Non-linear loss of suitable wine regions over Europe in response to increasing global warming

Abstract

Evaluating the potential climatic suitability for premium wine production is crucial for adaptation planning in Europe. While new wine regions may emerge out of the traditional boundaries, most of the present-day renowned winemaking regions may be threatened by climate change. Here, we analyse the future evolution of the geography of wine production over Europe, through the definition of a novel climatic suitability indicator, which is calculated over the projected grapevine phenological phases to account for their possible contractions under global warming. Our approach consists in coupling six different de-biased downscaled climate projections under two different scenarios of global warming with four phenological models for different grapevine varieties. The resulting suitability indicator is based on fuzzy logic and is calculated over three main components measuring (i) the timing of the fruit physiological maturity, (ii) the risk of water stress and (iii) the risk of pests and diseases. The results demonstrate that the level of global warming largely determines the distribution of future wine regions. For a global temperature increase limited to 2°C above the pre-industrial level, the suitable areas over the traditional regions are reduced by about 4%/°C rise, while for higher levels of global warming, the rate of this loss increases up to 17%/°C. This is compensated by a gradual emergence of new wine regions out of the traditional boundaries. Moreover, we show that reallocating better-suited grapevine varieties to warmer conditions may be a viable adaptation measure to cope with the projected suitability loss over the traditional regions. However, the effectiveness of this strategy appears to decrease as the level of global warming increases. Overall, these findings suggest the existence of a safe limit below 2°C of global warming for the European winemaking sector, while adaptation might become far more challenging beyond this threshold.

Keywords: Vitis vinifera L.; adaptation to climate change; climate change; general circulation model; phenological model.

FIGURE 1

Schematic of the methodology used for (a) the calculation of S _M; (b) the calculation of S _P; and (c) the calculation of S _Z. Following their definitions given in Section 2, the three indexes are plotted against (a) the day of maturity, (b) the dryness index (DI) (Equation 3) and (c) the hydrothermic index (Equation 4). The green regions indicate the conditions of full suitability, which are bounded by specific limits. For S _M, these limits coincide with the first day for which the following 30‐days mean temperature is below 22°C (T _OPT1) and the first day for which the following 30‐days mean temperature is below 15°C (T _OPT2). For S _P and S _Z the limits are based on those defined in Fraga et al. (2013).

FIGURE 2

Pattern of (a) the best suited grapevine variety and (b) the varietal diversity for the baseline period 1980–2009. The dark green contours in the upper panel identify the simulated suitable region for wine production over the baseline period, that is, the traditional region of wine production. Black contours mark the coastal outlines, including major estuaries. Light grey contours indicate the country's borders, included here for better geo‐referencing the suitability features. Map lines delineate study areas and do not necessarily depict accepted national boundaries.

FIGURE 3

Mean pattern of the best suited grapevine variety for: (left panels, i.e. (a) and (d)) the period 2010–2039; (middle panels, i.e. (b) and (e)) the period 2040–2069; (right panels, i.e. (c) and (f)) the period 2070–2099. Upper panels, i.e. (a), (b) and (c), are relative to climate projections under RCP4.5 scenario, while lower panels, i.e. (d), (e) and (f), are relative to climate projections under RCP8.5 scenario. Dark green contours identify the simulated suitable region for wine production over the baseline period. Black contours mark the coastal outlines, including major estuaries. Light grey contours indicate the country's borders, included here for better geo‐referencing the suitability features. Map lines delineate study areas and do not necessarily depict accepted national boundaries.

FIGURE 4

Mean pattern of the grapevine varietal diversity for: (left panels, i.e. (a) and (d)) the period 2010–2039; (middle panels, i.e. (b) and (e)) the period 2040–2069; (right panels, i.e. (c) and (f)) the period 2070–2099. Upper panels, i.e. (a), (b) and (c), are relative to climate projections under RCP4.5 scenario, while lower panels, i.e. (d), (e) and (f), are relative to climate projections under RCP8.5 scenario. Black contours mark the coastal outlines, including major estuaries. Light grey contours indicate the country's borders, included here for better geo‐referencing the suitability features. Map lines delineate study areas and do not necessarily depict accepted national boundaries.

FIGURE 5

Evolution of the absolute suitable area for wine production, that is, Vitis vinifera cluster, (a) within the traditional boundaries for wine production, that is, over the traditional wine regions identified over the baseline period; (b) outside the traditional boundaries for wine production; (c) over the whole of Europe.

FIGURE 6

Scatterplot of the simulated relative area loss for Vitis vinifera cluster over the traditional wine regions (%) versus the projected global temperature anomaly (°C) with respect to the pre‐industrial level. Circles, squares and triangles indicate the simulations performed with the CNRM‐CM5, the EC‐EARTH and the MPI‐ESM‐LR global models respectively. Empty and full symbols indicate dynamical downscaling with the CLMcom and SMHI regional models respectively. Violet symbols indicate the coupling with linear/non‐sequential phenological model; orange symbols indicate the coupling with linear/sequential phenological model; cyan symbols indicate the coupling with non‐linear/non‐sequential phenological model; and green symbols indicate the coupling with non‐linear/sequential phenological model. Solid black lines indicate the mean trends before and after the 2°C level of global warming. The latter has been evidenced through the dashed grey line.

FIGURE 7

The sources of uncertainty in (upper panels, i.e. (a) and (b)) the estimation of suitable area in the traditional regions and (lower panels, i.e. (c) and (d)) the relative loss over the traditional regions, for (left panels, i.e. (a) and (c)) RCP4.5 scenario and (right panels, i.e. (b) and (d) RCP8.5 scenario. The relative uncertainty of the three main components is expressed as the fraction of the approximated total variance ${\sigma }_{\mathrm{tot}}^{*}$ (Equation 6). Violet portions represent the relative uncertainty associated with Phenological Models; green portions represent the relative uncertainty associated with GCMs; and orange portions represent the relative uncertainty associated with RCMs.

FIGURE 8

Simulated percentage of area loss over the traditional wine regions for the five grapevine clusters, for (lower histograms) RCP4.5 and (upper histograms) RCP8.5 scenario. Green bars are relative to estimates for which cultivar turnover (i.e. grapevine variety replacement with a more climatically suited one) is included. Purple bars are relative to the estimates of the additional area loss when cultivar turnover is not included. Different shades of green and purple bars indicate different periods over which these estimates have been performed. These estimates regard the early varieties (first block), the early‐to‐mid varieties (second block), the middle range varieties (third block) and the mid‐to‐late varieties (fourth block). The estimates for late varieties are not included here because the difference between a scenario with varietal turnover and a scenario without varietal turnover would be null.