The trial was conducted at the Monforte Mosconi vineyard (Enrico Serafino SRL - Poderi Antonio Gentile SSA Monforte d Alba (CN)). The vineyard has a complete southern exposure and is located at 450 m.s.l. The monitoringperiodlasted 5 months (from 15/05/2016 to 15/10/2016). Three types of colored plastic net (Caliber: 0.32 mm; Mesh: 3 x 7 mm; Weight: 52 g m^-2, Height: 100 cm) were used for the experiment: Carbon black ; Gray and Black Green .

The reduction in solar light transmittance is 14 % for the Gray, 15% for the Green and 17 % for the Black nets.

A total of 75 Nebbiolo vine plants were involved. The nets were placed before May 30^th. The nets were placed at BBCH 57 phenological stage, when inflorescences were fully developed and flowers separating, and they were removed after the harvest. Five treatments were studied, and triple-randomized experimental blocks were set up, each with five adjacent plants. The three blocks were arranged at different elevations in the vineyard with an overall difference in height of about 18 m. Of the five experimental treatments, two were managed without nets and three with net coverage; one was managed by manually removing all the leaves and young shoots around the bunches just after flowering (A), and one was managed by keeping all the original leaves around the bunch but cutting the lateral shoots (B) (this is the traditional method used in the area). The other three treatments involved covering the vines with nets (C, D, E) and treating them as treatment A, that is, removing the leaves and shoots just after flowering (BBCH 71).

The C treatment involved covering the vines with a black net, while the D and E plots were covered with green and gray nets, respectively. How the experimental blocks were prepared is shown in Figure 1.

The phenological stages, the damage created by the sun or by hail and other parameters listed below were checked for each experimental plot.

Five plants were collected, in a randomized way, from each experimental block and 350 berries were cut with their pedicels. Of these 350 berries: 50 (10 berries/plant) were used to check the berry weight; 50 (10 berries/plant) were used to check the technological ripening parameters; 250 (50 berries/plant) were used to check the phenolic ripeness and phenol content. The main phenological stages were determined on the basis of: the number of canes/plant; the number of clusters/plant; the weight of the berries the % of sun burned berries; the % of berries damaged by hail; the green pruning operation timing (min/plant).

The technological ripening parameters of the juice were determined at harvesting:Sugars g/L (as the sum of glucose and fructose); titratable acidity as H₂T g/L; pH; Free Nitrogen Available for yeast (FAN); Malic Acid g/L. The latter two parameters were measured, together with the sugars, with a FOSS infrared instrument. The berry samples obtained with the pedicels, were manually pressed in a plastic bag and before the analysis the juice was filtered up to 5 NTU turbidity.

The flavonoids were analyzed at the CREA-ENO laboratory (Asti, Italy) using some of the most widely used spectrophotometric methods 14under optimized conditions for red wine analysis 15. The polar compounds were loaded and washed at a low pH with diluted sulfuric acid, to improve the recovery of such acidic phenols as gallic acid. The total phenol content (TP) was assessed by considering the reduction of the phosphor tungstic-phosphomolybdi acids(Folin-Ciocalteu s reagent) to blue pigments by phenols in an alkaline solution. Concentrations were determined by means of a calibration curve as (+)-catechin, mg/kg of grape. The proanthocyanidine index (PC) was evaluated by considering its transformation into cyanidin by means of the method of Di Stefano et al. 16, which uses iron salts as a catalyst, to increase the reproducibility of the cyanidin yield and replaces n-butanol with the optimal percentage of ethanol. Under such conditions, the average yield was estimated to be 20%, and the PC concentration (mg/kg of grape) was conventionally expressed as five times the amount of formed cyanidin, by means of a calibration curve with cyanidin chloride (ε = 34700). The catechins and proanthocyanidines reactive to the vanillin parameter (FRV) - were analyzed using the optimized and controlled vanillin-HCl method of Broadhurst and Jones 17, according to the conditions described by Di Stefano et al. 14 Concentrations were calculated as (+)-catechin (mg/kg of grape) by means of a calibration curve. An aliquot of 5 mL of red grape extract, diluted (5-20 times, to obtain a final reading in the 0.3-0.6 AU range) with 0.5M H₂SO₄ was loaded onto a conditioned Sep-Pak to establish the total anthocyanins index (TA). The column was washed with 2mL of 5mM H₂SO₄, and the red pigments were eluted with 3mL of MeOH into a 20mL calibrated flask. The extractable Anthocyanis (EA) were quantified directly based on their maximal absorbance in the visible range (536-542 nm), according to the methods described by Di Stefano et al. 14. The flavonoid index (F) was estimated spectrophotometrically based on their absorbance at 280 nm, according to the literature 14, and the data were presented as equivalents of (+)- catechin, mg/L.

Univariate analyses were conducted with the SAS v.9.0 software. Significant differences between the five treatments were assessed by means of PROC GLM with the LSMeans / PDIFF command, adjusted according to Bonferroni. Multivariate analyses were performed on the whole dataset of 32 examined variables, with XLSTAT software 18 The aim of such analyses is that of associating some more important compounds and characteristics with the respective elevation order in the vineyard on the hill, coded as 1=Low (bottom row), 2=Medium, 3=High (top row) over a total range of 18 m of elevation.

The reflectance spectra of the hand trimmed fresh skin and seeds aggregated in triple masses of 30 grains were acquired using an NIR SCiO^TM v. 1.2 device, 19 (Figure 2) operating over the 740-1070 nm range, protected by a magnetic spacer capsule. Three independent spectra were scanned from each analytical sample. A Random Forest algorithm inherent to the Lab-SCiO^TM software was used for the classification of the skin and seed of the five treatments; a probability vs. threshold of 20% was tested using the online version of Med-Calc for percentage comparisons. The NIR spectra were downloaded from the SCiO^TM repository, and then imported in WinISI II v1.04 (FOSS NIRSystem/Tecator, Infrasoft International, LLC) software compatible format. A calibration-cross-validation modified partial least squares method of the first derived reflectance spectra was fitted for the group average of each laboratory variable. The outliers were discerned after one passage at t>2.0. No math pretreatment was adopted, so that the Partial Least Square (PLS) models with 331 coefficients plus a constant could be used in new vine skin and seed datasets from spectra recorded by a SCiO^TM instrument and downloaded from a repository. To understand Whether any of the spectral positions in the five treatments overlapped, the average reflectance spectra were submitted to a principal coordinate analysis (PCoA) and subsequent agglomerative hierarchical clustering (CHA).

The timings necessary to perform the “Green” pruning operations in the treatment, (complete leaf and lateral shoots removal around the cluster) with and without a hail net application, are reported in Table 1.

As can be seen from the table, the average timing necessary to cut all the lateral shoots (B) on each plant is on average 33.7“(51 h/ha), where the timing necessary to cut the leaves around the bunches (A) is 48.6“(74 h/ha), which means a delta of 44 % compared to the previous case (B). The timing necessary to cut the leaves and shoots below the nets (C, D and E) is 60.1” (92 h/ha), which means an increase of 24 % compared to the same treatment managed without the nets (A).

The main phenological stages were evaluated from flowering to harvest. No differences were observed between the treatments up to the start of veraison. Only at 10 d post veraison was the ripening process in the “black net” treatment clearly in advance, with the berry coloring being completed a few days before the others.

A slight difference was observed, regarding sun burn, between the net treatments (C, D and E) and the reference treatments (A and B), although the damage was acceptable (< 5 %). No differences were observed between the net treated treatments (C, D and E).

Only in one case did a weak hailstorm affect the vineyard during the summer. A large amount of mechanical damage was observed in the “non-treated” treatments (A, B), but with damage always below 5%.

The cluster/cane rate of each plant was determined to verify whether the five treatments had the same production potential. The average cluster/cane results are reported in Table 2.

As can be seen in the table above, no significant differences emerge for cluster/cane rate.

It is therefore possible to exclude that the results were affected by the different production potential of the treatments.

The results pertaining to the average berry weight of the five treatments under study are reported in Table 3.

As can be seen in the table, only the black net (C) shows a significant difference (p<0,05) in the berry weight compared to the other treatments, with an average weight of 1.48 g/berry, while the others show an average of 1.70 g/berry (-13%). As far as the number of seeds in a berry is concerned, one result points out a high number under the “colored” nets (D, E), where an average increase of 102% (vs. A, B) was measured, while C decreases the seed number by 45%.

The only juice parameters that evidenced a significant difference between the treatments at ripening were the pH and the malic acid content (Table 4).

Considering the pH of the juice, the D (green net treatment) and E (gray net treatment) treatments show a higher average level (3,21) than the other treatments (+13%), with the lowest content found in the A (leaves and shoots removal) and C samples. This is partially consistent with the malic acid content, which shows a similar trend for the pH, for the D treatment, and shows the highest content (1.37 g/L= +28%) and for the A treatment, which shows the lowest content (0.93 g/L).

The berry skin did not show any significant differences in the polyphenol profiles of the raw or of the extracted matrices (Table 5).

Moreover, the anthocyanin profile of the skin extracts of the experimental berry samples was uniform (Table 6).

However, the net treatments showed significantly different effects for the extractable phenolic composition of the seeds (Table 7).

As can be seen above, the total extractable phenols (TP) are minimum (369 mg/kg) in treatment E (gray), while the level reaches a maximum in group B (+53% vs. E). On the other hand, the extractable proanthocyanidines (PC) of the seeds in the treatments with the green and gray nets are 33% greater than the others. The vanillin reactive flavans (FRV) appear uniform between treatments, but when related to the PC, the FRV/PC ratio results to be significantly reduced in treatment D (Black) and 30 % elevated in treatment B.

The effect of the elevation in three strata within the vineyard was globally considered through an overall set of 29 variables (Table 8). Increases in the flavans and extractable polyphenols can be noted for the higher elevation, especially those of the skins, and then in the polyphenols extractable from the seeds, probably in response to the greater temperature range at the top of the vineyard. On the other hand, in the valley part of the vineyard, the skin has been modified by an accumulation of flavonoids (the most negative PLS coefficient), total polyphenols and extractable proanthocyanidines. Figure 3

The average spectra of the skin and seeds are reported in Figure 4 and Figure 5. The reflectance of the green nets appear the lower in the skin and the higher in the seeds.

As far as the NIRS performances are concerned (Table 9), the result was elevated both for the skin, with an average RSQ of 0.74±0.11 and average RPD values of 2.1±0.46 as well as for the seeds, whose average values reach an RSQ of 0.81±0.07 and an RPD of 2.4±0.46. The highest RPD (3.0) was for the skin total flavonids (SkT_F). In the figure 6, figure 7, figure 8 are reported the scatterplots for: the skin extractable anthocyans percentage; the skin extractable total polyphenols and the seed extractable flavans reactive to vanillin to proanthocyanidines ratio.

Based on the NIR spectra the dissimilarities between the five these appear quite different in the two studied matrices. In the skin (Figure 9) one cluster associate the treatments B and C-black opposed to A, D-green and E-gray.

In the seeds (Figure 10) the treatment A looks quite original, while B is paired with C-black and are opposed to D-green clustered with E-gray

A slight difference was observed, regarding sun burn, between the net treatments (C, D and E) and the reference treatments (A and B), although the damage was considered acceptable (< 5 %). No differences were observed between the net treated treatments (C, D and E).

This result confirms those of other studies in literature, performed with the same variety of grapes (Nebbiolo) or other varieties, when berries are exposed to the sun very early in the season, just after flowering (pea size), which stimulates the berry skin to produce several substances (mainly Carotenoids) that are useful to protect the berry itself from the sunlight, even against strong UV exposure 20212223.

The berry weight is an important parameter for the modern viticulture, because the production of red wines is mainly related to the phenols in the skin and the aroma precursor extraction potential 24.

When the berry weight is reduced, there is a corresponding reduction in the berry volume. As a result, there is an increase in the skin/juice ratio (more skins per juice unit), which is positive from a quality perspective. The black net test (C) showed a reduction in the berry weight of about 13%. The reduction may be related to the reduced amount of light that the black net had on the new berries during the first phase of the growing period (from fruit setting to veraison) in which the final size of the berries is related to the photosynthetic performances of the berries. In fact, at this moment, the berry behaves like a leaf that is full of chlorophyll, and the increase in berry volume may decrease if a part of the total available sunlight is absorbed by C-black plastic. The other colored nets (green and gray) probably did not have a significant depressive effect on the photosynthetic activity of the berries. However, different colored nets, when applied early in the season, may affect the final number of seeds, with a significant rise (102%), while in our experiment the black nets contrasted flower fecundation by 45%. We can explain the reduction of the seed number and berry weight in the black net from a local warming up in flowering time according to Kliever 25 and to Leolini et al 26, nevertheless the cause of the seeds rises in the green and gray nets of the seed number, but not the berry weight, is an original result. The green colored net consistently increased the berry pH and keeps up the malic acid content. The malic acid content of berries is positively correlated with the pH of the juice and is generally preserved when the berries are not subjected to high temperatures during the ripening process. The green net (D) seems more able to prevent high temperatures in the berries, while the black net (C), consistently with the material type that absorb the higher solar energy amount, affecting the cluster microclimate and the experimental conditions (A), where all the solar UV rays are not filtered, promoted the berry temperature increase being in that case related to an increased malic acid catabolism, generally associated with a faster ripening process 2728.

The anthocyanin profile of the skin extracts was uniform between the treatments. This is an important parameter to consider because the anthocyanin profile of Nebbiolo, which is directly responsible for the wine color 2930, does not ensure the same technological stability or intensity as other famous grape varieties, and already needs to be protected in the field. However, the colored nets seem have a neutral effect on the synthesis and extractability of anthocyanins from the skin.Anthocyanins are the main compounds responsible for the color of red wines, and their interactions with other phenolic compounds, called co-pigments, allow the color stabilization of aged wines to be improved through co-pigmentation or stabilization reactions 3132.

As far as the anthocyanin profile is concerned, it is important to try to synchronize the technological ripening (sugar accumulation and fixed acid reduction) with the phenolic ripening (high skin color extraction and low tannin release from the seeds) as much as possible under warm environmental conditions 3334. In such a context, coverage with nets can help to reach this goal.

A limited reduction only occurred in the polyphenols extractable from the seeds in treatment E (gray), while the colored net favored a rise in the extractable proanthocyanidines (PC).

The fact that the net color can affect the seed physiology even by increasing their mass but not the skin phenolic composition, is a remarkably interesting result, which was certainly not expected. An increased amount of extractable polyphenol compounds from the seeds means the seed ripening is less advanced, and thus the seeds can release more tannins. The result of the seed extractable FRV/PC ratio, which showed a higher tannin polymerization ratio in the D (0.68) and E (0.73) treatments, is also interesting. High values of the FRV/PC ratio, close to 1, mean the tannins are on average less polymerized, and thus more reactive with proteins and probably more astringent. In short, the D (green net) and E (gray net) treatments showed more extractable tannins from the seeds, but which were more polymerized, than the other treatments.

Despite the limited range of variation in the dataset, where the CV ranged from 3 to 17%, the vibrational spectroscopy results were in line with the expectations. In a framework of highly variable (average CV 58%) woody vinery materials, Baca-Bocanegra et al 35, using a portable Micro-NIRS (VIAVI), observed an average RPD of 3.9±0.8 for six polyphenol compounds.

Two aspects need to be considered to explain the dissimilarities of the treatments in the NIR spectra. First, the NIR spectrum of the skin only partially represents the polyphenols, which, moreover, have varied very little thanks to standardized management conditions. Second, in the spectrum of the skins are contained the vibrations of many other organic substances present in the plant cell-walls 36 which could explain the spectral diversity observed between the treatments B and C-black compared to the other three groups. In the seeds, the chemical differentiation on the polyphenols caused by the colored nets has assumed a greater dimension, significantly expressed also by the laboratory analyzes and the NIR spectra have been modified as evidenced by the high predictability of the compounds analyzed. This does not mean that other undetermined substances may have modified consequently the dendrograms representative of the spectral associations between the groups, charged to the treatment A, then in suborder to B and C-black, and then to the D-green and E-gray nets. Notice that B paired to C-lack and D-green paired to E-gray even in the skins looked very similar.