The opportunity of using durum wheat landraces to tolerate drought stress: screening morpho-physiological components

Abstract. Local genetic resources could constitute a promising solution to overcome drought stress. Thus, eight (8) durum wheat landraces and one improved variety were assessed for drought tolerance in pots under controlled conditions. Three water treatments were tested: control (100 % of the field capacity (FC)), medium (50 % FC) and severe (25 % FC) stress. The assessment was carried out at the seedling stage to mimic stress during crop set-up. Results showed that increased water stress led to a decrease in biomass and morpho-physiological parameters and an increase in antioxidant enzyme activities. Severe water stress decreased the chlorophyll fluorescence parameters, relative water content (RWC) and water potential of the investigated genotypes by 56.45, 20.58, 50.18 and 139.4 %, respectively. Besides, the phenolic compounds content increased by 169.2 % compared to the control. Catalase and guaiacol peroxidase activities increased 17 days after treatment for most genotypes except Karim and Hmira. A principal component analysis showed that the most contributed drought tolerance traits were chlorophyll fluorescence parameters, RWC and electrolyte conductivity. Unweighted pair group method with arithmetic mean clustering showed that the landraces Aouija, Biskri and Hedhba exhibited a higher adaptive response to drought stress treatments, indicating that water stress-adaptive traits are included in Tunisian landraces germplasm.


Introduction
Durum wheat (Triticum turgidum ssp.Durum) is a crucial cereal crop, constituting the most widely worldwide staple food, providing high carbohydrate, protein content and calories for the human diet (Enghiad et al., 2017;Liu et al., 2017).The actual wheat yield should be increased by 70 % to meet the global population needs expecting to reach 9.7 billion people in 2050 (Alexandratos and Bruinsma, 2012;United Nations, 2022).
Drought constitutes the main limiting abiotic stress factor for wheat productivity.It affects around half of the world's wheat-sown fields of 230 million hectares (Pfeiffer et al., 2005).In those areas, the wheat yield might decrease by 50-90 % (Olivares-Villegas et al., 2007).This situation will become more serious in the future due to climate change scenarios.In fact, predictions of climate change involve rises in average temperatures of 2.6 and 4.8 °C by 2065 and 2100, respectively, associated with more frequent periods of drought and heatwaves (Allan et al., 2021).
Improvement of wheat yields is a complex and challenging objective mainly for cropping in dry areas where water is a limiting factor (Jia et al., 2017).Water availability is critical at all plant growth stages from early plant growth to grain filling (Helaly et al., 2017).In fact, stress events during vegetative growth significantly decreased wheat grain yield (Abid et al., 2018a;Sehgal et al., 2018;Yu et al., 2018;Cohen et al., 2021).Yield losses due to drought stress can reach 25 % under 40 % of water deficit and 32 % in the case of no irrigation at reproductive and grain-filling growth stages (Daryanto et al., 2016(Daryanto et al., , 2017;;Anwaar et al., 2020).Moreover, the effects of water deficiency are dependent on both the severity and duration of stress (Bista et al., 2018;Laxa et al., 2019).
Osmotic adjustment constitutes an important mechanism to lower osmotic potential and thus maintain leaf water status and cellular functions under low soil water content (Abid et al., 2018a, b;Alghory, 2019;Haider et al., 2020;Abeed et al., 2021).Osmotic adjustment is obtained through the synthesis and accumulation of different water-soluble organic compounds such as proline and soluble sugars (Yu et al., 2019;Wang et al., 2019Wang et al., , 2020)).Proline accumulation in plants is related to water stress tolerance (La et al., 2019;Moreno-Galván et al., 2020), and it is the most biochemical trait of drought tolerance (Singh et al., 2014).Lowering osmotic potential can help water uptake (Rani et al., 2022) and maintain cellular turgor (Abid et al., 2018a;Soni et al., 2021).This is particularly important since metabolic processes appear to be more closely linked to cell turgor than leaf water potential (Leuschner et al., 2019;Soni et al., 2021).
To quench the excess of ROS generated in cells, plants exhibit various enzymatic and non-enzymatic antioxidants and increasing the accumulation of osmo-protectants and compatible solutes (Sekmen et al., 2014), which help the plants to reach oxidative homeostasis (Rather et al., 2020a, b;Asgher et al., 2021).H 2 O 2 mitigates abiotic stress through upregulating the activity of antioxidant enzymes, defence proteins and transcription factors (Sun et al., 2016;Nazir et al., 2020;Asgher et al., 2021).
The present study aimed to identify drought-tolerant durum wheat landraces using morpho-physiological traits and enzymatic activities as screening parameters.The drought stress mechanism was also investigated for a better understanding and management of drought stress.

Plant material and growth conditions
The used plant material was constituted of nine durum wheat genotypes (T.turgidum ssp.Durum) including eight landraces (Mahmoudi, Hmira, Swabaa Aljia, Jneh Khotifa, Chili, Biskri, Hedhba and Aouija) from GenBank Tunisia and one improved variety (Karim) from central seed mutual society of Tunisia.
The seeds of the different genotypes were disinfected with 5 % sodium hypochlorite for 10 min, then rinsed three times with distilled water.The seeds were germinated on Wattman filter paper and placed in Petri dishes for 24 h.The germinated seeds were transplanted into plastic pots (2 L) containing a mixture subtract of peat:perlite (2:1; V/V) at the rate of three plants per pot.No additional fertilization has been added since the peat used contains the main nutrients the plant needs (nitrogen (N), phosphorus (P) and potassium (K)) in the following proportions: 14:10:18.A completely randomized experimental design was adopted with three replicates for each genotype.The plants were grown under controlled conditions of 21 ± 2 °C (night/day), a photoperiod of 16/8 h and an average humidity of 60 %.
After 15 days of growth (Z11) (Zadoks et al., 1974), water stress treatments were applied as 100 % of the field capacity (FC) (Control), 50 % FC and 25 % FC.Water irrigation treatments were monitored by weighing pots every 2 days.The stress was applied over a period of 18 days (Z14).

Morpho-physiological traits
A set of morpho-physiological traits were assessed.The aerial part length (APL) was measured by a caliper from the collar to the tip of the longest leaf according to the method described by Morard (1986).The fourth LA was measured using ImageJ software (version 1.46r) (Wayne Rasband, NHI, USA).
Dry matter rate (DM) (%) was recorded for aerial part of the plants.After measuring initial fresh weight (FW), they were placed in an oven at 80 °C for 48 h.The obtained material was weighted as dry weight (DW).The DM (%) was obtained according to the present formula: Canopy temperature was measured by infrared thermometry of the Helect (China) type.Besides, chlorophyll fluorescence components such as initial fluorescence (F 0 ), maximum fluorescence (F m ), variable fluorescence (F v ), maximum fluorescence efficiency (F v/m ) and leaf photosynthetic capacity (F v/o ) were evaluated using an Opti-Science30 (Opti-Sciences Inc., USA).
The relative water content (RWC) (%) was evaluated for the flag leaf according to Turner (1986).Thus, the leaves were cut at the base of the blade, weighed immediately to determine the FW, then placed in Vial tubes filled with distilled water and stored in the dark.After 24 h, the samples were weighed again to obtain the turgor weight.These same samples were then dried in an oven at 55 °C for 48 h before being reweighed to determine the DW.The calculation was made according to the following formula: The plant water potential (Ψ w ) was set up according to the method of Canny (1997) using a pressure chamber (PMS Instrument Co., Corvallis, OR, USA).The electrolyte conductivity (EC) was determined according to the method described by Ben Dkhil and Denden (2012).Indeed, leaf samples were washed with distilled water to remove surfaceadhered electrolytes and cut into discs of uniform size.The flag leaf discs were put in closed test tubes containing 10 mL of deionized water and incubated at room temperature (25 °C) for 24 h.Subsequently, the electrical conductivity of the solution (EC1) was recorded.Samples were then autoclaved at 120 °C for 20 min and the final electrical conductivity (EC2) was obtained after cooling the solution to room Chaouachi et al. -Durum wheat landraces to tolerate drought stress temperature.The EC was calculated as EC1/EC2 and expressed as a percentage.
The chlorophyll Index (CI) was determined using SPAD 502 chlorophyll meter, Konica Minolta brand, Zhejiang.The readings are provided in units called SPAD (Soil Plant Analysis Development).

Biochemical traits
Five biochemical traits were measured as proline content (PC), soluble sugars content (SSC), H 2 O 2 content (H 2 O 2 ), malondialdehyde content (MDA) and phenolic compounds.First, the leaf PC was determined using the method of Bates et al. (1973).The plant sample (100 mg) is treated with 40 % methanol and then heated to 85 °C for 1 h.Then 1 mL of extract is added to a mixture of distilled water, acetic acid and ninhydrin.Second, the SSC in leaves was also measured according to the method of Shields and Burnett (1960).Thus, 100 mg of leaf were ground with 10 mL of ethanol (80 %) and incubated in a water bath for 30 min at 70 °C.After cooling, the extract is centrifuged at 6000 rpm for 10 min.Subsequently, 50 μL of the supernatant are added to 5 mL of anthrone (2 %) and 2.45 mL of ethanol (80 %).Absorbance was measured at 640 nm.The H 2 O 2 content was determined by spectrophotometry according to Sergiev et al. (1997).Indeed, 100 mg of fresh plant material is ground in 1 mL of a 0.1 % TCA (trichloroacetic acid) solution.The ground material is then centrifuged at 12 000 rpm for 15 min at 4 °C.Then, a volume of 0.5 mL of the supernatant is incubated in the presence of 1 mL KI (1 M) and added to 0.5 mL of potassium phosphate buffer (KH 2 PO 4 /K 2 HPO 4 ; 10 mM; pH 7).The mixture is then homogenized and incubated for 5 min.The optical density is determined at 390 nm.Under conditions of oxidative stress, plants exhibit a chain of lipid peroxidation ultimately leading to the production of MDA, a reactive aldehyde capable of reacting with 2-thiobarbituric acid.MDA extraction and assay were performed according to the method of Ksouri et al. (2007).The last biochemical measure is the total phenolic compounds or polyphenols (Ph.C), which were determined using the Folin-Ciocalteu reagent according to the method of Singleton et al. (1999).

Antioxidant enzyme activity
Antioxidant enzyme activities were measured as guaiacol peroxidases (GPX) and catalase activity (CAT).The activity of GPX was measured according to the method reported by Egley et al. (1983).In fact, 500 mg of fresh plant material were ground in a mortar containing a volume of 5 mL of a phosphate buffer (50 mM; pH = 6.5).The ground material obtained is then centrifuged at 12 000 rpm for 20 min at 4 °C.The enzymatic activities as well as the protein contents are assayed in this extract.The measurement was carried out in a volume of 3 mL containing the phosphate buffer (50 mM; pH = 6.5), guaiacol (50 mM), H 2 O 2 (2 %) and 100 μL of enzymatic extract.Enzyme activity was monitored by spectrophotometry at 470 nm.
The CAT activity is determined according to the method described by Beers and Sizer (1952).Hundred milligrams of fresh plant material were ground in a mortar with 1.5 mL of phosphate extraction buffer (50 mM; pH = 7).The ground material is then centrifuged at 12 000 rpm for 5 min at 4 °C.The enzymatic activities as well as the protein contents are assayed in this extract.For a final volume of 1 mL, a reaction mixture consisting of 50 µL of supernatant and 950 µL of H 2 O 2 reagent (20 mM) was prepared.Absorbance is monitored at 240 nm.

Drought Stress Indexes
Drought Susceptibility Index (DSI) and Stress Tolerance Index (TSI) were calculated based on DM (%) using the following equations (Fischer and Maurer, 1978;Golbashy et al., 2010;Grzesiak et al., 2012): where C, D are dry matter of above ground part of the plant (DM) in control (C) and drought (D) treatments, respectively.xC, xD are average values for all examined genotypes of dry mater of aboveground part of the plant (DM) in control (C) and drought (D) treatments, respectively.

Statistical analysis of the data
For statistical analysis, R v.3.6.1 was used (Core Development Team R, 2020).The effect of drought stress treatments on the studied genotypes was determined by a multivariate analysis of variance (MANOVA) test.A post hoc analysis was set up using Tukey's multiple comparison test.All measured variables were used for the principal component analysis (PCA) which was carried out by using FactoMineR (Lê et al., 2008) and Factoextra (Kassambara and Mundt, 2017) libraries.To normalize data and to study the complex genotypic variability under water stress, the results were presented as Log 2 (Trait Fold Change (TFC)) = Log 2 C − Log 2 D, where C represents control value (100 % FC), and D refers to drought stress treatments (50 or 25 % FC).
In addition, an agglomerative hierarchical clustering, using unweighted pair group method with arithmetic mean (UPGMA) method, was performed to classify the genotypes according to their tolerance.

Effect of drought stress on morpho-physiological traits and biochemical traits
In this study, 16 morpho-physiological and biochemical traits were used to screen drought stress tolerance of durum wheat genotypes.All those parameters were under the significant effect of genotype × water stress treatments (G × T) [see Supporting Information-Table S1].
The APL was significantly reduced by water stress for most of the studied genotypes, except Hedhba, Jneh Khotifa and Hmira [see Supporting Information-Table S1].The highest rate of decrease was observed under 25 % FC for the genotype Chili (30.9 %) compared to the control, while the Jneh Khotifa was the least impacted.
Data showed that leaf temperature (LT) was not affected by water stress at early growth stage as for Aouija, Hedhba, Jneh Khotifa, Chili and Karim [see Supporting Information-Table S1].
Drought stress led to an accumulation of phenolic compounds that was proportional to the severity of water stress.The highest increase of phenolic compounds was noted for Mahmoudi under 25 % FC with a 5-fold increase compared to control [see Supporting Information-Table S1].
Soluble sugars changes were significantly (P < 0.001) affected by treatments [see Supporting Information-Table S1].Maximum SSC increase under severe stressed conditions (25 % FC) was observed in Mahmoudi, Swabaa Aljia and Jneh khotiffa with a sugar content increase about 15-fold [see Supporting Information-Table S1].These results were in accordance with Chaudhari et al. (2017) who observed that the soluble carbohydrate concentration in well-watered wheat plants was lower than that of stressed plants.
PC was significant (P < 0.001) under the interaction between genotypes and treatments [see Supporting Information-Table S1].The proline accumulation increased two times under high stress treatment (25 % FC).The maximum increase in PC was noticed in Mahmoudi content under 25 % FC [see Supporting Information-Table S1].The results showed that all tested genotypes tend to decrease their RWC under 50 % FC mainly for Aouija followed by Jneh Khotifa and Chili (Fig. 1A).Under 25 % FC, the RWC decreased significantly compared to the control plants for all the genotypes.Moreover, the lowest decrease was observed in Aouija, Biskri and Hedhba.The decrease in RWC was not significant between 50 and 25 % FC for Biskri and Hedhba (Fig. 1A).Likewise, the DM rate was greatly reduced by the drought stress (Fig. 1B).Under 50 % FC, the genotype Karim was the most sensitive to drought FC with 28.4 % decrease.Under 25 % FC, the genotype Mahmoudi showed a decrease of 55 % compared to control.In addition, a remarkable increase was seen in the H 2 O 2 content under drought stress (Fig. 1C).Under 50 % FC, the lowest rise was observed for Biskri (80 %) and Hmira (81.81 %).While under 25 % FC, Hedhba showed the lowest increase (91.67 %) in H 2 O 2 content (Fig. 1C).
The F v/o and the F v/m decreased significantly (P < 0.001) under drought stress [see Supporting Information-Table S1].The F v/o and F v/m ratios under drought stress decreased by 34.6 and 9 % under 50 % FC and by 56.45 and 20.58 % under 25 % FC, respectively (Fig. 1D and E).The lowest decrease of F v/m was noted in Hedhba (5.8 %) under 50 % FC and in Biskri (8 %) under 25 % FC compared to the control treatment (Fig. 1D and E).
Drought stress drastically affected the membrane stability of the tested genotypes increasing by the way of the EC.The highest rise in EC was noted for Hmira and Karim under both drought stress treatments (Fig. 1G).
Besides, drought stress conditions stimulated MDA accumulation.Maximum increase was observed for Aouija and Chili under both levels of drought stress (Fig. 1H).

Effect of drought stress on antioxidant enzyme activities
Data showed that GPX activity was higher when water stress levels increased from 50 to 25 % FC for most tested genotypes despite sampling dates (17 and 22 days after treatment (DAT)) (Fig. 2).
Under 50 % FC, the average GPX activity was higher at 22 DAT.While at 17 DAT, the highest increase was observed for Chili, Hedhba and Jneh Khotifa.Moreover, the lowest increase in GPX activity was observed for Hmira (85.25 %) on the 17 DAT.On 22 DAT, the GPX activity increased for all the genotypes.However, the most important increase was noted for Aouija, Chili and Swabaa Aljiya.The less increase was observed for Hmira (101.88 %) [see Supporting Information-Table S1].
sUnder 25 % FC, all the studied genotypes increase their GPX activity on the 17 DAT relative to the control plants.The highest increase was observed in Biskri and Jneh Khotifa.Moreover, the lowest increase in GPX activity, on the 17 DAT, was noted for both.The lowest GPX activity was noted for Hmira with an increase rate equal to 167.32 % compared to the control treatment [see Supporting Information-Table S1].
The results of this study showed that all the genotypes tend to increase their GPX activity on the 22 DAT compared to GPX activity on the 17 DAT.The rate of increase in GPX activity for most of the studied genotypes between the two sampling dates (17 DAT and 22 DAT) was higher under 25 % FC than under 50 % FC treatment (Fig. 2).
The results of CAT measured on nine genotypes under 50 % FC on the 17 DAT show that the CAT activity increased compared to the control treatment.The highest increase was measured in Aouija (Fig. 2C).
Under 50 % FC, the CAT activity increased on the 22 DAT compared to the 17 DAT in most of the tested genotypes.The highest increase was observed in Jneh Khotifa, Aouija and Hedhba (Fig. 2C).Moreover, the less increase was observed on Hmira with a rate of increase equal to 31.58 % compared to the control treatment [see Supporting Information-Table S1].
On the other hand, under severe drought stress (25 % FC), the CAT activity increases on the 22 DAT compared to the CAT activity on 17 DAT (Fig. 2D).Indeed, on the 17 DAT, the increase in CAT activity was observed on Jneh Khotifa, Aouija and Hedhba.And the Log 2 (FC CAT ) between 100 and 25 % FC values on the 17 DAT was equal, respectively, to −5.14, −6.07, and −7.9 (Fig. 2D).

Relative contribution of measured parameters to plant biosynthesis under stress conditions
To evaluate the relative contribution of the measured parameters to wheat drought stress tolerance, estimated by DM, we proceeded to a multiple regression analysis using the stepby-step procedure (Stepwise).The dependent variable was chosen as DM at different water regimes (100, 50 and 25 % FC) for all the tested genotypes.The independent variables were those chosen to best fit the model (Table 1).
For DM at 100 % FC, the best model (R 2 = 0.84) was obtained by the independent variables as CAT 17 DAT , SSC, MDA and RWC.Under moderate water stress (50 % FC), DM was best determined (R 2 = 0.94) by (CI), SSC, CAT 17 DAT , H 2 O 2 , CAT 22 DAT and GPX 17 DAT .Under severe water stress (25 % FC), the independent variables of LA, SSC and CI were among best model (R 2 = 0.81).The results of linear regression showed that SSC and CI are the most determining factors of DM under both moderate and severe water stress (50 and 25 % FC) (Table 1).
Under 50 % FC, the most important DSI based on the DM was observed for Karim and Chili, and the minimum was observed for Aouija (Fig. 3).Under 25 % FC, the most important DSI index was observed for Mahmoudi and Hmira, while lowest DSI was observed for Aouija.Results showed that Aouija is the most stable genotype based on DSI index with values below 1 under 50 % FC as well as under 25 % FC stress (Fig. 3).In addition, the TSI highest values were observed for Aouija, Hedhba, Biskri and Jneh Khotifa (Fig. 3).

Genotypic clustering of durum wheat under water stress
The PCA explained 64 % of the cumulative variance of measured parameters for durum wheat under water stress.The first principal component accounted for 52.3 % of the total variability; while the second component accounted for only 11.7 % (Fig. 4).
The traits that most contributed to PCA were chlorophyll fluorescence parameters, RWC and EC (Fig. 4A); while antioxidant enzymes (CAT and GPX), sugars and proline were the less PCA contribute traits (Fig. 4A).
In our study, the biplot derived from the PCA allowed the identification of four durum wheat genotype groups (Fig. 4B).
Indeed, the first and the second groups are formed by the different tested genotypes under 100 and 50 % FC.The third group is the genotypes Aouija, Hedhba and Biskri under 25 % FC.While the fourth group is constituted by the rest of tested genotypes under 25 % FC.PCA showed an ellipse of 95 % confidence of the genotypes Aouija, Hedhba and Biskri narrowing in the same direction (Fig. 4B).
The UPGMA showed that the tested genotypes were clustered in three groups depending on the irrigation levels.However, genotypes under 25 % FC were clustered in two groups: one with the most drought-tolerant genotypes, clustered with the genotypes under 100 % FC, and the second cluster with the sensitive genotypes.The dendrogram clustering showed that drought-sensitive genotypes, i.e.Karim and Hmira were located together.Similar trend was observed for drought resistant genotypes, i.e.Aouija, Biskri and Hedhba (Fig. 5).

Discussion
In the next years, climate change will continue to threaten the world's food supply.In these circumstances, the development of new high-yielding varieties adapted to diverse environments is crucial (Mammadov et al., 2018).To improve drought tolerance, plant breeders must first select the potential germplasm that contains large genotypic variability for drought tolerance (Baenziger, 2016).
For this purpose, 16 characters were used to assess the variability of 9 durum wheat genotypes under water stress conditions applied at the early juvenile growth stage.In fact, the early growth stages affect the grain yield because the photosynthetic reserves accumulated until flowering provided approximately 57 % of the final grain yield (Gallagher et al., 1976).
The MANOVA revealed significant differences among tested genotypes as well as for the genotype × treatment interactions for all assessed traits.The same findings were reported by Khodadadi et al. (2011) and Mwadzingeni et al. (2016).The use of PCA as a key solution for integrating all measured parameters to screen for drought stress has been successfully used in many species, including common bean (Ali et al., 2019), bread wheat (Ghasemi and Farshadfar, 2015) and the durum wheat (Kacem et al., 2017;Quagliata et al., 2023).
PCA allowed us to identify the traits that contributed most to the total variance.These traits were RWC, DM, H 2 O 2 content, chlorophyll fluorescence parameters (F v/m and F v/o ), Ψ w , EC and MDA.A significant association between CI, chlorophyll fluorescence parameters (F v/m and F v/o ), RWC, APL, DM and water potential parameter was observed as previously in Grzesiak et al. (2019).Furthermore, a strong and positive relationship among antioxidant system, enzymatic activities (CAT and GPX), non-enzymatic content (phenolic compounds) and EC illustrates production of ROS in response to changes in membrane stability, which triggers production of antioxidants (Sheoran et al., 2015;Caverzan et al., 2016).In  addition, PCA showed that landraces Aouija, Biskri, Hedhba and Jneh Khotifa were the most tolerant to drought stress genotypes, whereas the variety Karim was the most sensitive.Like related wild species (Maxted et al., 2008), landraces (Zeven, 1998;Ouaja et al., 2021) are considered a valuable source of germplasms for the breeding of drought-tolerant genotypes (Al Khateeb et al., 2017;Placido et al., 2020).These landraces exhibit a broader genetic background that results from both natural and farmer selection (Belay et al., 1995).Thus, using landraces such as Aouija, Biskri, Hedhba and Jneh Khotifa could be a way to improve wheat varieties exhibiting good nutritional characteristics and yield potential, notably under drought stress conditions (Al Khateeb et al., 2017).

Effect of drought stress on plant water status and photosynthesis activity
RWC is one of the most important physiological traits to check the cell hydration and it's probably the most meaningful measure of plant water status under water deficit (Silva et al., 2007;Qayyum et al., 2021).
Under water stress conditions, the RWC decreases independently of wheat's growth stage.In this sense, Bentahar and Ykhlef (2017) observed a significant reduction in RWC only under severe stress (20 % FC) in potted durum wheat under semi-controlled conditions.
The decrease in RWC of sensitive genotype seedlings leaves could be attributed to a higher transpiration rate, due to the rapid loss of water before the stomata close, than water uptake compared to tolerant genotypes (Chaves et al., 2009;Flexas et al., 2009;Osakabe et al., 2014;Mondini and Pagnotta, 2015).Under rising water shortages, plant species lose water mainly by transpiration and then they tend to control their stomatal closure.In addition, many studies showed that stomatal closure is a genotypic depending component (Hajiboland et al., 2015;Ahmad et al., 2018;El et al., 2019).
Our results were similar to the findings of Pour-Aboughadareh et al. (2017) and Reza Beighi and Bijanzadeh (2020), indicating that RWC can be used as an important secondary trait for selecting wheat genotypes under drought conditions.Indeed, plants maintaining RWC under drought conditions are meant to be resistant (Ali et al., 2019;Karimpour, 2019;Sattar et al., 2019).Our data showed that Aouija, Jneh Khotifa and Chili showed the ability to maintain the highest RWC under drought stress conditions.
Chlorophyll Index decreased significantly compared to unstressed plants for most of the tested genotypes.Reduction in CI due to drought stress has been documented by many studies (Alachew et al., 2016;Chen et al., 2016;Khayatnezhad and Gholamin, 2021).
Besides, chlorophyll fluorescence analysis revealed that drought results in decreased electron transfer to the PSII reaction centre due to variations in energy absorption, trapping, electron transport and dissipation per section, resulting in a reduced photosynthetic efficiency of PSII (Stirbet et al., 2018;Khatri and Rathore, 2019).According to ŽIVČÁK et al. (2014), water stress progressively decreases PSII electron transport and increases non-photochemical quenching in wheat leaves that supports the role of alternative sink electrons (of PSII or PSI) and the flow of cyclic electrons for photo-protection of PSII and PSI, which also generate the ATP needed to cope with water stress conditions.Furthermore, F v/m and F v/o are considered important parameters for assessing the integrity of the internal mechanism within a leaf during the photosynthetic activities and are considered as accurate methods for screening plant tolerance to drought stress (Sharma et al., 2015;Lee et al., 2016).Like other physiological traits, these parameters are highly affected by drought stress treatments [see Supporting Information-Table S1].A reduction in F v/m by water deficit suggests a possible inhibition of PSII photochemistry that may be due to inadequate energy relocation from the light-harvesting chlorophyll complex to the reaction centre (Ahmed et al., 2013).The minimum decrease of F v/o and F v/m was observed for Biskri and Aouija indicating that a higher defensive ability for PSII is a key resistance mechanism for wheat landraces, which agrees with previous studies for drought-treated durum wheat accessions (Hairat and Khurana, 2015).Those genotypes (Biskri and Aouija) are more tolerant to drought stress and could be an opportunity for plant breeders as potential sources of drought resistance.

Osmotic adjustment
Dehydration of plant tissues impairs various physiological processes, especially the changes in leaf water potential (Grzesiak et al., 2019).The same results were recorded in our assay, where Ψ w decrease was correlated with drought stress treatments.
Osmotic adjustment allows the plant to maintain turgor pressure at low water potential, which is important for maintaining metabolic functions (Izanloo et al., 2008;Bhutto et al., 2023).
We observed an increase in leaf's PC under drought conditions as previously reported (Nowsherwan et al., 2018;Ali et al., 2019).Proline accumulation is involved in maintaining plant growth under drought conditions through the osmoprotective and developmental functions of amino acids (Batista-Silva et al., 2019).The results showed that the genotypes Mahmoudi and Karim have the most important proline accumulation.High capacity of proline accumulation under water deficit is commonly associated with high drought stress tolerance (Bilal et al., 2015;Quagliata et al., 2023).However, the relationship between proline accumulation and enhanced or reduced tolerance to stress is still controversial in the literature, having been reported correlation of higher proline concentrations with both higher and lower stress tolerance (Arteaga et al., 2020).This amino acid is also involved in scavenging ROS, thus maintaining protein stability and pH homeostasis in the cytoplasm (Bandurska et al., 2017).
Under drought stress conditions, plants accumulate other osmotically active substances as sugars (Sami et al., 2016;Basu et al., 2016;Hussein et al., 2022;Chunyan et al., 2022).Our data showed a substantial increase in this solute correlated with water stress levels as previously reported (Sedaghat et al., 2020;Adrees et al., 2020).The increase in SSC was higher under 25 % FC than 50 % FC treatments and this trend was more pronounced in Jneh Khotifa than in the other tested genotypes.In fact, sensitive plants showed less of an increase in SSC than did tolerant plants (Abid et al., 2018a).

Cell membrane stability
Our data showed that electrolyte leakage progressively amplified with increasing drought severity.These results are in Chaouachi et al. -Durum wheat landraces to tolerate drought stress agreement with a set of previous studies (Chaudhari et al., 2017).Similar to our results, Hairat and Khurana (2015) and Pour-Aboughadareh et al. (2017) reported that different species have different effect on drought stress as some are less affected than others, while depending on their thylakoid membrane stability under drought stress.A strong correlation between cell membrane stability with growth and field performance of wheat seedlings has been previously reported (Bajji et al., 2001), which resembles our results.

Oxidative homeostasis
Photosynthetic activity is reduced under drought stress leading to ROS accumulation as a result of photosynthetic electron transport reactions under the situations of saturated electron flow (Reddy et al., 2004).Drought is among an excessive accumulation of H 2 O 2 and of MDA, a marker to measure the amplitude of oxidative damage in the stressed conditions (Porcel and Ruiz-Lozano, 2004).
Higher H 2 O 2 production under drought stress treatments might be associated with photosynthetic damage.The highly reactive H 2 O 2 damages the photosynthetic system, oxidizes proteins, lipids, nucleic acids and carbohydrates, and damages cell membranes (Reddy et al., 2004;Sabra et al., 2012).Our results showed that higher MDA concentrations in the drought-stressed plants were associated with higher H 2 O 2 content, especially under severe drought-stress conditions.Indeed, the maximum values of H 2 O 2 and MDA were found under limited water levels (50 and 25 % FC) in the genotype Chili (Fig. 1C).
Changes in antioxidant enzyme activities in response to water deficit are meant to maintain the fine balance between the generation and detoxification of ROSs at the intracellular level (Ahmadi et al., 2018b;Zhao et al., 2020).Tolerant genotypes are characterized by high antioxidant capacity to alleviate oxidative damage and maintain structural integrity of cell components (Ahmadi et al., 2018a).Plants have a developed antioxidative system that uses the enzymes superoxide dismutase, ascorbate peroxidase, catalase, GPX, peroxiredoxins, monodehydroascorbate reductase, dehydroascorbate reductase and glutathione reductase to reduce and eliminate ROS-mediated oxidative damage (Soares et al., 2019).However, changes in antioxidant enzyme activities under drought stress are dependent on plant species, cultivar, stress intensity and duration (Chmielewska et al., 2016;Soni et al., 2017).Indeed, the rate of increase in GPX activity for most of the studied genotypes between the two sampling dates (17 DAT and 22 DAT) was higher under 25 % FC than under 50 % FC treatment.
During the early days of drought, plants showed a rapid increase in CAT and GPX activities as it scavenged H 2 O 2 , produced after the dismutation of O 2 − by superoxide dismutase (Movludi et al., 2014).It is well-evident that the generation of H 2 O 2 is the primary signal for the initiation of antioxidant enzyme activities (Abid et al., 2018a) and therefore is an essential component of the ROS-mitigation system.
Higher CAT and GPX activities with lower H 2 O 2 accumulation in Biskri and Hedhba plants indicate the existence of an improved redox defense potential to drought stress (Fig. 1).
During the early days of stress, the activities of both CAT and GPX increased 17 DAT, above all, under 25 % FC, for most of the tested genotypes except Karim and Hmira.Those two genotypes have the lowest CAT and GPX activities (Fig. 2).These contexts may give the explanation for the lower capability of Karim and Hmira plants to build and maintain a high activity of antioxidants under drought stress as compared to Biskri and Hedhba plants.In contrast, the plants under Biskri and Hedhba displayed enhanced enzymatic activities of CAT and GPX and consequently lower H 2 O 2 production as compared to Hmira and Karim plants.Moreover, GPX activity was more important than CAT activity when the drought stress treatments reached 22 DAT for most of the tested genotypes.CAT and GPX of landraces wheat genotypes showed greater activities in response to water stress than cultivated wheats at seedling stage (Ahmadi et al., 2018b).

Durum wheat drought stress selection criteria
A several studies have recently identified biomass and antioxidant system as proxies for selection of wheat genotypes at the young seedling stage under drought (Mickky and Aldesuquy, 2017;Ahmadi et al., 2018b).During water deficit, plants reduce their vegetative growth by reducing cell proliferation and expansion to maintain reserves and thus sustain survival as growth adjustment mechanism (Nahar et al., 2015).
Among the various stress tolerance indices, TSI has been suggested as a suitable selection criterion during early growth because this parameter enables us to identify individuals with high performance and stress-tolerant potential under adverse conditions (Pour-Aboughadareh et al., 2017).Abid et al. (2018b) and Aberkane et al. (2021) used the Drought-Tolerant Index to investigate drought tolerance in durum wheat genotypes.The genotypes Aouija, Hedhba, Biskri and Jneh Khotifa had the highest TSI.Similarly, high TSI is associated with a limited decline in RWC, DM and chlorophyll content in response to water deficit (Pour-Aboughadareh et al., 2017).
Drought Susceptibility Index is another selection index for contrasting genotypes under drought stress conditions.In fact, the genotypes, which carry the lowest DSI values are marked as drought-tolerant wheat (Mwadzingeni et al., 2016).In our case, the lowest DSI value was observed under severe stress in Aouija.
Cluster analysis with all measured parameters and calculated indexes under three levels of irrigation (100, 50 and 25 % FC) showed that the genotypes Aouija, Biskri and Hedhba could be used in breeding programmes as potential tolerant genotypes.

Conclusions
This study showed that drought stress had negative effects on most durum wheat genotypes.Data revealed that RWC, CI, chlorophyll fluorescence parameters, water potential, EC, H 2 O 2 , MDA and DM are involved in drought tolerance and can be used as selective traits for screening durum wheat genotypes at seedling stage.Based on DSI and TSI under drought stress conditions, the landraces Aouija, Biskri and Hedhba showed drought stress tolerance aptitude.However, Karim and Hmira were drought sensitive genotypes.
Our results identified several tolerant genotypes that responded well to water stress and could be candidates for further studies of molecular mechanisms underlying other physiological and biochemical changes and subsequent yield improvements under drought stress.In addition, the identified landrace durum wheats represent valuable genetic resources that can be used for isolation of new drought resistanceassociated genes or alleles to develop new wheat varieties that produce high yields in drought-prone areas.
In terms of perspective, it is interesting to complete this study with the tools of molecular genetics.Indeed, the search for molecular markers, the study of the genetic determinism and the heritability of identified characters could lead to the screening of durum wheat genotypes tolerant to drought stress, as good indicators of water stress tolerance are needed to facilitate the use of these traits in breeding programs.Moreover, these results need to be confirmed at later stages of development.
The variation of Log 2 (TFC) of eight traits of the nine studied genotypes under three irrigation treatments, are presented in Fig. 1.Positive Log 2 (TFC) (C−D) values indicate a decrease in the trait value under drought stress (D) compared to control plants (C).Negative values of Log 2 (TFC) (C−D) indicate an increase in the trait value.

Figure 2 .
Figure 2. Variability of GPX and CAT activity content measured at 17th and 22nd days after drought of 50 and 25 % FC of nine durum wheat genotypes.

Figure 3 .
Figure 3. Drought tolerance indexes of nine genotypes under three treatments of irrigation, 100, 50 and 25 % FC based on their DM.The light grey colouration represents DSI or TSI determined for each genotype under 50 % FC stress.The dark grey colouring DSI or TSI calculated for each genotype under 25 % FC stress.

Figure 4 .
Figure 4. Principal component analysis of all the measured variables for the nine genotypes and three studied treatments (100, 50 and 25 % FC).(A) Visualization of the contribution of each variable to the variance of the model and the direction of the variation.(B) PCA visualization of individuals per genotype and per treatment.Ellipses with 95 % of confidence were represented to allow clustering among genotypes and treatments.

Figure 5 .
Figure 5. Dendrogram of the mean of all measured parameters and tested genotypes under three water stress treatments.