AGRONOMIC PERFORMANCE OF ‘SAUVIGNON BLANC’ VARIETY TRAINED IN Y-TRELLIS AND VERTICAL SHOOT POSITION TRELLIS IN A HIGH-ALTITUDE REGION OF SOUTHERN BRAZIL

The objective of the present work was to compare the effect of the training systems Y-trellis (Y) and Vertical Shoot Position trellis (VSP) on the production efficiency and chemical characteristics of grapes and wine from 'Sauvignon Blanc' cultivated in high altitude regions of Santa Catarina State, Brazil. The study was carried out in 2013, 2014 and 2015 harvests in a commercial vineyard located in the municipality of São Joaquim - SC (28º 17 ’38 ”S and 49º 55’54” O, altitude 1350 m). The obtained data were subjected to analysis of variance (ANOVA) with F Test (p ≤ 0.05). The relationship between the vegetative growth and the yield was influenced by the training system. For the ‘Sauvignon Blanc’ cultivar, the yield was 93% higher in the Y-trellis training system, resulting in better vegetative-productive balance. The technological and phenolic maturation of the grapes were similar in both training systems. The results of this study show that the Y-trellis training system is an alternative to use in Santa Catarina altitude regions because it provides an increased productivity on the ‘Sauvignon Blanc’ variety without compromising the composition of grapes, when comparing to the VSP training system.


INTRODUCTION
The high altitude regions of Santa Catarina State, Brazil, are characterized by having vineyards between 900 and 1400 meters above sea level, longer phenological cycles, greater solar availability and greater thermal amplitude in comparison to other Brazilian wine-growing regions (Brighenti et al., 2013). Due to these characteristics, the cultivation of the Vitis vinifera L. varieties is favored, reaching maturity indexes that allow elaboration of wines differentiated by their intense color, aroma and acidity (Marcon Filho et al., 2015).
The production of wine grapes in high altitude regions is based on vertical shoot position trellis (VSP), and the vines are pruned in spur cordon (Falcão et al., 2008). However, the use of VSP generally results in excessive vegetative growth in the vineyards due to the high water availability (Bem et al., 2016), combined with high concentrations of organic matter in the soil (Zalamena et al., 2013) and choice of vigorous rootstocks. According to Marcon Filho et al., (2015), in grapevines, as in most fruitful species, the balance between the fruit load (drain) and the leaf area (source) induce the quantity and quality of production. The balance between these two parameters is determinant for the composition and maturation of grape berries.
The grapevine has a great diversity of architecture in its canopy and perennial parts, constituting different training systems (Wurz et al., 2019). To overcome the problem of excessive vigor, changing the shape of the canopy can be an alternative to achieve a balance between vegetative growth and production. In addition, it can be done by the canopy division, which simultaneously increases the production and can improve grape composition (Reynolds and Vanden Heuvel, 2009). According to Clímaco et al., (2003), excessive vigor induces less productivity. One alternative of canopy division is the adoption of the Y-trellis, which has been used in many regions of Brazil (Falcão et al., 2008;Hernandes et al., 2013). This training system, despite having higher initial implantation cost than the VSP, has the advantage of increasing production without loss of grape quality, besides facilitating the installation of plastic cover for protected cultivation (Pedro Junior et al., 2015).
Among the varieties grown in areas above 900 m in Santa Catarina State, there is 'Sauvignon Blanc', the white variety that is well adapted to altitude conditions. In addition to better adaptation to edaphoclimatic conditions of high altitude regions of Santa Catarina, 'Sauvignon Blanc' wines presents high quality with specific features such as high aromatic complexity and quality (Brighenti et al., 2013). The objective of the present work was to compare the effect of the training systems Y-trellis and VSP on the yield efficiency and chemical characteristics of 'Sauvignon Blanc' cultivated in high altitude regions of Santa Catarina State, Brazil.

MATERIAL AND METHODS
The study was conducted during the 2013, 2014 and 2015 vintages in a commercial vineyard located in the municipality of São Joaquim -Santa Catarina (SC) (28º13'86"S and 49º81'14"W, altitude 1350 m above sea level). The climate of the region is humid mesothermic (Cfb) according to the Koppen classification (Peel et al., 2007), and the soil type is Cambisol, which is characterized by high levels of clay (485 g/kg) and organic matter (69 g/kg). A large amount of precipitation occurs in this region from October to March, during which the average rainfall is approximately 145 mm per month.
The variety 'Sauvignon Blanc' grafted on the rootstock Paulsen 1103 was used, with spacing of 3.0 m between rows and 1.5 m between plants. The experimental design was based on randomized blocks with five replications and five plants per plot.
The implantation of the vineyard was carried out in the winter of 2009. The training systems Y-trellis and VSP were compared. The vines were pruned in bilateral spur cordon with two buds for VSP and two to four buds for Y-trellis, which corresponded to an average of 45 and 74 buds per plant, respectively. Management practices (pruning, leaf removal, shoot trimming and phytosanitary treatments) were carried out by the company in accordance with the recommendations of the technicians in charge.
The grapes were harvested on 02/25/2013, 03/06/2014 and 02/12/2015, according to the winery standards. At harvest, five plants per plot were selected to obtain the number of clusters, production per plant and productivity per hectare. The production per plant was determined with an electronic scale, and the results were expressed in kg/plant. Estimated productivity (t/ha) was obtained by multiplying production per plant by planting density (2222 plants/ha).
The leaf area was estimated during the grape harvest. Ten shoots per plot, located in the middle third of the plant, were selected. The central vein length of all leaves of the shoot was measured using a ruler graduated in cm. Total leaf area per shoot was obtained according to mathematical models developed by Borghezan et al. (2010). The total leaf area per plant (cm²) was determined from the average leaf area of the shoot, multiplied by the average number of shoots per plant.
The measurement of balance between vegetative and productive growth was carried out through the ratio between production per leaf area (kg/m²), obtained by the ratio between the average values of production per plant and leaf area. The Ravaz Index (IR) was also calculated, which is the quotient between the average mass of clusters, produced by a given plant in kg, and the average mass of shoots in kg, produced by the same plant, measured during winter pruning.
At harvest, five clusters per plot were randomly sampled to carry out the physical analyzes: cluster length (cm), measured using a digital caliper; cluster weight (g) and rachis weight (g), measured with an analytical scale; number of berries per cluster, obtained by manual counting of berries.
Berry weight (g) was calculated from [(cluster weight -rachis weight) / Number of berries]. The cluster compactness index was obtained by the [(cluster weight) / (cluster length) ²] as proposed by Tello and Ibáñez (2014). The berry diameter (cm) was measured by transversal measurement of diameter from 20 berries per plot.
At harvest, 100 berries per plot were collected to determine grape technological and phenolic maturity. The berries were taken to the Enology Laboratory of Santa Catarina State University, where they were weighed and separated from skins.
From the must obtained from flesh maceration, soluble solids (ºBrix), total acidity (meq/L) and pH were determined, according to the methodology proposed by OIV (2009).
The extract solutions were obtain through the methodology described by Marcon Filho et al., (2015); the concentration of total polyphenols in the skins was determined by the spectrophotometry method of Singleton and Rossi (1965), using the Folin-Ciocalteu reagent (Vetec), and gallic acid as standard, with absorbance measurement at 760 nm.
The data were submitted to analysis of variance (ANOVA) with F Test (p ≤ 0.05) in a 2 x 3 factorial scheme, considering the vintage as a factor. The variables significantly influenced by the vintage were also submitted to principal components analysis (PCA).

RESULTS AND DISCUSSION
The production variables (number of clusters, production per plant and yield per hectare) of 'Sauvignon Blanc' grown in high altitude regions were influenced by the training systems (Table I). On average, 81 clusters per plant for Y-trellis and 54 clusters per plant for VSP were observed. Higher bud load left after winter pruning was related with higher number of clusters obtained in Y-trellis. A similar result was reported by Pedro Junior et al. (2015) when evaluating the same training systems for 'Cabernet Sauvignon'. Regarding the production and yield, Y-trellis presented, in all evaluated vintages, higher averages than VSP. Production ranged from 3.1 to 6.0 kg per plant in VSP and from 6.7 to 12 kg per plant in Ytrellis. The increase in production observed in both training systems from 2013 to 2015 is due to plant development, given that the evaluations started in a four-year vineyard. The production found in this work was higher than the values found by Borghezan et al., (2011). In a previous study that evaluated the productive behavior of 'Sauvignon Blanc' in a high altitude region, it was found production values of 2.7 kg per plant (Borghezan et al., 2011). As much as the work was carried out in the same growing region, this difference in production is due to vegetative conditions of each vineyard.
The plants trained in Y-trellis were on average 93% more productive than theplants trained in VSP, with an accumulated yield in three vintages of 59.1 t/ha, while VSP accumulated 30.5 t/ha. Higher productivity in training systems that allow canopy division were also reported by Hernandes et al. (2013) with 'Niágara Rosada' and by Pedro Junior et al. (2015) with 'Cabernet Sauvignon', both in Brazil. The Y-trellis is recommended because it favors higher number of buds and shoots, increasing yield and allowing better vegetative reproductive balance, without losses in grape quality (Marcon Filho et al., 2015).
The higher values of yield components (number of clusters, cluster weight and number of berries per cluster) observed in plants trained in Y-trellis explain the difference in productivity between the two training systems. Vines with divided canopy tend to have higher yields than those in undivided canopy, generally because of better leaf area exposure and light interception, as well as the higher number of buds that are left per length unit in the pruning line (Reynolds and Vanden Heuvel, 2009). Regarding clusters and berries physical characteristics, influence of the training systems on some evaluated variables was observed (Table II). The cluster weight and the number of berries per clusters were higher for Y-trellis, but the berry diameter and berry weight were not affected by training systems in all evaluated vintages. The compactness index did not differ by F Test (p ≤ 0.05) when comparing the training systems within the same vintage. However, in three vintages average there was a tendency that clusters produced in plants trained in Y-trellis to present higher compactness than those produced in VSP (Table 2).
In a previous study evaluating botrytis bunch rot in two training systems (VSP and Y) with 'Sauvignon Blanc', Bem et al. (2017) observed higher compactness in clusters produced in Y-trellis, which also resulted in higher incidence and severity of botrytis bunch rot. The Y-trellis tends to reduce vine vigor, leading to a better balance between vegetative growth and production (Reynolds and Vanden Heuvel, 2009); however, due to the higher nutrient flow to the berries, there is also an increase in cluster compactness (Hernandes et al., 2013). The cluster compactness index is considered an important factor to assess grape quality (Tello and Ibáñez, 2014). This condition can make plants more susceptible to fungal diseases, especially Botrytis cinerea (Evers et al., 2010;Bem et al., 2017).
Technological (soluble solids, total acidity and pH) and phenolic (total polyphenols) maturity of 'Sauvignon Blanc' at harvest was not influenced by the training systems in all evaluated vintages (Table  III). The soluble solids (SS) content and total acidity (TA) of the must were similar in 2013 and 2014 vintages, around 20 ºBrix and 106 meq/L, for both systems. However, lower values of SS and higher TA were found in 2015 vintage; such situation was due to the occurrence of hail in the vineyard, which caused injuries on the plants, forcing the company to anticipate the grape harvest.
However, it is important to emphasize that, in the first two evaluated years, the levels of SS and TA were appropriate for the production of quality wines. Borghezan et al. (2011) and Brighenti et al. (2013) also found similar values when they studied the variety 'Sauvignon Blanc' in high altitude regions of Brazil. It is noteworthy that, due to the cold climate of these regions, acids degradation will always be slower and, as a consequence, the levels of titratable acidity are high at harvest.  The freshness sensation of wines is directly related to pH and total acidity. In addition, wine pH between 3.3 and 3.6 guarantees better microbiological and physicochemical stability (Jackson, 2014). It was observed, in both training systems, that grape must pH was below 3.3. In this case, lower pH is preferred, because pH tends to increase during and after fermentation.
Regarding the content of total polyphenols in the grape skin, a variation betweenfrom 260.9 to 580.5 mg/L was observed for Y-trellis and from 303.0 to 542.4 mg/L for VSP. This variation is not directly related to the training system, since there was no difference between training systems for polyphenol content in each vintage. Therefore, this variation is ascribed to the effects of climate, nutritional and water status of the plant and grape ripening status at harvest time (Reynolds and Vanden Heuvel, 2009). The skin polyphenols contents observed for 'Sauvignon Blanc' were similar to those obtained by Brighenti et al. (2014) in white varieties like 'Vermentino', 'Verdicchio' and 'Prosecco' grown at altitude, and are higher than those found by Wurz et al. (2018) in the 'Sauvignon Blanc' variety grown at altitude. Similar results were observed by Bavougian et al. (2012), who did not verify the effect of training system on the accumulation of total polyphenols in grape berries.
The relationship between vegetative growth and production of 'Sauvignon Blanc' was influenced by the training systems (Table IV). Considering the results, an index between leaf area and grape production from 16.0 to 19.1 cm²/g for Y-trellis and from 24.9 to 34.6 cm²/g for VSP was observed. The literature reports an interval of 7 to 14 cm² of leaf area per gram of grape to reach berries maturation (Howell, 2001). In high-altitude vineyards, the ideal relationship between leaf area and production was 23 cm²/g for 'Merlot' (Borghezan et al., 2011) and 16 cm²/g for 'Syrah' (Silva et al., 2009).  For the index that considers the relationship between production and leaf area, values of 0.55 to 0.68 kg/m 2 were obtained for Y-trellis and 0.32 to 0.43 kg/m 2 for VSP. According to Kliewer and Dokoozlian (2005), in a study in California/USA, the grapevine was in physiological balance with a ratio between 0.5 to 1.2 kg/m 2 . In Italy, Intrieri and Filipetti (2000) recommended values from 1.0 to 1.5 kg/m 2 for the production of quality wines. In comparison with the parameters cited by the works above, it was found that Y-Trellis trainning system provides better vegetative and yield relationship, resulting in better vine balance.
Concerning the Ravaz index, values of 6.6 and 8.2 were observed for Y-trellis and 4.5 and 6.2 for VSP. According to Kliewer and Dokoozlian (2005), a Ravaz index between 5 and 10 is required to reach berries maturation. Ravaz index values below 5 do not support the increase of wine grape quality, and indicates that plants invested a bigger amount of energy in shoot growth rather than in cluster development .
The search for the best production and quality ratio is not always easy, especially in altitude conditions where the humid climate and fertile soil make growth control a challenge. However, it was observed that for Y-trellis the balance indices evaluated resulted in values closer to ideal. The results obtained indicate that grape production for winemaking is ensured by an adequate leaf area. However, the relation between leaf area and grape production is still outside the ranges reported in the literature; this suggests that it is possible to increase plant productivity in both systems without loss of grape quality.
The principal component analysis (PCA) was performed with the variables that showed statistical difference between the training systems in three vintages. Through the PCA, it was possible to observe that PC1 1 and PC 2 explained 78.38% of the total variance ( Figure 1). PC 1, which accounted for 47% of the total variance, clearly separated the averages of Y-trellis and VSP into two groups. Plants trained in Y-trellis are positively related to: production, number of clusters, number of berries, cluster compactness, Ravaz index and total acidity. The VSP system is related to the following variables: leaf area/ production and berry weight.
Although the results of univariate analysis have detected consistent effects of training systems, a multivariate approach contributes to characterize the effect of training systems in a general context of the evaluated vintages.

CONCLUSIONS
The ratio between vegetative growth and production was influenced by the training systems adopted in high altitude regions of Southern Brazil. Productivity of 'Sauvignon Blanc' was higher in Ytrellis, which resulted in a better vegetativeproductive balance.
The technological and phenolic maturity was similar in Y-trellis and VSP training systems.
The results of this study show that Y-trellis is an alternative to use in high altitude regions of Southern Brazil, as it provides increased productivity of 'Sauvignon Blanc' without compromising grape composition, when compared to VSP.