环境因素-碳通量类文献Environmental Factors Determining Carbon Isotope Discrimination and Yield

Environmental Factors Determining Carbon Isotope Discrimination and Yieldin Durum Wheat under Mediterranean ConditionsJ.L.Araus,*D.Villegas,N.Aparicio,L.F.Garcı´a del Moral,S.El Hani,Y.Rharrabti,J.P.Ferrio,and C.RoyoABSTRACT conditions.Although water status during growth dra-matically affects yield and⌬,a specific environmental The effect of environment on the relationship between grain carbonvariable responsible for the positive relationship usually isotope discrimination(⌬)and yield was studied for durum wheatfound between⌬and yield across growing conditions (Triticum turgidum L.var.durum)under Mediterranean conditions.A group of25genotypes was grown under contrasting water regimes(i.e.,trials)has not been unequivocally identified(Far-in two regions of Spain during three years.The first objective was to quhar and Richards,1984;Craufurd et al.,1991;Ace-determine the environmental factors responsible for the strong posi-vedo,1993;Stewart et al.,1995;Araus et al.,1999a,b). tive relationship previously observed between⌬and yield across trials.For breeding purposes,it is crucial to assess how the Environmental factors tested were total water input(W i),mean tem-growing environment affects the relationship between perature,accumulated reference evapotranspiration(ET0),and the⌬and yield across genotypes(Acevedo,1993;Condonratio W i/ET0during different periods of the crop cycle.Water inputand Richards,1993;Richards,1996;Araus et al.,1999b). during grain filling was the variable most strongly correlated withPatterns in this relationship may be masked by pheno-grain⌬and yield across all the trials,as well as across the subset oflogical differences among genotypes that may affect trials in northeastern Spain.In southeastern Spain,the most droughtyield and also⌬,especially in drought-prone environ-prone of the two regions,W i from sowing to heading explained themost variation in grain⌬and yield.The second objective was to study ments.For example,under Mediterranean conditions the effect of environment on the relationship between⌬and yield those genotypes with fewer days from sowing to heading across genotypes.No significant correlation was found for trials with or to anthesis show higher⌬values(Craufurd et al., a mean yield up to about2000kg haϪ1,but the strength of the relation-1991;Richards and Condon,1993;Richards,1996;Araus ship increased sharply and attained significance in trials yielding2500et al.,1998)probably because they attain grain filling kg haϪ1.When yield above2500kg haϪ1the correlation between⌬with more water in the soil,whereas the evapotranspira-and yield remained relatively steady and positive,with an r valuetive demand is lower.Nevertheless,in bread wheat(Trit-around0.5.It is concluded that breeding to raise durum wheat yieldicum aestivum L.,Sayre et al.,1995)and durum wheat in Mediterranean conditions could take advantage of selecting for(Araus et al.,1998)large genotypic variability in⌬, higher⌬only in relatively wet years or under supplementary irrigation.independent of phenology,has been reported.This study was performed on a large collection of D rought,defined as water deficit,and often com-Spain that differ in drought severity.The objective wasdurum wheat genotypes cultivated in a wide range of bined with high temperature stress,is one of theto determine whether there is a common specific envi-greatest constraints to cereal grain yield in Mediterra-ronmental variable which simultaneously affects the⌬nean areas.For crops such as durum wheat,grown underof mature kernels and yield.That information may indi-rainfed conditions,agricultural practices are not suffi-cate the basis of the widely reported strong positive cient to mitigate the effect of drought,and plant breed-relationship between yield and⌬across trials.An addi-ing has become the best tool for yield increases(Sriva-tional objective was to assess how growing conditions stava,1991;Acevedo,1991;Slafer et al.,1994;Ceccarelliaffect the strength and sign(positive or negative)of the and Grando,1996;Royo et al.,1998).relationship between⌬and yield across genotypes. Carbon isotope discrimination(⌬),when measuredin plant dry matter,integrates transpiration efficiency,the ratio of net photosynthesis to water transpired,over MATERIALS AND METHODSthe period during which the dry matter is assimilated,Plant Materials and Experimental Design⌬and transpiration efficiency being negatively related(Farquhar and Richards,1984;Condon et al.,1990).On A total of12field trials were performed from1997to1999in two Mediterranean regions,northeastern(NE)and dry matter basis,⌬has been proposed as a breedingsoutheastern(SE)Spain,and under two contrasting water criterion for increasing yield in temperate cereals andregimes(rainfed and irrigated)within each region.The trial other crops,under either favorable or drought stresslocations and soil characteristics are summarized in Table1,and environmental conditions are detailed in Table2.Appro-priate fertilization was provided to the seed bed,and trials J.L.Araus and J.P.Ferrio,Unitat de Fisiologia Vegetal,Facultat dewere top-dressed at the onset of jointing depending on the Biologia,Universitat de Barcelona,Diagonal645,08028Barcelona,Spain;D.Villegas,N.Aparicio and C.Royo,Area de Conreus Exten-sius,Centre UdL-IRTA,Alcalde Rovira Roure191,25198Lleida,Abbreviations:⌬,grain carbon isotope discrimination;CIMMYT, Spain;L.F.Garcı´a del Moral,S.El Hani and Y.Rharrabti,Departa-Centro Internacional de Mejoramiento de Maı´z y Trigo;DA,days mento de Biologı´a Vegetal,Facultad de Ciencias,Universidad defrom sowing to anthesis;DH,days from sowing to heading;ET0,ac-Granada,18071Granada,Spain.Received5Feb.2002.*Correspond-cumulated reference evapotranspiration;ICARDA,International ing author(josel@porthos.bio.ub.es).Center for Agricultural Research in the Dry Areas;NE-,northeastern; Published in Crop Sci.43:170–180(2003).SE-,southeastern;W i,total water input.170ARAUS ET AL.:ENVIRONMENT,GRAIN⌬,AND YIELD IN DURUM WHEAT171 Table1.Description and location of the sites used in the study.Cultivation was performed in two sites from NE Spain(Lleida province, Catalonia)and SE Spain(Granada province,East-Andalusia).Soil texture Site Region Water Regime Coordinates Altitude Soil type(USDA)Soil texture Soil pH Sand Silt Claym above sea level% Gimenells NE Spain Irrigated41؇40؅N;200Calcixerolic-Xerochrept fine-loamy8.131.039.030.00؇20؅EEl Cano´s NE Spain Rainfed41؇41؅N;440Fluventic-Xerochrept loamy-fine8.234.147.118.81؇13؅EGranada SE Spain Irrigated37؇21؅N;650Typic Xerofluvent silty clay8.040.049.610.43؇35؅WVentas de Huelma SE Spain Rainfed37؇10؅N;720Loamy Carcixerolic silty clay8.217.854.927.33؇50؅W Xerochreptwater status of the crop.When applied,supplementary irri-the grain was ripe,yield(kg haϪ1)expressed at a100g kgϪ1moisture content(Table2).gation was given during late winter and spring(Table2).Presowing subsoil moisture was not relevant for any of thetrials assayed.Environmental Parameters Trials consisted of25durum wheat genotypes grown inFor each trial,W i was calculated as the sum of rainfall and randomized complete blocks with four replicates in plots ofirrigation(when applicable)for the following periods:sowing 12m2(six rows,0.20m apart).These genotypes included fourto heading,sowing to anthesis,sowing to maturity,heading commercial Spanish cultivars(Altar-aos,Jabato,Mexa,andto anthesis,heading to one week after anthesis,heading to Vitro´n),and21genotypes from the eastern Mediterraneanmaturity,anthesis to maturity(grain filling),and from one basin,including commercial cultivars and advanced lines ofweek after anthesis to maturity.Mean temperature and refer-the CIMMYT/ICARDA durum wheat breeding programence ET0for the same periods were also assessed.ET0was (Awalbit,Bicrecham-1,Chacan,Chahra-1,Haurani,Korifla,Krs/calculated from the mean maximal and minimal average tem-Haucan,Lagost-3,Lahn/Haucan,Massara-1,Moulchahba-1,peratures of each period using the PC-program ETO(Sub-Mousabil-2,Omlahn-3,Omrabi-3,Omruf-3,Quadalete//Erp/zero Evapotranspiration)(Snyder and Pruitt,1991)version Mal,Sebah,Stojocri-3,Waha,Zeina-1,and Zeina-2).The ge-1.04(revised in February,1994).notypes were chosen to represent a wide range of geneticvariability in terms of agronomical characteristics.Moreover,Carbon Isotope Discriminationto minimize the potential interference of phenology on yieldand⌬,these genotypes were selected with a relatively narrow For each plot,a sample of about2g of mature kernels was range of variability in the number of days from planting to oven dried and finely ground(mesh diameter of0.5mm).The heading,days to anthesis,and days to physiological maturity13C/12C ratio of samples was subsequently determined by mass (a mean range of7and6d for heading and anthesis dates,spectrometry at the Serveis Cientı´fico-Te`cnics de la Universi-respectively).The times from sowing to heading(DH)and tat de Barcelona,Spain.Samples of0.7to0.9mg were com-anthesis(DA)were recorded when more than half the plants busted in an elemental analyzer(EA1108,Series1,Carlo Erba in a plot reached the stages55and65of Zadoks’scale(Zadoks Instrumentazione,Milan,Italy)and the13C/12C ratio was mea-et al.,1974),respectively.Thermal time was calculated in sured with an isotope ratio mass spectrometer(Delta C,Finni-growing degree-days(GDD)by summing the daily values of gan Mat,Bremen,Germany)operated in continuous flow mode. mean temperature,with a base temperature of0ЊC(Gallagher,A system check for elemental analyses was achieved with an 1979).Physiological maturity was recorded when most of the interspersed working standard of atropine.Stable carbon iso-plants had reached Zadoks’stage87,representing kernels in tope composition was expressed as␦13C values(Farquhar et hard dough stage.Plots were harvested mechanically whenal.,1989),where␦13C(‰)ϭ[(R sample/R standard)-1]ϫ1000,Table2.Growing conditions and main agronomical characteristics of the trials performed during this study.Days from sowing to either heading(DH)or anthesis(DA),total water input(Wi),reference evapotranspiration(ET0),carbon isotope discrimination of kernels(⌬).W iThermal Thermal Thousand Sowing time to time to Irrigation Mean Seasonal kernel Environment Year date DH DA heading anthesis Rainfall(times)Temperature ET0Wi/ET0⌬Yield weightd gdd Mm؇C mm‰kg haϪ1gNE Spainirrigated19973Dec.9613514312001300258150(3)11.54970.8218.36b4964c45.1e 199823Nov.9715015713101411285100(2)10.34140.9316.74e5192b46.7d199910Nov.9816217212311369255150(3)9.64820.8418.72a7009a54.0a NE Spainrainfed19973Dec.9613814412601334230010.84010.5714.59i2062g42.5f 199817Nov.971551631294140218309.83900.4714.63hi2531f47.8c19993Nov.9817518413511496256010.04730.5417.10d3820e51.7b SE Spainirrigated19975Feb.97869211491242173200(2)15.54870.7717.00d2624f37.7h 199811Dec.9712913612961406311100(1)13.35740.7217.48c4312d46.4d199915Dec.9812413811571356128300(3)13.15590.7716.34f3628e31.3i SE Spainrainfed199713Dec.96849311821326134016.04900.2715.51g1853h42.3f 199821Jan.9811612214341538180014.65080.3515.62g2053gh40.1g199925Nov.9814716613941702193012.35370.3614.74h2547f42.4f172CROP SCIENCE,VOL.43,JANUARY–FEBRUARY 2003Table 3.Percentages of the sum of squares obtained in the analy-and R is the 13C/12C ratio.Secondary standards of graphite,sis of variance for grain yield and carbon isotope discrimination sucrose,and polyethylene foil (IAEA,Vienna,Austria)cali-of mature grains (⌬)of 25durum wheat genotypes grown in brated against Peedee belemnite (PDB)carbonate were used two regions (NE and SE Spain)under two water regimes for comparison.The accuracy of the ␦13C measurements was (rainfed and irrigated)during three years (1997to 1999sea-Ϯ0.1‰.Following Farquhar et al.(1989),⌬was further calcu-sons).For the analysis days from sowing to heading has been lated from ␦13C as ⌬ϭ(␦a Ϫ␦p )/(1ϩ␦p ),where ␦a and ␦p considered as a covariant.refer to air and plant,respectively.On the PDB scale,free Source of variationdf Yield ⌬atmospheric CO 2,␦a ,has a current composition of approxi-mately Ϫ8‰(Farquhar et al.,1989).Days to heading 123.4*** 2.3***Year 20.4ns 1.6***Region113.0***9.0***Statistical AnalysisYear ϫRegion 2 5.0***22.7***WR†(Region)242.8***51.2***Analyses were done with the SAS-STAT package (SAS Year ϫWR (Region)4 1.8** 5.4***Institute Inc.,1996).An analysis of variance was performed Block (Year ϫWR ϫRegion)36 3.3*** 1.7***Genotype24 2.5*** 1.9***for grain yield and ⌬,considering days from sowing to heading Genotype ϫYear 48 1.5*** 1.0***or from sowing to anthesis as the covariant.Means of trials Genotype ϫRegion240.8**0.4*for yield and ⌬were compared by the LSD test at P ϭ0.05.Genotype ϫYear ϫRegion 48 1.6***0.7**Stepwise discriminant analysis was used to ascertain the re-Genotype ϫWR (Region)48 1.3**0.8***Genotype ϫYear ϫWR (Region)96 2.6*** 1.3**lationship between either ⌬or grain yield as the dependent Residual86412.57.2variable and three different periods of the crop cycle (sowing to heading,heading to anthesis and anthesis to maturity)for *,Significant at 0.05probability level.**,Significant at 0.01probability level.each environmental parameter as the independent variables.***,Significant at 0.001probability level.The relationship between Pearson’s correlation coefficients ns,Nonsignificant.of individual trials and mean grain yield was fitted by Table †WR,water regime.Curve (Jandel Co.,1994).Although variation in phenology was limited within tri-als,DH probably also reflects an environmental effect RESULTSas the trials differed dramatically in DH (Table 2).The two regions differed in rainfall,temperature,and Therefore,environmental conditions were responsible evaporative demand,SE Spain being markedly more by far for most of the variability observed in yield and arid,as inferred from the ratio W i /ET 0of the rainfed ⌬.The main environmental factor affecting yield and ⌬sites.Differences between irrigated and rainfed sites in was water regime,rainfed or irrigated,but the interac-this ratio were also evident (Table 2).The number of tion between year and region also had a strong affect days from sowing to either heading (DH)or anthesis on ⌬.Grain yield was doubled and ⌬was 2.5‰higher in (DA)was higher in NE than in SE Spain (Table 2).irrigated than in rainfed trials in NE Spain and 64%and Heading was at approximately 1235growing-degree 1.6‰higher,respectively,in SE-Spain trials (Table 2).days and anthesis at approximately 1340in all trials ex-cept two.The exceptions were both in one of the sites Effect of Environment on the Relationshipin SE Spain,and had a higher number of growing-degree between ⌬and Yield across Trialsdays than the other sites (Table 2).The effect of environment,understood as the combi-The ⌬values of mature kernels and grain yield were strongly (P Ͻ0.001)and positively correlated across nation of region and water regime,on grain yield and ⌬of grain was much higher than that of genotypic vari-all the trials (Fig.1).Correlations between these traits in trials across either NE or SE Spain were similar inability (Table 3).Also,DH had a strong effect on yield.Fig.1.Relationship across the whole set of trials between the carbon isotope discrimination of mature grains and grain yield.Each point represents a rainfed or irrigated trial (composed by 25genotypes and four replicates per genotype)grown in NE Spain or SE Spain.ARAUS ET AL.:ENVIRONMENT,GRAIN⌬,AND YIELD IN DURUM WHEAT173 Table4.Pearson’s correlation coefficients of the relationship across trials between both grain yield and carbon isotope discrimination of mature grains(⌬),and some environmental variables recorded for different periods of the crop cycle.Correlations have been calculated with the whole set of trials(nϭ12)used in this study and presented in Table2.W i†Mean temperature ET0†W i/ET0 Growing period Yield⌬Yield⌬Yield⌬Yield⌬Sowing to maturity0.75**0.78**Ϫ0.54Ϫ0.150.080.320.78**0.72** Sowing to heading0.370.47Ϫ0.65*Ϫ0.28Ϫ0.52Ϫ0.490.440.54 Heading to anthesis0.570.27Ϫ0.110.21Ϫ0.19Ϫ0.280.59*0.24 Heading to one week after anthesis0.460.18Ϫ0.36Ϫ0.15Ϫ0.17Ϫ0.280.510.16 One week after anthesis to maturity0.500.72**Ϫ0.16Ϫ0.020.330.60*0.440.55 Heading to maturity0.80**0.75**Ϫ0.180.050.240.460.81**0.65* Sowing to anthesis0.60*0.57Ϫ0.66*Ϫ0.30Ϫ0.51Ϫ0.540.64*0.63* Anthesis to maturity0.69*0.82**Ϫ0.140.010.330.59*0.69*0.73** *,Significant at0.05probability level.**,Significant at0.01probability level.†Wi,water input(rainfall plus irrigation);ET0,accumulated reference evapotranspiration.slope.To ascertain the basis of these relationships,Pear-in NE and SE Spain separately(Table5).Water input or son’s correlation coefficients were calculated across theW i/ET0from anthesis to maturity were the independent 12trials between these traits and several environmental variables selected first in the stepwise analysis for the parameters evaluated during the crop cycle(Table4)whole set of trials and the subset of NE Spain.In the plus DH and DA,these last being nonsignificant at the case of SE Spain,however,W i or W i/ET0from sowingto heading were the first choices in the analysis(Ta-Pϭ0.05level(DH,rϭ0.51and rϭ0.14;DA,rϭ0.49and rϭ0.11for yield and⌬,respectively).The ble5),and were also linearly related with yield and⌬(Fig.3).Moreover,W i and W i/ET0,either from heading strongest relationships observed with either yield or⌬were those that involved W i,for the whole crop cycle to maturity(Fig.3),or anthesis to maturity,were corre-or just from the last part of the cycle.The W i/ET0ratio lated with yield and⌬only for NE Spain.The slope of for the complete crop cycle,between heading and matu-all these relationships tended to be steeper in NE than rity,or only during grain filling,was also significantly in SE Spain.correlated with yield and⌬.For the interval from sowingto heading,W i and W i/ET0showed much weaker rela-Effect of Environment on the Relationship tionships with yield and⌬.In general,mean temperature between⌬and Yield within Trialsand accumulated evapotranspiration correlated lessTo minimize the potential interference of phenology strongly with yield and⌬regardless of the stage of theon the phenotypic relationship between grain⌬and yield, crop cycle.Significant relationships of either W i or W i/the set of genotypes chosen in this study had a relatively ET0with yield and⌬were linear(Fig.2and Fig.3).small variability in DH and DA within each trial.Analy-Stepwise analysis showed W i and W i/ET0were againsis of variance for⌬and yield using DH or DA as co-the parameters best correlated with yield and⌬.The com-variables confirmed that the effect of genotype on these bination of W i or W i/ET0for the three crop cycle periodstraits remained highly significant after subtracting the explained about70%of the total variability in yield and⌬across the12trials and about90%across the six trials effect of phenology expressed as DH or DA(Table3).Fig.2.Relationship across the whole set of trials between water input(including irrigation if applied)from sowing to maturity and either grain yield or carbon isotope discrimination.Each point represents either a rainfed or irrigated trial(composed by25genotypes and four replicates per genotype).174CROP SCIENCE,VOL.43,JANUARY–FEBRUARY2003Table5.Stepwise analysis taking yield and⌬as dependent variables,and three different periods of the crop cycle(sowing to heading, heading to anthesis and anthesis to maturity)for each environmental parameter as independent variables.Values are proportions (per one)of the total variability in grain yield and carbon isotope discrimination across the12trials attributable to either a given environmental variable or explained by the progressive combination of these variables.Period Yield Yield accumulated Period⌬⌬accumulatedWater input(W i)NE-and SE-Spain trials combinedAnthesis–maturity0.48*0.48Anthesis–maturity0.67**0.67 Heading–anthesis0.130.61Sowing–heading0.100.77 Sowing–heading0.060.67Heading–anthesis0.000.77W i/ET0ratio NE-and SE-Spain trials combinedAnthesis–maturity0.47*0.47Anthesis–maturity0.53**0.53 Heading–anthesis0.130.60Sowing–heading0.20*0.73 Sowing–heading0.110.71Heading–anthesis0.010.74Water input(W i)NE-Spain trialsAnthesis–maturity0.85**0.85Anthesis–maturity0.83*0.83 Heading–anthesis0.12*0.97Sowing–heading0.010.84 Sowing–heading0.010.98Heading–anthesis0.010.85Water input(W i)SE-Spain trialsSowing–heading0.600.60Sowing–heading0.560.56 Heading–anthesis0.100.70Anthesis–maturity0.330.89 Anthesis–maturity0.050.75Heading–anthesis0.050.94W i/ET0ratio NE-Spain trialsAnthesis–maturity0.90**0.90Anthesis–maturity0.69*0.69 Heading–anthesis0.020.92Sowing–heading0.050.74 Sowing–heading0.000.92Heading–anthesis0.010.75W i/ET0ratio SE–Spain trialsSowing–heading0.66*0.66Sowing–heading0.600.60 Heading–anthesis0.060.72Anthesis–maturity0.260.86 Anthesis–maturity0.010.73Heading–anthesis0.030.89*,Significant at0.05probability level.**,Significant at0.01probability level.Fig.3.Relationship across the subset of trials,from NE Spain or SE Spain,between water input from sowing to heading and(a)grain yield or(b)carbon isotope discrimination of mature grains(⌬),and between water input from heading to maturity and(c)grain yield or(d)carbonisotope discrimination.Each point represents either a rainfed or irrigated trial(composed by25genotypes and four replicates per genotype) grown in NE Spain or SE Spain.ARAUS ET AL.:ENVIRONMENT,GRAIN⌬,AND YIELD IN DURUM WHEAT175Fig.4.Relationship between mean trial grain yield and the Pearson’s correlation coefficient(r)of the relationship between grain yield and carbon isotope discrimination of mature kernels within the same trial(a).The correlation coefficients of the same relationships after subtracting the effect of DH are also shown(b).Each point represents a trial composed by25genotypes and four blocks per genotype.Trials were performed in NE Spain and SE Spain under two moisture regimes(irrigated and rainfed)and over3yr(1997–1999).Genotype effects,although only accounted for about no significant(PϽ0.05)relationships,but coefficientsof correlation increased sharply to attain significance 2%of total variability in both yield and⌬,were highlysignificant.On the other hand,these phenological traits for trials yielding about2500kg haϪ1.Above this yield, also had a highly significant effect,particularly on yieldthey remained steady at around0.5,except for a trial (Table3).However,the effect of phenology in the analy-under irrigation in SE Spain with rϭ0.69.Many of sis of variance may have incorporated an environmentalthese relationships remained almost unchanged after bias.Moreover,no significant correlation across geno-subtracting the effect of DH(Fig.4b),although the types was observed between DH or DA and either⌬correlation coefficient in the highest yielding trial was or grain yield in any of the trials.greatly decreased when the effect of DH was removed. For the12trials,mean trial yield was plotted againstThe slope of the linear regression between yield and⌬the Pearson’s correlation coefficient of the phenotypic showed the same pattern of changes as the growing relationship between⌬and yield within the same trialconditions of the trials improved.It increased from val-(Fig.4a).This coefficient was positive in all the trials.ues near zero in the lowest yielding trial(SE Spainrainfed1997)(Table2)to around1000kg haϪ1per1‰The relationship fitted an asymptotic function(PϽ0.01).Thus,trials yielding around2000kg haϪ1showed increase in⌬,for trials that yielded about2500kg haϪ1.176CROP SCIENCE,VOL.43,JANUARY–FEBRUARY2003Fig.5.Relationship between mean trial grain yield and the slope of the regression of the relationship between grain yield and carbon isotope discrimination of mature kernels within the same trial(a).Slopes of the same relationship after subtracting the effect of DH are also shown(b).Each point represents a trial composed by25genotypes and four blocks per genotype.Trials were performed in NE Spain and SE Spainunder two moisture regimes(irrigated and rainfed)and over3yr(1997–1999).Above these yields,slope values remained steady(Fig.(PϽ0.1)with yield and not correlated with⌬across5a).Moreover,the slopes of these relationships were trials.The lack of a stronger correlation between cropunaffected when the effect of DH was subtracted(Fig.phenology and yield and its absence for⌬could be due5b).From the set of25genotypes two groups of five to the dependence of DH and DA on temperature(Hay,genotypes each,having the extreme values of⌬at the1999and references herein).Indeed,mean temperaturetwo most productive trials were selected.Further,for from sowing to heading and from sowing to anthesiseach trial the percentage in yield as a result of picking was negatively correlated(PϽ0.05)with yield but notbased on these two groups was calculated as[mean yield with⌬.In our study,differences in DH and DA betweenhigh⌬group–mean yield low⌬group/mean yield low trials were mainly caused by regional differences in tem-⌬group]ϫ100.The average yield differential showed perature.Thus,the higher temperatures in the trials in a similar pattern to that on Fig.4(but PϽ0.10).Thus,SE Spain caused an accelerated accumulation of grow-the high⌬group consistently showed a higher yield ing-degree days and therefore a shorter DH and DA(about10%)for trials yielding about2500kg haϪ1than in those in NE Spain.and above.Role of Environmental Variables in the DISCUSSION Relationships between⌬and Yield across TrialsRole of Phenology in the RelationshipsOur results show that W i for the whole crop cycle between⌬and Yield across Trials and specifically that comprising the part from headingonwards was the main environmental variable affecting In spite of the large differences between trials in DHand DA,both parameters were only weakly correlated grain yield and⌬.Sharing a common environmentalARAUS ET AL.:ENVIRONMENT,GRAIN ⌬,AND YIELD IN DURUM WHEAT177Table 6.Environmental conditions (means over three years)dur-variable would explain the strong positive relationship ing the last part of the of crop culture (from heading to physio-between grain yield and ⌬across trials (Fig.1).Similar logical maturity).results have been reported in durum wheat (Araus et Wi/ET 0†TemperatureVPD‡al.,1999b)and other cereals (Condon et al.,1987;Roma-؇C mbar gosa and Araus,1991;Acevedo,1993).NE–Spain irrigated 0.7717.610.4Previous reports on barley (Hordeum vulgare L.)and NE–Spain rainfed 0.3116.912.4durum wheat under Mediterranean conditions have SE–Spain irrigated 0.4219.319.6found a strong dependence of grain ⌬on W i during the SE–Spain rainfed0.2119.921.2later stages (from heading and anthesis to maturity)of †Wi/ET 0,ratio of water input versus reference evapotranspiration.the crop cycle (Araus et al.,1999a).In the present study,‡VPD,vapor-pressure deficit at anthesis.W i explained the differences in ⌬across growing envi-ronments better than ET 0or even the ratio W i /ET 0(Ta-The pattern of the relationships between W i and yield ble 4)as well as other related variables such as ET 0–W i and ⌬suggest that for the same amount of water input,(data not shown).The strength of the relationships of a higher grain yield and ⌬were attained in NE Spain.W i /ET 0with yield and ⌬was almost comparable to that These regional differences could initially be explained obtained with W i .However,because ET 0alone related in terms of a higher ET 0or vapor-pressure deficit in the poorly to yield and ⌬,the good performance of W i /ET 0trials done in the south (Table 6)or by differences in soil seemed to be due only to the weight of W i in this ratio.In water-holding capacity and distinct seasonal patterns fact,the range of variability in accumulated ET 0across of rainfall.Canopy temperature depression at anthesis environments was smaller than that for W i .Stewart et measured by infrared thermometry was about 5ЊC higher al.(1995),working with wild plant communities along in SE than in NE Spain regardless of the water regime a natural rainfall gradient,also concluded that W i was assayed (Royo et al .,2002).This observation suggests the environmental variable best related to carbon iso-higher transpiration rates in the southern trials,irrespec-tope composition.Most of the studies on this topic,how-tive of whether stomatal conductance was lower than ever,deal with tree species and not annual plants (see in the northern trials as can be inferred from the lower references below).In general,these studies are not con-⌬values of SE Spain (Table 2).Thus,Condon et al.clusive in terms of agreement on a common,single envi-(1992)reported a decrease in ⌬in wheat which was at-ronmental variable.They relate ⌬to rainfall amounts,tributed to stomatal closure in response to increasing evapotranspiration,the W i /ET 0ratio,soil water status,VPD.However,complementary explanations may be or different drought stress indices either at the seasonal sought because the relationships between the W i /ET 0level or during a particular period of the growing season ratio and grain yield or ⌬in NE and SE Spain showed (Francey and Farquhar,1982;Freyer and Belacy,1983;the same pattern as those with W i illustrated in Fig.3.Leavitt and Long,1989;Dupouey et al.,1993;Guehl et A possible explanation could be a lower water holding al.,1995;Korol et al.,1999;Moore-Darrin et al.,1999).capacity of the sandy soils in SE Spain (Table 1).De-Water input during grain filling was not only the variable creased soil water availability results in lower values of best correlated with yield and ⌬for the whole set of ⌬in cereals (Farquhar and Richards,1984;Hubick and trials (Table 4),but also for the subset of NE Spain (Fig.Farquhar,1989;Condon et al.,1992)primarily because 3).However,whereas no significant correlations were of the decrease in stomatal conductance.Other soil char-attained,in SE Spain,W i from the first part of the crop acteristics,such as soil compaction,may also affect ⌬cycle (sowing to heading)tended to be the variable best in wheat (Masle and Farquhar,1988).related to yield and ⌬(Fig.3).On the other hand,no significant correlation was attained between W i from Role of Environment on the Relationshipssowing to heading and yield or ⌬in NE-Spain trials (Fig.between ⌬and Yield within Trials3).These differences between regions may have an envi-ronmental basis.The Mediterranean environment of SE For all the trials,the slope of the regression between Spain is characterized by more severe drought stress ⌬and grain yield remained positive.In fact,for wheat during the last part of the crop cycle than is that of and other cereals,⌬is frequently positively correlated NE Spain (Table 6).The presence of a higher level of with grain yield and/or total biomass not only under drought stress during grain filling in SE-Spain trials is well-irrigated but also under rainfed conditions (Con-further supported by the generally lower 1000-kernel don et al.,1987;Romagosa and Araus,1991;Kirda et weight in SE than in NE Spain (Table 2).Under the al.,1992;Araus et al.,1993,1998;Merah et al.,2002;growing conditions of SE Spain,grain yield and ⌬are Morgan et al.,1993;Sayre et al.,1995).Most of these defined earlier in the crop cycle because photoassimi-(and other studies reporting positive relationships)were lates produced before anthesis may play a major role conducted in Mediterranean or similar environments not only in defining the total number of grains per unit where there is strong reliance on “within-season”rain-land,but also in the filling of these grains (Slafer and fall (Condon et al.,2002).However,unlike the results Araus,1998;Royo et al.,1999).Interestingly,for two of Araus et al.(1998)in durum wheat,in the poor-of the three years studied,the ⌬for the rainfed site in yielding environments of our study (about 2000kg ha Ϫ1),SE Spain was higher than in that in the north in spite the correlation between grain ⌬and yield across geno-of the lower yield and greater drought in the former types was not significant,with correlation coefficients for two rainfed trials from SE Spain close to zero.Suchlocation (Table 2).。