Yield sustainability of winter wheat under three limited-irrigation schemes based on a 28-year field experiment

Ynmei Go,Meng Zhng,Zhimin Wng,Yinghu Zhng,*

a College of Agronomy and Biotechnology,China Agricultural University,Beijing 100193,China

b School of Life Science,Shanxi Normal University,Taiyuan 030031,Shanxi,China

c Ministerial and Provincial Co-Innovation Centre for Endemic Crops Production with High-quality and Efficiency in Loess Plateau,Shanxi Agricultural University,Taigu 030801,Shanxi,China

Keywords:Yield Sustainability Cultivars Climate change Phenology

ABSTRACT Sustainable intensification is an agricultural develo pment direction internationally.However,little is known about the yield sustainability of winter wheat(Triticum aestivum L.)under limited irrigation schemes on the North China Plain(NCP).A 28-year field experiment from 1991 to 2018 at Wuqiao Experimental Station was used to characterize long-term yield,evapotranspiration(ET),and water use efficiency(WUE)trends under three irrigation treatments(W1,irrigation just before sowing;W2,irrigation before sowing and at jointing stage;W3,irrigation before sowing,at jointing stage,and at anthesis).Yield gaps and the effects of genetic improvement,climate change,and climate variables on wheat yield and key phenological stages were estimated using the Agricultural Production Systems Simulator(APSIM)model.Grain yield and WUE of winter wheat increased during the 28 years under the three irrigation treatments,and the upward trend of WUE followed a saturation curve pattern.ET increased slightly.Simulation results showed that genetic improvement dramatically prolonged the phenological stages of vegetative growth period and contributed to yield increase by 0.03%-15.6%.The rapid increase in yield with lower water use was associated mainly with an increase in biomass with genetic improvement and partly with an increase in harvest index.A curvilinear relationship between WUE and yield emphasized the importance of obtaining high yields for high WUE.The yield gaps between potential yield and yield under W1 treatment increased from 1991 to 2018 but were relatively constant for the W2 and W3 treatments.Elevated atmospheric CO2 concentration offset the negative effects of temperature increase on yield,leading to minor(-2.3% to 0.3%)changes in yield under climate change.Thus,genetic improvement played a dominant role in yield increase,and limited-irrigation schemes(W2 and W3)can increase wheat yield and promote sustainability of crop production on the NCP.

1.Introduction

The North China Plain(NCP)is one of the largest wheat production regions in China and contributes 50% of the nation’s wheat grain[1].A winter wheat and summer maize(Zea mays L.)rotation is the main cropping system in this region.With an uneven distribution of annual precipitation,only 25%-40%(100-180 mm)of rainfall occurs during the winter wheat season,falling short of the water requirements of wheat.To maintain high crop yield,irrigation has become necessary to replenish soil water during the wheat season.Conventional irrigation for winter wheat consists of 4-5 irrigation events:before sowing,wintering stage,greening stage,jointing stage,and flowering stage,with total irrigation exceeding 250 mm.With irrigation consuming some 70%of global water resources[2,3],water supplies adequate to meet agricultural demands may not be available in many countries,in particular China[4].The irrigation program(W3)currently promoted[5]consists of three events:before sowing,jointing stage,and anthesis,and allows high yield and water use efficiency(WUE).But to further reduce irrigation amounts,one(W1,before sowing)or two(W2,before sowing and at jointing stage)irrigation events in the wheat growing season are irrigation trends for the future.Rainfed agriculture increases groundwater conservation,despite aggravating soil dryness and severely reducing wheat production[6].

With the explosive growth of global population,meeting the increasing demand for food is urgent[7].In response to this pressure,there is increasing focus on‘‘sustainable intensification”[8,9].This strategy for meeting the demand for food security is‘‘expected to rely on increases of crop production rather than on enlargement of crop production areas”[10].In the NCP,with the aim of increasing crop yields and address water scarcity,the three limitedirrigation schemes(W1,W2,and W3)have been proposed and applied for many years and at many sites.But their yield sustainability has not been evaluated and their relationship with crop water use is unclear.The relative contributions of each factor to crop yield and yield gap were estimated in previous studies[11-13]under sufficient water supply or rainfed conditions at a regional scale,at two extremes of irrigation.These conditions do not reflect actual production conditions or water scarcity in the NCP.Estimating the size of the current yield gap and how this gap has changed over the past few decades is essential for evaluating the yield sustainability of these three irrigation schemes.

During the last several decades,wheat yield in the NCP has been increased by high-yielding cultivar selection,improved crop management practices,and climatic changes[14-16].Most studies[12,13,17,18]have shown that yield has followed a declining trend under climate change,while changes in management practices(fertilization and cultivar use)have mitigated weather effects and dominated yield improvement in the past decades.However,a recent study[19]shows that climate and agronomy,not genetics,accounted for maize yield gains in favorable environments during 2005-2018 in the U.S.state of Nebraska.Xiao and Tao[13],using the Agricultural Production Systems Simulator(APSIM)model[20],reported that climatic change increased winter wheat yields insignificantly by-3.0% to 3.0% during 1980-2009 on the NCP.Shi et al.[7]and Yu et al.[21]drew similar conclusions using the DeNitrification-DeComposition(DNDC)[22]and Agro-C models[23].

Crop phenological stage has been one of the most reported and consistent biological characteristics in response to climate change[24].Crop phenology is co-determined by climate change and agricultural management.Agronomic management practices such as cultivar changes,fertilization,and irrigation over the past decades has prolonged the growth period of winter wheat[25,26].The effects of sowing date change on wheat growth periods have been small because wheat sowing date has been delayed or advanced only slightly by fewer than five days per decade[26].However,the influence of climate change on crop phenology has been inconsistent,and some studies have shown that climate change prolonged[27,28]or had insignificant effects on[29-31]the crop growth period.In contrast,most research has shown that excessive increase of temperature has resulted in acceleration of crop growth and shortening of growing seasons,irrespective of wheat type and time period[32,33].With respect to the responses of crop key phenological stages and yield to other climatic variables,such as precipitation,solar radiation,and atmospheric CO2concentration under different irrigation treatments,there is little information.Quantifying the contribution of cultivar change,climate change,and climatic variables on crop yield and phenology under different irrigation treatments would assist in addressing climate change and facilitate sustainable agriculture.

Compared with short-term trials,long-term experiments have advantages for studying crop yield trend over a long time and estimating agricultural sustainability[34].They represent the basic method for investigating the long-term effects of genetic improvement,crop management practices,climatic changes,and other factors on crop production[14].In this study,the objectives of a 28-year field experiment were to(1)characterize wheat annual yields,evapotranspiration(ET),and WUE trends and the correlation between crop yield and crop water use;(2)estimate the yield gaps by comparing potential and actual yields,and(3)use the APSIM model to jointly and separately evaluate the response of long-term wheat yield and the lengths of the key growth stages to cultivar and climate variables.

2.Material and methods

2.1.Experimental site and soil characteristics

A long-term field experiment was conducted from 1991 to 2018 at Wuqiao Experimental Station of China Agricultural University,Hebei province,in the NCP,China.The monsoon climate dominates this region,which has a mean annual precipitation of 515 mm,with about 78%falling during the summer maize season.Loam soil with a deep profile is the main soil type(Table S1).The soil water reservoir in a 0-200 cm soil layer at the site is 687 mm.The wilting point is 7.5%-9.0%(by weight)and the pH is approximately 8.0.The 0-20 cm topsoil contains 12.5 g kg-1organic matter,1.02 g kg-1total nitrogen(N),64.9 mg kg-1alkali-hydrolysable nitrogen,41.6 mg kg-1available phosphorus(P),and 112.3 mg kg-1available potassium(K).

2.2.Weather and crop management

Historical daily weather data from Wuqiao station,including temperatures(maximum,minimum,and average),sunshine hours,and precipitation from 1990 to 2018,were obtained from the National Meteorological Networks of the China Meteorological Administration.The sunshine duration was converted into daily solar radiation using the?ngstr?m formula[35].Atmospheric CO2concentration was obtained from the Mauna Loa Observatory,HI,USA(https://gml.noaa.gov/ccgg/trends/data.html).

At the experimental station,winter wheat was grown from early to mid-October to June under irrigated conditions.Summer maize was planted into the winter wheat field without tillage after wheat harvesting at a plant density of 55,000-75,000 plants ha-1.From 1990 to 2013,irrigation for maize was applied at sowing,V9,and silking or an earlier grain-filling stage depending on the seasonal rainfall situation.Basal dressing was applied at approximately 72 kg ha-1N,105 kg ha-1P2O5,and 120 kg ha-1K2O.An additional 108 kg ha-1N was applied with the irrigation at V9 for maize.From 2014 to 2018,approximately 75 mm irrigation was applied after maize planting to ensure germination if there was no rain during this period,and there was no irrigation during the maize growing season.All fertilizers were applied at sowing.The summer maize harvest occurred usually in late September,with yields ranging from 5.0 to 11.1 Mg ha-1during the 1990-2018 growing seasons.Afterwards,100 mm of irrigation water was supplied to ensure wheat emergence.Then the land was usually dried for 7-10 days to ensure that winter wheat could be sown,and the soil water content at sowing was sampled.The changes in wheat tillage practices,total chemical fertilizers,and cultivars from 1991 to 2018 at Wuqiao Experimental Station are listed in Table S2.The cultivars were all common cultivars widely planted in this region.Only one cultivar was grown per year and was identical for the three irrigation treatments.The amount of fertilizer applied remained unchanged from 1990 to 2018,and all fertilizer was applied at sowing at 165 kg ha-1N,138 kg ha-1P2O5,and 90 kg ha-1K2O.The tillage practices followed those of local farmers and changed with advances in tillage techniques,which can contribute to yield increase but were not accounted for in this study.The sowing date of winter wheat ranged from October 10 to 21 according to the harvest date of summer maize and the precipitation distribution in October.The seeding density was increased with the delay of sowing date to ensure the same population size(total stem number)before wintering stage.This practice was adopted for comparing yields at maturity under differing irrigation regimes.

In the 28-year experiment,winter wheat was sown in plots in a randomized replicated trial with three irrigation treatments.Each treatment was repeated three times.The treatments consisted of W1(100 mm irrigation before sowing),W2(100 mm irrigation before sowing and an additional 75 mm irrigation at jointing stage)and W3(100 mm irrigation before sowing and additional 75 mm irrigations at jointing and anthesis).A flow meter was installed in the tube(100 mm diameter)to record the irrigation amount in each plot.The irrigation regime,including the irrigation amount and irrigation time,remained unchanged from 1990 to 2018.The row spacing was 15 cm.The plot size was 60 m2(6 m×10 m).

2.3.Measurements and methods

Phenological data(sowing,anthesis,and maturity dates),aboveground dry matter(biomass),yield,ET,WUE,and crop management practices from 1990 to 2018 for wheat cultivation were obtained from the field research dataset.The key phenological stages,including sowing,wintering stage,jointing stage,booting stage,anthesis,and maturity,were recorded throughout the cycle using the Zadoks scale[36].At maturity,for each plot,grain yield was determined using plant samples from an approximately 3-m2area using standard protocols[37].In addition,plants from two adjacent inner rows of 0.5 m were manually cut at ground level.They were partitioned into different organs,and their weight was recorded after oven-drying for 48 h at 75 °C for aboveground biomass determination[38].Harvest index was calculated as the proportion of grain weight to biomass for two 50-cm inner row samples at maturity.

At sowing and maturity stage,soil samples were collected with a soil corer at 20-cm intervals to a depth of 2 m for each treatment,with three replicates.Soil samples were dried to constant weight in a forced-draft oven at 105 °C.Crop seasonal ET was calculated using the soil water balance equation[39].Capillary rise and water drainage were negligible,and no surface runoff occurred in the experimental plots.

where ET is crop evapotranspiration(mm),SWD is soil water depletion(mm),P is precipitation(mm),and I is irrigation amount(mm).Water-use efficiency(WUE,kg m-3)was calculated as grain yield divided by ET.

The daily reference evapotranspiration(ET0)represents the evaporative power of the atmosphere and does not consider other factors,such as soil,crop,and management practices.ET0was calculated with ET0calculator,which is software developed by the Land and Water Division of FAO(https://www.fao.org/land-water/databases-and-software/eto-calculator/en/).The ET0calculator estimates ET0from meteorological data using the FAO Penman-Monteith equation[40].

2.4.Crop model and simulation

The APSIM model version 7.9 r4044(https://www.apsim.info)was used to simulate wheat phenology,yield,biomass,ET,and WUE values.In APSIM,irrigation amount and irrigation date under W1,W2,and W3 treatments were set in the‘‘Irrigate on date”module in the management toolbox using the actual irrigation dates of jointing and flowering stages in the field.The APSIM model was calibrated and validated based on the field-measured wheat cultivation data from 1990 to 2018.The wheat phenology(flowering and maturity dates),yield,and biomass under W1 and W3 irrigation treatments in 1991-1993,1996,1998,2000,2002,2004-2007,2011,2012,2014,2015,and 2017 were used for the model calibration.The crop data collected in the remaining years were used for model validation.Nine wheat cultivars were used to calibrate the APSIM model according to actual field planting conditions from 1990 to 2018.These nine cultivars were‘‘planted”in sequence according to their cultivation years(Table S2).The difference between the simulated yield and limited irrigation yields in the same year was caused only by differing irrigation amounts.The nine wheat cultivars parameters required by APSIM include vernalization sensitivity,photoperiod sensitivity,kernel number per stem weight at the beginning of grain filling,maximum grain size,thermal time to end of juvenile stage,thermal time to floral initiation,thermal time to flowering,thermal time to start grain filling,and thermal time to the end of grain filling(Table S3).The detailed soil measurements in our study included soil water content(SWC)at sowing,bulk density(BD),saturated volumetric water content(SAT),drained upper limit(DUL),15-bar lower limit(LL15),and so on(Table S1).The comparison indexes included simulated and observed anthesis/maturity dates,biomass,yield,ET,and WUE.

2.5.Effects of cultivar change and climate variables on wheat yield

In Table 1,four scenarios were simulated and the effects of climate change and cultivar change on wheat yield were estimated following Lyu et al.[12].The four scenarios included(1)combined effects of climate change and cultivar change;(2)effects of cultivar change;(3)effects of climate change;and(4)a long-term baseline(LTB)yield that was used to illustrate various effects on the yields under conditions of detrended climatic factors,no new cultivars,and consistent crop practices.To disentangle the effects of climate variables,including temperature,precipitation,solar radiation,and CO2concentration on wheat yield,another four scenarios were simulated(Table 2).Detrended climatic variables were calculated by the following formula following Lobell et al.[41]and Ray et al.[42],and the reference year was 1991.

where Tdiis the detrended temperature for the ith year,Tiis the observed temperature for the ith year,Tpiis predicted temperature for the ith year,based on a linear fit for 1991-2018,with i ranging from 1991 to 2018.The detrended precipitation and radiation were defined similarly.

Table 1Four scenarios used to estimate the effects of climate change and cultivar change on yield.

Table 2Four scenarios that were used to isolate the effects of climate variables on wheat yield.

The relative yield changes under the seven scenarios were calculated as:

where yieldiis the simulated yield in the ith year under the three scenarios and yieldLTBiis the simulated LTB yield for the ith year.

2.6.Simulation of potential yield and yield gap

We define potential yield(Yp)as the yield in a given wheat cultivar grown in a favorable environment with no limit on irrigation.Rainfed potential yield(Yrainfed)is the yield of a given wheat cultivar grown with no additional irrigation.The limited-irrigation yields(YW1,YW2,YW3)are the actual field yields under the W1,W2,and W3 irrigation treatments.The nine wheat cultivars were used to simulate the potential yield and rainfed potential yield according to actual field planting conditions from 1990 to 2018 at Wuqiao Experimental Station(Table S2).Yield gaps(YG)were defined as follows:YG1,the difference between potential yield(Yp)and actual field yield under the W1 irrigation treatment(YW1)(Eq.(5));YG2,the difference between potential yield(Yp)and the actual field yield under the W2 irrigation treatment(YW2)(Eq.(6));YG3,the difference between potential yield(Yp)and actual field yield under the W3 irrigation treatment(YW3)(Eq.(7)).These yield gaps are written mathematically as:

2.7.Statistical analysis

The performance of the APSIM model in simulating anthesis,maturity,biomass,yield,ET,and WUE in wheat was evaluated using the correlation coefficient(R2),root mean square error(RMSE),normalized root mean square error(NRMSE),and Willmott’s index of agreement(D value)by comparing the observed and simulated values.The simulation capacity is very good if NRMSE＜10%,good if 10%＜NRMSE＜20%,and nonpredictive if 20%＜NRMSE＜30%.The D value can well reflect the quality of the model,especially when the regression fitting lines are close to the 1:1 line.The higher D value,the lower bias of the model between the observed and simulated variables[43].These statistical indicators were computed from observed and simulated variables.All figures were created using Origin Pro 2019(Origin Lab Corporation,Northampton,MA,USA).Analyses of variance(ANOVA)was fitted using the general linear model procedure in SPSS version 20.0(IBM,Armonk,NY,USA).Significant differences were identified using ANOVA and least significant difference(LSD)tests at 95% or 99% confidence levels.

3.Results

3.1.Performance of APSIM

The APSIM-Wheat model was calibrated using field-observation data from 1991 to 2018.APSIM provided very good estimates of days to anthesis and maturity(Fig.S1A,B).In these experiments,the simulated mean days to anthesis and maturity were 203 and 237 days,and the observed mean values were 205 and 237 days.For the simulated days to anthesis and maturity,R2values were 0.84 and 0.83,D values were 0.93 and 0.94,and NRMSE values were 2%and 1.2%,respectively,indicating that the model can simulate the growth period very well(Fig.S1A,B).The simulated grain yield and biomass at maturity also agreed well with the observed values(Fig.S1C,D).The NRMSE values were 9% for grain yield and 12%for biomass,and the D values were 0.90 and 0.86,respectively,indicating that the model can replicate the growth of crops well.The simulated ET and WUE also agreed well with the observed values(Fig.S1E,F).The NRMSE values were 8% for ET and 11% for WUE.The D values were 0.87 for ET and 0.65 for WUE,indicating that the model estimates the soil water reasonably well.

3.2.Temporal climate trends during the wheat growing season

Over the 28 years,the weather during the winter wheat growing season changed somewhat under the conditions of climatic warming.The mean seasonal daily minimum and average temperature increased by 0.4°C(P＜0.01)and 0.3°C(P＜0.05)per decade,respectively,from 1991 to 2018,whereas no change was found in the daily maximum temperature(P＞0.05;Fig.S2A).The increase in the mean temperature was caused mainly by the increase in the minimum temperature.The total precipitation in the wheat growing season increased by 16.8 mm per decade from 1991 to 2018(P＞0.05;Fig.S2B).Although this increase was not significant,large annual variations in precipitation were found during the 28 years in the wheat growing season,with a maximum value of 254.6 mm and a minimum value of 37.5 mm.The mean precipitation over the 28 years was 125.1 mm(Fig.S2B).Solar radiation in the wheat growing season increased 94.8 MJ m-2per decade from 1991 to 2018(P＜0.05;Fig.S2C).Atmospheric CO2concentration increased from 354.5×10-6in 1990 to 408.7×10-6in 2018(P＜0.001;Fig.S2D).ET0was relatively constant over the past 28 year(P＞0.05;Fig.S2E).The mean seasonal ET0for winter wheat was 576 mm.

3.3.Observed yield,ET,and WUE

Owing to inter-annual variation of climate,there was great variation in wheat yield among seasons.The observed grain yield of winter wheat showed clear(P＜0.001)increasing trends from 1991 to 2018 under the W1,W2,and W3 irrigation treatments(Fig.1A).The annual yield increases for winter wheat were respectively 49,68,and 79 kg ha-1under W1,W2,and W3 conditions.The yield-change trend could be divided into two phases:a lower and marginally fluctuating phase from 1991 to 1999 and a higher and highly variable phase from 2000 to 2018(Fig.1A).The yield increase for winter wheat occurred mainly from 1999 to 2004,when the cultivars were changed quickly(76 Xuanxi to Laizhou 95021 to Lumai 21 to Shijiazhuang 8),as shown in Table S2.The mean grain yields from 1991 to 2018 were respectively 6.05,7.09,and 7.69 Mg ha-1under W1,W2,and W3 treatments.The linear regression in Fig.S3A shows a positive relationship of grain yield with biomass of winter wheat.The yield increase was also affected by the harvest index.A positive relationship between grain yield and harvest index was also found(Fig.S3B).

The ET was relatively constant over the 28 years,although it increased slightly(P＞0.05;Fig.1B).The observed WUE from 1991 to 2018 under the three irrigation treatments also showed upward trends,and they followed a saturation curve pattern(Fig.1C).The increase in WUE occurred mostly during 1991 to 2004,reaching a plateau.The change in WUE of winter wheat could be divided into two phases:a lower,rapidly increasing,and highly variable phase from 1991 to 2004 and a higher and marginally variable phase from 2004 to 2018(except for 1999,2002,and 2007 to 2010 under W1 treatment)(Fig.1C).The mean WUE value measured from 1991 to 2018 was 1.68 kg m-3under the W1 condition and 1.76 kg m-3under the W2 and W3 conditions.The linear regression in Fig.S3C shows no linear clear relationship between WUE and ET(P＜0.05).However,the saturation curve regression in Fig.S3D shows a clear relationship(P＜0.001)between yield and WUE.WUE and yield both increased before plateau,and at the plateau,yield was highly variable and WUE was marginally variable(Fig.S3D).

3.4.Yield gap

The simulated mean potential yield(Yp)of winter wheat showed marked(P＜0.001)increasing trends of 0.09 Mg ha-1per year from 1991 to 2018(Fig.2A).The yield increases occurred mainly from 2000 to 2016 with lower fluctuation.The simulated rainfed potential yield was relatively constant over the past 28 years without large increases in precipitation(Figs.2A,S2A).The long-term mean yield gaps between the potential yield(Yp)and the yield under W1,W2,and W3 treatments over the period 1991 to 2018 were respectively 1.89,0.86,and 0.25 Mg ha-1.YG2and YG3from 1991 to 2018 statistically neither increased nor decreased,but YG1increased from 1991 to 2018(Fig.2B).

3.5.Effects of cultivar change,climate change,and climate variables on yield

The simulated yields under W1,W2,W3,and Yp(no-limit irrigation)treatments all increased significantly under the combined-effect scenario from 1991 to 2018(Fig.3B-E).The cultivar-change scenario significantly increased the yields under the W3 and Yptreatments(Fig.3D,E).A relatively constant trend in the yields under different irrigation treatments occurred in the LTB and climate change scenarios(Fig.3A-E).

To quantify the effects of cultivar change,climate change,and climate variables on wheat yields,relative changes in yields under the three irrigation treatments were calculated based on the scenario simulations in Table 1 and Table 2.Compared with the yield under the LTB scenario,the climate-change scenario insignificantly increased the yields under the three irrigation treatments,leading to positive relative yield changes of respectively 5.5%,2.9%,1.7%,1.1%,and 0.9% under rainfed,W1,W2,W3,and Yptreatments(Figs.3,4A).The effect of climate change on yield was due mainly to increasing temperature and elevated atmospheric CO2concentration(Fig.4B).Increasing temperature increased wheat yield by 2.6%and 0.2%under rainfed and W1 treatments,but reduced yields by 1.3%,1.9%,and 2.2%under W2,W3,and Yptreatments,although the effects were not significant.Elevated atmospheric CO2concentration increased yield by 2.5%-3.1% under different treatments,but the effect was not significant(Fig.4B).Owing to annual climate change,yield variation varied from-11.1%to 27.8%,and the amplitude of variation increased gradually from 1991 to 2018 under the three irrigation treatments.The amplitude of variation gradually decreased with the increase of irrigation time(Fig.S4A-E).

The increase in the yields under W1,W2,W3,and Yptreatments in the cultivar-change scenario was higher than that in the LTB scenario(Fig.3B-E).The relative yield changes as a result of cultivar change were respectively 3.6%,9.1%,14.7%,and 15.6% under W1,W2,W3,and Ypconditions(Fig.4A).The cultivar change showed no effect on yield increase under rainfed treatment(0.03%).The combined effects of climate change and cultivars change jointly increased yields by respectively 5.2%,4.4%,12.4%,16.6%,and 17.3% under rainfed,W1,W2,W3,and Ypconditions(Fig.4A).

3.6.Effects of cultivar change and climate variables on key growth stages

Fig.1.Temporal trends in observed grain yield(A),evapotranspiration(ET;B)and Water-use efficiency(WUE;C)under W1,W2,and W3 treatments during winter wheat growing seasons from 1991 to 2018 at Wuqiao Experimental Station.**,P＜0.01;***,P＜0.001.

Over the 28 years,the lengths of the simulated key growth stages of vegetative growth period(VGP,from sowing to anthesis)and whole growth period(WGP,from sowing to maturity)both increased,but that of the reproductive growth period(RGP,from anthesis to maturity)decreased under the combined-effect scenario(Fig.5A-C).Cultivar change significantly extended the mean time of VGP and WGP by 6.9 and 5.7 days and reduced the RGP by 1.2 days(Fig.5G-I).A relatively constant trend in growth-stage lengths occurred in the long-term baseline(LTB)scenario.However,the climate-change scenario shortened the growth stages of VGP and WGP,owing mainly to rising temperature(Fig.5D,E).Increased temperature shortened the mean lengths of VGP and WGP by 2.8 and 2.7 days,respectively.Precipitation,solar radiation,and CO2concentration showed little effect on the key phenological stages.The combined effects of climate change and cultivars change jointly increased the mean times of VGP and WGP by respectively 4.4 and 3.2 days and reduced the RGP by 1.2 days.

Fig.2.Temporal trends in simulated yield under rainfed and potential yield(Yp)treatments,and yield gap between potential yield and actual field yields under the W1,W2,and W3 treatments.*,P＜0.05.

4.Discussion

4.1.Relationship between wheat yield and water use

In our study,ET0and ET were relatively constant over the 28 years,increasing slightly but not significantly(Figs.S2E,1B).The proportion of increase in crop yield was not the same as that in ET.The former was much greater than the latter,suggesting that grain yield of crops could be increased without much increase in water use.Many studies have shown that a highly linear and relatively constant relationship between biomass and water consumption in a given species[44].But the increase in biomass was greater than the increase in ET for winter wheat.Zhang et al.[45]showed that new cultivars,an increase in chemical fertilizer application,and an increase in soil fertility may all contribute to an increase in biomass and grain yield with less water consumption.The fertilizer application rate(ranging from 144 to 184.5 kg ha-1)remained largely unchanged from 1990 to 2018.Thus,the increase of yield and biomass was due mainly to genetic improvement.The lower increase in water use with greater increase in grain yield could be partly attributed to increase in harvest index,earlier flowering,longer grain-filling duration,and improved management practices[45].The increase in biomass at maturity contributed more to grain yield improvement than did the increase in the harvest index(Fig.S3A,B).

Increasing WUE in agriculture is expected to increase food and water security in China.Tilman et al.[46]showed that unless WUE is increased,greater agricultural production will require increased irrigation.Under the conditions of water resource scarcity and limited irrigation water,achievable grain yield is dependent mainly on increases in WUE[3].This dependence highlights the importance of attaining relatively high yields to attain high WUE.However,in our study,the increase in WUE occurred mostly from 1991 to 2004,then reaching a plateau(Fig.1C).Although the simulated mean potential yield(Yp)of winter wheat increased continuously after 2004(Fig.2A),the actual yield improved slowly from 2004 to 2018.This finding indicates that WUE was the main factor limiting further yield improvement after 2004.

4.2.Effects of cultivar change and climate change on yield and key phenological stages

Genetic improvement accounted for 0.03%-15.6% of yield increase,and climate change showed insignificant effects on yield increase under different irrigation treatments(Fig.4A).Most studies[11-13,17,18]have shown that genetic improvement and crop management practices have offset the negative effects of climatic change on yield improvement and that the rapid increase in grain yield has been dependent mainly on genetic improvement.Zhou et al.[15]showed that genetic improvement in grain yield was due mainly to earlier anthesis date,higher kernel weight and harvest index,and lower plant height.Liu et al.[25]reported that crop management from 1981 to 2010 reduced the lengths of the VGP and WGP,but increased the length of RGP in both spring and winter wheat.Our results differed from those of previous studies.In our study,genetic improvement prolonged the phenological stages of VGP and WGP,and reduced the RGP(Fig.5A-C).This finding is consistent with the increasing trend in biomass at anthesis and maturity,kernel number per spike,and unchanged trend in kernel weight(Fig.S5).Prolonging phenological stages of VGP and WGP implied that in adaptation to climate warming,cultivars with a longer VGP and WGP requirement have been adopted to increase grain yield in the past three decades.However,recent studies have given much lower weight to genetic yield potential improvement and shed light on the yield increase and spatiotemporal changes of wheat phenology,as well as their drivers.Tao et al.[26]found that climate warming outweighed agricultural management in affecting wheat phenology across China during 1981-2018.This inconsistency may be due to the different methods and different growing environments studied.

In our study,yield varied greatly from one year to the next year under the effects of climate change(Fig.S4),indicating that improvements in cultivars did not reduce the yearly yield variation influenced by weather.To reduce the seasonal yield variation caused by weather,one effective measure might be to develop new cultivars that could perform better under a wide range of climate conditions[47].Our study also showed that the amplitude of fluctuation gradually decreased with the increase of irrigation times(Fig.S4).In many crops,the instability of yield has been considered to be one of the main factors causing yield gap,especially in arid environments[48,49].In our study,the yield gap between the potential yield and the yield under the W1 condition increased significantly from 1991 to 2018(Fig.2B).This increase was caused mainly by the changes in irrigation and climate(especially precipitation)during the past 28 years.These results indicate that the sustainability of crop yield was poor under the condition of reduced water supply.However,the yield gaps between the potential yield and the yield under W2 and W3 treatments from 1991 to 2018 statistically neither increased nor decreased(Fig.2B).Thus,the W2 and W3 irrigation schemes may lead to yield sustainability.

Fig.3.Temporal trends in simulated yield under the combined-effect,long-term baseline,cultivar-change,and climate-change scenarios.Numbers in square brackets refer to regression coefficients and R2.*,P＜0.05;**,P＜0.01;***,P＜0.001.

4.3.Effects of climate variables on yield and key phenological stages

Precipitation,solar radiation,and CO2concentration showed little effects on wheat growth periods(Fig.5D-F).Previous study[50]have shown a decrease in climate-driven yield over time associated with an upward trend in temperature.In general,temperature warming leads to a yield loss by shortening the reproductive phase,accelerating leaf senescence,and causing stomatal closure.In our study,an increase in mean temperature shortened the VGP and WGP,and reduced wheat yield under W2,W3,and Yptreatments,but increased yield under rainfed and W1 irrigation treatments,although the difference was not significant(Fig.4B).Thus,reduced irrigation can mitigate the negative effects of rising temperatures on yields.This finding is not consistent with those of previous studies.The effects of temperature on yield were increased under drought conditions[51,52]and alleviated under irrigated conditions[53].Besides the direct effects on plant physiology and photosynthesis,high temperatures increase water demand and reduced soil water supply via increased evapotranspiration,leading to elevated water stress that impairs crop growth and yield formation[54].The reason for the discrepancy may be that the climate in this study did not show extremes such as droughts and heat waves.

Fig.4.The mean relative change in yield under rainfed,W1,W2,W3,and Yp(potential yield)treatments from 1991 to 2018 as a result of climate change,cultivar change,combined effect(A),and climate variables(B).*,P＜0.05;**,P＜0.01;***,P＜0.001.

Fig.5.Simulated key phenological stages of the wheat vegetative growth period(VGP,from sowing to anthesis;A,D),reproductive growth period(RGP,from anthesis to maturity;C,F),and whole growth period(WGP,from sowing to maturity;B,E)during 1991 to 2018 as a result of climate change,cultivar change,and climate variables.Relative changes in growth days were calculated under climate change,cultivar change,combined-effect,and long-term baseline scenarios(G-I).*,P＜0.05.

The damage to plants caused by increased high temperatures may be somewhat offset by CO2fertilization[55].In the APSIM model,climatic variables include mainly temperature,solar radiation,and precipitation,while CO2concentration is generally considered to be a fixed value.Whether elevated CO2concentrations are considered in climate-change impact assessments will influence the simulation results.Li et al.[56]found that if the expected increase in CO2concentration was not considered,cotton yield would decrease by 2%-15%,whereas in the contrary case it would increase by 30%-53%.CO2concentration showed a larger effect on C3 plant growth via three mechanisms:radiation use efficiency,transpiration efficiency,and critical leaf N concentration.Observations of crops grown under elevated CO2concentration showed that a mean increase of 13% in yield and 5% reduction in ET can be expected[57].In our study,elevated atmospheric CO2concentration insignificantly increased wheat yield under the three irrigation treatments(Fig.4B).However,increasing wheat grain yield in response to elevated CO2had been achieved by improving kernel number per spike(Fig.S5).The internal mechanisms may be the increased net leaf photosynthetic rate and the availability of dry matter in the floret.Consequently,floret death rates decreased and grain number increased[58].Elevated atmospheric CO2concentration offset the negative effects of temperature increase on yield,leading to the slight but nonsignificant increase in yield under climate change(Fig.4).

5.Conclusions

Field experiment data from 1990 to 2018 at Wuqiao Experimental Station in the NCP,together with the APSIM mode,were used to characterize yield sustainability and clarify the relative contributions of cultivars,climate,and its drivers to winter wheat yields and key phenological stages.Genetic improvement dramatically prolonged the phenological stages of vegetative growth period and contributed to yield increase by 0.03%-15.6%.The rapid increase in yield with lower water use was associated mainly with an increase in biomass with genetic improvement and partly with an increase in harvest index.Elevated atmospheric CO2concentration offset the negative effects of temperature increase on yield,leading to negligible change in yield under climate change.W2 and W3 limited-irrigation schemes can narrow yield gaps,increase wheat yield,and increase the sustainability of crop production in the NCP.

CRediT authorship contribution statement

Yanmei Gao:Writing-original draft.Meng Zhang:Data curation,Software,and Methodology.Zhimin Wang:Writing-review& editing.Yinghua Zhang:Writing-review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This study was supported by the National Natural Science Foundation of China(31871563),China Agriculture Research System of MOF and MARA(CARS-3),and Ministerial and Provincial Co-Innovation Centre for Endemic Crops Production with Highquality and Efficiency in Loess Plateau(SBGJXTZX-44).We thank Professor Xiaoguang Yang for her advice on the revision of this manuscript.We also thank the editor,and the anonymous reviewers for their valuable suggestions that improved the manuscript.

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2022.04.006.