Location and treatments

The study was conducted in crop seasons 2011/12, 2012/13 and 2013/14 in a vineyard in the city of Agua Doce, located in the Midwestern region of the state of Santa Catarina (SC), southern Brazil (26°59'S latitude, 51°33'W longitude and 969 m altitude). The cultivar was Cabernet Sauvignon (Vitis vinifera) grafted onto the Paulsen 1103 rootstock. The vineyard was installed in 2003 and conducted in espalier system. Planting density was 2299 plants ha^-1, with a spacing of 1.50 m between plants and 2.90 m between rows. The climate of the region was classified as humid mesothermal (Cfb), according to Koppen classification, with an average annual temperature of 16.6 °C and rainfall ranging from 1000 to 1900 mm year^-1. There was a weather station nearby to collect data such as temperature and rainfall during the evaluation period (crop seasons). Non-residual herbicide was used to desiccate the vegetation in the planting rows. Between the planting rows, cover crops such as Paspalum notatum, Trifolium repens L. and Lolium multiflorum were predominant. Throughout the cycle of the grapevine, the cover crops were manually mowed and the residues deposited on the soil surface.

The soil was classified as Typic Dystrudepts (^{Soil Survey Staff, 2014}), located in smooth undulating relief with the following physical and chemical characteristics at 0.0-0.20 m prior to the installation of the experiment: 109.5 g kg^-1 of sand, 444 g kg^-1 of silt and 451.5 g kg^-1 of clay (^{Tedesco et al., 1995}); 62.45 g kg^-1 of organic matter (Walkley Black method), 6.14 pH in water (1:1 ratio); 2.32 and 122.45 mg kg^-1 of available P and K, respectively (both extracted by Mehlich-1); 0.0, 7.05 and 5.75 cmol_c kg^-1 of Al, Ca and Mg, respectively (extracted by KCl 1 mol L^-1) (Tedesco et al., 1995).

The experimental design was a randomized block with four replicates. Each replicate consisted of five plants. The three central plants of each replicate were evaluated. The treatments consisted of N doses (0, 20, 40, 80 and 120 kg N ha^-1 year^-1), whose source was urea (45% N), applied in the area of crown projection on the soil surface, without incorporation. The doses were applied annually at budding, as suggested by ^{Brunetto et al. (2009}).

Leaf collection and nitrogen analysis

At full flowering and veraison, ten fully expanded leaves opposite the first cluster (counted from the apex) were collected in branches of the year of each plant in each crop season. The leaves were dried in a forced air circulation oven at 65°C until constant dry mass was obtained and then ground in a Willey-type mill. The tissue was subjected to sulfur digestion and then distilled using a micro Kjeldahl N distiller (Te 0363, Tecnal, Sao Paulo, Brazil) for the determination of total N (^{Tedesco et al., 1995}).

Yield parameters and must composition

At harvesting, the number of clusters per plant was counted and all clusters were weighed to determine the yield per plant and yield. Berries were collected in the upper, middle and lower portions of five clusters per plant, and then crushed. The following chemical parameters were determined in the must: pH using pH meter (AD8000, ADWA, Szeged, Hungary) calibrated with buffer solutions of pH 4.0 and pH 7.0; total soluble solids (TSS) using a digital temperature-compensated refractometer (RTD 45, Instrutherm, Sao Paulo, Brazil); titratable acidity (TA) by titration (0.1 N of NaOH) using phenolphthalein (1%); and tartaric acid by high performance liquid chromatography (HPLC).

Statistical analyses

The results were submitted to analysis of variance, using Sisvar software, version 5.6 (^{Ferreira, 2003}). Regression analysis was performed when the differences were significant by the F test at 5%. Principal component analysis (PCA) based on the average of the variables (total N content in leaf, yield, average cluster weight, number of clusters per plant, pH, TSS, TA and tartaric acid) was also done using CANOCO software, version 4.5 (^{Ter Braak and Smilauer, 2002}). Equation 1 was used to determine maximum technical efficiency (MTE) of the dose:

Equation (1)

Where: b₁ = x, and b₂ = x²;

N concentration in leaf

The applications of N doses in the soil increased the total N contents quadratically in leaves of Cabernet Sauvignon grapevines collected at full flowering in crop seasons 2011/12 and 2013/14, and in leaves collected at veraison in 2012/13 and 2013/14 (Table 1). The increase in N concentration in leaves at full flowering occurred with the increase of N dose up to 80 kg N ha^-1 in crop season 2011/12 and up to 60 kg N ha^-1 in 2013/14. This increase in N in the leaf tissue indicates that addition of nitrogen fertilizer can increase leaf N content even in soils with high organic matter content. The response to N doses in the soil may be due to the low temperature of the air and soil, inhibiting microorganisms (^{Burns et al., 2016}), but also because of the physical protection of soil organic matter, since the soil presents 451.5 g kg^-1 of clay, which favors the formation of microaggregates. On the other hand, N added to the soil may be related to the high root emission capacity of grapevines during budding and especially at flowering (^{Radville et al., 2016}). This potentiates the uptake of water and nutrients, such as N present in soil, or of N forms resulting from the application of urea at budding (^{Brunetto et al., 2016}).

Table 1 Total N content in leaves collected at flowering and veraison, yield parameters and must composition in 'Cabernet Sauvignon' grapevines subjected to applications of 0, 20, 40, 80 and 120 kg N ha^-1 year^-1 during crop seasons 2011/12, 2012/13 and 2013/14. 

ns = not significant; * = significant with P < 0.05.

The increase of N forms in the soil can stimulate the emission of younger roots by the plants, increasing the volume of soil explored by the root system, which also potentiates nutrient uptake (^{Radville et al., 2016}). However, much of the N absorbed by the grapevines is typically derived from the mineralization of organic matter and decomposing residues of the grapevines themselves or of the plants that co-inhabit the vineyards (^{Brunetto et al., 2014}). This may happen because the addition of N doses via urea can stimulate the activity of the soil microbial population, increasing the mineralization of organic matter and plant residues. Thus, N forms derived from the fertilizer in soil and in solution can be diluted, which increases the probability of N uptake derived from sources other than fertilizer (Yang et al., 2017). Part of the N from the fertilizer, especially ammonium or even nitrate, can be adsorbed to reactive inorganic particles or complexed to the functional groups of organic compounds in soil, decreasing their concentration in soil solution (^{Barlow et al., 2009}).

During flowering, the leaves show intense cell division and elongation, which increases dry matter and, as a result, these organs become drains for nutrients (e.g., N) (^{Schreiner & Scagel, 2006}; ^{Brunetto et al., 2014}). Similarly, N distribution in the leaves can be stimulated by increased leaf area, which raises transpiration and transport of N forms to the leaves. Furthermore, a concentration gradient is generated near the root system, which potentiates the approximation of N forms, such as nitrate, through mass flow, increasing the probability of uptake by the root system (^{Barlow et al., 2009}).

The leaves collected at veraison were also sensitive in detecting the increase of N in the plants subjected to applications of increasing N doses in soil (Table 1), most likely derived from the fertilizer and other soil sources (^{Brunetto et al., 2014}; Brunetto et al., 2016). The increase of N content in the tissue at veraison increased even at doses of 76 and 59 kg of N ha^-1 in crop seasons 2012/13 and 2013/14, respectively. However, N concentrations determined in leaves collected at veraison in 2012/13 were lower than those found in leaves collected at flowering. This can be attributed to reduced cell division in the leaves, which decreases the need for nutrients (^{Marschner, 2012}; ^{Agnelli et al., 2014}). The lowest rate of uptake is related to transpiration, which can also reduce the flow of water and nutrients within the plant and future approximation of N forms to the outer surface of the roots (^{Radville et al., 2016}). However, the lowest N concentration in leaves collected at veraison can be explained by the redistribution of part of the N in the leaves to the clusters, which have intense cell division and increased dry matter (Brunetto et al, 2009). This is because some of the nitrogen compounds present in the leaves, such as amino acids (Marschner, 2012), can be degraded and N can be redistributed to other organs of the grapevine, such as branches of the year, stem, roots and especially clusters (Zapata

et al., 2004).

It should be noted that N concentrations in leaves collected at flowering or veraison were distinct among the crop seasons (Table 1). In crop season 2012/13, N concentration in leaves collected at flowering was higher than that found in 2013/14. The same was observed in the leaves collected at veraison in 2012/13 compared to 2013/14. The highest accumulated rainfall (1906 mm), which occurred in 2013/14 (mainly in October, November and December), may have contributed to this result (Figure 1). Rainfall favors the maintenance of soil moisture. Together with mild temperatures (25°C), it stimulates the action of extracellular enzymes such as urease, which degrade the urea molecule, increasing the infiltration of N forms in soil, both of the fertilizer and of the mineralization of organic matter or decomposing residues. This process reduces N losses by volatilization in the form of ammonia (^{Lorensini et al., 2012}), enhancing the uptake of N forms by the grapevines (^{Brunetto et al., 2016}). There was high rainfall (1400 mm) in crop season 2012/13, but it was 25% lower than in 2013/14. Thus, we believe that the greatest increase of N in the leaves collected in 2012/2013 is a consequence of the best soil water balance, which may have stimulated the emission of young roots by the grapevines, enabling the exploration of a greater volume of soil, increasing the potential for water and nutrient uptake (^{Radville et al., 2016}).

Yield parameters

The applications of N doses increased the number of clusters per plant and average cluster weight in a quadratic manner, which was reflected in increased yield per plant (Table 1) and yield in crop season 2012/13 (Figure 2a). The dose of maximum technical efficiency was obtained with the application of 46 kg N ha^-1 (Figure 2a). In grapevines subjected to doses greater than 40 kg N ha^-1 (80 and 120 kg N ha^-1), we found a decrease in the number of clusters per plant, average cluster weight, yield per plant (Table 1) and yield (Figure 2a). The response of grapevines to low N doses (20 and 40 kg N ha^-1) may be associated to the high organic matter content in soil at 0.0-0.20 m (62.45 g kg^-1).

Figure 1 Accumulated rainfall and average temperature in the experimental area in Água Doce-SC during crop seasons 2011/12, 2012/13 and 2013/14. 

Figure 2 Relationship between yield and nitrogen dose (a) and yield and total nitrogen content in fully expanded leaves collected at flowering (b), in grapevines subjected to the application of 0, 20, 40, 80 and 120 kg N ha^-1 year^-1 during crop seasons 2011/12, 2012/13 and 2013/14. 

Part of the soil organic matter and plant residues deposited on the soil may be mineralized by the microbial population, increasing the concentration of N forms in soil solution, which can be absorbed by the grapevines (^{Barlow et al., 2009}; ^{Burns et al., 2016}), reducing the use of N derived from the fertilizer (^{Agnelli et al., 2014}; ^{Brunetto et al., 2016}). However, part of the N derived from higher N doses may be lost in vineyards by surface runoff (^{Sorrenti & Toselli, 2016}) leaching (^{Lorensini et al., 2012}) and volatilization as ammonia and nitrous oxide (Lorensini et al., 2012; ^{Hannam et al., 2016}), reducing the amount of N derived from the fertilizer used by the plants (Brunetto et al., 2016).

In this study, we believe that the amounts of N derived from the fertilizer lost were small due to the smooth undulating relief of the vineyard, causing reduced runoff of solution on the soil surface. As the soil has high organic matter content (62.45 g kg^-1), most N forms derived from the fertilizer were possibly complexed by soil organic matter (Yang et al., 2017). This decreases leaching as well as volatilization losses, since the average temperatures throughout the year in the region are relatively low (Figure 1).

The average air temperatures in September, which coincided with the budding period in which the N doses were applied, were between 12 and 14 °C in crop seasons evaluated in this study. However, mild temperatures reflect the low release of N from soil organic matter (^{Burns et al., 2016}). In this study, the losses of N derived from the fertilizer in the vineyard were higher in the grapevines subjected to the highest N doses, which showed a reduction in the number of clusters per plant, average cluster weight, yield per plant (Table 1) and yield (Figure 2a). This may be associated to the incidence of diseases in these plants, since part of the N derived from the fertilizer absorbed by the grapevines and diagnosed by the analysis of leaves collected at flowering and veraison (Figure 2b) may have stimulated vegetative growth. This increased vegetative growth may have reduced the incidence of solar radiation in plants and air circulation, stimulating the occurrence of fungal diseases in flowers, during flowering (^{Duchêne et al., 2001}).

In crop seasons 2011/2012 and 2013/14, the application of N doses did not affect the number of clusters per plant, average cluster weight, yield per plant (Table 1) and yield (8.14 and 6.71 Mg ha^-1) (Figure 2a). Such yields are typically found in Cabernet Sauvignon vineyards conducted in espalier (^{Brunetto et al., 2009}). The low response of grapevines to N doses reinforces the possibility of the contribution of N forms derived from sources other than fertilizer, as well as N translocation from plant reserve organs. The mineralization of soil organic matter and the decomposition of grapevine residues (pruning material) or of decomposing cover crop residues deposited on the soil are also considerable sources of N contribution (Brunetto et al., 2014; ^{Burns et al., 2016}). Additionally, the presence of white clover (Trifolium repens L.) between the rows of the vineyards may have caused biological fixation of atmospheric N, increasing the concentration in soil and possibly being absorbed by the root system (^{Radville et al., 2016}). It should be noted that the soil has high levels of organic matter, which together with the absence of toxic aluminum, favors soil microbial activity.

Must composition

TSS contents increased in a quadratic manner with increasing N doses applied to grapevines in crop seasons 2011/12 and 2012/13, and were not affected by N doses in 2013/14 (Table 1). In crop seasons 2011/12 and 2012/13, even though we found a statistically significant difference between the values of TSS in the must of the grapevines subjected to N doses, the values were very close. This is expected and consistent, since there was an increase in cluster weight (Table 1) (^{Cheng et al., 2014}). This normally promotes the dilution of sugars due to the effect of accumulated rainfall and, consequently, of TSS values (^{Hannam et al., 2016}).

The applications of N doses increased pH in must quadratically in crop seasons 2011/12, 2012/13 and 2013/14 (Table 1). However, the increase of N doses in the soil reduced TA in must quadratically in 2012/13 and 2013/14 and tartaric acid in all three crop seasons. This was probably due to the relationship between pH and acidity, which depend on the ionization capacity and acid concentration of the medium (^{Brunetto et al., 2009}). In 2012/13, which had the lowest yield, TA values exceeded 200 meq L^-1, approximately 25% more than in the other two crop seasons.