The experiment was conducted at the Center for Ecological Studies at the Chinese Academy of Sciences, Sichuan in south-west China. Healthy and uniform, 2-year-old P. zhennan plants were collected from Sichuan Agricultural University, Sichuan province, and transplanted to 10 L plastic pots filled with approximately 4 kg of homogenized topsoil (pH 7.3, total nitrogen 0.19%, and carbon 2.67%). The pots were arranged in a complete randomized block design in a greenhouse (temperature range 18–32°C, relative humidity range 50–85%) and regularly watered. Light availability was homogeneous inside the greenhouse, and direct sunlight reduction due to covering was in the range of 6–9%. After growing for 2 months in the greenhouse, the plants were subject to three replicates of four treatments for 90 days: two water regimes (well-watered and water-stressed) and two levels of P fertilization (with and without P fertilization). Immediately prior to applying the treatments, total and available P in soils was first determined to be 0.89 g kg^-1 and 27.6 mg kg^-1, respectively. Available P was extracted with 0.5 M NaHCO₃ (pH 8.2) according to Olsen and Sommers (1982) and measured colorimetrically by the molybdate-ascorbic acid as described by Murphy and Riley (1962). Soil relative water content (SRWC) of the two water treatments (control: 80–85%; severe drought: 30–35%) was calculated using the weight method (Xu et al., 2009). The pots were weighed daily and watered up to their respective target SRWC, by replacing the amount of water transpired and evaporated. SRWC was expressed as:

where W_soil was the current soil weight, W_pot was the weight of the empty pot, DW_soil was the dry soil weight, and W_FC was the soil weight at field capacity.

Phosphorous fertilization was supplied as sodium dihydrogen phosphate (NaH₂PO₄, 25.5% P) with the dose of 129.3 mg P mixed in 200 mL water per pot every 30 days. In order to avoid systematic error produced by fluctuation in environmental conditions, pots were rotated after every 5 days during the experiment. Each treatment was replicated three times. Plant samples were collected at the end of the experiment.

Plant height (cm), stem diameter (mm), and leaf area (cm²) were measured by using a measuring tape, caliper, and a leaf area meter (CI 202, United States), respectively. After removing the plants from the soil, roots, stems, and leaves were separated and weighed. Samples were oven dried at 70°C for 24 h, to measure biomass.

Expanded leaves were collected from each plant and weighed to obtain fresh weight (FW). The leaves were then immediately dipped into distilled water at a temperature of 4°C and under dark conditions. After 12 h, leaves were weighed to obtain turgor weight (TW) and then dried for 24 h in an oven set at 70°C to determine dry weight (DW). The following equation was used to calculate leaves relative water content (LRWC).

The net CO₂ assimilation rate (P_n), stomatal conductance (G_s), and intercellular CO₂ concentration (C_i) were measured with a portable open-flow gas exchange system (LI-6400, LI-COR Inc., United States), between 9:00 and 11:00 h, on fully expanded leaves that were at similar stages of development. During this time, relative air humidity, CO₂ concentration, and photon flux density were maintained at 60–70%, 380 μmol mol^-1 and 800 μmol m^-2 s^-1, respectively. Intrinsic water use efficiency (WUE_intr) was calculated by dividing the instantaneous values of P_n by G_s. The maximum quantum efficiency of photosystem II (F_v/F_m) of the leaves was measured with a portable pulse amplitude modulated fluorometer (PAM-2100, Walz, Effeltrich, Germany), where the leaves were dark-adapted with clips for 20 min. After this time, minimal fluorescence (F_o) was measured under a weak pulse of modulating light over 0.8 s, and maximal fluorescence (F_m) was induced by a saturating pulse of light (8,000 mmol m^-2 s^-1) applied over 0.8 s. F_v/F_m was calculated, where F_v was the difference between F_m and F_o.

Chlorophyll a (Chl a), chlorophyll b (Chl b), and carotenoids (Car) were determined using 0.2 g FW leaf samples, with 80% acetone as a solvent. Leaf samples were placed in dark conditions for 36 h and absorbance was recorded at 662, 645, and 470 nm spectrophotometrically (Lichtenthaler and Buschmann, 2001).

Dried leaf samples (0.2 g DW) were mixed with 6 mL of 80% ethanol at 80°C for 30 min. The resulting extracted supernatant was analyzed for soluble sugars (SS) following the anthrone method using sucrose as a standard (Zhang and Qu, 2003). For NO₃^- concentration, 0.2 g of frozen leaves was homogenized in 5 mL of deionized water, while for NH₄⁺ concentration, 0.2 g of frozen leaves was homogenized in 2 mL of 10% HCl. The resulting supernatants were analyzed using a quantitative colorimetric method as described by Tang (1999). Proline was extracted with 2 mL of 10% acetic acid and 5 mL of 3% sulfosalicylic acid, respectively. The resulting supernatants were analyzed according to the method of Liu et al. (2014). Soluble proteins (SP) were determined using Bradford G-250 reagent.

Two measures of ROS were estimated. Firstly, production rate of superoxide anion (O₂^⋅-) was measured by monitoring nitrite formation from hydroxylamine, in the presence of O₂^⋅- (Elstner and Heupel, 1976). Fresh leaves (0.2 g) were homogenized with 2 mL of 65 mM phosphate buffer (pH 7.8) and centrifuged at 5000 × g for 10 min. The incubation mixture contained 0.9 mL of 65 mM phosphate buffer (pH 7.8), 0.1 mL of 10 mM hydroxylammonium chloride, and 1 mL of supernatant. After incubation at 25°C for 20 min, 17 mM sulphanilamide, and 7 mM α-naphthylamine were added to the incubation mixture and kept at 25°C for 20 min. Ethyl ether in the same volume was added and centrifuged at 1500 × g for 5 min. The absorbance wavelength for the aqueous solution was 530 nm.

Secondly, hydrogen peroxide (H₂O₂) concentration was determined by monitoring the absorbance of the titanium-peroxide complex at 410 nm (Patterson et al., 1984). Fresh leaves (0.2 g) were homogenized with 5 mL of acetone and centrifuged at 3,000 × g for 10 min. The reactive mixture, containing 0.1 mL of titanium reagent (50 μL of 20% titanium tetrachloride in concentrated HCl), 0.2 mL of ammonia, and 1 mL of supernatant, was centrifuged at 3,000 × g for 10 min. The resulting precipitate was washed five times with acetone, before being centrifuged at 10,000 × g for 5 min. The precipitate was solubilized in 3 mL of 1 M H₂SO₄ and the absorbance at 410 nm was measured. Lipid peroxidation was estimated by measuring malondialdehyde (MDA) content according to the thiobarbituric acid (TBA) test at 450, 532, and 600 nm, respectively (Zhou et al., 2007). For MDA assay, 0.25 g of fresh leaves were ground in 5 mL of 1% trichloroacetic acid (TCA) and centrifuged at 5,000 g for 10 min. Supernatant (1 mL) was added to 4 mL of 20% TCA (containing 0.5% TBA) and the mixture was heated at 95°C for 30 min, before being cooled in an ice bath. Absorbance was read using a spectrophotometer at 450, 532, and 600 nm (Zhou et al., 2007). The MDA concentration was calculated using the following equation:

Three measures of antioxidant stress were assessed. Superoxide dismutase (SOD) activity was determined using the nitroblue tetrazolium (NBT) method (Fu and Huang, 2001). One unit of SOD activity was defined as the amount of enzyme required for 50% inhibition of NBT reduction at 560 nm. Activities of catalase (CAT) and peroxidase (POD) were determined using the methods of Fu and Huang (2001). For CAT, the decomposition of H₂O₂ was determined by measuring the reduction in absorbance at 240 nm for 1 min. For POD, the oxidation of guaiacol was determined by measuring the increase in absorbance at 470 nm for 1 min. One unit of CAT and POD activity was defined as an absorbance change of 0.01 unit’s min^-1.

All measurements were repeated three times, and the data were organized using Microsoft Excel 2007 and presented as means ± standard errors (SEs). SPSS version 16.0 (Chicago, IL, United States) was used to run one-way analysis of variance (ANOVA) and Duncan’s multiple range tests at the 0.05 significance probability level were used to compare mean values. Prior to analysis, data were checked for normality and homogeneity of variances. Origin pro 8.5 was used for graphical presentation; error bars represent standard errors, and all data in the figures represent the means ± SEs.

Reduction in morphological traits was observed in drought-stressed plants compared with well-watered plants (Table 1). Under water deficit conditions, shoots and root biomass, leaf area and stem diameter significantly decreased (45.2, 39.8, 28.5, and 9.1%, respectively) compared with well-watered plants irrespective of P application. Root biomass in drought-stressed plants was significantly higher (14.4%) in fertilized plants than in unfertilized plants. There were no significant treatment differences for the other growth parameters.

In comparison with well-watered unfertilized plants there was a significant reduction in LRWC, P_n, F_v/F_m, C_i, and G_s (27.2, 72.6, 17.2, 34.72, and 73.94%, respectively) of drought-stressed unfertilized plants. Under drought stress conditions, LRWC, P_n, and F_v/F_m were significantly lower in unfertilized plants than in fertilized plants, while there were no significant effects of fertilizer on the other parameters in drought-stressed plants (Table 2). Water use efficiency (WUE_intr) showed an opposite trend and increased 44.13% under drought condition than well-watered plants, regardless of P application. However, there was no significant change in WUE_intr in P fertilized plants under water-stressed conditions. There was no effect of P fertilizer on LRWC or any of the photosynthetic and chlorophyll parameters in well-watered plants. Moreover, under drought condition water consumption rate was higher in P-fertilized plants than unfertilized plants but lower than well-watered plants (data not shown).

Concentration of Chl a and Chl b in non-fertilized plants was significantly lower (22.2 and 40.0%, respectively) in water-stressed than in well-watered plants (Figures 1A,B). We found that Chl a and Chl b concentration was significantly greater in drought-stressed plants that had been treated with fertilizer than in those that had been unfertilized. Neither water nor fertilizer treatment had any effect on carotenoid concentration (Figure 1C).

The concentration of NH₄⁺ was higher in well-watered plants than drought-stressed plants, while the opposite was found for proline concentration (Table 3). With the exception of SS concentration, which was greater in fertilized than unfertilized drought-stressed plants, we found no significant effect on biochemical parameters of fertilizer within well-watered or drought-stressed plants.

Regardless of P application, measures of ROS production (O₂- and H₂O₂) and lipid peroxidation (MDA) were significantly higher in plants under drought stress conditions than in well-watered plants (Figure 2). Addition of P had no significant effect on concentration of either both O₂- and H₂O₂ in the two water treatments or MDA in well-watered plants; however, it resulted in significantly lower MDA content in drought-stressed plants (Figures 2A–C).

Activity of POD and CAT was higher in drought-stressed plants than in well-watered plants, and SOD activity was highest in unfertilized, drought stress plants (Figure 3). Addition of P had no significant effect on the measures of antioxidant stress in either of the water treatments.

Drought stress is considered a major environmental stress that adversely affects tree growth and forest productivity around the world (Bartlett et al., 2012; Harfouche et al., 2014). We found that, irrespective of P application, biomass of the above and below ground plant parts of P. zhennan was significantly lower under drought stress conditions compared with well-watered conditions. Low water content in soil decreases mobility of available ions, nutrient availability, and microbial activities in the soil (Hu and Schmidhalter, 2001). Furthermore, root interaction with arbuscular mycorrhizas (AM) can increase plant tolerance to drought because the fungi improves plant water status by modulating ABA-mediated abiotic signaling pathway involving D-myo-inositol-3-phosphate synthase (IPS) and 14-3-3 proteins (Li et al., 2016). Moreover, in the plant it causes partial or total stomatal closure, drop in water potential, loss of cell turgor reduction of cell expansion, and if the dehydration is severe, the disruption of normal bilayer structure of the cell membranes through a reduction in synthesis, and possibly denaturation of cytosolic and organelle protein that leads to impaired cell metabolism (Mahajan and Tuteja, 2005). We found that application of P to plants under drought stress resulted in significantly greater root biomass, which may be attributed to several factors, including increased uptake of P, higher consumption rate of assimilates in root material, and enhanced hydraulic conductance of the root system (Garg et al., 2004). Phosphorous application plays an important role in root development thereby increasing accessibility to other nutrients in the rhizosphere (Liao and Yan, 2000; Razaq et al., 2017). Moreover, the root tip is responsible for sensing and signaling of P availability, and to initialize the reduction in root growth in P deficient environments and the proteins containing an SPX domain are important in regulating not only P uptake but also P distribution within subcellular compartments (Rouached et al., 2010). Higher root biomass also improves the ability of roots to extract soil moisture, contributing to drought tolerance (Meng and Yao, 2015); therefore, P application may enhance the drought tolerance of P. zhennan through the promotion of root biomass. This higher root biomass would be a further advantage in drought conditions where water is available in deeper soil profiles (Tardieu, 2012). We also found that P fertilization had no effect on the biomass of above and below ground plant parts in well-watered condition and suggest this may be due to the slow-growing nature of P. zhennan, and/or the sufficient availability of P in the soil to meet the functional requirements of the establishing seedlings. Curiously, our results contrast with studies that report increased growth rate and biomass accumulation in P fertilized plants (Graciano et al., 2005; Jones et al., 2005; Ram et al., 2006) but support other studies that suggest non-significant effects of P fertilization on growth parameters (Yin et al., 2012; Liu et al., 2015). The high variability in the responses of different plant species to P application may be because response to fertilization depend on the species, available nutrients, nutrient interactions, soil physical properties, water availability, among many other factors that modulate the response of the plant to an increase in particular nutrient concentration in the soil. Absorption of P takes place at the soil surface, and its lower diffusion rate and slower movement toward the root, compared with other nutrients, possibly affect its use efficiency (Schachtman et al., 1998; Grant et al., 2005). However, in greenhouse experiments, growing conditions are well controlled; therefore P use efficiency can be improved if P is mixed uniformly with the volume exploited by roots (Mitchell, 1957; Sample et al., 1980). However, P use efficiency varied within a species. For example in Hordeum vulgare the expression of the gene HVPT5, that can be used to estimate phosphate use efficiency, was higher under low P availability in a tolerant accession, but its expression did not change in the sensitive accession (Ren et al., 2016). Moreover, plants have inducible high affinity phosphate transporters and constitutive low affinity phosphate transporters encoded by Pht (phosphate transporter) gene family that ensure P uptake from the soil and distribution within different organelles of plant to sustain photosynthesis, respiration, and growth even under low P availability conditions (López-Arredondo et al., 2014). Member of Pht1 gene family encoded high affinity P transporters which are mostly expressed in epidermal and outer cortex of the root cells and have already been identified as mediators of P uptake when P is limited (Schunmann et al., 2004). However, members of the Pht2, Pht3, and Pht4 gene families were found to be associated mostly with P distribution within subcellular compartments (Versaw and Harrison, 2002; Guo et al., 2008). Moreover, recently, the role of microRNAs (miRNAs) has been revealed in the regulation of P homeostasis (Doerner, 2008). miRNA399 has been uncovered as a component of the shoot-to-root P deficiency signaling pathway, it moves via phloem and repress E2-conjugase which causes increase in the expression of root P uptake transporters and hence in the acquisition of P by the roots and its translocation and distribution to the shoot (Lin et al., 2008).

We found that leaf area of drought-stressed plants was significantly lower than in well-watered plants, irrespective of P application. Generally, water deficiency during the vegetative growth stage changes leaf turgidity and temperature, and reduces the supply of assimilates, thus inhibiting leaf growth. In drought tolerant plant species, reduction in leaf area is considered an adaptive strategy to reduce water loss through transpiration (Rostamza et al., 2011). However, the lack of difference in leaf area, but higher Pn rate and LRWC that we observed in P fertilized, drought-stressed P. zhennan, suggests better acclimatory response to drought stress conditions. Dehydration tolerance of a plant can be measured using the LRWC index. In our study, LRWC in plants under drought conditions was significantly lower than that in well-watered plants. Reductions in LRWC in response to drought-stressed conditions have also been observed in different plant species (Shubhra et al., 2004; Yang and Miao, 2010). We found that P fertilizer increased LRWC in drought-stressed plants, but not in well-watered plants. The higher LRWC in fertilized, drought-stressed plants in our study may be associated with the greater biomass of P fertilized plants. Several studies have also reported improved LRWC due to either an improved ability of root to extract water or an improved conservation of water in the plant tissues (Centritto et al., 1999; Garg et al., 2004; Shubhra et al., 2004; Sato et al., 2010). However, our results contradict several studies that showed non-significant effects of P fertilization on LRWC under water deficit condition (dos Santos et al., 2004; Singh et al., 2006; Liu et al., 2015). This variability in the effect of P in drought-stressed plant species may be due to interspecific differences in physiological, biochemical, and molecular mechanisms, such as gene expression and protein assimilation. Genes induced by drought stress have been shown to not only protect plant cells from dehydration but also regulate signal transduction of certain genes, many of them encode ion transport proteins which need ATP (P rich compound) and P is also an important element in various metabolic steps of protein synthesis (Bohnert and Jensen, 1996; Bohnert and Sheveleva, 1998). Moreover, a complex intercross between P and N availability in plant water use was demonstrated in the subtropical trees Eucalyptus grandis and Pinus taeda (Faustino et al., 2013; Graciano et al., 2016; Costa et al., 2017). Differences in dry mass partitioning as well as changes in morphology and physiology in different organs explain why fertilization can affect plant drought tolerance in different direction, accordingly with the environmental factors (soil texture and moisture, nutrient availability, stress intensity and duration) and capacity of the plant to make morphological and physiological acclimations to stressful conditions.

Drought stress reduces photosynthetic rate, due to a decrease in leaf expansion and associated damage to photosynthetic machinery (Wahid and Rasul, 2005). Our findings revealed that P_n significantly decreased under drought stress, but the drop was higher in unfertilized than in fertilized plants. Stomatal closure is considered to be the main factor in decreasing photosynthesis under water deficit conditions (Anjum et al., 2011). Stomatal closure in response to soil water deficit occurs because roots release high concentrations of abscisic acid (ABA) to the xylem and, as a result, the increased pH of xylem sap promotes ABA loading and subsequent transport to the shoots (Hartung et al., 2002). It is known that many drought-inducible genes respond to ABA level in leaves, for example, ABA-dependent and ABA-independent regulatory systems of gene expression can be regulated under drought stress (Zhu, 2002). Qin and Zeevart (2002) reported that protein dephosphorylation and farnesylation are responsible for ABA signaling, while Sauter et al. (2001) showed that ABA stimulates K+ ions efflux from the guard cells, resulting in loss of turgor pressure, and decrease G_s. Reduction of G_s limits gas exchange, decreases C_i concentration and rates of photosynthesis, due to decline in Rubisco activity (Reddy et al., 2004). Similarly, our results indicated a significant decline in G_s and C_i under drought stress, more sharply in unfertilized than in fertilized plants. There is still debate, however, about whether drought restricts photosynthetic rate through stomatal closure or metabolic impairment (Tezara et al., 1999). Our results revealed that lower P_n was not only associated with G_s limitation, but was also due to impaired photosynthetic apparatus as reflected by significant decrease in F_v/F_m during water deficit conditions. We also found that P application under drought stress conditions resulted in significantly higher P_n and F_v/F_m, but had no significant effect on stomatal conductance or transpiration rate. Thus, our results suggest an enhanced drought tolerance mechanism that conserves water, as indicated by increased LRWC, is stimulated by the addition of P. These results are consistent with previous studies that have reported enhanced photosynthetic activity in different plant species treated with P fertilizer under drought stress (Burman et al., 2009; Singh et al., 2013; Liu et al., 2015). In addition, other factors may have positive impacts on P_n, in response to P application, such as increased production of assimilatory products (ATP and NADPH) and carboxylation activities (Lawlor and Cornic, 2002). WUE_intr is considered to be an important component of adaptation to drought stress and in our study; WUE_intr showed an opposite trend to P_n, where it was significantly higher in drought-stressed plants, irrespective of P application. Therefore, plants under drought partially closed the stomata to reduce waters losses but the photosynthesis was affected proportionally in lesser extent. However, P application in our study had no significant effect on WUE_intr of water-stressed and well-watered plants, as has also been reported in other studies (Oliveira et al., 2014; Liu et al., 2015).

Reduction in chlorophyll concentration is a sign of oxidative stress or pigment photooxidation under drought stress (Zhang and Kirkham, 1996) and low levels of photosynthetic pigments limit the rate of photosynthesis, thus reducing primary production. Our results showed that water deficit caused significant damage to the photosystem by degrading chloroplast pigments. The degraded chloroplast pigments may have also contributed to the decreased P_n observed in our study. Similar findings have been observed in other studies that suggest drought stress damage photosynthetic pigments (Frosi et al., 2013; Rivas et al., 2013). Previous work has shown that Chl a and Chl b are susceptible to soil water deficiency (Farooq et al., 2009) Furthermore, soil dehydration has been shown to damage lamellae vesiculation and chloroplast membranes, inducing reductions in chlorophyll (Anjum et al., 2011). In our study, P application had a significant positive effect on Chl a and Chl b concentration, which may explain the higher P_n rate in P fertilized, drought-stressed plants, because high leaf chlorophyll concentration may allow for increased harvesting of light over shorter periods of time, as evidenced by the observed higher photosynthetic rates. Although there were changes in chlorophyll concentration, P fertilization did not change carotenoids concentration. This result is different of that found in the herbaceous Petunia hybrid, in which high P inhibited carotenoids biosynthetic genes (Nouri et al., 2015). Although our results were contradictory to some studies, in which negligible changes in chlorophyll concentrations due to P fertilization may be due to the duration and severity of drought (Zhang and Kirkham, 1996; Campbell and Sage, 2006; Singh et al., 2013). Nevertheless, our findings are supported by previous studies that suggest P application increases synthesis of photosynthetic pigments in plants under drought stress (Sharma, 1995; Sinha et al., 1995).

Plants adapt to drought environments by increasing the solute concentration of cells to maintain osmotic function and hydration (Ramanjulu and Bartels, 2002). Plants accumulate a variety of osmolytes in the cytosol, therefore the ability to increase osmotic pressure is considered to be a potential cellular drought tolerance mechanism, as it improves or maintains turgidity and continuation of plant growth. Apart from osmotic adjustment, osmolytes also help in ROS detoxification, membrane stabilization, as well as protecting macromolecules (Keunen et al., 2013). Our results showed that there was no significant increase in SS concentration under drought in unfertilized plants. However, P fertilization significantly increased SS concentration in drought-stressed plants compared with well-watered; this may be due to the inhibition of normal SS utilization and translocation during water stress or hydrolysis of starch (Shubhra et al., 2004). Accumulation of SS protects cells in water deficit environments by substituting the hydroxyl group for water, thus maintaining a hydrophilic interaction between proteins and membranes to retain membrane integrity (Hoekstra et al., 2001). This positive effect of P fertilization on SS accumulation and mobilization clearly indicates its role in improving drought tolerance of P. zhennan. We found that proline concentration was significantly higher in drought-stressed plants than in well-watered. Proline accumulation in low moisture environments is due to reciprocal regulation of two pathways: up-regulation of proline synthesizing enzymes and down-regulation of proline degrading enzymes activities (Peng et al., 1996). Reddy et al. (2004) reported that the accumulation of proline in response to drought stress was regulated by a rate limiting enzyme, PFC5, in higher plants. Previous studies conducted on different plant species showed varied response of proline accumulation to P application under drought stress. For example Liu et al. (2015) suggested that P application significantly decrease proline concentration in water-stressed Fargesia rufa; however, Al-Karaki et al. (1996) found that P application significantly increased proline accumulation in sorghum while the bean plants showed higher accumulation at low P level than at high P level. We found that P application neither decreased nor enhanced proline accumulation in drought-stressed plants. It clearly indicates that proline accumulation responses to P fertilization in drought-stressed plants are inconsistent, varying according to specific tolerance mechanisms and level of P applied. It is already understood that proline accumulation increase under drought stress in many plant species but not necessarily with P application.

Nitrogen is an important nutrient for plant growth as it is involved in the synthesis of chlorophyll, amino acids, nucleic acid, and proteins. Generally, water stress can reduce available N uptake, resulting in a decrease in the production of nitrogenous compounds (Lawlor and Cornic, 2002; Garg et al., 2004). Similarly, our results showed lower amounts of NH₄⁺, NO₃^-, and SP under limited water supply, regardless of P fertilization. Possible reasons for this decreased SP concentration in drought-stressed plants include an associated increased function of protease enzymes, proteolysis or decreased protein synthesis, as well as the lower P_n, i.e., less carbon to build any metabolite. Furthermore, our findings revealed that P application slightly up-regulated NH₄⁺ and NO₃^- levels in water-stressed plants and these may be attributed to changes in associated enzyme activities (Burman et al., 2004, 2009). Azcon et al. (1996) also observed a positive effect of P fertilization on the reduction and assimilation of nitrogenous compounds. It appears that P fertilization can enhance the drought tolerance of P. zhennan by accumulating osmoprotectors and enhancing nitrogenous compounds reduction and assimilation.

Response to drought is an inherent property of a plant, but it also depends on the length and severity of stress period. Long term drought stress causes a decline in the rate of photosynthesis, leading to an over-production of ROS (Foyer and Noctor, 2005; Carvalho, 2008). An increase in ROS level triggers protein degradation, lipid peroxidation, DNA fragmentation and may cause cell death (Apel and Hirt, 2004). ROS (O₂^⋅- and H₂O₂) production and MDA content were found to be significantly higher in drought-stressed plants compared with the well-watered, regardless of P application because of low photosynthetic rate and other physiological disruption within the cell. Induction of antioxidant enzyme activities is a general tolerance strategy to drought stress, as it helps plants to overcome oxidative stress and associated damage. The antioxidative enzyme SOD is responsible to dismutase O₂^⋅- into H₂O₂ in the chloroplast, mitochondrion, cytoplasm, and peroxisome, while POD and CAT play important functions in scavenging H₂O₂. Antioxidative (SOD, POD, and CAT) enzyme activities were also found to be significantly higher in drought-stressed plants compared with well-watered plants, irrespective of P application, suggesting a strong antioxidant defense mechanism in P. zhennan under drought conditions. These results indicate that antioxidative enzyme processing is substrate (ROS) inducible, leading to increased expression of genes encoding these enzymes. Reddy et al. (2004) reported that drought stress induced mRNAs corresponded to the genes of antioxidant enzymes. Similar results were also reported in other studies conducted on variety of plant species (Mittler, 2002; Murshed et al., 2013; Oliveira et al., 2014). In our study, P application resulted in slightly lower levels of O₂^⋅- and H₂O₂ under drought stress, while level of MDA was significantly lower. P application had no significant effect on antioxidant enzymes or their activities and remained higher under drought treatment. Our findings indicate P. zhennan has the potential to tolerate natural drought conditions and that P fertilization may have a positive role in maintaining the tolerance capacity and mitigating the effects of drought stress.