PHYSIOLOGICAL AND AGRONOMICAL RESPONSES TO ENVIRONMENTAL FLUCTUATIONS OF TWO PORTUGUESE GRAPEVINE VARIETIES DURING THREE FIELD SEASONS

Extensive agricultural losses are attributed to heat, often combined with drought. These abiotic stresses occur in the field simultaneously, namely in areas with Mediterranean climate, where grapevine traditionally grows. The available scenarios for climate change suggest an increase in the frequency of heat waves and severe drought events in summer, also affecting the South of Portugal. In this work we monitored several production- related parameters and evaluated the state of the oxidative stress response apparatus of two grapevine varieties, Touriga Nacional (TN) and Trincadeira (TR), with and without irrigation, during three field seasons (2010 to 2012). Overall, results point to a high correlation of most yield and stress-associated parameters with the specific characteristics of each variety and to each season rather than the irrigation treatments. In the season with the driest winter, 2012, the lack of irrigation significantly affected yield in TR, while in the two other seasons the impact of the irrigation regime was much lower. In 2012, the yield of TN was affected by environmental conditions of the previous season. The irrigation treatments significantly affected berry size rather than quality. ao calor, frequentemente combinado com seca. Esses stresses abióticos ocorrem simultaneamente no campo, nomeadamente em áreas com clima mediterrâneo, onde tradicionalmente se cultiva a videira. Os possíveis cenários de alterações climáticas apontam para um aumento na frequência de ondas de calor e eventos de seca prolongada no verão, afetando também o Sul de Portugal. Neste trabalho foram monitorizados vários parâmetros relacionados com a produção, e a resposta ao stresse oxidativo foi avaliada em duas variedades de videira, Touriga Nacional (TN) e Trincadeira (TR), com e sem irrigação, durante três épocas sucessivas (2010 a 2012). Em geral, os resultados obtidos apontam para uma elevada correlação da maioria dos parâmetros associados ao stresse e ao rendimento com as características específicas de cada variedade e com as condições ambientais de cada ano e não directamente com os tratamentos de irrigação. No ano cujo inverno foi mais seco, 2012, a falta de irrigação afetou significativamente o rendimento em TR, enquanto nos dois outros anos o impacto do regime de irrigação foi muito menor. Em 2012, o rendimento de TN foi afetado pelas condições ambientais da temporada anterior. Os tratamentos de irrigação não afetaram significativamente a qualidade da uva, apenas o seu tamanho.


INTRODUCTION
As sessile organisms, plants are constantly exposed to changes in the surrounding environment. Temperatures above the normal optimum are sensed as heat stress by living organisms. Heat stress disturbs cellular homeostasis and can lead to severe impairment in growth and development, and even to death. Worldwide, extensive agricultural losses are attributed to heat, often in combination with drought or other stresses. In fact, drought and heat represent an excellent example of two different abiotic stresses often occurring simultaneously (Knight and Knight, 2001), namely in regions of long-time grapevine production, such as the Mediterranean surrounding areas. Furthermore, the available scenarios for climate 2 change over the pending decades suggest an increase in aridity and shifts in the amount, seasonality and distribution of precipitation, affecting the Mediterranean region (Pinto et al., 2011). The predicted increase in the frequency of summer heat waves and the simultaneous increase of the duration of the dry season will lead to extended and severe drought events, with a concomitant overexploitation of water resources for agriculture purposes, increasing limitations to plant growth and fruit development and therefore to fulfilling their potential yields (Chaves, 2002;Chaves et al., 2003). Furthermore, it is expected that, as early as 2040, species such as grapevine will have moved their distribution northward and uphill, leading to changes in plant phenology, anticipating flowering and ripening (Fraga et al., 2016;Ramos et al., 2015).
The importance of water supplement in wine grapes was first studied in Portugal ca. twenty years ago (Lopes, 1994). In the future it will ensure the longterm sustainability of viticulture, with grapevine plants showing up normal physiological activity and an adequate balance between vegetative and reproductive development, while preserving yield quality (Rodrigues et al., 2008). Vines with no supplemental water or under deficit irrigation have less vegetative growth, smaller berries and lower yields than those with high water availability (Williams et al., 2009). However, grapevine irrigation, unlike in other crops, should usually be sub-optimal, to avoid a decrease in quality (Gaudillère et al., 2002).
The many abiotic stresses that significantly limit the distribution of grapes around the world and reduce crop yield, as is the case of water deficit, can also be used in a positive way to enhance berry flavor and quality characteristics (Chapman et al., 2005). Deficit irrigation with moderate water stress is in fact associated with increased fruit quality, especially for varieties used in wine production (Williams et al., 2009). The resulting reduced shoot vigor, competition for scarce carbon resources, reduced berry size, concentrating flavors and color will improve berry quality (Castellarin et al., 2007;Deluc et al., 2009). In fact, in red wine grapes, the occurrence of some water deficit during the growing season has been interpreted as beneficial for wine quality, leading to wines with more fruity and less vegetal aromas and flavors than vines with higher water status, that show a tendency for vegetal aromas, pepper flavor and astringency.
Typically, crop productivity is dependent on photoassimilates produced at the whole plant level. The decline in stomatal aperture normally observed under drought is accompanied by an adjustment of leaf area at the whole plant level, either through the inhibition of new leaf growth or through the earlier senescence of older leaves, in the case of prolonged stress. This causes a decrease of area available for transpiration but also to lower intercepted radiation throughout the growing season with consequences in biomass production (Pereira and Chaves, 1993). This decrease in foliage can also lead to excessive fruit exposure to sunlight (Pellegrino et al., 2005) with undesirable effects on berry production (sunburn). The plant tries to avoid dehydration and excessive irradiation by changes in the leaf angle, aiming at smaller angles, which protect against excess solar energy but also diminish carbon assimilation (Pinheiro and Chaves, 2011).
To study the response of the plant to abiotic stress, it is relevant to approach it under field conditions instead of growth-rooms since the most important feature of grapevine production lies in berry production and quality and even phenotypic characteristics are altered (Mishra et al., 2012). Thus, the study of stresses in an artificial environment can elucidate important mechanisms of resistance , but one must always consider the whole environment that a plant grows in to fully comprehend the 'stress resistance' mechanisms build up in the field. In this work we monitored the phenotypical evolution together with several production-related parameters and the status of the oxidative stress response machinery in two Portuguese grapevine varieties, Touriga Nacional (TN) and Trincadeira (TR), with and without irrigation, during three field seasons (2010 to 2012) and integrated the results obtained with the meteorological data registered in the field region. These varieties were chosen for their contrasting responses to abiotic stress, measured in controlled conditions (Carvalho et al., 2015, that characterized TN as tolerant to stress and adapted to withstand heat waves and combinations of several abiotic stresses while TR is more sensitive and its growth is impaired under a single abiotic stress. The choice of extreme irrigation methods also allowed the clear discrimination of the response behavior of both varieties upon water scarcity and abundance.

Location and climate
The experiments were conducted at a research vineyard (Centro Experimental de Pegões) located in Pegões (38º40'N; 8º36'W), 70 km east of Lisbon during three seasons (2010 to 2012). The climate is Mediterranean, with hot, dry summers and mild air temperatures with precipitation concentrated during autumn and winter, although the climate in Pegões is affected by its proximity to the sea and the mountain of Arrábida, which causes low temperatures in the night period. When comparing the 3 years, 2010 was found to be the hottest with maximum and minimum summer temperatures ca. 3 ºC higher than those measured in the following years ( Figure 1). The year 2012 was characterized by lower precipitation in the winter than the previous years and in 2011, spring and fall temperatures were higher and summer temperatures lower than in the other two years ( Figure 1). All these factors may have contributed to a less severe summer drought stress experienced by plants in 2011. Irradiation was measured throughout the seasons using a radiation sensor (LI-250 Light Meter, Li-Cor, Lincoln Nebraska, USA). The soil is derived from podzols, with a sandy surface layer and a clay rich (low permeability) horizon at a depth of ca 1 m. A mixture of clones resulting from polyclonal selection (certified material with a "standard" designation) of Touriga Nacional and Trincadeira were established in the vineyard. Both were grafted on 1103 Paulsen rootstock in 2002. The plants are spaced 2.5 m between rows and 1 m within rows, resulting in a density of 4000 plants ha -1 , and trained on vertical trellises each with a pair of movable foliage wires for upward shoot positioning.

Stress treatments in the Field
Irrigation water was applied with drip emitters, two per vine, positioned 30 cm from the vine trunk (32 L per vine twice a week for FI; Lopes et al., 2011). The water was supplied according to the crop evapotranspiration (ETc), corresponding to 100% ETc (in FI). ETc was estimated from ETo, using the crop coefficients (Kc) proposed by Allen et al. (1998). The two irrigation treatments were fully irrigated (FI) and non-irrigated (NI). Six plants in each treatment were pre-selected for yield analysis.
Samples for the quantification of abscisic acid (ABA), hydrogen peroxide (H 2 O 2 ), pigments, ascorbate and glutathione were taken on the 5 th August 2010, 18 th August 2011 and 23 rd August 2012, corresponding to the phenological state 35 (numeric scale of Eichhorn and Lorenz, 1977)

Meteorological measurements
Meteorological data from the area surrounding Pegões was supplied by the "Associação de Viticultores de Palmela" (AVIPE) in 2010 and in the following years was retrieved from Instituto Português do Mar e da Atmosfera (IPMA), at https://www.ipma.pt/pt/oclima/monitorizacao/.

Phenological analysis
The phenological state of the plants was analysed before the start of the irrigation and at regular intervals until harvest. The classification used was according to Baggiolini (1952), later transformed to the numeric scale of Eichhorn and Lorenz (1977) for an easier statistical comparison of the data.

Grapevine yield and vigor
At harvest, grape clusters were counted and weighted. Clusters were separated into clusters from count nodes and clusters from non-count nodes. The incidence of Botritis and sunburn was also quantified. Winter pruning was performed in December and all the pruned wood of each plant was weighted and separated into wood from the count and non-count nodes. The crop load calculation, in the form of the Ravaz index (Yield/Pruning Weight), where the yield from the current harvest is used against the pruning weight in the following dormant season, was also calculated.

Berry analysis
Once harvested and split into groups, determined by treatment and variety, the berries, with their petioles detached, were immediately taken to the laboratory to undergo berry analysis. Triplicate samples of 100 berries each were assessed for berry weight, must volume, pH, total acidity (g/L) and total solids (Brix Index). pH was measured by using a calibrated pH meter, with the must at room temperature. Total acidity was measured through the titratable method with sodium hydroxide, and expressed in terms of tartaric acid.

H 2 O 2 quantification
H 2 O 2 production was detected using a fluorometric horseradish peroxidase (HRP) linked assay (Amplex Red assay kit, Invitrogen). Leaf material (0.1 g) was collected at each time point and ground over activated charcoal in the presence of liquid N 2 as described by Creissen et al. (1999). Samples were centrifuged 10 min at maximum speed and the supernatants were kept on ice until measurements. H 2 O 2 concentrations in purified extracts were determined according to the manufacturer's instructions. Absorbance was then measured with a microplate reader at 570 nm. H 2 O 2 concentrations were expressed in µmol/g fresh weight.

Abscisic acid
The extracts for ABA quantification were carried out as described by Vilela et al. (2007). ABA was quantified through immunoassay by indirect enzymelinked immunosorbent assay (ELISA) with monoclonal antibodies, using a commercial kit (Olchemim Enzyme Immunoassay, Olomouc, Czech Republic), according to the manufacturers recommendations.

Antioxidative metabolite quantification
Reduced and oxidized glutathione and ascorbate concentrations were determined in leaves collected at the end of each stress treatment. Leaf material (0.5 g) was frozen in liquid N 2 . Each sample was homogenised in 5 mL of ice-cold 6% metaphosphoric acid (pH 2.8), containing 1 mM EDTA, in the presence of liquid N 2 . Homogenates were centrifuged at 27 000 g for 15 min at 4 ºC and the resulting acid extract was stored at -80 ºC.
Reduced (GSH) and oxidised (GSSG) glutathione were analysed colorimetrically by the 2-vinylpiridine method described by Anderson et al. (1995). GSH and GSSG concentrations were expressed in µmol/g fresh weight. Percentage of reduction corresponds to the percentage of total glutathione pool present as GSH and is defined as GSH/ (GSH + GSSG) x 100.
Ascorbic (AsA) and dehydroascorbic (DAsA) acids were assayed using a method adapted from Okamura (1980). To determine AsA and total ascorbate, 125 µL of the acid extract was neutralized with 25 µL of 1.5 M triethanolamine. After thorough mixing, 150 µL of 150 mM sodium phosphate buffer pH 7.4 were added. For the quantification of total ascorbate 75 µL of 10 mM DTT were added. This was followed by 15 min incubation at 25 ºC to reduce the DAsA present in the extract. To remove excess DTT, 75 µL of 0.5% (w/v) N-ethylmaleimide were added. The samples were then mixed and incubated 30 s at 25 ºC. For the quantification of AsA, water was added instead, so that the volumes of both samples were equal. To both samples the following reagents were added successively: 300 µL of 10% (w/v) trichloroacetic acid, 300 µL of 44% (v/v) phosphoric acid, 300 µL of 4% (w/v) 2,2'-dipyridyl in 70% ethanol and 150 µL of 3% (w/v) FeCl 3 . After mixing, the samples were incubated for 1 h at 37 ºC. Absorbance was recorded at 525 nm. The concentration of DAsA was calculated by subtracting the AsA concentration measured from the total ascorbate determined. Standard curves of AsA in the range of 10-60 µM were prepared in 5% metaphosphoric acid.

Statistical analysis
To study the influence of the water regime on indicators of production, for each variety and each season a completely randomized design with two replicates (each with three plants) was adopted. For data analysis the mean of the three plants of each plot was considered and the model included the effects of the water regime factor (two levels, FI and NI), the effects of the season factor (three levels, 2010, 2011 and 2012) and the respective interaction. ANOVAs were performed to assess the effect of the water regime and the seasons and all analyses were done separately per variety. The effects of the season, water regime and respective interaction were considered significant when the P-value of the test to the respective effects was lower than 0.05. Additionally, a Tukey test was performed to compare the means of each combination season/water regime and statistically significant differences were accepted for a p-value lower than 0.05. For several pairs of possible agronomically-related parameters, Pearson correlations were also performed. ANOVAs, the posthoc tests and the Pearson correlations were made in R (version 2.15.1 Copyright (C) 2012 The R Foundation for Statistical Computing).

Weather conditions and soil water status in the three seasons
In 2010 there was abundant rainfall in the winter and it rained until June and the maximum summer temperatures were the highest of the three seasons (33 ºC in August, Figure 1). The season of 2011, although without rain from June to August, had high levels of rainfall in the spring (Figure 1), allowing the soil water profile to be fully refilled in May ( Figure 2) and also had the mildest summer temperatures (average maximum temperature from July to September was 28 ºC, Figure 1). In fact, maximum temperatures in July and August of 2011 were below the 30 year average. The season with the lowest amount of rainfall was 2012, with 137 mm of rainfall until June, less than half the quantity of the two previous years. Comparing these data with the 30 year average, it is possible to note that, in fact, rainfall in 2012 was below the average expected for the area. Monitoring the soil water content through the quantification of the Fraction of Transpirable Soil Water (FTSW) in the 0-1.0 m soil profile during the growth season allowed to clearly separate the well watered treatment (FI) from the non-irrigated treatment (NI) in the field (Figure 2) in 2010 and 2012.
In a climate with Mediterranean characteristics, although with Atlantic influence, the most prominent individual abiotic stress in non-irrigated crops is drought. This was the case in the three studied seasons, particularly from June to August, where rainfall was almost absent. The maximum temperatures did not exceed 33ºC (July 2010) and light intensity reaching the canopy was, in average, 2500 µmol quanta m -2 s -1 . The three seasons monitored were, overall, typical representatives of the Mediterranean weather, with some specificities that made each unique, and thus the majority of the parameters quantified was more significantly affected by the season than by the irrigation treatments. In these three seasons it was possible to verify the different strategies of response to abiotic stress already analyzed in previous greenhouse experiments with the two varieties under study, namely a rapid and efficient reaction of TN, that was able to boost the buffering capacity of the cell's redox pool upon heat stress while TR was more intensely affected by stress for a longer period of time (Carvalho et al., 2015).   6 Furthermore TN showed a greater ability to withstand a combination of two or three abiotic stresses simultaneously when compared to TR .

Yield and vigor
The duration of the main phenological stages was unaffected by the irrigation regimes during the three growing seasons (Figure 3). Yield, however, experienced changes with the water regime, TR showing a significantly larger range of yield and number of clusters than TN, varying between 0.38 and 7.43 kg per vine and 4.7 and 19.7 clusters per vine (Table I). The parameters measured were yield per vine, number of clusters per vine, incidence of sunburn and of Botrytis, percentage of clusters in non count nodes, harvest weight, percentage of wood from non count nodes, the Ravaz Index. Significance levels of the factors in the ANOVA: *** 0.001; ** 0.01; * 0.05; "ns" not significant. Statistically significant differences after Tukey's multiple comparison tests for a p value lower than 0.05 are the following: * indicates significant differences between irrigation treatments in the same season; lower case letters indicate significant differences between seasons in the same irrigation treatment.
TR was more affected by season and water regime (p<0.001), with higher yield in FI plants in 2010 and 2012, the seasons in which soil water availability was lowest in NI. In 2011 the differences observed between FI and NI were not significant in both varieties. In TN the major factor influencing yield and number of clusters was the season (p<0.001), and there were no significant differences between irrigated and non irrigated vines. TN consistently produced more clusters than TR, a reflection of its higher production capacity, a well documented intrinsic varietal characteristic (IVV, 2011). The percentage of sunburn and Botrytis incidence were affected by the environmental conditions (p<0.01). The highest percentage of Botrytis incidence occurred in 2011, the season with lower maximum temperatures and higher humidity (Table I). Sunburn incidence affected TN in the drier warmer years while TR was mostly affected in 2011 (Table I) Typically, TN has lower bunch weight than TR, with smaller berries, although, in general its production capacity was higher and more regular (IVV, 2011). In fact TR is known for its low basal fertility and high vegetative vigor. Berry weight and volume was, in average, higher in TR through the three seasons and the number of clusters was higher in TN in both irrigation treatments. Pruning weight, however, was only higher in TR in 2010, a season with plentiful water and mild temperatures in the winter.
Contrary to what happens in roots, leaf growth is usually significantly impaired under drought conditions, due to a rapid decrease in the extensibility of expanding leaf cell walls (Hsiao and Xu, 2000). This drought-response process is ABA-mediated as the upkeep of leaf expansion is dependent on the presence of endogenous ABA, regardless of water status (Sharp et al., 1994). The increase of ABA content verified in 2012 in TR-NI was not accompanied by changes in pruning weight; however, the Ravaz index indicated excessive vegetative vigor. These differences in results obtained in the chemical signaling in response to water deficit may be caused by differences in the susceptibility of the varieties to water stress Rocheta et al., 2016) but environmental factors are also likely to intervene (Dodd, 2009;Romero et al., 2012), such as sporadic summer showers, temperature and evaporative demand in the area and the type of soil.
When irrigation is in excess the number of viable buds and their fruitfulness is reduced in Thompson seedless grapevines (Williams et al., 2009). In the current study, even though irrigation was set to compensate 100% ETc, cluster number was not affected by the irrigation regime in both varieties. Berry size (volume and weight) were most affected by irrigation in all seasons, showing that a late imposition of water deficit significantly affects berry size, possibly through a reduction of cell expansion.
Reports usually indicate that an earlier water deficit (post flowering water-stress) is more effective in influencing this berry characteristic, though the reduction of cell number as reported in berries of Syrah (Mccarthy, 1997). In Cabernet Franc, early water stress increased anthocyanin and phenolics content at harvest (Matthews and Anderson, 1988) and in Bobal, this increase was accompanied by a decrease of berry size (Salón et al., 2005). These decreases of berry weight and volume were more significant in TR, reaching 50% of the FI values in 2010, while TN-NI showed decreases typical of deficit irrigated vines (circa 80% the FI values) as the ones obtained by Oliveira et al. (2013)   The characteristics measured were berry volume and weight, brix index, pH and total acidity. Significance levels of the factors in the ANOVA: *** 0.001; ** 0.01; * 0.05; "ns" not significant. Statistically significant differences after Tukey's multiple comparison tests for a p value lower than 0.05 are the following: * indicates significant differences between irrigation treatments in the same season; lower case letters indicate significant differences between seasons in the same irrigation treatment.
9 (Conradie et al., 2002;Petrie and Sadras, 2008;de Cortázar-Atauri et al., 2009;Ramos et al., 2015). In Portugal, strong correlations between budburst and winter temperatures were found (Fraga et al., 2016). In the current work, a significant increase in the number of clusters also accompanied the anticipation of phenology, probably as a result of the abundance of water and mild spring temperatures of the previous season (2011).

Oxidative stress, antioxidative defence system and pigments
Plant water status was evaluated through the assessment of the pre-dawn leaf water potential (ψ pd ) at the moment of sampling for pigments, H 2 O 2 and status of the antioxidative defence system (Table III). Sampling was done on the phenological stage 35 (numeric scale of Eichhorn and Lorenz, 1977) and irrigated plants showed ψ pd values around -0.2 MPa in the three seasons, indicating a similar water supply throughout the experiment. In the non-irrigated treatments, 2012 was the season with the highest water deficit (ψ pd = -1.17 MPa in TN-NI) while in the previous seasons values were similar and ranging from -0.7 to -0.8 MPa, values consistent with severe water stress.

Table III
Effect of season (2010 to 2012) and irrigation treatment (FI/NI) on pre dawn leaf water potential (ψpd), Reduced (AsA) and oxidized (DAsA) ascorbate, reduced (GSH) and oxidized (GSSG) glutathione, abscisic acid (ABA) and hydrogen peroxide (H2O2) concentrations, and percentage reduction of ascorbate and of glutathione, in leaves of Touriga Nacional (TN) and Trincadeira (TR) Each value is the mean of four independent samples measured in triplicate (n=4). Significance levels of the factors in the ANOVA: *** 0.001; ** 0.01; * 0.05; "ns" not significant. Statistically significant differences after Tukey's multiple comparison tests for a p value lower than 0.05 are the following: * indicates significant differences between irrigation treatments in the same season; lower case letters indicate significant differences between seasons in the same irrigation treatment.
H 2 O 2 concentration in TN was significantly lower in FI-2010 and 2011 and increased from 2010 to 2012 in both irrigation regimes (Table III). In TR, H 2 O 2 concentration changed with the season in NI (p<0.1), with the highest values in 2010 and the lowest in 2011. ABA concentration in the leaves (Table III) was not significantly affected by the water regime in TN, only by the season (p<0.01) in NI, decreasing from 2010 to 2012. In TR, irrigation was the main factor of variation (p<0.01) of ABA contents, with NI showing significantly higher values than FI in 2011 and 2012.
The operational status of the mechanisms of defence against oxidative stress was assessed through the quantification of the oxidized and reduced forms of ascorbate and glutathione and of photosynthetic and protective pigments (Table III). In TN, AsA was affected by season (p<0.001) and not by irrigation, a trend similar to that of DAsA, GSSG and of the percentage reduction of glutathione (%redGSH).
There was no influence of the factors studied in the concentration of GSH and in the percentage reduction of ascorbate ( In both varieties, the content of all pigments was significantly affected by season (p<0.001) while the water regime did not affect anthocyanins and chla/chlb ratio in TN and in TR chlorophylls and anthocyanins were unaffected (Table IV). In TR-210 and TN-2012 chla/chlb ratio was higher in NI while in TN-2011 it was higher in FI.
Oxidative stress was highest in the most water stressed season (2012) in TR, with a concomitant response of the antioxidative response system, with repercussions in the shift of ascorbate to the oxidized form, as opposed to its unchanged status in TN, which showed high % reduction of both ascorbate and glutathione. These intrinsic differences between varieties, that are evident in imposed abiotic stress conditions , were not as striking with season variation although TN invested in increasing total glutathione and its redox status while TR invested in increasing its levels of ascorbate, especially in 2012. Also, photosynthetic pigments increased in the seasons of 2010 and 2012 in the NI treatments, indicating an attempt to optimize photosynthetic ability in an environment that favors stomatal closure.

Correlations between winter meteorological data and the following year's yield and between agronomically relevant quantifications
To investigate whether the environmental conditions, namely precipitation and temperature during the winter months, affected the following year's production, and if summer irrigation was able to overcome eventual production impairments, we performed correlations between meteorological data and yield. There were significant positive correlations between the maximum winter temperatures and the number of clusters in both varieties while the correlation between the accumulated precipitation in the same period and the number of clusters was significantly negative (Figure 4).  Pearson correlations between several production indicators and environmental parameters were performed in order to assess the influence of the environment on the production/quality of both varieties (Table V). It is interesting to notice that in TR ABA content was significantly correlated with ψ pd and with FTSW at the time of ABA quantification while in TN it was not. Also, in TN harvest weight was negatively correlated with FTSW while in TR there was no correlation between soil water content during the season and wood production but there was a positive correlation between winter rainfall and wood production. Expectably, in both varieties berry weight and volume were significantly influenced by water availability in the soil and by ψ pd , with TR showing larger and heavier berries per unit of water available. Each value is the mean of four independent samples measured in triplicate (n=4). Significance levels of the factors in the ANOVA: *** 0.001; ** 0.01; * 0.05; "ns" not significant. Statistically significant differences after Tukey's multiple comparison tests for a p value lower than 0.05 are the following: * indicates significant differences between irrigation treatments in the same season; lower case letters indicate significant differences between seasons in the same irrigation treatment.

Table V
Agronomically/physiologically relevant Pearson correlations between several indicators of production and/or stress and environmental conditions and the respective p-values in the three seasons (2010 to 2012) and two irrigation treatments (FI and NI) in Touriga Nacional (TN) and Trincadeira (TR) Indicators of production and/or stress are: yield per vine, number of clusters per vine, incidence of sunburn, percentage of clusters in non count nodes, harvest weight, Ravaz Index, berry volume and weight, abscisic acid (ABA), FTSW-Fen 35 (Fraction of Transpirable Soil Water in the phenological stage 35 of the numeric scale of Eichhorn and Lorenz, 1977), FTSW-Harv (Fraction of Transpirable Soil Water at harvest). Indicators of the environmental conditions are: predawn leaf water potential (ψ pd ), Winter temperature (average temperature between December and March of the previous season), Winter rainfall (accumulated precipitation between December and March of the previous season).

CONCLUSIONS
In TN, yield was significantly affected by environmental conditions during winter, especially in 2012. Summer irrigation was able to increase berry weight but was unable to increase the number of clusters and berries, as they had already been established well before irrigation started. The irrigation treatments affected berry size and volume significantly, while other berry characteristics changed with the severity of the season, but overall this variety can withstand lack of irrigation without decreasing quality, as already observed in the region of 'Dão' (Gouveia et al., 2012).
Overall, larger differences in plant behavior were obtained between seasons rather than among treatments within a variety and season. Such differences are due to the specific characteristics of each variety. Thus, when analyzing the effects of irrigation practices in field conditions, multi-year studies should be undertaken (Intrigliolo and Castel, 2010) and varieties should be carefully compared with one another in order to understand which are better suited for each specific environmental conditions.