This experiment was carried out in the greenhouse of Fruit Plant Nursery Area, Institute of Horticultural Sciences (IHS), University of Agriculture, Faisalabad (UAF), Pakistan. One and a half years old 30 uniform and healthy plants of two white flesh guava cultivars viz. “Gola” (round-shaped) and “Surahi” (pear-shaped) were selected in compliance with international, national, and/or institutional guidelines.

The plants were grown in plastic containers having 6 kg of soil material comprised of farmyard manure, sand, and silt (1:1:1). The selected plants of chosen guava cultivars were subjected to three field capacity levels, viz. T_o= 100% field capacity (control), T₁ = 75% field capacity, and T₂ = 50% field capacity, with one plant per pot in triplicate, in a Randomized Complete Block Design (RCBD). The plants were subjected to drought stress for 120 days during the summer season (April–July) and data were recorded. During the study period, the average climatic factors were measured as atmospheric temperature (27.2–34.4°C), relative humidity (28.8–59.6%), and day length (8.2–10.4 h). To maintain three-field capacity levels, irrigation intervals ranged from 6 to 25 days according to the environment, temperature, and evapotranspiration rate, and the exact amount of irrigation water was calculated by using the following formula:

The saturation percentage (SP) was calculated as;

The saturation percentage of growing media was 24%; hence, its half, that is, 12% was considered as 100% field capacity. The quantity of water required for 100, 75, and 50% field capacity levels in pots containing 6 kg of soil media were calculated as follows:

After 120 days of drought stress treatment, the leaf samples were collected and plants were irrigated as required for 45 days for the recovery from drought stress, and data were recorded.

Plant height (cm) was measured twice (before and after stress) using a ruler. The difference between both the readings was calculated as net plant height. The number of leaves was counted before and after stress and the difference between both readings was calculated as the net number of leaves. Leaf area (cm²) was calculated after multiplying the length and width of the leaf (before and after stress). The difference was calculated as net leaf area. Four leaves from each replicate were harvested and their fresh weight (g) was taken immediately using a digital weighing balance. The leaves were kept in an oven (Memmert-110, Schwabach, Germany) at 70°C for 72 h for dry weight (g).

The plants were shifted from the greenhouse (45–47°C) to the growth room (32°C). After 24 h, chlorophyll contents (CC) (μg/g), photosynthesis (A) (μmol CO₂ m⁻²s⁻¹), transpiration (E) (μmol H₂O m⁻²s⁻¹), water use efficiency (WUE), sub-stomatal CO₂ (Ci) (μmol mol⁻¹), stomatal conductance to water vapor (gs) (C, μmol m⁻²s⁻¹), and leaf temperature (Tch) (°C) were measured from the third recently matured young leaf from the apex by using portable infrared gas analyzer IRGA (LCi-SD, ADC; Bio-scientific Ltd., UK) in five replicates. All measurements were made during the daytime between 10:00 a.m. and 12:00 O'clock.

The activity of Superoxide Dismutase (SOD) (IU/mg of protein) was determined by measuring its ability to prevent the photo-reduction of nitroblue tetrazolium (NBT). Enzyme extract was prepared in potassium phosphate buffer (pH 5), vortexed, and centrifuged. The reaction mixture comprises enzyme extract (100 μl), potassium phosphate buffer (pH 5) (500 μl), methionine (200 μl), triton X (200 μl), NBT (100 μl), and distilled water (100 μl). The mixture was kept under UV light for 15 min and then 100 μl of riboflavin was added. The absorbance of reaction mixture was observed at 560 nm using an ELISA plate (Giannopolitis and Ries, 1977). The amount of enzyme restricting 50% of the NBT photo decline was considered as one unit of SOD. Peroxidase (POD) (IU/mg of protein) activity was measured using the solution of POD reaction comprising 20 mM guaiacol (100 μL), 50 mM phosphate buffer (pH 5) (800 μL), gum solution (0.1 ml), and 40 mM H₂O₂ (100 μL). Enzyme extract (100 μl) and reaction mixture (100 μl) were added and absorbance was recorded at 470 nm using ELISA plate (Liu et al., 2009). The catalase (CAT) (IU/mg of protein) activity was measured as the amount of H₂O₂ consumed and converted to water H₂O and oxygen O₂. Enzyme extract (100 μl), used for SOD determination, was taken and 100 μL of 5.9 mM H₂O₂ was added in it. The absorbance was recorded at 240 nm on ELISA plate (Liu et al., 2009). An absorbance change of 0.01 units per min was considered as one unit of POD and CAT. Proline contents (μg/g Fw) were estimated (Bates et al., 1973) from 0.5 g of fresh leaf tissue homogenized in 10 ml of 3% sulfosalicylic acid. Homogenate was filtered through Whatman No. 2 filter paper. Later, the filtrate (2.0 ml) was mixed with 2.0 ml of acid ninhydrin solution (Ninhydrin, 1.25 g) and then dissolved in 6 M orthophosphoric acid (20 ml) and glacial acetic acid (20 ml). The mixture was kept for 60 min on ice bath to cool. Finally, toluene (4.0 ml) was added to the solution and mixed vigorously by passing a continuous stream of air for 1–2 min. The absorbance was taken at 520 nm using spectrophotometer.

Total soluble proteins (TSP) (mg/g) were quantified by adding the Bradford reagent to the enzyme extract prepared for SOD analysis and absorbance was observed at 595 nm (Bradford, 1976) using a spectrophotometer. Total phenolic contents (TPC) (mg/g GAE) were estimated by using the Folin–Ciocalteu reagent (FCR) method (Ainsworth and Gillespie, 2007). The 200-μl F-C reagent was added to 100 μl of tissue extract followed by the addition of 800 μl of 700 mM Na₂CO₃ and incubation for 2 h at room temperature. The absorbance was measured at 765 nm using a spectrophotometer. Total flavonoid contents (TFC) (mg/g catechin standard) were determined by the previously described method (Dewanto et al., 2002). In brief, 1 ml extract, containing 0.01 mg/L of dry matter, was added in a 10-ml volumetric flask and then mixed with 5 ml of distilled water followed by the addition of 0.3 ml of 5% NaNO₂. In two consecutive intervals of 5 min each, 0.6 ml of 10% AlCl₃ and 2 ml of 1 M NaOH, respectively, were added and the absorbance was taken at 510 nm using spectrophotometer. Antioxidants capacity (DPPH radical scavenging assay) (%), of designated plant crude extract and its polar fractions, was evaluated by assessing their scavenging ability toward 1, 1-diphenyl-2-picrylhydrazyl stable radicals (DPPH) (Queiroz et al., 2009).

The leaf samples, from drought-stressed (50 and 75% field capacities) and control plants (100% field capacity) of both guava cultivars, were taken in liquid N₂ and stored at −80°C. Total RNA was extracted, quantified, and purified by using the previously described method (Jaakola et al., 2001). cDNA was synthesized with 5-Aminoallyl-dUTP (AA-dUTP). Before coupling unincorporated aa-dUTP molecules were removed. The coupling of the dyes (Cy3 and Cy5) to the aa-dUTP cDNA was performed as per manufacturer's instructions. Microarray spotted slides with 500 drought-specific ESTs from Gossypium arboreum and Gossypium hirsutum, showing >70% homology to Arabidopsis thaliana (NCBI database), were hybridized with the guava cDNA probe. The standard protocols established by ArrayIt (https://shop.arrayit.com/microarray_tools.aspx) provided by MicroGrid610 by Genomic Solutions® were followed. The detailed procedure was followed as described previously (Ahmed et al., 2020).

Microarray UC4 scanner (Genomic Solutions, USA) was used for scanning the slides. For each sample, separate hybridization, using a single dye, was carried out. The intensities of each sample were measured separately and the images were saved. The preliminary analysis of scanned images was carried out with the help of the microarray image analysis software (GeneTAC Integrator, Genomic Solutions, USA) (Saeed et al., 2003). Data were analyzed using GeneSpring GX software (Agilent Technologies, Santa Clara, CA) and the quality parameters of the extracted data were processed using customized technology created for in-house built experimental procedures. Percentile Shift method and Scaling options were used for the data normalization and to overcome inter-array differences. Normalized intensity values were used for Hierarchical Clustering using information of the differentially expressed sequences and keeping 1.5-fold change as the cutoff value.

The differentially expressed oligos were used to trace original/parent sequence IDs. For GO functional categorization and annotation, these ESTs were saved in FASTA format and run in Blast2Go Pro software (https://www.blast2go.com/) using Cloudblastx for mapping and annotation (Conesa et al., 2005). The functional categorization was done based on cellular components, biological processes, and molecular functions. The differentially expressed common ESTs, upregulated and downregulated, in both guava cultivars were separately categorized.

The data were analyzed using a three-way factorial arrangement. Means were compared through the LSD test at 0.05 level of significance in statistics 8.1.

The exposure of Guava cultivars “Gola” (round-shaped) and “Surahi” (pear-shaped) to three-field capacity levels (100, 75, and 50%) revealed distinct morphological alterations. The responses of both guava cultivars to all the field capacity levels remained non-significant in terms of plant height, number of leaves, leaf fresh weight, and dry weight (Table 1). However, leaf area was decreased as the field capacity level decreased from 100 to 75% and then to 50% (i.e., increasing drought stress) in “Gola” (−32.75 and −53.42%, respectively) and “Surahi” (−31.10 and −67.16%, respectively) cultivars. Comparatively, the “Surahi” cultivar showed a significantly higher leaf area than the “Gola” cultivar at 75% field capacity level (i.e., 25% drought stress) (Table 1). Taking both cultivars (“Gola” and “Surahi”) together, the mean values of plant height, number of leaves, and leaf area were found to be minimum at the maximum level of drought stress (i.e., 50%). Leaf fresh weight was reduced, significantly, under 75 and 50% field capacity as compared to the 100% level in contrast to leaf dry weight which remained non-significant across the three-field capacity levels (Table 1). These morphological reductions were translated into significant decrements in physiological parameters of guava cultivars in terms of minimal chlorophyll content, photosynthesis, transpiration rate, sub-stomatal CO₂, and stomatal conductance to water vapor ratio under the highest drought stress level (i.e., 50%). The transpiration rate of both the cultivars was decreased with the increasing drought stress, yet the “Surahi” cultivar showed a minimum transpiration rate (0.334 μmol H₂O m⁻²s⁻¹) as compared to “Gola” (0.543 μmol H₂O m⁻²s⁻¹) which enhanced the WUE of “Surahi” cultivar at maximum drought stress, that is, 50% field capacity (Table 2).

The antioxidant capacity (DPPH) and superoxide dismutase (SOD) activity were found to be non-significant in both guava cultivars (Table 3). Total soluble proteins decreased as the drought stress increased from 75 to 50% in “Gola” (−11.99 and −27.74%, respectively) and “Surahi” (−11.69 and −22.19%, respectively) with “Gola” cultivar showing higher values of total soluble proteins at all three field capacity levels (Table 3). In the same context, total phenolic contents were the maximum in the “Gola” cultivar (15.14%) under maximum drought stress (50% field capacity) (Table 3). Nevertheless, the peroxidase (POD) and catalase (CAT) activities along with proline and total flavonoids contents were observed to be maximum (402, 170.21, 116.75, and 22.23%, respectively) in the “Surahi” cultivar under 50% field capacity (Table 3). Proline contents and total flavonoid contents were also increased, significantly, in “Gola;” however, the “Surahi” cultivar accumulated maximum contents of both under maximum drought stress levels (Table 3). Altogether, the combined effect of drought stress decreased superoxide dismutase (SOD) activity and total soluble proteins and increased the peroxidase (POD) and catalase (CAT) activities which, somehow, elevated the antioxidant capacity (DPPH) of both guava cultivars (19.89% in “Gola” and 32.14% in “Surahi”) (Table 3).

Correlation analysis of morphological, physiological, and biochemical responses of guava cultivars to drought stress revealed high negative correlation of leaf temperature (Tch) and high positive correlation of chlorophyll content (CC) and photosynthesis (A) with plant height (PH₁), leaf number (LN₁), leaf area (LA₁), leaf fresh weight (LFW₁), and leaf dry weight (LDW₁) (Table 4). Interestingly, superoxide dismutase (SOD) exhibited a strong positive correlation with plant height (PH₁), leaf number (LN₁), leaf area (LA₁), leaf fresh weight (LFW₁), and leaf dry weight (LDW₁) (Table 5). Similarly, superoxide dismutase (SOD) was also positively correlated to the chlorophyll content (CC) and photosynthesis (A) (Table 6). Moreover, the correlation of total soluble proteins with sub-stomatal CO₂ (Ci), transpiration (E), and stomatal conductance to water vapor (gs) was also strong and positive (Table 6).

Principle Component Analysis (PCA) of morphological, physiological, and biochemical responses, of “Gola” (C₁) and “Surahi” (C₂) cultivars under drought stress, was carried out by classifying them into separate groups. The PCA plot developed was based on the first (PC1) and second (PC2) component factors (64.39 and 23.50%, respectively). Strains C₂T_o and C₁T_o, maintained under control conditions, along with C₂T₂ and C₁T₂, of maximum drought stress, were found to be the most divergent and outliers (Figure 1A). The dendrogram depicted the formation of two main groups (G₁ and G₂), where G₂ grouped higher stress treatments in both cultivars, and four subgroups (A-D) placing “Surahi” (C₂) with drought stress treatments T₁ and T₂ in subgroups C and D (Figure 1B).

Microarray analysis underpinned differential expression of 117 ESTs in the “Gola” cultivar, 82 of which were upregulated while 35 were downregulated. Similarly, out of 234 differentially expressed ESTs in the “Surahi” cultivar, 166 were upregulated, whereas 68 were downregulated under drought stress. There were 50 co-upregulated ESTs in “Gola” and “Surahi” cultivars under drought stress (Figure 2). Functional annotation of upregulated ESTs among cellular components of the “Gola” cultivar disclosed 23 sequences belonging to the functions related to the plastid (8), nucleus (8), and cytosol (7). Each chloroplast, plastid envelope, and mitochondrion contained four ESTs individually; whereas, every compartment such as the cytosolic part, vacuole, nuclear lumen, plastid stroma, and endopeptidase complex contained three ESTs each. On the other hand, there were 55 upregulated ESTs belonging to the nucleus (17), cytosol (25), and plastid (13), while 7 belonging to vacuole and 6 to chloroplast of the “Surahi” cultivar in response to drought stress (Figure 3A). Notably, 46% of ESTs, upregulated in cellular components of the “Gola” cultivar, belonged to plastid, nucleus, and cytosol. Similarly, out of the total ESTs upregulated in cellular components of the “Surahi” cultivar, almost 81% of ESTs were upregulated in plastid, nucleus, and cytosol (Figure 3A). Regarding functional annotation of differentially expressed downregulated ESTs, 33 ESTs were downregulated in different cellular components of the “Gola” cultivar under drought stress. Among these, a major component (such as 36%) accounted for the downregulation of five ESTs in nucleus, four in plastid, and three in cytosol. The other 64% cellular component of the “Gola” cultivar contained the downregulated ESTs (Figure 3B). Similarly, 48% of the total downregulated ESTs, in cellular components of the “Surahi” cultivar, belonged to the plastid, nucleus, and cytosol (Figure 3B). The remaining 52% of cellular components contained the other downregulated ESTs (Figure 3B).

In response to drought stress, cultivar “Surahi” showed a higher number of ESTs, upregulated (185) and downregulated (103) as compared to the “Gola” cultivar which contained 90 and 98 upregulated and downregulated ESTs, respectively, in most of the biological processes (Figures 4A,B). Among the upregulated ESTs in biological processes of the “Surahi” cultivar, 60 (32.4%) ESTs were involved in various metabolic processes taking place in the nucleus, RNA, cellular protein, phosphate-containing compounds, regulation of RNA, and regulation of nucleo-base RNA compounds. However, comparatively, the “Gola” cultivar expressed 28 (31.1%) upregulated ESTs in the biological processes of various metabolic systems such as cellular proteins, phosphate-containing compounds, nucleic acids, oxoacid, RNA, and cellular amide (Figure 4A). Likewise, 24 (23.3%) and 34 (34.7%) of the total downregulated ESTs in “Surahi” and “Gola” cultivars, respectively, were associated with drought stress-responsive biological processes (Figure 4B). Surprisingly, seven ESTs were observed downregulated, as a result of the stress response, in the biological processes of only the “Gola” cultivar (Figure 4B). A total of 54 and 71 ESTs were upregulated in “Gola” and “Surahi” cultivars, respectively, in the molecular functions resulting from drought stress (Figure 5A). Out of these, five ESTs in the “Gola” cultivar and 14 in “Surahi” were found to coordinate with metal ion binding capacity. Furthermore, there were 36 upregulated ESTs relevant to purine-related molecular functions in “Surahi” as compared to 28 ESTs in the “Gola” cultivar (Figure 5A). Among the downregulated ESTs in molecular functions, 48 belonged to the “Gola” cultivar, whereas only 28 belonged to the “Surahi” cultivar (Figure 5B). Out of 48 downregulated ESTs in the “Gola” cultivar, 18 belonged to purine-related molecular functions, while metal ion binding involved only 1 EST. In contrast, 12 and 7 ESTs were downregulated in molecular functions related to purine binding and metal ion binding, respectively, in the “Surahi” cultivar (Figure 5B).

Several important ESTs were identified in “Gola” and “Surahi” cultivars in response to drought stress. In addition to similar ESTs upregulated in both cultivars under drought stress (Table 7), ESTs encoding peroxidases (peroxidase-like, thioredoxin-dependent peroxidase 1, and Ascorbate peroxidase 1) and plant regulator RWP-RK family transcription factors were significantly upregulated in “Surahi” as compared to “Gola” cultivar. Sucrose synthase (SUS), alcohol dehydrogenase (ADH), and ubiquitin family genes were also upregulated in the “Surahi” cultivar (Table 8). The drought inducted genes including basic leucine zipper (bZIP) transcription factors were downregulated (Table 9) and putative zinc transporter 11 precursor (ZIP11) was upregulated in the “Surahi” cultivar (Table 8). Ca²⁺/H⁺ exchanger (CAX3) was also downregulated in the “Surahi” cultivar in response to drought stress (Table 9).