The experiment was carried out at the research field of the Bangabandhu Sheikh Mujibur Rahman Agricultural University (BSMRAU), Salna, Gazipur from November, 2011 to March, 2012 on an upland soil. It is located at the center of Madhupur Tract (24^˚05' North latitude and 90˚16' East longitude) at an elevation of 8.4 m above the sea level. The soil of the experimental field belongs to Salna series of Shallow Red-Brown Terrace soil type (AEZ 28) with silty clay texture in surface and silty clay loam in sub-surface region of the soil [17] [18] . The experimental site is situated in the sub-tropical region characterized by heavy rainfall during the months from July to September and scanty or no rainfall in the rest of the year.

Thirty-five wheat genotypes including most of the popular varieties, some advanced lines and some lines from abroad collected from Wheat Research Centre of Bangladesh Agricultural Research Institute, Nashipur, Dinajpur, Bangladesh were used in the present study.

The land was well prepared by ploughing and cross-ploughing four times with a power tiller. Laddering was done for breaking the clods and leveling the lands. The various stubbles were removed by hand from the experiment field just before preparing the plot. The individual plots were prepared by making ridges (8 - 10 cm high) around the each plot to restrict the lateral run off of fertilizer with irrigation water.

The experiment was carried out in a split-plot design comprising two water regimes in main plot and 35 wheat genotypes were placed randomly in sub-plot with three replications. The water regimes were 1) Control (four irrigations were applied at crown root initiation, booting, anthesis and grain filling stages), and 2) Water deficit stress (irrigation was stopped after crown root initiation stage i.e. 20 days after sowing and the crop was protected from rainfall by rainout shelter). Thirty-five wheat genotypes including most of the popular varieties and some advanced lines collected from Wheat Research Centre (WRC) of Bangladesh Agricultural Research Institute (BARI), Dinajpur were Prodip, Shatabdi, Sourav, Gourab, Sufi, Kanchan, Seri, Pavon, Barkat, Balaka, Aghrani, Akbar, BARI Wheat 26, Protiva, Ananda, Bijoy, BARI Wheat 25, BAW 1151, BAW 1157, BAW 1158, BAW 1159, BAW 1160, BAW 1161, BAW 1162, BAW 1163, BAW 1164, BAW 1165, BAW 1166, BAW 1167, BAW 1168, BAW 1169, BAW 1170, BAW 1171, BAW 1172 and BAW 1173. The unit plot size was consisted of 6 rows each of 2.5 m long having a row to row distance of 20 cm.

Wheat seeds at the rate of 120 kg・ha⁻¹ were sown in line by hand on November 24, 2011. Seeds were placed continuously in lines by making narrow and shallow furrows with iron rod and covered with soil by hand. After sowing of seeds light irrigation was given to ensure uniform germination of seeds. Fertilizers were applied @ 100-60-40-20-1 kg・ha⁻¹ N-P₂O₅-K₂O-S in the form of urea, triple super phosphate, muriate of potash and gypsum, respectively. Two-third of urea and total amount of other fertilizers were applied during final land preparation. The rest amount of urea was top dressed at crown root initiation stage (20 days after sowing) followed by first irrigation. Intercultural operations were done uniformly in each plot to ensure normal growth of the crop. Weeding and mulching were done simultaneously in the experimental plot for two times, firstly at 15 days after sowing (DAS) and secondly, at 35 DAS. Thinning was also done at 14 DAS.

Canopy temperature was measured with an infrared thermometer (Model THI-500, TASCO, Japan) at 12:30 pm on the day. The thermometer was held so that the sensor viewed only the canopy at an oblique angle above the horizontal; this position gave an elliptical canopy target [19] and prevented the thermometer from sensing the soil surface when the leaves were rolled. All canopy temperature measurements were made five places in a plot and in a south facing direction to minimize sun angle effects as suggested by Turner et al., [20] .

Stress tolerance and susceptibility indices including relative performance (RP), mean productivity (MP), geometric mean productivity (GMP), tolerance (TOL), stress susceptibility index (DSI), stress tolerance index (STI), and yield stability index (YSI) for water deficit environment were calculated based on grain yield under water deficit stress and control conditions. Stress tolerance attributes were calculated by the following formulae:

[21] (1)

Mean productivity (MP) and Tolerance (TOL) was calculated according to Gupta et al., [22]

Geometric mean productivity (GMP), stress tolerance index (STI) and stress susceptibility index (SSI) were calculated according to Fernandez [23] for yield of each genotype as follows

where,

Yws = mean yields of a given genotype in water stress (WS) conditions;

Yns = mean yields of a given genotype in non-stress (NS) conditions and;

Xns = mean of all genotypes under non-stress (NS) condition.

where,

Yws = mean yields of a given genotype in WS condition;

Yns = mean yields of a given genotype in NS condition;

DII = Drought intensity index.

The drought intensity index (DII) for each water regime was calculated as

where,

Xws = mean of all genotypes under WS condition;

Xns = mean of all genotypes under NS conditions.

where,

Yws = mean yields of a given genotype in WS condition;

Yns = mean yields of a given genotype in NS condition.

Recorded data were analyzed by statistically using the software MSTATC (Developed by the Department of Crop and Soil Sciences, Michigan State University, East Lansing, MI 48824 USA). Significance between treatments were tested by using least significant difference test (LSD) at p < 0.05 level.

The means and the ranges of relative value of ten characters of the 35 wheat genotypes subjected to water deficits are given in Table 1. The Relative value of plant height, calculated as the ratio of the height of stressed plants and that of non-stressed plants, ranged from 0.90 to 0.99 with an average of 0.95. Relative value of plant height of more than 0.90 was found in seven genotypes. Similarly, the relative value of flag leaf length ranged from 0.71 to 1.00 with an average of 0.87. Malik and Hasan [24] and Khanzada et al., [25] have earlier reported that shoot length of wheat genotypes significantly reduced under water stress condition. Reduction in plant height in wheat due to drought was reported by [26] [27] and [28] . Sangtarash [29] also reported that the flag leaf length of wheat was significantly affected by water stress.

Relative value of number of tillers, spike length, grains spike⁻¹, 1000 grain weight, grain yield and straw yield of thirty-five wheat genotypes are presented in Table 1. Relative value of number of tillers, calculated as the ratio of number of tillers of stressed plants to that of non-stressed plants, ranged from 0.42 to 0.73 with an average of 0.58. These results are in agreement with the findings of Bayoumi et al., [26] and Khakwani et al., [27] who observed that drought caused reduction in number of effective tillers plant⁻¹ by 36.3 and 35 percent, respectively.

Akram [30] observed that number of tillers m⁻² was affected significantly by different water stress treatments. The relative value of spike length and grains spike⁻¹ ranged from 0.84 to 0.96 and 0.75 to 0.91 with an average of 0.90 and 0.95, respectively. This result is in agreement with the findings of Mirbahar et al., [31] who observed that spike length of wheat decreased more in stress susceptible genotypes and less in stress tolerant ones. Khanzada et al., [25] and Qadir et al., [32] have earlier reported that water stress throughout vegetative and reproductive development caused a significant reduction in number of grains spike⁻¹ in wheat. Elhafild [33] demonstrated that drought stress resulted in reduced pollination and reduced the number of grains spike⁻¹. Water deficit also caused remarkable variation in 1000-grain weight and the relative value of 1000-grain weight ranged from 0.79 to 0.92 with an average of 0.84. This result is in agreement with those reported by Khan et al., [34] and Qadir et al., [32] who observed that 1000-grain weight in wheat was reduced mainly due to increasing water stress. The relative value of grain yield also varied significantly among the genotypes and ranged from 0.41 to 0.91 with an average of 0.59. The relative straw yield and harvest index values ranged from 0.50 to 0.83 and 0.85 to 1.14 with a corresponding mean of 0.64 and 0.95, respectively. The reason for lower grain yield under water stress condition was mainly due to reduction in number of effective tillers plant⁻¹, spike length, number of grains spike⁻¹ and 1000-grain weight. The average yield loss in wheat due to drought stress estimated by Bayoumi et al., [26] and Khakwani et al., [27] were 43.20 and 58% - 82%, respectively. Chandler and Singh [35] also reported that both grain yield and biological yield in wheat showed maximum sensitivity to moisture stress.

Thirty-five wheat genotypes were ranked on the basis of their yield reduction due to water deficit over control (Table 2). A hypothetical scale was made to categorize the genotypes in different rank order on the basis of yield reduction. Genotypes were ranked into four groups as tolerant (less than 30% yield reduction), moderately tolerant (30.01% - 40.00% yield reduction), moderately susceptible (40.01% - 50.00% yield reduction) and susceptible (above 50.01% yield reduction). Four genotypes were categorized in tolerant group because they were relatively more productive both under control and water deficit conditions, and exhibited low yield reduction due to water deficit stress. Similarly, six genotypes were found moderately tolerant as they gave lower yield than the tolerant ones but higher yield than the susceptible genotypes. A large number of genotypes tested in this experiment were grouped as moderately susceptible due to higher yield reduction in water deficit condition. Seven genotypes were ranked in susceptible group due to their very low yielding ability and very high yield reduction which ranged from 50.01% to 59.17% in water deficit condition. The average grain yield of wheat genotypes was reduced by 43.2% as reported by Bayoumi et al., [26] and by 50% as reported by Nouri-Ganbalani et al., [28] under the drought stress condition.

Canopy temperature measured at anthesis stage varied significantly among the genotypes due to water deficit presented in Figure 1. The canopy temperature in wheat genotypes ranged from 23.35˚C to 25.65˚C in control and it increased up to 25.50˚C to 29.47˚C under water deficit condition at anthesis stage. The highest increase in canopy temperature (25%) was recorded in the genotype BAW 1166, and it was close to that in BAW 1167 (24%), Seri (23%), Pavon (21%) and BARI Wheat 25 (21%). The lowest increase in canopy temperature was recorded in BARI Wheat 26 (6%) which was followed by those in BAW 1157 (8%), Sourav (8%), BAW 1169 (9%) and Gourab (9%). Canopy temperature in wheat genotypes increased under water deficit condition might have occurred due to increased respiration and decreased transpiration as a result of stomata closure. This result is in agreement with the findings of [36] who reported that leaf temperature in drought stressed wheat plant was higher than in well-watered plants at both vegetative and anthesis stages. They also reported that the plants that showed a lower leaf temperature also showed a higher photosynthetic rate. The lower photosynthetic rate in plants exposed to higher temperature might have resulted from increased respiration [37] . Winter et al., [38] also found significant differences in leaf temperature between drought stressed and irrigated plants, but not among the wheat genotypes.

The different stress tolerance indices used in this experiment are presented in Table 3. From the stress tolerance point of view, the minimum TOL values were recorded in genotypes BARI Wheat 26 (0.50), Sourav (1.22), BAW 1158 (1.25), BAW 1169 (1.57) and BAW 1170 (1.73). This result showed that, the smaller the TOL value, the lower is the grain yield reduction under stress conditions and consequently lower stress sensitivity. The stress susceptibility index (SSI) also followed the trend as like as TOL in the same genotypes. The maximum mean productivity (MP), geometric mean productivity (GMP) and stress tolerance index (STI) were recorded in the genotypes BARI Wheat 26, BAW 1158, BAW 1169 and BAW 1170. Similarly, the maximum yield stability index (YSI) was obtained in the genotypes BARI Wheat 26, BAW 1158, Sourav, BAW 1169 and BAW 1170. Based on MP, GMP and STI values recorded in this experiment, the genotypes BARI Wheat 26, BAW 1158, Sourav, BAW 1169 and BAW 1170 could be considered as relatively water deficit tolerant. An analysis of correlations between the various stress tolerance parameters used in this study provides interesting observations about the information reflected by each of them (Table 4). Yields in the normal irrigation were correlated with yields in the water deficit condition (r = 0.619^**). A significantly negative correlation was found between TOL and grain yield under stress conditions but this correlation is not confirmed under control conditions, suggesting that selection based on TOL will result in more yield reduction under water deficit condition or low yield under control condition. Similar results were reported by Clarke et al., [12] and Rosielle and Hamblin [39] showed that a selection based on TOL failed to identify the best genotypes. The lowest SSI was obtained in the genotype BARI Wheat 26 followed by that of the genotypes BAW1158, 1169, 1170 and Sourav. The stress susceptibility index (SSI) introduced by Fischer and Maurer [40] showed significant slightly negative correlation with yield under stress, and presented a lower positive correlation with yield in control condition. Having in mind the fact

Yp = Control yield, Ys = Water deficit yield, MP = Mean productivity, TOL = Tolerance, GMP = Geometric mean productivity, STI = Stress tolerance index, YSI = Yield stability index, SSI = Stress susceptibility index.

Yp = Control yield, Ys = Water deficit yield, MP = Mean productivity, TOL = Tolerance, GMP = Geometric mean productivity, STI = Stress tolerance index, YSI = Yield stability index, SSI = Stress susceptibility index, ^**Indicates correlation is significant at the 0.01 level, ^*Correlation is significant at the 0.05 level.

that a small value of TOL is desirable, selection for this parameter would tend to favor low yielding genotypes. A larger value of TOL and SSI show relatively more sensitivity to water deficit, thus a smaller values of TOL and SSI are favored. Several authors noticed that selection based on these two indices favors genotypes with low yield under non-stress conditions and high yield under stress conditions [41] . Likewise TOL and SSI, the YSI was positively correlated with the yield under water deficit but not related with the yield under control condition.

The present study indicated that there was a positive and significant correlation among MP, GMP, STI and yield under both water deficit and control conditions and hence they were better predictors than TOL, SSI and YSI (Table 4). The observed relations were in consistence with those reported by Fernandez [18] in mungbean, Farshadfar and Sutka [42] in maize and Golabadi et al., [41] in durum wheat. Rosielle and Hamblin [39] suggested that any tolerance index (TOL) is smaller, less sensitive to drought and the genotype would be desirable. So, to overcome this problem [23] presented stress tolerance index (STI) which had been able to identify high yield genotypes in both stress and non-stress conditions. Higher value of STI for one genotype is indicator of higher drought tolerance and the more potential yield for that genotype. Fernandez [23] and Christian et al., [43] introduced another suitable index namely geometric mean productivity (GMP). This index has more power to separate genotypes than MP, and accordingly [23] made GMP on the basis of STI.