RESPONSE OF GRAPEVINE CV. ‘BRANCELLAO’ AND ‘SOUSÓN’ TO SUPPLEMENTARY IRRIGATION: WATER RELATIONS, VINE GROWTH, YIELD AND BERRY AND WINE COMPOSITION

Growers are concerned about the negative impacts on vine yield and berry quality caused by global warming. Irrigation systems are increasingly being installed in vineyards in order to counteract those problems. Therefore, an efficient irrigation management is required. In this context, a field experiment was carried out over three years (2012-2014) on red Vitis vinifera (L.) cv. ‘Brancellao’ and ‘Sousón’ in order to assess the effects of supplementary irrigation on vine performance and must and wine composition. Rain-fed vines were compared with a treatment irrigated to 40% of potential evapotranspiration (ET0) from bloom to two/three weeks before harvest. Both


INTRODUCTION
Global warming predictions estimate reductions in rainfall and increasing evapotranspiration for the near future (IPCC, 2014). Modifications in the spatial and temporal distribution of rainfall (Mirás-Avalos et al., 2009) as well as temperature increases (Cruz et al., 2009) have already been reported in Galicia (NW Spain). These changes might cause that grapevine (Vitis vinifera L.) cultivars adapted to temperate and cool climates reach their maximum temperature threshold and lose their specific organoleptic 99 qualities. The negative impacts of water deficit on vine growth, yield and the need to secure and stabilize production (Gouveia et al., 2012) have raised vinegrowers concern for using irrigation in their vineyards, even in humid regions such as northwest Iberian Peninsula. In this sense, deficit irrigation strategies have been successfully adopted as management tools to ensure an adequate balance between vegetative growth and yield while preserving water resources and enhancing berry composition (Dry et al., 2001;dos Santos et al., 2003;Intrigliolo and Castel, 2009). Therefore, irrigation (amount and time) must be applied in accordance with the "terroir" and type of wine to be produced (Ojeda, 2008) in order to maintain an appropriate soil water content, according to grapevine phenological stage, and avoid potential negative impacts on vine vigour, berry composition and wine quality (Jackson and Lombard, 1993;Deloire et al., 2004). Furthermore, the need for customizing irrigation according to cultivar and region of production has been highlighted (Basile et al., 2012).
Even though a great effort on research has been devoted to determine the effects of irrigation and vine water status on yield and berry composition for several grapevine cultivars (e.g. Jackson and Lombard, 1993;Reynolds et al., 2007;van Leeuwen et al., 2009;Intrigliolo et al., 2012), the results were different due to the great variety of climate conditions and vineyard managements. These studies lack from a thorough assessment of the chemical composition of the musts and wines produced, so they do not reflect the possible influence that irrigation may exert on the volatile composition of the wines. Moreover, no studies about irrigation have been conducted on red grapevine cultivars growing in the northwest of the Iberian Peninsula, except for 'Touriga Nacional' (Gouveia et al., 2012), and the research effort has concentrated on white cultivars (Fandiño et al., 2012;Trigo-Córdoba et al., 2015;Cancela et al., 2016).
Although in Galicia white grapevine (Vitis vinifera L.) cultivars are mainly grown, red varieties are also spreading. Among them, 'Sousón' and 'Brancellao' represent a significant area of cultivation in Galicia (NW Spain) and north of Portugal. These two autochthonous cultivars are used for plurivarietal wines. 'Sousón' is a traditional red grapevine cultivar that is spreading over the Ribeiro Designation of Origin (DO), nowadays it is ranked first in production. This cultivar is characterized by its high polyphenolic potential and is used for providing colour to plurivarietal wines (Consello Regulador Ribeiro, 2016); however its growing cycle is long. 'Brancellao' is the third autochthonous red cultivar in the Ribeiro DO. Its growing cycle is medium and it is used as a complement for plurivarietal wines since it has low polyphenolic content but a good aromatic expression (Consello Regulador Ribeiro, 2016). The potential of 'Brancellao' as a possible alternative to the 'Mencía' cultivar for the elaboration of quality red wines, due to its maturation characteristics and phenolic composition has been already pointed out (Cortés and Díaz, 2011).
Therefore, the aim of the current study was to assess the effects of supplementary irrigation on the agronomic performance of two red grapevine cultivars, 'Brancellao' and 'Sousón', grown in Galicia. We also determined the influence of supplementary irrigation on must and wine composition of both cultivars, including amino acids and volatile compounds. The current study tried to determine whether it is necessary or not to apply supplementary irrigation to two red grapevine cultivars in Galicia in order to counteract the negative effects of climate change.

Description of the study site
The experiment was carried out, over three consecutive years (2012)(2013)(2014), in two adjacent 0.2-ha vineyards (Vitis vinifera L.) planted in 1998 with red cultivars 'Brancellao' and 'Sousón' grafted on 196-17C rootstock. The vineyards were located at the experimental farm of the Estación de Viticultura e Enoloxía de Galicia (EVEGA) in Leiro (Ourense, NW Spain), within the Ribeiro DO (42º 21.62' N, 8º 7.02' W, elevation 115 m). Vines were trained to a vertical trellis on a single cordon system (10-12 buds per vine). Spacing was 2.4 m between rows and 1.25 m between plants (3333 vines/ha). Rows were East-West oriented. Cultural practices were those commonly applied in the area and weed growth was controlled mechanically.
Climate is Atlantic with average annual rainfall of 900 mm, of which about 70% falls during the dormant period. It has been described as temperate, humid with cool nights (Fraga et al., 2014).

Experimental design
For each cultivar, two treatments were established: rain-fed and drip irrigation to 40% potential evapotranspiration (ET0). Weather data for 100 calculating ET0, using the Penman-Monteith equation (Allen et al., 1998), were recorded at a weather station located 150 m from the experimental site. Precipitation was subtracted from ET0 each week. The calculated amount of water was applied the following week. Irrigation was triggered in late June (just after bloom) and water was applied till mid-August, between two and three weeks before harvest.
Water was applied early in the morning, for minimizing evaporation losses, through two pressurecompensated emitters (4 L/h) per vine, located 25 cm on either side of the trunk. Frequency of water applications varied from 3 to 5 days per week, depending on the amount of water required. Irrigation water was of good quality, with pH of 6.5, electrical conductivity of 163.4 μS/cm and 0.4 mg/L of suspended solids. Total irrigation depth per season was 50, 79.3 and 50 mm, in 2012, 2013 and 2014, respectively. The experiment was laid out in randomised blocks, with three replicates. Each of the replicates consisted of three rows with 12 vines per row. The eight vines in the centre of the middle row were used for measurements and the rest acted as buffer vines. In summary, for each year and cultivar, two treatments and three replicates per treatment were considered.

Field measurements
Midday stem water potential (Ψs) was measured at noon (12:00 -13:00 h) on non-transpiring leaves that had been bagged with both plastic sheet and aluminium foil for at least 1 hour before measurements (Choné et al., 2001). Readings were taken fortnightly from mid-May until mid-September using a pressure chamber (Pump-Up, PMS Instruments, Albany, OR, USA). One leaf from two vines per replicate was used per date (6 leaves per treatment and cultivar). This modality of leaf water potential has been proven the most useful for assessing vine water status in Galician cultivars .
On the same days as Ψs, stomatal conductance was measured around solar midday using a leaf porometer (Model SC-1, Decagon Devices, WA, USA) on one fully-exposed and healthy leaf per vine and two vines per replicate (6 leaves per treatment and cultivar).
Chlorophyll concentration index (CCI) was estimated non-destructively with a portable chlorophyll meter (CCM-200, Opti-Sciences, Tyngsboro, MA, USA), which calculates a unitless CCI value from the ratio of optical absorbance at 655 nm to that at 940 nm. This value is correlated with total foliar extractable chlorophyll (Steele et al., 2008). These measurements were performed from veraison to harvest on three leaves per plant and three vines per replicate (9 vines per treatment). Major veins and areas of obvious visual damage or disease were avoided.
Yield was determined at harvest on each of the internal rows (8 vines/row) of each replicate. The number of clusters per vine was recorded. Average cluster weight was calculated by dividing yield per vine by the number of clusters. Berry fresh weight was determined on random samples of 200 berries per replicate.
Pruning weight (PW) and external leaf area (SA) were determined in six vines per replicate. External leaf area was estimated after veraison, when shoot growth had ceased, following the method proposed by Sánchez-de-Miguel et al. (2010), in which the width and height of the canopy are collected at five different spots along the vine using a measurement tape. Data on SA could not be obtained in 2013 because vines were trimmed prior to performing of these measurements.
Water productivity (WP) was assessed as yield (in kg per ha) divided to the amount of water received during the growing season (rainfall + irrigation).
Gross incomes were assessed using the prices issued by Xunta de Galicia (Diario Oficial de Galicia 16th July 2008) for the Ribeiro DO. Grape is paid according to its probable alcoholic grade. Hence, one kg of 'Brancellao' or 'Sousón' grapes with 13% vol. probable alcoholic grade would be paid at 1.39 €; whereas 1 kg with 11% vol. would be paid at 1.06€. In our case, both rain-fed and irrigated vines received the same management practices except for the additional costs of irrigation. Therefore, gross incomes might provide an indication of the sustainability of irrigation in the medium term.

Winemaking
Grapes from each cultivar and treatment were manually harvested and transported to the experimental winery in field boxes. Winemaking was performed at EVEGA separately on samples of about 40 kg from each field replication.
Grapes from each cultivar and treatment were separately destemmed and transferred to 35-L stainless steel containers. During grape processing SO2 (50 mg/L) was added to the mass in order to avoid oxidation and for microbiological control. Fermentations were carried out at room temperature (22-24 ºC). Excellence XR (Lamothe-Abiet, Bordeaux, France) yeast was added according to manufacturer's instructions. Density and temperature of fermentations were daily monitored. Wine lots were punched down daily until the end of alcoholic fermentation (8 days). Then, they were pressed, racked into new tanks and kept at room temperature for a couple of days. A tartaric stabilization was carried out at 4 ºC for one month, approximately. Finally, wines were filtered, sulphited to 35 ppm of free sulphur dioxide, bottled and stored.
Analytical determinations in the wines were performed just after alcoholic fermentation.

Must and wine basic attributes
The analyzed attributes of musts comprised total soluble solids content (SSC), pH, titratable acidity (TA) and malic and tartaric acid concentrations. They were determined using the official methods for must analysis (OIV, 2009) Wine alcohol content, pH, and TA were determined by Infrared Spectrometry with Fourier Transformation (IRFT) using a WineScanTM FT120 (FOSS®, Barcelona, Spain) calibrated according to official protocols (OIV, 2009). Malic and tartaric acids were determined by enzymatic methods using a LISA200 Analyzer System.
Wine colour characteristics, including total polyphenol index (TPI), anthocyanins and tannins concentrations were assessed with an ultravioletvisible Themo Helios Zeta spectrophotometer using the methods described by Glories (1984) and Zamora (2003). TPI was directly measured after diluting 1% the wines. Total anthocyanins were quantified according to the decolouration experienced by adding metabisulphite and were expressed as mg/L; total tannins were determined after wine treatment with concentrated hydrochloric acid.

Analysis of amino acids and aroma compounds
Amino acids concentrations in must and wine samples were assessed following the method proposed by Gómez-Alonso et al. (2007) and Garde-Cerdán et al. (2009), with slight modifications. Briefly, must and wine samples were subject to a 30minute derivatization reaction using DEEMM (diethylethoxymethylenemalonate) as derivatizing agent. Then, they were heated at 70 ºC for 2 hours in order to degrade the excess of DEEMM and reagent by-products. Finally, amino acids were eluted and their concentrations detected at 280 and 269 nm using the photodiode array detector of a High Performance Liquid Chromatography (HPLC) equipment (1100 series, Agilent Technologies, Palo Alto, CA, USA). Determinations were carried out on one sample per replicate (hence, three samples per treatment and year for each cultivar). Each of these determinations was performed three times in order to account for the reproducibility of the methods. Further details for these determinations, including a complete list of chemical reagents and elution conditions, can be found in Bouzas-Cid et al. (2015).
Higher alcohols and other volatile compounds were determined by gas chromatography (GC). Each wine sample from each cultivar, treatment replication and year was analyzed in triplicate. An Agilent (model 7890A) chromatograph equipped with a flame ionization detector was used. A capillary column CP-Wax57CB (Chrompack) (50 m x 0.32 mm x 0.20 μm) was employed. The operating conditions were the following: injector and detector temperatures, 230 ºC; column temperature, programmed from 40 ºC to 200 ºC at 4 ºC/minute and final isotherm for 15 minutes; carrier gas (Hydrogen) flow 3.3 mL/minute; auxiliar gas (Nitrogen) flow, 30 mL/minute. A 2 μL sample of the extract was injected in split mode (purge rate 5.2 mL/minute, split 125 mL/minute, attenuation of 1).

Statistical analysis
A two-way ANOVA was carried out on the agronomic and chemical data (R software v. 3.2.3; R Core Team, 2015) accounting for irrigation treatment, year and their interaction as factors. Differences were considered significant when p < 0.05, according to the Tukey's Honest Significant Difference (HSD) test. Cultivar was accounted for in the first analysis of the data set (reflected on Table I) in order to ascertain if both cultivars behaved differently; then, data from each cultivar were independently analyzed..

Climate conditions and vine water relations
The time course of daily ET0 and rainfall over the growing seasons (April to late-September) studied is shown in Figure 1. Rainfall was concentrated on the beginning of the growing season in 2012, whereas it was evenly distributed in 2014. The growing season of 2013 showed less rainfall events than the other two years, and they were mainly concentrated at the beginning of the season. Maximum ET0 values were similar among seasons; however, these records were higher at the beginning of the season in 2013 and 2014 than in 2012. A summary of the total values for the climate variables recorded at this site can be found in Trigo-Córdoba et al. (2015); briefly, the driest year was 2012 with 841 mm of annual rainfall, slightly lower than the 900 mm average for the area. However, the 2012 growing season (April to harvest) was the wettest one with 313 mm rainfall. In contrast, 2013 and 2014 were wet years with, approximately, 1300 mm rainfall; nevertheless, the growing seasons of these years were drier than that of 2012 (163 and

Precipitação (barras) e evapotranspiração de referência (linhas) diárias na finca experimental durante os anos 2012 a 2014. Se amostram as temperaturas máximas, mínimas e médias diárias para cada ano. As frechas (vermelhas para 'Brancellao' e pretas para 'Sousón') indicam as fases de brotação (B), floração (F), pintor (V) e colheita (H).
Irrigation allowed vines of both cultivars to maintain a higher water status than those rain-fed, as proven by the midday stem water potential values (Figure 2), which tended to decline progressively over the course of the growing season due to higher temperatures and solar radiation values that caused an increasing evaporative demand (as shown in Figure 1). This decreasing pattern reflects the decrease in soil water availability along the season. The lowest values of midday stem water potential were observed in 2013 and 2014 ( Figure 2) for both cultivars. In the case of 'Brancellao', minimal values of -0.88 MPa were measured in 2013 and 2014. In the case of 'Sousón', minimal stem water potentials reached -1.08 MPa in 2013. Both cultivars showed differences between rain-fed and irrigation conditions regarding water status. These differences were significant from day 200 (19th July), approximately ( Figure 2) and were detected on more occasions for 'Brancellao' than for 'Sousón', indicating that these cultivars present a different behaviour in regard to water status, as previously observed for two white cultivars . Therefore, this kind of studies must be conducted for each cultivar in order to schedule irrigation appropriately, as suggested by Basile et al. (2012). It is no doubt that vine response to water stress depends on the phenological stage (Jackson and Lombard, 1993); therefore, irrigation can be used as a tool to control vine water status as a function of phenological stage, water stress degree and the aims of the vineyard (Ojeda, 2008).  Both cultivars showed high values of stomatal conductance (Figure 3), greater than those reported by other authors (Williams and Araujo, 2002;Intrigliolo and Castel, 2009), probably caused by a sufficient soil water availability and a high relative humidity in the atmosphere during our study. The fact that the measurement interval of the equipment used is rather narrow, measurements greater than 0.7 mol H 2 O/m 2 /s should be taken with caution. However, comparisons between treatments are allowed because all the measurements were performed with the same equipment and conditions. Rain-fed vines had lower stomatal conductance values than irrigated vines on certain dates, especially for 'Sousón'; although no significant differences between treatments were observed in most of the measurement dates due to the high variability in these readings. In addition, the values observed in both rain-fed and irrigated vines did not reflect stomatal closure by the end of the season, suggesting that plant functioning was not altered by the treatment. Most of the published reports indicate an increase in stomatal conductance under irrigation conditions (e.g. de Intrigliolo and Castel, 2009); however, these studies were carried out under Mediterranean or semi-arid conditions, where water constrains were higher than those observed in our study; thus, the high values found in our study are somewhat expected. In addition, Williams and Trout (2005) and Teszlák et al. (2013) observed peak values of stomatal conductance greater than 0.8 mol H 2 O/m 2 /s in wellirrigated 'Thompson Seedless' and 'Riesling' vines, similar to our findings and those from a previous study on two white grapevine cultivars . Furthermore, midday stem water potential and stomatal conductance values measured in the current work are over the optima (-1.25 to -1.4 MPa and 0.12 to 0.15 mol H 2 O/m 2 /s, respectively) suggested by Romero et al. (2010) for 'Monastrell' cultivar under the semi-arid conditions of south east Spain, which might be the reason for the absence of differences among treatments in the present study. In 'Brancellao', CCI values were lower under rain-fed than under irrigation conditions only in 2013 ( Figure  4 a and c). However, 'Sousón' also showed significant differences for this variable on one date in 2014 (Figure 4 b and d). The absence of differences between treatments in 2014 might have been caused by the even distribution of rainfall over the growing season. Due to the fact that CCI is directly correlated to extractable chlorophyll content (Richardson et al., 2002), our results suggest that leaf chlorophyll in the two cultivars studied was very sensitive to slight changes in vine water status.

Vegetative growth and yield components
Most of the vegetative growth and yield variables measured in the current study were affected by cultivar and season (Table I). In contrast, treatment only exerted a significant effect on pruning and berry weights. The interaction between cultivar and treatment was significant for pruning, cluster and berry weights.
In the case of 'Brancellao', most of vegetative growth and yield components were unaffected by irrigation (Table II)

Table II
External leaf area, pruning weight, yield components, water productivity and expected gross incomes of rain-fed and irrigated vines during each season for the two cultivars considered in this study.

Superfície foliar externa, peso de poda, componentes da produção, produtividade da água e renda bruta esperada para as vinhas em sequeiro e regadas durante cada ano para as duas variedades estudadas. Parameter
Year Brancellao Sousón In the case of 'Sousón', neither SA nor PW were affected by irrigation (Table II). In contrast, all of the yield components accounted for in this study were negatively influenced by irrigation at least on one season. Hence, yield was lower in irrigated than in rain-fed vines in 2012 and 2013 due to lower number of clusters per vine and lower cluster weights, which could have been caused by a greater incidence of bunch rot detected on irrigated vines. On the contrary, berry weight was greater in irrigated vines in 2013.
Both cultivars showed higher PW in irrigated vines, although these differences were not always significant, especially for 'Sousón'. Overall, these results suggest that vine vegetative growth was stimulated by the water applied, as observed by Intrigliolo and Castel (2010) for 'Tempranillo'.
The lack of effects of irrigation on vine yield may be attributed to high water supply by rainfall, which was sufficient for fulfilling vine water requirements. These results are in disagreement with previous reports for other cultivars and climate conditions (Reynolds et al., 2007;Intrigliolo and Castel, 2010;Gouveia et al., 2012). However, Junquera et al. (2012) and Lanari et al. (2014) did not detect any difference in yield until the third year of experiments, suggesting that vineyard irrigation studies might be carried out for longer time periods. Moreover, under the conditions of the current study, rainfall events and the volume and characteristics of the soil explored by the root system may have buffered the yield response under irrigation shortage ensuring full canopy transpiration until July. Consequently, there was a very slow development of water stress, which only was mild to moderate (van Leeuwen et al., 2009), depending on the cultivar, over the three years of study. Hence, rain-fed vines reached lower plant water status than the irrigated ones only late in the season (mid-July, August). Moreover, it is important to take into account the different length of the growing cycle for the two cultivars studied; 'Sousón' has a more lengthy maturation period, thus it is not harvested till October or even November, when rainfall is already occurring in the region, whereas 'Brancellao' is harvested by the end of September. This has probably caused the lack of differences in yield and its components between treatments.
Water productivity was similar between treatments for 'Brancellao'. However, WP was greater for 'Sousón' rain-fed vines than for those irrigated (Table  II). These observations can be explained by the fact that WP index was developed for Mediterranean climates, where summer rainfall amounts are much lesser than those registered in our study, as previously suggested by Cancela et al. (2016).
Gross incomes were similar between treatments in the case of 'Brancellao'; however, they were higher for rain-fed vines than for those irrigated in the case of 'Sousón' (Table II). From the economic viewpoint, irrigation was not a viable practice for these red cultivars under the conditions of this study, especially for 'Sousón', due to the fact that gross incomes were not increased and higher costs must be faced when using irrigation.

Must and wine basic attributes
Cultivar and irrigation treatment influenced all the must attributes considered in the current study, except for tartaric acid concentration (Table I). Season affected the concentrations of tartaric and malic acids; whereas the interaction between cultivar and season was significant for pH and tartaric acid concentration.
In the case of 'Brancellao', significant differences between treatments were only observed for pH and TA (Table III). Rain-fed vines produced musts with lower values of TA in 2012 and 2014 than irrigated vines. Moreover, in 2012, musts from rain-fed vines showed a higher pH than those from irrigated vines. In the case of 'Sousón', must attributes remained unaffected by the irrigation treatment. However, in 2014, musts from rain-fed vines presented a higher SSC value than those from irrigated vines (Table III). These observations seem to indicate that berry composition of both cultivars was not affected by irrigation, and the values found for SSC and TA under irrigation were very similar to those of the rain-fed treatment, suggesting an excess of source compared to sink demand. External leaf area values are within the range considered optimal to ensure maturation of up to 12 t/ha (Schneider, 1989). Our yield ranged from 11 to 16 t/ha, depending on the cultivar and year, and thus differences in must composition is not expected. However, these results are in disagreement with reports for other cultivars grown in other regions under Mediterranean or semiarid conditions, where sink demands are not covered due to severe water restrictions (Intrigliolo and Castel, 2010;Romero et al., 2010), proving that irrigation might not provide any benefit under the current Galician climate conditions, as previously reported for two white grapevine cultivars .
Wine attributes depended on the cultivar (Table I). However, irrigation treatment only influenced alcohol content, TA, pH and the concentration of tartaric acid; whereas season only affected pH, TPI and anthocyanins ( Table I). The interaction between cultivar and season was significant for TA, pH, malic acid and anthocyanins.
'Brancellao' wine basic attributes were only affected by irrigation in the last year (Table IV). Wines from the rain-fed treatment showed higher values of pH and TPI, whereas wines from the irrigated treatment had a higher concentration of tartaric acid.
'Sousón' wine composition was very similar between treatments for the three years of study (Table IV). Significant differences were only detected for alcohol content and TPI.
Similarly to the observations in must composition, wines were mainly unaffected by irrigation. Nevertheless, those slight variations might be detected at the sensory level, as previously reported for white cultivars . In fact, wines from the irrigated vines tended to have lower alcohol concentrations and higher total acidities than those from the rain-fed treatment. This trend was observed for both cultivars and has been previously reported for Spanish red grapevine varieties (e.g. Intrigliolo and Castel, 2010).

Amino acids and volatile compounds
Concentrations of individual amino acids in the musts from both cultivars varied from year to year (Tables V and VI). In the case of 'Brancellao', the concentrations of seven amino acids remained unaffected by the year: arginine, alanine, tyrosine, isoleucine, phenylalanine, ornithine and lysine, as well as the ammonium ion (Table V). In the case of 'Sousón' musts, only three amino acids showed similar concentrations over the three studied years: glutamic acid, tyrosine and ornithine (Table VI). This strong effect of vintage on the amino acids concentrations in musts has been previously observed in Galician white cultivars ; and it can be explained by the different degree of berry maturation due to the particular weather conditions of a given year, as reported also for other cultivars (Ortega-Heras et al., 2014). In contrast, irrigation only affected the concentrations of certain amino acids in some of the studied years (Table V and VI). For instance, in the case of 'Brancellao', irrigation increased the concentrations of alanine, γ-aminobutyric acid and ammonium ion in 2013, and tryptophan in 2014 (Table V). Similarly, concentrations of amino acids in 'Sousón' musts were very similar between treatments in 2012 and 2013 (only tyrosine concentration was increased under irrigation conditions in 2013). However, in 2014, 'Sousón' musts from the irrigated treatment showed higher concentrations in asparagine, threonine, valine, cysteine, isoleucine, tryptophan and leucine that led to a greater concentration of total amino acids in musts from the irrigation treatment (Table VI). .06 * Different letters in the row for each year indicate significant differences between treatments at p < 0.05. The significance of the year as factor is expressed as ns = non-significant; * p < 0.05; ** p < 0.01; *** p < 0.001.

Table VI
Amino acids and ammonium concentrations (mg/L) in the musts from 'Sousón' red grapevine cultivar in the three vintages studied  339.46 360.86 194.20 165.48 148.10 a 189.24 b ** Different letters in the row for each year indicate significant differences between treatments at p < 0.05. The significance of the year as factor is expressed as ns = non-significant; * p < 0.05; ** p < 0.01; *** p < 0.001.

Concentração de aminoácidos e ião amónio (mg/L) nos mostos da variedade tinta 'Sousón' nas três vindimas estudadas
In the case of 'Brancellao', all the amino acids were present in the range of concentrations reported in the literature (Bell and Henschke, 2005), except for proline and methionine, which appeared in lower concentrations. In the case of 'Sousón', amino acids concentrations were low when compared to those previously reported, especially for glutamic acid, glutamine and proline. Both cultivars were arginine accumulators since their ratio proline/arginine was lesser than 1, as observed for other cultivars traditionally grown in this region , and other widely-grown cultivars such as 'Syrah' (Garde-Cerdán et al., 2009). In fact, proline and arginine metabolism are linked and the final concentration of one of them is influenced by the concentration of the other, with arginine as a precursor for at least some of the proline accumulated in the grapes (Kliewer, 1968). The most abundant amino acids found in 'Brancellao' and 'Sousón' musts were arginine, GABA, alanine, glutamine and glutamic acid although their concentrations changed over the study period. These amino acids constituted 64-71% of the total free amino acids in musts from 'Brancellao', and 56-61% in musts from 'Sousón'.
Irrigation did not affect the total concentration of amino acids in the wines (Figure 5), except for 'Brancellao' wines in 2013. Amino acids concentrations in 'Sousón' wines increased from year to year. Slight differences between treatments were detected for certain amino acids concentrations in some years for both cultivars (data not shown). In general, the concentrations of higher alcohols and other volatile compounds were affected by the year, as previously reported by Vilanova et al. (2012). In 'Brancellao', all compounds were affected except for methanol, ethyl acetate, 1-propanol, acetol and 2,3butanediol levo (Table VII). In contrast, concentrations of these compounds in 'Sousón' wines remained unaltered from year to year except for those of methanol, ethyl acetate, acetaldehyde, benzylic alcohol and ethyl lactate (Table VIII).
Irrigation only affected wines from 'Brancellao' in 2013. The concentrations of benzylic alcohol, 2phenylethanol, 2,3-butanediol levo and meso were significantly greater in the wines from the rain-fed treatment (Table VII). This suggests that irrigation might exert an effect on wine aroma compounds. In fact, acetaldehyde concentration was lower than the threshold for this compound, which causes wine oxidation (Swiegers et al., 2005). Moreover, irrigation reduced the concentration of 2phenylethanol, an aromatic compound that provides floral notes, which could lead to a reduction in the wine aroma complexity.
In the case of 'Sousón', irrigation influenced the concentrations of several of these compounds in the three years of the study. In 2012, acetol concentrations were lower in wines from irrigation than in those from the rain-fed treatment. In 2013, ethyl acetate and acetaldehyde showed, respectively, a greater and a lower concentration in wines from the rain-fed treatment. Finally, in 2014, wines from the irrigation treatment presented lower concentrations of ethyl lactate, 2,3-butanediol levo and meso and γβ-lactone than the wines from the rain-fed treatment (Table VIII).
Recently, Talaverano et al. (2016) found higher alcohols, C6 compounds and phenol volatiles concentrations in 'Tempranillo' wines from their rain-fed treatment than in those from the deficit irrigated treatments, whereas the highest concentrations of ethyl esters were detected in wines from the irrigated treatments. These findings are in accordance with the ones reported in the current study. 19 * Different letters in the row for each year indicate significant differences between treatments at p < 0.05. The significance of the year as factor is expressed as ns = non-significant; * p < 0.05; ** p < 0.01. LOD = Limit of Detection

Table VIII
Irrigation effects on volatile compounds concentrations (mg/L) in wines from 'Sousón'. The significance of the year is also shown. ns Different letters in the row for each year indicate significant differences between treatments at p < 0.05. The significance of the year as factor is expressed as ns = non-significant; * p < 0.05; ** p < 0.01. LOD = Limit of Detection.

CONCLUSIONS
Independently of the cultivar, irrigated vines showed less negative stem water potential values than vines under rain-fed conditions. However, stomatal conductance was not altered, indicating that water supplied by rainfall was sufficient for adequate plant functioning in both cultivars. Pruning weight was higher in irrigated vines, suggesting a greater vegetative growth under this treatment. In contrast, irrigation did not increase yield in these cultivars. Must and wine composition, including amino acids and volatile compounds concentrations, were negatively affected by irrigation. Moreover, water productivity was greater in rain-fed vines in the case of 'Sousón', whereas no significant differences were detected for 'Brancellao'. Consequently, gross incomes were not increased by the irrigation practice. The results from this study indicated that irrigation in vineyards from a humid region such as Galicia might still not be necessary, although research in other areas within the region and with other irrigation protocols (timeframes and amounts) may provide other results.