The experiment was conducted in August 2022 at the Tianfu New Area Experimental Station (30°27′N, 104°13′E; 500 m elevation) of the Sichuan Academy of Agricultural Sciences, China. The site experiences a subtropical monsoon climate characterized by humid summers. Based on data from the meteorological station near the experimental site, the mean annual climate condtitions over the past five years (2020–2024) were as follows: air temperature, 16.5 °C; relative humidity, 79%; sunshine duration, 1032.9 hours; and rainfall, 895.6 mm, the above data were retrospectively calculated to characterize the climatic context and were not available during the planning or execution of the experiment. Soil analysis revealed a medium loam texture with pH 6.97, organic matter content of 2.19%, and bulk density of 1.65 g·cm^-3.

Five-year-old ‘Shine Muscat’ (Vitis labruscana Bailey × V. vinifera L.) with uniform vigor were selected. Three trellis systems were established:

The trial was conducted following a completely randomized design with three replicates, where each replicate of the three trellis systems (VT, U-PT, HT) was established in an independent rain-shelter structure plastic film roofing and insect-proof mesh sidewalls, totaling nine structures. Each structure spanned 6 m (width) × 125 m (length), with 4.3 m ridge height and 2.5 m gutter height. Vines were planted in north-south oriented rows with 20 cm spacing between shoots on each side. All vines received uniform pruning, flower and fruit management, water-fertilizer regime, and pest and disease control, details are provided in the Supplementary Materials.

The experimental site experiences continuous high-temperature weather every summer. In this study, we define the first day in August with a daily maximum temperature exceeding 38°C as the start of the high-temperature period. On 2 August 2022 (defined as Day -6), the maximum temperature at the experimental site remained below 38°C according to weather forecasts. We therefore defined this day as the pre-stress reference timepoint and collected initial parameter data. On 8 August (defined as Day 1), when temperatures peaked at 39°C with meteorological projections of sustained heatwaves (>38°C), we initiated continuous investigation into plant responses to heat stress. Subsequent measurements were conducted at 3-day intervals (Days 3, 6, 9, and 15; 10, 13, 16, and 22 August respectively) until daily maximum temperatures subsided below 38°C. As shown in Figure 2, the pre-Day 1 period (before 8 August) had daily maximum temperatures below 38°C, while the high-temperature period (8–24 August) saw daily temperatures exceeding 38°C for at least 1 hour, with a peak temperature of 42°C.

Before experiment initiation, 10 uniformly growing vines in central rows of each rain-sheltered structure per treatment were tagged as fixed sample plants. The following sampling was performed within each structure: Six leaves at the 7^th node of newly developed shoots per treatment were selected for diurnal monitoring of photosynthetic parameters and chlorophyll fluorescence parameters at each designated timepoint (Day -6, 3, 6, 9, 15). During 12:00–14:00 at each designated time point, 10 leaves from the 6^th-8^th nodes per treatment were collected for chlorophyll content determination. Five 7^th-node leaves collected on Day -6 and Day 15 were analyzed for stomatal morphology and chloroplast ultrastructure. All samples were obtained from the same canopy orientation. Each treatment was replicated three times, meaning the above sampling was identically conducted in three independent structures.

For each trellis system, three representative canopy areas with uniform leaf distribution were selected. ZDR-U1W1S-T2 temperature and relative humidity (RH) loggers (Hangzhou Zheda Instruments Co., Ltd., Hangzhou, China; ± 0.5°C and ±3% RH accuracy) were mounted at the 4^th-5^th nodes of newly developed shoots (near fruit clusters) to record hourly data throughout August 2022 (1-31 August; total 744 hours). We calculated the daily and monthly mean, maximum, and minimum for both canopy temperature and RH. Additionally, we determined the cumulative durations and corresponding proportions (of the total 744-hour period) for the following parameters: temperatures ≥40°C and ≥45°C, RH within 60-80%, and RH >80%.

Leaf sunburn assessment was conducted on Day 15 according to the method of Li et al. (2011) with slight modifications. Briefly, ten pre-marked plants (constituting three biological replicates) per treatment were used. Six newly developed shoots per plant were randomly sampled to examine all expanded leaves. Sunburn severity was classified using a six-grade system based on thermal necrosis area percentage: Grade 0 (0%), 1 (≥0% and <5%), 2 (≥5% and <10%), 3 (≥10% and <25%), 4 (≥25% and <50%), and 5 (≥50%). Sunburn incidence rate and sunburn damage index were calculated as follows:

where a-f are Grade 0-5 leaf counts, and n is the total number of assessed leaves.

The observations by scanning electron microscopy were conducted by Wuhan Servicebio Technology Co., Ltd. (Wuhan, China), following their standard procedures (Servicebio, 2025; available online: https://www.service-bio.com/data-detail?id=4303&code=DJSYBG). Briefly, leaf segments (5 × 5 mm) adjacent to the primary veins were immediately fixed in 2% paraformaldehyde and 2.5% glutaraldehyde at 4°C. Samples were rinsed three times with 0.1 M phosphate buffer (PB, pH 7.4), post-fixed in 1% osmium tetroxide for 1-2 h, then rinsed three additional times with PB. Dehydration was performed through a graded ethanol series (30%, 50%, 70%, 80%, 90%, 95%, 100%, 100%; 15 min per step). Samples were transitioned through isoamyl acetate for 15 min, followed by critical point drying and gold sputter-coating.

Observations were conducted using an SU8100 field-emission scanning electron microscope (Hitachi, Tokyo, Japan). Stomatal density (number per mm²) and aperture ratio (open/total stomata × 100%) were calculated from 200× micrographs. Stomatal dimensions (length, width, aperture length, aperture width) were measured from 2500× micrographs. Quantifications were performed using Adobe Photoshop CC 2023 (version 24.0; Adobe Inc., San Jose, CA). For each treatment, five biological replicates were analyzed with three random fields of view per replicate.

Stomatal length was defined as the maximum distance between the poles of the guard cell pair. Stomatal width represented the maximum dimension perpendicular to the length axis. Aperture length was measured as the longest axis of the stomatal pore, with aperture width defined as the maximum dimension perpendicular to the aperture length axis.

The observations by transmission electron microscopy were conducted by Wuhan Servicebio Technology Co., Ltd. (Wuhan, China), following their standard procedures (Servicebio, 2025; available online: https://www.service-bio.com/data-detail?id=4300&code=DJSYBG). Briefly, leaf segments (2 × 5 mm) adjacent to the primary veins were immediately fixed in 2% paraformaldehyde and 2.5% glutaraldehyde at 4 °C. Samples were rinsed three times with 0.1 M phosphate buffer (PB, pH 7.4), post-fixed in 1% osmium tetroxide for 7 h, then rinsed three additional times with PB. Dehydration was performed through a graded ethanol series (30%, 50%, 70%, 80%, 95%, 100%, 100%; 1 h per step). Tissues were embedded in Spurr’s resin, transferred to microcentrifuge tubes, and thermally cured in a dry oven at 60 °C for 48 h. Embedded blocks were sectioned into 60-80 nm slices using an EM UC7 ultramicrotome (Leica Microsystems, Wetzlar, Germany). Sections were stained with 2% uranyl acetate for 8 min (protected from light), rinsed with ultrapure water, then counterstained with 2.6% lead citrate for 8 min in a CO₂-free chamber, followed by ultrapure water rinses. Samples were imaged using an HT7800 transmission electron microscope (TEM) (Hitachi High-Tech Corporation, Tokyo, Japan). Quantifications were performed using Adobe Photoshop CC 2023 (version 24.0; Adobe Inc., San Jose, CA) by measuring the length and width of chloroplasts and lipid droplets and by counting the number of lipid droplets per chloroplast. Five random fields of view per section were analyzed.

Chlorophyll content was determined according to the method of He et al. (2018) with slight modifications. Briefly, Leaf samples (100 mg fresh weight, FW) were dark-extracted in 10 mL 80% acetone for 24 h at 25°C. Absorbance measurements at 663 nm (A₆₆₃) and 645 nm (A₆₄₅) wavelengths were conducted using a UV-1800 UV-Vis spectrophotometer (Shimadzu Corporation, Kyoto, Japan). Chlorophyll contents were calculated as follows:

Where V is the extraction volume (mL), and W is fresh weight (g).

Leaf photosynthetic parameters, including net photosynthetic rate (P_n), stomatal conductance (g_s), transpiration rate (T_r), and intercellular CO₂ concentration (C_i), were measured at 2-h intervals from 08:00 to 18:00 using an LI-6400XT portable photosynthesis system (LI-COR Biosciences, Lincoln, NE, USA). The measurement procedure followed the method described by Rogiers et al. (2020). The leaf chamber was maintained at: 500 μmol·s^-1 airflow rate, 60% relative humidity, 400 μmol·mol^-1 reference CO₂ concentration, 25°C block temperature, and 1500 μmol·m^-2·s^-1 photosynthetic photon flux density (PPFD). The intrinsic water-use efciency (WUEi) was calculated according to Wang et al. (2020): WUEi=P_n/g_s.

Chlorophyll fluorescence parameters were recorded synchronously with gas exchange measurements using a PAM-2500 portable modulated chlorophyll fluorometer (Walz GmbH, Effeltrich, Germany). The measurement procedure followed the method described by Pollet et al. (2010). After the leaves were dark-adapted for 30 min, the minimum fluorescence (F₀) was first measured with a weak measuring light (~0.1 μmol·m^-2·s^-1); Subsequently, a saturating light pulse (>1500 μmol·m^-2·s^-1, 0.8 s duration) was applied to measure the maximum fluorescence (F_m). Afterward, the leaves were light-adapted under natural light for 30 minutes. Once the fluorescence signal stabilized, the steady-state fluorescence (F_s) was recorded. Another saturating pulse (>1500 μmol·m^-2·s^-1, 0.8 s) was then applied to determine the maximum light-adapted fluorescence (F_m′). The leaf was subsequently shaded for 3 seconds, followed by exposure to far-red light for 5 seconds to measure the minimum light-adapted fluorescence (F₀′). The following parameters were calculated according to Roháček (2002): maximum quantum yield of PSII [F_v/F_m = (F_m – F₀)/F_m]; effective quantum yield of PSII [√_PSII = (F_m′ – F_s)/F_m′]; photochemical quenching coefficient [qP = (F_m′ – F_s)/(F_m′ – F₀′)]; non-photochemical quenching coefficient [NPQ = (F_m – F_m′)/F_m′].

Data were analyzed by one-way ANOVA using SPSS 20.0 software (SPSS Inc., Chicago, IL, USA). For data collected across multiple time points, statistical comparisons were conducted in two ways: (1) Differences among the three trellis systems at each individual time point; (2) Differences across time points within each trellis system. In all cases, Duncan’s multiple range test was used for post-hoc comparisons at a significance level of p < 0.05. Prior to ANOVA, the assumptions of the model were evaluated. Homogeneity of variances was confirmed using Levene’s test (p > 0.05). Due to the limited number of biological replicates (n = 3), which reduces the statistical power of formal normality tests, the assumption of normality was assessed graphically by examining normal quantile-quantile (Q-Q) plots of the residuals (Blanca et al., 2017). As the data showed no severe deviations from normality and variances were homogeneous, ANOVA was considered robust and appropriate for analysis even for this sample size (Schmider et al., 2010). Correlation analysis and principal component analysis (PCA) were conducted on physiological parameters collected on Day 15. SigmaPlot 14.0 (Systat Software Inc., San Jose, CA, USA) generated graphical presentations. Correlation heatmaps and PCA plots were generated using the online platform Chiplot (https://www.chiplot.online/, accessed 11 April 2025).

Temperature monitoring during August 2022 (Figure 3A) showed that U-PT maintained more stable canopy temperatures with lower daily averages compared to other trellis systems. As detailed in Table 1, HT reached a maximum canopy temperature of 48.87°C, significantly exceeding U-PT by 3.84°C (p < 0.05). No significant difference in maximum temperature was observed between VT and U-PT (p > 0.05). Regarding cumulative heat exposure, HT and VT canopies endured ≥40°C temperatures for 14.07% and 13.78% of monitoring time respectively, equivalent to 1.86 and 1.82 times that of U-PT (7.57%) (p < 0.05 for both). More critically, durations ≥45°C reached 2.46% (HT) and 1.48% (VT), which were 18.92 and 11.38 times that of U-PT (0.13%) (p < 0.05 for both).

Further analysis of canopy RH (Figure 3B), conducted in August 2022, revealed a gradual decrease in humidity levels across the three trellis systems as the high-temperature stress progressed, followed by a gradual recovery after Day 15 (22 August) as canopy temperatures decreased. As detailed in Table 2, U-PT maintained the highest mean RH (68.31%), which was significantly higher than that of HT (67.54%) (p < 0.05), while VT showed an intermediate value (67.58%). In terms of humidity distribution, both U-PT and VT recorded a comparable percentage of time within the 60–80% RH range (approximately 25%) (p > 0.05), which was significantly longer than that of HT (20.68%) (p < 0.05 for both). Additionally, U-PT registered the highest proportion of time under >80% RH conditions (38.82%), significantly exceeding both VT and HT (approximately 36.4%) (p < 0.05 for both).

After 15 consecutive days of prolonged high-temperature stress (Figures 4A, B), all trellis systems showed leaf sunburn damage with distinct severity levels. Both VT and HT exhibited a higher number of sunburn-affected leaves with larger scorched areas. In contrast, U-PT retained fewer damaged leaves, mainly showing tip and margin scorching and curling.

Quantitative analysis (Figure 4C) confirmed HT had the highest sunburn incidence (53.45% leaves affected), significantly surpassing VT (42.82%) and U-PT (42.84%) (p < 0.05 for both). Severe damage (Grade 5) predominated in VT (20.80%) and HT (26.49%), followed by Grade 4 (8.61%, 15.33% respectively), with <15% low-grade injuries. U-PT displayed a bimodal distribution: 18.67% Grade 5 and 11.56% Grade 1. The sunburn damage index (Figure 4D) demonstrated U-PT’s superiority with the lowest value (26.57), showing 15.86% and 38.76% reductions compared to VT and HT respectively (p < 0.05 for both).

Figure 5 demonstrates significant morphological alterations in leaf stomata across three trellis systems following 15 consecutive days of high-temperature stress. Pre-stress observations revealed stomata protruding above the epidermis with turgid guard cells and expanded apertures (Figure 5, Day -6). Post-stress analysis showed stomata flush with the epidermis or sunken beneath it, accompanied by collapsed guard cells and substantially reduced aperture dimensions (Figure 5, Day 15 d). Epidermal anticlinal cell walls became more pronounced under heat stress.

As detailed in Table 3, all measured parameters (stomatal width, aperture length, aperture width, and aperture rate) exhibited significant reductions in all trellis systems after heat stress (p < 0.05), except for stomatal length in HT. Compared with their pre-stress values, VT, U-PT, and HT demonstrated respective decreases of 70.79%, 68.77%, and 76.80% in aperture length, with aperture width reductions reaching 79.91%, 76.02%, and 78.94%. Corresponding aperture rate declines measured 43.28%, 33.39%, and 45.17% respectively. Furthermore, U-PT maintained 16.25% and 21.09% higher aperture rates than VT and HT (p < 0.05 for both) respectively after heat stress. Notably, stomatal density remained unchanged relative to pre-stress levels (p > 0.05), with no statistical differences observed between trellis systems (p > 0.05), suggesting stomatal density may not be a key regulator in short-term heat adaptation.

Pre-stress observations revealed that all three trellis systems exhibited normal chloroplast architecture (Figure 6). The ellipsoidal chloroplasts adhered to cell walls and contained starch granules (S) and lipid droplets (L, osmiophilic granule aggregates) in the stroma. HT chloroplasts displayed notably larger lipid droplets and smaller starch granules compared to VT and U-PT even before stress imposition. As detailed in Table 4, the length of lipid droplets in HT was significantly greater than that in VT and U-PT by 58.59% and 80.46%, respectively, while the width was significantly greater by 39.56% and 58.76%, respectively.

Following 15 consecutive days of high-temperature stress, distinct ultrastructural responses emerged (Figure 6). VT and HT chloroplasts showed marked expansion with lipid droplets occupying most stromal space. HT chloroplasts progressed to spherical morphology with envelope integrity loss. and accompanied by complete starch granule depletion. Quantitative data (Table 4) on Day 15 showed that chloroplast length in VT and HT was significantly greater than in U-PT by 27.23% and 22.98%, respectively, and chloroplast width was significantly greater by 29.79% and 95.34%, respectively. The number of lipid droplets per chloroplast in VT and HT was significantly higher than in U-PT by 101.25% and 168.33%, respectively. Regarding lipid droplet size, no significant differences were observed between VT and U-PT. However, HT exhibited significantly larger lipid droplets than both VT and U-PT, with length greater by 12.65% and 27.21%, and width greater by 35.43% and 39.84%, respectively. In summary, in contrast to VT and HT, U-PT chloroplasts exhibited minimal volumetric expansion, alongside substantially smaller lipid droplets and larger-sized starch granules, indicating better structural integrity under prolonged heat stress.

No significant differences in chlorophyll content were observed among trellis systems at the pre-stress timepoint (Day -6) (p > 0.05; Figure 7). During summer high-temperature periods, Chl a and Chl b contents initially increased then declined in all trellis systems. VT and U-PT reached peak contents of Chl a and Chl b on Day 6, while HT peaked earlier on Day 3. From Day 6, chlorophyll retention ranked as U-PT > VT > HT. By Day 15, total chlorophyll decreased by 10.93% (U-PT), 14.80% (VT), and 18.79% (HT) relative to pre-stress levels. U-PT maintained 16.01% higher total chlorophyll than HT (p < 0.05), with VT showing intermediate values that were not significantly different from either U-PT or HT (p > 0.05 for both).

As shown in Figure 8, the diurnal patterns of P_n, G_S, and T_r were consistent across the three trellis systems, forming bimodal curves—with two exceptions: G_S in HT exhibited a “rise-fall-rise” trend from 8:00 to 18:00 on Day -6, Day 6, and Day 15; T_r in VT and HT showed a transient increase followed by a decrease on Day 3. On Day -6, the highest peaks of P_n, g_s, and T_r generally occurred at 12:00 for all trellis systems, except for G_S in HT leaves, which peaked at 10:00. From Day 3 to Day 15, the diurnal highest peaks advanced to 10:00 for all three parameters. Throughout the observation period, secondary peaks generally occurred at 16:00, while troughs occurred at 14:00. C_i in all trellis systems followed a declining-then-increasing diurnal trend, with minimum values at 12:00–14:00 from Day -6 to Day 3 and at 10:00–12:00 from Day 6 to Day 15.

From Day -6 to Day 15, P_n, g_s, T_r, and C_i across all three trellis systems exhibited a trend of increasing first and then decreasing, peaking on Day 3 or Day 6. During high-temperature periods, P_n followed the order U-PT > VT > HT, with differences between trellis systems progressively widening over time. By Day 15, the average P_n from 8:00 to 18:00 of U-PT, VT, and HT had decreased by 5.47%, 12.25%, and 36.42% respectively, relative to pre-stress levels. Additionally, the average P_n of U-PT was 14.21% and 76.22% higher than that of VT and HT, respectively (p < 0.05 for both). On Day 3 and Day 6, the average g_s, T_r, and C_i differed significantly among trellis systems, with values following HT > VT > U-PT (p < 0.05 for all comparisons). These disparities diminished over time: by Day 15, g_s and T_r showed no significant differences between U-PT, VT, and HT (p > 0.05 for all comparisons), except that the average C_i of HT was significantly higher than those of VT and U-PT (p < 0.05 for both).

Throughout the prolonged high-temperature stress period, the WUEi of leaves under the three trellis systems exhibited distinct dynamics (Figure 9). The WUEi of HT showed an overall decreasing trend, and by Day 15, it had decreased by 35.25% compared to the pre-stress level. In contrast, the WUEi of VT remained relatively stable over the course of the stress period. The WUEi of U-PT increased initially and then declined, peaking on Day 9 with an increase of 46.75% compared to the pre-stress level. By Day 15, it remained 25.06% higher than the initial value. Throughout the stress period, the WUEi values across the three trellis systems consistently followed the order: U-PT > VT > HT. On Day 15, there was no significant difference between U-PT and VT, (p > 0.05) but both were significantly higher than HT (p < 0.05 for both), with increases of 121.96% and 93.57%, respectively.

As shown in Figure 10, F_v/F_m and Φ_PSII exhibited minor diurnal variations across all trellis systems on Day -6 and Day 3, with only slight midday declines. As stress duration lengthened, both parameters across all trellis systems gradually declined: F_v/F_m developed a pronounced V-shaped diurnal pattern, reaching a minimum at 14:00, while Φ_PSII followed a “rise-fall-rise” pattern, peaking at 10:00 before declining to a trough at 14:00—a pattern consistent with qP diurnal variations observed from Day -6 to Day 9. By Day 15, qP transitioned to showing a morning maximum followed by a continuous decline to a 14:00 minimum. Notably, qP values transiently increased on Day 3 compared to pre-stress levels but declined thereafter as stress duration lengthened.

From Day -6 to Day 9, NPQ diurnal curves across all trellis systems exhibited an initial increase followed by a decrease, peaking between 14:00 and 16:00. However, by Day 15, NPQ trajectories shifted to a sustained upward trend. During the initial high-temperature stress (Day -6 to Day 3), the diurnal amplitudes of NPQ were relatively small. From Day 6, both the diurnal amplitudes and absolute values of NPQ increased progressively.

Comparative analysis identified HT as the most sensitive to heat stress. By Day 15, HT displayed significantly lower daytime-averaged F_v/F_m (0.62), Φ_PSII (0.19), and qP (0.32) compared to VT (0.73, 0.33, 0.48) and U-PT (0.76, 0.37, 0.52) (p < 0.05 for all comparisons). Conversely, HT’s average NPQ (1.05) exceeded both VT (0.92) and U-PT (0.85) (p < 0.05 for both). No significant differences in chlorophyll fluorescence parameters were observed between VT and U-PT (p > 0.05), except for NPQ (p < 0.05).

Pearson correlation analysis was conducted to evaluate relationships among 17 indicators, including leaf sunburn damage index, stomatal morphological traits, chlorophyll content, photosynthetic parameters, and chlorophyll fluorescence parameters (Figure 11). The results revealed positive correlations among five variables: sunburn damage index, g_s, T_r, C_i, and NPQ. Conversely, these five variables exhibited negative correlations with 12 other parameters, such as stomatal length, stomatal aperture ratio, chlorophyll content [Chl a, Chl b, and Chl (a+b)], P_n, F_v/F_m, Φ_PSII, and qP, while the latter group showed mutual positive correlations. Notably, F_v/F_m exhibited significant (p < 0.05) to extremely significant (p < 0.01) correlations with stomatal aperture, chlorophyll content, and photosynthetic and chlorophyll fluorescence parameters. This confirms that F_v/F_m can reflect plant heat tolerance by integrating information on stomatal status and photosystem stability.

Principal component analysis (PCA) was conducted to comprehensively evaluate the heat tolerance of different grape trellis systems under high-temperature stress, using the 17 indicators from the previous correlation analysis. The first four principal components (PCs) collectively accounted for 92.59% of the total variance (eigenvalues: 11.44, 2.18, 1.07, and 1.06) (Figures 12A, B). Notably, the first two PCs (PC1 and PC2) explained 67.32% and 12.79% of the variance respectively, with a cumulative contribution rate of 80.11% (Figure 12A), indicating their sufficiency in explaining dataset variability. In PC1, the sunburn damage index, g_s, T_r, C_i, and NPQ exhibited negative loadings (λ), whereas the remaining 12 variables showed positive loadings (Figure 12C). Among these, the sunburn damage index, P_n, F_v/F_m, Φ_PSII, Chl (a+b), and qP contributed most significantly to PC1 (absolute λ > 0.9). In contrast to PC1, PC2 showed distinct loading patterns: stomatal width, Chl a, P_n, T_r, F_v/F_m, Φ_PSII, and qP had negative loadings, while the remaining ten variables exhibited positive loadings (Figure 12C). Stomatal length and stomatal aperture width demonstrated the highest loadings on PC2 (λ > 0.8).

PCA-derived scores revealed distinct heat tolerance profiles among trellis systems. While U-PT and VT exhibited minimal separation in the PC score space, both diverged markedly from HT (Figure 12D). Following Peng et al. (2023) ‘s methodology, composite scores were calculated for each trellis system based on PC1 and PC2 (Table 5), where higher scores indicated superior heat tolerance. The heat tolerance ranking was: U-PT (1.82) > VT (0.94) > HT (-0.42), confirming U-PT’s enhanced capacity to maintain photosynthetic stability under high-temperature stress.

Under the dual pressures of global warming, high-temperature stress in vineyards has emerged as a critical challenge for sustainable viticulture. Trellis systems, as the structural framework of grape cultivation, can significantly influence plant heat tolerance (Liu et al., 2015; Gladstone and Dokoozlian, 2003). This study investigated the influence of trellis systems on the heat tolerance of ‘Shine Muscat’ grapevines.

In the rain-shelter cultivation, the top film blocks the outward escape of long-wave radiation emitted by the soil surface and plants (Chase et al., 1999). Hot air rises and accumulates at the top of the shelter, and heat dissipation can only rely on slow horizontal gas exchange between the sides of the shelter (which remain uncovered, unlike the film-covered top) and the outside environment (Lim et al., 2017). This low heat dissipation efficiency further exacerbates high-temperature stress inside the shelter during summer.

Our study demonstrated that different trellis systems exhibited significant differences in canopy temperature during high-temperature periods. Among them, HT had the highest canopy temperature and the most severe leaf sunburn damage. Its horizontal canopy is not only inherently prone to absorbing solar radiation and thus heating up, but more critically, it exacerbates the inherent structural limitations of the rain-shelter: the dense horizontal canopy intercepts solar radiation transmitted through the top film, and this radiation absorption—coupled with the heat-retention effect of the top film—collectively induces severe heat accumulation above the canopy (Florence, 2017). Despite the restricted air movement, the extreme heat under HT likely elevated the saturation vapor pressure deficit, which can intensify transpirational water loss from leaves while simultaneously reducing the relative humidity of the canopy air mass. VT showed intermediate canopy temperatures between HT and U-PT. Its V-shaped canopy helps reduce solar radiation interception per unit leaf area at noon, and the open canopy structure facilitates vertical convection. However, due to its smaller canopy spread angle (60°) and lower trunk height (1.0 m), horizontal air circulation remains restricted (Reynolds and Vanden Heuvel, 2009), thus leading to its higher canopy temperature compared to U-PT.

U-PT exhibited superior microclimate regulation performance. It was characterized by the lowest daily mean canopy temperature, minimal temperature variability, and the highest mean relative humidity among the trellis systems. Moreover, it endured a markedly shorter cumulative duration of extreme high temperatures (≥45°C), accounting for only 5.28% and 8.78% of that in HT and VT, respectively. This advantage is attributable to its unique “high trunk-wide opening-pendulous shoot” architecture: the elevated trunk (1.5 m) and wide canopy opening angle (130°) significantly enhance both horizontal and vertical air circulation within the rain-shelter, while the pendulous new shoots effectively reduce direct solar radiation interception. Correspondingly, U-PT showed the lowest leaf sunburn damage index among all tested trellis systems. Collectively, U-PT confers significant advantages for rain-shelter cultivation, making it particularly well-suited for grape-growing regions with high heat accumulation risks.

Stomata, as specialized epidermal gas-exchange channels in plants (Hetherington and Woodward, 2003), directly regulate leaf photosynthetic-transpiration coupling efficiency through their size, aperture, density, and spatial distribution (Chang et al., 2023). After 15 consecutive days of high-temperature stress, significant reductions in stomatal width, aperture length, aperture width, and aperture ratio were observed across all three trellis systems, while stomatal density remained unchanged. Short-term heat exposure typically induces stomatal opening to enhance transpirational cooling (Santanoo et al., 2022), whereas prolonged heat stress triggers stomatal closure or aperture reduction to balance transpirational water loss and leaf temperature (Cheng et al., 2023; Sarwar et al., 2019). Divergent reports on high-temperature effects on stomatal density—ranging from no significant impact (Yang et al., 2024) to increases (Jumrani et al., 2016) or decreases (Yang et al., 2021)—likely reflect species-specific genetic responses and experimental variability. Stomatal number, size, and spatial distribution define the limits of physiological regulation—higher density and larger apertures generally indicate enhanced heat tolerance (Driesen et al., 2020; Haworth et al., 2021). Despite the absence of inter-trellis differences in stomatal density or aperture dimensions post-stress, U-PT exhibited a significantly higher stomatal aperture ratio than VT and HT (p < 0.05), demonstrating its superior capacity for stomatal regulation under prolonged heat stress.

High-temperature stress exerts profound impacts on chloroplast structure and functionality. Excessive reactive oxygen species (ROS) accumulation under heat stress inhibits chlorophyll synthesis (Mustafa et al., 2021; Lim et al., 2007) and promotes its degradation (Mattila et al., 2018), accompanied by disintegration of the chloroplast envelope and thylakoid structures (Zhao et al., 2022), collectively driving chloroplast dysfunction. However, our study revealed transient increases in Chl a and Chl b contents during initial high-temperature stress across all trellis systems. We hypothesize that timely activation of antioxidant systems (such as SOD and APX) effectively scavenges early-stage ROS, delaying chlorophyll degradation while upregulating chlorophyll biosynthesis-related genes to compensate for photosystem damage-induced declines in light-harvesting efficiency. This phenomenon is consistent with chlorophyll dynamics observed in the herbaceous peony (Paeonia lactiflora) cultivars ‘Bo Baishao’ and ‘Fenyunu’ under natural high-temperature conditions (Wang et al., 2022).

Divergent responses emerged among trellis systems under prolonged heat stress. VT and U-PT maintained later chlorophyll content peaks (Day 6) with smaller subsequent reductions, whereas HT exhibited earlier peaking (Day 3) followed by rapid depletion, suggesting insufficient ROS scavenging efficiency leading to premature chlorophyll degradation in HT leaves (Wang et al., 2024). Chloroplast ultrastructure observations corroborated these findings: following 15 consecutive days of high-temperature stress, both VT and HT chloroplasts displayed abnormally swollen morphology, along with substantial lipid droplet accumulation—hallmark products of chloroplast membrane lipid peroxidation (Arzac et al., 2022; Jalil et al., 2023). Notably, HT also exhibited ruptured chloroplast envelopes. Prior to stress (Day -6), HT chloroplasts already showed larger lipid droplets and smaller starch granules compared to other systems (Figure 3), likely due to sufficiently high pre-stress canopy temperatures that caused early cellular damage. This precondition and the higher canopy temperature predisposed HT to more severe peroxidative damage upon subsequent heat exposure. In contrast, U-PT chloroplasts maintained structural integrity and significantly lower lipid droplet accumulation compared to other trellis systems.

Our data showed that prolonged heat stress triggered consistent responses in photosynthetic parameters (P_n, g_s, T_r, C_i) across all trellis systems, mirroring chlorophyll content dynamics with initial upregulation followed by progressive decline. The transient photosynthetic enhancement during initial high-temperature stress likely stemmed from short-term acclimation mechanisms, including increased stomatal conductance to enhance transpirational cooling and transient chlorophyll biosynthesis activation. This pattern aligns with observations by Dou et al. (2021) conducted in five grape cultivars, which showed temporary increases in P_n, g_s, T_r, and C_i under initial heat stress. However, prolonged high-temperature stress ultimately led to a reduction in photosynthetic rate during the later stages.

Under heat stress, the reduction in P_n can be attributed to both stomatal and non-stomatal limitations (Luis et al., 2016). In this study, reductions in g_s and C_i across all trellis systems indicated stomatal involvement in the P_n decline. However, severe chloroplast damage in VT and HT leaves suggested additional non-stomatal limitation, especially in HT. Analysis of WUEi further elucidated the predominant photosynthetic limitation in each system. The continuous decline in WUEi in HT indicated a greater reduction in P_n than in g_s, confirming non-stomatal limitation as the primary constraint. In contrast, stable WUEi in VT reflected a proportional decline in P_n and g_s, consistent with stomatal limitation. Notably, U-PT showed a substantial increase in WUEi, indicating maintained photosynthetic capacity under more efficient stomatal regulation. Thus, the P_n decline in U-PT likely resulted from a proactive stomatal strategy, protecting the photosynthetic machinery from severe heat damage.

Chlorophyll fluorescence analysis further validated these findings. F_v/F_m, a key probe for reflecting the degree of environmental stress, quantifies the potential maximum photochemical efficiency of the PSII reaction center (Doğru, 2021) and exhibits a positive correlation with plant heat tolerance (Wang et al., 2022), remaining stable under non-stress conditions (EI-Hendawy et al., 2019). qP represents the proportion of absorbed photons utilized for photochemical electron transport, while NPQ represents the proportion of absorbed photons dissipated as heat (Xu et al., 2020). In this study, F_v/F_m and NPQ of three trellis systems remained stable during initial stress, indicating an intact PSII reaction center. Concurrently, qP increased, consistent with trends in chlorophyll content and P_n. As heat stress continued, however, NPQ rose, suggesting plants mitigated photodamage by dissipating excess photons as heat—a regulatory pattern consistent with findings in lettuce seedlings (He et al., 2022) and squash seedlings (Qin et al., 2010). Across all trellis systems, qP, F_v/F_m, and Φ_PSII progressively declined in later stages of stress, indicating damage to the PSII reaction center and impairment of light energy conversion efficiency.

It should be noted that gas exchange measurements conducted under a standardized 25°C condition, an optimal temperature for photosynthesis (Uri et al., 2015), primarily reflect the potential photosynthetic performance of leaves after exposure to the distinct thermal environments induced by different trellis systems, rather than the in-situ photosynthetic rates of grapevines under actual field heat stress. The superior P_n and g_s values exhibited by U-PT under this controlled, standardized condition underscore its enhanced capacity to retain higher photosynthetic physiological potential following heat stress.

In our study, under identical high-temperature stress durations, U-PT and VT maintained higher F_v/F_m, Φ_PSII, qP, and P_n relative to HT. While U-PT showed no significant differences in F_v/F_m, Φ_PSII, or qP compared to VT (p > 0.05), it exhibited a higher P_n (p < 0.05). The PCA confirmed U-PT as the most heat-tolerant trellis system.

These photosynthetic superiority may be attributed to enhanced intrinsic acclimation mechanisms. In our parallel study conducted under identical experimental conditions (Luo et al., 2025), we found that the activities of key antioxidant enzymes—superoxide dismutase (SOD), peroxidase (POD), and catalase (CAT)—in all three trellis systems showed an initial increase followed by a decrease. However, U-PT exhibited a delayed activity peak, and its overall enzyme activity was higher than that in the VT and HT throughout the stress period, which was consistent with higher expression levels of the corresponding genes (VvSOD and VvCAT). Consequently, U-PT accumulated significantly lower levels of ROS (O₂^-· and H₂O₂) and malondialdehyde (MDA), coupled with the smallest increase in relative electrical conductivity.

These findings indicate that the superior canopy architecture of U-PT mitigates heat stress by creating a milder and more stable thermal environment, thereby avoiding the overwhelming impact of extreme heat on cellular defense systems that occurs in HT. This favorable microclimate allows the antioxidant system in U-PT to be effectively activated without being rapidly depleted, ensuring continuous scavenging of excess ROS and protecting photosynthetic apparatus (such as chloroplast integrity and the PSII complex) from oxidative damage, which ultimately contributes to the maintenance of higher photosynthetic efficiency in U-PT.

The results of this study are only applicable to ‘Shine Muscat’ grapevines under rain-shelter cultivation, and their applicability to other grape varieties, regional climates, or cultivation patterns requires further verification. Furthermore, the expression of chlorophyll content on a fresh weight basis is acknowledged as a limitation. It should be noted that the trellis systems represent a holistic treatment that concurrently affects vine vigor, water relations, and canopy microclimate; our study evaluated their integrated performance under heat stress. Future work should employ targeted experiments to dissect the contribution of these individual factors. Molecular biology techniques can be used to explore the molecular mechanisms underlying the enhanced heat tolerance of grapevines by U-PT. Long-term field experiments should be conducted to evaluate the effects of different trellis systems on grape yield and fruit quality, so as to clarify the economic feasibility of their application in production. Meanwhile, canopy structure manipulation experiments or microclimate simulation can be employed to verify the hypothesis of canopy heat accumulation in HT, thereby providing theoretical support for the optimization of trellis systems.