This study was conducted in a greenhouse at the Centro de Citricultura Sylvio Moreira, in Cordeirópolis, São Paulo, Brazil. The north-south-oriented greenhouse was covered with 150-μm thick transparent plastic and closed on all sides with an anti-aphid screen, with natural fluctuations of climatic conditions. Climate variables were monitored throughout the experiment using a weather station (Vantage Pro2; Davis Instruments, Hayward, CA, United States). Average, maximum, and minimum solar radiation in μmol m⁻² s⁻¹ were 914.9 ± 154.7, 1,664.5 ± 372.3, and 155.7 ± 76.7, respectively. Average, maximum, and minimum air temperature in °C were 22.6 ± 3.8, 31.8 ± 4.9, and 15.4 ± 3.2, respectively. Average, maximum, and minimum relative humidity (%) were 67.6 ± 9.5, 89.1 ± 6.5, and 40.6 ± 11.4, respectively.

Twelve-month-old sweet orange trees cv. Valência [Citrus sinensis (L.) Osbeck] grafted on citrumelo “Swingle” rootstock [Citrus paradisi Macfad. x Poncirus trifoliata (L.) Raf.] were obtained from a commercial nursery and transplanted into 12-L pots filled with a well-drained commercial pine-bark-based substrate (Terra do Paraíso®). The pots were wrapped in a reflective aluminized blanket to avoid substrate over-heating. At transplant, each tree had a single stem and nine leaves.

Fifteen days later, the trees were separated into two groups and either (i) bud grafted onto opposite sides of the main stem with two buds obtained from micrografted plants (healthy, –) or (ii) bud grafted onto opposite sides of the main stem with two budwoods obtained from CLas source plants maintained in our laboratory and previously tested for the presence of bacteria by real-time quantitative PCR (qPCR) (infected plants, +). In both groups of the plants, buds with lignified tissue presenting a length of ca. 4 cm were used.

The plants were pruned to allow for uniform growth of two branches per plant (first vegetative flush) 37 days after transferring them to the 12-L pots. Leaves (one per branch) closest to the main stem were tagged to measure central vein length. Three Soil Plant Analysis Development (SPAD) values were averaged in each leaf used for central vein measurements, using the SPAD-502 Chlorophyll Meter (Minolta Co. Ltd, Osaka, Japan). The number of leaves on each branch was also evaluated. Such morphological measurements were performed once a week for 13 weeks until the first vegetative flush of growth was considered completely expanded (127 days after transferring plants to the 12-L pots). At this point, six leaves of the first vegetative flush per individual plant (of a total of 16 plants) were sampled and used for DNA extraction and CLas quantification by qPCR as described below. One hundred twenty-seven days after transferring the plants to the 12 L pots, cycle threshold values (C_T) for HLB+ shoots were determined as indirect measures of CLas titer and were ca. 22.51 ± 0.55, showing that CLas inoculation was efficient (Ammar et al., 2011; Coletta-Filho et al., 2014; Canale et al., 2017).

All the pots were watered daily until saturation for 171 days. Thereafter (6 weeks after CLas quantification), both + and – plants were exposed to one of two water management treatments: (1) fully irrigated (FI): plants were irrigated 3 days a week to saturation; (2) deficit-irrigated (DI): plants were irrigated 3 days a week using 40% of the water applied to the respective FI treatment. The 40% deficit irrigation level was selected given the fact that we expected to induce mild physiological responses in the plants, as demonstrated by Pedroso et al. (2014). There were eight replicate trees in each of the four treatments (Figure 1). To determine the amount of water required to saturate the substrate, each FI pot was irrigated with a known volume of water to saturation. Outflow from the pots was collected in a graduated cylinder, and watering was stopped when water was observed to be leaking from the bottom of the pots. The difference between the volume of the drained water and the amount of water applied was the volume of water needed to saturate the substrate. The amount of water required to irrigate the DI plants was calculated as:

where V_DI is the volume of water required to irrigate the DI(+)- and DI(–)-treated plants, and V_FI is the average volume of water applied to either FI+ or FI–, respectively. The total water amount applied to each treatment was recorded. Volumetric substrate moisture was monitored every week at a depth of 0.15 m using a soil moisture probe (model number MO750; Extech Instruments, Nashua, NH, United States). Three values of volumetric soil moisture were averaged in each pot.

All the plants were fertilized monthly with a complete nutrient solution containing in g per pot: 3.4 Ca(NO₃)₂.4H₂O; 0.85 MgSO₄.7H₂O; 1.95 NH₄NO₃; 1.1 KH₂PO₄; 0.5 of NH₄SO₄; 0.5 of CaCl₂; 0.01 of NaMoO₄; 0.01 g of ZnSO₄; 0.01 of MnSO₄; 0.01 of CuSO₄; 0.02 of H₃BO₃; and 0.1 Fe-EDDHA containing 6% Fe. The volume and concentration of the nutrient solution were continuously adjusted according to each water management treatment for all the plants to always receive the same amount of nutrients, regardless of the volume of nutrient solution applied to each treatment.

A day after initiating water management treatments (172 days after transferring the plants to the 12-L pots), another two branches (second vegetative flush) were allowed to grow, one from the tip of each branch of the first vegetative flush (Figure 1). The same morphological measurements to characterize the first vegetative flush were performed in the second vegetative flush of growth. Therefore, one leaf per branch of the second vegetative flush that was most proximate to the first vegetative flush was tagged, so central vein length and SPAD measurements could be taken, as described above. Morphological measurements of the second vegetative flush were performed once a week for 11 weeks (225 days after transferring the plants to the 12-L pots) when all the leaves had completely expanded. At this point, the plants of all the treatments were assessed to measure a variety of physiological and growth traits, as well as to diagnose and quantify CLas in both fine roots and petiole of the leaves from the first and second vegetative flushes.

Leaf gas exchange measurements were taken on leaves of the second vegetative flush that were most proximate to the first vegetative flush. Light response curves of net photosynthetic rate (A_net) using a Li-Cor 6,800 (Li-Cor Inc., Lincoln, NE, United States) portable photosynthesis system were obtained at 08:00 and 11:00 a.m. and as described in Shapiro et al. (2004) taking into consideration the precautions of Pons and Welschen (2002). The system incorporated a CO₂ controller that was used to set the CO₂ concentration inside the leaf cuvette to 400 μl L⁻¹. A 6 cm²-cuvette was fitted with a red-blue light source. The A_net, stomatal conductance (g_s) and transpiration rate (E) at 800 μmolm⁻² s⁻¹ were determined from the light response curves with 24 levels of photosynthetic photon flux density (PPFD): 1,500, 1,200, 800, 500, 200, 100, 90, 80, 70 μmol m⁻² s⁻¹ and every five units between 70 and 0 μmol m⁻² s⁻¹. The 800-μmol m⁻² s⁻¹ light level was previously determined to be the light saturation point for citrus trees. The rate of respiration in the dark (R_dark) was taken as the value of A_net at 0 μmol m⁻² s⁻¹ PPFD.

The rate of respiration in the light (R_light) was estimated using the method originally described in Kok (1948), as the y-axis intercept of a first-order linear regression fitted to A_net–irradiance plots to measurements made over the 25–65 μmol m⁻² s⁻¹ irradiance range. All gas exchange data were corrected for an increase in intercellular CO₂ concentrations (C_i) with a decrease in irradiance, which can result in reduced rates of photorespiration and increased rates of carboxylation (Villar et al., 1994). The correction was applied by adjusting the R_light through iteration to minimize the intercept of photosynthetic electron transport (J) as a function of irradiance (Kirschbaum and Farquhar, 1987). J was calculated according to Farquhar and von Caemmerer (1982):

where Γ^* is the CO₂ compensation point in the absence of R_light (von Caemmerer and Farquhar, 1981; 38.6 at 25°C). The rates of oxygenation and carboxylation by ribulose-1,5-bisfosfato carboxilase oxygenase (Rubisco, EC 4.1.1.39 – V_o and V_c, respectively) were calculated at 800 μmol m⁻² s⁻¹ (light saturation point), according to Farquhar and von Caemmerer (1982):

The effects of varying atmospheric [O₂] or [CO₂] on oxygenation (V_o) and carboxylation (V_c) at each intensity of light used were also calculated according to Farquhar and von Caemmerer (1982):

and

Equations (5, 6) used the Rubisco kinetic constants (K_c = 404.9 μmol mol⁻¹¹; K_o = 278.4 mmol mol⁻¹) previously determined by Bernacchi et al. (2001) at 25°C, and calculated V_cmax as:

Γ ^* (the CO₂ compensation point in the absence of R_light) depends on the Rubisco specificity factor, O₂ partial pressure, and is calculated according to von Caemmerer and Farquhar (1981) for real leaf temperature:

In this study, we used the specificity presented in Cousins et al. (2010), ambient O₂ concentration, and temperature response function of Brooks and Farquhar (1985) to determine Γ^*. The values of V_o, V_c, and V_c:V_o ratio reported here were calculated at an ambient CO₂ concentration.

All the gas exchange measurements were taken at a relative humidity of approximately 45% by manipulating the amount of air passing through a drying column prior to entering the leaf cuvette. Air temperature in the Li-Cor cuvette was set to 30°C for all the measurements to avoid any influence of leaf temperature on gas exchange variables. In addition, the flow rate was kept at 500 μmol m⁻² s⁻¹.

Fast chlorophyll induction kinetics (measured for 1 s), typically known as OJIP, was measured with the Li-Cor 6800 (Li-Cor Inc., Lincoln, NE, United States) portable photosynthesis system. The measurements were taken between 06:00 and 7:00 a.m. after dark-adapting the same leaves used for gas exchange measurements overnight. Reflective aluminized foil was used to wrap the leaves to avoid any light exposure before the measurements. An induction flash intensity of 16,000 μmol m⁻² s⁻¹ was used to yield minimal (F₀) and maximal fluorescence (F_m) from which the maximum quantum yield of primary photochemistry was derived (Genty et al., 1989) as F_v/F_m.

The JIP test equations (Strasser and Strasser, 1995; Strasser et al., 1995, 2004) were applied to calculate: the effective antenna size of an active reaction center (RC) (ABS/RC); the maximal trapping rate of PS II (TR₀/RC); the electron transport in an active RC (ET₀/RC); the effective dissipation of an active RC (DI₀/RC); the electron transport probability (ET₀/TR₀); the quantum yield of electron transport (ET₀/ABS); the number of photons absorbed by an excited photosystem II (PSII) cross-section (ABS/CS₀); the maximal trapping rate in a PSII cross-section (TR₀/CS₀); the electron transport in a PSII cross-section (ET₀/CS₀); the fraction of active reaction centers per excited cross-section of the leaf (RC/CS₀); and the performance index (PI, energy cascade processes from the first light absorption event until plastoquinone reduction). To present the data in spider plots with the same scale for all parameters, the values obtained for the HLB– FI plants were considered as the standard (=1), whereas the values of the other treatments were plotted as the relative change in relation to this standard, based on real values.

Leaf water potential (Ψ) was measured immediately after leaf excision at pre-dawn (Ψ_pd) and midday (Ψ_md) according to Scholander et al., 1964, using a pressure chamber (model 1000; PMS Instrument Co., Albany, OR, United States). The drop (%) of Ψ_md in relation to Ψ_pd was also calculated (Ψ_drop). The measurements were performed on leaves adjacent to those used for gas exchange analysis (leaves of the second vegetative flush of growth).

Three leaves from the second vegetative flush from each tree (ca. 100 mg) were harvested at midday and added into liquid N₂ following the extraction of starch in 0.7 mol.L⁻¹ perchloric acid as described by Delatte et al., 2006. The insoluble material was washed once with water and three times with 80% (v/v) ethanol to remove residual soluble glucans and pigments. Starch in the insoluble fraction was digested with α-amylase (EC 3.2.1.1; Megazyme, Bray, Ireland) and amyloglucosidase (EC 3.2.1.3; Megazyme, Bray, Ireland). Glucose equivalents were determined by an enzymatic reaction using hexokinase (EC 2.7.1.1; Roche, Basel, Switzerland) and glucose-6-phosphate dehydrogenase (EC 1.1.1.49; Roche, Basel, Switzerland), which converts NAD to NADH in an equimolar ratio as described by Smith and Zeeman (2006). The increase in the absorption spectrum at 340 nm (OD340) was assayed in a spectrophotometer (Multiskan Go; Thermo Fisher Scientific, Waltham, MA, United States).

The supernatant was used to determine the concentration of soluble sugars. Glucose, fructose, and sucrose were quantified enzymatically according to Viola and Davies (1992) with modifications described by Thalmann et al. (2016). Briefly, 10 μl of the samples were added to 183 ml of 50 mmol L⁻¹ HEPES buffer (pH 7.5) containing 1 mmol L⁻¹ATP, 1 mmol L⁻¹ NAD, and 1 mmol L⁻¹ MgCl₂. To measure glucose, hexokinase (EC 2.7.1.1; Roche, Basel, Switzerland) and glucose 6-phosphate dehydrogenase (1.1.1.49; Roche, Basel, Switzerland) were used to convert glucose to 6-phosphogluconate with concomitant reduction of NAD to NADH, which was monitored spectrophotometrically (Multiskan Go; Thermo Fisher Scientific, Waltham, MA, United States) at 340 nm. Subsequently, phosphoglucoisomerase (EC 5.3.1.9; Roche, Basel, Switzerland) was added to determine the amount of fructose. Finally, invertase (EC 3.2.1.26; Sigma-Aldrich, St. Louis, MO, United States) was added to cleave sucrose into fructose and glucose. The further increase in OD340 represented sucrose.

Maltose was determined according to Revanna et al. (2013), with modifications described by Smirnova et al. (2017). A two-step assay was performed using a 96-well microplate. First, in a volume of 100 μl, the sample containing maltose, 100 mmol L⁻¹ citrate-NaOH (pH 6.5), and 2 U maltase (a-glucosidase, EC 3.2.1.20; Megazyme, Bray, Ireland) was incubated for 30 min. The absorbance at 340 nm was measured at the end of the first step with a spectrophotometer (Multiskan Go; Thermo Fisher Scientific, Waltham, MA, United States). Subsequently, a total of 100 μl of a solution containing 100 mM Tricine-NaOH (pH 7.8), 5 mmol L⁻¹ MgCl₂, 1 mmol L⁻¹ ATP, 1 mmol L⁻¹ NADP⁺, 0.5 U hexokinase, and 0.25 U glucose 6-phosphate dehydrogenase was added. The increase in the absorption spectrum at 340 nm specific for NADH was measured again.

For protein quantification and SOD activity, 0.4 g of fine leaf powder (from the second vegetative flush) was homogenized in 2 ml of 100 mmol L⁻¹ potassium phosphate buffer (pH 7.5), with 3 mmol L⁻¹ dithiothreitol, 1 mmol L⁻¹ EDTA, and 4% (w/v) PVPP (Capaldi et al., 2015). The suspension was centrifuged at 14,000 × g at 4°C for 35 min, and the supernatant was stored at −80°C for further analysis. Total protein content was determined using bovine serum albumin as a standard (Bradford, 1976). SOD activity was expressed as unities of SOD per mg of protein.

The activity of superoxide dismutase (SOD) in the spectrophotometer was determined with the method described by Giannopolitis and Ries (1975), in which the activity is measured by the ability of the enzyme to inhibit the photochemical reduction of the compound tetrazolium-nitroblue chloride (NBT). The reaction solution (3 ml) was composed of a 50- mmol L⁻¹ Na-phosphate buffer (pH 7.8), 75 μmol L⁻¹ NBT, 3 μmol L⁻¹ riboflavin, 13 mmol L⁻¹ methionine, 0.1 mmol L⁻¹ EDTA, and 50 μl of protein extract. The solution was added to glass tubes and illuminated with a 15 -W fluorescent lamp for 5 min. After the period of exposure, the solution was analyzed in a spectrophotometer model Genesis 10S UV-Vis (Thermo Fisher Scientific, Waltham, MA, United States) at 560 nm.

The measurement of H₂O₂ and lipid peroxidation content was performed on the same extraction, in which 0.3 g of frozen fine leaf powder (from the second vegetative flush) mass of the leaves was homogenized in 0.1% (w/v) trichloroacetic acid and 4% (w/v) polyvinylpolypyrrolidone (PVPP) centrifuged at 5,590 × g for 15 min at 4°C (Alexieva et al., 2001). For H₂O₂ content, the supernatant was mixed with a 100-mmol L⁻¹ potassium phosphate buffer (pH 7.0), and 1 mol L⁻¹ potassium iodide (1:1:4) and incubated at 4°C for 1 h in darkness and after 20 min at 25°C. The absorbance of the samples was measured at 390 nm (Multiskan Go; Thermo Fisher Scientific, Waltham, MA, United States) and calculated using a standard curve with known concentrations of H₂O₂. Lipid peroxidation was determined by the presence of malondialdehyde (MDA) according to Heath and Packer (1968). To the supernatant sample, 1 ml of a solution containing 20% (w/v) trichloroacetic acid and 0.5% (w/v) thiobarbituric acid was added and then incubated at 95°C for 30 min followed by quick cooling at 4°C to stop the reaction. The samples were re-centrifuged for 5 min at 12,100 × g, and the supernatant was measured at 535 and 600 nm (Multiskan Go; Thermo Fisher Scientific, Waltham, MA, United States). The absorbance of TBA reactive substances that formed was determined at 535 nm. The measurements were corrected for unspecific turbidity by subtracting the absorbance at 600 nm. Using an extinction coefficient of 155 mmol L⁻¹⁻¹ cm⁻¹, the amount of MDA was calculated.

Approximately 0.01 m of the tips of both branches of the second vegetative flush were dried in an air forced ventilation oven at 65°C for 72 h, ground in a mill grinder (Willy–MSSL-031). Thereafter, a range of ca. 2 mg of sample were weighed out in tin capsules. Thereafter, the carbon (¹³C/¹²C) isotope ratio of the samples was analyzed in an Isoprime isotope ratio mass spectrometer (Micromass, Wilmslow, United Kingdom) coupled to an elemental analyzer (Eurovector, Pavia, Italy). Isotopic C-ratio was calculated using the following standard δ notation:

where R = ¹³C/¹²C for carbon. The isotope ratios were calibrated against the international standards IAEA CH6 and IAEA CH7. δ¹³C results were referenced against Pee Dee Belemnite (PDB). Precision (the standard deviation of the set of standards analyzed in each batch) was 0.06‰.

Deoxyribonucleic acid extraction was performed according to the CTAB method originally described in Murray and Thompson (1980). Briefly, fine roots (ca. 2 mm) and the petiole of six fully expanded leaves were collected, cut into small pieces, and placed in 2-ml plastic tubes (200 mg each) containing 5-mm beads and 625 μl of buffer 1 (100 mmol L⁻¹ Tris [pH 8], 50 mmol L⁻¹ EDTA, and 500 mmol.L⁻¹ NaCl). The plant tissue was ground in a homogenizer TissueLyzer II (Qiagen, Germantown, MD, United States) at 30 Hz for 120 s, and then 725 μl of buffer 2 [5% CTAB, 10% Sarcosyl (Sigma-Aldrich, St. Louis, MO, United States), and 10 mmol L⁻¹ B-mercaptoethanol] was added to the tube. This was held at 65°C for 30 min and then centrifuged for 5 min at 3,500 rpm. The supernatant was transferred to a new 1.5-ml microtube, extracted with chloroform/isoamyl alcohol 24:1, and precipitated with isopropanol; the total DNA was suspended in 400 μl of 1/10TE + RNase (pancreatic ribonuclease, EC 3.1.27.5). DNA concentration was quantified by spectrophotometry at 260/280 OD ratio using the nanodrop system (Thermo Fisher Scientific, Wilmington, NC, United States). The DNA was standardized to a 100-ng μl⁻¹ concentration.

Real time quantitative PCR (q-PCR) was performed with a total volume of 13.5 μl [3 μl of total DNA, 6.75 μl of 2x Maxima® Probe/Rox qPCR Master mix buffer (Fermentas, St. Leon-rot, Germany), 0.4 μmol L⁻¹ of each primer, 0.2 μmol L⁻¹ of FAM /Iowa Black FQ label probe (Integrated DNA Technology, IDT, Coralville, IA, United States) using an ABI7500 (Applied Biosystems, Foster City, CA, United States) under default cycling conditions, totaling 40 cycles. The primers and probes used were based on the TS elongation fraction (efTs) of CLas (Hong et al., 2010). All the reactions used internal control primers (GAPDH gene) according to Boava et al. (2015). The results were expressed as the number of PCR cycles, in which reporter emission reaches a value higher than the threshold. The C_T values are inversely proportional to the concentration of bacteria in the sample, and it has been shown that C_T values lower than 36 are considered CLas-positive (infected, +) (Ammar et al., 2011; Coletta-Filho et al., 2014; Canale et al., 2017). Samples with C_T values between 33 and 36 were reread to certify the positive diagnosis. For the negative control, tissues of the healthy plants were used, with Milli-Q autoclaved (Merck Group, Darmstadt, Germany) water as the mock sample.

All the leaves were collected, and their area was measured using a portable area meter (LI-3000C; Li-Cor, Lincoln, NE, United States). Root volume was measured using a 2-L graduated cylinder containing a known volume of water (V1). The roots were then submerged into the water, and the new volume was recorded (V2). Root volume was calculated as the difference between V1 and V2.

The leaves, stems, and roots were dried at 65°C for 72 h and weighed (± 0.001 g) for dry weight determination. Leaf, stem, and root dry weight values were then used to calculate both the biomass allocation (%) and the relation between shoot (leaf dry weight + stem dry weight) and root dry weights (root: shoot ratio).

A complete randomized design was set up in a split plot scheme to assess the effect of HLB disease condition in citrus and its interaction with water management regime during plant growth. Two levels of HLB [healthy (HLB–) or infected (HLB+)] were the main plot factor, and two levels of management supply [fully irrigated (FI) or deficit-irrigated (DI)] were the subplot factor. Treatment combinations were repeated eight times, totaling 32 experimental units in this study. Dependent variables were tested by ANOVA followed by Student's unpaired t-test at 5% probability. Statistical methods were conducted using the software ASSISTAT 7.0 BETA.

HLB infection reduced growth. The HLB+ FI plants developed smaller leaf area and lower root volume, root: shoot ratio, and stem, shoot, and root dry weights than the HLB– FI plants (Table 1). These trends were not observed for the HLB+ DI and HLB– DI plants; therefore, no differences were observed between the two treatments for all variables shown in Table 1.

The use of DI reduced leaf and shoot dry weight values of both the HLB+ and the HLB– plants (Table 1). The HLB– DI plants also exhibited lower values of leaf area, root volume, and stem and root dry weights than the HLB– FI plants (Table 1). Nonetheless, no differences between the HLB+ FI and the HLB+ DI plants were observed for leaf area, root volume, and stem and root dry weight (Table 1). Moreover, the HLB+ FI plants exhibited a lower root: shoot ratio than HLB+ DI (Table 1). HLB– FI also showed a reduced root:shoot ratio compared with HLB– DI (Table 1). Phenotype images are shown in Supplementary Figures 1, 2.

The HLB+ FI plants required less water to saturate the substrate than the HLB– FI plants throughout the experiment (Supplementary Figure 3A). The HLB+ FI plants consumed an average of 16.6% less water than the HLB– FI plants (Supplementary Figure 3A). Similarly, less water was applied to the HLB+ DI plants in relation to those of HLB– DI (Supplementary Figure 3A).

Both Ψ_pd and Ψ_md were higher (less negative) in the HLB+ plants, regardless of water management treatment (Figure 2). On the other hand, DI reduced both Ψ_pd and Ψ_md, regardless of HLB infection conditions (Figure 2). Compared with the corresponding FI treatment, DI reduced the Ψ_md of HLB+ and HLB– by 28.7 and 56.3%, respectively (Figure 2B). Moreover, Ψ_drop was larger in HLB– DI, whereas HLB+ DI presented the lowest values of Ψ_drop (Figure 2B).

Higher A_net, g_s, and E were observed in the HLB– FI plants in relation to both HLB+ FI and HLB– DI (Figures 3A,C,D). Nonetheless, no differences between HLB+ DI and HLB– DI were observed for A_net, g_s, and E (Figures 3A,C,D). Regarding the HLB+ treatments, no differences were observed for A_net, whereas HLB+ FI presented higher values for both g_s and E in relation to HLB+ DI (Figures 3A,C,D). The DI plants were less depleted in ¹³C (higher δ¹³C, lower discrimination against ¹³C) than the FI plants (higher discrimination against ¹³C) (Figure 3B). HLB infection did not change δ¹³C, since no differences were observed between treatments with the same water management treatment (HLB+ FI x HLB– FI and HLB+ DI x HLB– FI; Figure 3B).

The HLB+ FI plants had higher values for both R_dark and R_light than for both HLB+ DI and HLB– FI (Figures 4A,B). DI reduced leaf respiration rates in both the healthy (HLB–) and the infected (HLB+) plants (Figures 4A,B). Overall, the FI plants exhibited higher V_c than the DI plants, regardless of infection conditions (Figure 4C). The HLB+ FI plants presented lower V_c and V_o rates than the HLB– FI plants (Figures 4C,D). Although DI reduced V_c in the HLB– plants, this trend was not observed between the HLB+ treatments (Figure 4C).

Chlorophyll a fluorescence traits associated with phenomenological fluxes, such as ABS/CS₀ and TR₀/CS₀, were higher in the HLB+ FI plants (Figure 5B). Nonetheless, the HLB+ FI plants also exhibited reduced values for ET₀/ABS and ET₀/TR₀ (Figure 5A). Although HLB infection did not deactivate reaction centers (refer to RC₀/CS₀ and TR₀/CS₀; Figure 5B), the HLB+FI plants presented higher energy dissipation rates per RC (DI₀/RC) and per CS₀ (DI₀/CS₀) (Figure 5). Changes in both DI₀/RC and DI₀/CS₀in the HLB+ DI plants in relation to the standard (HLB– FI) were lower (Figure 5). Thereby, the HLB+ plants presented reduced values for PI, especially under FI conditions (Figure 5B). Overall, we did not observe differences between HLB– FI and HLB– DI (Figure 5).

The HLB+ FI plants presented lower SOD activity, associated with higher contents of MDA and H₂O₂ in relation to both HLB+ DI and HLB– FI (Figure 6). No differences were observed between HLB+ DI and HLB– DI for MDA and H₂O₂ contents and SOD activity (Figure 6). The same was observed between HLB– FI and HLB– DI (Figure 6).

The HLB+ plants exhibited higher leaf starch content than the corresponding HLB– plants (Figure 7). Both DI treatments reduced leaf starch content in relation to the corresponding FI treatments (Figure 7). Thereby, both DI treatments had higher glucose, fructose, and sucrose contents than the respective FI treatments (Figures 8A–C). HLB+ DI presented higher maltose content than HLB+ FI, whereas the opposite was observed between the HLB– DI and the HLB– FI plants (Figure 8D). Although no differences were observed for glucose and fructose, we measured higher sucrose and lower maltose contents in HLB+ FI in relation to HLB– FI (Figure 8). Finally, the HLB+ DI plants exhibited higher sucrose and maltose contents than the HLB– DI plants (Figures 8C,D).

At 225 days after transferring plants to the 12 L pots, the CLas bacterial titer in the roots, estimated by C_T values, was not affected by water management treatments and was lower than the values observed in the leaves (Figure 9). Higher C_T values (i.e., lower bacteria content) were observed in HLB+ DI in both first and second vegetative flushes (Figures 9A,B). Nonetheless, C_T values were lower than 24 in both HLB+ FI and HLB+ DI treatments, which indicates high pathogen concentration.

We observed that the HLB+ FI plants grew less and consumed a reduced volume of water despite presenting higher volumetric substrate moisture in relation to the HLB– FI plants (Supplementary Figure 3). Such results indicate that the HLB+FI plants presented limited water uptake capacity. It has been reported that limitations to water movement in HLB-infected plants are associated to root growth impairment (Graham et al., 2013; Kadyampakeni et al., 2014; Hamido et al., 2016), as observed in this study (Table 1). Nonetheless, we demonstrated that physiological traits other than root growth are also involved in this process. Indeed, water transport in the HLB+ FI plants was limited not only by reduced growth but also by lower g_s and E (Figures 3C,D). Leaf area was also smaller in the HLB+ FI than in the HLB– FI plants (Table 1), which reduces total shoot transpiration (Galvez et al., 2011; Pedroso et al., 2014). Therefore, stomatal effects (Figures 3C,D), in combination with a reduced root system and leaf area (Table 1), decreased water uptake in the HLB+ FI plants (Supplementary Figure 3A). These results, associated with higher Ψ_pd and Ψ_md in HLB+ FI (Figure 2), strongly support that the HLB+ plants exhibit reduced hydraulic conductance compared with the HLB– plants (Ribeiro et al., 2009). Thus, substrate water content reduces slowly in plants with limited hydraulic conductance (Pérez-Pérez et al., 2009), as observed in the HLB+ FI treatment (Table 1; Figures 2, 3; and Supplementary Figure 3A).

Stomatal closure is likely a direct effect of the interplay between microorganisms and plant compounds secreted during the plant-pathogen interaction, but limited to epiphytic species and related to high concentrations of bacteria (Gudesblat et al., 2009). Considering this is not the case of CLas, a systemic pathogen directly introduced in the plant by the vector, we argue that the variations in stomatal conductance observed in our study occurred in parallel to changes in the hydraulic conductance in infected plants. Similarly, g_s decreased in citrus during the infection with xylem-limited bacteria Xylella fastidiosa, which caused xylem occlusion and restricted water flows to the leaves (Habermann et al., 2003).

Plants can undergo stress when water uptake of roots fails to compensate evapotranspiration rates (Fereres and Soriano, 2007), so we would expect that HLB– FI had an improved water status compared with HLB+ FI (Etxeberria et al., 2009; Johnson et al., 2014; Kumar et al., 2018). However, lower values of leaf area (Table 1), E, and g_s (Figures 3C,D) also reduced water loss in the HLB+ FI plants, mitigating the damages to water status in this treatment, as supported by the higher values of both Ψ_pd and Ψ_md in relation the HLB– FI plants (Figure 2). Such reductions in g_s and E were proportional to the decrease in A_net (Figure 3A). Therefore, we did not observe significant changes in δ¹³C (Figure 3B). There is a good relationship among δ¹³C, A_net, and g_s (Chaves et al., 2007), so δ¹³C has been used to determine the water use efficiency of plants grown under different environmental conditions (Farquhar and Richards, 1984; Busch et al., 2020). Considering that δ¹³C integrates photosynthetic activity and water relationships (Aranjuelo et al., 2009), throughout the period over which the tissue was synthesized (Salazar-Parra et al., 2015), we can assume that HLB infection did not affect water use efficiency (Figure 3D). In addition, respiration can contribute to the depletion of δ¹³C (Badeck et al., 2005; Chaves et al., 2007; Busch et al., 2020). Thus, despite the low g_s of HLB+ FI (Figure 3C), δ¹³C did not vary between FI treatments given the increase in both R_light and R_dark caused by HLB infection (Figures 4A,B).

Both DI treatments exhibited lower (more negative) Ψ_pd and Ψ_md than the respective FI treatments (Figure 2). In addition, the DI plants invested more energy in root than in shoot growth, since a higher root:shoot ratio was observed in relation to the respective FI treatment (Table 1). Higher investments in root growth in response to reduced water availability (Lei et al., 2006) can improve water uptake (Galvez et al., 2011). The negative effects of DI on plant water status were not intensified by HLB+ infection; but on the contrary, both Ψ_pd and Ψ_md were higher in the HLB+ DI than in the HLB– DI plants (Figure 2). Ψ_drop was larger in HLB– DI, whereas HLB+ DI had the lowest values of Ψ_drop (Figure 2B). Both DI treatments also had lower discrimination against δ¹³C (Figure 3B), which is a positive response of the DI plants since it represents improved water use efficiency (Pérez-Pérez et al., 2009; Panigrahi et al., 2014; Romero-Conde et al., 2014). Altogether, our results show that the irrigation of HLB-infected plants need to be accurately designed to aid the best irrigation management to HLB-infected trees, and that future research is required to determine the best irrigation level for HLB+ trees while also considering its effects on fruit quality and yield.

The HLB+ DI plants displayed higher C_T values in both the first and second vegetative flushes of leaf growth (Figures 9A,B). Considering that the first vegetative flush of growth had tested positive for HLB before the initiation of water management treatments, DI contributed to alleviate CLas population increase in relation to HLB+ FI. This response is likely associated with the high leaf respiration rates observed in the HLB+ FI plants (Figures 4A,B). Citrate, a subproduct of the TCA cycle, has been presumed to be a main source of energy for CLas (Cruz-Munoz et al., 2018), and high R values could accelerate CLas growth. Nonetheless, conditions that decrease both R_light and R_dark, such as deficit irrigation (Figures 4A,B), can contribute to hindering bacterial growth (Figures 9A,B). However, the HLB+ DI plants still presented high concentrations of bacteria, and no effects were observed in the roots (Figure 9).

A 9% reduction in CLas population was associated with less starch content in HLB+ DI in relation to HLB+ FI (Figure 7), even though the process associated with starch synthesis/breakdown in the HLB-infected trees has not been fully elucidated (Etxeberria et al., 2009; Gibon et al., 2009; Fan et al., 2010). Our results also provide evidence that both starch synthesis and breakdown are impaired in the HLB+ plants, but were partially alleviated by DI (Figure 7).

Higher starch content in HLB+ FI (Figure 7) was observed despite the highest R values (Figures 4A,B) and lower A_net and V_c in relation to HLB– FI (Figures 3A, 4C). R impacts plant carbon balance since a variable proportion of CO₂ assimilated by photosynthesis is released back to the atmosphere through respiration (Ayub et al., 2014). Changes in this process can result in substantial variation in C availability for plant maintenance and growth (Flexas et al., 2010; Griffin and Heskel, 2013). Taking into account that less C was available in HLB+ FI in relation to both HLB+ DI and HLB– FI (Figures 3A, 4A–C), most of the C was converted into starch in the HLB+ FI plants (Figure 7).

The accumulation of lower biomass in HLB+ FI in relation to HLB– FI would suggest that more starch was accumulated in HLB+ FI since less carbon was used to promote plant growth (Table 1; Figure 7). Although it can contribute to starch accumulation given the damages HLB causes on the transport of photosynthates to growing tissues (Etxeberria et al., 2009; Graham et al., 2013, Johnson et al., 2014; Kadyampakeni et al., 2014; Hamido et al., 2016), overall, biomass production did not differ between both HLB+ treatments (Table 1), so starch synthesis was upregulated in the HLB+ FI plants (Figure 7).

The HLB+ FI plants had lower maltose in leaves in relation to HLB– FI (Figure 8D), and the HLB+ DI plants exhibited higher sucrose and maltose content compared with the HLB– DI plants (Figures 8C,D). Since maltose is the main product of starch degradation (Zeeman et al., 2004; Fan et al., 2010; Aritua et al., 2013), it is clear that starch breakdown was also downregulated by HLB, which was alleviated by DI. Indeed, the HLB+ FI plants exhibited lower glucose, fructose, sucrose, and maltose content compared with the HLB+ DI plants (Figure 8).

Carbohydrate reserves decrease in storage organs, as plants are exposed to reduced water availability (Meltcafe et al., 2010). Because of reduced photosynthetic capacity under DI conditions, soluble sugars from starch degradation can be used to maintain plant metabolic activity (Dickson and Isebrands, 1991), so increases in soluble sugar contents were observed (Chaves et al., 2009; Pedroso et al., 2014; Puglisi et al., 2019), as shown in HLB+ DI (Figures 7, 8).

Nonstructural sugars, such as sucrose, are important signaling molecules and play important roles in adaptive mechanisms, such as sucrose induction of stress defense under reduced water availability conditions (Hanson and Smeekens, 2009; Ramel et al., 2009; Pinheiro and Chaves, 2011). The increase in glucose, fructose, sucrose, and maltose content in the HLB+ DI plants (Figure 8) could have triggered a series of protection events against impairments in the physiology of the HLB+ plants (Ramel et al., 2009; Pinheiro and Chaves, 2011). Although the HLB+ FI plants had lower soluble sugar contents in leaves (Figure 8), they also exhibited a reduced capacity to move electrons into the electron transport chain beyond QA⁻, since most of the absorbed (ABS/CS₀) and trapped (TR/CS₀) energy was dissipated (DI₀/CS₀ and DI₀/RC) (Figure 5). Even though this process was not associated with the deactivation of reaction centers, it was followed by reduced values of PI (Figure 5B).

This apparent blockage of electron transfer between the acceptors (Shu et al., 2012) may have damaged the initial stage of photosynthetic activity of a non-impaired RC complex (Strauss et al., 2006; Zlatev, 2009) in the HLB+ FI plants (Figure 5). Significant reductions of proteins associated with photosystem II stability have been reported in HLB-infected plants (Nwugo et al., 2013). Nonetheless, such changes were not observed in HLB+ DI plants (higher soluble sugar contents, Figure 8), such that trends in chlorophyll a fluorescence traits in the HLB+ DI plants were similar to those in the HLB– plants (Figure 5). Thus, our results strongly indicate that the photochemistry of citrus trees is resistant to DI (Pedroso et al., 2014; Romero-Conde et al., 2014). In addition, DI activated specific ROS-scavenging systems triggered by higher soluble sugar contents (Figure 8), which reduced oxidative damages and photoinhibition (Ramel et al., 2009; Pinheiro and Chaves, 2011) caused by HLB+.

The increase of photorespiration (V_o) in HLB+ DI over HLB+ FI (Figure 4D), supports an avoidance mechanism against oxidative damages and photoinhibition (Hochberg et al., 2013). The limitation of CO₂ assimilation rates (A_net, Figure 3A), accompanied by an increase in the activity of photorespiration (Figure 4D) as another sink for the absorbed energy, reduced the decline in electron transport (Harbinson et al., 1990; Wingler et al., 1999; Pinheiro and Chaves, 2011) for the HLB+ DI plants (ET₀/ABS, ET₀/TR₀, ET₀/RC; Figure 5). Such trends in V_o were followed by higher values of SOD activity in HLB+ DI compared with HLB+ FI (Figure 6A).

Since the activity of SOD metalloenzymes is responsible for alleviating oxidative stresses caused by ROS (Alscher et al., 2002; Foyer and Noctor, 2009), we also observed lower MDA and H₂O₂ contents in the HLB+ DI leaves in relation to the HLB+ FI plants (Figures 6B,C). Higher MDA values in HLB+FI in relation to HLB+DI plants (Figure 6B) is another indication that the ROS scavenging system was less efficient in alleviating oxidative stress (Dourado et al., 2014; Capaldi et al., 2015) caused by the disease. Although HLB caused a series of oxidative stresses in the plants (Figures 5, 6), the use of DI partially alleviated such damages.

Although HLB infection reduced growth, A_net, and g_s (Figures 3A,C), we observed that DI did not aggravate these negative effects, since there were no differences between the A_net of HLB+ FI and HLB+ DI (Figure 3A). Therefore, leaf area, root volume, and stem and root dry weights did not vary between both HLB+ treatments (Table 1). It is important to note that the higher carboxylation rates of HLB+ FI compared with HLB+ DI (V_c; Figure 4C), did not improve A_net (Figure 3A), so growth was not substantially modified (Table 1). Much of the carbon assimilated by HLB+ FI was likely consumed by R (Figures 4A,B). Nonetheless, the lower V_c and V_o values in the HLB+ plants (Figures 4C,D) are a strong indication of reduced Rubisco or total activity (Aranjuelo et al., 2005; Pinheiro and Chaves, 2011), and are likely due to chloroplast disruption caused by starch accumulation (Stetler et al., 2009; Fan et al., 2010), especially in the HLB+ FI plants (Figure 7).