Zhiqin Wng,Dojin Gu,Srh S.Beeout,Ho Zhng,Lijun Liu,Jinhng Yng,*,Jinhu Zhng

aJiangsu Key Laboratory of Crop Genetics and Physiology,Co-Innovation Center for Modern Production Technology of Grain Crops,Yangzhou University,Yangzhou 225009,Jiangsu,China

bCrop and Environmental Sciences Division,International Rice Research Institute(IRRI),Metro Manila,Philippines

cDepartment of Biology,Hong Kong Baptist University,Hong Kong,China

Keywords:Rice(Oryza sativa L.)Dry direct-seeding Alternate wetting and drying Furrow irrigation Water productivity Methane emission

A B S T R A C T Dry direct-seeded rice grown in raised beds is becoming an important practice in the wheat–rice rotation system in China.However,little information has been available on the effect of various irrigation regimes on grain yield,water productivity(WP),nitrogen use efficiency(NUE),and greenhouse gas emission in this practice.This study investigated the question using two rice cultivars in 2015 and 2016 grown in soil with wheat straw incorporated into it.Rice seeds were directly seeded into raised beds,which were maintained under aerobic conditions during the early seedling period.Three irrigation regimes:continuous flooding(CF),alternate wetting and drying(AWD),and furrow irrigation(FI),were applied from 4.5-leaf-stage to maturity.Compared with CF,both AWD and FI significantly increased grain yield,WP,and internal NUE,with greater increases under the FI regime.The two cultivars showed the same tendency in both years.Both AWD and FI markedly increased soil redox potential,root and shoot biomass,root oxidation activity,leaf photosynthetic NUE,and harvest index and markedly decreased global warming potential,owing to substantial reduction in seasonal CH4 emissions.The results demonstrate that adoption of either AWD or FI could increase grain yield and resource-use efficiency and reduce environmental risks in dry direct-seeded rice grown on raised beds with wheat straw incorporation in the wheat–rice rotation system.

1.Introduction

Rice(Oryza sativa L.)is one of the most important food crops in the world,consumed by ＞3 billion people[1,2].China is the main producer of rice,contributing ＞28%of total global rice production[3,4].Historically,transplanting is the primary method for rice establishment under puddled and flooded conditions worldwide[5,6].Two major challenges,however,limit the development of transplanted and flooded rice(TFR).One is water shortage.TFR is estimated to consume ＞50%of the fresh water diverted for human uses[5,7].Fresh water for irrigation,however,is becoming increasingly scarce because of population growth,increasing urban and industrial development,and decreasing availability resulting from pollution and resource depletion[8–10].The other is labor shortage.The process of conventional manual transplanting has a large labor requirement,but China is facing a labor shortage as more and more farmers seek higher-paying work in urban areas[6,11].Mechanized transplanting of rice can save labor,but raising seedlings and handling them during transplanting still require much labor.Furthermore,flooded rice fields have been identified as an important source of atmospheric methane(CH4),one of the major potent greenhouse gases(GHG),and contribute approximately 15%–20%of global total anthropogenic CH4emission [12, 13]. Although puddling favors rice–rice cropping systems by creating hardpans and reducing water loss through percolation,it can negatively affect non-rice upland crops that follow rice in the rotation,by dispersing soil aggregates,reducing permeability in subsurface layers,and forming hardpans at shallow depths[6,14].

To counter water shortage and/or labor shortage,several water-saving and/or labor-saving techniques have been developed,including dry direct-seeded rice(DDSR)[3,6,15],irrigation with alternate wetting and drying(AWD)[10,16–18],and furrow irrigation(FI)[14,19–21].In DDSR,seeds are sown in nonpuddled and unsaturated soil,soil is kept under aerobic conditions until the four-or five-leaf stage,and then the water level is allowed to fluctuate between 5 and 10 cm throughout the rice-growing season[3,15].The technique can save substantial water and reduce labor costs relative to TFR,but its grain yield varies with location and cultivar type[3,6,15]. The AWD technique, characterized by alternation of periods of soil submergence with periods of nonsubmergence during the growing season,can substantially reduce irrigation water consumption and lead to improved water productivity(WP)[8,10,17,18,22].However,whether the technology can achieve the dual goal of increasing grain yield and saving water is debated[8,10,18,23–26].In FI,irrigation is applied by running water down the furrows remaining from previous crops such as wheat,maize,and soybean[14,19,21].FI can increase WP and suppress false smut disease,although it shows yield penalties in comparison with continuous flooding irrigation[14,19,21].To date,little work has been done to compare simultaneously the effects of DDSR,AWD,and FI on rice yield and WP,especially for rice grown on raised beds.

The wheat–rice rotation system is one of the major crop systems in China,especially in the lower reaches of the Yangtze River, a region characterized by an irrigated environment of intensive winter wheat–summer rice double cropping over an annual planting area of 13 Mha[14,27].In most situations in such a system,wheat is grown in raised beds,which drain faster after rain,produce higher yields,and require less irrigation water than flat fields under conventional tillage without raised beds[14,27].Rice,in contrast,is transplanted and grown in puddled and flooded flat fields[5,10,28].To save labor and irrigation water and reduce tillage cost,in recent years more and more farmers are sowing rice seeds in raised beds.However,increasing grain yield and saving labor and irrigation water in direct-seeded rice grown in raised beds remains a challenge[14,15,21].

The disposal of wheat residue preceding a rice crop is another challenge in the rice–wheat cropping system.Farmers always burn the crop residue, particularly when they want to establish the rice crop rapidly while labor is limited.This practice leads to loss of most organic carbon and large losses of nitrogen(N),phosphorus(P),and potassium(K)as well as severe air pollution and death of beneficial soil fauna and microorganisms[11,29].For this reason,incorporation of crop residues into the field soil is currently enforced in China as a measure to promote sustainable agricultural production[11,18].However,this practice provides readily available carbon and N substrate,inducing greater CH4release from rice paddy fields and also influencing N2O emissions[11,30–32].Although it is reported[6,11,14]that DDSR,AWD,and FI can decrease CH4emissions,little information is available on the effects of various irrigation regimes on CH4emissions in direct-seeded rice grown on raised beds and when wheat residues are incorporated into rice fields.

The main objective of this study was to compare the effects of CF,AWD,and FI on grain yield,WP,nitrogen use efficiency(NUE),and greenhouse gas(GHG)emissions in dry direct-seeded rice grown in raised beds and when wheat straw is incorporated into field soil.Some agronomic and physiological traits that are closely associated with rice growth, namely proportion of productive tillers, leaf area duration(LAD),leaf photosynthetic rate(Pr),leaf photosynthetic nitrogen use efficiency(PNUE),nonstructural carbohydrate(NSC)in the stem and its remobilization during grain filling,root biomass,and root oxidation activity(ROA),were investigated to elucidate the biological process by which irrigation regimes affect rice yield,WP,and NUE.Such a study was expected to provide useful information about the use of dry direct seeding of rice to achieving higher grain yield and resource use efficiency and reducing environmental risk by adopting appropriate irrigation regimes.

2.Materials and methods

2.1.Plant materials and cultivation

Field experiments were conducted at a research farm of Yangzhou University,Jiangsu,China(32.30′N,119.25′E,21 m altitude)during the rice growing season(June–October)of 2015 and 2016.The soil was a sandy loam(Typic Fluvaquent,Entisols,U.S.classification)that contained 24.2 g kg?1organic matter, 101 mg kg?1alkali hydrolyzable N, 32.4 mg kg?1Olsen-P,and 65.6 mg kg?1exchangeable K.The field capacity soil moisture content was 0.190 g g?1and the bulk density of the soil was 1.35 g cm?3. The average air temperature,precipitation,and sunshine hours during the rice growing season across the two study years,measured at a weather station close to the experimental site,are shown in Fig.S1.

The previous crop was wheat.Before wheat sowing,the field was plowed,harrowed,and leveled,and then beds were raised using a tractor-mounted furrow opener (Yangzhou Agricultural Machinery Group,Yangzhou,China).The width of the beds(mid-furrow to mid-furrow)was 1.50 m,with 1.30 m wide flat tops and 0.20 and 0.15 m furrow width and depth,respectively.During the wheat harvest,wheat straw was chopped to approximately 5 cm in length with a combine[YanMar AW82GR(4LZ-28),Lianyungang Agricultural Machinery Co.Ltd.Lianyungang,China].Before rice sowing,all the beds were dry-plowed and harrowed without puddling.Fresh wheat straw was incorporated to a soil depth of 0–12 cm in the beds during tillage.The amount of wheat straw incorporated was approximately 6300 kg ha?1dry weight(DW)containing 35 kg N,6.0 kg P,and 70 kg K ha?1on average.After wheat straw incorporation,the beds from the preceding season were reshaped and furrows were cleaned using a tractor-mounted bed planter(Yangzhou Agricultural Machinery Group).Two rice (Oryza sativa L.) cultivars, the indica/japonica hybrid Yongyou 2640(YY-2640)and the japonica inbred Huaidao 5(HD-5),currently planted in local production,were used.In both years,dry seeds were sown manually in the beds on June 10 in 0.20-m-wide rows,with seven rows in each bed.The seeding rate was 30 kg ha?1for YY-2640 and 60 kg ha?1for HD-5.The reason for the smaller seeding rate for YY-2640 than for HD-5 was that YY-2640 has more spikelets per panicle and needs fewer panicles per unit area than HD-5 to achieve a comparable grain yield.On the first day of sowing,the beds were flooded and then kept moist but not saturated until the 4.5-leaf stage(July 2)when irrigation treatments were applied. A combination of herbicides (bensulfuron methyl (30.0 active ingredient(AI)g ha?1) and pretilachlor(450 g AI ha?1)was applied pre-emergence on June 11 followed by pyribenzoxim application at 45 g AI ha?1at approximately the 3-leaf-stage on June 24.In both years,N(48 kg ha?1as urea),P(30 kg ha?1as single superphosphate)and K(40 kg ha-?1as KCl)were applied and incorporated one day before sowing.Nitrogen as urea was also applied at early seedling(14 days after sowing(DAS)(36 kg ha?1),early tillering(32 DAS,48 kg ha?1),panicle initiation(51 DAS,60 kg ha?1)and the initiation of spikelet differentiation(71 DAS,48 kg ha?1).The total N application was 240 kg ha?1,within the recommended range. Both cultivars (50% of plants) headed on September 4–5(86–87 DAS),and were harvested on October 22(134 DAS)in both years.

2.2.Irrigation treatments

The experiment was laid out in a randomized complete block design under split plot arrangements with four replications.Three irrigation regimes:continuous flooding(CF),alternate wetting and drying(AWD),and furrow irrigation(FI),were assigned to main plots,with the two rice cultivars(YY-2640 and HD-5)constituting subplots(6 m×5 m).Main plots were separated by an alley 1 m wide with plastic film inserted into the soil to a depth of 0.50 m to form a barrier.Subplots were separated by a bund 0.40 m wide wrapped with plastic film.Irrigation treatments started from 4.5-leaf-stage(July 2)until one week before the final harvest of plants(October 14–15).In the CF regime, the water level was allowed to fluctuate between 3 cm and 5 cm on the raised beds during the treatment period,following recommended farming practice.In the AWD regime,fields were not irrigated until soil water potential reached ?15 kilopascal(kPa)at 15–20 cm depth in the center of beds.In the FI regime,irrigation was applied by running water down the furrows until the beds were wetted(soil water potential=0) to the center. Water was then withheld until the soil water potential in the center of beds reached approximately ?5 kPa at 15–20 cm depth.Soil water potentials of ?15 kPa in the AWD regime and ?5 kPa in the FI regime were chosen because our earlier work[20,25,28]showed that a mild soil-drying regime(soil water potential?15 kPa at 15–20 cm depth)in the AWD regime and ?5 kPa in the FI regime during the growing season did not reduce grain yield relative to the CF regime. Soil water potential was monitored at 15–20 cm soil depth in the bed center with a tensiometer consisting of a sensor of 5 cm length(Institute of Soil Science,Chinese Academy of Sciences,Nanjing,Jiangsu,China).Four tensiometers were installed in each subplot,and readings were recorded at 11:30 a.m.each day.When soil water potential reached the threshold,beds in AWD plots were flooded to 2.0–2.5 cm depth,and irrigation was applied to FI plots through furrows until wetting in the centers of beds.The amount of irrigation water was monitored with a flow meter (LXSG-50 flow meter, Shanghai Water Meter Manufacturing Factory, Shanghai, China) installed in the irrigation pipelines. Both irrigation and drainage systems were built between the main blocks.Each plot was irrigated or drained independently.

2.3.Measurements of plant physiological traits and soil redox potential(Eh)

Leaf water potentials of the topmost fully expanded leaves on stems were measured at 11:30 a.m.with clear sky at 45(D1)and 95(D2)DAS in 2015 and at 46(D1)and 96(D2)DAS in 2016 when soil water potentials in the bed centers were approximately ?15 kPa in the AWD regime and ?5 kPa in the FI regime,and at 47(W1)and 97(W2)DAS in 2015 and at 48(W1)and 98(W2)DAS in 2016 when plants were rewatered.The corresponding growth stages were close to panicle initiation(D1 and W1)and early grain filling(D2 and W2).Two pressure chambers(model 3000,Soil Moisture Equipment Corp.,Santa Barbara,CA,USA)were used for leaf water potential measurement,with eight leaves for each treatment.

On the same dates,the photosynthetic rate(Pr)and N content of the topmost fully expanded leaves were measured.Four gas exchange analyzers(LI-COR 6400 portable photosynthesis measurement system,LI-COR,Lincoln,NE,USA)were used to measure leaf Pr.The measurement was performed during 9:00–11:00 a.m.when photosynthetic active radiation above the canopy was 1300–1500 μmol m?2s?1.For the same measurement,the difference in Pr was ＜5%among the four gas exchange analyzers.Each gas exchange analyzer was used in one replicate in a treatment, and the data from four replications (four analyzers) were then averaged for each treatment, with the aim of minimizing errors between treatments caused by measurement differences among different gas exchange analyzers.Eight leaves were measured for each treatment. N content in leaves was determined following Yoshida et al.[33].Photosynthetic NUE(PNUE)of leaves was calculated from the photosynthetic rate and the specific N content of leaves;thus,PNUE(μmol g?1s?1)=Pr(μmol m?2s?1)/specific leaf N content(g m?2).

Twenty plants in each subplot were tagged for recording tiller number.Counts were made at early tillering(31–32 DAS),jointing stage(50–51 DAS),heading time(86–87 DAS),and maturity(133–134 DAS).The proportion of productive tillers was defined as the number of panicles developed from tillers as a percentage of the number of tillers at the jointing stage.

Total aboveground biomass and leaf area index(LAI)were measured at early tillering, panicle initiation (51 DAS),heading time,and maturity.Plants from 10 plants from each subplot were sampled from the third row to minimize border effects. All plant samples were separated into green leaf blades,stems(culms+sheaths), and panicles(at heading time and maturity). Dry matter of each component was determined after drying at 70°C to constant weight.The amount of nonstructural carbohydrate (NSC) in the stem(culm+sheath)was determined at heading time and maturity.The method for extraction and determination of NSC(starch+soluble sugars)was modified following Yoshida et al.[33].Briefly,the sample was dried in an oven and ground into fine powder.In a 15 mL centrifuge tube,100 mg of ground sample was added to 10 mL of ethanol(density 630 g L?1)followed by immersion in a water bath at 80°C for 30 min.The tube was then centrifuged at 5000×g for 20 min after cooling.The supernatant was collected and the extraction was repeated three times.The supernatant was heated in a water bath at 80°C until most of the alcohol was removed and the volume was reduced to approximately 3 mL.The sugar extract was then diluted to 25 mL with distilled water.The concentration of sugars in the extract was then analyzed as described by Yoshida et al.[33].

The residue left in the centrifuge tube was dried at 80 °C for starch extraction. Two milliliters of distilled water was added to the tube containing the dried residue. The tube was then shaken in a boiling water bath for 15 min. Two milliliters of 9.36 mol L?1HClO4was added to the tube after cooling. The solution was shaken for 15 min. The extract was then made up to about 10 mL and centrifuged at 5000 ×g for 20 min. The supernatant was collected and a further 2 mL of 4.68 mol L?1HClO4was added to the residue. The extraction was repeated as above. The supernatants were combined and made up to 50 mL with distilled water. The starch was analyzed following Yoshida et al. [33]. The NSC in stems was expressed as g kg?1DW, and NSC per m2was expressed as g m?2[NSC content(g kg?1) × stem weight (kg m?2)].

Crop growth rate (CGR) and NSC remobilization were calculated using the following formulas:

where W1and W2are the first and second measurements of shoot biomass(g m?2),respectively,and t1and t2represent the first and second times(days)of measurement.

The area of green leaf blades from ten plants in each subplot was measured with an area meter(LI-3050C,LI-COR).Leaf area duration(LAD)and net assimilation rate(NAR)were calculated using the following formulas:

where W1and W2are the first and second measurements of shoot biomass(g m?2),L1and L2are the first and second measurements of leaf area(m2m?2),respectively,t1and t2represent the first and second times(days),of measurement.

Root dry weight and root oxidation activity(ROA)were measured on the same dates as leaf photosynthesis measurement(D1,D2,W1,and W2).To maintain canopy conditions,the vacant spaces left after sampling for measurements of root and shoot biomasses were immediately replaced with plants taken from the borders and these replanted plants were not subjected to further sampling.

For each root sampling,a 20 cm×20 cm×20 cm cube of soil around each individual row was removed with a sampling core.Such a cube contains approximately 95%of total root biomass[34,35].Each sample contained 4–5 plants(YY-2640)or 7–8 plants(HD-5)and was used for each measurement.The roots in each cube of soil were carefully rinsed with a hydropneumatic elutriation device(Gillison's Variety Fabrications,Benzonia,MI,USA).After fresh weight was recorded,portions of each root sample were used for the measurement of ROA or root dry weight.ROA was determined by measuring oxidation of alpha-naphthylamine(α-NA)following Zhang et al.[36],and expressed as μg α-NA per gram DW per hour(μg α-NA g?1DW h?1).Root DW was determined after drying in an oven at 70°C to constant weight.

Soil Eh was monitored at 10 cm soil depth using Pt-tipped electrodes(Hirose Rika Co.Ltd.,Yokogawa,Japan)and an oxidation-reduction potential meter with a reference electrode(Toa PRN-41),with four replications for each treatment.

2.4.Measurement of GHG emissions

Fluxes of CH4and N2O were measured from sowing to maturity using a static vented flux chamber technique described previously[11].Briefly,the chamber included a permanent base that was inserted into the soil(with rice plants growing inside); extensions of varying length to accommodate the growing plants;and a lid equipped with vent tube,fan and thermocouple wire.The base was made of a PVC frame(0.5 m×0.5 m)and buried to a depth of 15 cm to leave about 10 cm above the soil line.Holes drilled in the base above and below the soil line allowed for relatively free root and water movement.During sampling,holes above the water line were plugged with rubber stoppers when the water level was below the holes to ensure that the chambers were airtight.One chamber was employed in each subplot and positioned at least 0.5 m inside the subplot,and sampling locations were connected using boardwalks to prevent soil disturbance during sampling.The chamber was wrapped with layers of sponge and aluminum foil to minimize air temperature changes inside the chamber during the period of sampling.

Gas flux measurements were made at 2-day intervals for 10 days after each fertilization and at 5-day intervals during other growth periods. For each flux measurement, gas samples were collected from 9:00–11:00 a.m.with a 20-mL syringe at 0,10,20,and 30 min after chamber closure and were analyzed for CH4,N2O,and CO2concentrations with a gas chromatograph(Agilent 7890A,Agilent Technologies,Palo Alto,CA,USA)equipped with two detectors.N2O was detected with an electron capture detector(ECD),and CH4with a hydrogen flame ionization detector(FID).CO2was reduced with hydrogen to CH4in a nickel catalytic converter at 375°C and then detected by the FID.The carrier gas was argon–methane(5%)at a flow rate of 40 mL min?1.The temperatures of the column and ECD detector were maintained at 40°C and 300°C,respectively.The oven and FID were operated at 50°C and 300°C,respectively.Total CH4(TCN4)and total N2O(TN2O)emissions were calculated as follows:

where Didenotes the number of days,Fithe measured flux in the ith sampling interval,and n the number of sampling intervals.

The global warming potential(GWP)of N2O and CH4was calculated in mass of CO2equivalents(kg CO2eq ha?1)over a 100-year time horizon.A radiative forcing potential relative to CO2of 298 was used for N2O and 25 for CH4[31].Greenhouse gas intensity(GHGI)was expressed as grain yield over GWP(grain yield/GWP,kg CO2eq kg?1grain).

2.5.Measurements of grain yield,WP,and NUE

Plants were hand-harvested on October 22 in both years.Measurement of grain yield and yield components followed Yoshida et al.[33].Plants in two rows on each side of the subplot were discarded to avoid border effects.Grain yield was determined from a harvest area of 6.0 m2(3 m×2 m)in each subplot and adjusted to 14%moisture.Aboveground biomass and yield components,i.e.the number of panicles per square meter,number of spikelets per panicle,percentage of filled kernels,and 1000-kernel weight,were determined from plants in an area of 0.48 m2(0.4 m×1.2 m,excluding the border ones) sampled randomly from each subplot. The proportion of filled kernels was defined as the filled kernels(with specific gravity ≥1.06 g cm?3)as a percentage of total number of spikelets.

Aboveground plants sampled at maturity were separated into straw,filled and unfilled kernels,and rachis.Dry weight of each component was determined following oven-drying at 70°C to constant weight.Tissue N content was determined by micro-Kjeldahl digestion, distillation, and titration to calculate aboveground N uptake[33].Internal N use efficiency(IEN), harvest index and WP were calculated using the following formulas:

2.6.Statistical analysis

Analysis of variance was performed using the SAS/STAT statistical analysis package(version 9.2,SAS Institute,Cary,NC,USA).SAS PROC general linear models(GLM)for split-plot design were used.Sources of variation included year,block(year),irrigation, year×irrigation, irrigation×block (year), cultivar,irrigation×cultivar,year×cultivar,and year×irrigation×cultivar.Data from each sampling date were analyzed separately.Means were tested by least significant difference at P ＜0.05(LSD0.05).Because there was no significant difference between the two study years for all determined parameters, data were averaged over the two years.Data for two cultivars were also averaged if determined parameters showed no significant difference between the two cultivars.

3.Results

3.1.Soil and leaf water potentials and amount of irrigation water

If there was no rain,it took 3–6 days to reach a soil water potential of ?15 kPa in the AWD regime and 2–3 days to reach a soil water potential of approximately ?5 kPa in the FI regime,depending on plant growth stage(Fig.S2-A and B).The CF regime received 19–21 applications of irrigation,whereas the AWD and FI regimes received respectively 6–7 and 11–13 applications,during the irrigation treatments.

Fig.1 shows midday leaf water potentials when soil water potentials in the bed center were approximately ?15 kPa in the AWD regime and ?5 kPa in the FI regime(D1 and D2)and when plants were rewatered(W1 and W2).At D1 and D2,leaf water potentials were ?0.82 to ?0.89 MPa in the AWD regime,and were significantly lower than those(?0.53 to ?0.59 MPa)in the CF regime or lower than those(?0.57 to ?0.64 MPa)in the FI regime,whereas they showed no significant difference between CF and FI regimes(Fig.1).At W1 and W2,there was no significant difference in leaf water potentials among the three irrigation regimes.

The total amount of irrigation water varied with irrigation regime(Fig.S3).It was 340 mm for the AWD and 312 mm for the FI regime,or respectively 75.1%and 68.9%of that for the CF regime(453 mm)on average.

Fig.1–Leaf water potential of direct-seeded rice grown on raised beds in various irrigation regimes.CF,AWD,and FI represent continuous flooding,alternate wetting and drying irrigation,and furrow irrigation,respectively,during the growing season.Measurements were made on the topmost fully expanded leaves at midday(11:30 a.m.)when soil water potential in the bed center was approximately ?15 kPa in the AWD regime or ?5 kPa in the FI regime(D1 and D2)and when plants were rewatered in the AWD regime or beds were wetted in the FI regime(W1 and W2).Vertical bars represent±standard error of the mean(n=16).Different letters under columns indicate statistical significance at the 0.05 probability level within the same measurement date.NS,non-significant at the 0.05 probability level.

3.2.Leaf photosynthetic rate(Pr),PNUE,soil Eh,and ROA

The Pr varied with irrigation regimes and measurement dates(Fig.2-A).Within comparison with that in CF,Pr in AWD was significantly reduced at D1 and showed no significant difference at D2,and in FI showed no significant difference at D1,and was significantly increased at D2.Both AWD and FI showed greater Pr than CF at W1 and W2.On average,both AWD and FI had greater Pr than CF(Fig.2-A).

Because there was no significant difference in specific leaf N content among the three irrigation regimes at the same measurement date(Fig.2-B),changes in PNUE were very similar to those in Pr,and as a result,both AWD and FI showed higher PNUE than CF,on average(Fig.2-C).

As shown in Fig.S4,soil Eh in CF was very low under the condition of wheat straw incorporation.Both AWD and FI significantly increased soil Eh relative to CF either at soil drying(D1 and D2)or at rewatering(W1 and W2),especially at D1 and D2(Fig.S4).Similar to soil Eh,root dry weight and ROA were significantly higher in AWD or in FI regimes than in CF regimes either at soil drying or at rewatering(Fig.3-A and B).When compared HD-5,YY-2640 had higher root dry weight and ROA at the same measurement date.The ROA was higher at W1 or W2 than D1 or D2 for the same cultivar,especially at W1(Fig.3-B),indicating a rewatering effect.

3.3.Tiller number,LAI,and shoot dry weight

The number of tillers in various irrigation regimes was shown in Table S1.At the early tillering stage(ET)and maturity(MA),the number of tillers including main stems was smaller in CF than in AWD or than in FI regimes.It showed no significant difference among the three irrigation regimes at the jointing stage(JS),and was significantly greater in CF than in AWD or than in FI regimes at heading time(HT).Both AWD and FI had a greater proportion of productive tillers than CF(Table S1).Differences in tiller number and proportion of productive tillers were not significant between AWD and FI regimes,and YY-2640 showed fewer tillers and main stems at each measurement time and a higher proportion of productive tillers than HD-5 in the same irrigation regime(Table S1).

Except for heading time,when LAI showed no significant difference among the three irrigation regimes within the same cultivar,LAI was significantly greater in AWD or in FI than in CF at all growth stages(Table S2).Both AWD and FI also showed greater LAD than CF from ET to MA(Table S2).Differences in LAI and LAD were not significant between AWD and FI for the same cultivar. However, YY-2640 showed greater LAI and LAD than HD-5 at all measurement times and in the same irrigation regime(Table S2).

Consistently with LAI,shoot dry weight at ET,panicle initiation(PI),and MA and CGR from ET to PI and from HT to MA were significantly greater in AWD or in FI than in CF(Table 1).Except that shoot dry weight at HT showed no significant difference between CF and AWD and CGR from PI to HT was significantly smaller in AWD than in CF,both shoot dry weight and CGR were highest in FI,intermediate in AWD,and lowest in CF.At the same growth stage or during the same growth period and in the same irrigation regime,YY-2640 showed greater shoot dry weight and CGR than HD-5(Table 1).

Similar to CGR,NAR was highest in FI,intermediate in AWD,and lowest in CF in each growth period(Fig.S5).

Fig.2–Leaf photosynthetic rate(A),specific leaf nitrogen content(B),and leaf photosynthetic nitrogen use efficiency(C)of direct-seeded rice grown on raised beds in various irrigation regimes.CF,AWD,and FI represent continuous flooding,alternate wetting and drying irrigation and furrow irrigation,respectively,during the growing season.D1 and D2 indicate when soil water potential in the bed center was approximately ?15 kPa in the AWD regime or ?5 kPa in the FI regime.W1 and W2 denote when plants were rewatered in the AWD regime or beds were wetted in the FI regime.The mean is an average of the data from D1,D2,W1,and W2.Vertical bars represent±standard error of the mean(n=16).Different letters above the column indicate statistical significance at the 0.05 probability level within the same measurement date,and NS means not significant at the 0.05 probability level.

3.4.Grain yield,NSC remobilization,NUE,and WP

Fig.3–Root dry weight(A)and root oxidation activity(B)of direct-seeded rice grown on raised beds in various irrigation regimes.CF,AWD,and FI represent continuous flooding,alternate wetting and drying irrigation,and furrow irrigation,respectively,during the growing season.YY and HD are two rice cultivars:Yongyou 2640 and Huaidao 5,respectively.D1 and D2 indicate when soil water potential in the bed center was approximately ?15 kPa in the AWD regime or ?5 kPa in the FI regime.W1 and W2 denote when plants were rewatered in the AWD regime or beds were wetted in the FI regime.Vertical bars represent±standard error of the mean(n=8).Different letters above the column indicate statistical significance at the 0.05 probability level on the same measurement date.

Grain yield was highest in FI,intermediate in AWD,and lowest in CF(Table 2).It was increased by 11.8%–16.1%in FI and 6.14%–9.66%in AWD relative to that in CF regimes.More panicles per m2, greater proportion of filled kernels, and higher kernel weight contributed to higher grain yield in FI and in AWD regimes.Although the number of spikelets per panicle for YY-2640 was decreased in AWD or in FI,the total number of spikelets per m2was increased relative to that in the CF regime,suggesting that an increase in panicle number per m2outweighed the decrease in spikelet number per panicle in these two irrigation regimes.A higher grain yield in FI than in AWD was attributed mainly to a greater percentage of filled kernels and a heavier kernel weight in FI.Relative to HD-5,YY-2640 produced a higher grain yield in the same irrigation regime, a result attributed to the greater number of spikelets per panicle in YY-2640 than in HD-5(Table 2).

Compared with CF,both AWD and FI showed a higher harvest index,in good agreement with the greater remobilization of NSC from stems to grains in these two irrigation regimes(Table 3).Generally,AWD or FI did not significantly decrease,or even significantly increased,total N uptake in plants,in comparison with CF(Table 3).For both cultivars,internal N use efficiency(IEN)was significantly higher in AWD or FI than in CF.It was increased by 5.76%–9.01%in AWD and by 7.67%–12.3%in FI(Table 3).Like IEN,WP was significantly increased in AWD (14.5%–18.6%) and in FI (22.6%–28.2%)compared with C.In the same irrigation regime,YY-2640 showed a higher N uptake and a higher WP than HD-5,whereas IENshowed no significant difference between the two cultivars(Table 3).

Regression analysis showed that most of the agronomic and physiological traits measured,including average tiller number,LAI,CGR,root dry weight,ROA,NSC remobilization,and harvest index,from each cultivar and each year were positively and significantly or very significantly correlated with grain yield and WP(Table S3),indicating that improved agronomic and physiological performance contributed to higher grain yield and WP in both AWD and FI regimes.Tiller number,proportion of productive tillers,shoot dry weight,leaf PNUE,ROA,NSC remobilization,and harvest index were also positively and significantly or very significantly correlated with IEN(Table S3), suggesting that improvement in these traits in the AWD and FI regimes led to an increase in IEN.

Table 1–Shoot dry weight and crop growth rate of direct-seeded rice grown in raised beds in various irrigation regimes.

3.5.GHG emissions and GHG intensity

Fluxes of CH4were highly dependent on water management(Fig.4-A and B).Methane emissions in CF regimes showed a very high peak at 39–40 DAS,which may be attributed to higher temperature(refer to Fig.S1-A and B)and vigorous rice growth during this period,and thereafter CH4fluxes remained at a low level.They were markedly decreased during the soildrying period in AWD regimes and during the entire growth period in FI regimes.The emission of CH4peaked at 34–35 DAS in AWD regimes,and the peak was much smaller than that in CF regimes.There were no obvious peaks of CH4emissions in FI regimes(Fig.4-A and B).

Nitrous oxide emissions were influenced mainly by both N fertilizer application and soil drying(Fig.4-C and D).Two large peaks of N2O emissions appeared at 5–6 DAS and 19–20 DAS,coinciding with N application at ?1 DAS and 14 DAS,respectively,and with unsaturation of raised beds during these periods.Nitrous oxide emissions in AWD and FI regimes also spiked after N application at early tillering(32 DAS)and at mid-season(52 DAS)and when soil water potentials were approximately ?15 kPa in the AWD and ?5 kPa in the FI regime(Fig.4-C and D).

During the growth period,the GWP of CH4and N2O under various irrigation regimes varied between 3.08 and 13.5 t CO2-eq ha?1(Table 4).Compared with that in CF regimes,GWP decreased by on average,62.3%in the AWD and by 74.9%in the FI regime. As shown in Table 4, both AWD and FI significantly increased N2O emission,and N2O emission was more increased in FI regimes.However,the GWP of N2Oaccounted for 1.0%, 5.6%, and 11.7% of the total GWP,respectively,in CF,AWD,and FI regimes,indicating that CH4emissions were the dominant part in the total GWP from rice fields in the various irrigation regimes.Both AWD and FI significantly decreased GHGI by 65.1%and 77.7%,respectively,in comparison with CF(Table 4).In the same irrigation regime,GWP and GHGI were significantly lower for YY-2640 than for HD-5(Table 4).

Table 2–Grain yield and yield components of direct-seeded rice grown on raised beds in various irrigation regimes.

Table 3–Nonstructural carbohydrate(NSC)remobilization,harvest index,nitrogen(N)uptake,internal N use efficiency(IEN),and water productivity(WP)of direct-seeded rice grown in raised beds in various irrigation regimes.

Fig.4–Fluxes of CH4(A and B)and N2O(C and D)on raised beds in various irrigation regimes during the rice-growing season.YY-2640 and HD-5 are two rice cultivars of Yongyou 2640 and Huaidao 5,respectively.CF,AWD,and FI represent continuous flooding,alternate wetting and drying irrigation,and furrow irrigation,respectively,during the growing season.Arrows with dash lines indicate the start of irrigation treatments,and arrows with solid lines show the days of nitrogen application.Vertical bars represent±standard error of the mean(n=8).

4.Discussion

The present results across the two years demonstrated that,in comparison with CF,FI could not only save irrigation water,but also increase grain yield in both japonica inbred and indica/japonica hybrid cultivars(Fig.2,Table 2).The discrepancies between our results and those from previous reports showing decreases in grain yield under FI may be attributed to many factors, including differences in climate conditions, soil textures,and genotypes[14,19,21].Two reasons,however,may contribute greatly to the higher grain yield in the FI regime in this study.First,FI in previous experiments was applied mostly according to the water depth or water disappearance in furrows[14,19,21].Such water management could not exactly control the soil drying degree in the beds. In contrast, FI in the present study was precisely imposed by monitoring the soil water potentials in the center of beds;that is,irrigation was applied by running water down the furrows until the beds had wetted to the center(soil water potential=0),and rewatered when soil water potential in the centers of beds reached approximately ?5 kPa.Under such conditions,midday leaf water potentials in FI regimes were not significantly lower than those in CF regimes(Fig.1).Furthermore, leaf photosynthetic rate was significantly greater in FI than in CF,especially during rewatering(Fig.2),indicating that the FI regime in this study promotes plant growth.Second,crop residues from the preceding season were removed from the field in previous experiments,whereas in the present study wheat straw was incorporated into the soil.A long period of flooding in the field can produce high concentrations of toxic reduction products such as Fe2+,H2S,and organic compounds,especially when wheat straw has been incorporated into soil,and could severely inhibit root growth[11,32,37,38].In the present study,soil Eh and ROA were much lower in CF regimes than in FI regimes(Figs.3 and S4),indicating a strong soil reduction and weak root growth in CF regimes following wheat straw incorporation although rice seeds were dry direct-seeded and rice was grown on raised beds.On the other hand,the FI regime significantly increased soil redox potential,root biomass and ROA,which could also promote shoot growth,as evidenced by increases in tiller number at early growth stages and larger LAD and CGR during the growing season(Table 1,Tables S1 and S2).These traits were positively and significantly or very significantly correlated with grain yield and WP(Table S3),suggesting that the FI regime could improve agronomic and physiological performance,leading to higher grain yield and WP.

Prior to this study,little information was available describing the effect of AWD on grain yield and WP in direct-seeded rice grown on raised beds, although that of AWD on transplanted rice has been well documented[10,17,24,26,36–40].Our results show that,like FI,AWD can produce both higher grain yield and higher WP than CF(Tables 2 and 3).Improved soil redox conditions,larger root biomass,stronger ROA,more tillers at the early tillering stage,longer LAD,greater CGR and NAR,more remobilization of NSC from stems to kernels,and higher harvest index contributed to the higher grain yield in AWD regimes(Tables 2 and 3,Figs.2 and 3,Tables S1 and S2,Fig.S5).It has been proposed[10,17]that soil-drying conditions in AWD are an important factor in grain yield and that a moderate wetting and drying regime could not only save water,but also increase grain yield.In the present study, the threshold of soil water potential for rewatering in the AWD regime was ?15 kPa at 15–20 cm depth in the centers of beds and the midday leaf potential was approximately ?0.82 to ?0.89 MPa(Fig.1).We suggest that these values could be used as indices for AWD to achieve higher yield and higher WP in direct-seeded rice grown in raised beds.

Table 4–Seasonal CH4 and N2O emissions,global warming potential(GWP),and greenhouse gas intensity(GHGI)from raised beds in direct-seeded rice in various irrigation regimes.

In the comparison of AWD with FI,FI resulted in higher grain yield in both cultivars(Table 2).The ecological and physiological mechanisms are unclear.A possible explanation is that soil drying was more severe(soil water potential?15 kPa) in AWD regimes than in FI regimes (soil water potential approximately ?5 kPa) for the threshold of rewatering.Under such soil water potentials,leaf photosynthesis showed some inhibition in AWD,whereas it showed no reduction or even an increase in FI regimes(Fig.2).The results suggest that there is still room to improve the AWD technique to achieve higher grain yield and WP in direct-seeded rice grown on raised beds in the wheat–rice rotation system.

Although there are reports[6,11,41]that DDSR,AWD,and FI could decrease CH4emissions,little information is available on the comparison of the effect of various irrigation regimes on GHG emissions in direct-seeded rice grown on raised beds and when wheat residues were incorporated into rice fields.The present results show that both AWD and FI substantially decreased seasonal CH4emissions and thereby reduced GWP and GHGI in comparison with CF(Table 4 and Fig.4).An interpretation is that the AWD and FI regimes could greatly improve soil redox conditions (Fig. S4), preventing CH4formation by inhibiting methanogenic bacteria and thus reducing CH4emissions[11,32].The reduced CH4emissions in the AWD and FI regimes may be due partly to greater abundance of such oxidants as Fe(III),given that oxidized forms of iron directly and indirectly inhibit methanogenesis[42,43].The finding that root dry weight and ROA were negatively and significantly correlated with CH4emissions(Table S4),suggest that the reduction in CH4emissions in the AWD and FI regimes is closely associated with these two improved plant traits. More studies are needed for full understanding of the mechanism by which AWD and FI regimes reduce CH4emissions.

It has been proposed[44–46]that the adoption of AWDbased technologies could reduce total cumulative plant N and NUE by stimulating N losses through increases in nitrification and denitrification.Both AWD and FI significantly increased N2O emissions in comparison with CF(Table 4,Fig.4).On average,0.47 kg and 0.87 kg N2O-N ha?1more were lost in AWD and FI regimes,respectively,than in CF regimes(Table 4).But these amounts account for only 0.20%and 0.36%of the total N application rate of 240 kg ha?1,suggesting that N loss due to N2O emissions is negligible in both AWD and FI regimes.Furthermore,the total N uptake was not significantly different between CF and AWD,whereas it was significantly greater in FI than in CF(Table 3).Thus,AWD did not decrease,but FI increased,total plant N absorption in direct-seeded rice grown in raised beds.This finding may be attributed to a greater absorption of N from the soil,as evidenced by larger root biomass and higher ROA in these two irrigation regimes(Fig.3).

It should be noted that gas flux measurements in the present study were conducted at 2-day intervals for 10 days after each fertilizer application and at 5-day intervals during the other growth periods.Although such measurements are common practice, they may miss some spikes of N2O emissions,given that N2O spikes last for only hours after fertilization and/or during soil drying-wetting alternation[37,41,46,47].The same is the case for CH4release from the soil after drainage or drying [37, 41, 46, 47]. More accurate measurement of GHG emissions is needed for a clear picture of C and N dynamics in AWD and FI regimes.

We have described only fluxes of CH4and N2O,although CO2fluxes were measured. There are three reasons for omitting CO2:(i)CO2emissions are estimated to contribute＜1%to the GWP of agriculture[41];(ii)the net balance between C respiration and fixation in a cropping system is difficult to detect in short-term experiments[41,47,48];and(iii)the rice–wheat rotation system at the present experiment site has been conducted for ＞30 years.Under such a system,AWD or FI in rice season may not alter soil organic C,as soil C stocks are already degraded[31,41,47].

Interestingly,both AWD and FI showed higher IENthan CF(Table 3).The mechanism underlying a higher IENin the two irrigation regimes is unexplained. Several interpretations,however, invite attention. First, both AWD and FI could increase the proportion of productive tillers(Table S1),which spend less N on redundant vegetative growth[40].Second,a higher leaf PNUE was observed in both AWD and FI regimes(Fig.2),which may contribute to a higher IENbecause a higher PNUE corresponds to lower investment of N in nonphotosynthetic components and higher investment in dry matter production[49–51].Third,both AWD and FI showed greater remobilization of NSC from stems to grains and a higher harvest index(Table 3)than CF.This difference could indicate a lower use of N to produce vegetative tissues and higher use of N to produce starch and protein in kernels[39,40],leading to a higher IEN.Further research is needed to elucidate the mechanism by which AWD and FI yield a higher IEN.

It should be noted that the indica/japonica hybrid YY-2640 produced higher grain yield,greater WP,and lower GWP and GHGI than the japonica inbred HD-5 in the same irrigation regime(Tables 2–4).The higher grain yield of YY-2640 was due mainly to its larger panicle with more spikelets,while a stronger ROA(Fig.3)may contribute to the reduction in CH4emissions,leading to a smaller GWP[11,41].Both of these differences could result in a higher WP and a smaller GHGI.However,the number of spikelets per panicle for YY-2640 was significantly decreased in both AWD and FI relative to that in CF,whereas it showed no significant difference for HD-5 among the three irrigation regimes(Table 2),suggesting that YY-2640 is more sensitive to soil drying than HD-5 during panicle development.Further study may reveal the highyielding mechanism of YY-2640 and the cultivar difference in response to irrigation regimes.

During the rice–wheat annual rotation cycle,the net CH4flux is significant during the rice-growing season but negligible during the wheat-growing season;in contrast,most of the N2O is emitted during the wheat-growing season[31].In the present study,we did not measure GHG emissions during the wheat-growing season. However, we observed that, on average,wheat grain yield was increased by 3.2%and 2.8%,respectively,in AWD(7.21 t ha?1)and FI(7.18 t ha?1)plots in comparison with that in CF (6.98 t ha?1) plots, although irrigation treatments were imposed only in the rice-growing season and all crop management practices were the same in the wheat-growing season.This result indicates that the AWD and FI imposed during the rice season can also increase wheat productivity.Further study may reveal the effects of water management during the rice-growing season on GHG emissions during the fallow period and wheat-growing season and the mechanism underlying the effects on CH4and N2O fluxes and wheat productivity.

5.Conclusion

In the system of dry direct-seeded rice grown in raised beds and when wheat straw was incorporated into the soil,both AWD and FI significantly increased grain yield,WP,and IENand reduced CH4emissions in comparison with CF.Improved soil redox conditions,an increase in tiller number at the early tillering stage,greater root and shoot biomass,ROA,LAD,CGR,NAR,and PNUE,more remobilization of NSC from stems to kernels and a higher harvest index contributed to higher grain yield and resource use efficiency and reduced environmental risk.An indica/japonica hybrid showed a higher grain yield and smaller GWP than a japonica inbred,owing mainly to the former's larger panicle with more spikelets, greater root biomass,and higher ROA than those of the latter.Soil water potentials of approximately ?15 kPa in AWD and at ?5 kPa in FI could be used for the threshold for rewatering. The mechanism underlying higher IENin both AWD and FI regimes and the cultivar difference in response to irrigation regimes merits further investigation.

Acknowledgments

We are grateful for grants from the National Key Research and Development Program of China (2016YFD0300206-4), the National Natural Science Foundation of China(31461143015,31471438), the National Key Technology R&D Program of China(2014AA10A605),the Priority Academic Program Development of Jiangsu Higher Education Institutions(PAPD-2015-01), the Top Talent Supporting Program of Yangzhou University (2015-01), and the Hong Kong Research Grant Council(14122415,14160516,14177617,AoE/M-05/12,AoE/M-403/16).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2018.05.004.