Water availability affects the capability of reflectance indices to estimate berry yield and quality attributes in rain-fed vineyards

Citation: Serrano, L.; Gorchs, G. Water Availability Affects the Capability of Reflectance Indices to Estimate Berry Yield and Quality Attributes in Rain-Fed Vineyards. Agronomy 2022,12, 2091. https:// doi.org/10.3390/agronomy12092091 Academic Editors: José Ramón Rodríguez-Pérez and Francisco Manzano Agugliaro Received: 16 June 2022 Accepted: 28 August 2022 Published: 1 September 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). agronomy Article Water Availability Affects the Capability of Reflectance Indices to Estimate Berry Yield and Quality Attributes in Rain-Fed Vineyards Lydia Serrano * and Gil Gorchs Departament d’Enginyeria Agroalimentària i Biotecnologia, Universitat Politècnica de Catalunya, 08860 Castelldefels, Barcelona, Spain *Correspondence: [email protected]; Tel.: +34-935-522-132 Abstract: Remote sensing methods are known to provide estimates of berry quality. However, previous studies have shown that the Normalized Difference Vegetation Index (NDVI) failed to predict berry quality attributes in rain-fed vineyards. This study explores the association of several reflectance indices with vine biophysical characteristics and berry yield and quality attributes and their temporal stability. The study was conducted in rain-fed Chardonnay vineyards located around Masquefa (Penedès region, Catalonia, Spain) over four years. Canopy reflectance, fractional Intercepted Photosynthetic Active Radiation, predawn water potential and canopy temperature at midday were measured at veraison whereas berry yield and quality attributes were determined at harvest. Water availability and vine biophysical attributes showed large temporal stability whereas berry quality attributes were not temporally stable. The capability of reflectance indices to estimate berry quality attributes was subject to the timing and extent of water deficits. The Photochemical Reflectance Index (PRI), the NDVI and the Water Index (WI) provided estimates of berry quality attributes under mild, moderate and severe water deficits, respectively. These results might have potential applications in precision viticulture activities such as selective harvesting according to grape quality attributes and the assessment of ripening. Keywords: berry yield and quality attributes; rain-fed vineyards; reflectance indices; water availability; NDVI; WI; PRI 1. Introduction A main current interest to viticulturists and wine industries is to identify zones of distinct grape quality for differential harvest in order to produce wines with different characteristics and properties [ 1 – 3 ]. In grapevines, berry quality largely depends on sugar/acid balance at harvest. Indeed, a certain amount of sugar content (i.e., Total Soluble Solids, TSS) is necessary in order to produce enough alcohol during fermentation whereas acidity (Titratable Acidity, TA) has to be reduced during ripening to a level that will be palatable. Thus, the ratio TSS/TA (maturity index, IMAD) indicates the sugar/acid balance. Both TSS and TA are commonly measured by repeated sampling during the ripening process to assess berry quality and maturity and, consequently, to determine the optimal harvest date [4,5]. Environmental conditions (i.e., climate and soil) as well as vineyard management practices influence the composition and, thus, the quality of grapes [ 6 ]. Despite the complex effects of environmental conditions on grape quality, vine water status has been long recognized as the factor most comprehensively determining berry ripening and composition [ 7 – 11 ]. Indeed, changes in vine water status, particularly at critical phenological stages, have a direct effect on grape composition by influencing vegetative growth, canopy microclimate and fruit growth and metabolism [ 7 , 8 , 10 , 11 ]. In addition, vine water status Agronomy 2022,12, 2091. https://doi.org/10.3390/agronomy12092091 https://www.mdpi.com/journal/agronomy Agronomy 2022,12, 2091 2 of 16 also influences fruit composition through an indirect effect on berry size, which decreases in vines subjected to water deficits [10,12]. The extent and seasonal timing of water deficits is largely determined by site variables (i.e., soil type and topography) as well as by water supply (rainfall and irrigation) [ 7 , 13 , 14 ]. Soil type and topography are spatially variable whereas water supply varies both in space and in time, particularly in Mediterranean rain-fed vineyards where rainfall is irregular and highly unpredictable. Therefore, in rain-fed vineyards, berry quality attributes might be expected to show large variability both within a year and across years (i.e., low temporal stability). Indeed, previous studies conducted at the field level have reported low temporal stability in berry quality both under rain-fed [15] and irrigated conditions [16]. Since berry quality attributes relate to the timing and extent of water deficits—and given that water deficits vary both spatially and temporally—it is inefficient to employ conventional methods to measure berry quality. As an alternative, remote sensing offers the possibility of a rapid assessment of large vineyard areas, avoiding the need for a large number of measurements of individual samples that are cost and time consuming. Previous studies have shown the capability of remote sensing methods to estimate vine biophysical variables such as size and vigour, potential indicators of berry quality and yield while providing opportunities for cost-effective generation of spatial data amenable to precision viticulture activities [ 17 , 18 ]. Vegetation indices derived from the red and near infrared bands have been used to map relative differences in vine canopy vigour and to estimate differences in fruit yield and quality attributes [ 4 , 17 , 19 ]. However, studies conducted under rain-fed conditions found that vegetation indices, such as the Normalized Difference Vegetation Index (NDVI), failed to predict berry quality parameters [ 20 – 22 ] whereas González-Flor et al. [23] showed that the capability of NDVI to estimate TSS was subject to the timing of occurrence of water deficits (i.e., before or after veraison). Thus, characterization of canopy structure appears less useful at estimating the effects of water stress on berry composition. Water deficits decrease the synthesis and accumulation of sugars and acids in berries not only through decreases in leaf area but also by inducing stomatal closure (which limits photosynthesis). Therefore, spectral indices capable of assessing plant water status or their effects on photosynthetic functioning might provide information potentially linked to fruit quality. Indeed, previous studies have shown the capability of the Photochemical Reflectance Index (PRI) [ 24 , 25 ], an indicator of the epoxidation state of xanthophyll pigments and, thus, of photosynthetic efficiency, to estimate berry quality [ 26 ] under mild to moderate water deficits. Similarly, the Water Index (WI), an indicator of plant water status [ 27 , 28 ], successfully estimated berry quality in grapevines experiencing water deficits [ 21 , 23 ]. Therefore, it appears that for in rain-fed vineyards, the timing and extent of water deficits might be critical in determining the capability of the spectral indices to assess berry yield and composition. Field spectroscopy might be useful at identifying reflectance indices to estimate berry yield and quality attributes. Indeed, in previous studies, we explored the capability of narrow-band reflectance indices at estimating berry yield and quality attributes at the canopy level while considering separately moderate to severe water deficits and mild to moderate water deficits [ 21 , 23 ]. This spatial scale (i.e., multifield or regional level) is of interest for wineries (industry) attempting to make decisions about harvest date at the regional scale for multiple vineyards simultaneously rather than single vineyard [ 22 ]. Herein, the capability of reflectance indices to estimate berry yield and quality attributes is further reassessed by considering single vines of different fields spreading over an area representative of the Designation of Origin (D.O.) Penedès through differing water availabilities. The main objective of this work was to examine opportunities to predict berry quality at harvest in rain-fed Chardonnay vineyards using reflectance indices of canopy structure and physiological status in order to: Agronomy 2022,12, 2091 3 of 16 (i) determine the temporal stability of vine vigour and water status and berry yield and quality attributes (ii) examine the association of reflectance indices to vine vigour, water status and berry yield and quality attributes (iii) study the impact of the timing and severity of water deficits on the capability of reflectance indices to estimate berry yield and quality attributes at harvest. 2. Materials and Methods 2.1. Study Site The study was conducted in ten rain-fed commercial vineyards of Vitis vinifera L. cv. Chardonnay plants (K5V1 clone) located in the Designation of Origins (D.O.), Penedès region (Catalonia, Spain), over four years (2007, 2008, 2009 and 2011). The location of the vineyards studied is provided in Table 1. Table 1. Location and area of the vineyards studied. Coordinates are provided in decimal degrees for a single vine at each vineyard. Sampled vines were less than 5 m apart from each other. Vineyard Municipality Area (ha) X Coordinate (◦) Y Coordinate (◦) Batista Masquefa 3.8 1.79620 41.49797 Hostal Masquefa 2.3 1.78292 41.50378 Valencians Piera 9.8 1.76192 41.52602 Les Planes Piera 13.8 1.76615 41.47577 La Plana Piera 3.2 1.75697 41.46499 Masover Piera 1.4 1.74533 41.48836 Isidro St. Llorenç d’Hortons 0.9 1.82534 41.47678 Casa St. Llorenç d’Hortons 1.2 1.81994 41.47524 Barraca St. Llorenç d’Hortons 0.9 1.81656 41.47701 La Creu St. Sadurní d’Anoia 3.7 1.79590 41.42612 The region has a Mediterranean climate with an average annual temperature of 15 ◦ C and mean annual rainfall of 550 mm. Weather data during the study period were obtained from a weather station located at Hostalets de Pierola (41 ◦ 31 0 59 00 N; 1 ◦ 48 0 31 00 W), which belongs to the Catalan Meteorological Network. Daily potential evapotranspiration (ET 0 ) was calculated by the Penman–Monteith equation using data from the same meteorological station. Weather water balance was calculated as the precipitation minus reference evapotranspiration (P − ET 0 ) for the entire growing cycle (from 1 November to 30 October) and at two different stages: pre-veraison (from budburst to veraison) and post-veraison (from veraison to harvest). Vines were planted between 1990 and 2001 at variable density, ranging from 2083 to 3703 stock ha –1 , and the training system was Double Royat. Soil textures are loam and loamy-silt and soil depth ranges from 0.35 m to 2.0 m among vineyards. A more detailed description of the vineyards studied in terms of soil properties and plantation characteristics is provided in Serrano et al. [ 29 ] and González-Flor et al. [ 23 ]. Thirty vines (3 vines at each vineyard) with contrasting vigour were chosen to carry out measurements at the stage of veraison, when vines attained full canopy expansion. This phenological stage was chosen on the basis of previous studies aimed at estimating berry yield and composition from remote sensing data [ 30 ]. Field data collection at veraison was carried out over the third and fourth week of July, whereas harvest was carried out around the second week of August. Agronomy 2022,12, 2091 4 of 16 2.2. Vine Water Status Predawn water potential ( Ψp ) was measured on a randomly selected mature leaf of the outer part of the canopy (1 leaf per vine) using a pressure chamber (Soilmoisture 3005, Soil Moisture Corp., Santa Barbara, CA, USA) at the stage of veraison. The canopy to air temperature difference (T canopy − T air ; ∆ T m ) of field vines was determined at midday (solar noon) using a hand-held infrared thermometer (Scheduler, SN 870401 in 2007 and ST Pro Plus, Raytek Corp., Santa Cruz, CA, USA in 2008), positioned at 20 cm of the canopy and held at an angle of 60◦so that the field of view was approximately 0.25 m2. 2.3. Vine Development and Vigour Fractional intercepted Photosynthetic Active Radiation (fIPAR) was determined using a ceptometer (Accupar, Decagon Devices Inc., Pullman, WA, USA). Data were gathered by placing the ceptometer at ground level. Measurements were taken at regular intervals (3 parallel and 4 perpendicular to the row) with the individual vine located in the central portion and avoiding border vines. Incident radiation readings were taken above the vines. Measurements were taken at midday (solar noon) within an interval of less than 1.5 h at the stage of veraison. 2.4. Spectral Measurements Canopy radiance was measured over each vine using a spectroradiometer UNISPEC (PP Systems Ltd., Havervill, MA, USA) with a 2.3 mm diameter bifurcated fibre optic (model UNI410, PP Systems, Havervill, MA, USA) fitted with a 12 ◦ field of view foreoptics (UNI-710, PP Systems Ltd., Havervill, MA, USA). The detector samples 256 bands at roughly even intervals (average band-to-band spacing 3.3 nm) within a 400–1100 nm effective spectral range. The radiometer was mounted on a tripod and held in a nadir orientation at ~0.75 m above the canopy. Four scans were internally averaged for each vine. Measurements were expressed as apparent reflectance after standardizing by the irradiance determined using a cosine corrected detector lens (UNI-685, PP Systems Ltd., Havervill, MA, USA) positioned above the canopy. Data were collected on cloudless days between 11:00 h and 13:00 h (i.e., solar noon) in order to minimize disturbances from the atmosphere and changes in solar elevation. Several hyperspectral indices related to canopy structure, water content and pigment content (i.e., chlorophyll and carotenoid) were calculated from each collected spectra. We herein report on those indices that provided significant relationships with either vine canopy structure and water status or berry quality parameters based on previous results [ 21 , 23 , 31 ]. The number of indices has been reduced in order to avoid redundancy of data. The Normalized Difference Vegetation Index (NDVI) [ 32 ] was calculated to provide canopy structure estimates (Leaf Area). In addition, The Water Index (WI) [ 28 , 33 ] was chosen to estimate vine water status. The Normalized Phaeophytinization Index (NPQI) [ 34 ] was determined to provide estimates of chlorophyll degradation. In addition, the carotenoid to chlorophyll ratio was assessed using the Structure Independent Pigment Index (SIPI) [ 35 ]. Finally, the Photochemical Reflectance Index (PRI) [ 24 , 25 ] was also determined. Indices were calculated from reflectance narrow bands as follows: NDVI = (R900 −R680)/(R900 + R680) WI = R900/R970 PRI = (R531 −R570)/(R531 + R570) SIPI = (R800 −R450)/(R800 −R650) NPQI = (R415 −R435)/(R415 + R435) Agronomy 2022,12, 2091 5 of 16 2.5. Berry Yield and Quality Attributes In the years of study, harvest took place over the second and third weeks of August when berries were at optimum ripeness for cava (sparkling wine) elaboration. Experimental vines were hand harvested, and total yield per vine was weighed. Berries were then carried to the lab in coolers and pressed, and the must, after filtration, was analysed to determine quality attributes for each sample (vine). Total soluble solids (TSS, ◦ Brix) were determined by refractometry (WM-7, ATAGO Co., Ltd., Tokyo, Japan). Total acidity (TA) was determined by titrimetry with 0.1 M NaOH to an end point of pH 8.2 using phenolphthalein indicator solution and was expressed in g tartaric acid L −1 . Maturity index (IMAD) was calculated as the ratio between TSS and TA. Berries for cava elaboration are considered at optimum ripeness at TSS ~ 18 ◦ Brix and TA ~ 10 g tartaric acid L −1 . In addition, as a surrogate for berry size, berry weight (W 100 ) was determined on a subsample of 100 berries. 2.6. Statistical Analyses Data were analysed using the statistical package IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA). Differences in the variables studied were evaluated by analysis of variance (ANOVA) and means were compared using the Tukey’s test. Variability (i.e., the degree of variation of a specific attribute within a year) in vine vigour, water status and berry yield and quality attributes was assessed using coefficients of variation. Temporal stability (i.e., whether a specific attribute systematically presents a similar spatial pattern over the years) was determined using the Kendall’s coefficient of concordance (W) for the variables studied. Kendall’s coefficient varies from 0 in case of total disagreement (i.e., no temporal stability) to 1 in case of total agreement [ 15 ]. In addition, in order to identify the association and rank the contribution of vine canopy structure and water status to berry yield and quality attributes, we carried out a series of Principal Component Analyses (PCA) where we considered several variables in conjunction: fIPAR, Ψp , berry yield and quality attributes (TSS, TA, IMAD) as well as reflectance indices (NDVI, WI, PRI SIPI, and NPQI) for the whole data set as well as for separate data sets according to the timing of occurrence (i.e., preor post-veraison) and severity (i.e., mild, moderate and severe) of water deficits. 3. Results 3.1. Weather Conditions Temperatures and precipitation over the entire growing cycle—from 1 November to 30 October—(Table 2) were close to the long-term average (i.e., mean annual temperature 15 ◦ C and annual precipitation 550 mm) in 2007 and 2008. In contrast, in 2009 and 2011, mean temperature was ca. 2 ◦ C higher than the long term average and cumulative precipitation was above the mean annual rainfall (163 mm and 188 mm higher than the long-term average for 2009 and 2011, respectively). A detailed description of weather conditions and, particularly, of the weather water balance over the years of study can be found in Serrano et al. [ 21 ] and González-Flor et al. [ 23 ]. Briefly, in 2007 and 2009, and according to the weather water balance, the water restriction was particularly severe at pre-veraison (i.e., from bud-break to veraison) with P − ET 0 = − 275 mm and P − ET 0 = − 216 mm in 2007 and 2009, respectively, whereas over the ripening period (post-veraison), there was ample water availability (Table 2). In contrast, in 2008 and 2011, water constraints had a larger incidence at post-veraison with P − ET 0 = − 131 mm and P − ET 0 = − 150 mm in 2009 and 2011, respectively, whereas water availability at pre-veraison was abundant (Table 2). In summary, in 2007 and 2009, water availability was scarce at pre-veraison whereas water restrictions had a larger incidence at post-veraison in 2008 and 2011. Agronomy 2022,12, 2091 6 of 16 Table 2. Average temperature (T mean ), precipitation (P) and reference evapotranspiration (ET 0 ) for the entire growing cycle, and weather water balance (P − ET 0 ) from bud break to veraison (pre-veraison) and from veraison to harvest (post-veraison). Data are from the meteorological station of Els Hostalets de Pierola (41◦3105900 N, 1◦4803100 W). Year Tmean (◦C) P (mm) ET0(mm) P−ET0(mm) Pre-Veraison P−ET0(mm) Post-Veraison 2007 15.0 510.8 851.4 −274.5 −41.1 2008 14.6 533.1 767.6 −97.0 −130.6 2009 17.5 687.6 928.2 −216.4 −88.7 2011 17.2 662.6 1031.8 −89.1 −150.0 3.2. Vine Vigor and Water Status There were significant differences (p< 0.05) in vine vigour and vine water status at veraison among years. Vine vigour, as indicated by fIPAR values, significantly differed among years ranging from 0.46 in 2007 to 0.70 in 2009 (Table 3). In addition, Ψp and ∆ T m at veraison showed significant differences among years. Predawn water potential ranged from − 0.88 MPa in 2007 to − 0.26 MPa in 2011 with intermediate values in 2008 and 2009, whereas ∆ T m ranged from − 4.62 ◦ C to 0.92 ◦ C (Table 3). Within a year, the coefficients of variation (CV) of fIPAR ranged from 11% in 2009 to ca. 28% in 2007. In addition, CV for Ψp predawn water potential ranged from ~24% in 2008 to ~41% in 2011, whereas ∆ T m showed large variation, particularly in 2007 (CV = 268%) and 2008 (CV = 110%). Table 3. Descriptive statistics for predawn water potential ( Ψp ), canopy to air temperature difference ( ∆ T m ) and fractional intercepted PAR (fIPAR): number of vines (n), minimum (Min), maximum (Max) and Mean values, standard deviation (SD) and coefficient of variation (CV). Year n Min Max Mean * SD CV (%) Ψp(MPa) 2007 30 −1.40 −0.30 −0.88 a 0.28 32.1 2008 27 −0.95 −0.40 −0.67 b 0.16 24.1 2009 30 −0.73 −0.20 −0.45 c 0.12 27.6 2011 30 −0.60 −0.13 −0.26 d 0.11 40.9 ∆Tm(◦C) 2007 30 −3.30 5.00 0.92 a 2.46 268 2008 27 −6.55 1.40 −1.96 b 2.14 110 2009 30 −7.15 −1.85 −4.62 c 1.59 34.4 2011 30 −5.50 0.30 −2.47 b 1.31 53.0 fIPAR 2007 30 0.16 0.66 0.46 a 0.13 27.7 2008 27 0.30 0.80 0.61 b 0.11 18.4 2009 30 0.54 0.86 0.70 c 0.08 11.0 2011 30 0.47 0.86 0.66 bc 0.09 13.3 * Different letters indicate significant differences (p< 0.05) among years according to Tukey’s test. 3.3. Berry Yield and Quality Attributes Table 4presents summary statistics for berry yield and quality attributes. Yield ranged from 3.4 kg vine −1 in 2007 to 5.8 kg vine −1 in 2011 and was significantly higher in 2011 than in the other years of study. Similarly, TSS showed significant differences among years with lower values in 2007 (17.1 ◦ Brix) than in the other years of study when TSS ranged from 18.8 ◦ Brix to 20.1 ◦ Brix. In contrast, there were no significant differences among years in TA and IMAD with an average value (calculated from annual means over the four years) of 10.2 g L−1of tartaric acid for TA and 1.92 for IMAD. Agronomy 2022,12, 2091 7 of 16 Table 4. Minimum (Min), maximum (Max) and mean (Mean) values, standard deviation (SD) and coefficient of variation (CV) observed on Yield, Total Soluble Solids (TSS), Total acidity (TA) and maturity index (IMAD) for the years of the study. Year n Min Max Mean * SD CV (%) Yield (Kg vine−1) 2007 30 1.19 5.56 3.41 a 1.33 39.0 2008 27 1.50 6.93 3.77 a 1.27 33.7 2009 24 1.92 7.28 4.22 a 1.49 35.3 2011 29 1.25 11.51 5.82 b 2.25 38.6 TSS (Brix) 2007 30 11.37 20.84 17.13 a 2.74 15.6 2008 27 16.40 22.03 19.32 b 1.58 8.2 2009 24 15.10 23.20 20.10 b 2.20 10.9 2011 29 15.50 21.80 18.78 b 1.55 8.3 TA (g tartaric acid L−1) 2007 30 6.30 14.40 9.73 a 1.85 19.0 2008 27 7.22 16.00 10.49 a 2.35 22.4 2009 24 7.39 13.13 10.02 a 1.66 16.6 2011 29 8.60 13.79 10.67 a 1.52 14.3 IMAD (TSS/TA) 2007 30 1.00 3.31 1.85 a 0.59 31.9 2008 27 1.03 2.87 1.94 a 0.51 26.2 2009 24 1.37 2.80 2.06 a 0.42 20.4 2011 29 1.34 2.54 1.81 a 0.34 18.8 * Different letters indicate significant differences (p< 0.05) among years according to Tukey’s test. Yield showed a large variation with CV ranging from 33.7% to 39.0%, whereas quality attributes showed lesser variation. Indeed, coefficients of variation ranged from 8.2 in 2008 to 15.6 in 2007 for TSS, whereas TA and IMAD showed larger variation with CV ranging from 14.3% to 22.4% for TA and from 18.8% to 31.9% for IMAD. 3.4. Temporal Stability Over the entire period of the study, the temporal stability in vine vigour and water status as well as in berry yield and quality attributes was assessed by the Kendall’s coefficient of concordance (W) (Table 5). The W evaluates to which extent a variable might be considered as time stable. Variables linked to vine water status presented high W values with Ψp (W = 0.90), ∆ T m (W = 0.69) and W 100 (W = 0.62), whereas variables linked to canopy vigour showed intermediate values with W = 0.53 for fIPAR and W = 0.422 for NDVI. Yield and TSS showed a lesser degree of concordance over the years of study (W = 0.22 and W = 0.19 for yield and TSS, respectively), whereas TA and IMAD did not show any temporal stability (i.e., W ~ 0). Table 5. Kendall’s coefficient of concordance (W) computed over the years of study. Variables are predawn water potential ( Ψp ), canopy to air temperature difference ( ∆ T m ), fractional intercepted PAR (fIPAR), Normalized Difference Vegetation Index (NDVI), Yield, weight of 100 berries (W 100 ), Total Soluble Solid (TSS), Total acidity (TA) and maturity index (IMAD). Variable W χ2Significance (pValue) Ψp0.899 72.823 0.001 ∆Tm0.686 55.572 0.001 fIPAR 0.530 42.911 0.001 NDVI 0.422 29.087 0.001 Yield 0.223 13.380 0.004 W100 * 0.616 33.267 0.001 TSS 0.188 11.303 0.010 TA 0.055 3.300 0.348 IMAD 0.019 1.000 0.801 * W100 was not measured in 2008. Agronomy 2022,12, 2091 8 of 16 3.5. PCA Analysis Principal Component Analysis was carried out to examine the association among the reflectance indices and vine attributes (vine vigour and water status, and berry yield and quality parameters). For the sake of clarity, we herein report the principal component (PC) results that accounted for at least 10% of variance and for those variables with large factor loadings (i.e., absolute coefficient’s magnitude > 0.4). When the overall data set was considered (Table 6), Component 1 (PC1) showed a strong association between NDVI and PRI, Ψp and fIPAR and represented 31% of variation. In addition, PC1 was associated with WI and yield and was negatively correlated with ∆ T m . Component 2 (PC2) associated TSS and IMAD and was negatively correlated to TA and represented 21% of variation. Component 3 (PC3) showed a strong association between NPQI and PRI accounting for 14% of variation. Overall, these three axis explained most of the variance in the data, accounting for 66% of the total variability. Table 6. Eigenvectors for measured vine vigour (fractional intercepted PAR, fIPAR), water status variables (predawn water potential, Ψp ; and canopy to air temperature difference, ∆ T m ), berry yield and quality attributes (Total Soluble Solids, TSS; total titratable acidity, TA; and maturation index, IMAD) and reflectance indices (Normalized Difference Vegetation Index, NDVI; Water Index, WI; Photochemical Reflectance Index, PRI; Structural Independent Pigment Index, SIPI; and Normalized Phaeophytinization Index, NPQI). Principal Component Analysis results are shown for the whole data set (4 years, n= 101) and for separate sets considering the timing of occurrence of water deficits (preand post-veraison, n= 51 and n= 50, respectively). Bold characters indicate variables with large factor loadings. Vine biophysical attributes are highlighted in green, berry yield and quality attributes in red and reflectance indices in blue. Variables 4 Years Pre-Veraison Post-Veraison PC1 PC2 PC3 PC4 PC1 PC2 PC3 PC1 PC2 PC3 PC4 Ψp0.752 0.393 −0.040 0.310 0.764 0.471 −0.036 0.486 0.648 0.294 0.291 ∆Tm−0.748 −0.087 0.447 0.014 −0.854 −0.109 0.247 −0.553 −0.009 0.347 −0.535 fIPAR 0.662 0.207 −0.236 0.184 0.835 0.083 −0.212 0.030 0.324 0.473 0.008 Yield 0.710 −0.296 0.249 0.254 0.619 −0.429 0.004 0.791 0.325 0.125 − 0.098 TSS 0.079 0.794 −0.295 0.085 0.274 0.817 −0.195 −0.726 0.275 0.018 0.265 TA 0.485 −0.702 −0.112 −0.125 0.487 −0.697 −0.142 0.722 −0.467 −0.179 − 0.236 IMAD − 0.364 0.896 −0.070 0.144 −0.236 0.952 −0.018 −0.828 0.413 0.107 0.317 NVDI 0.582 0.162 −0.312 −0.452 0.810 0.053 0.136 0.083 0.196 −0.654 0.363 WI 0.645 −0.178 −0.061 0.321 0.524 0.017 0.147 0.803 0.243 0.078 0.322 PRI 0.552 0.372 0.580 −0.304 0.466 0.145 0.751 −0.005 0.766 −0.479 − 0.300 SIPI − 0.397 −0.271 −0.409 0.670 −0.470 0.012 −0.696 0.121 −0.460 0.682 0.331 NPQI 0.169 0.136 0.797 0.422 −0.579 0.052 0.577 0.194 0.723 0.381 − 0.340 Eigenvalue 3.687 2.506 1.662 1.266 4.467 2.511 1.589 3.608 2.513 1.764 1.159 Variance (%) 30.7 20.9 13.9 10.5 37.2 20.9 13.2 30.1 20.9 14.7 9.7 Vines experiencing severe water deficits mostly showed negative factor scores for PC1 whereas those experiencing moderate and mild water deficits had positive factor scores (Figure 1a). In addition, PC3 separated vines that experienced mild or moderate water deficits according to the timing of occurrence of water deficits. Thus, in vines experiencing mild water deficits at post-veraison, PC3 had positive factor scores whereas in vines experiencing moderate water deficits at pre-veraison showed negative factor scores (Figure 1b). Agronomy 2022,12, 2091 9 of 16 1 (a) (b) Figure 1. Principal component (PC) scores for the overall data set of ( a ) PC1 and PC2 and ( b ) PC1 and PC3. Each symbol corresponds to a single vine under severe pre-veraison water deficits (orange symbols, n= 29), moderate pre-veraison water deficits (red symbols, n= 26), moderate post veraison water deficits (green symbols, n= 22) and mild post veraison water deficits (blue symbols, n= 24). When timing of occurrence of water deficits was considered, the PCA extracted three (71% of variance) and four (80% of variance) principal components for preand postveraison water deficits, respectively. Under pre-veraison water deficits, PC1 accounted for 37% of variance and was associated with fIPAR, NDVI, ∆ T m and Ψp , whereas PC2 accounted for 21% of variance and was mainly associated with TSS and IMAD with positive factor loadings (Table 6). In addition, PC2 was associated with yield and TA with negative factor loadings. Under post-veraison water deficits, PC1 was associated with berry quality attributes (TSS, TA, and IMAD) as well as WI and yield (30% of variance) and PC2 accounted for 20% of variance and was mainly associated with PRI, NPQI and Ψp(Table 6). When the data set was partitioned according to the severity of water deficits (i.e., mild, moderate and severe), the number of extracted PC was variable. Five, four and three PC were extracted for mild, moderate and severe water deficits, respectively (Table 7). Under mild water deficits, PC1 was primarily associated to berry yield and quality attributes (TSS, TA and IMAD) and accounted for 27% of variance. The second principal component (PC2) was related to yield, PRI, SIPI and NDVI (21% of variance), whereas the third component was associated to fIPAR, WI and NDVI (14% of variance). The fourth principal component (PC4) was related to Ψp , ∆ T m , WI and fIPAR (10% of variance). Under moderate water deficits, PC1 was related to yield, berry quality attributes (TSS, TA and IMAD) and NDVI, and accounted for 31% of variance. The second principal component (PC2) was related to WI, PRI and NPQI and to TSS (19% of variance), whereas the third component was associated to SIPI, PRI and ∆ T m (16% of variance). The fourth principal component (PC4) was related to Ψp and ∆ T m (11% of variance). Under severe water deficits, PC1 accounted for 32% of variance and was related to ∆ T m , WI, yield, NDVI, TA and IMAD, whereas PC2 (27% of explained variance) was mainly associated with PRI, SIPI, NDVI, Ψp and TSS. The third principal component (PC3) was mainly related to TSS and SIPI (11% of the accounted variance). Agronomy 2022,12, 2091 16 of 16 50. Zarco-Tejada, P.J.; Guillén-Climent, M.L.; Hernández-Clemente, R.; Catalina, A.; González, M.R.; Martín, P. Estimating leaf carotenoid content in vineyards using high resolution hyperspectral imagery acquired from an unmanned aerial vehicle (UAV). Agric. For. Meteorol. 2013,171–172, 281–294. [CrossRef] 51. Dzikiti, S.; Verreynne, J.S.; Stuckens, J.; Strever, A.; Verstraeten, W.W.; Swennen, R.; Coppin, P. Determining the water status of Satsuma mandarin trees [Citrus Unshiu Marcovitch] using spectral indices and by combining hyperspectral and physiological data. Agric. For. Meteorol. 2010,150, 369–379. [CrossRef] 52. Sun, P.; Wahbi, S.; Tsonev, T.; Haworth, M.; Liu, S.; Centritto, M. On the use of leaf spectral indices to assess water status and photosynthetic limitations in Olea europaea L. during water-stress and recovery. PLoS ONE 2014,9, e105165. [CrossRef] 53. Ballester, C.; Zarco-Tejada, P.J.; Nicolás, E.; Alarcón, J.J.; Fereres, E.; Intrigliolo, D.S.; Gonzalez-Dugo, V. Evaluating the performance of xanthophyll, chlorophyll and structure-sensitive spectral indices to detect water stress in five fruit tree species. Precis. Agric. 2018,19, 178–193. [CrossRef] 54. Suárez, L.; Zarco-Tejada, P.J.; González-Dugo, V.; Berni, J.A.J.; Sagardoy, R.; Morales, F.; Fereres, E. Detecting water stress effects on fruit quality in orchards with time-series PRI airborne imagery. Remote Sens. Environ. 2010,114, 286–298. [CrossRef] 55. Stagakis, S.; González-Dugo, V.; Cid, P.; Guillén-Climent, M.L.L.; Zarco-Tejada, P.J.J. Monitoring water stress and fruit quality in an orange orchard under regulated deficit irrigation using narrow-band structural and physiological remote sensing indices. ISPRS J. Photogramm. Remote Sens. 2012,71, 47–61. [CrossRef] 56. Suárez, L.; Zarco-Tejada, P.J.; Sepulcre-Cantó, G.; Pérez-Priego, O.; Miller, J.R.; Jiménez-Muñoz, J.C.; Sobrino, J.A. Assessing canopy PRI for water stress detection with diurnal airborne imagery. Remote Sens. Environ. 2008,112, 560–575. [CrossRef] 57. Pérez Peña, J.E.; Tarara, J. A portable whole canopy gas exchange system for several mature field-grown grapevines. Vitis 2004 , 43, 7–14. 58. Sims, D.A.; Gamon, J.A.J. Estimation of vegetation water content and photosynthetic tissue area from spectral reflectance: A comparison of indices based on liquid water and chlorophyll absorption features. Remote Sens. Environ. 2003 ,84, 526–537. [CrossRef] 59. Yu, R.; Brillante, L.; Martínez-Lüscher, J.; Kurtural, S.K. Spatial Variability of Soil and Plant Water Status and Their Cascading Effects on Grapevine Physiology Are Linked to Berry and Wine Chemistry. Front. Plant Sci. 2020,11, 790. [CrossRef] [PubMed] 60. Ripoll, J.; Urban, L.; Staudt, M.; Lopez-Lauri, F.; Luc, P.R. Bidel, L.P.R.; Bertin, N. Water shortage and quality of fleshy fruits— making the most of the unavoidable. J. Exp. Bot. 2014,65, 4097–4117. [CrossRef]