Introduction

Grapevine is among the most widespread and economically important crops suffering environmental constraints such as increased levels of thermo-radiation and incidences of drought and salt stress (Keller, 2010; Amato et al., 2020; Marín et al., 2021). Soil salinity is increasingly threatening agriculture globally and it is predicted to impact 50 % of cultivated land by 2050 (Hu and Schmidhalter, 2023). Hence, winegrowers will increasingly face environmental constraints to the extent that saline waters (e.g., Na₂SO₄ + MgSO₄, NaCl) would be used as alternative water resources for irrigation in Mediterranean grapevine (Martínez-Moreno et al., 2022). It is worth noting that some authors (Mirás-Avalos and Intrigliolo, 2017; Martínez-Moreno et al., 2023) report minimal impacts of salty water on vine performances in the short period. Moreover, the quality of groundwater is decreasing due to salt intrusion and pollutants from various sectors (Weerasekera, 2017). Following this, research activities improving the understanding of the salinity stress response can support identifying genetic resources and management solutions to challenging environments.

Under salinity stress, plant response generally follows a two-stage pattern. The initial stage is dominated by osmotic stress due to increased salt concentration in the soil (Isayenkov and Maathuis, 2019), while the concentration of critical ions (e.g., Na⁺) in the leaf remains unchanged. Thereafter, with persistent salinity stress, plants accumulate a toxic level of these ions in the older leaves and shoots, causing necrosis and premature canopy senescence (Munns and Tester, 2008).

The initial (leaf) ion-independent response to salt stress is increasingly attracting interest in research studies focussing on the physiological limitations induced by salt accumulation into the soil, even though leaves do not accumulate ions above the non-stressed level. For example, in grapevine, initial exposure to salinity triggered both stomatal and metabolic impairments of photosynthesis and, in turn, reduced shoot growth while ion concentrations in leaves remained comparable to that of control (Montanaro et al., 2022). The above-mentioned stomatal closure is a common response to initial salt stress and it conceivably determines the reduction of transpiring flux to reduce the rate of water uptake and thus of salts (Munns and Tester, 2008). In addition, improving understanding of the early response to salinity stress response would support the screening of plant populations for tolerant individuals. For example, Atieno et al. (2017) explored the genetic variation for salinity tolerance in chickpeas under ion-independent stress.

Nowadays, plant stress recognition is increasingly pursued through high-throughput plant phenotyping as an innovative, non-invasive, image-based technology. In line with this, the response of grapevine to environmental constraints (e.g., drought, pathogens) has recently been the subject of studies employing image-based datasets in which different parts of the plant (e.g., whole canopies or single leaves) are interrogated through different image-acquisition systems (e.g., from drones or robotised platforms) (Briglia et al., 2019 and 2020; Ouyang et al., 2021). Particularly, red, green and blue (RGB) colour bands and related colorimetric indexes (e.g., Dark Green) have been revealed as promising digital proxies of drought stress also because RGB cameras are easily accessible even through smartphones (Reynolds et al., 2018; Ninomiya, 2022). However, image-based phenotyping of grapevines subjected to salt stress has not received much attention. Although various management approaches have been developed to face soil salinity (e.g., surface flashing, drip irrigation, mechanical removal of salts surface crust, use of organic soil amendments, introduction of tolerant plants, priming) (see for review Panagea et al., 2016), digital tools for reliable and real-time monitoring of salt-stress need to be developed to determine plant adaptation capacity and support new breeding strategies (Costa et al., 2019). Moreover, a digital tool might support the scoring of symptoms of salt injury, which are often based on human visual assessment (e.g., Doulati Baneh et al., 2014). Hence, additional knowledge on smart plant phenotyping in grapevine under salinity stress would create the basis for an early assessment of that stress, identifying digital proxies of traits enabling the transition toward digital agriculture. In plant phenotyping, a method to build up a dose–response curve has been proposed accounting for traits related to leaf mass (or leaf area) (Poorter et al., 2010). However, to the best of our knowledge, till now, a dose–response curve has not been used to describe the response of an image-based colorimetric trait to soil salinity.

This study aimed to test the hypothesis that RGB-based imaging technology can parametrise the early response to salt stress in grapevine. To test this hypothesis, gas exchanges, fluorescence and stem water potential, leaf ions and pigments were analysed in grapevines (Aleatico/SO4) irrigated with saline water (100 mM NaCl) along with RGB images pictured at a robotises phenotyping platform.

Materials and methods

3. RGB imaging

A group of five vines per treatment was imaged around midday (from 12:00-13:00 h) on days 0 (just before salt treatment was initiated), 1, 2, 3, 6 and 9 using a LemnaTec 3D Scanalyzer phenotyping platform (LemnaTec GmbH, Aachen, Germany). The platform is located at the ALSIA-CRMA laboratories, part of the European Plant Phenotyping Network and infrastructures (https://emphasis.plant-phenotyping.eu/about). Vines were automatically conveyed into the imaging chamber for image acquisition. The chamber for the RGB image was equipped with two visible light cameras (2 Mpixel, Basler Scout scA1600-28gc, Colour, 1624 × 1234 pixels, ICX274 1/1.8" CCD, 28 fps, C Mount, Gigabit Ethernet). The lighting inside the chamber employed fluorescent tubes Osram T5FH 21 W 865 HE. Two images were acquired for each vine, one from each lateral orthogonal direction (0° and 90°). The image segmentation and analyses were carried out using the software LemnaGrid v5 following Arvidsson et al. (2011). For the image analysis, according to Briglia et al. (2020), the vine canopy was partitioned into the following regions of interest: bottom (nodes 1–5 from the base of the current season's growth), middle (nodes 6–11) and top (nodes > 11). For each region of interest, the LemnaGrid software v5 (LemnaTec GmbHc, Germany) operated a colour classification of the RGB images in Dark Green, Green, Yellow and Brown colour classes according to Acosta-Gamboa et al. (2017). Details on the RGB reference values (anchor points) per colour class are reported by Briglia et al. (2019). The mean pixel values of R, G and B were also determined. Additional RGB-based indexes were also determined according to Montanaro et al. (2023). A schematic workflow of the image segmentation and analysis is reported in Figure 1.

Figure 1. Schematic workflow of image segmentation and data analysis: after the picture acquisition (a) and segmentation (b), the three regions of interest (ROI) (top, middle, bottom) were identified. Thereafter, for each ROI, the mean pixel values of red (R), green (G) and blue (B) were determined. On each ROI, the fraction of the yellow (Y), G, dark green (DG) and brown (Br) were determined as the ratio between the number of pixels falling in the specific colour band and the total number of pixels.

Save View full size Expand inline Collapse inline

Results

This study reports physiological and phenotypic changes induced by a 10-day salt stress, causing an increase in soil EC up to about 8 dS m^-1 (Figure 2). At the end of the experiment, leaves from the bottom, middle and top regions of the main stem in salt-stressed vines showed concentrations of K⁺ and Na⁺ comparable to those in CTRL vines (Figure 3). In contrast to this, the mean concentration of the Chl-a was significantly reduced by the salt stress being approx. 0.68 ± 0.04 mg g^-1 DW, while it was approx. 1.06 ± 0.05 mg g^-1 DW in CTRL vines (Figure 4). Similarly, the concentration of carotenoids was significantly lower than that in CTRL vines, being about 0.48 ± 0.06 mg g^-1 DW; on the contrary, the Chl-b concentration did not show significant variation (Figure 4).

Figure 2. Variations in soil electrical conductivity in control (CTRL) salt-stressed (SALT) vines.

Save View full size Expand inline Collapse inline

Comparing treatments at the same time, * indicates statistically significant differences (p < 0.05). Values are mean ±SE (n = 14) and are visible only when larger than the symbol.

Figure 3. Concentrations of (A, C, E) potassium and (B, D, F) sodium detected in leaves of control (CTRL) and salt (100 mM NaCl) stressed (SALT) vines sampled on the last day (Day 9 after the salinity stress imposition) from the top, middle and bottom regions of the vine. Values are means ±SE (n = 3). NS = not-significant difference comparing regions (Top, Middle, Bottom) within the same mineral element and treatment (CTRL or SALT); ns = not-significant difference comparing treatments within the same mineral element and canopy region. Tests were carried out at α = 0.05 (Kruskal–Wallis’ test).

Save View full size Expand inline Collapse inline

Figure 4. Boxplot of the concentration of (left) chlorophyll-a, (middle) chlorophyll-b and (right) carotenoids measured at the end of the experiment in leaves of vines irrigated with (CTRL) fresh and (SALT) salty water. The grey dots represent the single analytical determinations. Comparing treatments within the same pigment, the different letter indicates significant differences (Kruskal–Wallis’ test, p < 0.05).

Save View full size Expand inline Collapse inline

1. Physiological traits

The response of leaf gas exchanges to salt stress is presented in Figure 5. Results show that g_s in CTRL vines was roughly stable during the experiment after an initial decline to 0.15–0.18 mol m^-2 s^–1 detected on Day 3, likely due to a declined VPD (Figure 6). Contrary to this, as early as the day after the salt application, stomatal conductance (g_s) declined to about 0.15 ± 0.01 mol m^-2 s^–1 while it was at about 0.25 ± 0.01 mol m^-2 s^–1 in CTRL vines (Figure 5A). Thereafter, g_s continued to decline, reaching 0.013 ± 8.1 × 10^-4 mol m^-2 s^–1 on Day 9 after the start of salt application (Figure 5A).

Stomatal limitations of gas exchanges induced by salt stress significantly influenced transpiration and photosynthetic rates. Starting from Day 1 after salt imposition, transpiration rate (E) in salt-stressed vines progressively declined, reaching about 1.7 ± 0.18 mmol H₂O m^-2 s^–1 on Day 4 (Figure 5B). Thereafter, E further declined, reaching an average value of 0.51 ± 0.03 mmol H₂O m^-2 s^–1 at the end of the experiment. Similarly, the net assimilation rate (A) in stressed vines steeply declined from the mean value of ~ 12.6 ± 1.03 µmol CO₂ m^-2 s^–1 (Day 0) to around 4.78 ± 0.14 µmol CO₂ m^-2 s^–1 on Day-3 (Figure 5C). During the following days, the mean value of A continued to progressively decline towards the minimum of 0.83 ± 0.06 µmol CO₂ m^-2 s^–1 at the end of the experiment (Figure 5C).

Figure 5. Variations in (A) stomatal conductance, (B) transpiration and (C) photosynthetic rate measured in control (CTRL, black line) and salt-stressed (SALT, red line) vines. Comparing treatments at the same time, * indicates statistically significant differences (p < 0.05). Values are mean (n = 18) ± SE and are visible only when larger than the symbol.

Save View full size Expand inline Collapse inline

Figure 6. Daily course of vapour pressure deficit (VPD), air temperature (°C) and photosynthetically active radiation (PAR) measured in the greenhouse during the experiment.

Save View full size Expand inline Collapse inline

The physiological parameters examined (A, E and g_s) declined smoothly with the increasing EC and were (negatively) significantly correlated with EC, showing the Spearman rank test (rho, ρ) ranging from –0.92 to – 0.99 (Figure 7A–C). Similarly, the stem water potential (SWP) declined with increasing soil EC (Figure 7D). However, after an initial prompt decline recorded at soil EC values of about 2 dS m^-1, the SWP remained stable at approx. –0.55 MPa despite the increasing soil EC up to 6 dS m^-1. Values of EC beyond ~6 dS m^-1 further decreased the SWP (Figure 7D).

Figure 7. Relationship between soil electrical conductivity (EC, dS m-1) and (A) stomatal conductance, (B) transpiration, (C) net photosynthetic rate and (D) stem water potential measured in control vines (CTRL, black dots) and salt-stressed vines (SALT, red dots). Valued are mean (n = 18) ± SE and are visible when larger than the symbol. All data were pooled in each panel before determining the Spearman rank correlation test (ρ) and exponential decay fitting.

Save View full size Expand inline Collapse inline

The maximum potential photosynthetic efficiency (F_v/F_m) in CTRL vines was, on average, ~0.76 throughout the experiment, whereas in leaves under salt stress, it significantly declined to ~0.63 (Day 9) if a transient increase recorded on Day 6 is excepted (Figure 8A).

During the salt stress period, the soil EC (Figure 2) progressively increased from about 0.80 dS m^-1 (starting point) to about 8.0 dS m^-1 by the end of the trial, whereas it remained almost stable at around 0.75 dS m^-1 in CTRL vines. Similarly to gas exchanges, the F_v/F_m showed a significant negative correlation (ρ = -0.86) with the increasing soil EC (Figure 8B).

Figure 8. Variation in chlorophyll-a fluorescence (Fv/Fm) measured in vines irrigated with tap water (CTRL) and under salt stress (SALT). Comparing treatments at the same time, * indicates statistically significant differences (p < 0.05). Values are mean (n = 6) ±SE. (B) Correlation between soil electrical conductivity (EC, dS m-1) and the Fv/Fm measured during the experiment; data were pooled before determining the Spearman rank correlation test (ρ) and linear fitting.

Save View full size Expand inline Collapse inline

2. Image-based salt-stress phenotyping

In our study, RGB images have been employed for phenotyping the effects of salt stress through analysis of variations in mean pixel values of R, G and B and in other RGB-based colorimetric indexes. At the end of salt stress, the bottom leaves of vines had visible colour changes, while the middle and top ones remained visually comparable to those in CTRL (Figure 9).

The map of the colour band fractions determined across the salt stress period is reported in Figure 9. The Dark Green was the most prominent colour band, allowing easy recognition of its variation. From Day 3, after salt stress imposition, the bottom region of salt-stressed vines showed a significant reduction of the Dark Green fraction compared to CTRL ones, reaching a significant ~43 % lower value on Day 9 (Figure 9). The fraction of the Brown colour band (bottom region) in CTRL vines was stably about 5 % throughout the experiment, while it significantly increased in salt-stressed ones, reaching 14 % and 38 % on Day 6 and Day 9, respectively (Figure 9).

Figure 9. Left: map of the fraction of the colour bands observed during the experiment in control (CTRL), left column) and salt-stressed (SALT, right column). Data refer to the Top, Middle and Bottom regions of the vines, comparing treatments within the same day and colour band fraction, * indicates significant differences (p < 0.05). Right: phenotypic variation of grapevine irrigated with fresh water (CTRL) and exposed to salt stress (SALT); RGB image of each vine has been pictured separately from each other, then they were composed for illustrative purposes. Salt stress caused leaf discolouration (mainly at the bottom region of the canopy) compared to control vines; salt also stunted growth. Due to the size of the RGB chamber, the newly growing apical shoots in the CTRL vine were not secured to the woody supporting pole, preventing them from being cut out from the frame.

Save View full size Expand inline Collapse inline

The response of Dark Green was correlated to increasing soil EC, showing a high coefficient of determination (R²) equal to 0.71 and 0.73 in the middle and bottom regions, respectively (Figure 10). In addition, the EC value of 4.2 dS m^-1 was determined as a threshold value corresponding to the change of 1 % of the maximum value of the fitting model of the Dark Green in the bottom region.

Figure 10. Correlation between soil electrical conductivity (EC) and Dark Green fraction measured in vines under increasing salt stress conditions. The non-linear fitting curve employed the function Y=Ym/[1+(EC/EC50)p], where Y is the Dark Green fraction calculated at EC = x, Ym is the maximum asymptote of the Dark Green fraction when soil EC=0,p is the slope at the steepest part of the curve, EC50 is the value of the EC when Dark Green fraction equals 50 % of Ym. Fitting lines were drawn when p < 0.05. The arrow indicates the threshold value of EC (ECt) corresponding to a ~1 % decline of Dark Green (Ym) measured at the lower part of the vine.

Save View full size Expand inline Collapse inline

Increasing soil EC had an erratic covariation with the mean pixel values of the R, G, B and RGB-based indexes to the extent that the Spearman’s rank test (rho, ρ) ranged from –0.74 to 0.75 (Figure 11A). For those colorimetric variables having a high ρ, we had a roughly good linear correlation with EC values detected at EC > 4.2 dS m^-1 (Figure 11B). For example, in the case of R and CI indexes, the values of R² were considerably >= 0.8 (Figure 11B).

Figure 11. (A) Heat-map of the Spearman's rank test employed to measure the correlation between soil electrical conductivity (EC) and R, G, B and the various RGB-based colorimetric indexes measured in the bottom part of the canopy in salt-stressed vines. Note that the variables have been sorted according to ascending ρ values from the minimum –0.74 (NB, Normalised Blue) to the maximum 0.75 (CI, Colouration Index) and that the legend scale has been drawn within the same range. (B) Scatterplot between soil EC and R, G, B and some RGB-based colorimetric indexes measured in salt-stressed vines. Note that solid symbols refer to paired values measured at EC > 4.2 dS m-1 (see Figure 10 for details on the EC threshold). The values of the coefficient of determination (R2) refer to a linear model Y ~ a*x +b, where the x was the EC data belonging to solid points, Y was the colorimetric variable and a and b were the slope and the Y-axis intercept, respectively. For full details on the colorimetric index, see Montanaro et al., 2023.

Save View full size Expand inline Collapse inline

Discussion

The present study reports on image-based proximal sensing of salt stress in grapevine employing RGB images, focussing on the early stage of the stress, which is independent of the toxic accumulation of ions (Munns and Tester, 2008). Accordingly, the critical ion (i.e., Na⁺) did not accumulate above the control level in the salt-stressed leaves; additionally, it was approx. 15–30 % of those reported in a 10-day (Mohammadkhani and Abbaspour, 2017) and 2-week (Fozouni et al., 2012) salt exposure. Similarly, the concentration of K⁺ was not influenced by salt stress and was consistent with that reported in Amorim et al. (2023) for the same SO4 rootstock after two weeks of salt stress. The unincreased leaf concentration of K⁺ and Na⁺ in stressed vines supports this in the early (osmotic) stage of salt stress (Munns and Tester, 2008). Although the significance of treatments (CTRL, SALT) and vine region (Top, Middle, Bottom) was appropriately evaluated using a non-parametric test (Kruskall–Wallis), the number of chemical analyses for K⁺ and Na⁺ should be increased to improve the robustness of the comparison.

In this study, we did not check for Cl^- accumulation as Na⁺ is the most prominent factor causing damage in grapevines under salt stress (Dag et al., 2015). Moreover, the role of Cl^- in causing salt-related impairments (e.g., photosynthesis decline) remains controversial (Baggett et al., 2021). Despite correlative information on paired accumulation of Na⁺ and Cl^- are available also showing similar increases (Walker et al., 1997; Meggio et al., 2014), measuring the concentration of Cl^- in response to the osmotic stage of salt stress would provide a more complete picture.

Although we found no significant change in leaf Na⁺ and K⁺ concentrations, salt stress caused several physiological responses. That is, soon after salty water application values in maximum potential quantum efficiency of Photosystem II (F_v/F_m) as well as in the other examined leaf functional traits (A, E and g_s) showed a prompt deviation of about 9 % (F_v/F_m), 40 % (A, g_s) and 15 % (E) from those recorded in CTRL vines.

Clearly, stomatal and metabolic impairments of photosynthesis occurred within a few days after salt stress application, confirming the sensitivity of Vitis spp. to salt exposure (Maas and Hoffman, 1977; Prior et al., 1992; Cramer et al., 2006; Meggio et al., 2014) and that they were uncoupled from colour-based changes in phenotype. Following this, it could be noted that approx. 50 % reduction of the pre-stress level of gas exchange parameters and F_v/F_m occurred when soil EC approached about 4 dS m^-1 (Figure 7,8B), whilst colour-based traits were still comparable to those in CTRL (Figure 9). Similarly, Upadhyay et al. (2018) reported a time lag between the impairment of key physiological parameters, including gas exchanges and the onset of visual symptoms. Hence, image-based phenotyping protocols should combine proxies based on physiological parameters (e.g., F_v/F_m) which quickly respond to salt stress. The impact of early salt stress on growing parameters (e.g., plant size, internode elongation) has been recognised in various species, including chickpea and grapevine (Atieno et al., 2017; Upadhyay et al., 2018; Montanaro et al., 2022).

The present study was conducted on potted vines embedding some potential divergences from field experiments. However, the response of EC to salty water application is comparable to that recorded in a field experiment in a commercial vineyard (Ben-Asher et al., 2006), including the following effect on SWP (Yu and Kurtural, 2020).

In our study, the image-based phenotyping of the salt-stressed vines employed two sets of measurements: (i) the pixel fraction of specific colour bands (Yellow, Green, Brown and Dark Green) and (ii) the mean pixel value of R, G and B and other RGB-based colorimetric indexes. The choice of colour bands employed to feed the LemnaGrid clustering algorithm to differentiate and calculate the relative colour classification was based on own previous research on drought stress (Briglia et al., 2019), considering its similarity with the initial stage of salt stress. In addition, leaf yellowish and browning were also associated with salt stress in a salt responsive transcriptome analysis of grapevine (Das and Majumder, 2018).

Among the specific colour bands examined, results show that the pixel fractions of the Dark Green at the lower and middle regions of the vines were the most prominent and informative on the progress of the salt stress. The present study did not test whether the Dark Green fraction (or other colorimetric indices) is specific to (early) salt stress or common to all osmotic stresses. However, it was found that the variation in the Dark Green fraction within a dose-response context occurred as early as three days after salt imposition, suggesting its suitability for the detection of that stress. In Briglia et al. (2019), Dark Green was the best predictor of SWP in a drought stress experiment, which is in favour of its potential suitability as a predictor in both early salt and drought stress, but this remains to be specifically tested.

The upper region of the vine shoot (the youngest region) was less informative, likely because of the dichromatism of young (copper-coloured) leaves in the cultivar we tested, as it was pointed out in a drought-related study (Briglia et al., 2020). Moreover, to prevent a misinterpretation of the effect of the stress caused by that dichromatism, top-view images were not considered, and only two side-view images (at 90° from each other) were employed to capture colour-based phenotypic traits according to phenotyping studies in grapevine (Briglia et al., 2019) and other crops (Klukas et al., 2014; Hairmansis et al., 2014; Petrozza et al., 2014). Hence, considering that the colour of young leaves is a cultivar-dependent trait (see, for example, code 051 in OIV, 2009), it is advisable not to capture young leaves within the image frame, thus avoiding misinterpretation.

Following this, the response of the Dark Green pixel fraction to increasing exposure to salt was evaluated separately in each plant region, showing a close correlation with soil EC (Figure 10). Data on the Dark Green fraction were then fitted through a dose-response curve belonging to Hill functions, which are useful functions for retrieving the threshold value at which the response variable is halfway between the minimum and maximum responses (Khan et al., 2003; Gadagkar and Call, 2015). The sigmoidal dose-response type equations have been employed in various contexts (Gadagkar and Call, 2015; Xue et al., 2017), including the responses of seedling growth to saline conditions in barley and maise (Khan et al., 2003; Xue et al., 2017).

The threshold value of EC (ECt) corresponding to the initial decline (~ 1 % of maximum) of the response variable has been considered a reliable parameter for comparing various levels of salt tolerance (Martínez-Cob et al., 1987). In line with this idea, it should be noted that the relative 50-60 % adaptive reduction in some gas exchange parameters in response to the salt stress occurred when soil EC approached the ECt threshold (Figure 10). However, values of ECt, as well as the relationships between colorimetric indicators and salt stress intensity (Figure 11), need to be parametrised in relation to factors affecting vine sensitivity to early (osmotic) salt stress, such as genetic diversity. For example, an in vitro study tested up to 40 grapevine rootstock accessions for osmotic-stress tolerance, confirming that genetic diversity induces a different morphometric response (Peiró et al., 2020). Hence, the influence of the genetic diversity of rootstock (and cultivar) on RGB-based colorimetric indexes remains to be examined.

Considering that similar leaf concentrations of K⁺ and Na⁺ were present among the treatments (Figure 3), variations in the pixel fraction of Dark Green are plausibly attributable to decreases in Chl-a and carotenoids concentrations, which is a common response to salinity (e.g., Carter and Knapp, 2001; Fozouni et al., 2012a; Fozouni et al., 2012b). The Chl-b concentration tended to decrease, but it was not significantly reduced compared to CTRL, likely due to the duration of stress and/or the limited number of observations. The decrease in carotenoids might be explained by considering that the main hormone involved in salt stress regulation is abscisic acid (Ryu and Cho, 2015), whose biosynthesis involves carotenoids as precursors (Seo and Koshiba, 2002). By contrast, leaf carotenoids would be stable in case of chlorosis symptoms triggered, for example, by N deficiency (Rustioni et al., 2018). In addition, considering that reflectance spectroscopy supports the identification of chlorophylls and carotenoids (and their relative change) in grapevine leaves (Rustioni et al., 2018), image-based salt stress phenotyping might be integrated with that signal. Following this, it can reasonably be said that image-based phenotyping (namely the Dark Green colour band) is a promising phenotyping tool for tracking the early stage of salt stress. This would also be supportive for proximal and remote vineyard phenomics in a precision agriculture domain (Meggio et al., 2010; Taskos et al., 2015).

Nowadays, image-based and affordable phenotyping is extensively employing RGB cameras (Reynolds et al., 2018; Ninomiya, 2022). Accordingly, the use of R, G and B and related indexes would be of interest in phenotyping salt stress as most of these indexes have been associated with changes in leaf pigment concentrations (see Montanaro et al., 2023 and the references therein), which are sensitive to salt stress as reported by Panda et al. (2013) in Oryza sativa and Sohrabi et al. (2017) in grapevines. In line with this, correlative information collected in our study supports that idea, at least for some colorimetric indexes (Figure 11A). In addition, salt-induced changes of the RGB colorimetric indexes were, in most cases, linearly coupled to EC values beyond 4 dS m^-1, which fits with the soil ECt threshold discussed above. Particularly, it should be noted that the mean pixel value of the CI colorimetric index and R channel were able to explain a large part of EC variance when greater than ECt (R² > 0.80) (Figure 11B). Considering that the employment of RGB-based colorimetric indexes as predictors of plant/fruit traits are increasingly used as input variables of artificial neural networks (ANN), achieving a considerable improvement in their predicting capability (Picón et al., 2022; Wei et al., 2022), the combination of RGB-based indexes and ANN in a salt stress phenotyping context deserves future research. Moreover, as the osmotic stress promptly reduces stomatal conductance and, in turn, changes the leaf temperature (Jones et al., 2002), it is advisable to combine RGB-based imaging with thermal ones also in view of its affordable low cost (Reynolds et al., 2019; Ninomiya, 2022).

The present study documented that RGB-based imaging (namely the fraction of the Dark Green colour band) is a promising phenotyping tool for tracking the early stages of salt stress, easily supporting the identification of critical thresholds of soil EC. This study also pointed out the robustness of some RGB-based colorimetric indexes to serve as a proxy within affordable phenotyping of salt stress. In view of the increasing need for alternative water resources, such as saline waters for irrigation in grapevine (Martínez-Moreno et al., 2022), this study contributes to identifying smart proxies potentially enabling the management of vineyard under salt stress in a digital viticulture domain.

Acknowledgements

This study was carried out within the Agritech National Research Center and received funding from the European Union Next-GenerationEU (PIANO NAZIONALE DI RIPRESA E RESILIENZA (PNRR) – MISSIONE 4 COMPONENTE 2, INVESTIMENTO 1.4 – D.D. 1032 17/06/2022, CN00000022) and by Programma Operativo Nazionale «Imprese e Competitività» 2014-2020 FESR of Ministero Sviluppo Economico (Project n. F/200058/01-02/X45, Decree n. 0110536 dated 15.04.2020), Project E.A.Sy. (Ecological sustainability in Agriculture Systems). This manuscript reflects only the authors’ views and opinions; neither the European Union nor the European Commission can be considered responsible for them. The authors thank Mr. A. Mossuto (Natura Informatica) and Dr. D. Amato for technical and lab assistance. This study was part of the initiatives of Phen-Italy, the Italian node of the European Infrastructure for multi-scale Plant Phenomics and Simulation for Food Security in a Changing Climate EMPHASIS.

References

• Acosta‐Gamboa, L., Liu, S., Langley, E., Campbell, Z. C., Castro-Guerrero, N., Mendoza‐Cózatl, D. G., & Lorence, A. (2017). Moderate to severe water limitation differentially affects the phenome and ionome of Arabidopsis. Functional Plant Biology, 44(1), 94. https://doi.org/10.1071/fp16172
• Amato, D., Montanaro, G., Summerer, S., Briglia, N., Attia, F., Challet, E., & Nuzzo, V. (2020). The effects of calcite silicon-mediated particle film application on leaf temperature and grape composition of Merlot (Vitis vinifera L.) vines under different irrigation conditions. OENO One, 54(4), 1007–1020. https://doi.org/10.20870/oeno-one.2020.54.4.4020
• Amorim, T. L., Santos, H. R. B., Neto, J. B., Hermínio, P. J., Silva, J. R. I., Silva, M. M. A., Simões, A. N., De Souza, E. S., & Ferreira‐Silva, S. L. (2023). Resistant rootstocks mitigate ionic toxicity with beneficial effects for growth and photosynthesis in grapevine grafted plants under salinity. Scientia Horticulturae, 317, 112053. https://doi.org/10.1016/j.scienta.2023.112053
• Arvidsson, S., Pérez‐Rodríguez, P., & Mueller‐Roeber, B. (2011). A growth phenotyping pipeline for Arabidopsis thaliana integrating image analysis and rosette area modeling for robust quantification of genotype effects. New Phytologist, 191(3), 895–907. https://doi.org/10.1111/j.1469-8137.2011.03756.x
• Atieno, J., Li, Y., Langridge, P., Dowling, K., Brien, C., Berger, B., Varshney, R. K., & Sutton, T. (2017). Exploring genetic variation for salinity tolerance in chickpea using image-based phenotyping. Scientific Reports, 7(1). https://doi.org/10.1038/s41598-017-01211-7
• Baggett, J. P., Habibsadeh, S., Toups, H. S., Cochetel, N., Ghan, R., Robinson, M. L., Barrios-Masias, F. H., & Cramer, G. R. (2021). Is foliar Cl- concentration the cause of photosynthetic decline in grapevine during mild salinity? OENO One, 55(4), 33–48. https://doi.org/10.20870/oeno-one.2021.55.4.4795
• Baneh, H. D., Hassani, A., & Shaieste, F. G. (2014). Effects of salinity on leaf mineral composition and salt injury symptoms of some Iranian wild grapevine (Vitis vinifera L. ssp. sylvestris) genotypes. OENO One, 48(4), 231. https://doi.org/10.20870/oeno-one.2014.48.4.1692
• Ben-Asher, J., Tsuyuki, I., Bravdo, B., & Sagih, M. (2006). Irrigation of grapevines with saline water: I. Leaf area index, stomatal conductance, transpiration and photosynthesis. Agricultural Water Management 83, 13–21 https://doi.org/10.1016/j.agwat.2006.01.002
• Briglia, N., Montanaro, G., Petrozza, A., Summerer, S., Cellini, F., & Nuzzo, V. (2019). Drought phenotyping in Vitis vinifera using RGB and NIR imaging. Scientia Horticulturae, 256, 108555. https://doi.org/10.1016/j.scienta.2019.108555
• Briglia, N., Williams, K., Wu, D., Li, Y., Tao, S., Corke, F., Montanaro, G., Petrozza, A., Amato, D., Cellini, F., Doonan, J., Yang, W., & Nuzzo, V. (2020). Image-Based assessment of drought response in grapevines. Frontiers in Plant Science, 11. https://doi.org/10.3389/fpls.2020.00595
• Carter, G. A., & Knapp, A. K. (2001). Leaf optical properties in higher plants: linking spectral characteristics to stress and chlorophyll concentration. American Journal of Botany, 88(4), 677–684. https://doi.org/10.2307/2657068
• Choné, X., Van Leeuwen, C., Dubourdieu, D., & Gaudillère, J. P. (2001). Stem Water Potential is a Sensitive Indicator of Grapevine Water Status. Annals of Botany, 87(4), 477–483. https://doi.org/10.1006/anbo.2000.1361
• Costa, C., Schurr, U., Loreto, F., Menesatti, P., & Carpentier, S. (2019). Plant Phenotyping Research Trends, a Science Mapping approach. Frontiers in Plant Science, 9. https://doi.org/10.3389/fpls.2018.01933
• Cramer, G. R., Ergül, A., Grimplet, J., Tillett, R. L., Tattersall, E. a. R., Bohlman, M. C., Vincent, D., Sonderegger, J., Evans, J. R., Osborne, C., Quilici, D. R., Schlauch, K., Schooley, D. A., & Cushman, J. C. (2006). Water and salinity stress in grapevines: early and late changes in transcript and metabolite profiles. Functional & Integrative Genomics, 7(2), 111–134. https://doi.org/10.1007/s10142-006-0039-y
• Dag, A., Ben‐Gal, A., Goldberger, S., Yermiyahu, U., Zipori, I., Or, E., David, I., Netzer, Y., & Kerem, Z. (2015). Sodium and chloride distribution in grapevines as a function of rootstock and irrigation water salinity. American Journal of Enology and Viticulture, 66(1), 80–84. https://doi.org/10.5344/ajev.2014.14019
• Das, P., & Majumder, A. L. (2018). Transcriptome analysis of grapevine under salinity and identification of key genes responsible for salt tolerance. Functional & Integrative Genomics, 19(1), 61–73. https://doi.org/10.1007/s10142-018-0628-6
• Fozouni, M., Abbaspour, N., & Baneh, H. D. (2012a). Short term response of grapevine grown hydroponically to salinity: Mineral composition and growth parameters. Vitis: Journal of Grapevine Research, 51(3), 95–101. https://doi.org/10.5073/vitis.2012.51.95-101
• Fozouni, M., Abbaspour, N., & Baneh, H. D. (2012b). Leaf water potential, photosynthetic pigments and compatible solutes alterations in four grape cultivars under salinity. Vitis: Journal of Grapevine Research, 51(4), 147–152. https://doi.org/10.5073/vitis.2012.51.147-152
• Gadagkar, S. R., & Call, G. B. (2015). Computational tools for fitting the Hill equation to dose–response curves. Journal of Pharmacological and Toxicological Methods, 71, 68–76. https://doi.org/10.1016/j.vascn.2014.08.006
• Goudriaan, J., & van Laar, H. (1994). Modelling potential crop growth processes. In Current issues in production ecology. https://doi.org/10.1007/978-94-011-0750-1
• Hairmansis, A., Berger, B., Tester, M., & Roy, S. J. (2014). Image-based phenotyping for non-destructive screening of different salinity tolerance traits in rice. Rice, 7(1). https://doi.org/10.1186/s12284-014-0016-3
• Hu, Y., & Schmidhalter, U. (2023). Opportunity and challenges of phenotyping plant salt tolerance. Trends in Plant Science, 28(5), 552–566. https://doi.org/10.1016/j.tplants.2022.12.010
• Isayenkov, S. V., & Maathuis, F. J. M. (2019). Plant salinity Stress: Many unanswered questions remain. Frontiers in Plant Science, 10. https://doi.org/10.3389/fpls.2019.00080
• Jones, H. G., Stoll, M., Santos, T., Sousa, C. de, Chaves, M. M., & Grant, O. M. (2002). Use of infrared thermography for monitoring stomatal closure in the field: Application to grapevine. Journal of Experimental Botany, 53(378), 2249–2260. https://doi.org/10.1093/jxb/erf083
• Keller, M. (2010). Managing grapevines to optimise fruit development in a challenging environment: a climate change primer for viticulturists. Australian Journal of Grape and Wine Research, 16, 56–69. https://doi.org/10.1111/j.1755-0238.2009.00077.x
• Khan, A. A., Rao, S. A., & McNeilly, T. (2003). Assessment of salinity tolerance based upon seedling root growth response functions in maize (Zea mays L.). Euphytica, 131(1), 81–89. https://doi.org/10.1023/a:1023054706489
• Klukas, C., Chen, D., & Pape, J. (2014). Integrated Analysis Platform: An Open-Source information System for High-Throughput plant phenotyping. Plant Physiology, 165(2), 506–518. https://doi.org/10.1104/pp.113.233932
• Lichtenthaler, H. K., & Wellburn, A. R. (1983). Determinations of total carotenoids and chlorophylls a and b of leaf extracts in different solvents. Biochemical Society Transactions, 11(5), 591–592. https://doi.org/10.1042/bst0110591
• Maas, E. V., & Hoffman, G. J. (1977). Crop Salt Tolerance—Current Assessment. Journal of the Irrigation and Drainage Division, 103(2), 115–134. https://doi.org/10.1061/jrcea4.0001137
• Marín, D., Armengol, J., Carbonell‐Bejerano, P., Escalona, J. M., Gramaje, D., Hernández‐Montes, E., Intrigliolo, D. S., Martı́nez-Zapater, J. M., Medrano, H., Mirás-Avalos, J. M., Palomares‐Rius, J. E., Romero-Azorín, P., Savé, R., Santesteban, L., & De Herralde, F. (2021). Challenges of viticulture adaptation to global change: tackling the issue from the roots. Australian Journal of Grape and Wine Research, 27(1), 8–25. https://doi.org/10.1111/ajgw.12463
• Martínez‐Cob, A., Aragüés, R., & Royo, A. (1987). Salt tolerance of barley (Hordeum vulgare L.) cultivars at the germination stage: Analysis of the response functions. Plant and Soil, 104(1), 53–56. https://doi.org/10.1007/bf02370624
• Martínez‐Moreno, A., Pérez‐Álvarez, E. P., López-Urrea, R., Intrigliolo, D. S., González-Centeno, M. R., Teissèdre, P., & Gil‐Muñoz, R. (2022). Is deficit irrigation with saline waters a viable alternative for winegrowers in semiarid areas? OENO One, 56(1), 101–116. https://doi.org/10.20870/oeno-one.2022.56.1.4910
• Martínez-Moreno, A., Pérez-Álvarez, E. P., Intrigliolo, D. S., Mirás-Avalos, J. M., López-Urrea, R., Gil-Muñoz, R., Lizama, V., García-Esparza, M. J., Álvarez, M. I., & Buesa, I. (2023). Effects of deficit irrigation with saline water on yield and grape composition of Vitis vinifera L. cv. Monastrell. Irrigation Science, 41(4), 469–485. https://doi.org/10.1007/s00271-022-00795-x
• Meggio, F., Prinsi, B., Negri, A., Di Lorenzo, G. S., Lucchini, G., Pitacco, A., Failla, O., Scienza, A., Cocucci, M., & Espen, L. (2014). Biochemical and physiological responses of two grapevine rootstock genotypes to drought and salt treatments. Australian Journal of Grape and Wine Research, 20(2), 310–323. https://doi.org/10.1111/ajgw.12071
• Meggio, F., Zarco‐Tejada, P. J., Núñez, L., Sepulcre-Cantó, G., Rebollo, M. Á. G., & Martín, P. (2010). Grape quality assessment in vineyards affected by iron deficiency chlorosis using narrow-band physiological remote sensing indices. Remote Sensing of Environment, 114(9), 1968–1986. https://doi.org/10.1016/j.rse.2010.04.004
• Mirás-Avalos, J. M., & Intrigliolo, D. S. (2017). Grape Composition under Abiotic Constrains: Water Stress and Salinity. Frontiers in Plant Science, 8. https://doi.org/10.3389/fpls.2017.00851
• Mohammadkhani, N., & Abbaspour, N. (2017). Absorption kinetics and efflux of chloride and sodium in the roots of four grape genotypes (Vitis L.) differing in salt tolerance. Iranian Journal of Science and Technology Transaction A-science, 42(4), 1779–1793. https://doi.org/10.1007/s40995-017-0445-0
• Montanaro, G., Briglia, N., Lopez, L., Amato, D., Panara, F., Petrozza, A., Cellini, F., & Nuzzo, V. (2022). A synthetic cytokinin primes photosynthetic and growth response in grapevine under ion-independent salinity stress. Journal of Plant Interactions, 17(1), 789–800. https://doi.org/10.1080/17429145.2022.2102259
• Montanaro, G., Petrozza, A., Rustioni, L., Cellini, F., & Nuzzo, V. (2023). Phenotyping key fruit quality traits in olive using RGB images and back propagation neural networks. Plant Phenomics, 5. https://doi.org/10.34133/plantphenomics.0061
• Munns, R., & Tester, M. (2008). Mechanisms of salinity tolerance. Annual Review of Plant Biology, 59(1), 651–681. https://doi.org/10.1146/annurev.arplant.59.032607.092911
• Ninomiya, S. (2022). High-throughput field crop phenotyping: current status and challenges. Breeding Science, 72(1), 3–18. https://doi.org/10.1270/jsbbs.21069
• OIV (2009). OIV Descriptor List for Grape Varieties and Vitis Species, 2nd ed.; OIV: Paris, France. Available on-line at https://www.oiv.int/public/medias/2274/code-2e-edition-finale.pdf
• Ouyang, J., De Bei, R., & Collins, C. (2021). Assessment of canopy size using UAV-based point cloud analysis to detect the severity and spatial distribution of canopy decline. OENO One, 55(1). https://doi.org/10.20870/oeno-one.2021.55.1.3078
• Panagea, I., Daliakopoulos, I. Ν., Tsanis, I. K., & Schwilch, G. (2016). Evaluation of promising technologies for soil salinity amelioration in Timpaki (Crete): a participatory approach. Solid Earth, 7(1), 177–190. https://doi.org/10.5194/se-7-177-2016
• Panda, D., Ghosh, D.C., & Kar, M. (2013). Effect of Salt Stress on the Pigment Content and Yield of Different Rice (Oryza sativa L.) Genotypes. International Journal of Bio-resource and Stress Management.4(3), 431-434. https://ojs.pphouse.org/index.php/IJBSM/article/view/410
• Peiró, R., Jiménez, C., Perpiñà, G., Soler, J. X., & Gisbert, C. (2020). Evaluation of the genetic diversity and root architecture under osmotic stress of common grapevine rootstocks and clones. Scientia Horticulturae, 266, 109283. https://doi.org/10.1016/j.scienta.2020.109283
• Petrozza, A., Santaniello, A., Summerer, S., Di Tommaso, G., Di Tommaso, D., Paparelli, E., Piaggesi, A., Perata, P., & Cellini, F. (2014). Physiological responses to Megafol® treatments in tomato plants under drought stress: A phenomic and molecular approach. Scientia Horticulturae, 174, 185–192. https://doi.org/10.1016/j.scienta.2014.05.023
• Picón, A., Bereciartua-Pérez, A., Eguskiza, I., Romero-Rodriguez, J., Jimenez-Ruiz, C. J., Eggers, T., Klukas, C., & Navarra-Mestre, R. (2022). Deep convolutional neural network for damaged vegetation segmentation from RGB images based on virtual NIR-channel estimation. Artificial Intelligence in Agriculture, 6, 199–210. https://doi.org/10.1016/j.aiia.2022.09.004
• Poorter, H., Niinemets, Ü., Walter, A., Fiorani, F., & Schurr, U. (2010). A method to construct dose–response curves for a wide range of environmental factors and plant traits by means of a meta-analysis of phenotypic data. Journal of Experimental Botany, 61(8), 2043–2055. https://doi.org/10.1093/jxb/erp358
• Prior, L. D., Grieve, A., & Cullis, B. R. (1992). Sodium chloride and soil texture interactions in irrigated field grown sultana grapevines. I. Yield and fruit quality. Australian Journal of Agricultural Research, 43(5), 1051. https://doi.org/10.1071/ar9921051
• Reynolds, D. B., Baret, F., Welcker, C., Bostrom, A., Ball, J., Cellini, F., Lorence, A., Chawade, A., Khafif, M., Noshita, K., Mueller-Linow, M., Zhou, J., & Tardieu, F. F. (2019). What is cost-efficient phenotyping? Optimizing costs for different scenarios. Plant Science, 282, 14–22. https://doi.org/10.1016/j.plantsci.2018.06.015
• Rustioni, L., Grossi, D., Brancadoro, L., & Failla, O. (2018). Iron, magnesium, nitrogen and potassium deficiency symptom discrimination by reflectance spectroscopy in grapevine leaves. Scientia Horticulturae, 241, 152–159. https://doi.org/10.1016/j.scienta.2018.06.097
• Ryu, H., & Cho, Y. (2015). Plant hormones in salt stress tolerance. Journal of Plant Biology, 58(3), 147–155. https://doi.org/10.1007/s12374-015-0103-z
• Seo, M., & Koshiba, T. (2002). Complex regulation of ABA biosynthesis in plants. Trends in Plant Science, 7(1), 41–48. https://doi.org/10.1016/s1360-1385(01)02187-2
• Sohrabi, S., Ebadi, A., Jalali, S., & Salami, S. A. (2017). Enhanced values of various physiological traits and VvNAC1 gene expression showing better salinity stress tolerance in some grapevine cultivars as well as rootstocks. Scientia Horticulturae, 225, 317–326. https://doi.org/10.1016/j.scienta.2017.06.025
• Taskos, D., Koundouras, S., Stamatiadis, S., Zioziou, E., Νικολάου, Ν., Karakioulakis, K., & Theodorou, N. (2015). Using active canopy sensors and chlorophyll meters to estimate grapevine nitrogen status and productivity. Precision Agriculture, 16(1), 77–98. https://doi.org/10.1007/s11119-014-9363-8
• Upadhyay, A., Gaonkar, T., Upadhyay, A. K., Satisha, J., Shinde, M., Kadoo, N. Y., & Gupta, V. S. (2018). Global transcriptome analysis of grapevine (Vitis vinifera L.) leaves under salt stress reveals differential response at early and late stages of stress in table grape cv. Thompson Seedless. Plant Physiology and Biochemistry, 129, 168–179. https://doi.org/10.1016/j.plaphy.2018.05.032
• Walker, R. R., Blackmore, D. H., Clingeleffer, P. R., & Iacono, F. (1997). Effect of salinity and Ramsey rootstock on ion concentrations and carbon dioxide assimilation in leaves of drip-irrigated, field-grown grapevines (Vitis vinifera L. cv. Sultana). Australian Journal of Grape and Wine Research, 3(2), 66–74. https://doi.org/10.1111/j.1755-0238.1997.tb00117.x
• Weerasekara, P. C. (2017). The United Nations World Water Development Report 2017 Wastewater: the Untapped Resource. Future of Food: Journal on Food, Agriculture and Society, 5(2), 80–81. http://www.futureoffoodjournal.org/index.php/journal/article/view/311
• Wei, X., Wu, L., Ge, D., Yao, M., & Bai, Y. (2022). Prediction of the maturity of greenhouse grapes based on imaging technology. Plant Phenomics, 2022. https://doi.org/10.34133/2022/9753427
• Xue, W., Yan, J., Zhao, G., Jiang, Y., Cheng, J., Cattivelli, L., & Tondelli, A. (2017). A major QTL on chromosome 7HS controls the response of barley seedling to salt stress in the Nure × Tremois population. BMC Genomic Data, 18(1). https://doi.org/10.1186/s12863-017-0545-z
• Yu, R. and Kurtural, S.K. (2020). Proximal sensing of soil electrical conductivity provides a link to soil-plant water relationships and supports the identification of plant water status zones in vineyards. Frontiers in Plant Science 11, doi: 10.3389/fpls.2020.00244