Interannual and spatial variability of grape composition in the Rioja DOCa show better resilience of cv. Graciano than cv. Tempranillo under a warming scenario

Weather conditions affect vine development and grape composition, although the response may be different depending on the variety and altitude. Under scenarios of climate change, the knowledge of the relationships between grape composition and climate is therefore important to know the suitability of a given cultivar. In this research, the variability of the grape composition of two red varieties with different phenological timings (Vitis vinifera L. cv. Tempranillo and Graciano) cultivated in Rioja DOCa (Spain) and the climatic variables that have a higher effect on the response of each of them were analysed. Grape composition of both cultivars at technical maturity (considered when a probable volumetric alcoholic degree (PVAD) = 13 o was reached) was analysed during the period 2008–2020 in areas located at different elevations and related to the weather conditions recorded in those areas. The results show the effect of temperature and water availability in different periods during the growing cycle on grape composition and the potential benefits of cultivating at a higher elevation, under higher water availability. The anthocyanins were affected by the maximum temperatures recorded in the period before veraison and during ripening (period veraison to technical maturity), decreasing their concentrations with increasing temperatures. In addition, higher water availability gave rise to an increase in acidity. Graciano is shown as a variety more suitable than Tempranillo to be cultivated under warmer conditions.


INTRODUCTION
Climate plays an important role in berry composition (Sadras et al., 2013;Webb et al., 2012), but under similar climatic conditions, each cultivar can present a different response as each variety is adapted to a given range of temperatures (Jones, 2012). Temperature is considered as one of the main drivers of the evolution of the growing cycle and the final maturity and berry composition (Sadras et al., 2007;Ovadia et al., 2013;Greer and Weedon, 2014). In addition, water availability is another factor influencing grape development and berry composition (Castellarin et al., 2007;Chalmers et al., 2010;Tramontini et al., 2013;Bonada et al., 2018;Pérez-Álvarez et al., 2021). Nevertheless, the response of each cultivar could be different due to their differences in the phenological timing, which means reaching maturation under different temperatures and different moisture conditions. High temperatures negatively affect titratable acidity (Buttrose et al., 1971;Barnuud et al., 2014), which in addition decrease by dilution when the berry size increases under wet conditions. Contrary, an increase in temperature give rise to an increase in sugar concentration (Coombe, 1987). The concentration of anthocyanins are also affected by temperature (Tarara et al., 2008;Greer and Weedon, 2014). Under high temperatures, lower accumulation and higher degradation can occur (Mori et al., 2007) and an unbalanced ratio between anthocyanins and sugars can take place (Martínez De Toda and Balda, 2015). This fact has been observed in varieites such as Cabernet franc, Shyraz or Maturana Tinta or Tempranillo (Sadras et al., 2013;Martínez De Toda and Balda, 2015;Martínez de Toda and Ramos, 2019). However, the effect could be different depending on the cultivar. In this respect, the comparative analy neyards. Its name comes from the Spanish 'temprano' meaning 'early' and it does in fact ripen quite early. It is very versatile from an oenological viewpoint being capable of producing wines that can withstand long ageing periods, with a good balance of alcohol content, colour, and an honest, smooth, fruity mouthfeel that turns velvety as it ages . With respect to agronomic performance, it sets well but is highly sensitive to pests and disease and not very resistant to drought or high temperatures.
The other red varieties approved in the area are Grenache, Mazuelo, Maturana Tinta and Graciano. The last one, Graciano, is an autochthonous variety in Rioja. In France, it is known as Morrastel and in Portugal as Tinta Miúda. It had been traditionally used as a complement for Tempranillo in the ageing process. That special contribution as a complement to Tempranillo has made it a very interesting variety for Rioja, where the area of cultivation has recovered in recent years, accounting for 1.8 % of the total vineyard surface area. It has therefore overcome the danger of disappearing that threatened it during the whole of the 20th century. It is fairly resistant to diseases such as downy mildew and powdery mildew. It is late-ripening, so it needs to be grown in sufficiently warm areas. It offers wines with a marked acidity and polyphenolic content, ideal for ageing, with a unique aroma that is much more intense than those of other varieties in Rioja.
Under the projected changes in climate, the growing cycle of the vines can suffer an advance of the phenology and a shortening of the growing cycle, and the varieties with earlier phenology could be more negatively affected. Previous studies on the Tempranillo (Ramos and Jones, 2018;Ramos and Martínez de Toda, 2020;Ramos et al., 2021) have pointed out that it is a variety that for its characteristics (earlier phenology and low acidity), could be negatively affected under warmer conditions. This suggests the need or adopting measures or looking for alternatives to maintain the suitability and quality of the wines, in particular for areas were the main cultivars could be threatened. Different studies are being carried out in the area exploring practices to mitigate the effect of climate change on the vines (Gutiérrez-Gamboa et al., 2021;Zheng et al., 2017). Other studies suggest the introduction of new plant material as an alternative to mitigate the effects (van Leeuwen et al., 2019). For that reason, exploring in more detail the behaviour or other varieties under similar climatic conditions could be important for the suitability of the area as the changes can be different depending on the variety. This research was planned under this hypothesis and its objective was to analyse the potentiality of one minority variety that is autochthonous in the region and that offers additional characteristics. The properties that offer Graciano vs. Tempranillo regarding its marked acidity and its later phenology could make this variety an optimal candidate for the area when temperature increases. To develop this objective, a comparative analysis of the effect of the weather conditions on phenology and grape composition for Graciano and Tempranillo was done in three zones located at a different elevation within the Rioja DOCa during the period 2008-2020.

Vineyard features
The research was conducted in vineyards planted with Vitis vinifera cv. Tempranillo and Graciano, in three zones, located at about 325, 430 and 565 m.a.s.l. in the municipalities of Aldeanueva de Ebro-Alfaro (Z-325), Fuenmayor-Logroño (Z-430) and Ocón (Z-565). The information used in this research referred to two vineyards in each zone ( Figure 1). The vines were planted between 1985 and 1996, following a pattern in agreement with the regulations established by the Consejo Regulador of Rioja (3000 vines/ha with an average pattern of 2.8 m × 1.2 m) and trained in trellis (double cordon). The vines were managed under rainfed conditions.
The dates of the phenological stages flowers separated, veraison and maturity were analysed for each area and variety during the period 2008-2020. The phenological survey for flowers separated and veraison was carried out in 10 plants in three different parts of each plot, at periodical intervals that ranged between four and seven days depending on the period analysed. The dates at which the vines in the survey plots reached 50 % of the stage (stages H and M according to Baillod and Baggiolini, 1993) were considered. The date of technical maturity was based on the date at which a given probable alcoholic degree (PVAD = 13 °), was reached. For grape composition, the sampling was carried out collecting berries in 40 randomly selected plants, at a rate of one cluster per plant and five berries per cluster from different parts (up, central and lower parts of the cluster).
The grape composition [berry weight (BW), total soluble solids, expressed as probable alcoholic degree (PVAD), titratable acidity (AcT), malic acid (AcM), anthocyanins (AntT), polyphenols (IPT) and colour intensity (CI)] were evaluated between veraison and maturity for each variety and location during the period 2008-2020. The values corresponding to the defined technical maturity was considered in the further analysis. All vine information was supplied by the Consejo Regulador of Rioja DOCa and the grape parameters were measured following the OIV official methods (OIV, 2004). The changes of each variable during ripening were analysed.

Climatic data and analysis
The weather conditions during the period of analysis were evaluated using daily data recorded in weather stations located in Alfaro, Logroño and Ausejo, which were near and at similar elevations (315, 450 and 550 m.a.s.l, respectively) than the analysed vineyards ( Figure 1). These weather stations belong to La Rioja Government. The meteorological information included daily maximum and minimum temperatures (Tmax and Tmin), precipitation (P) and potential evapotranspiration (ETO) estimated according to Penman-Monteith. Crop evapotranspiration was then estimated using the crop coefficients proposed by Allen and Pereira (2009), and the precipitation minus crop evapotranspiration (P-ETc) was considered as an index of water availability. The daily data were averaged for the growing season (GS: budbreak-harvest) and periods between phenological stages (BB-BL: budbreakbloom; BL-V: bloom-veraison; V-H: veraisonharvest). These variables were used to analyse the response the vine under different weather conditions.

Variability in the vine response and its relationship with climatic variability
To evaluate the relationship between grape composition and climate variability, a multivariate analysis was done using the average grape composition recorded for each variety at each location in each year. Among the multivariate techniques, a hierarchical cluster analysis was applied. This technique allows the classification of observations taking into account its similarity, and it has been applied for different purposes, such as the analysis of the similarity of the phenolic complex in berries in different varieties (Levchenko et al., 2021) or to differentiate between grape cultivars according to their phenolic contents and antioxidant properties (Sridhar and Charles, 2018), or the identification of vineyards in different locations which give rise to similar grape composition . The cluster analysis was applied using the Euclidean distance to measure the similarity between the objects and using two methods (Ward and Group average) to optimise the number of clusters to be retained. All the data were standardised. The number of clusters to be retained was defined by taking into account the agglomeration distance. A cut-off point was established when the distance between one step and the next was greater than twice the average distance. Once the number of the cluster was determined, the centroids were evaluated (average values of the grape parameters) and the characteristics of the years that gave rise to a similar response were analysed. In addition, the grape composition was related to the climate variables that showed differences between clusters to know the differences in the effect that temperature and water availability could have in grape composition for each cultivar.

Spatial and temporal variability in the weather conditions
The weather conditions recorded at the three zones of Rioja included in this research in the period 2008-2020, during the growing season (usually mid April-mid August), are summarised in Figure 2. The average TmaxGS during the period analysed varied between 21.8 and 25.3 °C in the coolest area (Z-565) and between 24.1 and 27.2 °C in the warmest area (Z-325), which means differences between years higher than 3 °C between the coolest and the warmest years, and also differences of the same order of magnitude between areas. For TminGS, the differences between zones were smaller than for TmaxGS, with TminGS ranging between 11.4 and 12.8 °C in the coolest area and between 11.8 and 13.6 °C in the warmest one. Regarding precipitation, high variability from year to year could be observed during the period analysed, with PGS (growing season precipitation) ranging between 126 and 440 mm, and with greater values in the zones Z-430 and Z-565 than in the zone Z-325 for most of the years analysed. Figure 3 shows a summary of the phenological dates (separated flowers, veraison and maturity) recorded at each location. Differences between years of up to 18 and 30 days in stage H, up to 33 and 28 days for stage M, and up to 39 and 40 days for maturity, were found respectively for Graciano and Tempranillo. In addition, some differences in the average phenological dates were observed between the three zones analysed for both varieties, with the earliest phenological dates in the area Z-325 and the latest in the zone Z-565. For maturity, considered as the date at which a given probable alcoholic degree (PVAD = 13 º) was reached, it can be observed the advance of that stage in the area located at a lower elevation, for both varieties, and also the earlier dates for Tempranillo compared to Graciano within each location.  to warm years, and cooler and wetter years, respectively. The differences in the phenological dates between both varieties were greater for stages M and for maturity rather than for flowering. In years like 2013, one of the coolest and wettest of the series, however, the differences in the harvest dates between both varieties were of 19 days in the zone Z-325 and about 7 days in the other two zones with harvest in mid-October in those zones while in Z-325 took place in the last third of September.

Grape composition variability
Regarding the grape composition, the mean values for each variety analysed at each zone are shown in Table 1. As it was aforehand mentioned, in the area located at a lower elevation, maturity was reached before, and the grapes were usually harvested with a higher probable alcoholic degree (higher than 14º), while in other areas that value was not reached in most of the analysed years. When the threshold PVAD = 13° was reached, higher AcT was observed for Graciano than for Tempranillo at each location, while it was opposite for AcM. The concentration of AntT, the TPI and the CI were also higher for Graciano than for Tempranillo, at each location while berry weight was higher for Tempranillo than for Graciano, although with high variability from year to year.
The relationship between some grape parameters, such as AcT, AcM and AntT and the probable alcoholic degree during ripening for both varieties and the three locations is shown in Figure 4.
It can be observed that the AcT, not only reached lower values in Tempranillo than in Graciano but they were reached faster. It can be observed, that the evolution of AcT with an increasing probable alcoholic degree for Tempranillo was smaller in the area located at lower elevation (Z-325) than in the other two zones, while for Graciano the differences were smaller, although the final AcT was much lower in that area where the probable alcoholic degree was higher (Figure 4a).
For AcM, as it was previously commented, the values were higher for Tempranillo than for Graciano. For Tempranillo, there was higher variability in the AcM values reached for a given probable alcoholic degree and lesser differences were observed from the beginning to the end of the ripening period.

For
Graciano, a greater concentration range was recorded, with a similar evolution in the three zones but reaching smaller values at harvest in the zone located at lower elevations as the grapes were harvested with a higher probable alcoholic degree (Figure 4b).
For AntT, the increase in AntT with an increasing probable alcoholic degree was similar in the three areas for Tempranillo and similar to Graciano at the lowest elevation, but for that variety, the increase was greater in the other two zones, where higher anthocyanin concentrations were recorded at harvest (Figure 4c).

Relationship between grape composition and weather conditions
The differences in grape composition and its relationship with the weather conditions between years were analysed with the establishment of groups of a similar response. The cluster analysis performed on the grape composition at technical maturity recorded at each location allowed for the establishment of those groups. The classification of the years into groups are presented in Figure 5 and the centroids and the characteristics of the years included in each group are shown in Tables 2  and 3, for Graciano and Tempranillo, respectively.
In the lowest elevation (Z-325), for Graciano, the years that showed the highest berry weight, the highest AcT and AcM, and the lowest PVAD at maturity were clustered in cluster C1 (2008-2013-2014). Those years were the coolest and wettest in the series. They had the smallest average temperature (both minimum and maximum temperatures) in all periods between phenological stages. On the contrary, the years included in cluster C4 (2011,2012,2016,2017) recorded the lowest berry weight, lower AcT and the lowest anthocyanin concentration and colour intensity. The years included in cluster C4 recorded the highest temperatures, reaching average maximum temperatures in the ripening period that were on average 4.4 °C higher than in the years included in cluster C1 and minimum temperatures 3 °C higher, on average. Regarding the other two clusters, there were some differences between them, with higher titratable acidity, but lower malic acid and lower concentration of AntT and TPI in cluster C2 (2009-2010-2019) than in cluster C3 (2015-2018-2020), being also berry weight smaller in cluster C2. The years included in those clusters recorded average growing season temperatures between 1.0 and 1.7 °C higher than years included in cluster C1, but with higher differences in the temperatures in periods between phenological stages. Tmax in the period BL-V and V-Mat, which were 1.3 and 2 °C, respectively, were higher than those recorded in years in cluster C1, and the Tmin in the same periods were 1 and 2.3 °C higher than in cluster C1. In addition, there were some differences in water availability, which was smaller in cluster C2 and C3 than in cluster C1, in particular in the period BL-V (Table 2).
For Tempranillo in the same area (Z-325), the warmest years were grouped in cluster C3 (2009-2012-2016-2017), quite similar to what was obtained for Graciano (cluster C4), but the cooler years were distributed into three clusters: C1 (2008-2014-2014-2015-2018-2019-2020), C2 (2013) and C4 (2010-2011). It can be marked that the lowest BW was recorded in cluster C4, which grouped the years in which the highest PVAD was reached at harvest, and those years recorded the highest concentration of AntT, and the highest TPI and CI. The lowest AntT, TPI and CI were reached in the years included in cluster C1, years in which the lowest PVAD was reached at harvest. The years included in the clusters C1 and C2 recorded the lowest maximum and minimum temperatures during all growing seasons and those of cluster C1-1 recorded also lower temperatures than those included in the other clusters, in two of the three periods considered in this analysis.     There were, however, differences of about 1 °C in the maximum temperatures between the subgroups of cluster C1. An additional difference between clusters was referred to as the water deficits. The cluster C2 presented the lowest water deficits during the whole growing season, with the highest water deficits in the period BL-V. The differences in water deficits between the other clusters were mainly recorded during the ripening period (Table 3).
For the other two zones, a similar influence of temperature and water availability on grape parameters (both acidity and phenolic composition) was observed. In zone Z-430, for Graciano, cluster C1 (2008Z-430, for Graciano, cluster C1 ( -2013 grouped the years with the highest AcT and AcM, while cluster C3 (2009-2012-2015-2017) grouped the years that recorded the lowest AcT, which were the ones that reached the highest PVAD and the highest concentration of AntT, TPI and CI. Regarding the characteristics of the years of each group, cluster C1 grouped the wettest and coolest years of the series, with average growing-season temperatures about 1.7 °C lower than the average in the series and up to 2.4 °C lower than the average of the warmest years, included in cluster C3. The years included in cluster C3 recorded particularly high temperatures (both maximum and minimum temperatures) in the period BL-V and also higher water deficits in that period than the rest of the years. Cluster C2 (2010-2011-2019-2020) grouped years with similar characteristics regarding temperatures to those of cluster C4 (2014-2016-2018), with average maximum temperatures in the period BL-V, which were about 1.3-1.4 °C higher than those corresponding to cluster C1, and about 1.5-1.4 °C lower than the ones corresponding to cluster C3 (Table 2).
For Tempranillo in zone Z-430, cluster C1 (2008-2010-2013-2015-2019) grouped years with the highest AcT and AcM, and concentrations of AntT, CI and TPI slightly higher than the years included in the other clusters. The years included in cluster C3 (2009-2011-2012-2014-2016-2017) recorded the lowest AcT and AcM and also a lower concentration of AntT compared to cluster C1. However, the lowest concentration of AntT was reached in the years included in cluster C2 (2018-2020), which were the ones with the lowest PVAD and the highest berry weight. Although the average growing season maximum temperatures of the years included in the three clusters were similar, there were differences in the average temperatures in some specific periods between phenological stages.
Thus, the years grouped in cluster C1 recorded lower temperatures during the ripening period than the years grouped in other clusters (about 2 °C lower than in cluster C2 and 2.1 °C lower than in cluster C3). The average minimum temperatures during ripening were also lower in cluster C1 (about 1.4 °C lower than in cluster C2 and cluster C3). In addition, the years included in cluster C3 recorded on average higher water deficits (Table 3).
In the area located at the highest elevation (Z-565), the cluster analysis gave rise to a year classification that differed in some way from the ones obtained in the other two zones. For Graciano, the years were classified into five clusters. Cluster C1 (2008-2013-2014) grouped the years with the highest AcT and AcM and BW. The years grouped in that cluster were the ones that recorded the lowest maximum temperatures during the whole growing season. The years included in cluster C3 (2019) and cluster C5 (2017) recorded slightly lower acidity than the years of cluster C1, but differed in the berry weight (being the lowest BW recorded in 2019) and in the concentration of AntT (higher AntT in cluster C3 than in C1 and C5), TPI (higher TPI in C5 than in C1 and in C3) and CI (lower CI in C3 than in C5 and in C1).
The year 2019 recorded the highest temperatures (both maximum and minimum) in the period BL-V, but lower values during ripening (similar to those recorded in the years groped in cluster C1, which were the coolest years). On the other side, the lowest AcT were recorded in the years included in cluster C2 (2010, 2018 and 2020) and cluster C4 (2009-2011-2012-2015-2016), but both clusters differed in the concentration of AntT, which was higher in years included in C2 than in cluster C4. The years grouped in these two clusters recorded higher temperatures than the ones grouped in cluster C1, but lower than in clusters C3 and C5, in particular during the period BL-V. In addition, water deficits were also higher in those years than in the years included in the other clusters (Table 2).
For Tempranillo at the highest elevation (Z-565), the year that recorded the highest acidity was separated in cluster C3 (2013), while the rest of the years that reached similar AcT values were classified into three clusters. Among them, cluster C2 (2010-2019-2020) recorded slightly lower AcT higher AcM and the highest AntT TPI and CI. The years in clusters C1 (2008-2009-2014-2018) and C4 (2011-2012-2015-2016-2017) showed intermediate values of all parameters.
The maximum and minimum temperatures of the year 2013 (cluster C3) were lower in almost all periods considered than in the rest of the years, and the climatic characteristics of the years included in the different clusters differed in the maximum and minimum temperatures in some specific periods. Thus, the years included in cluster C2 recorded the highest maximum and minimum temperatures in the period BL-V, but the lowest minimum temperatures during the ripening period, while the years included in cluster C4 recorded the highest maximum temperatures and high minimum temperatures in the ripening period. Water deficits in that area located at higher elevation were smaller than in other areas located at a lower elevation, being the years included in cluster C4 the ones with the highest values, in particular in the period BL-V (Table 3).

Phenology variability between cultivars under different weather conditions
Despite the high variability in the weather conditions recorded during the period analysed, the results allowed confirming differences between the zones located at a different elevation (between 325 and 565 m.a.s.l.) and extracting information about the vine response of the studied cultivars under a wide range of temperature and precipitation values.
The averages dates for stage H were quite similar for both varieties, with higher variability for Tempranillo than for Graciano ( Figure 2). Higher differences existed for veraison between both varieties, with earlier dates for Tempranillo than for Graciano, being the advance higher at the lowest elevation. The earlier veraison in that area implies that ripening occurs under very hot and dry conditions, and maturity was reached also earlier.
The average date at which the 13 º were reached, did not differ so much between both varieties but the variability was higher for Graciano than for Tempranillo and the advance was usually greater in the two zones located in Rioja Oriental (Z-325 and Z-565) than in the zone located in Rioja Alta.
These results confirmed the different responses in the zones with the Atlantic and Mediterranean influence and, despite the variability that existed from year to year, Graciano is shown as the variety that could maintain the delay at maturation, which seems to be higher at the highest elevation.
Under hot conditions, the differences in the response gave rise to the very earlier phenology of Tempranillo. A representative year of that situation was the year 2017. In that year harvest of Tempranillo took place in mid-August in Z-325 while for Graciano took place 14 days later. The same difference in the dates was found between both varieties in the other zones but harvesting taking place 14 days later. The differences in phenology between years with different temperatures agree with the average advances at the beginning of bloom, veraison and harvest for an increase of 1 °C (3.1. 5.2 and 7.4 days, respectively) given by Ruml et al. (2016) and results of other authors (Fraga et al., 2016;Ramos et al., 2015;van Leeuwen and Darriet, 2016;Ramos and Martínez de Toda, 2019), who linked phenological dates and temperature and indicated that ripening under warmer conditions and higher water deficits can affect the grape composition. It is important, however, to mark the differences in the response of both cultivars.

Grape composition variability
The results found during the period of analysis allowed confirming some differences in grape composition between both cultivars. For a given PVAD, the final titratable acidity values (AcT) reached in Graciano were higher than in Tempranillo, and the decrease of AcT was slowly for Graciano than for Tempranillo when the PVAD increased ( Figure 4a). AcM, however, reached lower values in Graciano than in Tempranillo. The AcT presented an increasing gradient with elevation. For Tempranillo, however, although AcM increased from the zone at the lowest (Z-325) to the highest (Z-565) elevation in Rioja Oriental, the maximum values were recorded in the zone located in Rioja Alta (Z-430), which was at an intermediate elevation. This could be justified by the effect of the higher temperatures in zone Z-325 recorded in the period BL-V. Sugiura et al. (2020) related acidity to temperatures in sensitive periods after flowering, and taking into account that the increase of AcM usually takes place just before veraison (Volschenk et al., 2006) the increase in the high temperatures in that period will give rise to lower final acid concentrations in the grapes. The differences in acidity found between the clusters and the average climatic values that characterised each of them indicated that higher maximum temperatures in the whole growing season give rise to a decrease in acidity and in particular the maximum temperature in the period veraison to maturity.
These results point in the same direction as those of Costa et al. (2020), who found strong correlations with the temperatures recorded in the months that included part of the ripening period, and agree with that pointed out by other authors regarding the effect of high temperatures on acidity (Coombe, 1987;Sadras et al., 2013;Keller, 2015;Barnuud et al., 2014;Vršič et al., 2014). The effect was slightly higher for AcT for Tempranillo than for Graciano in zones Z-430 (-33 vs. -0.14 g AcT/L per 1 ºC) and in Z-565 (-0.36 vs. 0.31 g AcT/L per 1ºC) and similar in Z-325 (-0.15 vs. 0.19 g AcT/L per 1 ºC). For AcM, the effect was negative and slightly greater for Tempranillo than for Graciano in the three zones: Z-325 (-0.26 vs. -0.22 g AcM/L per 1 ºC); Z-430 (-0.26 vs. 0.16 g AcM/L per 1 ºC) and Z-565 (0.29 vs. 0.21 g AcM/L per 1 ºC). These results mean that under warmer conditions, Tempranillo will have even lower acidity values and that, although Graciano could also decrease in acidity, its decrease will be smaller.
The average concentration of AntT was also higher in Graciano than in Tempranillo and higher in both zones of Rioja Oriental (Z-325 and Z-656) than in the zone located in Rioja Alta (Z-430), for both varieties. This confirms again the different responses of the vines in Rioja Alta and Rioja Oriental. However, in that case, higher values were found in zone Z-325 than in zone Z-565. The gradient in the levels of AntT between locations were in agreement with the sugar contents reached, which had the lowest values in zone Z-430 and the highest in Z-325, for both varieties.
The results were consistent with the smaller berry weight recorded in zone Z-325 and the highest in zone Z-565, which in addition was smaller for Graciano than for Tempranillo. Berry weight was mainly driven by water availability, as confirmed by the higher values recorded in wet years and it was also in agreement with the higher berry weight in the zones located at higher elevations where the index P-ETc was higher. The effect of temperature and water availability on the concentration of anthocyanins has been indicated by other authors in different varieties (Mori et al., 2007;Barnuud et al., 2014;Pastore et al., 2017;de Rosas et al., 2017;Shinomiya et al., 2015). Increasing temperatures could produce a decrease in the concentration of anthocyanins, but water deficits could favour its increase (Hochberg et al., 2015;Castellarin et al., 2007). In the study case, the effect of temperature and water deficits pointed out in the same direction for both varieties, although with differences between them and between zones.
The effect of increasing temperatures during ripening was significant and higher in zone Z-430 for Tempranillo than for Graciano (-36 vs. -21 mg/L AntT per 1 ºC increase in Tmax during ripening), while it was not significant for the other two zones. The effect of water availability was appreciable in the zone located at a higher elevation, indicating that in that area the concentration increased when the water deficit increased, and in higher proportion for Graciano than for Tempranillo (1.48 mg/L per 1 mm vs. 0.52 mg/L AntT per 1 mm (P-ETc).
Regarding the other two parameters related to phenolic composition, TPI and CI, there were also greater for Graciano than for Tempranillo and there was not a clear trend with elevation although, for both cultivars, the highest values were found in Z-565 and the lowest in Z-430. Nevertheless, the response differed between years and the relationship between those variables and the climatic variables presented some differences between zones and between both varieties. While for Graciano, TPI was positively influenced by maximum temperature during ripening in both zones located on Rioja Oriental (Z-325 and Z-565), its effect was not significant for that variety in zone Z-430. For Tempranillo, however, the effect was significant and opposite in zone Z-430 (located in Rioja Alta) and not significant in the other two zones. The effect of temperature on CI was only appreciable for Tempranillo and higher in the warmest zone (Z-325) than in the zone located at intermediate elevation (Z-430) (-1.8 vs. 0.65 colour units per 1 ºC increase) and not significant at the highest elevation.
In addition to the direct effects of the climatic conditions on acidity and on the phenolic composition, which could be different between both varieties, a significant effect on the decoupling between anthocyanins and sugar might be remarked. The observed ratio AnT/PVAD was smaller for Tempranillo than for Graciano in the three zones (average values for Graciano 43.5, 40.8 and 49.4, respectively, for the zones Z-325, Z-430 and Z-565, while for Tempranillo the averages were 36.7, 36.0 and 35,7, respectively). The accumulation of sugar in Tempranillo seemed to reach the maximum value more quickly than in Graciano (Figure 4), which could suppose that Graciano continued to accumulate anthocyanins at high temperatures while in Tempranillo the accumulation of anthocyanins stopped. This effect was also confirmed when the responses in the coolest and the warmest zones were compared.
It was also confirmed that an increase in Tmax during the ripening period would produce a decrease in the ratio, greater for Tempranillo than for Graciano (2.4 vs. 2.11 for 1 ºC increase), and greater at the lowest elevation, which means even higher decoupling between anthocyanins and sugars. Thus, the comparative analysis of the response of Graciano and Tempranillo in the analysed zones of the Rioja DOCa confirmed, that the impacts on acidity and on the decoupling between anthocyanins and sugars that are expected under warmer conditions, could be smaller for Graciano than for Tempranillo, which enhance the potentiality the minority variety.

CONCLUSIONS
The results of this research, carried out in the zones of the Rioja DOCa during a 13 year period, showed the high variability in the response of different well-adapted varieties cultivated in the area, due not only to the weather conditions but also to the location. The results confirmed the later phenological timing of Graciano vs. Tempranillo and between zones at different elevations. In addition, the grape composition regarding acidity and anthocyanins that offer Graciano compared to Tempranillo and the lower decoupling between anthocyanins and sugars that it suffers with the temperature in all analysed zones, show the potentiality of Graciano as an autochthonous variety to be considered under warmer scenarios. All the observed results represent significant treats under the climate change perspective that could be relevant for winemakers in the area.