Yield and water-production functions of two durum wheat cultivars grown under different irrigation and nitrogen regimes

Transcription

1 agricultural water management 96 (2009) available at journal homepage: Yield and water-production functions of two durum wheat cultivars grown under different irrigation and nitrogen regimes Fadi Karam a, *, Rabih Kabalan b, Joêlle Breidi b, Youssef Rouphael c, Theib Oweis d a Lebanese Agricultural Research Institute, Department of Irrigation and Agro Meteorology, P.O. Box 287, Zahleh, Lebanon b Lebanese Agricultural Research Institute, Department of Plant Breeding, P.O. Box 287, Zahleh, Lebanon c Department of Crop Production, Faculty of Agricultural and Veterinary Sciences, Lebanese University, Dekwaneh-Al Maten, Lebanon d International Center for Agricultural Research in the Dry Areas, P.O. Box 5466, Aleppo, Syria article info Article history: Received 29 July 2008 Accepted 27 September 2008 Published on line 5 November 2008 Keywords: Irrigation strategy Supplemental irrigation Triticum durum L. Vapor pressure deficit Water use efficiency abstract Wheat (Triticum durum L.) yields in the semi-arid regions are limited by inadequatewater supply late in the cropping season. Planning suitable irrigation strategy and nitrogen fertilization with the appropriate crop phenology will produce optimum grain yields. A 3-year experiment was conducted on deep, fairly drained clay soil, at Tal Amara Research Station in the central Bekaa Valley of Lebanon to investigate the response of durum wheat to supplemental irrigation (IRR) and nitrogen rate (NR). Three water supply levels (rainfed and two treatments irrigated at half and full soil water deficit) were coupled with three N fertilization rates (100, 150 and 200 kg N ha 1 ) and two cultivars (Waha and Haurani) under the same cropping practices (sowing date, seeding rate, row space and seeding depth). Averaged across N treatments and years, rainfed treatment yielded 3.49 Mg ha 1 and it was 25% and 28% less than half and full irrigation treatments, respectively, for Waha, while for Haurani the rainfed treatment yielded 3.21 Mg ha 1, and it was 18% and 22% less than half and full irrigation, respectively. On theotherhand,nfertilizationof150and200 kg N ha 1 increasedgrainyieldinwahaby12%and 16%,respectively,incomparisonwithNfertilizationof 100 kg N ha 1,whilefor cultivarhaurani the increases were 24% and 38%, respectively. Regardless of cultivar, results showed that supplemental irrigation significantly increased grain number per square meter and grain weight with respect to the rainfed treatment, while nitrogen fertilization was observed to have significant effects only on grain number per square meter. Moreover, results showed that grain yield for cultivar Haurani was less affected by supplemental irrigation and more affected by nitrogen fertilization than cultivar Waha in all years. However, cultivar effects were of lower magnitude compared with those of irrigation and nitrogen. We conclude that optimum yield was producedfor bothcultivarsat 50% ofsoilwater deficitassupplementalirrigation and N rate of 150 kg N ha 1. However, Harvest index (HI) and water use efficiency (WUE) in both cultivars were not significantly affected neither by supplemental irrigation nor by nitrogen rate. Evapotranspiration (ET) of rainfed wheat ranged from 300 to 400 mm, while irrigated wheat had seasonal ET ranging from 450 to 650 mm. On the other hand, irrigation treatments significantly affected ET after normalizing for vapor pressure deficit (ET/VPD) during the growing season. Supplemental irrigation at 50% and 100% of soil water deficit had approximately 26 and 52 mm mbar 1 more ET/VPD, respectively, than those grown under rainfed conditions. # 2008 Elsevier B.V. All rights reserved. * Corresponding author. Tel.: ; fax: address: fkaram@lari.gov.lb (F. Karam) /$ see front matter # 2008 Elsevier B.V. All rights reserved. doi: /j.agwat

2 604 agricultural water management 96 (2009) Introduction Water is considered the most limiting factor for cereal production in the central plains of the Bekaa Valley of Lebanon with typically dry Mediterranean climate. Environmental conditions are characterized by adequate amounts of rainfall during winter (December to February) while few precipitation events are registered in spring from mid-march to mid-may. However, the unfavorable distribution of rain over the growing season and the year-to-year fluctuations constitute a major constraint to wheat growth and yield. Under these conditions, wheat plants generally suffer a midseason drought stress that can reduce grain number per spike, while grain weight may suffer a terminal stress caused by high temperatures at the end of the cropping season (García del Moral et al., 2003). As a result, soil water in the root zone often does not satisfy crop water needs over the whole season, especially in spring months where the chances of rain to occur become less, and where most of the crop growth occurs (Loss and Siddique, 1994). Therefore, irrigation late in the season is required to match soil water stress and to stabilize yields (Campbell et al., 1993; Oweis et al., 1999). Wheat constitutes almost 50% of the area cropped with cereals in Lebanon and most of the cultivated lands are in the Bekaa Valley, which by itself accounts for 42% of the total agricultural land in the country (Karam and Karaa, 2000). Data reported by the FAO (2005) indicated that despite a slight increase in cereal yields in Lebanon between 1992 and 2004 (from 2.1 to 2.5 Mg ha 1 ), average annual growth rate of cereal production per capita was found to decrease drastically between the two periods from 4.6% to 2.5%. Therefore, a challenge the Lebanese agriculture has to face in the coming years is to increase cereal yields to higher levels to satisfy the food requirements of the population and to reduce the gap between the rapid growth rate and food requirements. Supplemental irrigation is an alternative to increase and stabilize yields of crops grown in rainfed areas (Howell et al., 1975; Zhang et al., 1998; Oweis et al., 1999). It can be defined as the addition to essentially rainfed crops of small amounts of water during times when rainfall fails to provide sufficient water for normal plant growth and optimal yield (Duivenbooden et al., 1999; Oweis et al., 1999). Wheat s most sensitive growth stages to water stress with respect to grain yield are stem elongation and booting, followed by anthesis and grain filling (Blum and Pnuel, 1990; Shpiler and Blum, 1991; García del Moral et al., 2003). Water deficit around anthesis may lead to a loss in yield by reducing spike and spikelet number and the fertility of surviving spikelets (Giunta et al., 1993), while water deficit during grain-filling period reduces grain weight (Royo et al., 2000). Moreover, Edmeades et al. (1989) pointed out that the lack of rainfall in spring time caused a water deficit for rainfed wheat around anthesis that increases in severity throughout the grain-filling period. Likewise, Giunta etal. (1993) and Zhong-hu and Rajaram(1994) found that kernels per spike and spike number per square meter were the yield components most sensitive to drought, while kernel weight remains relatively stable due to the high translocation of assimilates stored during the pre-anthesis period. Understanding the effect of water stress on yield formation becomes an essential step for planning a suitable irrigation strategy for wheat. The amount of water may be scheduled at booting-anthesis and grain-filling and in dry years irrigation may be needed as early as at stem elongation stage to ensure rigorous canopy development (Oweis et al., 1999; Chen et al., 2003). On the other hand, matching nitrogen (N) supply to plant water availability is essential for a successful grain yield. In that sense, Tilling et al. (2007) demonstrated that the response of wheat to nitrogen fertilization is heavily reliant on rainfall distribution. Moreover, the results of some research have shown that the first developmental processes that occur at early growth stages depends mainly on water and nitrogen availability (Simane et al., 1993). Efforts to optimize combinations of supplemental irrigation and nitrogen fertilization of wheat were conducted in many parts of the Mediterranean (Harmsen, 1984; Ryan et al., 1991; Oweis et al., 1998). The objectives of this study were to determine the effect of supplemental irrigation and nitrogen fertilization and their interaction on yield, evapotranspiration and water use efficiency of two durum wheat cultivars, Waha and Haurani, widely used by farmers in a typical rainfed Mediterranean environment, like the central plains of Bekaa valley of Lebanon. 2. Materials and methods 2.1. Site description Wheat (Triticum durum L.) seeds were sown under field conditions at Tal Amara research station in the central Bekaa Valley ( N lat., E long. and 905 m above sea level). The experiment was performed during the , and growing seasons from November to June. Tal Amara has a well defined hot and dry season from May to October and cold extending for the remainder of the year. Main average rainfall is 592 mm, with a maximum of 145 mm in January. The soil of the study area is characterized by high clay content and relatively low organic matter content. Field slope is less than 0.1% and total available water within the top 90 cm of soil profile is 170 mm Cultural practices and experimental design Two durum wheat cultivars (Waha and Haurani) were selected for this experiment. Waha is a high-yielding cultivar mainly due to its higher spikes per square meter and grains per spike than other cultivars (García del Moral et al., 2003). Waha has a short straw, straight leaves, short spikes and long grain, while Haurani has longer straw and spike, making it sensitive to lodging. Moreover, Haurani is a drought-tolerant variety. Waha and Haurani were sown using a mechanical plot drill planter with 0.20 m row spacing. The seeding rate was adjusted for a density of seeds m 2, according to the standard practices in the central Bekaa Valley. In this experiment, the effects of delayed sowing dates were avoided by sowing the crops before the end of November (Table 1). Seeds were planted into 8 10 cm furrows with 4 5 cm soil cover above the seeds in a 3200 m 2 -experimental field (64 m NS 50 m WE).

3 agricultural water management 96 (2009) Table 1 Some agronomic and management practices carried out during the experiments. Observation Sowing date (d.o.y.) a 5 November 2000 (310) 10 November 2001 (314) 22 November 2003 (326) Cultivars Waha and Haurani Waha and Haurani Waha and Haurani Seeding depth (cm) Row space (cm) Seeding rate (plants m 2 ) Cultivated area (m 2 ) 3200 (64 m NS 50 m WE) 3200 (64 m NS 50 m WE) 3200 (64 m NS 50 m WE) Effective cultivated area 2400 (48 m NS 50 m WE) 2176 (34 m NS 50 m WE) 2000 (40 m NS 50 m WE) Harvest (d.o.y.) 15 June 2001 (166) 22 June 2002 (173) 27 June 2004 (179) Growing period (days) a Day of year (dates are given in parenthesis). Three water supply treatments (rainfed and two treatments irrigated at 50% and 100% of soil water deficit) were coupled with three nitrogen rates (100, 150 and 200 kg N ha 1 ). Water was applied using a line-source sprinkler at preanthesis and by gravity from anthesis onwards when the soil water content dropped below 50% of the total available water in the upper 90 cm of the soil depth. In and cropping seasons (normal years) irrigation was scheduled at booting-flowering and grain-filling stages. In cropping season (dry year) a drought was recorded early in the growing season and irrigation was supplied at stem-elongation, booting-flowering and grain-filling stages. A flow meter was used to measure the amount of applied irrigation water. Irrigation was applied at 50% (IRR1) and 100% (IRR2) of soil water deficit (SWD), while a rainfed treatment (IRR0) was maintained under no irrigation throughout the growing season. Irrigation dates and depths are given in Table 2. In this experiment, nitrogen deficiency was avoided by applying N fertilization at rates 100 kg N ha 1. Therefore, three N rates were applied at 100 (NR1), 150 (NR2) and 200 kg N ha 1 (NR3). All treatments received at sowing fertilization as NPK ( ) broadcasted mechanically and incorporated into the upper 10-cm of soil layer at a rate of 50 kg N ha 1. Then, ammonium nitrate (NH 4 NO 3, ) was applied in two splits, where an equal amount of 25 kg N ha 1 was given to all treatments at stem elongation, and different amounts of 25, 75 and 125 kg N ha 1 were then given to treatments NR1, NR2 and NR3, respectively, at booting stage. Dates and amounts of N fertilization are given in Table 3. The experimental design was a split plot design. Years were assigned to blocks and cultivars to main plots and the combinations (IRR NR) to sub-plots. Three water supply levels and three nitrogen rates were randomly distributed within the main plots in three replicates each. In total, 54 subplots of 20 m 2 area each (5 m NS 4 m WE), separated by rows 2 m wide, representing all combinations (IRR NR) Crop phenology Regular observations were made of phenology in terms of days after sowing (DAS) and sum of temperature-day (8C), assuming Table 2 Irrigation dates (day of year) and depth (mm) of wheat treatments. Date of irrigation Growth stage Day of year Water depth (mm) IRR0 IRR1 IRR March 2001 Booting April 2001 Anthesis May 2001 Dough stage Total irrigation Total rain (1 November 2000 onwards) Total (rain + irrigation) April 2002 Anthesis May 2002 Dough stage Total irrigation Total rain (1 November 2001 onwards) Total (rain + irrigation) March 2004 Booting April 2004 Anthesis May 2004 Dough stage Total irrigation Total rain (1 November 2003 onwards) Total (rain + irrigation)

4 606 agricultural water management 96 (2009) Table 3 Dates (days after sowing) and amounts (kg N ha S1 ) of nitrogen fertilization of wheat treatments. Date of fertilization Growth stage Days after sowing Fertilizer source N application (kg N ha 1 ) NR1 NR2 NR November Sowing 0 NPK (17%) March Stem elongation 144 NH 4 NO 3 (34%) April Booting 167 NH 4 NO 3 (34%) November Sowing 0 NPK (17%) March Stem elongation 126 NH 4 NO 3 (34%) April Booting 170 NH 4 NO 3 (34%) November Sowing 0 NPK (17%) March Stem elongation 119 NH 4 NO 3 (34%) May Booting 157 NH 4 NO 3 (34%) the base temperature (T base ) of development and growth for wheat crop is equal to 6 8C (Rawson and Gómez MacPherson, 2000). The length of vegetative period was calculated as days from sowing to anthesis (growth stage 65 in the Zadoks scale), where as the grain-filling period was calculated as days from anthesis to physiological maturity (growth stage 91 in the Zadoks scale) (Zadoks et al., 1974). The dates of the most important growth stages of wheat crop were observed when 50% of the plants attained a given stage, i.e., tillering, stem elongation, booting, anthesis, soft-dough stage, and grain stiffening (Doorenbos and Kassam, 1980; Zhang and Oweis, 1999; Beuerlein, 2001). Ambient weather data were daily recorded from the automated weather station of the Institute (AURIA 12E, DEGREANE, France), 50 m from the experimental site. Data were used to compute vapor pressure deficit (VPD) at hourly basis from maximum and minimum air temperatures, assuming relative humidity was 100% at the daily minimum air temperature (Allen et al., 1998). Mean daily VPD was then calculated by averaging hourly VPD and the growing season mean VPD was calculated by dividing the sum of mean daily VPD to the length of the growing season, in days (Chen et al., 2003) Soil water monitoring Soil water content in the plots was measured using a Sentry 200-AP TDR (Time Domain Reflectometry, Sentry 200-AP, 1994). The TDR was calibrated to the soil at Tal Amara over a wide range of soil water content. In all years, access tubes were installed in the central sub-plot of each treatment to measure soil water content in 0.15-m increments for the first 0.3 m, then in 0.30-m increments down to a 1.2 m depth. The TDR was calibrated in the field, and readings were then converted to volumetric soil water content (u v ), using the following calibration equation: Y ¼ 0:0079X þ 1:948 (1) where Y represents u v (in %); X is the TDR measurement; and are the coefficients of the calibration equation. The standard error of the regression model estimation was m 3 m 3, and the coefficient of determination was TDR readings were used to estimate seasonal evapotranspiration (ET) in the plots using a water balance model as the difference between inputs and outputs within the soil profile, assuming drainage (D r ) and runoff (R o ) in the layer 0 90 cm equal to zero: P þ I D r R o ET ðs e S b Þ¼0 (2) where P is precipitation, I is irrigation, D r is drainage, R o is runoff, S e is the soil water content at the end of a time interval, S b is the water content at the beginning of the same time interval. All terms in Eq. (2) are expressed in mm Yield analyses and water use efficiency Harvest date was determined at grain moisture of 15% and ranged from mid to late June. Yield was determined in sampling areas of 1 m 2 from the central rows of each sub-plot, where the number of grains, 1000-grain weight, grain yield and aboveground biomass were measured. The number of grains per square meter was determined by counting the grains from all spikes in the harvest area using a seed counter (Contador, Pfeuffer, Germany). Mean 1000-grain weight was calculated from the weight of five sets of 1000 grains each from the sampling area. Moisture content in the grains was determined using Inframatic 8100 (PerCon, Germany). Harvest index (HI), defined as the ratio of grain weight per mature weight of aboveground parts (Cox and Jolliff, 1986; Moser et al., 2006) was also calculated. Water use efficiency (WUE) of grain produced was calculated as the ratio of grain yield at 0% humidity (in kg ha 1 ) after passing it in the oven for 72 h at 105 8C to crop evapotranspiration (in mm) (Caviglia and Sadras, 2001) Statistical analysis All data were statistically analyzed by ANOVA using the PROC MIXED procedure of SAS (SAS Institute, 1997). Mean separation was performed only when the F-test indicated significant (P < 0.05) differences among the treatments, according to the Fisher s protected LSD test. The interactions IRR NR were also reported and significant differences were analyzed at P < 0.05.

5 agricultural water management 96 (2009) Results and discussion 3.1. Climatic conditions Annual precipitation totaled 450, 580 and 670 mm in the , and cropping years, compared to historical average of 592 mm ( ). However, the rainfall pattern showed monthly variability between the three growing years (Table 4). In and about 95% of seasonal rain occurred between September and February and 5% fell between March and May, where a competition for limiting resources, mainly water, between vegetative and reproductive organs may occur for wheat (Miralles et al., 2000). In , 60% of the rain occurred between September and February, while 40% of the rain fell between March and May, with more frequent rain during the vigorous growth period. Moreover, in 2002 rainfall recorded in March was mm, while it was 15.6 mm in March 2001 and 6.5 mm in March 2004, out of an historical average of 81 mm for this month. The drought recorded in March of years 2001 and 2004 reoccurred in April, where rain was 5.1 mm in 2001 and 12.4 mm in 2004 compared to long term average of 41 mm, but in May rain was below the long average (17 mm) in all three years (Table 4). Weather conditions that prevailed at Tal Amara were generally cooler in than in and cropping seasons. When calculated over the whole year, average air temperature was 1.1 and 0.7 8C warmer in and , respectively, and 0.9 8C cooler in than the annual historical average (14.2 8C). In growing year, temperatures from December to June were higher than the long-term averages (Table 4). Consequently, a drought was recorded early in the season at stem elongation stage. In , lower air temperatures and more frequent rain were observed during the growing season than in Moreover, in March and April, with booting-anthesis stage, average air temperature in all years was C warmer than the long run averages. This has led to less frequent periods during which the soil surface was wet. On the other hand, the relatively warmer weather conditions that prevailed in and cropping years also have increased seasonal and mean daily VPD compared to the growing year. Indeed, total growing season mean daily VPD from November 1st to June 30th totaled 403 mbar in , 368 mbar in and 355 mbar in , thus giving mean daily vapor pressure deficit of 1.66, 1.52 and 1.48 mbar day 1 in , and cropping years, respectively. Moreover, mean daily VPD followed the same general pattern in all growing years (Fig. 1). However, there was greater scatter in the and data than for the data set. Higher mean daily VPD values were observed late in the season (June July) where the highest values of potential evapotranspiration were recorded at Tal Amara and in the central Bekaa Valley in general (Aboukhaled and Sarraf, 1970; Karam et al., 2007) Effects of supplemental irrigation on yield and its components Averaged across years, grain yield of cultivar Waha irrigated at 100% of SWD (IRR2) was 4470 kg ha 1, showing 50 kg ha 1 Table 4 Mean daily temperature and total rainfall prevailed during the experiments, compared to the long-run means ( ). September October November December January February March April May June July August Average/tot Mean air temperature (8C) Rain (mm) Mean air temperature (8C) Rain (mm) Mean air temperature (8C) Rain (mm) Mean air temperature (8C) Rain (mm)

6 608 agricultural water management 96 (2009) Fig. 1 Daily evolution of vapor pressure deficit (VPD) at noon during the three cropping years. more grains than irrigation treatment at 50% of SWD (IRR1) (Table 5). For cultivar Haurani, similarly to Waha, yield was slightly higher with irrigation at 100% of SWD than at 50% of SWD. Mean grain yield of IRR2 treatment was 3900 kg ha 1 and was 120 kg ha 1 higher than IRR1 treatment, while rainfed treatment (IRR0) in cultivar Haurani had the lowest yield (3210 kg ha 1 ). However, the magnitude of the differences between the rainfed and irrigated treatments was observed to be smaller in Haurani than in Waha. Indeed, while for cultivar Waha average increases in grain yield in response to irrigation were 27% and 28% in IRR1 and IRR2, respectively, compared to IRR0, the increases in the same treatments for cultivar Haurani were 18% and 22%. This may reflect a higher aptitude of cultivar Waha to water supply than cultivar Haurani. Supplemental irrigation was observed to increase grain number per unit ground area in all years. For cultivar Waha, the increases in grain number in IRR1 and IRR2 were 20% and 12% (2001), 6% and 10% (2002) and 27% and 19% (2004), respectively, with respect to IRR0. When compared to the rainfed treatment, irrigation across years at 100% of SWD induced less increase (13%) in grain number per unit ground area than irrigation at 50% of SWD (17%). For cultivar Haurani, mean increases in grain number due to irrigation were 15% and 6% (2001), 7% and 21% (2002) in IRR1 and IRR2, respectively. In 2004, grain number was observed to increase by 9% in IRR1 while no increase was recorded in IRR2 treatment. Averaged across years and irrigation treatments, supplemental irrigation in cultivar Haurani induced increases in grain number per unit ground area by 11% and 9% in IRR1 and IRR2, respectively, in comparison with the rainfed treatment (8130 grains m 2 ). In contrast, the 1000-grain weight seemed to be less affected than grain number for both cultivars by supplemental irrigation and nitrogen rate. Indeed, the percentage of increase of 1000-grain weight in cultivar Waha due to irrigation was 12% (2001) in IRR1 treatment, while no increase was observed in IRR2 treatment in comparison with the rainfed treatment (43.97 g). In 2002, the percentages of increase in 1000-grain weight were 10% and 16% in IRR1 and IRR2, respectively, with respect to IRR0 (38.98 g), while in 2004 the increases were 3% and 12%, respectively, in comparison with IRR0 treatment (43.77 g). Across years, supplemental irrigation at 50% of SWD increased 1000-grain weight by 8% against an increase of 13% with supplemental irrigation at 100% of SWD in comparison with the rainfed treatment (42.24 g). For cultivar Haurani, even though the weight of 1000-grain was slightly higher than for cultivar Waha, the magnitude of increase of 1000-grain weight in response to irrigation was observed smaller than in cultivar Waha. Indeed, averages across years and irrigation treatments, the increases in 1000-grain weight were 7% and 11% in IRR1 and IRR2, respectively, in comparison with IRR0 treatment (43.57 g). García del Moral et al. (2003) demonstrated that kernel number per spike has a significant contribution to grain yield, especially under drought conditions, while in cooler environments the compensatory effects among yield components were almost absent, probably because of the relative availability of water and nitrogen during the critical phases of plant development Effect of N fertilization on yield and its components A trend of increasing yields with N rates was observed for both cultivars with the lowest yield occurring at 100 kg N ha 1. For cultivar Waha, even though there were differences among nitrogen treatments, yield at 200 kg N ha 1 (4380 kg ha 1 ) was slightly higher than yield at 150 kg N ha 1 (4230 kg ha 1 ), while nitrogen treatment of 100 kg N ha 1 had the lowest yield (3770 kg ha 1 ). For cultivar Haurani, nitrogen treatment of 100 kg N ha 1 in Haurani had the lowest yield (3010 kg ha 1 ) while treatments of 150 and 200 kg N ha 1 yielded 3730 and 4150 kg ha 1, respectively. However, the differences between the low (NR1) and high (NR3) nitrogen rates were smaller in Waha than Haurani. In all three years, grain number per ground area was influenced by increased N applications. Increasing the rate of N fertilization from 100 (NR1) to 150 (NR2) and 200 kg N ha 1 (NR3) has stimulated the production of additional grains per unit ground area by 12% and 8% (2001), 5% and 16% (2002) and 18% and 24% (2004) in NR2 and NR3, respectively, in comparison with NR1. Averaged across years and N treatments, grain number per unit ground area was observed to increase in cultivar Waha by 11% and 17% in NR2 and NR3,

7 agricultural water management 96 (2009) Table 5 Main effects of supplemental irrigation (IRR) and nitrogen rate (NR) on grain number (GN), grain weight (GW), grain yield (GY), aboveground biomass (AB), and Harvest index (HI) of the two wheat cultivars during the three cropping years. Growing year Treatment Waha Haurani GN (m 2 ) GW (g) GY AB (Mg ha 1 ) (Mg ha 1 ) HI GN (m 2 ) GW (g) GY AB (Mg ha 1 ) (Mg ha 1 ) Irrigation level IRR0 7, , IRR1 8, , IRR2 8, , NR1 7, , NR2 8, , NR3 8, , HI F-tests Irrigation level IRR NR ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** *** * ** * * *** * ** * * Irrigation level IRR0 9, , IRR1 9, , IRR2 9, , NR1 8, , NR2 9, , NR3 10, , F-tests Irrigation level IRR NR ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** *** * ** * * *** * ** * * Irrigation level IRR0 8, , IRR1 10, , IRR2 9, , NR1 8, , NR2 10, , NR3 10, , F-tests Irrigation level IRR NR ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** *** * ** * * *** * ** * * * Significant at P < ** Significant at P < *** Significant at P < Averages across years Averages across years Irrigation level IRR0 8, , IRR1 9, , IRR2 9, , NR1 8, , NR2 9, NR3 9, , respectively, in comparison with NR1 treatment (8341 grains m 2 ). Moreover, N fertilization of 150 and 200 kg N ha 1 has increased significantly (P < 0.01) grain number per unit ground area in cultivar Haurani by 30% and 44% (2001), 24% and 33% (2002), and 16% and 26% (2004) in NR2 and NR3, respectively, compared to N fertilization of 100 kg ha 1. When averaged across years, grain number per unit ground area was found to increase by 24% and 39% (P < 0.01) in NR2 and NR3, respectively, in comparison with NR1 (6575 grains m 2 ). On the contrary, nitrogen fertilization was observed not to have significant effects (P < 0.05) on 1000-grain weight. Indeed, supply of 150 and 200 kg N ha 1 did not induce in cultivar Waha any increase in 1000-grain weight almost in all the three cropping years, except in cropping year where the increases were 3% and 6% in NR2 and NR3, respectively, with respect to NR1 (41.08 g). In 2001 and 2004, the application of N fertilization of 150 and 200 kg ha 1 slightly lowered 1000-grain weight by 2 5%, with respect to N fertilization of 100 kg ha 1 (48.03 g in 2001 and g in 2004). When averaged across years, N fertilization of 150 kg ha 1 did not stimulate any increase in 1000-grain weight with respect to N fertilization of 100 kg ha 1 (46.34 g), while N fertilization of 200 kg ha 1 was observed to reduce this parameter by 2% (45.63 g) with respect to NR1. As a result, cultivar Waha exhibited grain yield increases in response to N application by 12% and 16% (P < 0.05) in NR2 and NR3, respectively, in comparison with NR1 (3770 kg ha 1 ), while for cultivar Haurani the increases were 24% and 38% in NR1 and NR2, respectively, in comparison with NR1 (3010 kg ha 1 ).

8 610 agricultural water management 96 (2009) The relatively small increases in grain yield observed in NR2 and NR3 treatments of cultivar Waha in comparison with N fertilization with 100 kg ha 1 could be attributable to the haying off effect, which can occur when N is applied excessively too, encouraging the crop to produce excessive biomass and use extra water, reducing water availability during the grain-filling process (Van Heraarden et al., 1998). Indeed, across years, the production of aboveground biomass in response to N fertilization was 9.03, 9.77 and Mg ha 1, in NR1, NR2 and NR3, respectively, exhibiting thus higher difference between the low (100 kg N ha 1 ) and high (200 kg N ha 1 ) N application than for grain yield. As a consequence, Harvest index (HI) was observed to decrease in the high N application (0.39), in comparison with the low (0.42) and medium (0.43) N applications (Table 5). These values were very close to those obtained by Przulj and Momčilović (2003) for a series of 20 wheat cultivars in Eastern Europe. For cultivar Haurani, aboveground biomass at harvest was 8.57 Mg ha 1 in NR1, 9.78 Mg ha 1 in NR2 and Mg ha 1 in NR3, giving thus HI values of 0.35 (NR1), 0.38 (NR2) and 0.40 (NR3). Royo et al. (1999) demonstrated that in favorable seasons during the growth period, plants accumulate sufficient amounts of dry matter for various biological functions, and a part of the accumulated dry matter is reserved to grain growth. Moreover, Caviglia and Sadras (2001) showed that translocation process that takes place in wheat plants between anthesis and maturity is influenced by the level of N rate. In this experiment, greater dry matter production at harvest resulted in greater grain yield. Indeed, in all three years, the highest grain yield obtained with the high N rate (200 kg N ha 1 ) was accompanied by the highest aboveground biomass production, with greater values recorded for cultivar Waha. However, in the least favorable years (2001 and 2002), grain yields were lower by 10% and 20%, respectively, in the high N level (NR3) than the low N level (NR1). This could be explained by the loss of a significant amount of dry matter for maintaining a large quantity of vegetative mass (Austin et al., 1977). In growing seasons with unfavorable weather conditions during the vegetative period, plants accumulate dry matter during the grain-filling period for growth of their vegetative parts as well for grain development (Bidinger et al., 1977) Effect of cultivars on grain yield The effect of irrigation was significant for the two wheat cultivars, but the effect of nitrogen was not always significant (Table 5). In all years, irrigation nitrogen interaction was significant, illustrating varietals differences of yield response to irrigation. Comparing separate means among cultivars at each irrigation treatment indicate that cultivar Waha had higher yield than cultivar Haurani at all water supply levels and in all years, but for both wheat cultivars the lowest yield was observed in the rainfed treatment. García del Moral et al. (2003) evaluated grain yield and its components of six ICARDA- CIMMYT wheat genotypes under Mediterranean conditions and they found that Waha had the highest yield due to is higher spikes per square meter and grain number per spike, in comparison with other cultivars. However, there were no significant differences in grain yield among nitrogen treatments of the two wheat cultivars in all three years. Across irrigation treatments, nitrogen application of 100 kg ha 1 resulted in grain yield of 3.77 Mg ha 1 in Waha against 3.01 Mg ha 1 in Haurani, while nitrogen applications at 150 and 200 kg ha 1 resulted in grain yield of 4.23 and 4.38 Mg ha 1 in Waha against 3.73 and 4.15 Mg ha 1 in the same treatments in cultivar Haurani. Moreover, data analyses for the 3-year experiment showed that in spite of significant differences (P < 0.05) in grain number per unit ground area between the two cultivars at all irrigation and nitrogen treatments, 1000-grain weight was found to not differ significantly between the two cultivars, neither in response to supplemental irrigation nor to nitrogen fertilization. Averaged across years and irrigation treatments, grain number per unit ground area in cultivar Waha was 8284, 9697 and 9368 grains m 2 in IRR0, IRR1 and IRR2, respectively, against 7472, 8250 and 8130 grain m 2 in the same treatments in cultivar Haurani. Across years and N treatments, grain number per unit ground area in cultivar Waha was 8341, 9290 and 9718 grains m 2 in NR1, NR1 and NR3, respectively, against 6575, 8125 and 9153 grain m 2 in the same treatments in cultivar Haurani Evapotranspiration and water use efficiency Data reported in Table 6 give evapotranspiration of wheat in all three years according to the water balance (Eq. (2)). Table 6 showed that ET of rainfed wheat ranged from 373 mm in to 433 mm in and to 493 mm in and these values were closely related to the amounts of rainfall registered during the three cropping seasons. In similar experiments, Zhang and Oweis (1999) pointed out that evapotranspiration depend on the seasonal rainfall under rainfed conditions and on the combined amount of water (irrigation and rainfall) under irrigation conditions. When averaged over the whole growing season, daily values of evapotranspiration of rainfed wheat were 1.7 mm day 1 ( ), 1.9 mm day 1 ( ) and 2.2 mm day 1 ( ). Moreover, supplemental irrigation increased markedly ET of wheat plants and the range of measured ET values varied from 450 to 650 mm, following to the level of applied water. Indeed, average daily evapotranspiration of wheat plants irrigated at 50% of soil water deficit varied from 2.0 mm day 1 in to 2.4 mm day 1 in and to 2.6 mm day 1 in , while supplemental irrigation at 100% of soil water deficit increased theses values to 2.6 mm day 1 ( ), 2.9 mm day 1 ( ) and 3.0 mm day 1 ( ). Moreover, results reported in Table 6 showed that under both rainfed and irrigation conditions wheat plants accounted by anthesis for 65 70% of seasonal evapotranspiration, while the remaining 30 35% is accumulated during the grain-filling stages. Irrespective of the rate, N fertilizer increased evapotranspiration in all three years. Averaged across years, the applications of 150 and 200 kg N ha 1 increased ET by and mm, respectively, in comparison with nitrogen application of 100 kg ha 1 (Table 6). Similar results observed by Caviglia and Sadras (2001) showed that N fertilization increased evapotranspiration of wheat plants in spite of reducing evaporation from soil.

9 Table 6 Measured evapotranspiration of wheat treatments according to water balance model (Eq. (2)) during the three cropping years. Growth stage Rain (mm) Irrigation (mm) Rain + irrigation (mm) RD (cm) SW (mm) ET (mm) IRR0 IRR1 IRR2 IRR0 IRR1 IRR2 IRR0 IRR1 IRR2 NR1 NR2 NR November Establishment December Seedling January Tillering February Stem elongation March Booting April Anthesis May Dough stage June Maturity Total November Establishment December Seedling January Tillering February Stem elongation March Booting April Anthesis May Dough stage June Maturity Total November Establishment December Seedling January Tillering February Stem elongation March Booting April Anthesis May Dough stage June Maturity agricultural water management 96 (2009)

10 612 agricultural water management 96 (2009) The growing season mean daily VPD totaled 403 mbar in , 368 mbar in and 355 mbar in After normalizing ET for vapor pressure deficit (ET/VPD) during the growing season, supplemental irrigation at 50% and 100% of soil water deficit had approximately 26 and 52 mm mbar 1 more ET/VPD than rainfed treatment, while N rates of 150 and 200 kg N ha 1 had 4 and 12 mm mbar 1 more ET/VPD than N rate of 100 kg N ha 1. Averaged across years, water use efficiency (WUE) of cultivar Waha under rainfed conditions was 8.1 kg ha 1 mm 1, while irrigated treatments at 50% and 100% of SWD had WUE values of 8.3 and 7.1 kg ha 1 mm 1, respectively. For cultivar Haurani, rainfed treatment resulted in WUE of 7.4 kg ha 1 mm 1, while a slight decrease in WUE was observed in IRR1 (7.1 kg ha 1 mm 1 ) and IRR2 (6.2 kg ha 1 mm 1 ), respectively. Analysis of variance for the combined data showed that WUE in all years was not affected either by supplemental irrigation or nitrogen rate (Table 7). In similar environmental conditions, Oweis et al. (1999) demonstrated that WUE of rainfed wheat was kg ha 1 mm 1 for grain yields < 3Mgha 1 and seasonal rainfall of 330 mm, while for irrigated wheat and grain yield >3Mgha 1 WUE was kg ha 1 mm 1. Values obtained in this experiment were slightly lower than those obtained by Zhang et al. (1998) and Oweis et al. (1999) mainly because seasonal rainfall registered at Tal Amara in the central Bekaa Valley was higher than that registered at the experimental site in northern Syria. In both cases, however, WUE were lower than the maximum value of 15 kg ha 1 mm 1 obtained by Siddique et al. (1990) for wheat in the Mediterranean region. Averaged across irrigation treatments, WUE at grain basis over the three seasons increased from 7.0 kg ha 1 mm 1 for cultivar Waha and 5.6 kg ha 1 mm 1 for cultivar Haurani to 7.6 Table 7 Main effects of supplemental irrigation (IRR) and nitrogen rate (NR) on evapotranspiration (ET), and water use efficiency of the two wheat cultivars during the three cropping years. Cropping year Treatment ET (mm) VPD (mbar) ET/VPD (mm mbar 1 ) Waha Haurani GY (Kg ha 1 ) WUE (kg ha 1 mm 1 ) GY (Kg ha 1 ) WUE (kg ha 1 mm 1 ) Irrigation level IRR IRR IRR NR NR NR F-tests a Irrigation level IRR NR ** ns ** ** ns ** ns * ns * ** ns ** ns * ns * * ns ** ns Irrigation level IRR IRR IRR NR NR NR F-tests Irrigation level IRR NR ** ns ** ** ** ** ** ** ns ** ** ** ** ** * ns * * * ** * Irrigation level IRR IRR IRR NR NR NR F-tests Irrigation level IRR NR ** ns ** ** ** ** ** ** ns ** ** ** ** ** * ns * * * ** * Averages across years Irrigation level IRR IRR IRR NR NR NR a ns, *, **non significant or significant at P < 0.05 or P < 0.01, respectively.

11 agricultural water management 96 (2009) Fig. 2 Relationship between grain yields (GY) and the total evapotranspiration (ET), for Waha and Haurani cultivars wheat over the three cropping years. and 7.5 kg ha 1 mm 1 for the former, and 6.7 and 7.1 kg ha 1 mm 1 for the latter by applying 50 (IRR2) and 100 kg ha 1 (IRR3) more N than N application of 100 kg ha 1. Slightly higher values of WUE occurred in N treatments of cultivar Waha in comparison with cultivar Haurani, and there was no significant increase in WUE between 100 and 150 kg N ha 1 applications. On average, no significant increase in WUE occurred in the three growing years for N application more than 150 kg ha Relationship between grain yield and evapotranspiration The relationship between grain yield (GY) and seasonal evapotranspiration (ET) is presented in Fig. 2. The relationships indicate that for each 10 mm increase in ET, there was a corresponding grain yield increase of 50 kg for Waha and 35 kg for Haurani. This reflects a greater response of Waha to increase in ET than Haurani. Similar linear relationships of grain yield to seasonal ET were established for rainfed and irrigated wheat (Zhang and Oweis, 1999). However, the value of the slope obtained by Zhang and Oweis (1999) was higher (11.6) than those obtained in this experiment. Although the threshold for the first gain of grain yield was 373 mm for Waha and Haurani (estimated from the regression equations), this value was smaller than the threshold for the first grain yield increment for winter wheat of 200 mm obtained by Musick et al. (1994) in the US southern plains and Zhang and Oweis (1999) in northern Syria. Moreover, there was a considerable scatter between grain yield and ET for rainfed data, probably due to the variation in rainfall distribution within the growing season and temperature differences between the three growing seasons. Although the rainfall in the and growing years was less than the long-term average, the crops may have benefited from the favorable inter-season distribution of rainfall. Crops may suffer from water stress during a long dry spell lasting from mid-march to early May in the and growing seasons. However, irrigation late in the two seasons improved soil water status during the grain-filling period, which apparently improved grain yield. In addition to seasonal crop water use, vapor-pressure deficit (reflecting temperature influence) during the grain-filling stage may play an important role in determining grain yield. In this experiment, more precipitation was distributed in the spring in 2002 than in 2001 and 2004 (total of 228 mm in 2002 against 31 mm in 2001 and 22 mm in 2004 between March and June), a higher slope of grain yield vs. ET could be produced in year 2002 than in 2001 and 2004, indicating thus more efficient use of water in this year in comparison with the other two cropping years. 4. Conclusions In all the three growing seasons, the N treatments had little influence on grain yield and irrigation regime being the predominant limiting factor. The significantly higher yields obtained from the irrigated plots in 2002 was attributed to the higher spring rainfall recorded in this year. This indicates that normal distribution of monthly rainfall, especially during the spring months, may affect positively grain yield of wheat rather than total seasonal rainfall. Analyses of variance for grain yield and its components revealed that these characters were affected mostly by supplemental irrigation and nitrogen rate. In fact, the effects of irrigation regime were observed for grain number, grain weight, grain yield but not for Harvest index (HI) and water use efficiency (WUE). effects were of lower magnitude compared with those of irrigation, though they were statistically significant for all traits. On the other hand, grain yield was greater in the cooler than in the warmer year, a consequence of more grains per square meter, heavier grains, and a longer plant cycle. Rainfed conditions caused reductions in grain yield estimated at 25 35% in comparison with irrigation treatments. Grain number per square meter was the yield component most sensitive to drought effects and was