Soil characteristics and soil water storage capacity
The soils of the analysed plots have clay loam textures, with 30.7 and 36% of sand, a clay content that ranges between 26.4 and 19.6% and a percentage of coarse elements (gravels) that ranged between 17.6 and 10.4%, respectively, in plots P1 and P2. The organic matter contents were relatively low (between 1.5 and 1.8%). Based on this information, the soil water content at field capacity ranged between 28 and 31% while that corresponding to the wilting point was 12 and 15%, respectively. Taking into account this information, the average ASW was estimated. It was observed that the ASW varied along the growing cycle, reaching very low values (< 5% of the SWC) at the end of the growing cycle (corresponding to the period veraison to maturity (Figure SM2).
Phenology variability during the period analysed and its relationship with the weather conditions
Vine phenology referring the stages H (flowers separated), M (veraison) and harvest presented high variability between years. The dates at which the stage H was reached varied between May 17th and June 5th (average: May 25th ± 5 days); veraison dates ranged between August 2nd and 31st (average: Aug. 19th ± 6 days) and harvest dates varied between September 1st and October 12th (average: Sep. 30th ± 9 days). There were no significant differences between both plots within a given year. The major differences were observed in the harvest date under the most extreme conditions, as it was the case of the years 2008 or 2017, with earlier harvest (about 10 days earlier in plot P2 than in plot P1). An advance trend was observed for veraison and harvest (0.93 and 0.83 days per year, respectively).
The observed trends in the phenological dates were mainly driven by the differences in temperatures recorded in the previous period. As it was indicated before, there was high variability in the dates at which the stage H was reached, and although there was a slightly advance with increasing Tmin values in the previous period, the relationship was not significant. The date at which veraison was reached was affected by temperature recorded in the period between bloom and veraison (with an advance of about 2.7 days per an increase of 1 °C in the mean temperature recorded in that period), but there was also an accumulated effect of the temperatures recorded before bloom (as denotes the correlation with the temperatures recorded in that period). Similarly, maturity was also affected by temperatures recorded in the previous period (between veraison and maturity) with an advance of about 3.65 days per an increase of 1 °C in the average temperature recorded in that period, but it was also affected by the temperatures recorded in the previous periods (Table SM1). Despite the fact that later phenological dates were observed in the wettest years, which was in agreement with the sign found in the correlations with ASW in different periods (Table SM1), its effect on the phenological timing was smaller than that of the temperature. The wettest years recorded also lower temperatures and it seemed that the temperature was the main driver of the variability in the phenological timing.
In addition to the previous analysis and in order to deepen the effect of temperature on phenology of this variety, the thermal requirements to reach each phenological stage were analysed. The accumulated 100 chill units, considered the threshold to start to accumulate heat units, was reached on average between the 15th and 25th March, which was in agreement with results in other zones in the Rioja DOCa and in other close viticultural areas in Spain (Ramos and Jones 2018; Ramos and Martínez de Toda 2020a). Based on this information, heat (GDD) was accumulated from March 15th and the base temperature for each stage was then estimated. The base temperatures were 7.4, 6.3 and 5.2 °C, for flowers separated, veraison and maturity, respectively, and taking into account those temperatures, the average GDD needed to reach the corresponding stages was 1350, 1690 and 2480 GDD, respectively. The goodness of the fit was analysed using the root mean square error (RMSE) and d-Willmott index values (RMSE: 6.82, 5.16 and 8.71; d-Willmott index: 0.57, 0.87, 0.68, for flowers separated, veraison and maturity, respectively), which according to the criteria given by Moriasi et al. (2007) indicated moderate to good agreement between simulated and observed dates. The average GDD values were used to project the changes in the phenological dates under warmer scenarios.
Grape composition variability during the period analysed
During the period of analysis, there were differences in the grape composition from year to year and also some differences between both plots, although the values followed similar trend in both plots. Grape composition at harvest in each plot and in each analysed year is shown in Table 1. The sugar content at harvest was slightly greater in plot P2 than in plot P1 while the average berry weight was higher in plot P1 than in plot P2. The titratable acidity presented high variability among years, with slightly greater values in P1 than in P2. Higher differences existed in the malic acid concentration between both plots. The anthocyanin concentrations, TPI and CI at maturity also showed high variability among years, with higher values in the dry and warm years (such as 2009, 2011, 2012 or 2017) than in years with other climatic characteristics, but with higher values in plot P2 than in plot P1. The differences in acidity, sugar content and anthocyanin concentrations were coherent with the different berry weight found between the plots, which could be related to the differences in the available water.
The analysis of the relationships between grape composition and climate variables obtained with the factor analysis after the varimax rotation is shown in Table 2. Four factors were retained, which explained 85% of the variance. It can be observed that the highest weight in the first factor, which explained 37.35% of the variances, was for the variables related to the phenolic components, with positive sing, and for the berry weight and malic acid, with negative signs. However, the coefficients for the variables related to temperature and precipitation were very small. The highest values in the second factor, which explained 18.18% of the variance, corresponded to berry weight, titratable acidity and malic acid and the variables that represented the ASW in the periods SF-V and V-H, being all of them positive. The third factor, which explained 16.27% of the variance, was mainly driven by the temperatures recorded in the ripening period, and the highest values in the fourth factor, which explained 13.75% of the variance, were related to titratable acidity (negative values) and temperatures referred to the period SF-V, which had positive sign.
The additional stepwise forward regression analysis done considering these variables showed that both acidity and anthocyanins were related to the ASW (Table SM1). This result agrees with that observed for acidity in the PCA (factor 2), but the effect on anthocyanins seemed to linked to the dilution effect that produces the higher berry weight associated to the higher water availability. However, the effect of temperature on titratable acidity found in factor 4 (which was the one that explained lower percentage of the variance) seemed to be hidden by the effect of the water effect. The ASW in the period V-H gave significant fits for AcT, while malic acid and anthocyanins were related to ASW in the period FFS-V. For AcT and AcM, the correlation coefficient was positive, while for AntT, the coefficient was negative. Nevertheless, the explained variance was quite small 13, 29 and 11%, for AcT, AcM and AntT, respectively).
Projected changes in climatic variables and their impacts
Projected changes in temperature and precipitation
The changes in monthly maximum and minimum temperature and precipitation, projected using the ensemble of models for 2050 and 2070 under the scenarios analysed, are shown in Table SM2. During the months corresponding to the growing season (April–October), the increase in the average Tmax under the scenario RCP4.5 could be of about 1.4 °C for 2050 and 1.8 °C for 2070, and under the RCP8.5 scenario could be near 2 and higher than 3 °C, for 2050 and 2070, respectively. The projected increase in Tmin during the growing season by 2050 could range between near 1 °C under the RCP4.5 scenario and 1.5 °C under the RCP8.5 scenario, and it could be of up to 2.5 °C by 2070. A decrease in precipitation is projected under both scenarios, which regarding the growing season could range between 10 and 16% by 2050, respectively under the RCP4.5 and RCP8.5 scenarios, and up to 30% by 2070, under the warmest scenario. This means that the growing season precipitation, which is already scarce at present, will be quite small, leading to greater water deficits.
Projected changes in vine phenology for Carignan in the study area
Based on the thermal requirements observed during the period 2008–2020 to reach each phenological stage (GDD values) and the projected changes in temperature, an advance in the phenological timing was projected (Table 3). Under both scenarios, the projected advance appears to be greater for maturity and veraison than for flowers separated. For flowers separated, the advance by 2050 could vary between 4 and 7 days respectively under the RCP4.5 and RCP8.5 scenarios, while veraison is projected to be advanced 6 and 12 days, and maturity could be advanced between 11 and 17 days, respectively, under the same scenarios. By 2070 and under the warmer scenario, the advance could be near double.
Projected changes in grape composition
Taking into account the relationship between grape composition and the climatic variables that presented significant fits and the projected changes in temperature and precipitation under different emission scenarios, change projections in grape composition were made. The projected changes in berry weight, acidity and phenolic compounds are shown in Table 3. Berry weight could suffer reductions of between 26 and 33 g/100 berries by 2050 under the RCP4.5 and RCP8.5 scenarios, respectively, which represent a reduction of about 12.9 to 16.7%, depending on the scenario. That projected change, nevertheless, is smaller than the variability observed during the period analysed. Based on the reduction on precipitation along the growing cycle and the observed relationship between acidity and water availability, titratable acidity may suffer reductions between 0.26 and 0.34 g/L while malic acid concentration could undergo r very small reductions (between 0.1 and 0.14 g/L by 2050, and between 0.14 and 0.28 g/L by 2070 respectively under the RCP4.5 and RCP8.5 scenarios). The concentration of anthocyanins could, however, increase slightly due to increasing water deficits (in about 3.7 and 5.7 mg/L by 2050 under the RCP4.5 and RCP8.4 emission scenario, respectively.