The effective mitigation of greenhouse gas emissions from rice paddies without compromising yield by early-season drainage
S. F. U. Islam, J. W. van Groenigen, L. S. Jensen, B. O. Sander, A. de Neergaard
TThe effective mitigation of greenhouse gas emissions from rice paddieswithout compromising yield by early-season drainage
Syed Faiz-ul Islam a,b,c, ⁎ , Jan Willem van Groenigen a , Lars Stoumann Jensen b ,Bjoern Ole Sander c , Andreas de Neergaard b a Department of Soil Quality, Wageningen University, Droevendaalsesteeg 4, Building 104, 6708 PB Wageningen, The Netherlands b Department of Plant and Environmental Sciences, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark c Climate Change Group, Crop and Environmental Sciences Division, International Rice Research Institute (IRRI), Los Baños, Philippines
H I G H L I G H T S • The effects of timing and duration ofdrainage in rice soils amended withresidue were studied. • Early-season drainage (ED) in combina-tion with midseason drainage reducedCH emission up to 90%. • Yield-scaled GWPs were reduced up to87% compared to conventional continu-ous fl ooding. • ED results in stabilisation of carbon earlyin the season, restricting potential formethanogenesis. • ED is an effective option for small-scalefarmers to reduce emissions, water usewhile maintaining yield. G R A P H I C A L A B S T R A C T a b s t r a c ta r t i c l e i n f o
Article history:
Received 11 June 2017Received in revised form 3 September 2017Accepted 3 September 2017Available online 25 September 2017Editor: D. Barcelo
Global rice production systems face two opposing challenges: the need to increase production to accommodatethe world's growing population while simultaneously reducing greenhouse gas (GHG) emissions. Adaptations todrainage regimes are one of the most promising options for methane mitigation in rice production. Whereas sev-eral studies have focused on mid-season drainage (MD) to mitigate GHG emissions, early-season drainage (ED)varying in timing and duration has not been extensively studied. However, such ED periods could potentially bevery effective since initial available C levels (and thereby the potential for methanogenesis) can be very high inpaddy systems with rice straw incorporation. This study tested the effectiveness of seven drainage regimes vary-ing in their timing and duration (combinations of ED and MD) to mitigate CH and N O emissions in a 101-daygrowth chamber experiment. Emissions were considerably reduced by early-season drainage compared toboth conventional continuous fl ooding (CF) and the MD drainage regime. The results suggest that ED + MDdrainage may have the potential to reduce CH emissions and yield-scaled GWP by 85 –
90% compared to CFand by 75 –
77% compared to MD only. A combination of (short or long) ED drainage and one MD drainage episodewas found to be the most effective in mitigating CH emissions without negatively affecting yield. In particular,compared with CF, the long early-season drainage treatments LE + SM and LE + LM signi fi cantly ( p b Keywords:
MethaneNitrous oxideDrainage timing and durationRedox potentialRice straw managementGlobal warming potential Science of the Total Environment 612 (2018) 1329 – ⁎ Corresponding author at: Department of Soil Quality, Wageningen University, Droevendaalsesteeg 4, Building 104, 6708 PB Wageningen, The Netherlands
E-mail address: [email protected] (S.F. Islam).http://dx.doi.org/10.1016/j.scitotenv.2017.09.0220048-9697/© 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Contents lists available at ScienceDirect
Science of the Total Environment he season, thereby reducing available C for methanogenesis. Overall N O emissions were small and not signi fi -cantly affected by ED. It is concluded that ED + MD drainage might be an effective low-tech option for small-scalefarmers to reduce GHG emissions and save water while maintaining yield.© 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
Rice (
Oryza sativa ) is the most important agricultural staple for morethan half of the world's population and is grown in 114 countries over atotal area of around 153 million ha, which is 11% of the world's arableland (FAOSTAT, 2011). Rice production must increase by 40% by theend of 2030 to meet rising demand from a growing world population(FAO, 2009). However, rice cropping systems are considered to beamong the major anthropogenic sources of methane (CH ) and nitrousoxide (N O). Estimates of global CH emissions from paddy soils rangefrom 31 to 112 Tg y − , accounting for up to 19% of total emissions,while 11% of global agricultural N O emissions come from rice fi elds(US-EPA, 2006; IPCC, 2007). GHG emissions from rice cropping systemshave therefore raised serious concerns (van Beek et al., 2010; W. Zhanget al., 2011a; G. Zhang et al., 2011b). There is an urgent need to reduceGHG emissions to the atmosphere to mitigate the adverse impacts of cli-mate change. Therefore rice cropping systems in future will need tocombine increased rice yields with decreased GHG emissions.Methane is the dominant GHG emitted from rice systems in terms ofglobal warming potential (GWP). It is an end product of organic matterdecomposition under anaerobic soil conditions (Conrad, 2002; Linquistet al., 2012a, 2012b), therefore the two strategies most often proposedto reduce CH emissions are to limit the period of soil submergence(i.e. draining the fi eld) and reduce carbon inputs (through residue man-agement). Of these, water management is the management option mostoften studied. Several fi eld studies have shown that the drainage of wet-land rice once or several times during the growing season effectively re-duces CH emissions (Wassmann et al., 1993; Yagi et al., 1997; Lu et al.,2000; Towprayoon et al., 2005; Itoh et al., 2011). Through mid-seasondrainage (MD) and intermittent irrigation, the development of soil re-ductive conditions can be prevented, leading to reduced CH emissions.MD can reduce the total CH emission during the rice-growing periodby 30.5% (Minamikawa et al., 2014). Additional reported bene fi ts in-clude the reduction of ineffective tillers, the removal of toxic substancesand the prevention of root rot, leading to increased yields and reducedwater use (Zou et al., 2005; Itoh et al., 2011).With respect to residue management, according to the InternationalRice Research Institute (IRRI) about 620 million tons of rice straw areproduced annually in Asia alone and this quantity is increasing everyyear (IRRI, 2016). In most places, these rice straws have no commercialvalue and are disposed of in various ways. Burning these residues in the fi eld is the most common practice, especially in Asia, because it elimi-nates numerous pathogens, kills weeds and is less laborious thanstraw incorporation (Mendoza and Samson, 1999; Kutcher and Malhi,2010). Both burning in the fi eld (resulting in air pollution and associatedhealth risks) and soil incorporation (resulting in methane emissions)are the cause of environmental concerns. In intensive cropping systems,where two or three crops are grown each year, straw production is sohigh and the time for residue decomposition so short that farmers tradi-tionally opt to burn the straw. The air pollution associated with thisburning is severe because combustion is often incomplete, resulting inthe emission of large amounts of pollutants such as SO and NOx aswell as toxic gases such as carbon monoxide (CO), dioxins and furans,volatile organic compounds (VOC) and carcinogenic polycyclic aromatichydrocarbons (PAH) (Jenkins et al., 2003; Oanh et al., 2011). Gadde et al.(2009) has calculated that 1 kg of rice straw burnt directly on the fi eldemits 1.46 kg of CO , 34.7 g of carbon monoxide (CO) and 56 g of dust. The Chinese government and some states in India have alreadybanned straw burning and in the near future many other Asian coun-tries will do likewise because of its association with serious humanhealth hazards, the huge carbon cost (the CO price), and loss of essen-tial nutrients such as N, P, K and S (Dobbermann and Fairhurst, 2002;Duong and Yoshiro, 2015). There is therefore likely to be an increasedfocus on straw incorporation in future, but the application of ricestraw usually increases CH emissions more than that it counteractsN O emissions in terms of GWP (Xu et al., 2000; W. Zhang et al.,2011a; G. Zhang et al., 2011b); Xia et al., 2014; Yuan et al., 2014).However, these studies also suggest that the effects of the straw ap-plication on CH emissions strongly depend on management ( fl oodingmanagement facilitating aerobic conditions during decomposition) orclimatic conditions. The effect of MD on CH emissions by altering soilredox conditions is well established, but there is still limited under-standing of the effect of the timing and duration of drainage on CH emissions. In particular, drainage during the early season (ED) could po-tentially be effective as the system still contains large amounts of avail-able C from rice straw. ED (as well as MD) could lead to aerobicdecomposition and stabilisation of this available C. However, little isknown about the effects of ED on either GHG emissions or yield. Fromthe studies of mid-season drainage (MD) and intermittent irrigation, itis evident that although increasing drainage frequency and durationcould strengthen the mitigation effect on CH emissions (Wassmannet al., 2000), it may have adverse effects on rice plant growth, resultingin a reduction in grain yield (Lu et al., 2000). However, their impacts onrice yield have been found to be mixed, e.g. signi fi cantly negative(Towprayoon et al., 2005; Li et al., 2011; Xu et al., 2015), signi fi cantlypositive (Qin et al., 2010) or unaffected (Minamikawa and Sakai,2005; Wassmann et al., 2000; Itoh et al., 2011; Yang et al., 2012;Pandey et al., 2014; Vu et al., 2015). The contradictory effects of drain-age regimes on rice yield might be attributed to the difference in drain-age duration and frequency, the level of water stress during thedrainage periods, rice variety and crop management (Belder et al.,2004; Feng et al., 2013). Therefore to ensure food security for theworld's increasing population, the impact of any mitigation strategyon GHG emissions and yield should be investigated simultaneously.Moreover the potential trade-offs between CH and N O emissionsresulting from MD have been well documented in paddy fi elds (e.g.Cai et al., 1997; Zou et al., 2005). Johnson-Beebout et al. (2009) haveshown in a pot experiment without rice plants that enhanced N O emis-sions could potentially outweigh the bene fi t of reduced CH emissionsunder the alternate wetting and drying (AWD) strategy in terms ofGWP. It is not yet clear what the effect of ED would be on N O emissions.An assessment of the effect of alternative water management strategieson GHG emissions should therefore focus on emissions of N O as well asCH .This study tested for the fi rst time the effect of combinations of EDand MD, varying in timing and duration, on emissions of CH and N Oand on yield in a growth chamber experiment with rice straw incorpo-ration. The overall objective was to investigate whether the combina-tion of ED and MD suppresses CH emissions from paddy systemsincorporated with rice straw while maintaining yield and without cor-respondingly raising N O emissions. The speci fi c research questionswere: i) Does a combination of ED and MD drainage reduce CH emis-sion without signi fi cantly increasing N O emission? ii) Is the length ofthe drainage period important in terms of GHG emissions? iii) Does
S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 – arly-season drainage affect grain yield and yield-scaled GHG emis-sions? and iv) What is the relative contribution of different alternativewater management regimes to CH and N O in terms of GWP? The spe-ci fi c hypotheses were: i) in the mid-season drainage treatment, early-season drainage further reduces CH emissions without increasingN O emissions ii) longer early-season drainage treatments furtherdecrease CH emissions but increase N O emissions, irrespective of theduration of MD, iii) early-season drainage treatments do not affectgrain yield but decrease yield-scaled GHG emissions, and iv) early-season drainage treatments decrease GWP without increasing the rela-tive contribution of N O emissions.
2. Materials and methods
The experiment was performed in a growth chamber at the Facultyof Science in the University of Copenhagen, Denmark from April toJuly 2014. The temperature was 28 °C/22 °C day/night with a 12-hday/12-h night light regime (450 μ E m − s − ). The experimental lay-out was a completely randomised block design with seven treatmentsand three replicates.Rice was grown in cylindrical plexiglass soil columns (inner diame-ter 14 cm, height 30 cm) on which cylindrical headspace chambers(60 cm tall) were mounted for gas sampling and made air-tight with aring frame with a water seal (Fig. 1). The headspace chambers were suf- fi ciently tall for the Vietnamese early maturing local inbred rice variety(75 –
85 cm height) used in the experiment. At the bottom of the soil col-umn, a nylon mesh was mounted to enable drainage through a smalltube, closed with a simple valve. Two electric fans (4 cm, 12 V DC)were installed inside the gas-sampling chamber, one at the top andthe other at the bottom facing opposite directions to ensure an homog-enous air mixture. Gas samples were collected through a rubber septumplaced in the top of the chamber. To measure the temperature inside thechamber a thermometer probe was inserted through another rubberseptum. The air-tightness of the chamber was tested using infraredgas analysis (IRGA) (WMA-2 CO analyser, PP Systems, UK) (Ly et al.,2013) and CO leakage was found to be minimal ( b Rice seedlings were grown in a nursery bed in the growth chambersfor 14 days and then transplanted, with two seedlings per soil column.The soil used for growing rice paddies was sandy loam (Luvisol, FAO/Al fi sols, USDA soil Taxonomy), which has highly favourable physicalcharacteristics (high water-holding capacity) for tropical rice produc-tion, collected from a NPK-fertilised plot in the Crucial Long-TermField Trial (Magid et al., 2006). In previous years, approximately 100 –
150 kg ha − inorganic N and between 10 and 50 kg ha − P and Kwere applied each year. Detailed soil characteristics can be found inTable 1.The soil was homogenously mixed and 21 columns were fi lled withan equivalent of 3 kg dry soil each. The water regime varied according tothe treatments, which are depicted in Fig. 2. The seven treatments were:i) continuous fl ooding (CF), ii) short mid-season drainage (SM), iii) longmid-season drainage (LM), iv) short early-season drainage (SE) andshort mid-season drainage, v) short early-season drainage and longmid-season drainage (LM), vi) long early-season drainage (LE) andshort mid-season drainage and vii) long early-season drainage andlong mid-season drainage. After packing the soil, water was added toeach column to saturate the soil and then kept saturated for threedays. One day before transplanting and before the morning addition ofwater, basal applications of organic substrates (rice straw) were appliedand homogenously mixed with the soil in the columns and then 10 mmwater added. On the transplanting day (0 days after transplanting(DAT)), fl oodwater was drained from the soil columns but kept moistfor transplanting. Transplanting was performed early in the early morn-ing (7:00 – fl ooded (i.e. watertable near the soil surface) in all ED (early-season drainage) treatments,while 10 mm standing water above the soil surface was maintained forthe rest of the treatments. For the establishment of the seedlings in theED treatments, a few drops of water were added around the rice shoottwice (morning and afternoon) at 1 DAT. The water level was increasedover time during the rice season; the water regime was 10 –
30 mm forthe fi rst 16 DAT, 40 mm for 17 –
24 DAT and 50 mm from 25 DAT untilten days before harvest. During drainage, fl oodwater was allowed toevaporate from the soil columns to avoid nutrient loss through drainageand the water maintained at 5 –
10 mm above the soil surface for just oneday before the two periods of fertilizer top dressing. The soil columnswere fl ooded again with demineralised water after fertilisation or drain-age periods. All treatments received identical applications of mineral fertilizer(N, P and K) and straw. Urea (46% N) was applied as nitrogen fertilizerat the rate of 0.54 g per soil column (corresponding to 160 kg N ha − )in three split doses: 30% of the N fertilizer was applied at 7 DAT, 35% Fig. 1.
Schematic drawing of soil column and gas chamber (adapted from Lindau et al.,1991) used to collect CH and N O emissions.
Table 1
Soil properties.Properties ValueSoilCoarse sand (%) 26Fine sand (%) 36Silt (%) 17Clay (%) 19pH (1 M KCl) 6.41CEC (c mol kg − ) 3.72Total N (g 100 g − ) 0.18Total C (g 100 g − ) 2.17Total P (mg g − ) 0.56Total K (mg g − ) 173.7 1331 S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 – t 35 DAT and subsequently the last 35% at 60 DAT. Single superphos-phate (7.2% P) and Muriate of Potash (50% K) were used as P and Kfertilisers at the rate of 90 kg ha − and 60 kg ha − respectively. Fullrates of P and K fertilisers were applied during transplanting. The ricestraw applied was collected from Vietnamese rice fi elds of a local inbredvariety. Before application the rice straw was cut into 1 – − ). The applied rice straw contained 0.6% N, 0.1% P, 0.07% S,1.7% K, 5% Si and 41% C, and had a C:N ratio of 68. Redox probes (METTLER TOLEDO) were permanently installed in fi ve of the seven treatments, namely CF, LM, SE + LM, LE + SM, LE+ LM, because of the limited availability of probes. The redox potential(Eh, mV) readings were taken on each sampling day. At 101 DAT theplant attained physiological maturity with N
80% ripe grains. At thatpoint, the aboveground plant biomass and grains were harvested,oven-dried and weighed.
Gas samples were collected on 1, 3, 5, 7, 14, 21, 26, 28, 30, 32, 34, 37,44, 54, 61, 71, 81, 91, 101 DAT during rice growth, always during thedaytime between 8.00 am and 11.30 am. After placing the top chamberon the base, gas samples were taken at 20-min intervals at 0, 20, 40 and60 min using 10 ml syringes. Gas samples were removed through therubber septum with a gas-tight syringe and stainless steel hypodermicneedle. Collected gas samples were immediately transferred into evacu-ated 3-ml vial (12.5 mm diameter, Labco limited, UK).The concentrations of CH and N O were analysed using a gas chro-matograph (Bruker 450-GC 2011) equipped with a separate electroncapture detector (operated at 350 °C for N O analysis) and fl ameionisation detector (operated at 300 °C for CH analysis). The oven tem-perature was set at 50 °C. Helium (99.99%) and Argon +5% CH wereused as carrier gases of CH and N O respectively at a fl ow rate of60 ml min − . Certi fi ed reference CH and N O gases were used for cal-ibration and quality control during every batch of gas analyses. The CH and N O fl uxes were calculated according to Smith and Conen(2004) and Vu et al. (2015). All emissions were converted to CO -equivalents (Hou et al., 2012),with GWP for CH set at 34 relative to CO (based on a 100-year timehorizon), and set at 298 for N O (IPCC, 2013; Wang et al., 2015). Thearea-based net GWP of the combined emission of CH and N O was cal-culated following Ahmad et al. (2009). Yield-scaled emissions were cal-culated, as de fi ned by Van Groenigen et al. (2010), as a ratio of growingseason net GWP and rice grain yield, expressed in kg CO -equivalentsper kg of grain yield. Statistical analyses were performed using Statistical Analysis System(SAS) Proprietary Software version 9.4 (SAS Institute Inc., Cary, NC,USA). Data were checked for independence, normality and homogenei-ty of variance. Analysis of variance (ANOVA) with repeated measures onCH and N O fl uxes were performed separately using the Mixed Proce-dure in SAS. The dependent variables of grain yield, yield components,biomass, seasonal CH and N O emissions, GWP and yield-scaled GWPwere analysed using the GLM Procedure in SAS. The type of response(linear, quadratic or cubic) was identi fi ed from the level of signi fi canceand an F-test of the difference in R . All differences were considered tobe signi fi cant at the 95% level ( P b fi cant difference) test.
3. Results
Table 2 shows yield components, grain yield and aboveground bio-mass. The rice plants in the drainage treatments generally showedmore vigour than the plants with CF. The total aboveground biomasswas lowest in the CF and highest in the treatment with a long early-sea-son drainage (LED) regime (LE + SM and LE + LM). In contrast, thegrain yield was highest in the CF treatment followed by LE + LM and
Fig. 2.
Illustration of treatments CF =
Continuous fl ooding , SM = Short Mid - season drainage , LM = Long Mid - season drainage , SE = Short early - season drainage , LE = Long early - seasondrainage .1332 S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 – E + SM and the lowest from the LM treatment. However, in terms ofgrain yield no signi fi cant difference was found between the treatmentsexcept with LM. Similarly, no signi fi cant difference was found in termsof yield components between the treatments. fl uxes The dynamics pattern of CH fl uxes over the whole rice-growingperiod was strongly affected by the water regime (Fig. 3). After thesoil columns were waterlogged, CH emissions increased steadily untilthe peak fl uxes were attained within the third week after transplanting.The highest CH peak was found in the CF treatment at 96 mg m − h − at 26 DAT, while the LE + SM and LE + LM treatments involving the LEDdrainage regime had the lowest peak with 34 mg m − h − and30 mg m − h − respectively. Methane fl uxes were drastically reducedby the mid-season drainage (SM and LM) period at the end of the tiller-ing stage (from 28 to 35 DAT), which varied between four and sevendays based on the treatments. In the late stages of the rice-growingperiod (54 DAT), small secondary CH fl ux peaks occurred in the CFtreatment that were absent from early-season drainage treatments.Fluxes from the CF treatment were consistently higher than thosefrom the other treatments. A one-way ANOVA showed that CH emis-sions from rice plants were signi fi cantly affected by water regime andthat there was a signi fi cant difference between the treatments.When comparing CF with traditional mid-season drainage varying inlength (SM and LM), the peak of SM and LM was observed to be slightlylower than CF (Fig. 3a), after which there was a signi fi cant drop in emis-sions in both SM and LM during the mid-season drainage period. Whenthe short early-season drainage (SED) treatment was included (treat-ments SE + SM and SE + LM, Fig. 3b), there was an additional decreasein the height of the fi rst peak. This drop was even greater after a longer(7-day) early-season drainage (LED) treatment (LE + SM and LE + LM;graph 3c). Moreover, the post-drainage depression lasted much longerthan the mid-season drainage period in the ED treatments dependingon the SED/LED treatments. fl uxes N O fl uxes varied considerably as a result of water regime and timeof fertilizer addition in the rice-growing period (Fig. 4). Nitrous oxide fl uxes were found to be two orders of magnitude lower than methane fl uxes. For all treatments, three N O emissions peaks were observedthat were all associated with N fertilizer application (7 DAT, 35 DATand 60 DAT respectively). Treatment CF had the lowest N O emissions,whereas the treatment with long mid-season drainage (LM) had thehighest N O emission. The SED treatments SE + SM and SE + LMshowed average N O fl uxes of 0.25 mg m − h − and 0.31 mg m − h − respectively, about 1.5 and two times higher than that in CF. For theLED treatments LE + SM and LE + LM, the average N O fl uxes were0.22 mg m − h − and 0.25 mg m − h − respectively, about 1.4 and1.5 times higher than those from the conventional CF treatment. Despite the increase in total drainage duration, the LED treatments LE + SM andLE + LM resulted in reduced N O emissions compared to their counter-parts SE + SM, SE + LM and LM. fl uxes The cumulative CH emissions from the paddy soils during the over-all rice-growing season showed very signi fi cant differences between allthe treatments (Fig. 5a). The highest cumulative CH emission was re-corded from treatment CF with 868 mg per soil column (correspondingto 563 kg CH ha − ). Long early-season drainage reduced CH remark-ably, with the LE + SM and LE + LM treatments having the lowest emis-sions of all the treatments with the cumulative 102 (66 kg ha − ) and 88(57 kg ha − ) mg CH per soil column respectively. Compared with CF,cumulative CH emission was signi fi cantly different ( p b fi cant differences between LE + SM and LE + LM.However, CH emissions from SM and LM were found to be signi fi cantlyreduced by 53% and 55% respectively compared to CF. SED treatmentsresulted in 8% and 20% reduction of CH emissions when added withthe SM and LM treatments, while the LED treatments resulted in a 34%and 35% reduction respectively.A signi fi cant effect ( p b O emissions (Fig. 5b). The lowest total cumulativeemission was 2.14 mg per soil column (1.39 kg ha − ) from the CF treat-ment. The highest total cumulative N O emissions were observed fromLM, followed by SE + LM, with emissions rates of 3.83 (2.49 kg ha − )and 3.77 (2.45 kg ha − ) mg per soil column respectively. The totalcumulative N O emissions from SM and LM were signi fi cantly higher( p b fi cantly higher (p b fi cant difference in emissionreduction when added to SM, but signi fi cantly (p b O emission when added to LM (LE + LM) by 42%.
The redox potential (Eh) of the soil during the rice-growing period ofthe fi ve treatments ranged from 78 to −
255 mV for CF, 65 to −
246 mVfor LM, 72 to −
225 mV for SE + LM, 70 mV to −
175 mV for LE + SMand 75 mV to −
168 for the LE + LM water treatment (Fig. 3d). Gener-ally the redox potential of the soil increased when the soil columns weredrained and decreased when the soil columns were re- fl ooded. Averagesoil Eh was found to decrease gradually after fl ooding until the soil col-umns were drained. After re- fl ooding, the soil Eh remained negativeuntil the soil columns were drained before harvesting. The redox poten-tial of the LED treatments were signi fi cantly different from the CFtreatment. In this study, a direct correlation between redox potentialand methane fl ux was found in all the treatments throughout the rice-growing season. Table 2
Biomass, grain yield and yield components affected by different alternate water management regimes.Irrigation Treatments Panicles column − Spikelets panicle − Grain fi lling (%) Grain weight (mg) Grain yield g column − Biomass g column − CF 9.3 ± 0.5 a a a a a c SM 9.3 ± 0.8 a a a a a c LM 9.1 ± 0.9 a a a a b b SE + SM 9.3 ± 1.0 a a a a a ab SE + LM 9.1 ± 0.9 a a a a ab a LE + SM 9.3 ± 1.0 a a a a a a LE + LM 9.3 ± 0.5 a a a a a a Data shown are the means ± standard deviation of three replicates. Within the column, the values with different letters are signi fi cantly different at p b fl ooding, SM = Short Mid-season drainage, LM = Long Mid-season drainage, SE = Short early-season drainage, LE = Long early-season drainage. 1333 S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 – .6. GWP and yield-scaled GWP There were signi fi cant ( p b + N O) was 14,507 kg CO equiv. ha − from the CF treatment,which was signi fi cantly higher than the other water treatments.Compared with SM and LM, treatments with SED showed a signi fi cantreduction in CH emission but increased N O emissions. However, theLED treatments (LE + SM and LE + LM) resulted in a remarkable reduc-tion of net GWP (CH + N O) by reducing both CH and N O emissionscompared to SM and LM. Consequently, the signi fi cantly lowest netGWP was found in LE + LM (1921 kg CO equiv. ha − ), followed bythe LE + SM (2123 kg CO equiv. ha − ) and SE + LM water treatments(4228 kg CO equiv. ha − ).Similarly, yield-scaled GWP was highest for the CF treatment (3.4 kgCO equiv. kg − rice grain), and signi fi cantly ( p b fi cantly (p b fi cant differencewas found between LE + SM and LE + LM. In contrast, mid-seasondrainage treatments (SM and LM) reduced yield-scaled GWP on aver-age by just 37%. Moreover, the LED treatments reduced yield-scaledGWP on average by 77% compared to mid-season drainage.
4. Discussion emissions The rate of CH emission gradually increased with the age of plantsduring the fi rst three weeks and showed two peaks: one at the early til-lering stage at around 21 –
26 DAT and another at the panicle initiationstage at around 54 DAT in treatments with CF and MD (SM and LM).Treatments with ED (SE + SM, SE + LM, LE + SM and LE + LM) showedonly the fi rst peak (Fig. 3). This fi rst emission peak was in line with pre-vious studies, where a fi rst peak was also found between 14 and 30 DAT(Wang et al., 1999; Ly et al., 2013; Vu et al., 2015; Wang et al., 2015).This is predominantly associated with the development of anaerobicsoil conditions, readily degradable C from the rice straw amended inthe soil, and the rapid growth of rice plants that facilitates the plant-mediated transport of CH .For the CF, SM and LM treatments the soil was continuously fl oodeduntil 28 DAT, including the fi rst peak. During continuous fl ooding ofpaddy soil, trapped O is rapidly respired and the soil undergoes reduc-tion processes (Takai and Kamura, 1966). The presence of readilyavailable organic substrates from rice straw in the fl ooded soil furtherenhances the reduction process by supplying electron donors, therebycreating an anaerobic environment (Wassman and Aulakh, 2000)leading to methanogenesis (Le Mer and Roger, 2001).The treatments with ED (SE + SM, SE + LM, LE + SM and LE + LM)underwent a 4 – (Woese et al., 1978). More importantlyit stabilises reactive C (from amendments and soil) by aerobic decom-position, which is likely to progress more quickly during the oxygenatedstages, resulting in less substrate for methanogens after subsequent fl ooding (Pandey et al., 2014). Plant growth was clearly affected byearly season drainage, most markedly by larger vegetative biomass. Fur-thermore, rice plants under aerobic soil conditions have been shown tohave less developed aerenchyma compared to those under anaerobicconditions (Kludze et al., 1993), which might have further reducedCH transportation and emissions. Therefore in terms of emissions, thedifference between treatments with early-season drainage and thosewith non-early-season drainage is clearly visible due to the delay inthe onset of emission and the suppression of the particular CH emissionpeaks that occur early in the cultivation season by ED. Fig. 3.
Temporal pattern of CH emissions as affected by different alternate watermanagement regimes compared to baseline irrigation regime CF and their relationshipwith soil redox potential. CF = Continuous fl ooding , SM = Short Mid - season drainage , LM = Long Mid - season drainage , SE = Short early - season drainage , LE = Long early - seasondrainage . Error bars indicate 1 S . E . M . ( n = ).1334 S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 – ethane emissions in all treatments peaked in the tillering stageand then decreased gradually due to the complete drainage of thepaddy fi eld at 28 –
35 DAT. According to Zhu (2006), this complete drain-age in the later stage of tillering is an important management practicethat is used to fi nish the tiller process of rice and supply rice roots with O to prevent sul fi de toxicity (Kanno et al., 1997) and also acciden-tally helps to reduce CH emissions. This temporal variability is consis-tent with patterns observed in other studies of CH fl ux in paddy fi elds, which have similarly found a decrease in CH emissions as a re-sult of mid-season drainage. This could be attributed to the increasinglyaerobic conditions in the sediments that suppress methanogenesis andthus CH emissions (Jia et al., 2001; Tsuruta, 2002; Towprayoon et al.,2005; Ali et al., 2013; Kim et al., 2013; Singh et al., 2003).Along with CF, two mid-season drainage treatments (SM and LM)reached a second peak at around 54 DAT. This second CH emissionpeak is in line with previous fi ndings (Schutz et al., 1989; Neue et al.,1997; Ly et al., 2013; Vu et al., 2015). This second peak is attributed tothe decay of crop organic matter, such as dead roots and root exudates(Schutz et al., 1989; Neue et al., 1996; Chidthaisong and Watanabe,1997) during the later stage of rice growth, as well as the slow decom-position of straw under continuous fl ooding, as reported by most of theauthors (Kimura et al., 2004; Gaihre et al., 2011). Furthermore, plant-mediated transport of CH is particularly ef fi cient at this stage of plantgrowth because of the well-developed aerenchymatous system (Wanget al., 2015).The major difference in fl ux pattern between the treatments withand without early-season drainage after mid-season drainage was theabsence of a second CH peak in the early-season drainage treatments.Despite the re- fl ooding of the paddy pots after mid-season drainage atthe fi nal tillering stage, CH emissions remained very low, whereasthey increased markedly in the non-early season drainage treatments.This could be attributed to further stabilisation of reactive C during themid-season drainage, which already had a smaller amount of availablecarbon due to early-season drainage.The reviews of Minami (1995) and Yan et al. (2009) report that sea-sonal CH emissions range from 2.7 to 1059 kg ha − for paddy fi eldsaround the world. Hence, the seasonal CH emissions measured in thepresent study (57 to 563 kg ha − ) are within the range of the publishedvalues. The seasonal methane emissions of this study's SED and LEDtreatments ranged from 57 to 314 kg ha − , which were comparable torecently reported results (18 –
320 kg ha − ) for intermittent irrigationacross China (Xie et al., 2010). In this study, higher soil Eh values were observed during the drain-age period compared to the soil columns that were continuously fl ooded. CH ef fl uxes increased as the soil Eh decreased, and decreasedrapidly after the columns were drained out as soil Eh increased. This is in Fig. 4.
Temporal pattern of N O emission as affected by different alternate water management regimes. Error bars are omitted for improved clarity. TD represents top dressing of Nfertilizer. CF =
Continuous fl ooding , SM = Short Mid - season drainage , LM = Long Mid - season drainage , SE = Short early - season drainage , LE = Long early - season drainage . Fig. 5. a) Total accumulated CH emissions as affected by different alternate watermanagement regimes b) Total accumulated N O emissions as affected by differentalternate water management regimes. CF =
Continuous fl ooding , SM = Short Mid - seasondrainage , LM = Long Mid - season drainage , SE = Short early - season drainage , LE = Longearly - season drainage . Error bars indicate 1 S . E . M . ( n = ). Different letters indicatesigni fi cance ( p b . ) of treatments rice production ( small letters ). 1335 S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 – greement with previous studies (Wang et al., 1993; Minami, 1994;Tyagi et al., 2010). According to Masscheleyn et al. (1993) methane isusually formed only after the soil Eh is reduced to below −
100 mVwith a near neutral pH of the fl ooded soil. However, other studieshave differentiated between critical soil Eh values for laboratory versus fi eld studies and indicated critical redox potentials of −
150 mV and −
100 mV respectively (Wang et al., 1993; Hou et al., 2000; Tyagi etal., 2010). The onset of signi fi cant CH emissions was found to startwhen the soil redox potential dropped below −
127 mV for the CF treat-ment. Overall, the critical Eh value for signi fi cant CH emissions rangedbetween −
155 to −
245 mV for the different water treatments, whichis consistent with previous results for laboratory experiments by Tyagiet al. (2010). However, during the post drainage period in the ED treat-ments, soil Eh was found to pass the theoretical threshold of −
100 mV,yet the emissions were not signi fi cantly different from zero. This mayindicate the absence of C substrate in the system. O emissions N O emissions were low compared to CH emissions in terms ofGWP. This is in line with previous studies (Abao et al., 2000; Ly et al.,2013; Vu et al., 2015). Bronson et al. (1997) have reported that N Oemissions are rarely detected during the rice season except directlyafter fertilisation. Yao et al. (2012) have reported negligible N O emis-sions during three continuous years under fl ooded conditions. Signi fi -cant N O peaks were evident in all the treatments only after Nfertilizer application. This is in line with previous studies (Pathak etal., 2002; Zou et al., 2005; Pandey et al., 2014). Readily available N sub-strate after topdressing of mineral N might have enhanced nitri fi cation in aerobic zones and subsequent denitri fi cation in anaerobic zones ofthe rhizosphere, resulting in induced N O emissions (Pandey et al.,2014). Based on the literature, fl uxes were measured directly afterfertilisation, as the peaks were associated with fertilisation and slightchanges in sampling day or time of day would have signi fi cant effectson the total cumulative emission. In future studies, higher temporal res-olution of sampling after fertilisation will bene fi t the validity of the com-mutative fl ux. Furthermore sampling was not undertaken every day, soa small number of measurements account for almost the entire seasonalcumulative fl ux sampling. There is therefore a possibility that N O fl uxeswere underestimated and the cumulative N O fl ux should beinterpreted with care. However, the low variation in the present datasuggests that some general conclusions can be drawn. The lowest andhighest N O emissions were observed in the CF and LM treatments re-spectively. Prolonged fl ooding promotes the development of strong an-aerobic conditions in soils, reducing any N O produced in the paddy fi elds to N (Ussiri and Lal, 2013). Long mid-season drainage could cre-ate prolonged partly anaerobic conditions in the soil, favouring simulta-neous nitri fi cation and denitri fi cation and resulting in considerable fl uxes of N O (Davidson et al., 2000; Pathak et al., 2002). These fi ndingsare in line with greenhouse experiments by Johnson-Beebout et al.(2009), who report that alternate wetting and drying increases N Oemissions from paddy soils relative to a continuously fl ooded treatment.Some fi eld studies have also found that MD and intermittent irrigationincrease nitrous oxide (N O) emissions compared to the CF treatment(e.g. Yan et al., 2000; Nishimura et al., 2004; Towprayoon et al., 2005;Jiao et al., 2006; Zou et al., 2005, 2009). However, N O emissionsaccounted for b
15% of the relative total annual emissions and did noteliminate the overall reduction in global warming potential (Tsurutaet al., 1998; Kurosawa et al., 2007; LaHuea et al., 2016). Despite thehigher total soil aeration duration in LED treatments, these treatmentsshowed the potential to emit less N O compared to the MD treatmentsSM and LM alone. When soil is well aerated, the oxidation, i.e. nitri fi ca-tion, of available N dominates and NO is the most common gas emittedfrom soil instead of N O (Davidson et al., 2000). As the rice-growing sea-son progresses, N is taken up by the plant or otherwise leached(Cassman et al., 1998), resulting in low N O emissions (Xing et al.,2002; Pandey et al., 2014). When taking into account both CH andN O fl uxes, the LED treatments had 7.5 and 4.5 times lower total globalwarming impacts than CF and MD respectively. In light of the likely in-crease in straw additions and the practical challenges of drainage, theLED strategy will facilitate the mineralising of C that can have long-term effects on emission and is likely to be a better option than the CFand MD strategy in terms of the total GHG budget. The area-based GWP of CH and N O emissions in the CF treatmentwere 21,705 and 637 mg per soil column (corresponding to 14,094 and414 kg CO equiv. ha − ) respectively. This is in agreement with themeta-analysis by Feng et al. (2013) who report an average GWP ofCH and N O emissions of 14,331 and 699 kg CO equiv. ha − Table 3
Cumulative emissions of CH and N O over the WRGS, grain yield and total CO -e GWP over the 100-year time horizon as affected by organic matter and water management.Irrigation treatments Grain yield(mg g soil − ) CH emission(mg g soil − ) N O emission( μ g g soil − ) GWP CO -e (mg g soil − )2007 IPCC, AR4 GWP CO -e (mg g soil − )2013 IPCC, AR5CF 6.81 ± 0.1 a
334 ± 7.9 a c a a SM 6.40 ± 0.1 a
179 ± 2.9 b bc b b LM 6.07 ± 0.2 b
186 ± 5.3 c a b b SE + SM 6.45 ± 0.1 a
123 ± 7.1 d b c c SE + LM 6.17 ± 0.1 ab
83 ± 6.3 e ab d d LE + SM 6.72 ± 0.1 a
39 ± 4.1 f bc e e LE + LM 6.81 ± 0.1 a
34 ± 5.3 f bc e e Data shown are means ± standard deviation of three replicates. Within the column, the values with different letters are signi fi cantly different at p b are 25(AR4)/34(AR5) and for N O it's 298 times higher than CO over the 100-year time horizon, respectively. Fig. 6.
Yield-scaled GWPs as affected by different alternate water managementregimes. CF =
Continuous fl ooding , SM = Short Mid - season drainage , LM = Long Mid - season drainage , SE = Short early - season drainage , LE = Long early - season drainage . Error bars indicate 1 S . E . M . ( n = ). Different letters indicate signi fi cance ( p b . ) oftreatments of rice production ( small letters ).1336 S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 – espectively. The long early-season drainage treatment showed veryhigh mitigation potential, reducing the area-based GWP of CH andN O to 2203 and 755 mg per soil column (corresponding to 1431 kgand 490 kg CO equiv. ha − ) respectively. A more relevant measuremight be yield-scaled emissions, linking GHG emissions directly to agri-cultural production (Pathak et al., 2010; Van Groenigen et al., 2010;Linquist et al., 2012a, 2012b; Venterea et al., 2011). In the presentstudy, plant growth was clearly affected by ED treatments, most mark-edly by larger vegetative biomass and plant vigour, which may havebeen due to better root growth and development (Table 2). Despitethe inclusion of conditions that could trigger high CH emissions, theLED treatment LE + LM showed potential to reduce yield-scaled GWPby 87% with only 0.43 kg CO equiv. kg − rice grain (430 kg CO equiv. Mg − ). This yield-scaled GWP from the LE + LM treatment was62% lower than the yield-scaled GWP (1146.3 kg CO equiv. Mg − ) re-ported by Pathak et al. (2010) and 34% less than the yield-scaled GWPfrom the global rice production (657 kg CO equiv. Mg − ) reported byLinquist et al. (2012a, 2012b).However, the present study was limited by controlled laboratoryconditions and relatively small soil columns and therefore yield andyield-scaled GHG data are not directly comparable to fi eld conditions.Given the very high importance of yield and yield-scaled data to farmersif they are to adopt new mitigation options, yield and yield-scaled emis-sion are represented here. Despite being a highly controlled growthchamber study, these results are considered to be highly relevant for fu-ture fi eld studies in different geographical regions. First, the measuredemissions (area as well as yield-scaled) were found to be within thelevels found in fi eld studies, as reported in recent meta-analyses (Yanet al., 2009; Feng et al., 2013). Second, conditions were chosen thatclosely mimic those in paddy fi elds. Reports from rice production sys-tems indicate that differences in CH emissions are the main contribu-tors to signi fi cant differences in yield-scaled GWP. Several factors cansigni fi cantly affect CH emissions from paddy fi elds, mainly soil temper-ature, crop residue management and water management. This studycarefully mimics the tropical rice fi eld conditions in climate chamberswhere soil temperature closely resembles topical soil temperaturesalong with a large input of rice straw (10 t ha − ). Similarly, the Nfertilisation dose was carefully selected following the guidelines fromthe meta-analysis by Feng et al. (2013), which showed that the largestreduction in yield-scaled GWP occurred at the rate of 150 –
200 kg N ha − . Some previous studies (Towprayoon et al., 2005; Li etal., 2011; Xu et al., 2015) report that MD and intermittent irrigation re-sults in a reduction in rice yield, indicating the importance of consider-ing drainage regimes both for the impacts on rice yield and GHGemissions, which guided the design of the current study. In future fi eldstudies, it might also be of interest to extend the calculation of netGWP to include the soil carbon balance (Mosier et al., 2006). Rice production uses approximately 40% of the world's irrigationwater. Close to one third of these areas experience water shortages, ne-cessitating water-saving strategies without compromising yield. Mid-season drainage alone can reduce emissions by up to one third com-pared to CF, but practices such as alternate wetting and drying (AWD)have shown greater water-saving and emission reduction potential(Sander et al., 2015). However, to practise AWD, farmers must fi rst beable to allow their fi elds to dry, and then must have a reliable sourceof water to rewet their fi elds as soon as it is needed and repeat this ina more or less continuous cycle. Therefore in order to implementthese practices, farmers need reliable control over irrigation water andusually also require small, well-levelled fi elds to avoid pockets thatdry excessively in the distant part of the fi eld that would impact riceyields with repeated drying cycles. In many developing Asian countries,full-scale AWD is therefore often not feasible because farmers have limited technical ability to suf fi ciently drain and re- fl ood their fi eldsduring the rainy season. In the dry season, farmers who rely on surfaceirrigation systems have a tendency to be reluctant to interrupt irrigationwhen water is available because of uncertainties around water availabil-ity when it is needed. In some of these locations, early-season drainageplus mid-season drainage could be an effective means of reducingmethane emissions, since a delay to the start of fl ooding at the startand a single long mid-season drawdown may still be feasible. However,emission reductions alone do not motivate the adoption of these watermanagement techniques since they do not directly bene fi t farmers. Inareas where farmers receive water through gravity-driven irrigation,these farmers rarely bene fi t fi nancially from reducing their water usebecause they do not pay for the quantity of water they use. In contrast,many farmers who rely on pump-driven irrigation do directly bene fi tfrom saving water, providing a potential incentive to reduce the dura-tion of fl ooding. To date the low adoption of mitigation strategies, e.g.AWD, indicates the importance of incentives to increase their adoption.Incentives such as carbon payments as a form of compensation for re-ducing emissions compared to traditional practice and subsidised agri-cultural inputs, e.g. good quality seeds, fertilisers etc. would encouragefarmers to test new mitigation techniques. No signi fi cant reduction inyield was found between the conventional CF and LED drainage re-gimes, meaning that even without incentives the LED regime appearsto be economically competitive with lower input costs, e.g. water sav-ings and equivalent yield. However, it is recognised that the high poten-tial of ED treatments may not be fully reached under fi eld conditions asour plant growth chamber experiment was conducted under near-opti-mal conditions. In particular, it is dif fi cult to maintain homogenouswater level conditions and homogeneous levels of applied residue inthe fi eld. Therefore our results should be validated in fi eld experimentsin different geographical regions with differing plant growth conditions.
5. Conclusions
Our results suggest a strong potential for early-season drainage toreduce the total GHG budget from rice paddy systems. Although the re-sults should be validated in the fi eld under realistic conditions, the LEDtreatments appeared to effectively mitigate seasonal CH emissions rel-ative to conventionally managed CF and MD water regimes while stillmaintaining grain yield. They also signi fi cantly reduced net yield-scaledGHG emissions. No signi fi cant difference was found between the twoLED treatments (LE + SM and LE + LM), con fi rming the fact that the du-ration of mid-season drainage is not important when the duration ofearly-season drainage is suf fi ciently long. These results also stronglysuggest that LED and SED can facilitate a decrease in net GWP fromrice paddy fi elds. Therefore, short and long early-season drainage treat-ments are proposed in addition to one mid-season drainage episode asan effective mitigation measure. In regions where farmers have limitedtechnical ability to drain their fi elds on a regular basis and face uncer-tainty in water availability, too much wet season water or uneven fi elds,the full-scale practice of AWD is not feasible. Simple alternate drainageregimes such as ED + MD might be an effective low-tech option forthose small-scale farmers to reduce GHG emissions and save waterwhile maintaining yield. Acknowledgements
This work has been conducted as part of a Ph.D. fellowship projectsupported by the Agricultural Transformation by Innovation (AGTRAIN),Erasmus Mundus Joint Doctorate Programme, funded by the EACEA(Education, Audiovisual and Culture Executive Agency) of the EuropeanCommission (agreement no. 2013-008). This work was further supportedby the Climate and Clean Air Coalition (CCAC) and the CGIAR ResearchProgram on Climate Change, Agriculture and Food Security (CCAFS),which is carried out with support from CGIAR Fund Donors andthrough bilateral funding agreements. For details please visit https://
S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 – cafs.cgiar.org/donors. The views expressed in this document cannot betaken to re fl ect the of fi cial opinions of these organisations. References
Abao, E.B., Bronson, K.F., Wassmann, R., Singh, U., 2000. Simultaneous records of methaneand nitrous oxide emissions in rice-based cropping systems under Rainfed condi-tions. Nutr. Cycl. Agroecosyst. 58, 131 – fi eld under different rice tillage systems in central China.Soil Tillage Res. 106, 54 – – – – fl ux in a fl ooded Rice soil: I. Residue, Nitrogen, and WaterManagement. 61, pp. 981 – fi elds as affected by nitrogen fertilisers and watermanagement. Plant Soil 196, 7 – fi ciency from improved resource manage-ment in irrigated rice systems. Field Crop Res. 56, 7 – fl ooded ricesoil incorporated with 13C-labeled rice straw. Soil Biol. Biochem. 29, 1173 – fi elds. Nutr.Cycl. Agroecosyst. 64, 59 – – – fi elds inChina: a meta-analysis. Agric. Ecosyst. Environ. 164, 220 – fi eld burning in India, Thailand and the Philippines. Environ. Pollut. 157,1554 – fl uxes from irrigated lowland rice fi elds.Philipp. Agric. Sci. 94, 335 – – and N O emis-sions in response to water management of paddy fi elds located in Southeast China.Chemosphere 89, 884 – fi elds by prolonging midseason drainage. Agric. Ecosyst. Environ. 141, 359 – – production, transport, oxida-tion and emission in rice paddy soil. Plant Soil 230, 211 – fl uenc-ing methane and nitrous oxide emissions from rice fi eld in relation to soil redox andmicrobial community. Commun. Soil Sci. Plant Anal. 37, 1889 – – fi elds in all of Japanese prefecture: relationship between emission rates and soil char-acteristics, water treatment and organic matter application. Nutr. Cycl. Agroecosyst.49, 147 – – fi eld ecosystems in the contextof input, decomposition and translocation of organic materials and the fates of theirend products (CO2 and CH4). Soil Biol. Biochem. 36, 1399 – – fi eld cultivation. J. Life Cycle Assess. Jpn. 3, 232 – – – – – fi eld.Plant Soil 136, 195 – – – – fi cation (SRI) under a rain-fed low-land rice ecosystem in Cambodia. Nutr. Cycl. Agroecosyst. 97, 13 – fi eld trial with urban fertilizers - is recycling of nutrients from urban areas to peri-urban organic farms feasible? In: Raupp, J., Pekrun, C., Oltmanns, M., Köpke, U.(Eds.), Long-term fi eld experiments in organic farming. Stollfuß Verlag Bonn GmbH& Co. KG, Berlin, pp. 59 –
78 (Scienti fi c Series/ISOFAR)Masscheleyn, P.H., DeLaune, R.D., Patrick Jr., W.H., 1993. Methane and nitrous oxide emis-sions from laboratory measurements of rice soils suspensions: effect of soil oxidationreduction status. Chemosphere 26, 251 – fi cient Agricultural Production (REAP), Canada,pp. 1 – – fl ooded rice. Fertil. Res. 40, 71 – – – – fi elds. Atmos. Environ. 30, 1751 – fi elds. Nutr. Cycl. Agroecosyst. 49,111 – fi eld with Japanese conventional water managementand fertilizer application. Glob. Biogeochem. Cycles 18. http://dx.doi.org/10.1029/2003GB002207.Oanh, N.T.K., Bich, T.L., Tipayarom, D., Manadhar, B.R., Prapat, P., Simpson, C.D., Liu, L.-J.S.,2011. Characterization of particulate matter emission from open burning of ricestraw. Atmos. Environ. 45, 493 – – – – S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 – in, Y., Liu, S., Guo, Y., Liu, Q., Zou, J., 2010. Methane and nitrous oxide emissions from or-ganic and conventional rice cropping systems in Southeast China. Biol. Fertil. Soils 46,825 – fl uence of daytime, season, and fertilizer treatment onmethane emission rates from an Italian rice paddy. J. Geophys. Res. 94, 16405 – fi elds. Environ. Int. 29, 547 – – fi elds. Chemosphere 59, 1547 – fi elds. Paper Pre-sented at World Congress of Soil Science, Bangkok Thailand, 14 – fi elds on atmospheric and aquatic envi-ronments. Proceedings of the Third International Conference on Eco Balance,pp. 209 – — a tool for methane mitiga-tion from irrigated paddy fi elds. Sci. Total Environ. 408, 1085 – greenhouse gas emissions: 1990 – fi elds. In: Ussiri, D., Lal, R. (Eds.),Soil Emission of Nitrous Oxide and Its Mitigation. Springer, Netherlands, pp. 213 – O emissions: a case study for arable crops.Eur. J. Soil Sci. 61, 903 – – fi elds on Vietnamese smallholder livestock farms. Nutr. Cycl.Agroecosyst. 103, 329 – – fi elds as affected by organic amendment, water regime, crop establishment, andrice cultivar. Environ. Monit. Assess. 57, 213 – – – – fi elds in Asia. III. Mitigation options and futureresearch needs. Nutr. Cycl. Agroecosyst. 58, 23 – – fi t in a rice – wheat cropping system inChina. Agric. Ecosyst. Environ. 197, 118 – emission from rice fi elds: multi-site fi eld observations. Plant Soil 326, 393 – – emission from rice cultivation. Aust. J. Soil Res. 38,1 – – – – fi eld as affected by water-saving irrigation. Phys. Chem. Earth A, Parts A/B/C53 –
54, 30 – O and CH emissions from a sandy loam paddy during rice seasons as affected by different nitro-gen application rates. Agric. Ecosyst. Environ. 152, 1 – – – production and emission from permanently fl ooded rice fi elds. Nutr.Cycl. Agroecosyst. 89, 81 – fi cation. China Agricultural Science and TechnologyPress, Beijing.Zou, J., Huang, Y., Jiang, J., Zheng, X., Sass, R.L., 2005. A 3-year fi eld measurement of meth-ane and nitrous oxide emissions from rice paddies in China: effects of water regime,crop residue, and fertilizer application. Glob. Biogeochem. Cycles 19. http://dx.doi.org/10.1029/2004GB002401.Zou, J., Liu, S., Qin, Y., Pan, G., Zhu, D., 2009. Sewage irrigation increased methane and ni-trous oxide emissions from rice paddies in southeast China. Agric. Ecosyst. Environ.129, 516 – S.F. Islam et al. / Science of the Total Environment 612 (2018) 1329 ––