The experiment was carried out at the experimental site of the Technical Institute of Field Crops (ITGC) of Setif 36°09" N; 05°22" E; 981 meters above sea level (masl) during the 2021 to 2022 agricultural season. The experiment was set up on 14^th December 2022, in a randomized complete block design (RCBD) with three replications. Each plot consisted of six lines 5 m long, spaced 0.2 m apart, which made up 6 m² of plot dimension. The plant materials used consisted of ten durum wheat genotypes shown in (Table 1).

Table 1

The soil is calcareous (Calcisol) with a silty clay texture, and organic matter content is 1.4% on the surface. The amount of monthly rainfall, temperature (min, max, mean) are presented in (Table 2).

Table 2

The chlorophyll content index (CCI) of each flag leaf was measured using a digital Chlorophyll Content Meter Model CCM-200 Plus. The relative water content (RWC) was determined at the heading stage according to Pask et al. (2012) method, five fresh leaves were weighted to record fresh mass (FM). The leaves were placed in distilled water for 24 h and weighed to get a turgid mass (TM). Samples were oven-dried at 65 °C for 24 h to record dry mass (DM). Relative water content was calculated as follows the equation 1.

The relative electrolyte leakage (REL%) of leaf tissues was measured using the method developed by Bajji et al. (2001), two leaves were collected, washed with tap water then with distilled water, and cut into 1 cm length segments. The samples were placed in tubes with 10 mL of distilled water and incubated for 24 h at room temperature in the laboratory. Subsequently, the first reading (EC1) was carried out. The final conductivity (EC2) was measured after placing tubes in a boiling water bath at 100 °C for 1 h.

The relative electrolyte leakage (REL%) was calculated as follows the equation 2.

The Canopy temperature (CT) measurements were taken on a sunny day using a portable infrared thermometer (Fluke Corporation. Everett.WA. USA). Readings were taken on sunny days between 11:00 to 14:00 hours.

Flag leaf area (FLA) was determined according to Spagnoletti-Zeuli and Qualset (1990). Five fresh leaves were collected, Leaf length (L) and wide (l) were measured and the area was calculated as follows the equation 3.

At maturity, data were collected on grain yield (GY) (t ha^-1), thousand kernel weight (TKW, g), number of spikes per m² (NSm^-2, spike), and number of grains per spike (NGS, grain).

An analysis of variance was performed for a measured trait at the 5% probability level to test the differences among genotypes, and the linear correlations were done to study the different associations among variables using Costat software (6.400, 1998). A principal component analysis (PCA) was done using the R Core Team.

Numerous agronomic characters that have been widely explored in wheat improvement programs influence grain yield. The data presented in Table 4 shows the genotype effect was significant for grain yield and the number of spikes per m². The highest-yielding genotypes were G3, G2, and G5 (3.52, 3.48, and 2.89 t ha^-1, respectively), with an overall mean of 2.54 t ha^-1. For thousand kernel weight was significantly higher in Boutaleb, G2, and G4 (34, 32.72, and 31.8 g, respectively) though it was significantly lowest in genotype Jupare C 2001 (28.4 g). The number of spikes for the genotypes evaluated was recorded from 320 to 556.66 spikes per m²; the genotype G2 recorded the highest value, while the genotypes G1, G3, G4, and G5 exhibited the lowest values with 346. 66, 358.33, 346.66, and 320 spikes. The mean values for NG/S varied from 21 grains for the introduced genotype Jupare C 2001 to 43 grains for genotype G5, with a mean of 30.48 grains overall for all genotypes. Grain yield is a complex characteristic determined by three components: the number of spikes per area, grain number per spike, and grain weight. MajidiMehr et al. (2024) stated that water stress is a crucial environmental factor that decreases grain yield in bread wheat. Liu et al. (2015) proved that durum wheat genotypes were better adapted to water deficits and were able to maintain their grain numbers in unfavorable environments, which contributed to a smaller decrease in grain yield. Under stressful conditions in arid and semi-arid regions, the major purpose of wheat breeding programs is to develop durum wheat cultivars with high grain yields. It has been reported that wheat yield improvements are principally due to increases in grain weight and grain number per spike (Feng et al. 2018; Hu et al. 2022). Nouri et al. (2011) mentioned that the relative yield performance of genotypes in drought-stressed and favorable conditions helps to select the desirable genotypes. Hence, the development of high-yielding genotypes with acceptable stability and adaptability is a suitable method for improving durum wheat yield in drought conditions (Pour-Aboughadareh et al. 2020).

Table 4

Table 5 shows the correlations between different traits and grain yield. Grain yield had positive and significant correlation with number of grains per spike (0.43*) and a non-significant association with number of spikes NS (0.27^ns). Fellahi et al. (2019) supports this finding, and several researchers agree, suggesting that higher numbers of grains per spike and number of spikelets increase grain yield (Würschum et al. 2018; Wolde et al. 2019). The number of kernels per spike has been suggested as a useful trait for improving wheat grain yield, especially under drought conditions (Bogale and Tesfaye 2016). By contrast, the findings of Iqbal et al. (2017) revealed a non-significant relationship with these traits. Grain yield also had a significant positive correlation with thousand kernel weights (r=0.44 ns), while the study of Ullah et al. (2021) suggested a significant association between grain yield and thousand kernel weights. On the other hand, Boudersa et al. (2021) suggested that all yield components, such as grain weight, number of grains per spike, and biomass, have a considerable contribution to grain yield and that any direct or indirect disturbance affecting any of the yield components inevitably affects the grain yield. Hence, grain yield improvement has been significantly associated with increased thousand-kernel weight; it expresses the grain size and considerably enhances the final yield of wheat (Iqbal et al. 2015). Ullah et al. (2021) conclude that for bread wheat, increased grain weight directly contributed to improved grain yield. Canopy temperature showed significant negative relationships with grain yield; Singh et al. (2022) stated similar findings. Low canopy temperatures in durum wheat lines were associated with higher grain yields (Sohail et al. 2020). Furthermore, Oulmi et al. (2020) noticed that a high canopy temperature causes a decrease in grain yield. Chlorophyll content did not correlate significantly with grain yield while showing a negative correlation with the number of grains per spike (r=-0.38). Similar findings were reported by Mohammadi et al. (2018). The flag leaf area exhibited a significant relationship with the number of grains per spike (r=0.43*) and chlorophyll content (r=-0.37*). This result agrees with the findings of Nor et al. (2015), who found that leaf area showed a significant positive correlation with the number of grains per spike. Wang et al. (2022) also stated similar results and observed that wheat genotypes with a larger flag leaf tend to produce more kernels per spike.

Table 5

The principal component analysis (PCA), one of the methods of multivariate analysis, elucidates among a set of traits which ones are decisive in genotypic differentiation and selection (Ara et al. 2018). The data shown in Table 6 revealed that three components exhibited an eigenvalue near or higher than one. These three PCs accounted for 72.36% of the total variation. Furthermore, an increase in the number of PCs was correlated with a decrease in Eigenvalues. Based on the results shown in Figure 1, PC1 was highly correlated with grain yield and thousand kernel weights. genotypes G2 and G3 were positively correlated with PC1, suggesting that they had high productivity. While genotypes Jupare C 2001, Bouatleb, and G1 were negatively correlated with PC1, which was described as a low-yielding genotype, this result is in accordance with the previous results of Frih et al. (2021) and Guendouz et al. (2021), who stated that grain yield and thousand kernel weights were associated with the two first components. The second component was a physiological axis, with relative electrolyte leakage and relative water content as the major contributors. The genotype G4 was positively connected with this axis and consequently was classified as a drought-tolerant line. By contract varieties, Oued El Bared and Boussallem were negatively correlated with this axis, suggesting that they were the most susceptible to drought conditions. PC3 was negatively connected with flag leaf area; both genotypes G5 and G6 were negatively associated with this axis, which was characterized by a large flag leaf area.

Table 6

Figure 1