Abstract
Background and goals Cold hardiness in dormant grapevine buds depends on water dynamics during winter. Understanding winter bud responses and their effects on summer leaf physiology is critical for developing cultivars adapted to harsh winters, while maintaining productivity. Semi-automated differential thermal analysis (DTA) is widely used to determine cold hardiness. In this study, we evaluated DTA alongside high-throughput sensor technologies (including hyperspectral and multispectral sensing) to assess their potential for cold-hardiness evaluation. The objective was to explore visible and infrared hyperspectral sensing as proxy traits to assess bud and leaf water content in grapevine.
Methods and key findings Over three years, bud cold hardiness (measured for two years, 2023 and 2024) and summer physiology (data collected in 2024 and 2025) were evaluated across 86 F1 progenies from a cross between Vitis hybrid NY84.0101.03 and Vitis amurensis PI 588634. Bud absorbance analysis identified consistent predictive bands at 1874, 1875, and 1888 nm across varying winter severities. Summer leaf reflectance at 1868 and 1873 nm showed strong correlations with winter bud hardiness. Among physiological traits, all but preveraison leaf-to-air vapor pressure deficit and chlorophyll fluorescence at veraison correlated positively with cluster and berry size. Stomatal conductance emerged as a key trait, showing both positive and negative associations with DTA values, depending on winter severity in 2023 and 2024.
Conclusions and significance Bud absorbance, which is related to water content, can serve as a reliable proxy for bud cold hardiness, traditionally measured by DTA. In contrast, hyperspectral measurement of the anthocyanin reflectance index in leaf tissues may be used as a selection trait during summer. Twelve progenies were identified as stable performers, offering valuable material for pre-breeding efforts to combine cold tolerance with vine productivity.
- cold hardiness
- differential thermal analysis
- high throughput phenotyping
- morpho-physiology
- spectral indices
- water content
Introduction
Grapevine cultivation is vulnerable to cold injury and yield loss due to sporadic frost and low-temperature events (De Rosa et al. 2021, Sharma et al. 2024). In recent years, wild species have been increasingly used in the development of hybrid cultivars tolerant to various abiotic stressors, including cold and freezing temperatures (Atak 2024). Cold-tolerance alleles from wild germplasm have been introgressed successfully into recently developed cold-hardy grape hybrids (Renzi et al. 2022). Among the wild Vitis species, Vitis amurensis accessions have been a primary focus for cold-stress response studies (Xin et al. 2013, Xu et al. 2014) and are considered promising breeding materials for developing cold-hardy cultivars (Ren et al. 2023, Atak 2024). Among the 70+ wild grape species, V. amurensis is notably resilient, capable of surviving extremely low temperatures, down to −40°C (Fennell 2004, Zhao et al. 2020).
Grapevine cold hardiness refers to the crop’s ability to withstand critical, lethal low temperatures during winter (Ferguson et al. 2011). In grapevines, damage caused by low temperatures, particularly to primary bud tissues, is highly detrimental to berry cluster production (Gao et al. 2014, Londo and Kovaleski 2017). Cold-induced mortality of primary buds can result in severe reductions of berry yield, with losses ranging from 27 to 100% in red wine grape varieties (Wolf and Miller 2001). Additionally, source-sink adjustments occurring during winter, and their carryover effects on summer physiological performance and yield-related traits, are well documented (Martinez-Lüscher and Kurtural 2021). However, no studies to date have been explicitly designed to investigate the relationship between winter cold-tolerance traits and subsequent summer physiological performance in the context of overall vine productivity and berry yield in perennial fruit crops.
Differential thermal analysis (DTA) of buds is a long-established method for quantifying grapevine cold hardiness, based on measuring the ability of dormant buds to avoid freezing through a supercooling mechanism (Wolf and Pool 1987). Despite its widespread use, DTA is limited by variations introduced during sample preparation (Londo et al. 2023), by the challenges of processing large sample sizes, and by the subjective nature of data interpretation. These limitations highlight the need for alternative, high-throughput, and automated phenotyping methods to assess cold hardiness.
An extensive review of high-throughput plant phenotyping technologies for estimating plant water status included multispectral and hyperspectral imaging, particularly within the near-infrared (780 to 1400 nm) and shortwave infrared (1400 to 3000 nm) regions (Carvalho et al. 2021). Specific spectral reflectance bands such as R1300 (Jones et al. 2004), R1450 (Seelig et al. 2009, Wang et al. 2009), and R1956 (Yang et al. 2021) have gained significant attention due to their stronger associations with leaf water content, compared to earlier indices in the 780 to 1400 nm range (Peñuelas et al. 1993, 1997, Prince et al. 2016). A combination of higher bud water content (free or bound) and below-freezing temperature is lethal to grapevine buds during mid-winter. Here, our study explored the potential of reflectance spectra to monitor water content in the context of bud cold-hardiness properties of dormant buds in grapevine canes. We leveraged spectral information collected on buds during winter and on vine canopy during summer to compare growth and yield morpho-physiological performance against cold-hardiness features of grapevines.
Hyperspectral data from buds were collected from 87 progenies derived from a cross between a cold-sensitive parent (NY81) and a cold-tolerant V. amurensis accession (PI 588634). The goal was to investigate water dynamics across winter and summer and to identify cross-seasonal traits in grapevines that integrate screening for cold hardiness, vine physiology, and berry yield. We used hyperspectral sensors (400 to 2500 nm wavelength range) and multispectral sensors (400 to 1200 nm range) to compute indices related to bud and leaf-specific water content and morpho-physiological traits. These measurements were used to establish relationships between cold hardiness and overall grapevine performance and productivity. The specific objectives of this study were to validate the use of spectrometry for quantifying grapevine bud cold hardiness during mid-winter, to identify spectral indices associated with bud and leaf water content that confer freezing tolerance and promote improved vine physiology and performance, and to determine key spectral indices that enable selection of beneficial traits across seasons to enhance vine adaptability and vineyard productivity.
Materials and Methods
Plant material and experimental site
A segregating F1 family consisting of 252 progeny vines was generated by crossing the French-American Vitis hybrid breeding line NY84.0101.03 with the cold-hardy V. amurensis pollen donor PI 588634. NY84.0101.03 is derived from a cross between NY62.0136.01 (Seibel 14665 × Seneca) and Ravat 34 (Plantet × unknown). These vines were planted in 2015 at the USDA-ARS research vineyard in Geneva, NY (42°89′N; 77°01′W; 161 m asl) in rows spaced 2.74 m apart with 1.82 m inter-vine spacing. For this study, 86 progeny vines were randomly selected for comprehensive evaluation of mid-winter cold hardiness and summer vine performance. Grapevine buds were collected during winter 2023 and 2024 for cold-hardiness evaluation using DTA. To explore the relationship between total bud water content and cold-hardiness responses observed through DTA, we employed hyperspectral reflectance sensors to compute indices associated with bud water status in the context of freezing tolerance.
To further examine water content dynamics across tissue types and seasons, commonly known water-sensitive hyperspectral signatures were analyzed in bud tissue (measured in the winters of 2023 and 2024). To establish the relationship between specific spectral signatures and actual bud water content, total water content in buds was measured by calculating the difference between fresh weight and dry weight. Buds were collected during winter 2024 and oven-dried at 60°C for 2 days. To assess potential carryover effects between seasons and to identify proxy traits for selecting stress resilience and vine productivity, in summer 2024, both key leaf physiological traits were measured and canopy multispectral imaging was performed at pre- and postveraison. Seasonal weather data including air and soil temperature, soil tension, solar radiation, and total precipitation were obtained from the Network for Environment and Weather Applications (www.newa.cornell.edu) for the station nearest to the experimental site (Supplemental Figure 1).
DTA and hyperspectral imaging of grapevine tissues
During mid-winter (December) 2023 and 2024, cold-hardiness data (low-temperature exotherms, or the lethal temperatures at which intracellular freezing occurs in buds, causing 50% or more tissue death [LT50], referred to hereafter as BudLTE) were collected from freshly excised dormant buds (eight to 10 per sample) obtained from representative canes of the selected progenies. Bud sample handling, processing, and analysis were conducted on the same day of collection using a programmable freezer, following the standard DTA method (Londo and Kovaleski 2017). Hyperspectral reflectance data of the dormant buds were determined using the automated Phoenix 6000 NIR Research Analyzer (Blue Sun Scientific, LLC). Hyperspectral absorbance data were recorded across wavelengths ranging from 400 to 2500 nm, at 1 nm resolution. During sample analysis, freshly excised dormant buds (n = 10 to 15; collected from the aforementioned representative canes) were placed closely together, with the bud cushion facing downward into small transparent cups (4 cm in diameter) that were sealed with white plastic caps to minimize light scattering outside the sample cup. The cups were then loaded onto the inbuilt sample carrier of the system, which includes embedded wavelength and reference standards for autocalibration (see https://www.bluesunscientific.com/phoenix-6000-tw). The carrier moves both laterally and rotationally during scanning, enabling comprehensive hyperspectral absorbance data collection from the bud samples, while minimizing sampling errors and improving measurement repeatability.
Reflectance values were converted into absorbance values using the formula Absorbance = log (1 / Reflectance). Spectral data were extracted using the built-in BlueSun diagnostics software. Full specifications of the instrument can be found at https://www.bluesunscientific.com/_files/ugd/1738b7_1a0c7b58f83240eaadf0ff136b978b26.pdf.
Measurement of leaf physiological traits in relation to water dynamics
The progenies were evaluated for morpho-physiological traits at both pre- and postveraison during summer 2024. Measurements were taken on the youngest fully expanded leaf from two representative shoots, selected from the proximal and distal ends of the left and/or right cordons of each vine, resulting in either four or eight readings per progeny vine on two consecutive sunny days (Supplemental Figure 1). At preveraison (early June 2024), physiological data were collected using an LI-600 porometer/ fluorometer (LI-COR Environmental). The parameters measured included stomatal conductance (gs; mol H2O/m2/sec), evapotranspiration (ET; mmol H2O/m2/sec), leaf-to-air vapor pressure deficit (VPD; kPa), leaf water content (LW; %), leaf humidity (LH; %), leaf temperature (LT; °C), and electron transport rate (ETR; μ mol/electrons m2/sec). At veraison (late July to August 2024), photosynthetic efficiency parameters were determined using the PhotosynQ MultispeQ v2.0 device (https://www.photosynq.com/product-page/multispeq-v-2-0), employing the ‘Photosynthesis RIDES’ protocol. Measurements were taken on the youngest fully expanded leaf from two representative shoots, selected from the proximal and distal ends of the left and/or right cordons of each vine, resulting in either four or eight readings per progeny vine over two consecutive sunny days (Supplemental Figure 1). Parameters included leaf temperature differential (Ltdiff); chlorophyll fluorescence traits such as maximum fluorescence (Fm), minimum fluorescence (Fo), steady-state fluorescence (Fs); photochemical maximum quantum efficiency (Fv/Fm); and photosystem II quantum yield (Phi2) using PhotosynQ. Morphological traits were also scored using a modified version of the International Plant Genetic Resources Institute’s grapevine descriptors (IPGRI 1997). Cluster size was visually categorized as: 1, small (average cluster length <5.1 cm); 2, medium (5.1 to 10.2 cm); or 3, large (>10.2 cm). Berry size was scored as: 1, small (average diameter <10 mm) or 2, medium (average diameter >10 mm).
At veraison, multispectral imaging of selected whole-vine canopies was conducted using a DJI Matrice 100 quadcopter equipped with a five-channel MicaSense RedEdge-M camera, capturing spectra from blue (475 nm) to near-infrared (840 nm) on a sunny day. Flights were conducted at 90 m altitude on clear days between 1030 and 1100 hr using the Pix4DCapture app, with flight parameters set to 75% side and front overlap, yielding ~20 cm/pixel resolution. Orthomosaic maps were constructed using Pix4D Mapper, with reflectance values calibrated from panel readings taken before and after flight, and ground control points for spatial accuracy. After the segmentation of vine canopies, downstream analyses were performed using the Fiji image analysis platform (Schindelin et al. 2012). Five images were merged into a stack and aligned using Linear Stack Alignment with the SIFT registration tool. To address canopy overlaps, regions of interest were manually defined using the rectangle tool, and mean greyscale pixel values (0 to 1) were recorded for each vine. Plant growth and pigment-related indices were computed from the leaf tissue, including normalized difference vegetation index (NDVI; Rodríguez-Pérez et al. 2007), structure independent pigment index (SIPI; Peñuelas et al. 1995), anthocyanin reflectance index (ARI; Gitelson et al. 2001), and carotenoid reflectance index (CRI; Peñuelas et al. 1993).
Association of spectral reflectance between tissues and seasons
We compared the associations between spectral indices computed across different tissues (bud and vine canopy) and seasons (winter and summer) to establish links within reflectance data and to identify suitable indices and physiological traits that can serve as proxies for selecting stress resilience and productivity in grapevines. Furthermore, spectral indices associated with key physiological traits such as gs and VPD, both of which are indicators of vine performance and productivity, were validated using a subset of progenies selected based on contrasting gs values measured during summers 2024 and 2025. Cluster compactness was visually scored based on number of full berries per cluster and was categorized as 1, very low; 2, low; or 3, high. Based on the relationship between the BudLTE and leaf physiological measures collected, we identified key functional traits that can be used to select for both winter hardiness and overall performance in grapevines.
Statistical analysis
Descriptive statistics and assessments of phenotypic trait normality were conducted using SAS JMP Pro ver. 18.1.0 (JMP Statistical Discovery LLC). Bootstrap forest predictive modeling was used to extract hyperspectral absorbance predictors with more than 10% contribution. Our primary focus was on the wavelength range between 1450 to 1950 nm to identify signatures related to bud water content and their previously established significant association with plant tissue water content (Peñuelas et al. 1997). To explore the significant relationships among leaf morpho-physiological attributes and computed hyperspectral indices, principal component analysis (PCA) was performed on the correlation matrix in SAS JMP Pro with trait means and inferred at significance levels p < 0.05 and 0.1.
Results
Phenotypic variation for bud cold hardiness and water content absorbance indices
Temperature fluctuations near freezing during winter 2023 to 2024 were mild in contrast to winter 2024 to 2025, which included a severe mid-winter freezing event, as indicated by the marked temperature drops in the climatic data (Supplemental Figure 1). The progeny vines developed from the cross between NY81 and V. amurensis appeared to have adapted their bud cold hardiness in response to seasonal variation, as reflected in the BudLTE values (derived by using LT50) between the 2023 and 2024 mid-winter seasons (Figure 1A). In 2023, LT50 values ranged from −6.24 to −22.08°C, with a mean of −11.27°C. In contrast, the 2024 LT50 values ranged from −6.0 to −32.38°C, with a significantly lower mean of −28.14°C, highlighting a harsher winter and greater cold adaptation.
A) Average low-temperature exothermic data (BudLTE) by differential thermal analysis reflecting cold-hardiness levels of the progeny plants tested in two winter seasons and B) bud hyperspectral absorbance profiles within the wavelength range from 1544 to 1950 nm in winters 2023 to 2024 and C) 2024 to 2025.
To identify specific spectral absorption features associated with water content, the bootstrap forest method was used to determine influential absorbance bands within the 1450 to 1950 nm range. Key predictors of bud water content and cold hardiness identified during the 2023 winter included wavelengths at 1544, 1609, 1879, 1949, and 1950 nm. For 2024, important predictors included 1868, 1873, 1874, 1883, 1887, and 1888 nm. In both years, the wavelength ranges from 1544 to 1940 nm showed reproducibility as proxies of BudLTE data, irrespective of the severity of winter differences (Figure 1B and 1C). A comparison between the two winters revealed stable predictors of bud cold hardiness across years. Specifically, in 2023, absorbance at 1874, 1875, and 1888 nm was ranked 7th, 8th, and 11th, respectively. In 2024, absorbance at 1874, 1875, and 1888 nm was ranked 2nd, 7th, and 3rd, respectively (Table 1). These bands demonstrate consistent predictive power regardless of seasonal variation in cold severity (Figure 2).
List of hyperspectral absorbance (ABS) wavelengths identified as better predictors for bud cold hardiness in each winter season.
Specific bud hyperspectral (HS) absorbance signatures identified as better predictors of bud cold hardiness evaluated at the absorbance wavelengths of 1874, 1875, and 1888 nm across two winter seasons: 2023 to 2024 (23_24; mild winter) and 2024 to 2025 (24_25; colder winter).
To establish the relationship between the specific spectral signatures identified and actual bud water content, total bud water content (BudTWC) was measured during the winter of 2024 to 2025. This analysis revealed that all eight spectral signatures previously associated with BudLTE (1868, 1874, 1879, 1883, 1887, 1888, 1949, and 1950 nm) were also significantly related to BudTWC (Table 2). These findings further support the potential application of these spectral signatures as effective tools for screening cold tolerance in grapevines.
Relationship of hyperspectral predictors for bud cold hardiness, water content, and bud exotherm temperature. ABS, hyperspectral absorbance; BudTWC, total bud water content; BudLTE, low temperature exotherm; ns, not significant.
Variability of leaf physiology traits and cluster and berry size
Phenotypic variability in physiological traits measured at pre- and postveraison is presented (Supplemental Table 1). The mean gs values measured across the population were 0.21 mol H2O/m2/sec preveraison and 0.14 mol H2O/m2/sec postveraison. During preveraison, vine leaves were physiologically active, exhibiting a higher ET rate (mean = 3.92 ± 1.64 mmol H2O/m2/sec) and elevated ETR (Pre-ETR; mean = 179.41 ± 64.15), in contrast to postveraison (Table 2).
The ETR values measured at veraison (Ver-ETR) declined, indicating that vines were beginning to lose foliage but had not yet lost all leaves. This physiological shift is consistent with other fluorescence-related characteristics measured in the study, particularly Fv/Fm, which showed relatively high values (mean = 0.71 ± 0.06) suggesting that the photosynthetic apparatus was still functionally active. Among the physiological traits measured, all parameters assessed at preveraison (except for Pre-VPD) were positively associated with cluster and berry size. All chlorophyll fluorescence traits related to photosystem activity measured at veraison were likewise positively correlated with cluster and berry size, underscoring the importance of sustained photosynthetic efficiency during ripening. The correlation matrix showing the significance based on p values among traits measured is provided (Supplemental Figure 2).
Link between leaf physiology, cluster and berry size, and spectral indices
Significant associations between leaf physiology, berry visual ratings, and vine canopy spectral signatures were revealed through pairwise correlation analysis (Figure 3). PCA of physiological traits, chlorophyll fluorescence parameters, canopy spectral indices measured during summer 2024, and winter bud cold-hardiness measures (BudLTE) indicated that PC1 and PC2 explained 20% and 16% of the total phenotypic variation, respectively (Figure 4). Interestingly, VPD was found to be positively and negatively associated with LW and gs, respectively, at both phenological stages. Among the canopy spectral indices, NDVI was strongly associated with gs measured at preveraison. In contrast, CRI and SIPI showed significant positive correlations with gs measured at veraison. Notably, ARI was positively associated with berry and cluster size and showed strong correlations with all chlorophyll fluorescence parameters (Fm, Fo, Fs, Fv/Fm, and Phi2), highlighting its potential as a spectral indicator of photosystem efficiency and vine productivity in grapevine.
Correlation matrix showing the relationship between leaf physiological traits measured pre- and postveraison, canopy spectral indices (structure independent pigment index [SIPI], carotenoid reflectance index [CRI], anthocyanin reflectance index [ARI], and normalized difference vegetation index [NDVI]), and cluster and berry size measured during summer 2024 and average low-temperature exothermic data (BudLTE) measured (based on differential thermal analysis) in winters 2023 to 2024 (2023_24) and 2024 to 2025 (2024_25). Ver, physiological measures observed at veraison; Pre, physiological measures observed at preveraison; gs, stomatal conductance (mol H2O/m2/sec); ET, evapotranspiration (mmol H2O/m2/ sec); VPD, leaf-to-air vapor pressure deficit (kPa); LW, leaf water (%); LH, leaf humidity (%); LT, leaf temperature (°C); ETR, electron transport rate (μ mol/electrons m2/sec); Ltdiff, leaf temperature differential; Fm, maximum fluorescence, Fo, minimum fluorescence; Fs, steady state fluorescence; Fv/Fm, photochemical maximum quantum efficiency; Phi2, photosystem II quantum yield.
Principal component analysis to show the relationship between physiological measures observed preveraison (Pre) and at veraison (Ver), canopy spectral indices measured (normalized difference vegetation index [NDVI], structure independent pigment index [SIPI], anthocyanin reflectance index [ARI], and carotenoid reflectance index [CRI]), and bud cold hardiness (average low-temperature exothermic data [BudLTE]) from two winter seasons, 2023 to 2024 (2023_24) and 2024 to 2025 (2024_25). gs, stomatal conductance (mol H2O/m2/sec); ET, evapotranspiration (mmol H2O/m2/sec); VPD, leaf-to-air vapor pressure deficit (kPa); LW, leaf water (%); LH, leaf humidity (%); LT, leaf temperature (°C); ETR, electron transport rate (μ mol/electrons m2/sec); Ltdiff, leaf temperature differential; Fm, maximum fluorescence, Fo, minimum fluorescence; Fs, steady state fluorescence; Fv/Fm, photochemical maximum quantum efficiency; Phi2, photosystem II quantum yield.
Association between leaf physiology and BudLTE data
Forty-four progenies were selected based on their cold hardiness, ranked by lower (more negative) BudLTE values for each winter and by their corresponding physiological attributes measured at preveraison and veraison during summer 2024. Notably, due to variation in winter severity between the two seasons, gs at both developmental stages from summer 2024 showed both positive and negative associations with LT50 values in 2023 and 2024. This pattern suggests a complex relationship between stomatal regulation and cold tolerance, highlighting the potential of using gs as a selection criterion. Understanding carryover effects is essential for managing vine stress and improving cold tolerance. To address the reproducibility of leaf physiological responses observed in a single season, 12 progenies were selected based on gs and cluster compactness characteristics at veraison during summers 2024 and 2025 (Table 3 and Figure 5). This approach aims to support the development of grapevines with both enhanced cold tolerance and maintained productivity. Based on this approach, 12 progenies (9-37, 9-53, 9-56, 9-60, 9-74, 10-37, 11-4, 11-9, 12-60, 12-68, 12-69, and 12-73) were selected as prebreeding materials for combining traits associated with cold tolerance (Table 3) and berry yield under varying winter conditions. In this subset, it was noteworthy that leaf ARI remained consistent between the two summers (R2 = 0.433) and showed a strong relationship with BudLTE in winters 2023 to 2024 and 2024 to 2025. Additionally, ARI significantly influenced other leaf physiological traits, including ET, VPD, LW, and LT, with correlation coefficients of −0.56, −0.63, −0.62, and −0.65, respectively (p < 0.05). Given the reproducibility of these results, ARI may serve as a reliable proxy for cold-hardiness selection during summer.
List of selected progenies with consistent bud cold hardiness and their leaf physiological traits measured at veraison. BudLTE, low temperature exotherm; gs, stomatal conductance; VPD, leaf-to-air vapor pressure deficit.
Bud hyperspectral absorbance profiles recorded between 1874 and 1888 nm wavelengths of 12 progenies selected based on their cold hardiness across two winter seasons evaluated in A) 2023 to 2024 and B) 2024 to 2025.
Discussion
This study was designed to investigate bud cold-hardiness traits in a cold-hardy wild grape species, V. amurensis, used in developing a mapping population with a French American Vitis vinifera hybrid line. This population was selected due to the well-documented cold tolerance of V. amurensis (Zhao et al. 2020, Ma et al. 2022). DTA is a traditional and widely used method for quantifying bud cold hardiness in dormant grapevine buds (Wolf and Pool 1987, Londo and Kovaleski 2017, Wang et al. 2022). However, DTA has inherent limitations particularly related to sample preparation, throughput, and subjective data interpretation (Londo et al. 2023). A recent review emphasized the potential of sensor technologies, particularly those operating in the near-infrared (780 to 1400 nm) and shortwave infrared (1400 to 3000 nm) spectra, for estimating water content in plant tissues (Carvalho et al. 2021). While previous studies focused primarily on identifying spectral indices related to LW (Peñuelas et al. 1993, 1997, Jones et al. 2004, Seelig et al. 2009, Prince et al. 2016, Yang et al. 2021), to our knowledge, none explored this approach for bud tissues. Because bud water content is a key physiological determinant of cold hardiness in dormant grapevine buds, we leveraged bud hyperspectral data to identify specific absorbance wavelengths associated with water content. These spectral features were then linked to BudLTE data, a traditional way of measuring cold hardiness using LT50. Based on our findings, we propose a reliable, high-throughput phenotyping method for assessing water content in overwintering buds. This approach offers a promising alternative to traditional DTA measurements and an efficient alternative for evaluating bud cold hardiness in grapevines.
Bud spectral absorbances as predictors of cold hardiness
We focused on the band depth and area of absorption features at ~1450 and 1950 nm, as this wavelength range is well established for its association with water content in plant systems (Seelig et al. 2009, Wang et al. 2009, Yang et al. 2021). In winter 2024 to 2025, we observed a significant relationship between actual BudTWC and water absorbance signatures (Table 2), which also influenced BudLTE across both winters. This promising result further supports the carryover effects of winter temperatures on vine physiology. Bud water dynamics have been identified as a key freezing tolerance strategy in conifers (Westman and Price 1988), figs (Knipling 1970), and grapevines (Chinnusamy et al. 2007, Wang et al. 2022, Ralser et al. 2024). Although the severity of winter conditions varied between the two years of the study, the most sensitive absorbance bands for determining bud water content consistently fell within the 1860 to 1950 nm range. Similar bands at 1912 and 1980 nm have previously been reported as key determinants of LW in wheat (Yang et al. 2021). Specifically, absorbances at 1874, 1875, and 1888 nm were identified as strong predictors of bud cold hardiness in grapevine. A similar absorption range (1830 to 2080 nm) was reported as a key spectral window for detecting water content, with high correlation in plant tissues that is largely unaffected by tissue structure (Zhang et al. 2012).
Leaf physiological and canopy spectral indices
The progenies of the mapping population maintained mean Fv/ Fm values above 0.6 during veraison, indicating the functional integrity of photosystem II (PSII) reaction centers and a greater capacity for non-radiative energy dissipation (Osmond and Grace 1995). A reduction in Fv/Fm values, as observed in the grapevine cultivar Tempranillo, has been associated with a decrease in the plastoquinone pool and reduced PSII efficiency (Maroco et al. 2002). Interestingly, progenies with V. amurensis background exhibited a unique physiological behavior: minimizing stomatal opening during veraison while maintaining high ET rates to support fruit development. This adaptation reflects a critical balance between water conservation and productivity. All chlorophyll fluorescence parameters measured in this study were strongly related to photosystem activity during veraison and showed positive associations with cluster and berry size.
These progenies appear to be water-efficient, capable of adapting to mild-to-severe summer conditions by effectively regulating the VPD between leaf and air while maintaining functional gs. VPD has been reported as a limiting factor that can disrupt the balance between vine growth and yield under environmental stress (Palliotti et al. 2009).
Among the canopy spectral indices measured, NDVI was closely associated with gs at preveraison, while CRI and SIPI were strongly correlated with gs during veraison. These findings are supported by previous research linking similar spectral indices with ground truth data on canopy morphometrics, leaf chlorophyll content, water status, vine vigor, and berry yield in grapevines (Gutiérrez-Rodríguez et al. 2004, Pou et al. 2022, Sharma et al. 2024). Likewise, pigment- and water-specific reflectance features in leaf tissue have been widely used to assess grapevine morphophysiological properties (Matese et al. 2022).
It is well established that carotenoid content is represented by the CRI index, which plays a key role in photoprotection, particularly under stress conditions. Carotenoids are central to the xanthophyll cycle, which dissipates excess light energy during partial stomatal closure (Demmig-Adams and Adams 1996, Zhou et al. 2017). Among the canopy spectral indices assessed, ARI was positively associated with berry and cluster size, as well as with all chlorophyll fluorescence parameters, highlighting its relevance to photosystem efficiency. Among all the leaf spectral indices evaluated in this study, ARI was found to be the most stable proxy for selecting cold tolerance across varying summers. Notably, increased anthocyanin accumulation in grapevine leaves has also been reported as a photoprotective response under water-limited conditions (Qin et al. 2011).
Selection for cold hardiness and vine performance
PCA conducted on 44 progenies selected based on both their cold-hardiness values across two winters, as well as their physiological attributes, revealed that gs was associated with BudLTE values measured at both preveraison and veraison. Interestingly, BudLTE values in winters 2023 to 2024 and 2024 to 2025 showed both positive and negative associations with gs, reflecting a complex and possibly environment-dependent relationship. The connection between summer physiological traits and winter bud responses is not novel: previous studies of grapevines have shown that disruptions in winter physiology can carry over into the subsequent growing season (Burakowski et al. 2022). Based on this premise, we selected 12 progenies with gs values close to the population mean at both preveraison and veraison. These were identified as promising breeding materials to combine traits for cold tolerance and berry yield improvement in grapevine.
This selection strategy may have implications for overall vine architecture, as hardy perennial plants often exhibit relatively compact growth forms, likely due to increased carbon allocation toward stress mitigation (Alpert 2006, Prats and Brodersen 2020). This observation is consistent with the energy demands associated with the supercooling ability of cold-hardy buds, which involves the breakdown of energy reserves to produce osmolytes that lower the freezing point of cellular water during winter (Pearce 2001, Holmlund 2021, Kovaleski et al. 2023). Such trade-offs between plant traits aimed at maintaining fitness under unpredictable environmental conditions have been documented (Dwivedi et al. 2021), further supporting the integrative approach of selecting for both physiological resilience and agronomic performance.
Conclusions
In this study, hyperspectral parameters based on multi-year field experiments that were conducted across different seasons were used to improve the throughput and accuracy of monitoring bud water dynamics in dormant grapevine buds during mid-winter. The results have important theoretical and practical implications for understanding bud water content in relation to cold hardiness, and for identifying key traits relevant to grapevine breeding efforts aimed to enhance cold tolerance while maintaining high berry yield. Our findings demonstrate that bud water content in grapevines can be effectively tracked through absorbance measurements and that hyperspectral signatures can serve as reliable proxies for bud cold hardiness, traditionally measured using DTA. Additionally, we identified two leaf spectral reflectance bands (1868 and 1873 nm, measured during summer) which showed strong correlation with BudLTE values and can potentially be used as proxy traits for screening cold-hardiness in grapevines. Based on these insights, we propose that the selected progenies identified in this study may serve as promising prebreeding materials for the development of grapevine cultivars with improved cold tolerance, enhanced vine performance, and sustained berry yield.
Supplemental Data
The following supplemental materials are available for this article in the Supplemental tab above:
Supplemental Table 1 Data distribution statistics of leaf physiological traits and canopy multispectral indices measured in summer 2024. SD, standard deviation.
Supplemental Figure 1 Monthly averages of solar radiation and soil tension (A); total precipitation (B); and average, maximum, and minimum air temperatures and soil temperature (C) at the field site in winter and spring 2023 to 2024; and summer, fall, and winter 2024 to 2025. Red arrows indicate the respective temperature drop during the sample collection in mid-winter of both year ranges; yellow arrows designate the leaf physiology measurement time windows at preveraison (PV; early June) and veraison (V; between July and August), allowing to quantify the possible carryover effects of winter 2023 to 2024 on the subsequent season’s growth and yield. Different background colors demarcate different seasons in each year on the graphs.
Supplemental Figure 2 Correlation matrix showing the significance (based on p value) of relationships among leaf physiological traits measured pre- and postveraison, canopy spectral indices (structure independent pigment index [SIPI], carotenoid reflectance index [CRI], anthocyanin reflectance index [ARI], and normalized difference vegetation index [NDVI]), and cluster and berry size measured during summer 2024 and average low-temperature exothermic data (BudLTE) measured (based on differential thermal analysis) in winters 2023 to 2024 (2023_24) and 2024 to 2025 (2024_25). Ver, physiological measures observed at veraison; Pre, physiological measures observed at preveraison; gs, stomatal conductance (mol H2O/m2/sec); ET, evapotranspiration (mmol H2O/m2/sec); VPD, leaf-to-air vapor pressure deficit (kPa); LW, leaf water (%); LH, leaf humidity (%); LT, leaf temperature (°C); ETR, electron transport rate (μ mol/electrons m2/sec); Ltdiff, leaf temperature differential; Fm, maximum fluorescence, Fo, minimum fluorescence; Fs, steady state fluorescence; Fv/Fm, photochemical maximum quantum efficiency; Phi2, photosystem II quantum yield.
Data Availability
All data underlying this study are included in the manuscript and its supplemental information.
Footnotes
The authors thank Lukas Rood, Payton Kliesch, and Kathleen Deys for help collecting field measurements. Special thanks to the Gold Lab (CALS, Cornell University, NY) for providing the SVC-HR1024i portable spectroradiometer for respective measurements. We appreciate the support provided by Tyler Gordon (PGRU, USDA-ARS), Jason Londo, and Awais Khan (Cornell University) by sharing resources on hyperspectral, DTA, and multispectral data sensors. This work was supported by the USDA-ARS project: Genetic Improvement of Grape Quality and Adaptation to Diseases and Abiotic Stress: 8060-21220-008-000-D.
Weerarathne V, Paudel A, Sehgal A, Tegtmeier R, Ikeogu U, Ahn E et al. 2025. Rapid determination of bud and leaf water content using hyperspectral sensors to monitor cold hardiness in grapevine. Am J Enol Vitic 76:0760027. DOI: 10.5344/ajev.2025.25019
By downloading and/or receiving this article, you agree to the Disclaimer of Warranties and Liability. If you do not agree to the Disclaimers, do not download and/or accept this article.
- Received April 2025.
- Accepted September 2025.
- Published online November 2025
This is an open access article distributed under the CC BY 4.0 license.
References
Vol 76 Issue 2
Jump to section
Related Articles
Cited By...
More from this TOC section
Similar Articles
