The experiment was carried out at Maoxian Mountain Ecosystem Research Station (103°53′ E, 31°41′ N; 1820 m asl), Chinese Academy of Sciences in southwestern China. In March 2013, the healthy and uniform plants (2-year-old, height 40 ± 5 cm) of dwarf bamboo (F. rufa) were selected from the nursery at Wanglang National Nature Reserve, and then transplanted into 50 L pots filled with 25 kg homogenized topsoil from the experimental site. Each pot had one standard plant having 4–5 ramets. Thereafter, all plants were placed in a semi-controlled greenhouse with a day/night temperature range of 15–33 and 10–15°C and relative humidity of 50–85%, and watered regularly with water from a nearby stream. Four months after the transplanting, the drought treatments were initiated. The pots were divided into three groups for drought treatments. One group was kept well-watered (WW) as the control [80% relative soil water content (RSWC)] during the whole experiment, and the other two groups were subjected to moderate drought (MD, 50% RSWC) and severe drought (SD, 30% RSWC) respectively for 30 days. The RSWC of each treatment was controlled using the weight method (Xiao et al., 2009; Xu et al., 2009). During the experiment, pots were weighed every other day and rewatered to their respective target RSWC by replacing the amount of transpired water. Evaporation from the soil surface was prevented by enclosing the soil with plastic bags which were tied at the base of each plant. Thereafter, the drought treatments were watered regularly as control for 15 days for recovery. In each treatment, four replications, each including five pots, were used for our experiments. The youngest fully expanded leaves at the same developmental stage were used to analysis various physiological and biochemical parameters.

Leaf relative water content was determined as described by Galmés et al. (2007a) and calculated according to the equation: LRWC = [(FW – DW)/(TW – DW)] × 100, where FW is leaf fresh weight; DW is leaf dry weight after drying at 70°C for 48 h, and TW is turgid leaf weight after soaking in deionized water for 12 h at room temperature.

Gas exchange and chlorophyll fluorescence were simultaneously performed using a Li-6400 portable photosynthesis system equipped with a 6400-40 fluorometer chamber (LI-Cor, Inc., Lincoln, NE, USA). The net CO₂ assimilation rate (P_n), stomatal conductance (G_s), intercellular CO₂ concentration (C_i), light-adapted maximum (F_m′), minimum (F_o′) and steady-state fluorescence yield (F_s) were measured between 9:00 and 11:00 am. Environmental conditions in the chamber used for leaf measurements consisted in a photosynthetic photon flux density (PPFD) of 800 μmol m^-2 s^-1, a vapor pressure deficit of 1.0–1.5 kPa, a flow rate of 500 μmol s^-1, an air temperature of 25°C and ambient CO₂ concentration of 380 ± 5 μmol mol^-1, respectively. A quantum yield of PSII reaction center photochemistry (Φ_PSII) and photochemical quenching (q_p) were calculated as (F_m′ – F_s)/F_m′ and (F_m′ – F_s)/(F_m′ – F_o′), respectively (Silva et al., 2015). After the measurements of the light-adapted parameters, the leaves were darkened using leaf-clips for 40 min. Then a saturation pulse of 8000 μmol m^-2 s^-1 was applied for 0.8 s, the maximal fluorescence yields (F_m) and the intrinsic quantum efficiency of photosystem II (PSII) photochemistry (F_v/F_m) was recorded. Non-photochemical quenching was calculated as F_m/F_m′ – 1. Basing on Φ_PSII + Φ_NPQ + Φ_f,D = 1, the quantum efficiency of photochemical energy dissipation (Φ_PSII = 1 -F_s/F_m′), ΔpH- and xanthophyll-mediated thermal dissipation (Φ_NPQ = F_s/F_m′-F_s/F_m) and fluorescence and constitutive thermal dissipation (Φ_f,D = F_s/F_m) were calculated according to Hendrickson et al. (2004). The flux of energy dissipation via each process (J_PSII, J_NPQ, and J_f,D) was calculated by multiplying the respective quantum efficiency with irradiance and leaf absorption coefficient (α), respectively (Harley et al., 1992; Hendrickson et al., 2004). The α was calculated from the rate of the photosynthetic carbon reduction cycle and the fluorescence yield under non-photorespiratory (2% O₂) conditions according to the method of Miyake and Yokota (2000).

The flux of electron transport through PSII (J_PSII) was determined as described by Harley et al. (1992). The rate of oxygenation by Rubisco (V_o) was measued according to von Caemmerer and Farquhar (1981). The rate of carboxylation by Rubisco (V_c) was estimated following Miyake and Yokota (2000). Under atmospheric conditions, the electron fluxes in the two cycles can be expressed as J_c = 4 ×V_c and J_o = 4 ×V_o, respectively. An alternative flux (J_a) aroused by electrons that are not used by the carboxylation and/or oxygenation cycles on the total electron flux driven by PSII was estimated from J_PSII – J_c – J_o (Miyake and Yokota, 2000). O₂-dependent J_a was determined from the difference between J_a (21% O₂) and J_a (2% O₂). Then, O₂-independent J_a was measured from the difference between J_a and O₂-dependent J_a.

Pigments from xanthophyll cycle (V, violaxanthin; A, antheraxanthin; Z, zeaxanthin; L, lutein) were determined as described by Thayer and Björkman (1990). Briefly, fresh leaf tissue (0.3 g) was extracted with 80% acetone, filtered through a 0.45 μm membrane and quantified by HPLC (Waters 2695, USA). A Spherisorb C18 column (5 μm, 250 mm × 4 mm) was used with a flow rate of 1.5 mL min^-1. Elution was conducted with acetonitrile/methanol (75:25, v/v) and methanol/ethyl acetate (70:30, v/v) as the A and B mobile phase. The mobile phase gradient was used as follows: start with 100% A for 7 min, increase to 100% B within 2 min, and then maintained for 23 min. The column was re-equilibrated with 100% A for 5 min prior to the next injection. The 10 μL sample was injected, and the pigments were detected by absorption measurements at 445 nm. The de-epoxidation state (DEPS) of xanthophyll cycle pool (VAZ) was calculated as (0.5A + Z)/(VAZ). Chlorophyll a+ b (Chla+b) and total carotene (Car) were extracted in the same way and measured by spectrophotometry at 662, 645, and 470 nm, respectively (Sükran et al., 1998).

The producing rate of superoxide anion (O2•−) was measured by monitoring the nitrite formation from hydroxylamine in the presence of O2•− (Zhou et al., 2004). Fresh leaf tissue (0.2 g) was 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 sulfanilamide 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 of the aqueous solution was read at 530 nm.

Hydrogen peroxide was determined by monitoring the absorbance of the titanium-peroxide complex (Zhou et al., 2004). Fresh leaf tissue (0.2 g) was homogenized with 5 mL of acetone and centrifuged at 3000 g for 10 min. The reactive mixture contained 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 and centrifuged at 3000 g for 10 min. The resulting precipitate was washed five times with acetone and centrifuged at 10,000 g for 5 min. The precipitate was solubilized in 3 mL of 1 M H₂SO₄, and the absorbance was read at 410 nm.

Lipid peroxidation was estimated by analyzing MDA content according to the thiobarbituric acid (TBA) test. Fresh leaf tissue (0.2 g) was homogenized with 2 mL of 50 mM phosphate buffer (pH 7.8) and centrifuged at 12,000 g for 20 min. One milliliter of supernatant was mixed with 3 mL of 20% trichloroacetic acid (TCA) solution containing 2% TBA. The reactive mixture was heated in a water bath at 95°C for 30 min and centrifuged at 15,000 g for 10 min. The absorbance was read at 532 and 600 nm. The amount of MDA was calculated using an extinction coefficient of 155 mM^-1 cm^-1 (Heath and Packer, 1968).

Organelles were isolated from leaves by differential and density-gradient centrifugation, according to the method of Mittova et al. (2000) and Song et al. (2009). Fresh leaf tissue (10 g) was chopped using a blender (HR-2826, Philips, China) with five volumes of medium per g FW in a medium containing 50 mM HEPES (pH 7.5), 5 mM γ-caproic acid, 0.3% BSA, 0.3 M sucrose, 10 mM NaCl, 5 mM Na-AsA, 10 mM β-mercaptoethanol, 2 mM EDTA and 1% PVP. The homogenates were filtered through four layers of gauze. The crude chloroplast fraction was sedimented by centrifugation at 1000 g for 5 min, purified by a Percoll discontinuous gradient (10, 40, 70, and 90%) and then recentrifuged at 4700 g for 15 min. An intact chloroplast layer was obtained from between 40 and 70% Percoll fraction. The harvested supernatant at 1000 g was recentrifuged at 12,000 g for 15 min, and then the pellets were collected. The collected pellets were resuspended in a medium containing 20 mM HEPES-KOH (pH 7.5), 330 mM sorbitol, 10 mM NaCl, and 2 mM EDTA. In this isolation procedure, the harvested supernatant at 12,000 g was considered to be the cytosol fraction. The collected pellets at 12,000 g were fractionated by a sucrose discontinuous gradient (25, 37, 45, and 57%) at 68,000 g for 3.5 h, and then an intact mitochondrial layer was obtained from between 37 and 45% sucrose fractions. The intactness of isolated chloroplasts and mitochondria were detected using the ferricyanide method and Cyt c method (Song et al., 2009), and found in the range of 80–90 and 75–85%, respectively. Interactive contamination (%) of the isolated organelles was calculated by dividing the activity of the respective marker enzyme (CCO, mitochondria; CAT, peroxisomes or chlorophyll content, chloroplasts) in the isolated organelle by its total activity (or amount) in the whole sucrose gradient. Although slight cross contaminations (in the range of 3–9%) of chloroplasts and cytosol by mitochondria, mitochondria and cytosol by chloroplasts, and chloroplasts, mitochondria and cytosol by peroxisomes were found, they were within an acceptable range (Mittova et al., 2000), suggesting the organelles were well-isolated. Finally, these isolated organelles were used for the following enzymes analyses.

Superoxide dismutase (EC 1.15.1.1) activity was assayed as described by Giannopolitis and Ries (1977). The reactive mixture contained NBT solution, which consisted of 50 mM phosphate buffer (pH 7.8), 13 mM methionine, 63 μM NBT, 1.3 μM riboflavin, 0.1 mM EDTA and supernatant. One unit of SOD activity was defined as the amount of enzyme required to cause a 50% inhibition in the rate of p-nitro blue tetrazolium chloride reduction at 560 nm. Ascorbate peroxidase (EC 1.11.1.11) activity was estimated by monitoring the rate of AsA oxidation at 290 nm according to Nakano and Asada (1981). The reactive mixture contained 25 mM phosphate buffer with 0.1 mM EDTA (pH 7.0), 0.25 mM AsA, 1 mM H₂O₂ and supernatant. Glutathione reductase (EC 1.6.4.2) activity was determined by monitoring a decrease in absorbance at 340 nm caused by NADPH oxidation as described by Madamanchi and Alscher (1991). The reactive mixture contained 25 mM HEPES with 0.2 mM EDTA (pH 7.8), 0.12 mM NADPH, 0.5 mM GSSG and supernatant. Monodehydroascorbate reductase (EC 1.6.5.4) activity was measured by monitoring a decrease in absorbance at 340 nm due to NADH oxidation according to Arrigoni et al. (1981). The reactive mixture contained 25 mM HEPES with 0.2 mM EDTA (pH 7.8), 0.1 mM AsA, 0.5 unit AsA oxidase, 0.1 mM NADH and supernatant. Dehydroascorbate reductase (EC 1.8.5.1) activity was assayed by following the formation of AsA from DHA at 265 nm as described by Dalton et al. (1986). The reactive mixture contained 25 mM HEPES with 0.1 mM EDTA (pH 7.0), 0.4 mM DHA, 3.5 mM GSH and supernatant.

The reduced (AsA) and oxidized (DHA) ascorbate contents were determined according to the method of Law et al. (1983). Fresh leaf tissue (0.2 g) was extracted with 2 mL of 5% TCA and centrifuged at 15,000 g for 15 min. For total ascorbate (AsA + DHA) determination, 0.2 mL of supernatant was mixed with 0.5 mL of 150 mM phosphate buffer (5 mM EDTA, pH 7.4) and 0.1 mL of 10 mM DTT for 10 min and then with 0.1 mL of 0.5% N-ethylmaleimide. For AsA determination, 0.2 mL of supernatant was mixed with 0.5 mL of 150 mM phosphate buffer (5 mM EDTA, pH 7.4) and 0.2 mL of deionized H₂O. Color was developed in both reactive mixtures adding 0.4 mL of 10% TCA, 0.4 mL of 44% orthophosphoric acid, 0.4 mL of 4% 2, 2′-bipyridyl and 0.2 mL of 3% FeCl₃. Then, the mixtures were incubated at 40°C for 40 min and read the absorbance at 525 nm. DHA was calculated from the difference between AsA + DHA and AsA.

The reduced (GSH) and oxidized (GSSG) glutathione contents were measured following Zhou et al. (2007). Fresh leaf tissue (0.2 g) was extracted with 2 mL of 6% metaphosphoric acid and centrifuged at 12,000 g for 20 min. For total glutathione (GSH + GSSG) determination, the reactive mixture contained 1.6 mL of 100 mM phosphate buffer (pH 7.5), 0.1 mL of 0.6 mM DTNB, 0.1 mL of 0.2 mM NADPH, 0.1 mL of 50 U mL^-1 GR and 0.1 mL of supernatant and quantified at 412 nm. GSSG was assayed according to the same method after removal of GSH by 0.03 mL of 2-vinylpyridine derivatizations at 25°C for 1 h. GSH was estimated by subtraction of GSSG from GSH + GSSG.

All variables within the same stage (drought and subsequent rewatering) were subjected to a one-way ANOVA due to the different treatment times (30 and 15 days), and the means of four replicates were compared by Duncan’s test at P < 0.05 level. Before ANOVA, the data were checked for normality and homogeneity of variances, and when needed, log-transformed to correct deviations from these assumptions. Linear regression was used to investigate the relationship among CO₂ assimilation, lipid peroxidation, and energy partitioning. Statistical tests were performed using SAS 9.1 program (SAS Institute, Gary, NC, USA).

The LRWC significantly decreased in drought-stressed compared with WW plants (Table 1). Especially under SD condition, the plants showed more severe dehydration. After rewatering, LRWC of previously stressed plants returned to control level. Moreover, drought stress obviously decreased P_n, G_s, F_v/F_m, Φ_PSII, and q_P. Thereinto, P_n and F_v/F_m respectively decreased by 40.1 and 10.0% in MD plants as well as by 61.3 and 16.3% in SD plants. On the contrary, NPQ increased by 38.5% in MD plants and by 93.6% in SD plants. It is important to report that C_i was only enhanced in MD plants. After rewatering, P_n, G_s, C_i, Φ_PSII, and q_P in SD plants restored to MD levels, which were still lower than those in WW plants. Differently, F_v/F_m in SD plants was 3.9 and 7.4% lower than that in MD and WW plants, respectively, whereas NPQ in SD plants was 26.3 and 29.0% higher than that in MD and WW plants.

The J_PSII and J_f,D significantly decreased after drought stress, especially under SD condition (Figures 1A,C). In contrast, J_NPQ increased by 22.4% in MD plants and by 51.9% in SD plants compared with WW plants (Figure 1B). After rewatering, J_PSII and J_f,D in stressed plants fully recovered, whereas J_NPQ in SD plants was still 15.7% higher than that in WW plants (Figure 1).

For WW plants, J_c, J_o, O₂-dependent J_a and O₂-independent J_a accounted for approximately 53.6, 23.1, 18.2, and 5.1% of J_PSII, respectively (Figure 2). Drought stress significantly decreased the proportion of J_c and J_o, and increased the proportion of O₂-dependent J_a and O₂-independent J_a. For example, in MD plants, J_c, J_o, O₂-dependent J_a and O₂-independent J_a accounted for approximately 31.2, 9.4, 35.9, and 23.5% of J_PSII, respectively (Figure 2). Especially, the proportion of O₂-independent J_a in MD plants was 8.0% higher than that in SD plants (Figure 2D). After rewatering, the proportions of these parameters in J_PSII in stressed plants returned to control levels (Figure 2).

As compared with WW plants, the content of xanthophyll cycle pigments (VAZ) obviously increased in MD plants, but decreased in SD plants (Table 2). Moreover, DEPS remained constant in MD plants although both V and Z contents increased. In comparison, DEPS increased dramatically by 22.3% in SD plants mainly due to significant decrease in V content. The higher L content was observed in drought-stressed than in WW plants. Also, the content of photosynthetic pigments (Chla+b and Car) significantly decreased in SD plants. After rewatering, all these parameters were restored completely except higher Z content and DEPS in SD plants.

The higher levels of O2•− and H₂O₂ were detected in drought-stressed compared with WW plants, resulting in lipid peroxidation (given by MDA accumulation) (Figure 3). The MDA content increased by 30.2% in MD plants and by 105.4% in SD plants (Figure 3C). After rewatering, the levels of O2•− and H₂O₂ in stressed plants were still higher than those in WW plants (Figures 3A,B). Comparatively, MDA content in MD plants fully recovered, but its content in SD plants was 51.6% greater than that in WW plants (Figure 3C).

Drought stress induced a general increase in the activities of ROS-scavenging enzymes localized in the chloroplasts, mitochondria, and cytosol. However, the activities of ROS scavenging enzymes in stressed plants returned to control levels after rewatering (Figure 4). In the chloroplasts, SOD and APX activities in drought-stressed plants were greater than those in WW plants (Figure 4A). However, MDHAR, DHAR, and GR activities only increased in SD plants. All enzymes activities fully recovered after rewatering, except that MDHAR activity was lower in MD than in WW plants. In the mitochondria, the activities of SOD, APX, and DHAR in stressed plants were greater than those in WW plants (Figure 4B), whereas MDHAR and GR activities increased only in SD plants. After rewatering, SOD and GR activities in stressed plants were still higher than those in WW plants, while DHAR activity was lower than that in WW plants. In the cytosol, all enzymes activities in stressed plants were higher than those in WW plants (Figure 4C). After rewatering, their activities completely recovered, except that GR activity was lower in SD than in WW plants. Moreover, DHAR activity was much higher than MDHAR activity in all cellular fractions of stressed plants. The mitochondria of stressed plants had higher SOD, DHAR, and GR activities than those observed in the chloroplasts and cytosol fractions (Figure 4).

The content of antioxidants (i.e., AsA + DHA, AsA, GSH + GSSG, and GSH) was not affected by drought stress, except AsA + DHA and AsA in SD plants, which respectively increased by 40.3 and 33.6% compared with WW plants (Table 3). However, drought stress significantly decreased the redox states of ascorbate and glutathione (i.e., AsA/DHA and GSH/GSSG). After rewatering, only AsA content and GSH/GSSG in SD plants were still higher than those in WW plants.

Correlation analysis of P_n, MDA and energy partitioning (Figures 5, 6) showed that P_n was positively correlated with J_PSII, J_f,D, J_c, and J_o, and was negatively correlated with J_NPQ after drought stress (P < 0.001) (Figures 5A–E). P_n only had a positive correlation with J_c after rewatering (P < 0.05) (Figure 5D). Contrastingly, MDA was negatively correlated with J_PSII, J_f,D, J_c and J_o, and was positively correlated with J_NPQ after drought stress (P < 0.01) (Figures 6A–E). MDA only had a positive correlation with J_NPQ after rewatering (P < 0.01) (Figure 6B). Regardless of drought or rewatering phase, there were no significant correlations between P_n and J_a, as well as MDA and J_a (Figures 5F, 6F).

Photosynthesis is one of the most sensitive physiological processes to stressful environments such as drought and high temperature, and it is severely affected in all its phases by such stresses. Furthermore, PSII is considered to play an important role in the response of photosynthesis to environmental interferences (Tian et al., 2013). Depression of photosynthetic capacity is therefore found one of the key indicators of the decrease in PSII activity (Kong et al., 2015). Our study showed that the reduction of P_n in F. rufa plants under drought was primarily attributed to impairment of photosynthetic apparatus, especially PSII, as indicated by the decreases in F_v/F_m and Φ_PSII (a measurement of the functional status of PSII) with higher C_i (Table 1). Moreover, the decrease of P_n under drought might be partly responsible for stomatal closure, which was demonstrated by a decline in G_s with decreasing LRWC. The photosynthetic capacity of C₃ plants cannot be restored when G_s drops below 50–100 mmol H₂O m^-2 s^-1 (Flexas et al., 2004, 2006), and when P_n is reduced by over 80% (Cornic and Fresneau, 2002). With full recovery of LRWC, the reduced P_n in MD plants (<80%) improved substantially due to complete recovery of PSII activity and partial recovery of G_s after rewatering (Table 1). Thus, stomatal closure may account mainly for the delayed recovery of P_n in MD plants. In contrast, SD plants displayed the only partial recovery of photochemistry suggesting a persisting metabolic impairment, at least partially inactive PSII units. However, Liu et al. (2014) found complete recovery of P_n in Fargesia denudata plants from SD condition. Thus, different dwarf bamboo species have certain differences in the photochemical response to drought stress.

In our study, photosynthetic pigments in F. rufa leaves maintained constant levels under MD, but declined under SD; and recovered to normal levels after rewatering (Table 2). This corroborated previous studies on Arbutus unedo (Munné-Bosch and Peñuelas, 2004) and Jatropha curcas (Silva et al., 2012). The adjustment of photosynthetic pigments will contribute to some degree of photoprotection under SD condition. Niyogi (1999) speculated that Chl and Car change to balance the absorption and utilization of light energy when plants are subjected to drought. Moreover, photosynthetic pigments loss may be a regulatory mechanism geared to reducing the amount of energy absorbed by leaves during drought; thus, decreasing energetic pressure at the PSII level (Munné-Bosch and Alegre, 2000). The effect of drought on photosynthetic pigments may vary with plant adaptation to habitat (Galmés et al., 2007a).

Drought decreases a plant’s capacity to assimilate CO₂ thereby decreasing the demand for reducing equivalents (Wujeska et al., 2013), which creates an imbalance between the absorption and utilization of radiant energy that eventually results in excess excitation energy (Flexas and Medrano, 2002). Thermal dissipation (NPQ) involving the xanthophyll and presumably lutein cycles is one of the efficient strategies for the safe removal of excess energy. The capacity of thermal dissipation of F. rufa plants strongly increased while P_n was suppressed, especially under SD condition, as indicated by higher NPQ and DEPS (e.g., their correlation r = 0.85, P < 0.01) as well as more abundant L (Tables 1, 2). This can also be confirmed by increased J_NPQ (nearly half of the absorbed energy, 43–53%) in stressed plants concomitantly with decreases in J_PSII and J_f,D (Figure 1). In this case, decrease of F_v/F_m eventully occurred, which may be a consequence of drought-induced P_n decline rather than its cause because J_PSII (Galmés et al., 2007a). After rewatering, thermal dissipation was still operated largely under SD condition though F_v/F_m did not recover. This redistribution of absorbed energy under drought conditions helps protect the photosynthetic apparatus from photoinhibition and accelerate its recovery once drought is relieved.

In addition, the contents of xanthophyll cycle pigments (VAZ) also contribute to NPQ. Previous studies showed that the VAZ increases or remains constant in parallel with DEPS in different plant species subjected to varying drought intensities (Munné-Bosch and Peñuelas, 2004; Peñuelas et al., 2004; Gallé et al., 2007; Galmés et al., 2007a). In our study, the accumulation of VAZ-pool in F. rufa plants was found under MD condition (Table 2), as can facilitate NPQ induction and thus play a photoprotective function in this case. Interestingly, under SD condition and subsequent rewatering, we observed a decreased or constant VAZ-pool accompanied by an increase in DEPS that was attributable to activation of violaxanthin de-epoxidase (VDE) by the acidification of the thylakoid lumen. This is consistent with the experimental result observed in Pistacia lentiscus plants (Munné-Bosch and Peñuelas, 2003). Thus, the role VAZ-pool in NPQ induction may have a certain relationship with (rely on) the intensity of stress.

The reduction in J_PSII in F. rufa plants was lower than the decline in the capacity of CO₂ assimilation (P_n and J_c) under drought condition, suggesting an alternative sink such as the water–water cycle (Miyake, 2010). We observed an increase in O₂-dependent J_a that is driven by the water–water cycle, especially under MD condition (Figure 2C), and also O₂-dependent J_a/J_PSII was greater than the values previously reported for other C₃ plants subjected to drought (Biehler and Fock, 1996; Lovelock and Winter, 1996). These results suggest that the water–water cycle can effectively operate by F. rufa plants to dissipate excess excitation energy, although its function is limited under SD condition. In contrast, Driever and Baker (2011) demonstrated that the water–water cycle is not a major alternative electron sink for dissipation of excess excitation energy when CO₂ assimilation is restricted. Furthermore, A higher proportion of O₂-independent J_a in stressed plants, especially under MD, may be used as a candidate for nitrate assimilation or a cyclic flow of electron within PSII (Miyake and Yokota, 2000), but the specific nature of this alternative electron sink is not known.

Enhanced photorespiration also serves as a safety valve to dissipate excess excitation energy during mild to moderate drought when C_i and G_s (>150 mmol H₂O m^-2 s^-1) rather than Rubisco activity limit photosynthetic capacity (Guan et al., 2004; Galmés et al., 2007a; Silva et al., 2010, 2015; Abogadallah, 2011). However, lower J_o value and J_o/J_PSII in our study indicated that photorespiration is not a major energy dissipation strategy in stressed F. rufa plants, as observed results in some Mediterranean plants (Nogués and Alegre, 2002). Hence, whether photorespiration plays a protective role may depend on differential inhibition of photosynthesis under stress conditions.

Restriction in CO₂ assimilation under drought inevitably increases ROS in different processes of electron transport. Within plant cell organelles, the chloroplasts and mitochondria are the two main sites of ROS generation, while the cytosol acts as a sink for H₂O₂ leaked from other cellular compartments (Møller et al., 2007; Noctor et al., 2014; Kasote et al., 2015). The oxidative damage occurred in stressed F. rufa plants, especially under SD, as shown by obvious ROS (O2•− and H₂O₂) accumulation and lipid peroxidation (MDA) (Figure 3). Meanwhile, the activities of ROS-scavenging enzymes in isolated organelles were activated substantially (Figure 4). Wherein, the activities of SOD and the enzymes involved in AsA-GSH cycle (APX, DDHAR, MDHAR, and GR) in mitochondria exhibited the most significant increases, which partly decrease the ROS accumulation. Thus, the differential responses of enzymes in different organelles of F. rufa plants may display an novel solution for removing harmful ROS.

Furthermore, AsA and GSH can also detoxify ROS (Gallé et al., 2007; Silva et al., 2010), and their levels increased in reponse to drought-induced ROS accumulation (Silva et al., 2012, 2015). In our study, DHAR activity was much higher than MDHAR activity in isolated organelles suggesting that AsA is recycled mainly via GSH oxidation, and this is more apparent in mitochondria than in chloroplasts and cytosol. Accordingly, a gradual increase in AsA content was observed after drought although GSH remained constant, resulting in lower ratios of AsA/DHA and GSH/GSSG (Table 3). This suggests that AsA in F. rufa plants plays a certain role in decreasing oxidative damage. After rewatering, SD plants still kept high levels of AsA reiterating its role in preventing oxidative damage.

The capacity of CO₂ assimilation or lipid peroxidation is closely related to absorption and allocation of light energy in leaves (Zhou et al., 2004, 2007). Therefore, we conducted correlation analysis between P_n or MDA and different energy partitioning processes in F. rufa plants at different phases. Our study showed highly positive correlations between P_n and J_c or J_o at the drought and rewatering phase, but there were no significant correlations between P_n and J_a (Figures 5, 6). This result suggests that a drought-induced decrease in J_c was mostly compensated by J_a rather than by J_o, resulting in improvement of P_n upon rewatering. Hence, J_a can be used as a sink for excess electrons in stressed F. rufa plants. When P_n was restricted at the drought stage, the ROS levels and lipid peroxidation did not clearly indicate decreased photorespiration as evident from negative correlation between MDA and J_o. However, this is not surprising as oxidative stress is determined not only by photorespiration but also by other electron transport chains and overall changes in the redox status (Foyer and Noctor, 2005; Noctor et al., 2014). Moreover, highly correlations between P_n and J_NPQ as well as MDA and J_NPQ at the drought and rewatering phase showed that thermal dissipation could efficiently regulates energy utilization for CO₂ assimilation to alleviate oxidative damage.