields have been done without application of straw, so that full partitioning of the origin of carbon flux into SOM, ROC and RS was not possible [4]. However, application of RS should be taken into account, since RS may not only be used as substrate for CH 4 production, but might also enhance CH 4 production from other carbon sources [20,21].
The partitioning of the CH 4 production from different sources of organic carbon (SOM, ROC, RS) can be achieved, if these have different isotopic signatures. However, a major difficulty during partitioning the sources of CH 4 is caused by the carbon isotopic fractionation during the conversion of organic matter to CH 4 , which is typically 10-70% [22]. Nevertheless, the relative contribution of acetoclastic versus hydrogenotrophic methanogenesis to CH 4 production has been determined successfully in environments such as rice field soil [23] and lake sediments [24], after the isotopic fractionation factors in both methanogenic pathways were determined. The d 13 C values of CH 4 from the two pathways are substantially different, since the isotopic fractionation factors of the two pathways are largely different [22,24,25]. Analogously, it is possible to partition the sources of CH 4 if the d 13 C of CH 4 derived from each carbon source in the rice field soil is known. Normally, the CH 4 derived from SOM, ROC and RS has similar d 13 C values, since all the organic matter has eventually been derived from rice plant material [23,26]. However, this problem may be solved by cultivation of rice in soil amended with 13 C-labeled RS.
The aim of this study was to determine the partitioning of the carbon flux involved in methanogenic degradation of carbon sources by determining the d 13 C of CH 4 derived from ROC. We therefore prepared rice microcosms with two treatments of 13 Clabeled RS, both having the same amount of RS (5 g kg 21 soil, equals about 5 t ha 21 ) but different content of 13 C. We determined the produced CH 4 and CO 2 by collecting soil cores and incubating samples anoxically [27].

Materials and Methods
Planted and unplanted rice microcosms Soil samples were provided by the Italian Rice Research Institute in Vercelli. Soil was taken from a drained paddy field in spring 2009 and was air dried and stored at room temperature. The soil was sieved (,2 mm) prior to use. The characteristics of the soil have been described previously [28]. Planting pots (upper diameter = 19 cm; lower diameter = 14 cm; height = 16 cm) were filled with 2 kg dry soil and turned into a slurry with demineralized water.
For planted rice microcosms, in total 48 pots were prepared, 16 pots for the unamended control, and 16 pots each for RS treatment I and RS treatment II. Fertilizer solution (50 ml of a solution containing per liter: 10 g urea, 7.6 g KH 2 PO 4 ) was added to each pot as basal fertilizer. For both RS treatments, 10 g powder of RS was added to each pot and mixed thoroughly into the soil slurry. The d 13 C values of RS added in treatment I and II were 213.0% and 474.7%, respectively. These d 13 C values were obtained by adding desired amount of 13 C-labeled (d 13 C = 1859.9%) and unlabeled (d 13 C = 227.6%) RS separately into each pot. The 13 C-labeled RS was prepared by growing rice plants in the greenhouse until the late vegetative stage. The plants were covered with a 18-L acrylic chamber, 1% 13 CO 2 (final concentration; 99 atom%, Sigma, Germany) was added to the headspace, incubated for 5 days (12 h light, 25uC), and then harvested. The unlabeled RS was from rice plant grown in the same manner without feeding on 13 CO 2 . These rice plants were dried and ground to powder. After 3 days of incubation in the greenhouse, all the pots were planted with one 12-day old rice seedling (Oryza sativa var. KORAL type japonica), and were flooded with demineralized water to give a water depth of 5 cm above the soil surface. The water depth was maintained throughout the experimental period. The rice microcosms were incubated in the greenhouse with a relative humidity of 70%, a 12h photoperiod and a 28/22uC day/night temperature cycle. The day of transplantation was taken as day zero. On day 21, a second dose of 30 ml fertilizer solution was added to each microcosm. At each sampling time (day 41, 55, 70 and 90), 12 rice microcosms were sacrificed (4 replicates for control and for each treatment). For unplanted microcosms, the preparation was the same as for planted ones, but without rice plant in the pots. In total, 12 pots were prepared with 4 pots each for the unamended control, RS treatment I and RS treatment II.

CH 4 flux, soil pore water and plant parameters
Rates of CH 4 emission was measured on day 41, 55, 70 and 90 of incubation in the greenhouse as described previously [27]. For flux measurements, planted rice microcosms were covered by flux chambers, and gas samples were taken every 30 min for 2 h. CH 4 emission rates were determined from the slope of the linearly increasing CH 4 mixing ratio and expressed in mmol CH 4 m 22 h 21 .
Samples for the determination of the isotopic signature (d 13 CH 4 ) of the emitted CH 4 were taken in glass containers (100 ml). The first sample was taken directly after closure of the chambers, the second sample was taken at the end of the 2-h closure period. The isotopic signature of the emitted CH 4 was calculated according to [27].
Pore water samples were collected into Venoject bloodcollecting tubes (Terumo Europe N.V., Belgium) from the rhizosphere (3 cm depth) and bulk (9 cm depth) soil of rice microcosms using Rhizon pore water samplers (Rhizosphere Research Products, the Netherlands). After heavy shaking by hand, the headspace of the tubes was sampled using a pressure lock syringe and directly analyzed for CH 4 and CO 2 and d 13 C. The CH 4 and CO 2 concentration in the soil pore water was calculated as described previously [27].
Plant height, tiller number and aboveground biomass were determined. For dry weight determination, samples were dried for 48 h at 60uC.

CH 4 and CO 2 production
Production rates of CH 4 and CO 2 and respective d 13 C values were determined by collecting soil core samples in rice microcosms on day 41, 55, 70 and 90 of incubation in the greenhouse [27]. After cutting off the rice plant, the surface water layer was removed. Soil cores were taken in each pot with stainless steel corer (Ø 22 mm, 210 mm in length). Two to three soil cores (about 100 g in total) were collected from each pot and transferred into a 250-ml bottle. The soil samples were turned into slurry using N 2 -gassed deionized sterile water so that the ratio of dry weight of soil to water was 1:1. After flushing the samples with N 2 , the bottles were sealed with butyl rubber stoppers and, after shaking, flushed again with N 2 to remove residual O 2 and CH 4 . Incubation was performed statically at 25uC in the dark for 24 h. Headspace samples were taken every 12 h after shaking the bottles, and analyzed for concentration of CH 4 and CO 2 and their d 13 C. The CH 4 and CO 2 production from planted soil microcosms was due to decomposition of SOM plus ROC (unamneded control) or of SOM, ROC plus RS (RS treatments). CH 4 production rates were calculated by linear regression of the CH 4 increase with incubation time, and expressed in nmol CH 4 g dw 21 h 21 of soil. The CO 2 production rates were determined analogously.
For unplanted soil microcosms, the methods for collection and incubation of soil core samples were similar, but these pots were not sacrificed, but at each sampling day (day 41, 55, 70 and 90), a 60-g soil core was taken from the pot. After removal of the soil core the residual soil in the pot was compacted, and water was added to maintain a water level of 5 cm depth. Using this procedure about 2.1% of the total amount of soil in the pot was collected during each sampling. The CH 4 and CO 2 production from unplanted soil microcosms was only due to decomposition of SOM (unamneded control) or of SOM plus RS (RS treatments).

Analytical techniques
The gas samples were analyzed for CH 4 and CO 2 using a gas chromatograph (GC) equipped with flame ionization detector (FID) [29]. Stable isotopic analysis of gas samples (CH 4 and CO 2 ) from pore water and soil core incubation were performed directly using the GCC-IRMS, samples from flux measurements (low in CH 4 ) were preconcentrated on a Precon (Finnigan, Bremen, Germany). The principal operation of the GCC-IRMS has been previously described [30,31]. The isotope reference gas was CO 2 (99.998% purity; Messer-Griessheim, Düsseldorf, Germany) calibrated with the working standard methyl stearate (Merck). The latter was intercalibrated at the Max-Planck-Institute for Biogeo-chemistry, Jena, Germany (courtesy of Dr. W.A. Brand) against NBS 22 and USGS 24, and reported in the delta notation vs. V-PDB: d 13 C = 10 3 (R sa /R st 21), with R = 13 C/ 12 C of sample (sa) and standard (st), respectively. The precision of repeated analysis was 6 0.2%, when 1.3 nmol CH 4 were injected [23]. The determination of the stable isotopic signatures of dried plant and soil samples was carried out at the Institute for Soil Science and Forest Nutrition (IBW) at the University of Göttingen, Germany.

Calculations
1. Fraction of CH 4 production from ROC (f ROC ). The fraction of CH 4 derived from ROC (f ROC ) can be determined from the following mass balance equation: where d 13 C CH4 = d 13 C of CH 4 produced (or dissolved) in the planted rice microcosms at each sampling time; d 13 C CH4-ROC = d 13 C of CH 4 formed from ROC (determination see below); d 13 C CH4-SOR = d 13 C of CH 4 formed from SOM plus RS, i.e. the CH 4 produced (or dissolved) in the unplanted soil treated with RS. The equation can be transformed into the following two equations for RS-treatment I and II, respectively: Since f ROC and d 13 C CH4-ROC should be the same in treatment I and II, d 13 C CH4-ROC can be calculated by solving equations (2) and (3): Then, f ROC can be calculated from either equation (2) or (3). 2. Fraction of CH 4 production from RS carbon (f RS ). The d 13 C values of the CH 4 produced (or dissolved) in the two RS treatments are given by the following two mass balance equations: with f RS , f SOM and f ROC denote fractions of CH 4 produced from RS, SOM and ROC, respectively; d 13 C RS-I and d 13 C RS-II are d 13 C of the rice straw carbon in treatment I (213.0%) and II (474.7%), respectively; d 13 C SOM and d 13 C ROC are d 13 C of SOM (225.8%) and of the plant biomass ( Fig. 1), respectively; DCH 4 designates the overall isotopic fractionation factors involved in CH 4 production from these organic matters, in case of dissolved CH 4 also those involved in oxidation and transfer of CH 4 from soil to the atmosphere.
Since the terms f SOM d 13 C SOM , f ROC d 13 C ROC and DCH 4 should be the same in treatment I and II, combination of equations (5) and (6) and solving for f RS results in: of which the d 13 C can be determined experimentally. Here, d 13 C CH4-I and d 13 C CH4-II were determined experimentally, and d 13 C RS-I and d 13 C RS-II were mixtures of labeled and unlabelled RS, of which the d 13 C were determined experimentally (see above). Finally, the fraction of CH 4 production from SOM (f SOM ) can be calculated, since Analogous equations are valid for the fractions of CO 2 produced from ROC, SOM and RS in rice field soil.

Statistical analysis
The significance of differences between treatments over time for various variables were determined by one-way analysis of variance (ANOVA) followed by multiple comparisons (Duncan post hoc test) using SPSS 13.0. To test the significance of the differences between contributions to produced and dissolved CH 4 or CO 2 , two-tailed independent t-tests were applied using Microsoft Excel 2007.

Stable carbon signature of rice plants
The d 13 C of rice plants in the control and RS treatments were almost constant with time ( Fig. 1). Rice plants in the RS treatments were enriched in d 13 C by about 5% compared with the control. The d 13 C of rice plants was consistently higher in treatment II than in treatment I, but the difference was not significant.

Rates and d 13 C of CH 4 emitted from planted microcosms
In the rice microcosms without addition of RS, CH 4 emission rates increased from the tillering stage (day 41) to the booting stage (day 55) and peaked at the flowering stage (day 70), then decreased again till the ripening stage (day 90) (Fig. 2A). Application of rice straw increased CH 4 emission rates throughout the growth period, but particularly during tillering and booting stage (Fig. 2A). The d 13 C of the emitted CH 4 became gradually more negative during the cultivation period in all the treatments (Fig. 2B). The d 13 C of CH 4 was substantially higher in RS treatment II . RS treatment I . control, especially during the tillering stage (Fig. 2B).

Concentrations and d 13 C values of CH 4 and CO 2 dissolved in pore water
Concentrations and d 13 C values of dissolved CH 4 and CO 2 were similar in the pore water sampled from 3 cm and 9 cm soil depth. Therefore, only the data from the 9-cm soil layer are shown (Fig. 3, 4A and B). In the planted microcosms, CH 4 concentrations increased steadily from the beginning until the ripening stage. Application of rice straw resulted in elevated CH 4 concentrations in the beginning but subsequently became similar to the control (Fig. 3A). The d 13 C values of the CH 4 dissolved in planted and unplanted microcosms were similar with each other in both RS treatments at tillering stage (Fig. 4A). However, while d 13 C values decreased with time in the planted microcosms, they did not decrease much in the unplanted microcosms. The d 13 C of the dissolved CH 4 was consistently higher (less negative) in RS treatment II . RS treatment I . control for both planted and unplanted microcosms (Fig. 4A). The d 13 C values of the dissolved CH 4 in planted microcosms (Fig. 4A) were similar to those of the emitted CH 4 (Fig. 2B).
In the planted microcosms, dissolved CO 2 concentrations were between 4.0 and 5.5 mM independently of the treatment and the vegetation period (Fig. 3B). The d 13 C of the dissolved CO 2 exhibited a temporal pattern similar to that of CH 4 and was again consistently higher (less negative) in RS treatment II . RS treatment I . control (Fig. 4B). However, d 13 C of dissolved CO 2 was in general higher (less negative) than that of CH 4 .

Rates and d 13 C of CH 4 and CO 2 produced in planted and unplanted microcosms
At each time of sampling, soil cores were collected from microcosms with and without rice plants, in order to determine the rates and the d 13 C of the CH 4 and CO 2 produced. Depending on the microcosm tested, CH 4 and CO 2 were produced from ROC (planted microcosms), SOM (all microcosms) and RS (RS-treated microcosms). In the planted control without RS treatment, CH 4 production rates increased steadily during the vegetation period (Fig. 5A). However, treatment with RS resulted in further increase of CH 4 production rates. In the unplanted microcosms, CH 4 production rates were also enhanced by RS treatments but were lower than in the planted microcosms with RS treatment. The d 13 C of produced CH 4 was similar in the planted and unplanted control microcosms without RS (Fig. 4C). Treatment with RS resulted in increase of d 13 C values of produced CH 4 , which was higher in treatment II than treatment I. However, the increase was less in the planted than in the unplanted microcosms (Fig. 4C).
The rates of CO 2 production were constant over the vegetation period in the planted microcosms and were similar for the treatments with and without RS, but were at least twice as high in planted as in unplanted microcosms (Fig. 5B). The d 13 C values of CO 2 exhibited a similar pattern with respect to vegetation period and treatment as that of CH 4 , but the values were generally higher (Fig. 4D).

Partitioning CH 4 and CO 2 produced in rice microcosms
For calculation of f ROC , first of all the d 13 C of the CH 4 and CO 2 produced from ROC had to be determined. The data, which were calculated using eq. (4), are shown in Table 1. The d 13 C of CH 4 produced from ROC was about 260% on average (range of 267 to 249%) during the whole vegetation period, though fluctuations on individual sampling dates, at tillering stage in particular, were rather high ( Table 1). The d 13 C values of CO 2 produced from ROC were about 231% at tillering stage and increased to around 211% to 24% subsequently ( Table 1). Values of f ROC were then calculated using eq. (2) and (3). Both equations gave similar values, but those obtained with eq. (2) showed higher standard deviations than those obtained with eq. (3). Only the latter values are shown in Fig. 6 and 7. ROC was found to make a major contribution (41-63%) to CH 4 production over the entire vegetation period (Fig. 6A). For CO 2 production, ROC had even a higher importance (43-76%) (Fig. 7A). The fractions of CH 4 and CO 2 produced from RS (f RS ) were calculated using eq. (7). Values of d 13 C were obtained from the CH 4 (Fig. 4C) and CO 2 (Fig. 4D) produced in soil samples from planted microcosms. Values of f RS were determined to be in a range of 12-24% for CH 4 production (Fig. 6B) and 11-31% for CO 2 production (Fig. 7B).
Finally, f SOM was calculated by difference to f ROC and f RS , being in a range of 23-35% of CH 4 (Fig. 6C) and 13-26% of CO 2 production in soil from planted and straw-treated microcosms (Fig. 7C).

Partitioning CH 4 and CO 2 dissolved in rice microcosms
Similarly as for the production of CH 4 and CO 2 (see above), the gases dissolved in the rice microcosms were also used for determination of the partitioning of their origin from ROC, RS, and SOM using the equations described above. In this case, values of d 13 C were from the CH 4 and CO 2 dissolved in pore water of planted and unplanted microcosms ( Fig. 4A and B). The d 13 C of CH 4 derived from ROC was 230% at tillering stage when calculated with d 13 C of CH 4 in pore water (Table 2), substantially more positive than that calculated with d 13 C of produced CH 4 ( Table 1). The resulting f ROC for CH 4 was only 13% (Fig. 6A). In  contrast, the relative contribution of RS (f RS ) to CH 4 dissolved was significantly higher than that for CH 4 produced at the tillering stage (Fig. 6B). However, the relative contributions of each carbon source to dissolved and produced CH 4 were nearly the same at later season (Fig. 6). For CO 2 , the d 13 C of CO 2 derived from ROC was 249% at tillering stage, more negative than that calculated with d 13 C of produced CO 2 (231%), but there was no significant difference between the relative contributions of each carbon source to dissolved and produced CO 2 (Fig. 7).

Partitioning of methane production
Our study comprehensively determined the partitioning of CH 4 and CO 2 production in a rice ecosystem considering all three major carbon sources (i.e., ROC, RS, SOM). In planted and straw-treated rice microcosms, more than 60% of the CH 4 was produced from root organic carbon, except on the first sampling date (tillering stage) when it was 41%. Thus, plant photosynthesis was the most important driver of CH 4 production. The same was the case for CO 2 production. The results are consistent with the observation that CH 4 and CO 2 production rates were at least twice as high in microcosms with than without rice plants ( Fig. 5A  and 5B). At the same time, the substantial lower d 13 C of CH 4 and CO 2 produced in planted versus unplanted microcosms also indicated that ROC-derived CH 4 and CO 2 production diluted the CH 4 and CO 2 produced from labeled rice straw (Fig. 4C and 4D). Our results are consistent with two earlier experiments reporting 40-60% of the CH 4 production being due to plant derived carbon. These experiments were based on pulse-labeling and FACE techniques [11,13], which potentially influence carbon flux partitioning in a different way than our approach. For instance, pulse-labeling may only account for part of the plant-derived carbon flux and FACE treatment may stimulate carbon flux [13,14]. Nevertheless, the determined relative contribution of plant derived carbon to production of CH 4 and CO 2 was rather similar despite the different approaches. Hence, the results that plant-derived carbon is the most important carbon source for CH 4 production in flooded rice fields is a rather robust finding.
In contrast to ROC, straw contributed only about 20% to CH 4 production. A similar low percentage has previously been found in Japanese rice soil microcosms after 50 days of incubation [4]. Immediately after application of the straw, however, its contribution to CH 4 production and emission reached almost 100% [4]. This was likely also the case in our experiments. This conclusion is supported by the following observations: (1) On day 41, d 13 C of the produced CH 4 was ,150% albeit the applied rice straw carbon had a d 13 C of 474.7% (Fig. 4C). The difference is much more than theoretically possible from isotope discrimination during methanogenesis. Therefore, we have to assume that the CH 4 produced immediately after straw application had a much higher d 13 C as it was derived from straw to a large extent. (2) The analogous observation was made with the produced CO 2 (Fig. 4D), although isotope discrimination is much smaller for production of CO 2 than of CH 4 . (3) Still after day 40, d 13 C of the produced CH 4 and CO 2 tended to decrease with vegetation time. Hence, we conclude that contribution of decomposition of straw to CH 4 production was very high after straw application and then progressively decreased as the carbon compounds of the straw became increasingly less decomposable. Future studies should further refine the seasonal change in flux partitioning. This will help improving the predictions of CH 4 emission rates from rice fields by process-based modeling.

Contribution of different carbon sources to the dissolved CH 4 and CO 2
Previous studies reported that d 13 C values of pore water CH 4 and emitted CH 4 were relatively poor proxies for those of produced CH 4 [32,33]. This assessment is plausible, since in rice field soil pore water CH 4 and emitted CH 4 are not only affected by CH 4 production, but also by CH 4 oxidation [34][35][36] and CH 4 transport [37][38][39], which all undergo carbon isotopic fractionation. Therefore, we primarily used the CH 4 produced in soil samples for determining flux partitioning. However, we found that not only the data of the produced CH 4 but also of the dissolved CH 4 allowed determination of flux partitioning and resulted in similar values. Thus, more than 60% of the CH 4 and CO 2  The values were calculated using d 13 C of CH 4 and CO 2 produced in rice field soil; means 6 SD (n = 4). doi:10.1371/journal.pone.0049073.t001 dissolved in soil pore water were derived from root organic carbon after tillering stage, nearly the same as for produced CH 4 and CO 2 ( Fig. 6 and 7). At tillering stage, however, the relative contribution of ROC to the dissolved CH 4 was significantly lower and that of RS significantly higher when compared to the contribution to the produced CH 4 . The difference was probably due to the gas transport limitation of rice plants at the early vegetative stage [32,40]. The residence time of CH 4 in pore water at tillering stage can amount to several days. Therefore, at day 41 the pore water was probably still highly enriched in 13 CH 4 which had been produced from RS at earlier time. This conclusion is consistent with the substantially higher d 13 C values of the dissolved CH 4 than those of the produced CH 4 at day 41 (Fig 4A and 4C). As a result, the relative contribution of RS to dissolved CH 4 was higher than to produced CH 4 at day 41 and that of ROC was lower (Fig. 6B).
In contrast, at later growth season, the residence time of CH 4 in pore water of planted soil was much shorter (several hours) [32], this was consistent with the rapid decrease of d 13 C values of dissolved CH 4 and CO 2 after tillering stage. Furthermore, the d 13 C values of dissolved and produced CH 4 were similar with each other after the tillering stage ( Fig. 4A and 4C). Therefore, the relative contributions of each carbon source to dissolved and produced CH 4 were similar to each other (Fig. 6). This suggested that pore water CH 4 could be used as a proxy for produced CH 4 and could be suitable for partitioning the CH 4 production after tillering stage.

Stable carbon isotope fractionation during CH 4 production from ROC
The d 13 C of the CH 4 produced from ROC (d 13 C CH4-ROC ) were in a range of 267% to 249%. These values are similar to d 13 C CH4 values observed in rice field soil or in incubations of soil slurries [23,33]. Theoretically the value of d 13 C CH4-ROC should be equal to the d 13 C of ROC plus the overall isotopic enrichment factor (e ROC,CH4 ) for the conversion of ROC to CH 4 . The d 13 C ROC should be similar to the d 13 C of the rice plant biomass (Fig. 1). Using these values and the d 13 C CH4-ROC , the overall enrichment factor e ROC,CH4 was in a range of about 224% to 242%. This is a rather large range, but has been observed before (about 220% to 275%) during anaerobic decomposition of straw in paddy soil [41] or anoxic incubations of rice roots [42]. The overall enrichment factor e ROC,CH4 is composed of (1) the enrichment factors involved in the conversion of ROC to the methanogenic substrates (i.e., acetate and H 2 /CO 2 ) and (2) in the enrichment factors involved in the conversion of the methanogenic substrates to CH 4 . The latter enrichment factors are the larger ones, in particular those involved in the production of CH 4 from H 2 /CO 2 [23,43]. Whereas acetoclastic methanogenesis has relatively moderate enrichment factors (210% to 225%), those of hydrogenotrophic methanogenesis are often very large (225% to 290%) [22]. Our data suggest that CH 4 production from ROC is dominated by hydrogenotrophic methanogenesis, which is consistent with earlier observations studying CH 4 production on rice roots [42,44,45].
The d 13 C of the CO 2 produced from root organic carbon was in a range of 231% to 24% ( Table 1). The overall isotopic Figure 6. Percentage contribution of (A) ROC, (B) SOM and (C) RS to produced and dissolved CH 4 in planted microcosms with RS treatment; means ± SD (n = 4). The differences between contributions to produced and dissolved CH 4 were tested by two-tailed independent ttests, indicated by * when P,0.05. doi:10.1371/journal.pone.0049073.g006 Figure 7. Percentage contribution of (A) ROC, (B) SOM and (C) RS to produced and dissolved CO 2 in planted microcosms with RS treatment; means ± SD (n = 4). The differences between contributions to produced and dissolved CH 4 were tested by two-tailed independent ttests, indicated by * when P,0.05. doi:10.1371/journal.pone.0049073.g007 enrichment factors involved in CO 2 production from organic matter were thus about 26% to +21%. These enrichment factors are much smaller than those involved in CH 4 production. Nevertheless, the range is similarly large, which may be due to carbon isotopic fractionation during CO 2 consumption by hydrogenotrophic methanogenesis [23] and also during reactions between gaseous CO 2 and bicarbonate/carbonate [46].

Practical considerations
Our study demonstrated the possibility to determine the partitioning of CH 4 and CO 2 flux from degradation of straw, soil organic matter, and plant root-derived carbon, by treating soil with 13 C-labeled rice straw. The procedure is more practical than labeling of the rice plants with 13 CO 2 that requires cumbersome incubation techniques or expensive FACE treatment. For calculation of f ROC , it was important that the d 13 C of the two RS applications were sufficiently different from each other, and in addition were sufficiently different from the d 13 C of both ROC and SOM. This was achieved by two RS treatments using the same amount of RS but 13 C-labeled to different extent. As a result, the d 13 C of emitted CH 4 (Fig. 2B), d 13 C of dissolved and produced CH 4 and CO 2 (Fig. 4) were substantially higher than the control without RS, and of course they were always higher in treatment II than treatment I.
Calculation of f RS was simply achieved by using the d 13 C values of the applied RS and the CH 4 derived from the two RS treatments (Eq. 7) assuming that ROC was not differently affected by the two RS treatments. This assumption was in agreement with the observation that the 13 C values of the rice plants in the two RS treatments were not significantly different (Fig 1). Notably, these values were significantly higher than those in the control microcosms without RS, probably because some of the RS carbon was assimilated (probably via CO 2 ) by the plants [20,21]. However, the difference was only a few permil and did not prevent computation of flux partitioning, since the difference to the d 13 C of the labeled RS was quite large.
In summary, application of labeled RS may be a convenient technique to determine flux partitioning in rice fields on a routine basis. The determination requires in total three planted field plots and three unplanted ones, i.e., two RS treatments and one untreated control, everything with appropriate replication. Technical installation is not required. Hence, it should be feasible to increase the data basis on the partitioning of CH 4 production from ROC, RS and SOM on a regional and seasonal scale. This will help improving process-based modeling of CH 4 emission from rice fields. The values were calculated using d 13 C of CH 4 and CO 2 dissolved in pore water; means 6 SD (n = 4). doi:10.1371/journal.pone.0049073.t002