FLUXES AND POOLS OF METHANE IN WETLAND RICE SOILS WITH VARYING ORGANIC INPUTS

Transcription

1 FLUXES AND POOLS OF METHANE IN WETLAND RICE SOILS WITH VARYING ORGANIC INPUTS R. WASSMANN 1'2, H. U. NEUE 1, M.C.R. ALBERTO l, R.S. LANTIN 1, C. BUENO l, D. LLENARESAS 1, J.R.M. ARAH 1'3, H. PAPEN 2, W. SEILER 2 and H. RENNENBERG 2 l International Rice Research Institute, PO. Box 933, 99 Manila, Philippines; 2 Fraunhofer Institute for Atmospheric Environmental Research, Garmisch-Partenkirchen, Germany; 3 The Edinburgh School of Agriculture, Edinburgh, U.K.; currently at Institute for Terrestrial Ecology, Edinburgh, U.K. Abstract. Measurements of methane emission rates and concentrations in the soil were made during four growing seasons at the International Rice Research Institute in the Philippines, on plots receiving different levels of organic input. Fluxes were measured using the automated closed chambers system (total emission) and small chambers installed between plants (water surface flux). Concentrations of methane in the soil were measured by collecting soil cores including the gas phase (soil-entrapped methane) and by sampling soil solution in situ (dissolved methane). There was much variability between seasons, but total fluxes from plots receiving high organic inputs (16-24 g CH4 m -2) 2 always exceeded those from the low input plots (3-9 g CH4 m- ). The fraction of the total emission emerging from the surface water (presumably dominated by ebullition) was greater during the first part of the season, and greater from the high organic input plots (35-62%) than from the low input plots (-23%). Concentrations of dissolved and entrapped methane in the low organic input plots increased gradually throughout the season; in the high input plots there was an early-season peak which was also seen in emissions. On both treatments, periods of high methane concentrations in the soil coincided with high rates of water surface flux whereas low concentrations of methane were generally associated with low flux rates. I. Introduction Fertilizer use has been found to be a crucial factor for the magnitude of methane emission from rice fields. The increase of methane emission rates by organic amendments observed in various field studies, e.g. in China (Wassmann et al., 1993b) and Japan (Yagi and Minami, 1990), has been attributed to a stimulation of methane production in the soil. The methane released from flooded soils is only a fraction of the total methane production; more than 50% of the methane locally produced is retained in the soil and oxidized by aerobic methanotrophic bacteria (Holzapfel-Pschorn et al, 1985; Sass et al., 1991). Aerobic methanotrophs grow exclusively on C1 compounds and sequentially oxidize CH4 to CO2; oxygen is essential for these eubacteria, but the partial pressure may be low (Higgins et al., 1981). The main location for methane oxidation in rice soils is thought to be the thin aerobic layer at the soilwater interface; bacterial oxidation at the root-soil interface appears to be of minor importance (de Bont et al., 1978). Environmental Monitoring and Assessment 42" , (~) 1996 Kluwer Academic Publishers. Printed in the Netherlands.

2 164 R. WASSMANN ET AL. The present study intended to combine investigations of methane fluxes to the atmosphere and methane pools in the soil in order to achieve a holistic view of the effects of organic amendments on the methane budget of rice soils. Total methane emission was monitored using automated closed chambers installed over plants; emission through the surface water (presumably largely ebullition) was measured using small chambers installed between plants. Measurements of pools encompassed the determination of the entire stock of methane in the soil and the fraction of the methane pool dissolved in soil solution. All parameters were recorded in soils fertilized by rice straw (high organic input) and in soils which received the stubble from the previous harvest as sole organic amendment (low organic input). 2. Methods and Fields Design The soil in the fields used for the present study was an Aquandic Epiaqualf with a high clay content (Neue et al., 1994). The rice cultivar planted was IR72; the fields remained flooded throughout the growing seasons. All fields received 120 kg N ha-lwhich was applied in the form of rice straw (high organic input) or urea (low organic input). All methane measurements were made using gas chromatographs equipped with Porapack N columns (0/120 mesh) and flame ionization detectors (FID). Method 2.1 involved direct sample transfer via a sample loop; methods 2.2, 2.3 and 2.4 depended on manual injection using syringes METHANE EMISSION The automatic measuring system described in detail in Wassmann et al. (1994) had the following basic features: 9 16 closed chambers (l:lm, w: lm, h: 1.2 m) distributed in the field according to a complete block design with 4 replicates; 9 a pneumatic system for automatic opening and closing of the chambers; 9 a sampling system providing direct air transfer from the inner volume of the chambers to a sample loop and a direct injection of aliquots into the GC; 9 an analytical system (GC plus integrator) linked to a data acquisition device. The measurements were conducted continuously in a 2-h cycle of operation comprising one emission measurement with a 24-min closing interval for each box (see Wassmann et al., 1994) METHANE FLUX FROM SURFACE WATER The quantity of methane emerging through the water column was collected in small Plexiglas chambers (1:40 cm, w:20 cm, h:20 cm) placed in position for 24 h between the plant rows. The upper part of the chambers was covered by aluminium foil to reflect sunlight. A septum permitted gas sampling from the inner volume

3 METHANE IN WETLAND RICE SOILS 165 of the chamber with pre-evacuated glass tubes were collected by syringe for GC analysis. These measurements were conducted once a week with 4 chambers per treatment. Measurements in the dry seasons 1993 and 1994 were started already before transplanting as soon as the fields were flooded ENTRAPPED METHANE The device for collecting soil cores consisted of a Plexiglas tube (1:25 cm, d:4.4 cm) sealed at one end by a rubber stopper. A needle penetrating through the rubber stopper linked the inner volume of the tube via a hose to a gas sampling bag. The open end of the tube was pushed vertically into the flooded soil while the gas sampling bag was gradually filled by air from the inner tube. Thus, emerging gas bubbles were trapped in the head space of the tube and in the gas bag. When the soil core sampler reached the plough pan, the gas bag was detached from the soil sampler. High clay content and high bulk density of the plough pan provided an immediate sealing of the lower side of the sampler by a thin layer of compact soil, so that the sampler could be retrieved while the soil core remained intact. The lower end of the sampler was then sealed with a rubber stopper. The core samplers containing soil, floodwater, and a small gas head space were vigorously shaken for 1 h. Methane previously entrapped in the soil matrix was released into the gas head space. The methane concentrations in the head space of the tube as well as in the gas sampling bags were determined by manual injection into a GC with a syringe. The gas volume of the head space in the tube and the sampling bag were determined by withdrawal into graduated syringes. The soil samples were oven-dried for determination of the dry weight and moisture content. The total amount of methane entrapped in the soil was computed by using the following equation, which neglects the small fraction of methane remaining in solution at the end of the shaking period: entr. CH4 = (cb~ - Camb) * Vbag + Ccore * Vc... DWsoil Cbag = CH4 concentration in the sample bag [mg/1]; Camb = CH4 concentration in the ambient air [mg/l]; Vbag = gas volume in the sample bag [1]; Ccore = CH~ concentration in the core head space [mg/1]; Vcore = gas volume in the core head space [1]; DWsoil = dry weight of the soil [rag]. The measurements were repeated in all fields in weekly intervals METHANE IN SOIL SOLUTION Soil solution was sampled through tubing of porous ceramic exposed at different depths of the soil column. Tubes were fixed on a PVC frame permitting the collection of soil solution samples by connecting a pre-evacuated glass tube ( ml volume) to a hose attached to each of these membranes and by extracting approximately 5 ml soil solution per sample. The remaining volume of the vacutainer tube

4 166 R. WASSMANN ET AL. Table I Cumulated water surface flux and total emission of methane and relative contribution of water surface flux during 1992 and 1993 wet seasons (WS) and 1993 dry season (DS) Season Low organic input High organic input Water Total Contribution Water Total Contribution surface emission of surface surface emission of surface flux flux flux flux (gch4m-2) (gch4m-2) (%) (gch4m-2) (gch4m-2) (%) 1992 WS DS WS was then filled with ambient air. The methane concentration in the head space of the vacutainer tubes was determined after vigorous shaking. The concentration of methane dissolved in soil solution was computed as follows: dissolved CH4 = C~(Ga~ + ~ KV~II - Camb* V~ Cvac = CH4 concentration in the vacutainer head space (after equilibration) [mg/l]; Vvac = volume of the vacutainer head space [1]; a = water/air partition coefficient (= 0.03 at 25 ~ Vsol = volume of soil solution [1]; Camb = CH4 concentration in the ambient air [mg/l]. The soil solution samples were collected once a week from 4 different depths; individual concentration values were averaged for each treatment. 3. Results 3.1. SURFACE FLUX AND TOTAL EMISSION Methane emission rates are shown against time in Figures 1 (urea treatment) and 2 (rice straw treatment) for the 1992 wet season; the cumulated methane fluxes of three consecutive seasons are listed in Table I. Cumulated methane emission rates showed pronounced variation between wet (WS) and dry season (DS) as well as between the two wet seasons (Table I). Emissions in the field treated with urea were consistently lower than the corresponding emissions in the field treated with rice straw. However, the ratio between these values was not constant during the course of this study. The emission in the field with organic amendments exceeded the corresponding values of the field treated with urea by factors of 2.0 (1992 WS), 8.8 (1993 DS), and 4.7 (1993 WS). The surface flux in the field with organic amendments surpassed the corresponding flux in the field treated with urea not only in absolute values but also in the relative contribution to the overall emission (Table I). Emission through the water surface accounted for 35-62% and for - 23% in fields fertilized with urea. During the first half of 1992 WS emission rates

5 METHANE IN WETLAND RICE SOILS CH 4 (mg.m-2.day-1) i Emission Ebu tion C-input: low 1~g2 WET S~SON 0 (] Treatment: UREA ~ i 50 9 m (] 5(] 60 7(] days after transplanting Figure 1. Methane flux rates of total emission (line) and water surface flux (bars) during the 1992 wet season in fields treated with urea. 3OO 250 CH 4 (rag. m-2.day- 1 ) JB EmissionJ i t:t,.,.llll;^. C-input: high 1992 WET SEASON 0 (] days after transplanting Figure 2. Methane flux rates of total emission (line) and water surface flux (bars) during the 1992 wet season in fields treated with rice straw.

6 168 R. WASSMANN ET AL. in the fields with low organic inputs were generally lower than 30 mg CH4 m -2 d-1 whereas the second half was characterized by pronounced variation of the emission rates between 30 and 170 mg CH4 m -2 d -1 (Figure 1). Emission rates in the fields with high organic amendments were generally higher than 0 mg CH4 m -2 d-1 throughout the 1992 WS and encompassed irregular fluctuations (Figure 2). Comparison of total emission and water surface flux rates in 1992 WS (Figures 1 and 2) shows that the latter was the predominant transport path in the initial phase of the growing season when plants were small. In both fields, the rates of the total flux and water surface flux were in an identical range during the early stage of the growing season. During mid- and end-season the records of emission rates were considerably higher than the simultaneous flux through the water surface which could be observed in both fields. The seasonal courses of the water surface flux rates during four consecutive seasons from 1992 WS to 1994 DS are displayed in Figures 3a, b,c,d and 4a,b,c,d. The patterns observed in these seasons corresponded to the variations described for the 1992 wet season, i.e. a gradual increase and bi-modal pattern of the flux rates in the fields with low and high organic amendments, respectively. The surface flux before transplanting was negligible (Figures 3c,d; 4c,d) ENTRAPPED AND DISSOLVED METHANE The records for entrapped methane in the 1992 WS and the I993 DS are displayed in Figures 3a,b,c,d. In fields with low organic inputs the methane pool was generally small during the first half of the season and increased gradually until the latter stage of the plant growth. The methane pool in the fields with high organic amendments built up right after transplanting, decreased during the middle of the season and increased again before harvest. At the end of the season the methane pools in both fields were practically identical around 400 #g CH4 g(soil) -1 9 The temporal course of dissolved methane in the soil during the 1993 WS and 1994 DS (Figure 4) showed a similar pattern as the entrapped methane. In the fields with low organic inputs, the methane concentration in the soil solution increased as the vegetation period proceeded. In fields with high orgarnic inputs, the methane concentrations in the soil solution followed a bi-modal pattern with an early-season peak and a late-season peak. 4. Discussion The results of this study clearly demonstrate a close relationship between pools and fluxes of methane in rice fields. The methane locally produced is temporarily retained in the soil in form of gaseous and dissolved pools, and these pools are the primary source of fluxes. Periods of high methane concentrations in the soil coincided with high flux rates through the water surface whereas low concentrations of methane were generally associated with low flux rates (Figures 3 and 4).

7 Figure 3(a,b,c,d). Water surface flux of methane (lines, left y-axis) and concentrations of soil entrapped methane (bars, right y-axis) during the 1992 wet season and the 1993 dry season r > z t~ t" z t" wet season mggh4 m-2 d-1 ugch4 g(soil)-i a -~Ebutlition~Entrapped {20! I 1993 dry season 800mgOH4 m-2 d-1 ugch4 g(soil)-i I C -'-Ebull " ~ 400 Low C-input 5. A d Iligh C-input lligh C-input Days after transplanting O0 120 Days after transplanting

8 Figure 4(a,b,c,d). Water surface flux of methane (lines, left y-axis) and methane dissolved in soil solution (bars, right y-axis) during the 1993 wet season and the 1994 dry season. > Z Z -] > 1993 wet season mgch4 m-2 d-1 ugch4 ml(sol'n)-i 8O0 20 a ~Ebu111tionB Dissolved 1994 dry season mgch4 m-2 d-1 ugch4 ml(sol'n)-i C ~-EbullitlonBBDlssolved 400 Low C-input 400 Low C-input i o ~ ~ 5 b High C-input 400 = ~ O0 Days after transplanting Days after transplanting

9 METHANE IN WETLAND RICE SOILS 171 The storage capacity of the soil is an important factor in regulating the actual flux rates of methane to the atmosphere. However, the net effect of a retention of methane by the soil on the total amount of methane emitted is not clear. The retention could represent a mere shift in the emergence pattern. The intense release of methane after harvest (Wassmann et al., 1994) indicated that the emergence of the gaseous methane pool in the soil contributed considerably to the overall emission. On the other hand, methane oxidation is a significant process in wetland rice soils which has a high potential to reduce the overall emission (Holzapfel-Pschorn et al. 1985) Apart from the availability of oxygen, methane oxidation is limited by the concentration of CH4 in the active surface layer (Conrad and Rothfuss, 1991). A long retention time of methane in the soil increases the methane pool size and accelerates the replenishment of this substrate for methanotrophs at micro-sites with aerobic conditions. The incorporation of organic material into the rice soil clearly increased both, fluxes and pools of methane. This impact reflected the stimulation of methane production by an ample availability of substrates for methanogenesis originating from the organic inputs. The stimulation of methane production by manure encompassed a maximum effect at the early stage of the vegetation period while methane production was low in the fields without organic amendments. Apparently, the impact of organic amendments on the methane budget of the rice fields subsided when pool sizes and surface flux rates decreased during days after transplanting. Methane concentration and flux rates increased during latter plant stages, indicating high methane production rates in this period. Since the late-season peaks of pools and fluxes were also observed in fields without organic amendments, a retarded impact of rice straw applications can be excluded. In the latter stage of the season, methanogenesis is based on newly formed autochthonous material; the fraction of the incorporated rice straw available for methanogenis was already consumed during the early stages. Methane can be emitted through the water surface by ebullutive or diffuse flux and can be transferred through the plant aerenchyma. Derived from the results presented, plant mediated flux can be determined indirectly by comparing total emission and water surface flux. A discrimination of fluxes through the water surface into gas bubbles and diffusion is not possible, but the latter is generally negligible in flooded rice fields due to the low solubility of methane and methane oxidation in the surface layer (Wassmann et al., 1993a). Thus, the flux rates through the water surface can be treated as dominated by ebullition. However, the accuracy of the determination of surface flux is considerably lower than the automatic records of the total emission due to the smaller size of the boxes. During the early stage of the growing season, virtually all the emission can be attributed to surface flux whereas plant mediated flux was the predominant path during latter plant stages (Figures 1 and 2). The relative contribution of surface flux to the overall emission was higher in the field with organic amendments (35-62%)

10 172 R. WASSMANN ET AL. as compared to the field with urea fertilization (-23%) (Table I). Apparently, the ample supply of substrates by organic amendments led to an intense methane formation which exceeded the storage capacity of the soil, so that a large portion of this methane was emitted as gas bubbles. In the field with low organic input, the newly formed methane was mainly retained in the soil building up the methane concentration. An improvement in the estimates of regional methane source strengths demands feasible strategies to determine the magnitude of methane emissions at various sites. The results obtained in this study indicate that the methane pool size in rice soils is an important factor in regulating the overall emission. The determination of methane pools in rice soils should be considered as a tool to compare the emission potential of a large number of rice soils. However, it remains unclear whether the close relationship between flux rates pool sizes found here in a clayey soil can be generalized to other soil types, e.g. sandy soils with a presumably lower capacity to entrap methane. Agricultural practice such as mid-season drainage could also diminish the correlation between methane concentration in the soil and flux rates. Therefore, the determination of the methane source strength from rice fields cannot rely on records of methane soil concentrations only, but has to include supplementary flux measurements. Measurements of water surface flux, which are relatively simple as compared to the determination of overall flux rates, could provide a realistic figure of the magnitude of methane emission rates during the early stage of the growing season. Methane emission rates during the latter stages of the growing season are dominated by plant-mediated flux, making field measurements of the overall flux indispensable for estimating emission rates. References de Bont, J. A. M., Lees, K. K. and Bouldin, D. F.: 1978, 'Bacterial Oxidation of Methane in a Rice Paddy', Ecol. Bull. 26, Conrad, R. and Rotfuss, F.: 1991, 'Methane Oxidation in the Soil Surface Layer of a Flooded Rice Fields and the Effect of Ammonium', Biol. Fertil. of Soils 12, Higgins, I. J., Best, D. J., Hammond, R. C. and Scott, D. C.: 1981, 'Methane-Oxidizing Microorganisms', Microbiol. Rev. 45, Holzapfel-Pschorn, A., Conrad, R. and Seiler, W.: 1985, 'Production, Oxidation, and Emission of Methane in Rice Paddiess', FEMS Microbiol. Ecol. 31, Neue, H. U., Lantin, R. L., Wassmann, R., Aduna, J. B., Alberto, M. C. R. and Andales, M. J. F.: 1994, 'Methane Emission from Rice Soils of the Philippines' in: Minami, K. Mosier, S. R. and Sass, R. L. (eds.), CH4 and N20 - Global Emissions and Controls from Rice Fields and Other Agricultural and Industrial Sources, NIAES Series 2, Tokyo, Japan, pp Sass, R. L., Fischer, F. M., Harcombe, P. A. and Turner, E T.: 1991, 'Methane Production and Enission in a Texas Rice Field', Global Biogeochem. Cycles 4, Wassmann, R., Papen, H. and Rennenberg, H.: 1993a, 'Methane Emission from Rice Paddies and Possible Mitigation Options', Chemosphere 26,

11 METHANE IN WETLAND RICE SOILS 173 Wassmann, R., Wang, M. X., Shangguan, X. J., Xie, X. L., Shen, R. X., Papen, H., Rennenberg, H. and Seiler, W.: 1993b, 'First Records of a Field Experiment on Fertilizer Effect on Methane Emission from Rice Fields in Human province (PR China)', Geophys. Res. Lett. 20, Wassmann, R., Neue, H. U., Lantin, R. S., Aduna, J. B., Alberto, M. C. R., Andales, M. J., Tan, M. J., Denier van der Gon, H. A. C., Hoffmann, H., Papen, H., Rennenberg, H. and Seiler, W.: 1994, 'Temporal Patterns of Methane Emissions from Wetland Ricefields Treated by Different Modes of N Application', J. Geophys. Res. 99, Yagi, K. and Minami, K.: 1990, 'Effects of Organic Matter Applications on Methane Emission From Japanese Paddy Fields', Soil Sc. Plant Nutr. 36,