Nutrient Management Effects on Wine Grape Tissue Nutrient Content

With limited research supporting local nutrient management decisions in North Carolina grape (Vitis vinifera) production, field studies (2015–17) were conducted to evaluate late season foliar nitrogen (N) application on leaf and petiole N concentration and yeast assimilable N (YAN) in the fruit. Foliar urea (1% v/v) was applied at different rates and application times beginning pre-and post-veraison. Compared to soil applied N, late season foliar N substantially enhanced petiole N and grape YAN. Smaller split N applications were generally more effective in increasing YAN than single larger N rates. These data demonstrate the value of assessing plant N content at full bloom with petiole N analysis or remote sensing to guide foliar N management decisions. Additional field studies (2008–11) were conducted to evaluate pre-bud soil applied phosphorus (P) and potassium (K) effects on petiole P and K nutrient status. Fertilizer P and K were initially broadcast applied (0–896 kg P2O5 ha−1; 0–672 kg K2O ha−1) prior to bud-break in 2008–09 and petiole P and K at full bloom soil test P and K were monitored for three to four years after application. Soil test and petiole P and K were significantly increased with increasing P and K rates, which subsequently declined to near unfertilized levels over the sampling time depending on site and P and K rate applied. These data demonstrate the value of annually monitoring petiole P and K levels to accurately assess plant P and K status to better inform nutrient management decisions.


Introduction
Wine grape production is a rapidly growing industry in North Carolina (NC), which ranks 10th in the United States (US) in wine grape production, 7th in wine production, and represents a $2 billion total economic impact [1,2]. The dominant wine grapes produced in the western Piedmont and mountain regions of NC are V. vinifera, which includes the European varieties and both French and American hybrids [3]. As the NC wine industry expanded, the demand increased for research-based information to support management decisions. However, soil fertility and plant nutrition research supporting soil and plant diagnostic criteria were limited. Despite these limitations and driven by increasing demand for information, guidelines have been established based on research from other regions [4,5], which may not be appropriate for NC conditions.
Although soil testing traditionally is used to assess the potential for nutrient stress in annual crops, with perennial fruit crops, tissue analysis is considered a more reliable diagnostic tool to assess nutritional status [6]. Soil tests are essential to determine soil nutrient status prior to vineyard establishment. However, plant analysis is critical to monitoring plant nutrient supply each growing season. Both forms of analysis should be used to maintain optimum nutrient availability [7]. Most grape production regions recommend use of petiole sampling at full bloom, although if nutrient stress symptoms appear later in the growing season, veraison sampling guidelines have also been established [8,9]. For In one of the early studies of late-season foliar N on wine quality, Lacroux et al. [32] demonstrated soil applied N increased vigor and Botrytis incidence, whereas, foliar N improved vine N status and enhanced aroma characteristics of Sauvignon blanc without increasing vigor or Botrytis susceptibility. Other recent studies confirm the positive effects of foliar N on increased YAN and wine aromatics [33][34][35]. In particular, foliar N was more effective in increasing juice YAN compared with early season soil applied N [36] These results confirm significant improvements in the aromatic profile and intensity of wines made from Tempranillo grapes treated with foliar N. Lower aromatic intensity and pronounced herbaceous flavors were observed in wines not treated with foliar N. In addition, foliar N increased grape amino acid concentrations, which improved must N composition and enhanced fermentation kinetics. The above studies demonstrate that the use of 1-2% (v/v) urea foliar applied at veraison shows considerable promise compared to traditional soil applied N.
In soil and plant nutrient surveys of numerous NC vineyards Havlin et al. [17] reported over nearly 70% of full-bloom petiole samples were below critical N levels, while approximately 20% and 30% were below critical P and K levels, respectively. The low N status of V. vinifera grown in NC is related to the minimal or no N applied by growers, a common practice used to reduce vine vigor and associated disease pressure. It is common to observe N deficiency symptoms on V. vinifera grapevines, especially from pre-veraison through harvest growth stages.
While only 30% or less of NC vineyards are P and/or K deficient, P and K fertilizer management decisions should be based on local research supporting interpretation of soil and plant analyses data [7]. When soil test and/or plant analysis levels are above established critical levels, no additional P or K fertilizers are needed. Unfortunately, few data are available to assess the relative plant tissue P/K response to P/K applications. With P, Janat et al. [37] reported two-year average petiole P increased from 0.21% to 0.30% with 100 kg P 2 O 5 ha −1 applied annually. Using two V. vinifera varieties grown on P deficient soils, initial soil application of 300 kg P 2 O 5 ha −1 increased full bloom petiole P from 0.8% to 1.2% in the first year, increasing to 1.8% in the second year, and decreasing to 1.5% in the third year after P application [38]. In contrast, higher K rates are generally used compared to P, although excessive K applications can reduce wine quality by increasing must pH or decreasing titratable acidity [39,40]. For example, concern for excessive K effects on wine quality in Virginia resulted in a reduction in soil test critical K levels and subsequent fertilizer K recommendations [41]. Using a French hybrid (cv. Foch) grown medium soil test K, sandy soil, full bloom petiole K increased from 2.8% to 3.2% K (average over five years) with annual applications of 600 kg K 2 O ha −1 [42]. In a survey of 60 vineyard growers in central India, Naraboli et al. [43] reported 2.50-2.75% petiole K at full bloom was associated with optimum yield and was achieved with an average 1000 kg K 2 O ha −1 . After three years of annual applications of up to 200 kg K 2 O ha −1 to cv. Cabernet sauvignon in Brazil, Mehlich 1 K increased nearly 10-fold (50-464 mg kg −1 and full bloom petiole K from 2.0% to 3.6% [44]. Due to limited research resources available to establish critical N, P, and K plant tissue levels under NC conditions, critical nutrient levels (NC) were adopted from those established in other regions [4,5]. Despite this limitation, research information can be provided to assess relative plant nutrient response to N, P, and K application. Therefore, using established plant nutrient critical levels, vineyard managers can decide if additional nutrients are needed and should expect applied nutrients to increase plant nutrient levels to or above the critical levels. Therefore, the objectives of these studies were to: (1) evaluate late season foliar N application on YAN in the fruit, and (2) quantify soil and plant nutrient response to soil applied P and K. Soil properties (check plots) were typical of vineyard soils (relatively high P, K, micronutrients) in the Yadkin Valley Appellation (Surry Co., NC) with a previous history of manure applications (old dairy farm; Table 1). Soil pH (0-20 cm) was optimum (Site A) for V. vinifera grape production; however, at Site B soil pH was below optimum. At Site A, soil samples were also collected from the soil applied N plots, although results were not significantly different from soil test data obtained from the untreated plots, thus residual soil applied N was not detected in at either site (data not shown). Monthly rainfall and temperature data were relatively normal, although 2017 rainfall was 30% higher than the 50-year average ( Table 2). Leaf and petiole N analyses at full-bloom (prior to foliar N application) were necessary to establish plant N levels that could be used to assess foliar N need. At all sites in each year petiole and leaf N contents at full bloom were below established critical levels of 1.2-1.6% petiole N and 2.5-3.5% leaf N [3,45], although in each year leaf N was at or slightly above 3.5% N in site A (Table 3). No significant differences in plant N or NDVI were detected between treatment areas since foliar N applications did not begin until pre-veraison. Despite 100 kg N ha −1 soil applied at bud break, plant N was not affected in 2015. In 2016 with 200 kg ha -1 soil applied N, leaf N at both sites and petiole N at site A were slightly but significantly increased, although petiole N (site B) was not affected (Table 3). In 2017, soil applied N at both sites slightly but significantly increased petiole N, but no difference was observed in leaf N (Table 3). Compared to 2015, where a lower N rate was soil applied, the plant N data illustrate that soil N applied at relatively high rates may only slightly increase plant N content at full bloom. In order to develop foliar N recommendations based on UAV-acquired imagery, NDVI measurements should relate to leaf or petiole N content. There was no clear correlation between petiole N and NDVI in both 2015 (site A, data not shown) and 2016 (site A and B), which was expected since reflectance occurs dominantly on the leaf surfaces. In contrast, leaf N and NDVI were correlated, indicating a potential use for multispectral remote sensing to asses plant N status ( Figure 1). However, due to the planophile leaf structure of most V. vinifera and the tendency for NDVI and other similar vegetative indexes to saturate as biomass increases, the ability to resolve differences at higher N concentrations and later in the season decreases [46]. This highlights the importance of the timing of flights and variability of leaf N. Comparatively, when only small differences in tissue N are present, UAV-based measurements are of less value and leaf or petiole sampling is required.
A contributing factor related to the difficulty in remotely measuring N content in 2017 was the limited grape leaf area within the canopy and an early morning flight time. After classifying the pixels into classes representing the grape canopy, soil, and surrounding grass, only 8% of the area was identified as grape canopy, the remaining 92% was bare soil or grass. Consequently, variations in leaf angle and leaf orientation had a greater effect on the resultant NDVI value compared to data collected in earlier years.

Foliar N Studies: Wine Grape Quality
In 2015, soil applied N (pre-bud break) had little effect on wine grape quality parameters compared to no N applied (Table 4). In contrast, foliar N significantly increased YAN and malic acid in grapes at harvest. A small increase in juice pH was also observed. These parameters generally increased with increasing N rate (0-44.8 × 1). However, the two treatments that resulted in the highest YAN levels were the 22.4 and 11.2 × 4 rates. The 44.8 and 22.4 × 2 treatments resulted in severe and moderate leaf edge burn, respectively, thus both were discontinued in 2016-17.

Foliar N Studies: Wine Grape Quality
In 2015, soil applied N (pre-bud break) had little effect on wine grape quality parameters compared to no N applied (Table 4). In contrast, foliar N significantly increased YAN and malic acid in grapes at harvest. A small increase in juice pH was also observed. These parameters generally increased with increasing N rate (0-44.8 × 1). However, the two treatments that resulted in the highest YAN levels were the 22.4 and 11.2 × 4 rates. The 44.8 and 22.4 × 2 treatments resulted in severe and moderate leaf edge burn, respectively, thus both were discontinued in 2016-17. In contrast to 2015, the higher soil applied N rate in 2016-17 increased malic acid and YAN compared to the 0 N treatment (2016-site A and B; 2017-site A), whereas soil N did not significantly increase YAN in 2017 (site B) ( Table 5). Jreij [47] reported maximum YAN with both soil and foliar applied N. No significant N treatment effects on Brix, pH, and TA were observed at either stie in 2016-17 (data not shown).  At both sites, increasing foliar N rate significantly increased YAN and malic acid compared to the "0" treatment, and were also significantly greater than with soil applied N ( Table 5). Compared to the 0 N treatment, the 11.2×4 split N treatment increased 2016 YAN 58% and 49% at site A and B, respectively, while increasing YAN at both sites by nearly two-fold over the soil applied N treatment, respectively, in 2017. Similar differences between foliar and soil applied N on YAN were recently reported for V. vinifera cv. petit manseng [48]. The split N treatments generally exhibited a larger response in YAN and malic acid than single N application rates. For example, in 2016 the 11.2 × 2 treatment increased YAN by 24% and 12% over the single 22.4 rate at site A and B, respectively. In contrast, foliar N with these same treatments increased YAN by 15% and 17% at site A and B, respectively, in 2017. In both years, similar trends in malic acid were observed with these same N treatments, although the effects were greater at site A than site B. Using 15N, Lasa et al. [34] demonstrated that split N applications on Merlot and Sauvignon Blanc, especially post-veraison, were more effective in increasing N in grapes berries compared to single applications of low foliar N rates and soil applied N. Foliar N treatment effects on malic acid were similar to those on YAN, although the increases were not as pronounced (Table 5). Similar studies in British Columbia using three split applications of 12 kg N ha −1 each over three years on seven vineyards and five varieties consistently increased (35-245%) YAN [23]. Bavaresco et al. [49] also documented split applications of foliar urea during the growing season increased YAN compared to a single N application. The results of our foliar N experiments concur with many other recent studies documenting foliar N was effective in increasing diverse flavonoid contents in grapes and wines from N deficient vineyards [50][51][52][53].

P and K Studies
The vineyards included in the P and K nutrient response studies were selected for potential wine grape response to soil applied P and K ( Table 6). The P sites were at or below the 28-32 mg kg −1 critical soil test P level, while the K sites were at or below the 120-130 mg kg −1 soil test K critical level [54]. Monthly growing season rainfall and temperature data were relatively normal, although at the Surry Co. site (1P09/1K09) rainfall  (Table 7). In each year, growing season rainfall and temperature were relatively normal at 2P09 (Moore Co. site), while at 2K08, growing season weather conditions were similar to 50-year averages except in 2010 and 2011 where rainfall was 18% and 34% higher, respectively.   At both P sites (1P09 and 2P09), soil test P increased substantially in 2010 following P application in 2009 ( Figure 2). With 896 kg P 2 O 5 ha −1 applied in 2009, soil test P increased approximately four-and five-fold above initial soil test P levels at the 2P09 and 1P09 sites, respectively. At 1P09, 448 and 896 kg P 2 O 5 ha −1 were needed to reach or exceed the critical soil test P level in 2010; whereas by 2011 soil test P exceeded the critical P level with only the highest P 2 O 5 rate. Since the initial soil test P at 2P09 was at the critical level, soil test P exceeded 28-32 mg P kg −1 with all P rates in 2010, and subsequently declined by nearly 50% by 2011 (Figure 2). Although the soil test P response to applied P 2 O 5 was curvilinear (Figure 2), approximately 17 kg and 9.5 kg P 2 O 5 ha −1 were required to increase soil test P 1 mg kg −1 at 1P09 and 2P09 sites, respectively. These data were similar to those reported for similar Ultisol soils in NC [55,56].
Although initial soil test P was below critical P level at 1P09, petiole P concentration was at the critical petiole P range with no P applied and significantly increased with increasing P rate ( Figure 2). In 2009 at 1P09, petiole P concentration increased two-, three-, and four-fold above the petiole P concentration with no P applied. In contrast at 2P09, petiole P concentration in 2009 was below the critical P range with the 0 and 224 kg P 2 O 5 ha −1 treatments, although initial soil test P was at the critical level. Using similar P rates, Janat et al. [37] and Conradie et al. [57] reported similar increases in petiole P with soil applied P. With 448 kg and 896 kg P 2 O 5 ha −1 applied, petiole P increased nearly 1.5-and 2-fold above the critical petiole P range in 2009, decreasing by nearly 50% in 2010, and by 2011 petiole P had declined to the critical P range. Thus, even at 896 kg P 2 O 5 ha −1 petiole P was just slightly above critical petiole P three years after P application.
At both K sites, soil test K significantly increased with increasing K rate (Figure 3). With the 672 kg K 2 O ha −1 treatment, soil test K increased approximately two-and threefold above the 0 K treatment at 2K08 and 1K09 sites, respectively. Similar to the P sites, soil test K response to applied K 2 O was curvilinear (Figure 3), where an average of 5.6 kg and 6.5 kg K 2 O ha −1 were required to increase soil test K 1 mg kg -1 at 1K09 and 2K08 sites, respectively. As with P, the K rate needed to increase soil test K by 1 mg kg −1 is highly variable depending type and quantity of soil clay minerals. A recent summary reported a range of 4-20 kg K 2 O ha −1 to increase soil test K by 1 mg kg −1 [58]. At 1K09, 336 kg K 2 O ha −1 were needed to reach or exceed the critical soil test K level; whereas only 168 kg K 2 O ha −1 were needed at the 2K08 site. Two years after K application at 1K09, soil test K had declined to or below the critical K level with all treatments except for the highest K rate which remained slightly above the critical K level. While soil test K declined following K application at the 2K08 site, soil test K remained above the critical K level two years after K application. However, in the third year, soil test K was at or below the critical level with all K treatments. petiole P concentration in 2009 was below the critical P range with the 0 and 224 kg P2O5 ha −1 treatments, although initial soil test P was at the critical level. Using similar P rates, Janat et al. [37] and Conradie et al. [57] reported similar increases in petiole P with soil applied P. With 448 kg and 896 kg P2O5 ha -1 applied, petiole P increased nearly 1.5-and 2fold above the critical petiole P range in 2009, decreasing by nearly 50% in 2010, and by 2011 petiole P had declined to the critical P range. Thus, even at 896 kg P2O5 ha -1 petiole P was just slightly above critical petiole P three years after P application.

Year
Regression  At both K sites, soil test K significantly increased with increasing K rate (Figure 3). With the 672 kg K2O ha -1 treatment, soil test K increased approximately two-and threefold above the 0 K treatment at 2K08 and 1K09 sites, respectively. Similar to the P sites, soil test K response to applied K2O was curvilinear (Figure 3), where an average of 5.6 kg and 6.5 kg K2O ha -1 were required to increase soil test K 1 mg kg -1 at 1K09 and 2K08 sites, respectively. As with P, the K rate needed to increase soil test K by 1 mg kg -1 is highly variable depending type and quantity of soil clay minerals. A recent summary reported a range of 4-20 kg K2O ha -1 to increase soil test K by 1 mg kg -1 [58]. At 1K09, 336 kg K2O ha -1 were needed to reach or exceed the critical soil test K level; whereas only 168 kg K2O ha -1 were needed at the 2K08 site. Two years after K application at 1K09, soil test K had declined to or below the critical K level with all treatments except for the highest K rate which remained slightly above the critical K level. While soil test K declined following K application at the 2K08 site, soil test K remained above the critical K level two years after K application. However, in the third year, soil test K was at or below the critical level with  1 were needed to reach or exceed the critical soil test K level; whereas only 168 kg K2O ha -1 were needed at the 2K08 site. Two years after K application at 1K09, soil test K had declined to or below the critical K level with all treatments except for the highest K rate which remained slightly above the critical K level. While soil test K declined following K application at the 2K08 site, soil test K remained above the critical K level two years after K application. However, in the third year, soil test K was at or below the critical level with all K treatments.   Increasing K rate significantly increased petiole K above the critical K range at both sites in 2009, although the response in petiole P was greater at 1K09 compared to 2K08. For example at the 672 kg K2O ha -1 rate, petiole K increased two-and three-fold above petiole K concentration with no K applied at 1K09 and 2K08, respectively (Figure 3). This difference may be due to higher initial soil test K level at 2K08. Similar responses were reported by Neilsen et al. [42] and Ciotta et al. [44]. In subsequent years, petiole K at 1K09 decreased by approximately 50% in 2010, while in 2011 petiole K had declined to the critical range with 168 kg K2O ha -1 treatment and remained slightly above the critical K range at the higher K rates. Similarly at site 2K08, petiole K increased to or above the critical K range in 2009, and although still above the critical range, petiole K declined by approximately 50% by 2010. By 2011, petiole K had declined to the critical level or slightly below with all treatments. Increasing K rate significantly increased petiole K above the critical K range at both sites in 2009, although the response in petiole P was greater at 1K09 compared to 2K08. For example at the 672 kg K 2 O ha −1 rate, petiole K increased two-and three-fold above petiole K concentration with no K applied at 1K09 and 2K08, respectively (Figure 3). This difference may be due to higher initial soil test K level at 2K08. Similar responses were reported by Neilsen et al. [42] and Ciotta et al. [44]. In subsequent years, petiole K at 1K09 decreased by approximately 50% in 2010, while in 2011 petiole K had declined to the critical range with 168 kg K 2 O ha −1 treatment and remained slightly above the critical K range at the higher K rates. Similarly at site 2K08, petiole K increased to or above the critical K range in 2009, and although still above the critical range, petiole K declined by approximately 50% by 2010. By 2011, petiole K had declined to the critical level or slightly below with all treatments.

Foliar N Studies
Based on previous nutrient survey studies in NC vineyards, a vineyard was selected that exhibited low petiole and leaf N concentrations at full bloom and veraison [17]. The experimental sites were located at Shelton Vineyards in Surry Co., NC. In 2015, the site was on a Fairview sandy clay loam (fine, kaolinitic, mesic Typic Kanhapludults), and in 2016-17, an additional site was added on a similar soil type (five site-years).  (Table 8). For the soil N treatment, 100 kg N ha −1 as urea was broadcast applied in early March, prior to budbreak. In subsequent years (2016-17) soil applied N was increased to 200 kg N ha −1 , while foliar N treatments deviated slightly due to salt damage to foliage observed in 2015 with the 44.8 kg N ha −1 rate applied once or in two split applications. In early February, soil samples were collected prior to bud-break (early-mid March) from each site at 0-10 and 10-20 cm depths. Four cores were randomly collected from each untreated plot and composited, from which a subsample was air dried and sent to the NC Department of Agriculture and Consumer Services Laboratory for analysis [54].
Prior to application of the foliar N treatments, plant tissue (petiole and leaf) samples (40)(41)(42)(43)(44)(45)(46)(47)(48)(49)(50) were collected at full bloom (pre-treatment) from opposite the first or second cluster from the bottom of the shoot in each treatment area or plot. Petioles were immediately separated from lamina and both placed in separate labeled paper bags or envelopes. Tissue samples were dried in a forced air oven at 27 • C to 32 • C for 24 h, and submitted to the NC Department of Agriculture and Consumer Services Laboratory for total nutrient analysis (54).
The potential for remotely measuring plant N status using a Normalized Difference Vegetative Index (NDVI) was also evaluated. An Unmanned Aerial Vehicle (UAV) was used to capture multispectral imagery between 2015 and 2017 [59]. Flights were conducted using a DJI Phantom 3 drone (Da-Jiang Innovations, Shenzhen, China) and Parrot Sequoia (Parrot, Paris, France) four-band multispectral sensor. The Sequoia sensor measures radiance in the green: 530-570 nm, red: 640-680 nm, red-edge: 730-740 nm, and near-infrared: 770-810 nm wavelengths.
Images were collected using pre-programmed flights 50 m above ground level with 70% side-overlap and 80% forward overlap. Ground control points were placed in the vineyard before each flight and used during post-processing to enhance the geometric accuracy of the orthomosaics. Orthomosaics were developed using photogrammetric software (Agisoft Metashape, St. Petersburg, Russia).) using default processing parameters and the recommended processing workflow. Flights were conducted on 20 May 2015; 24 May 2016; and 11 May 2017, on the same day as plant tissues were collected for analysis.
The georeferenced orthomosaics were used to calculate the fractional green-leaf area and an average NDVI for each plot. Plot boundaries were developed in a Geographic Information System (ESRI-ArcGIS 10.8, Redlands, CA, USA) [60] using the measured dimensions of each plot and the orthomosaic. A supervised maximum likelihood classifier [61] was used to segment and classify pixels into grass, bare soil, shadows, and grape-leaves. Pixels classified as grape-leaves were summed and used to calculate the fractional green-leaf area for each plot and used as a proxy to estimate canopy 'vigor' and biomass during analysis. NDVI was calculated using pixels classified as green-leaf area to assure NDVI values were representative of the vinifera leaf-canopy alone. NDVI is based on the measurement of red and near-infrared (IR) wavelengths described by: These two wavelengths are known to correlate to both plant biomass and N content and are often related to plant 'vigor' [62][63][64]. Descriptive statistics for NDVI were calculated using the raster analysis and zonal statistics available in ArcGIS. NDVI values from 2016 and 2017 represent uncalibrated radiometric surface reflectance and consequently analyzed independently by year.
Foliar N treatments (Table 8) were applied with a backpack CO 2 sprayer (R and D Sprayers, Inc., Opelousas, LA, USA) equipped with 4-80 • C flat spray nozzles on 0.30 m spacing. The 1.2 m spray boom was held vertically along each side of the treatment row to facilitate optimum canopy coverage. Urea solution (20% N) was diluted with variable amounts of distilled H 2 O to prepare 1650 mL of final N solution applied to each treatment.
At harvest,~8-10 clusters were collected from each plot and immediately placed in a cooler under dry ice. Samples were delivered immediately after harvest to the Appalachian State Chemistry and Fermentation Service Lab, where grape juice was analyzed for pH, total acidity (TA), Brix, malic acid, and YAN [65]. Juice YAN was determined by summing primary amino acid-N obtained by HPLC analysis and ammonia-N by an enzymatic assay [66]. Plant N and fruit quality data were analyzed using General Linear Models routine in SAS 9.4 (SAS Institute, Cary, NC) [67]. Treatment effect means for measured parameters were compared using Fisher's protected LSD (p < 0.05).

Soil P and K Studies
Field studies were initiated to evaluate the response in wine grape petiole P and K concentrations to soil applied P and K. Based on soil and plant nutrient survey data collected in 2006-08 [17], four vineyard sites were identified with potential vine response to applied P and K (Table 9). In 2008, one site was selected to represent the potential for low soil K at Hanover Park Vineyard (Yadkinville, Yadkin Co., NC, USA) and in 2009 three additional vineyards were identified (two P sites (Round Peak Vineyard, Dobson, Surry Co. NC, USA; Black Rock Vineyard, Sanford, Moore Co. NC, USA) and 1 K site (Round Peak Vineyard, Dobson, Surry Co., NC, USA). Four additional sites were used in the four-year P and K study. However, these data are not included since the initial soil test P and K levels exceeded established critical levels [44]. Monthly growing season (March-October) rainfall and temperature data were obtained from nearest weather stations in Mt. Airy, NC (1P09, 1K09), Sanford, NC (2P09), and Yadkinville, NC (2K08) [US Climate Data 2021, version 3.0]. At each site-year, 20 m rows were used per treatment with treatments arranged in a randomized complete block design with four replications. Plots were one vine row in a 3 m × 20 m plot, where ground cover between rows was tall fescue (Festuca arundinacea) with 1 m bare surface maintained under the vine canopy. In February of the 2008 and 2009 studies, soil samples (0-20 cm depth) were collected in each plot prior to treatment application, and again in the amended region each subsequent year after application to evaluate influence of treatments on soil test P and K. Soil samples were analyzed using standard soil analysis methods [54]. At the P sites, triple super phosphate (0-46-0) was applied at 0, 224, 448, and 896 kg P 2 O 5 ha −1 . At the K sites, potash (0-0-60) was applied at 0, 168, 336, and 672 kg K 2 O ha −1 . After soil samples were collected, P and K treatments were broadcast applied about 40 cm on either side of the row without incorporation.
Plant tissue (petiole and leaf) samples opposite the first or second flower cluster from the bottom of the shoot were collected at full bloom (20-28 May) (~40 petiole/leaf subsamples). Petioles were immediately separated from leaf blades and both placed in separate labeled paper bags or envelopes. Tissue samples were dried in a forced air oven at 80 to 90 • F for 24 h and submitted to the NC Department of Agriculture and Consumer Services Laboratory for total nutrient analysis [54]. Since NC vineyards utilize petiole analysis to assess vine nutrient status only the petiole P and K results are shown.
Analysis of variance for soil test and plant tissue data was performed using the General Linear Model procedure in SAS Version 9.1 (SAS, Cary, NC, USA) [67]. Mean comparisons were performed with Least Significance Difference (LSD) at a probability level of 0.05.

Conclusions
The purpose of these studies was to quantify the effect of foliar applied N on selected N parameters in wine grapes related to wine quality, and to evaluate petiole P and K response to soil applied fertilizer P and K.
Petiole N content was below the critical level (1.2-1.6%) at each site used in the foliar N studies, which suggest that foliar N applied pre-and post-veraison could significantly improve grape N content and other parameters critical to enhancing flavor compound concentrations in wine grapes. Increasing foliar N rates generally increased YAN, while split N applications generally increased wine grape quality parameters to a greater extent than single foliar applied N rates. Pre-bud break soil applied N had little or no effect on wine grape quality unless applied at elevated rates. These preliminary data also demonstrate the potential use of UAVbased remote sensing in assessing N status in the vineyard. Although UAV's provide a unique opportunity to capture images at resolutions needed to detect plant N status, sensitivity of the measurements can be affected early in the season when there is little growth. As this analysis suggests, if this technology is to detect early-season plant N in V. vinifera, timing of flights should correspond with adequate canopy development to providing robust, reliable measurements. Therefore, identifying N deficient grape plants at full bloom by either plant sampling/analysis or through remote sensing can direct the vineyard manager to initiate late-season foliar N management to improve wine grape quality.
The P and K response studies were located on soils testing at or below critical soil test P or K levels. At each site, increasing P or K rate increased soil test and petiole P and K levels. In the two-or three-years after application, petiole P and K declined to at or below critical levels; however, at the highest P or K rates petiole P or K levels remained at or above the critical nutrient range. Since plants accumulate nutrients throughout the rooting depth, it is difficult to base P or K management decisions on soil test P or K. Therefore, these data illustrate the importance of annual petiole analysis to determine when petiole P or K levels decline to or below established critical levels warranting application of P or K. Although additional P and K response studies representing a wider range in clay content and other soil properties are needed to establish specific P and K rates required to increase petiole P or K above established critical levels. Based on these limited data, growers should regularly monitor petiole P/K, and if below the critical levels soil applied 200-300 kg P 2 O 5 ha −1 and 300-400 kg K 2 O ha −1 . Subsequent P or K rates should be adjusted depending in petiole P/K the year of or following application.