In all experiments, potted grapevines (V. labrusca) cv. Niagara Rosada grafted on IAC-766 [Vitis riparia × (Vitis cordifolia × Vitis rupestris) × Vitis caribaea] rootstocks were used. The plants were grown in pots (7 L) containing sterilized substrate (clay soil, manure and, sand at a ratio of 1:1:1) and kept in a greenhouse at an average air temperature of 25°C. After bud break, the plants were reduced to a single stem, and each pot received 200 mL of water daily. The plants were fertilized monthly with 20 g of NPK (4:14:8) fertilizer.

Grapevine diseased leaves containing urediniospores of P. euvitis were immersed in 200 mL of distilled water and leaves abaxial surface were scraped with a brush. Spore concentration was determined using a hemocytometer and adjusted to 10⁵ urediniospores mL⁻¹. Serial dilutions were performed to obtain 10⁴, 10³ and 10² urediniospores mL⁻¹. Urediniospore suspensions of P. euvitis were sprayed on the abaxial leaf surface until run-off. Inoculated plants were kept in a dark inoculation room for 24 h under 100% RH and 25°C. A humidifier was used to reach 100% RH in the inoculation room. Maintenance of the inoculum was carried out through re-inoculation of urediniospores onto healthy plants every month. This method of inoculation was used in all experiments. Plants in the control treatment were sprayed with water and kept under the same conditions. In all experiments, the inoculations were performed 1 month after bud break, when plants had seven fully expanded leaves.

Two experiments were performed in a growth chamber (Conviron, Winnipeg, Canada) to determine how rust affects gas exchange of grapevine leaves. The conditions in the growth chamber were 25°C (± 2°C) with a photoperiod of 12 h, with photosynthetic active radiation (PAR) of 400 μmol m⁻² s⁻¹. Five treatments were applied in the first experiment: inoculations with 0, 10², 10³, 10⁴ and 10⁵ urediniospores of P. euvitis mL⁻¹. In the second experiment, three treatments were applied: inoculation with 0, 10,3 and 10⁵ urediniospores of P. euvitis mL⁻¹. The fourth fully expanded leaf (from the plant base) was used for evaluation of gas exchange in both experiments. Five and ten replications were used in the first and second experiments, respectively. The experimental design was completely randomized.

Different levels of inoculum were used to obtain a wide range of levels of disease severity. The evaluations were carried out twice a week on the same leaf area of 2 cm². The net CO₂ assimilation (A), stomatal conductance (g_s), intercellular CO₂ concentration (Ci), and transpiration (E) were estimated in diseased and healthy leaves using a portable infrared gas analyser (LI-6400XT, LI-COR Inc., Lincoln, NE, USA). The air CO₂ concentration (Ca) during the measurements was 400 μmol mol⁻¹, and gas exchange was measured under PAR of 800 μmol m⁻² s⁻¹. The leaf areas selected for evaluations of gas exchange were photographed, and digital images were processed with Quant software (Vale et al., 2001) to estimate the disease severity in each assessment. The pustules and the eventual yellowish or brown halo that surrounded the pustules were considered the diseased area.

Disease severity on the leaf areas assessed for gas exchange was related to the relative CO₂ assimilation rate by non-linear regression according to the model:

where Px is the net photosynthetic rate of a leaf with rust severity x, and Po is the average photosynthetic rate of healthy leaves. The β-value corresponds to the effect of disease severity on the green leaf area adjacent to the lesion. The data from gas exchange experiments described in the previous section were used to estimate the β-values.

Values of g_s, Ci, and E were transformed in proportions relative to the average value of each variable in healthy leaves (g_sx/g_so, Cix/Cio, and Ex/Eo) in both experiments. Bastiaans model (Equation 1) was also used to fit the data of g_sx/g_so and Ex/Eo (dependent variables) vs. disease severity (independent variable). Linear regressions were performed between relative Ci and disease severity. Statistica 6.0 software was used to perform non-linear regressions.

Two experiments were carried out to determine the photosynthetic limitations of plants infected with P. euvitis. Three plants were inoculated with P. euvitis at a concentration of 10⁴ urediniospores mL⁻¹, and three plants were sprayed with water. After inoculation, plants were kept at greenhouse at temperature of 25°C (± 2°C). Six evaluations were made, two per day, beginning on the 15th day after inoculation. The evaluations were carried out with a LI-6400XT equipped with a fluorimeter (6400-40, LI-COR Inc., Lincoln, NE, USA). Some photosynthetic variables were estimated from curves of photosynthesis response to increasing chloroplastic CO₂ concentration.

Photosynthesis measurements were begun with Ca of 400 μmol mol⁻¹ that was gradually reduced (250, 150, 100) to 50 μmol mol⁻¹ and then gradually increased (400, 600, 800, 1,400) up to 2,000 μmol mol⁻¹. Respiration (R) represents the intercept of the linear regression from initial values of the A/Ci curve. Mesophyll conductance (g_m) was calculated as:

where Γ* is the photosynthetic compensation point, i.e., the CO₂ concentration at which the photorespiratory efflux of CO₂ is equal to the CO₂ photosynthetic assimilation rate; and J is the transport of electrons from chlorophyll fluorescence assessments. The CO₂ concentration at the site of carboxylation in the chloroplast (Cc) was obtained with the following equation:

A/Cc curves were obtained and V_cmax and J_max were estimated (Farquhar et al., 1980; Sharkey et al., 2007; Flexas et al., 2008):

where K_c and K_o are the Michaelis-Menten constants of Rubisco for carboxylation and oxygenation, respectively, and O is the internal O₂ concentration, considered equal to the external O₂ concentration (Flexas et al., 2007). V_cmax and J_max were estimated by non-linear regressions with software STATISTICA 6.0.

From A/Cc curves, the A_max and g_smax parameters that corresponded to the maximum amount of CO₂ assimilation and the maximum stomatal conductance, respectively, were also obtained. Stomatal limitation (Ls) was calculated considering the A-values at Ca of 400 μmol mol⁻¹ and at Ci of 400 μmol mol⁻¹ (Farquhar and Sharkey, 1982). The average values of photosynthesis-related variables of healthy and diseased plants were compared by Student's t-test (P ≤ 0.05).

Two experiments were carried out in a greenhouse for the histopathological analyses of P. euvitis in cv. Niagara Rosada. Seven plants were inoculated with urediniospore suspensions of P. euvitis at a concentration of 10⁴ urediniospores mL⁻¹, and seven plants were sprayed with water.

Five diseased leaves and five healthy leaves were sampled at dawn of the 14th and 40th days after inoculation for light microscopy analyses. Inoculated and healthy samples were fixed in Karnovsky solution modified with phosphate buffer pH 7.2. During fixation, a vacuum pump was used to remove air from the tissues, and then samples were dehydrated in an ethanol series (10, 20, 30, 40, 50, 60, 70, 80, 90, and 100%) and infiltrated in hydroxyethyl methacrylate (Leica Historesin®), as recommended by the manufacturer. The blocks were sectioned at 5–7 μm thick in a rotational microtome (Leica RM 2045) with the aid of a steel blade (type C). Next, the slides were deposited on a hot plate at 40°C to dry and fix the sections on the blade. To detect starch grains, zinc-chloride iodine (Strasburger, 1913) was used on the sections on the blades that were mounted on the reagent itself. Anatomy images were taken by using a Leica® DC video camera attached to a Leica® microscope 300F (Leica Microsystems Heerbrugg GmbH, Heerbrugg, Switzerland).

An area of 25 cm² was removed from the same leaves used for anatomy to determine the starch content. The samples were dried with forced air circulation (65°C) until they reached constant weight. Subsequently, the starch contents in the leaf fragments were determined using the enzymatic method proposed by Amaral et al. (2007). The contents of starch in healthy and diseased leaves were compared by Student's t-test (P ≤ 0.05).

Samples of diseased leaves were collected 22 days after inoculation and immediately fixed in a buffer containing 3% glutaraldehyde and 0.2 M cacodylate, pH 7.25, for transmission electron microscopy analyses. The samples were air-dried in a vacuum chamber for 5 min, post-fixed in 1% osmium tetroxide for 2 h, dehydrated through a graded series of acetone solutions (30, 50, 70, 90, and 100%), and embedded in Spurr's resin. The blocks were sliced with a Leica UC6 ultramicrotome. The sections were treated using 5% uranyl acetate and 2% lead citrate for 30 min each for contrast (Reynolds, 1963). Electron micrographs were taken using a Gatan 830.J46W44 video camera attached to a Jeol transmission electron microscope (JEM-1011) at an acceleration voltage of 60 kV.

Two experiments were conducted, with five treatments: inoculation with 0, 10², 10³, 10⁴, and 10⁵ urediniospores mL⁻¹ of P. euvitis. The experimental design was completely randomized with five replications. The plants were kept in a greenhouse for 45 days after inoculation. Rust severity (Angelotti et al., 2008) and leaf area (Li-3100, LI-COR Inc., Lincoln, NE, USA) were estimated for all leaves of each plant. The stems, trunk and roots of each plant were collected and dried in an oven with forced air circulation (65°C) until they reached constant weight. The concentrations and total contents (concentration * dry mass) of carbohydrates in roots were determined. Soluble carbohydrates were extracted with a methanol:chloroform:water (MCW) solution, according to Bieleski and Turner (1966). The starch concentration was quantified in 10-mg samples from the insoluble fraction obtained from the soluble carbohydrate extraction. The contents of sucrose, total soluble sugars and starch were determined according to Dubois et al. (1956), van Handel (1968) and Amaral et al. (2007), respectively.

Linear regressions were performed on the relationships between leaf area and root biomass (dependent variables) and disease severity (independent variable). The relationships between the proportion of sucrose and starch contents (relative to the average of healthy plants) and disease severity were described by the power model

where y is the relative carbohydrate content, x is disease severity and a and b are function parameters. The relationship between relative total soluble sugar content and disease severity was described by a negative exponential model

where y is the relative sugar content, x is disease severity and a and b are function parameters. Statistica 6.0 software (StatSoft Inc., Tulsa, OK, USA) was used in all statistical analyses (Equation 7). Non-linear regressions were performed with STATISTICA 6.0 software.

The incubation period for P. euvitis was 7 days in all experiments. Rust severity reached 45 and 90% in experiments 1 and 2, respectively (Figure 1, Supplementary Figure 1). The A-values in healthy plants ranged from 8 to 15 μmol m⁻² s⁻¹, with an average of 12 μmol m⁻² s⁻¹. The g_s-values in healthy plants ranged from 0.06 to 0.61 mol m⁻² s⁻¹. The average values of Ci and E in healthy plants were, respectively, 245.6 μmol mol⁻¹ and 3.72 mmol m⁻² s⁻¹. High variability in the relationships between disease severity vs. gs, Ci, and E were observed (Figure 1). All regressions were significant, though the coefficients of determination were below 0.5. Slight decreases in relative values of g_s (g_sx/g_so) and E (Ex/Eo) were noticed when the severity of rust increased (Figures 1A,C). Both variables decreased according to equation 1 that estimated reductions for gs and E of 97 and 88%, respectively, for rust severity of 80% (Figures 1A,C). Relative Ci (Ci/Co) was directly proportional to disease severity, and an increase of 1.3% in disease severity corresponded to a 1% increase in intercellular CO₂ concentration (Figure 1B).

A steep decrease in the relative net photosynthetic rate was observed even at low rust severity, and null values of relative photosynthesis were observed in severities higher than 40% in experiment 1 and over 20% in experiment 2 (Figure 2). Low variability was observed in the relationships between disease severity and relative net photosynthetic rate, which was reflected by the high coefficient of determination (r² = 0.90 and standard error = 0.35). The value of the β parameter, estimated by the model of Bastiaans, was 5.78. The estimated reduction of the net photosynthetic rate was 46 and 73% for disease severities of 10 and 20%, respectively. According to the model, the relative net photosynthetic rate becomes zero when rust severity reaches 40%.

Plants infected with P. euvitis showed lower photosynthetic activity than healthy plants. V_cmax, J_max, g_m, g_smax, and A_max were significantly higher in healthy plants than in diseased ones (Table 1). The reduction in these parameters in diseased plants (average of disease severity = 7.2%) ranged from 35 to 67%, and g_m was the photosynthetic variable most affected by rust. There was no significant difference in Ls between healthy and diseased plants (Table 1).

Leaves infected with P. euvitis showed low starch contents in the mesophyll underlying the pustules (Figures 3A,B). However, there was a significant increase in the accumulation of starch in the chlorophyll parenchyma in regions adjacent to pustules (Figures 3A,C) compared to healthy leaves (Figures 3E–G). The starch concentration was significantly higher (P = 0.02) in diseased (33.1 mg g⁻¹) than in healthy (22.9 mg g⁻¹) leaves, which corresponds to an increase of 44% in starch concentration in leaves with rust symptoms. The spongy parenchyma of diseased leaves showed reduction of intercellular spaces due to cell hypertrophy surrounding the pustules (Figure 3D) when compared with the spongy parenchyma of healthy leaves (Figures 3E–G). Chloroplast degeneration due to disruption of plastid membranes was observed in chlorophyll parenchyma cells infected with P. euvitis (Figure 4). All degenerated chloroplasts were close to the pathogen haustoria (Figure 4).

Total leaf area of healthy plants ranged from 735 to 1712 cm². Total leaf area decreased linearly with increasing rust severity, due to defoliation (Figure 5A, Supplementary Figure 2). According to the linear model, an increase of 1% in disease severity accounts for 11% reduction in leaf area. A linear reduction in root dry matter with increasing rust severity was also observed, although the variability was higher than that in the relationship between disease severity and leaf area (Figure 5B, Supplementary Figure 3). At 80% disease severity, the root dry matter decreased from 18.51 to 4.11 g. No correlation was found between disease severity and dry matter of trunks or stems in grapevine (Figures 5C,D).

The average concentrations of total soluble sugars, sucrose, and starch in healthy plants were, respectively, 55.4, 43.2, and 118 mg.g⁻¹ of root dry matter, and the average contents of total soluble sugars, sucrose and starch were, respectively, 773.3, 659, and 2,850 mg per root (data not shown). The relative contents of these carbohydrates in roots decreased with increasing rust severity (Figure 6). The availability of sucrose in diseased plants (rust severity higher than 20%) was half the sucrose availability in healthy plants (Figure 6B). A reduction of 60% in starch availability was observed in plants with rust severities between 20 and 80% (Figure 6C).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer MJ and handling Editor declared their shared affiliation.