X. Chone 1-2, C. Van Leeuwen 1 ' 2, P. Chery 2 and P. Ribereau-Gayon 1

Transcription

1 Terroir Influence on Water Status and Nitrogen Status of non-irrigated Cabernet Sauvignon (Vitis vinifera). Vegetative Development, Must and Wine Composition (Example of a Medoc Top Estate Vineyard, Saint Julien Area, Bordeaux, 1997) X. Chone 1-2, C. Van Leeuwen 1 ' 2, P. Chery 2 and P. Ribereau-Gayon 1 1) Faculte d'oenologie de Bordeaux, Universite Bordeaux 2 Victor Segalen, 351 Crs de la Liberation, Talence Cedex, France 2) ENITA de Bordeaux, 1 Crs du General de Gaulle, Gradignan Cedex, France Submitted for publication: March 2000 Accepted for publication: January 2001 Key words: Terroir, Nitrogen, water status, Vitis vinifera, Cabernet Sauvignon, vigour, anthocyanin, phenolic content, must nitrogen content, leaf water potential Terroir effect on vine vigour, must composition and wine quality was investigated for Cabernet Sauvignon during a rainy vintage in a top estate of the Medoc area (Bordeaux, France). Soil was the only variable in this survey. Vine water status was determined by means of leaf water potential. Mild water deficits were observed only on a gravelly soil with a shallow root zone. Vine nitrogen status was determined by total must nitrogen content. Nitrogen status varied from deficient to unlimited. Nitrogen deficiency reduced vine vigour to a greater extent than did mild water deficits. The smallest berries, as well as the highest phenolic content for both must and wine, were observed under nitrogen deficiency. Both early mild water deficits and a nitrogen deficiency throughout the growth period were demonstrated to have beneficial effects on the phenolic content of berries and on wine quality. Two combinations of vine water status and vine nitrogen status led to the most highly appreciated wines: a low nitrogen status without water deficits and a medium nitrogen status accompanied by mild water deficits. For any given vineyard terroir can be described as the combination of pedoclimatic conditions, vine components such as cultivar, clone and rootstock as well as viticultural techniques (pruning system, canopy management, etc.). Given this, terroir must be seen as a complex system with multiple variables: it is impossible to study all the variables in one experiment. Within a single estate (10 to 100 ha) in Bordeaux, there is generally no climatic variation but several types of soil. It is well known that the effect of the soil on vine behaviour is mediated through varying watercontent levels and their effects on vine water status (Seguin, 1970; Seguin, 1986; Van Leeuwen & Seguin, 1994). Water deficits occur when transpiration exceeds the ability of the root system to supply water to the transpiring leaves. Mild water deficits are known to have positive effects on reducing berry size (Smart, 1974) and on berry skin anthocyanin and tannin content in red grape varieties (Matthews & Anderson, 1988; Van Leeuwen & Seguin, 1994; Koundouras et al, 1999). Schultz and Matthews (1993) observed that in potted vines water deficits caused embolism in the xylem shoot apex, which stopped shoot growth. This has been confirmed in field studies (Matthews et al, 1987; Van Leeuwen & Seguin, 1994; Naor & Wample, 1996). Given this consideration, it can be said that under mild water deficits vegetative growth is no longer in competition with reproductive development as a sink of photosynthesis resources. Hence fruits are primary sinks. This can partly explain the richer must and wine constitution obtained from vines having undergone mild water deficits. Without the addition of nitrogen fertiliser, vine nitrogen status depends upon soil organic matter content, its mineralisation rate and the C/N ratio. It has been shown for Merlot (Delas et al, 1991) and for Thompson seedless and White Riesling (Kliewer and Cook, 1971; Spayd et al, 1994) that nitrogen fertilisation increases vigour. In the Bordeaux oceanic climate rainfall can vary to a considerable extent from one season to another. In a dry season, such as 1990, high enological berry potential and therefore wine quality are strongly linked to mild water deficits (Van Leeuwen and Seguin, 1994). In a rainy season, this is less likely to occur even though a soil effect on wine quality is generally admitted to exist despite the rainy conditions. As a better understanding of the terroir effect can help growers in their vineyard management and grape selection, the effect of soil variation on vine vigour, berry constitution and wine quality was investigated in this survey during a rainy vintage. Vine water status and vine nitrogen status were measured by means of physiological indicators. In the case studied soil was the only variable differentiating blocks planted with the same cultivar and rootstock and with the same vine spacing and subject to the same canopy management. Conradie (1986) demonstrated that vine nitrogen uptake from bud break through fruit maturity is progressively less utilised for vegetative growth and increasingly more utilised by the fruit. This was confirmed by Soyer et al (1995), who showed that vine berry nitrogen content closely reflected mineral nitrogen availability in the soil. Thus, berry total nitrogen content can be used as a physiological indicator of vine nitrogen status. However, the levels of total nitrogen berry content as an indicator of vine nitrogen status is poorly documented. MATERIALS AND METHODS Experimental location and plant material The study was conducted in Leoville Las Cases vineyard. Saint- Julien, Haut-Medoc, Bordeaux area, France. Soil mapping of this

2 Terroir Influence on Waier and Nitrogen Status of Cabernet Sauvignon vineyard revealed an unexpected soil variation: nine pedological types spread over only 100 ha. Four blocks planted with Vitis vinifera cv. Cabernet Sauvignon grafted on Vitis riparia L. rootstock were studied. Vine age was between 24 and 31 years. The four blocks were identified as 1G, 3 A, 4H and 5P. Each block was set up on a different soil type. To avoid soil variation inside a block, only 100 vines were studied per site, all located around the soil pit studied. The soil type, texture, percentage of organic matter in fine soil, occurrence of water table and root zone depth are presented in Table 1. The four blocks were less than 500 m apart. The altitude of the four blocks varied from 10 to 20 m and row orientation was identical. Hence it was accepted that no climatic variation occurred among the four blocks. This was confirmed by the vine phenology in 1997, which was similar for all the terroirs studied (bloom: May 25; veraison: July 21). The vines were spaced 1 x 1.1 m apart and double Guyot pruned; trunk height was 0,4 m from the soil surface. During vegetative development, vine canopy was mechanically cut at 1.1 m height for every block. Water potentials Pre-dawn leaf water potential (dawn^) and leaf water potential (leatf) were measured with a pressure chamber (Scholander et al, 1965) equipped with a digital manometer (SAM Precis 2000, Gradignan, France). DawnY was measured at the end of the night on uncovered mature leaves. LeaPf was measured on mature leaves exposed to sunlight, whether they were transpiring or not (Turner, 1981). Nitrogen status As in the preceding years, no nitrogen fertiliser was added in Hence vine nitrogen uptake resulted mainly from mineralisation of organic matter. Vine nitrogen uptake was estimated weekly, from veraison through harvest, by the berry total nitrogen content (Bell et al, 1979). Vine vigour Dry leaf mass was measured on eight vines per terroir just before harvest. Leaf area per vine was deduced from the correlation obtained. Average total bunch mass per vine was measured on 30 vines per block on the day of harvest. Average pruning mass per vine was determined on 25 vines per block. Must characteristics from veraison through harvest From veraison through harvest one thousand berries were sampled weekly at random from 100 vines on each site. These berries were immediately counted and weighed to determine average fresh berry mass. Berries were pressed and the juice was gently centrifuged to remove suspended solids. Sugar content was measured by refractometry. Titratable acidity was measured manually with Bromothymol blue. The titration was done with 0.1 N NaOH to an end point of ph 7.0. Malic acid was determined using an enzymatic kit from Boehringer Mannheim. On 200 berries anthocyanins were determined by the method of Ribereau-Gayon & Stonestreet (1965). The measurement of absorbency was conducted at a wavelength of 520 nm (optical distance: 1 cm). Tannins were measured by the 1PT: "indice des polyphenols totaux" (Ribereau-Gayon, 1970). The wavelength used was 280 nm (optical distance: 1 cm). Micro vinifi cation Microvinification was conducted identically for all lots. For every terroir 50 kg of fruit were harvested on the same day (September 27). The fruit from each block was crushed and stems were removed. For every lot 40 kg were put into a SOL stainless steel tank. In order to have similar wine alcohol content for tasting, 1G, 3A and 5P must were chaptalised to obtain the same sugar content as found in 4H. Alcoholic fermentation and malolactic fermentation were enhanced with a selected yeast strain and a selected bacteria strain, respectively. All the wine-making parameters (fermentation arrd maceration temperature, pumping over frequency, draining away, etc.) were similar. Wine tasting Wines were tasted blindly by a panel of 15 professional tasters. They were noted by positive preference order from 1 to 4. The scores of all judges for a wine were summed. Least significant difference (LSD) was determined by a modified Chi 2. Statistics Water potential and vigour indicators data were analysed with the ANOVA procedure and the Newman Keuls test to determine least significant difference (Statbox Pro Software). RESULTS Water status Soil water content was at a field capacity on all the soils studied in the spring (April), which is usual in the French oceanic area. In the Medoc 1997 was a rainy vintage: the rainfall in May, June and August was far above the average calculated over a 39-year span. Conversely, rainfall was a bit less than average in July and September (Table 2). From bloom through veraison total rainfall TABLE 1 Soil and rooting characteristics of the studied plots. 1G 3A 4H 5P Type of soil Gravel content Water table Texture of fine soil in limits of root zone depth Neoluvisol on gravely soil 75% no Sandy clay Planosol sedimorphe with heavy clay subsoil 15% no Heavy clay Redoxisol with heavy clay subsoil 50% yes Sandy clay Podzosols 15% no Sandy Organic matter (% in fine soil) in the top 60 cm Root zone depth (meter)

3 20 lerroir influence on Water and Nitrogen Status oj l^atternel bauvignon TABLE 2 Monthly rainfall in Saint Julien in 1997, compared with the longterm average. Rainfall 1997 (mm) Average rainfall (mm) May June July August September was 18 mm and from veraison to harvest 113 mm. From July through September P dawnt data varied from to MPa, whereas over the same period IG dawn data varied from to MPa (Fig. 1). At three different stages (July, August and September), dawn*f data were significantly lower for IG than for the other blocks. On August 23 midday leaf water potential showed no differences among the 4 soils (Fig. 2). Yet IG leaf water potential was significantly more negative than that of the other locations at the end of the afternoon, being the beginning of the recovery period of vine water status. This confirmed the lower pre-dawn leaf water potential measured on IG during the same period. Climatic conditions were identical for the four blocks studied. Thus water status differences observed in 1997 reflected mostly soil ability to supply water to the vine. Hardie & Considine (1976) and Van Leeuwen et al (1994), showed that dawny of MPa reflected a mild water deficit, which induced slackening of shoot growth. Given this, IG can be considered as the only block where vines underwent mild water deficits in As IG received as much rainfall as 5P, where no water deficit occurred, IG water deficits reflected low water-storage capacity, due to a high proportion of gravel in the shallow root zone (Table 1). Conversely, on 5P the root zone was twofold deeper than on IG, whereas the gravel content in the root zone was fivefold lower. These conditions were reflected in the high dawnt data collected from July to September on 5P. On 4H, despite the shallow root zone, no water deficits were detected. This was due to a permanent water table linked to the subsoil clay depression. Conversely, on 3A the subsoil clay is convex. Thus no water table can be formed ant the root zone is deeper (Table 1). Nitrogen status From veraison through harvest, berry total nitrogen content for 4H was threefold lower than for 5P and twofold lower than for IG (Fig. 3). Berry total nitrogen content of 3A was between that of 4H and of IG. Throughout this period the difference between berry total nitrogen content appeared to remain constant from one block to another. Lower total nitrogen content for 4H reflected the low organic matter content in this soil (Table 1). Conversely, 5P soil organic matter percentage was fourfold higher than for 4H, with total berry nitrogen being threefold higher than for 4H. Thus, under the conditions of this survey, berry total nitrogen content appeared to reflect soil organic matter content closely. Vine vigour A strong correlation between leaf area (measured with a planimeter) and dry leaf mass was established. Total leaf area per vine (Table 3) was not significantly different for IG, 3A and 4H, indicating that total leaf area per vine was not affected by water status for these 3 blocks. Hence, IG mild water deficits were only due to low soil-water content available in the root zone depth. Moreover, significantly larger total leaf area per vine did not enhance water deficits on 5P. This confirmed the higher soilwater content in the root zone depth of 5P. The significantly lower total leaf area per vine for 4H, in comparison to 5P was not due to water status, which was similar for both terroirs. The lower vine nitrogen status on 4H appeared to explain its weaker vigour. Furthermore, the total pruning mass per vine and the total cluster mass per vine were also significantly lower on 4H than on 5P. As with total leaf area per vine, 4H and IG showed no significant difference for total pruning mass per vine, nor for total leaf area per vine. Two vigour indicators, total bunch mass per vine -0,2-- (0-0,25 Q) -0,35 JULY AUGUST SEPTEMBER FIGURE 1 Dawn leaf water potential evolution from July through September 1997, Data are means of 8 measurements on 8 adjacent vines. Error bars indicate SE. S. Afr. J. EnoL Vitic., Vol. 22, No. 1, 2001

4 Termir Influence an Water and Nitrogen Status of Cabeme! Saavignon II 0,0 d) «-1,0 -o~ 1G leaf water potential - - 3A leaf water potential -D-4H leaf water potential -*c 5P leaf water potential -1,6 02:24 03:36 04:48 06:00 07:12 08:24 09:36 10:48 12:00 13:12 14:24 15:36 16:48 18:00 19:12 FIGURE 2 Diurnal leaf water potential on August 23, Data are means of 8 measurements on 8 adjacent vines. Error bars indicate SE j" f SEPTEMBER FIGURES Berry total nitrogen content from veraison through harvest (1997). TABLE 3 Vigour indicators: Total leaf area, total pruning mass and total bunch mass per vine. Total leaf area per vine (m2) Total pruning mass per vine (kg) Total bunch mass per vine (kg) 1G 3A 4H 5P 1.01 a' 1 ) 0.24 a b a b 0.96 a 0.31 b 1.09b 0.8 a 0,19 a a 1.46 a 0.43 c 1.103b (') Values within rows followed by the same letter do not differ significantly. and total shoot mass per vine, showed that 3A was significantly more vigorous than 4H, but no more than 1G. Thus the difference in nitrogen supply between 3A and 4H, which had the same water status, seemed to have had more influence on vigour than did vari ations in water status between 1G and 4H or between 1G and 3A Total leaf area per vine and total pruning mass per vine for 51 were significantly higher than for the 3 other terroirs, while 51 total cluster mass was significantly higher than 4H only. As tht trellising and pruning system were the same on the four sites, thii showed that low nitrogen status on 4H reduced total bunch mas: per vine to a greater extent than did the mild water deficits on 1G This further emphasised that low vine nitrogen status had ; greater influence on vigour than did mild water deficits befort veraison. Fresh berry mass From veraison through harvest 5P fresh berry mass was constant ly 30% higher than that of 4H, whereas it was only 10% to 15 9i higher than that of 1G and 3A (Fig. 4). Fresh berry mass and tota cluster mass per vine showed a strong linear correlation (R 2 =0.92

5 12 Terroir Influence on Water and Nitrogen Slalus of Cabeme! Sauvignon A -1G -4H -5P AUGUST SEPTEMBER FIGURE 4 Fresh berry mass evolution from veraison through harvest (1997). 230 y- AUGUST SEPTEMBER 440,0 FIGURES Must-reducing sugar content evolution from veraison through harvest (1997) Q> JULY AUGUST SEPTEMBER FIGURE 6 Must titrable acidity evolution from veraison through harvest (1997).

6 Terroir Influence on Water and Nitrogen Status of Cabernet Sauvignon 13 SEPTEMBER FIGURE? Must malic acid (meq/l) content evolution from two weeks post-veraison through harvest (1997) "' C 3A 1 G AUGUST SEPTEMBER FIGURES Must anthocyanin content evolution from two weeks after veraison through harvest (1997). n=4), which demonstrated that flower abortion did not explain the yield difference between the sites. Low fresh berry mass on 4H confirmed the low vigour previously observed. Lower fresh berry mass on 1G and 3 A, compared to 5P, can be attributed to mild water deficits and lower nitrogen status respectively, even though these are not reflected by significant differences in total cluster mass per vine. Berry composition through ripening Must from 4H and 1G exhibited the highest sugar content and the lowest acidity, whereas the reverse was true on 5P (Figs. 5, 6 and 7). Must sugar content was less for 1G than for 4H. Even though titratable acidity on 1G and 4H was similar, malic acid content was nevertheless lower on 1G. As already shown by Van Leeuwen and Seguin (1994), vine water deficits enhance low must malic acid content. For the four sites a strong linear correlation (R 2 =0.97; n=4) between average dawn*jk data (from July through September) and berry malic acid content at the harvest was observed. Anthocyanins, tannins and wine appreciation Anthocyanin content for 4H was almost twofold higher than for 3A and 5P, whereas the difference between 4H and 1G was less pronounced (Fig. 8). IPT differences followed the same hierarchy, with 4H and 1G exhibiting the highest tannin content (Table 4). The professional panel of tasters distinguished 4H and 1G as the best wines (Table 4). The most appreciated wines were also the highest in phenolic content. Wine phenolic content correlated closely with phenolic content (Table 4). Must phenolic content for 3A, 4H and 5P suggested that low nitrogen status without water deficits was linked to high enological potential for a red grape vari-

7 TABLE 4 Must phenolic content at the harvest, wine phenolic content and wine-tasting appreciation. Must anthocyanin content at harvest (mg/l) Must tannin content at harvest (IPT) (1) Wine anthoeyanin content (mg/l) Wine tannin content (IPT) Wine-tasting sum rate 1G 3A 4H 5P (i) IPT = "indice des polyphenols totaux" a (2) 39 b a < 2 > Values within rows followed by the same letter do not differ significantly. ety such as Cabernet Sauvignon. This confirmed the result of Keller and Hrazdina (1998) on continuously irrigated, potted Cabernet Sauvignon. They observed that excessive nitrogen fertilisation decreased win phenolic content. In a rootstock trial Delas et al. (1991) observed an increase of phenolic grape content for some of the rootstocks where nitrogen fertilisation was not applied. Hence, both early mild water deficits and lower nitrogen status throughout the growth period had beneficial effects on total berry phenolic contents and wine quality. This appeared to be linked to the limiting effect on the vine vigour of both of these factors. DISCUSSION With the same water status for both sites, lower nitrogen status on 4H than on 5P induced a decrease in vigour. The total leaf area per vine, the total pruning shoot mass per vine and the total cluster mass per vine were respectively 55%, 44% and 30% lower on 4H than on 5P. Considering these vigour indicators, 4H underwent a clear nitrogen deficiency in The total cluster mass per vine represented a calculated yield of 6.5 tons per ha for 4H, suggesting that nitrogen deficiency level was moderate, as it allowed an acceptable yield for a high-quality wine-producing vineyard. Bell et al (1979) reported on continuously irrigated Thompson seedless that for a total nitrogen content of must at harvest ranging from 287 to 754 mg/l no significant difference occurred in crop mass per vine. Kliewer and Cook (1971) showed on potted vines that leaf area and trunk circumference increased linearly with increasing concentrations of nutrient solutions ranging from 0 to 4 mm of NOs, whereas they remained constant with concentrations ranging from 4 to 8 mm of NOs. This suggests that the Bell et al (1979) range of must total nitrogen content indicated no lack of nitrogen, because it was superior to the threshold value where nitrogen begins to have an effect on vigour. In P must total nitrogen content (290 to 370 mg/l from veraison through harvest) fell within this range. Given this, 5P must total nitrogen data reflected no limiting vine nitrogen status. Moreover, 3A must total nitrogen content (185 to 230 mg/l) corresponded to an intermediate level of vine nitrogen status, between those of 5P and of 4H. This was emphasised by the fact that 3A vigour, estimated by the total pruning mass per vine and the total cluster mass per vine, was significantly lower than that of 5P and significantly higher than that of 4H, with no water status difference c Conradie (1980, 1986) demonstrated that most of the vine nitrogen uptake occurs from bud burst through veraison. During this period nitrogen is increasingly allocated to fruit. From veraison through harvest Conradie (1991) observed a translocation on Chenin blanc of nitrogen from shoot, leaves and permanent structure to bunches. These observations were confirmed by constant or slightly increasing must total nitrogen content on the 4 sites in 1997, from veraison through harvest, despite different rates of berry diameter increase (Figs. 3 and 4). Low nitrogen status for 4h can thus be considered as reflecting limited vine nitrogen uptake since bud break. This could explain why the low nitrogen status of 4H reduced vigour more than did 1G mild water deficits for 1G. 4H low nitrogen status was constant through the vegetative and the reproductive periods (Fig. 3), whereas 1G mild water deficits appeared at the end of fruit set and were not permanent until harvest (Fig. 1). Further investigations will be necessary to define more accurately the range of total nitrogen content in must leading to the best must and wine quality. Different total nitrogen contents data on 1G and 4H indicated that a nitrogen status varying from low to medium favoured the production of highly appreciated wine in this Bordeaux top estate. Yet 1G and 3A grape total nitrogen content data suggested that a medium nitrogen status was adequate to obtain a high-quality wine, if associated with mild vine water deficits. Conversely, under the conditions of this survey, without vine water deficits only nitrogen deficiency on 4H (90 to 150 mg/l of total must nitrogen) had an obvious positive effect on must and wine phenolic content. This positive effect on wine quality suggested that nitrogen deficiency should be maintained in the vineyard and that nitrogen deficiency in must should be corrected in the tank before and/or during the fermentation process. This practice is already followed with success in several Bordeaux vineyard areas. CONCLUSIONS This terroir survey, carried out during a rainy season in the Bordeaux area, revealed a soil effect on vine nitrogen status measured by must total nitrogen content. Despite the rainy vintage, we observed on a gravelly soil (1G) mild water deficits linked to low soil water content in the shallow root zone. Low nitrogen status was found on soil having a particularly low organic matter content (4H). We observed that low nitrogen status reduced vine vigour more than did mild water deficits. Yet when vines were subject to one of these limiting factors, the wines produced in this Bordeaux top estate were of significantly higher quality. Low vine nitrogen status induced high berry tannin and anthocyanin content, accompanied by low berry mass. It also reduced yield, but this may be acceptable in a very high-quality grape-producing vineyard. Considering the benefits of low vine nitrogen status on the phenolic content of must and wine, it remains to define how this low nitrogen status can be managed more accurately in order to conserve the equilibrium between the highest possible wine quality and a decreased but economically acceptable yield. LITERATURE CITED Bell, A.A., Ough, C.S. & Kliewer, W.M., Effects on must and wine composition, rates of fermentation and wine quality of nitrogen fertilisation of Vitis vinifera var. Thompson seedless grapevines. Am. L Enol. Vitic. 30, Conradie, W.J., Seasonal uptake of nutrients by Chenin blanc in sand culture: I. Nitrogen. S. Aft. J. Enol. Vitic. 2, S. Afr. J. Enol. Vitic., Vol. 22, No. 1,2001

8 Terroir Influence an Water and Nitrogen Status of Cabemet Sauvignan 15 Conradie, W.J., Utilisation of nitrogen by the grapevine as affected by time of application and soil type. S. Afr. J. Enol. Vitic. 2, Conradie, W.J., Distribution and translocation of nitrogen absorbed during early summer by two-year-old grapevines grown in sand culture. Am. J. Enol. Vitic. 42, Delas, J. Molot, C. & Soyer, J.P., Effects of nitrogen fertilisation and grafting on the yield and quality of the crop of Vitis vinifera cv. Merlot. Proceedings of the International symposium on nitrogen in grapes and wine, Seattle, pubbshed by The American Society for Enology and Viticulture Hardie, W.J. & Considitie, J.A., Response of grapes to water deficit stress in particular stages of development. Am, J. Enol, Vitic. 2, Keller, M. & Hrazdina, G., Interaction of nitrogen availability during bloom and light intensity during veraison. II. Effects on anthocyanin and phenolic development during grape ripening. Am. J. Enol. Vitic. 49, Kliewer, W.M. & Cook, J.A., Arginine and total free amino acids as indicators of the nitrogen status of grapevines. Am. J. Enol. Vide. 96, Koundouras, S., van Leeuwen, C., Seguin, G. & Glories, Y., Influence of water status on vine vegetative growth, berry ripening and wine characteristics in Mediterranean zone (example of Nemea, Greece, variety Saint George, 1997), J. Inter. Vigne Vin. 33, Matthews, M.A., Anderson, M.M. & Schultz, H.R., Phenologic and growth responses to early and late season water deficit in Cabernet franc. Vitis 26, Matthews, M.A. & Anderson, M.M., Fruit ripening in Vitis vinifera L.: Responses to seasonal water deficits. Am. J. Enol. Vitic, 39, Naor, A. & Wample, R.L., Diurnal internode growth rate in field-grown 'Concord' (Vitis labrusca) grapevines. J. Hortic. Sci. 71, , Ribereau-Gayon, P., Le dosage des composes phenoliques totaux dans les vins rouges. Chim. Anal. 52, Ribereau, P. & Stonestreet, E,, 1965, Le dosage des anthocyanes dans le vin rouge. Bull. Soc. Chim. 9, Scholander, RE, Hammel, H.J., Bradstreet, E.D. & Hemmingsen, E.A., Sap pressure in vascular plants. Science 148, Schultz, H.R. & Matthews, M.A., Xylem development and hydraulic conductance in sun and shade shoots of grapevine (Vitis vinifera L.), evidence that low light uncouples water transport capacity from leaf area. Planta 190, Seguin, G,, Les sols des vignobles du Haut-Medoc. Influence sur 1'alimentation en eau de la vigne et sur la maturation du raisin. These de doctoral d'etat, Faculte des Sciences, Universite de Bordeaux. Seguin, G., "Terroirs" and pedology of wine growing. Experientia 42, Smart, R.L., Grapevine response to furrow and trickle irrigation. Am. J. Enol. Vitic. 25, Soyer, J.P., Molot, C., Bertrand, A., Gazeau, O., Lovelle, B.R. & Delas, J., Influence de I'enherbement sur 1'alimentation azotee de la vigne et sur la composition des moflts et des vins. In: Oenologie 95, Actes 5 e Symp. Inter. Oenologie, Bordeaux. Tec Doc ed., Paris Spayd, S.E., Wample, R., Evans, R.G., Stevens, R.G., Seymour, B.J. &Nagel, C.W., Nitrogen fertilisation of white Riesling grapes in Washington. Must and wine composition. Am. J. Enol. Vitic. 45, Turner, N.C., Techniques and experimental approaches for the measurement of plant water status. Plant Soil 58, Van Leeuwen, C. & Seguin, G., Incidences de I'alimentation en eau de la vigne, appreciee par 1'etat hydrique du feuillage, sur le developpement de 1'appareil vegetatif et la maturation du raisin (Vitis vinifera L. cv. Cabernet franc, Saint Emilion, 1990). J. Inter. Vigne Vin. 28,'

9 Relationship Between Microclimatic Data, Aroma Component Concentrations and Wine Quality Parameters in the Prediction of Sauvignon blanc Wine Quality J. Marais 1, F. Calitz 2 and P. D. Haasbroek 3 1) ARC Infruitec-Nietvoorbij, Private Bag X5026, 7599 Stellenbosch, South Africa 2) ARC Biometry Unit, Private Bag X5013, 7599 Stellenbosch, South Africa 3) ARC Agromet, Private Bag X5013, 7599 Stellenbosch, South Africa Submitted for publication: September 2000 Accepted for publication: April 2001 Key words: Microclimate, temperature, radiation, Sauvignon blanc quality, prediction model Sauvignon blanc grape chemical and wine sensory data, as well as meteorological data (temperature and visible light radiation), collected in three climatically different wine regions in South Africa over three seasons and from two different canopy treatments were statistically analysed. A model for the prediction and/or definition of Sauvignon blanc wine quality was developed. The model utilises above- and within-canopy radiation and can explain 68.8% of the variation in the cultivar-typical vegetative/asparagus/green pepper intensity of Sauvignon blanc wine. Other significant correlations, e.g. between temperature and monoterpene concentrations, were also obtained. Further research is necessary to test and refine this model for application under different environmental conditions. Over many years research has been aimed at developing a simple model or recipe that could define and predict wine quality. Examples are sugar/acid ratio ( B/TTA) (Du Plessis & Van Rooyen, 1982), the glycosyl-glucose (G-G) assay (Francis et al, 1998) and the red wine colour index (Holgate, 2000). Such a model should also be useful as a method to compensate the grape producer according to grape quality. The aim further is that the model should incorporate parameters that are easily measurable under field conditions, instead of parameters such as aroma impact volatiles, the quantification of which is dependent on complicated and expensive gas chromatographic and mass spectrometric analyses. Due to the complexity of factors affecting grape and wine composition and quality, attempts to date have achieved varying degrees of success and a foolproof, simple model still has to be developed. Nevertheless-, earlier attempts were valuable and all contribute something to the eventual solution of this problem. Significant correlations between microclimatic parameters and wine sensory characteristics were reported earlier (Noble, Elliot- Fisk & Alien, 1995). Models using light radiation in the prediction of photosynthesis under field conditions, were also developed (Blackburn & Proctor, 1983; Goudriaan, 1986; Spitters, 1986; Haasbroek, Myburgh & Hunter, 1997). Photosynthesis is dependent on light radiation and temperature and is indirectly an important factor in grape composition development. The approach in this study was to investigate the possible use of microclimatic parameters as a quality predictor, which can be easily measured over the whole grape-ripening period. The rate of enzymatic and chemical reactions in grapes, in terms of development and/or degradation of components, is dependent mainly on temperature and light radiation. Therefore these two climatic parameters were used as variables. Canopy density, which is a function of different climatic and viticultural factors, largely determines the effects of temperature and light radiation. The purpose of this study was to correlate microclimatic data (temperature and radiation) with aroma component concentrations and wine quality parameters of Sauvignon blanc and to develop a model for predicting wine quality. MATERIALS AND METHODS Grapes, regions and seasons Vitis vinifera L. cv. Sauvignon blanc grapes, grown in three climatically different regions, namely Robertson (Region IV, degree days), Stellenbosch (Region IE, degree days ) and Elgin (Region II, degree days) (Le Roux, 1974), were used. A single canopy manipulation treatment, which altered microclimate in a natural way to obtain a higher degree of shading, was applied in each region (Archer & Strauss, 1989; Marais, Hunter & Haasbroek, 1999). Control vines were not manipulated. Three replications, consisting of 15 vines per replica, were used. The general viticultural practices followed were described by Marais et al. (1999). Grapes were obtained over three seasons, namely 1997, 1998 and Results of the 1997 and 1998 seasons were published by Marais et al. (1999) and trials during the 1999 season were a repetition of the first two seasons. Grape samples (whole bunches, 2kg per sample) were collected weekly at random from each treatment and replicate over three weeks between approximately 16 B (close to veraison) and 21 B (ripeness). Grapes were harvested at ripeness (fourth ripening stage) for wine production, following standard Nietvoorbij practices for small-scale white wine production (Marais et al., 1999). Meso- and microclimatic measurements MCS data loggers, which continually measured temperature and visible light radiation ( nm) over the whole ripening period on an hourly basis, were installed within selected vine canopies in the vicinity of the clusters, as well as above the canopies. Each radiation sensor consisted of 10 individual sensors, fitted in parallel onto a metre-long rod, thus giving a more S. Afr. J. Enol. Vitic., Vol. 22, No. 1,

10 A Model for Predicting Sauvignon blanc Wine Quality 23 reliable average reading within the canopy (Marais et a/., 1996; 1999). These measurements were done in each experimental unit. Aroma component analyses Grape samples from each treatment, region, season and replicate were analysed for free and bound monoterpenes and bound Ci3- norisoprenoids by gas chromatography, and for 2-methoxy-3- isobutylpyrazine (ibmp) by gas chromatography / mass spectrometry (Marais et al., 1996; 1999). Sensory analysis Wines of each treatment, region, season and replicate were sensorially evaluated for fruitiness and vegetative/asparagus/green pepper intensities by a panel of six experienced judges. A line method was used, i.e. evaluating the intensity of each characteristic by making a mark on an unstructured, straight 100 mm line (Marais et al., 1999). Statistical procedures Seventy-two independent data sets consisted of three seasons, three localities, two canopy treatments and four ripening stages. Only 70 data sets were used, due to two missing plots. A 15-vine plot was considered as an experimental unit. For each experimental unit the microclimatic measurements, such as the mean maximum, mean minimum and average temperatures during a ripening week, were calculated and served as independent variables. The average light radiations above and within the canopies were also recorded as independent variables. The grape and wine measurements (averages of three replicates), such as aroma volatiles (monoterpenes, norisoprenoids and ibmp) and sensory data (fruitiness intensity and vegetative/asparagus/green pepper intensity) were recorded as dependent variables. Pearson's correlation coefficients were calculated between the above-mentioned independent and dependent variables and scatter plots were examined for trends. After examining the scatter plots, it was clear that the radiation variable showed a non-linear trend. Therefore three additional terms were added (i.e. the squares of the above- and within-canopy light radiation and the interaction between them) before the data were subjected to the stepwise regression procedures. For each dependent variable a stepwise regression was performed with a specified significance level of at least 5% for inclusion in the models, using SAS (Version 6.12 ) statistical software (SAS, 1990). For the above-mentioned analyses, the data were handled in two ways. Firstly, all the data were statistically processed. Secondly, the data obtained at ripeness (fourth ripening stage) only were statistically processed, since wines were produced from grapes at this stage. RESULTS AND DISCUSSION The final models produced by the stepwise regression procedures are given in Table 1 and Figures 1 to 3 for the complete data set, and in Table 2 and Figures 4 to 6 for the data of harvest week four only. These coefficients correspond to significant correlations between dependent and independent variables (data not shown). The figures presented were selected with the main purpose of this study in mind, i.e. to predict wine quality from easily-measurable microclimatic parameters. Although ibmp was significantly described by the interaction of above- and within-canopy radiation, only 43.4% of the variation could be explained by the model (Table 1, Fig. 1). For fruitiness intensity 33.3% of the variation was explained by maximum temperature and the above-canopy radiation (Table 1, Fig. 2). The vegetative/asparagus/green pepper intensity (complete data set) showed the best coefficient of determination, i.e. R 2 = 68.8% (Table 1, Fig. 3). The above-canopy radiation showed a greater effect than the within-canopy radia- TABLE1 Coefficients of independent variables selected for model and standard errors. Models fitted to all the data (N=70). Variable Intercept T min. T max. T aver. A W AxW A 2 W 2 R 2 Mono! ± ± % sl<0.01 Noris ± ± % sl=0.03 ibmp ± ± ± ± % sl< ± ± ± % sl<0.01 Vagp ± ± ± ± % sl<0.01 Monot Noris ibmp Fi Vagp A W T min. T max. T aver. _ = Total free monoterpenes {relative concentration) = Tota! bound monoterpenes and Cn-norisoprenoids (relative concentration) = 2-Methoxy-3-isobutylpyrazine concentration (ng/l) = Fruitiness intensity (%) = Vegetative/asparagus/green pepper intensity (%) = Visible light radiation above the canopy (MJ) = Visible light radiation within the canopy (MJ) = Minimum temperature ( C) Maximum temperature ( C) = Average temperature ( C) = Not selected

11 24 A Model for Predicting Sauvignon blanc Wine Quality TABLE 2 Coefficients of independent variables selected for model and standard errors. Models fitted to the data from harvest week four only (N= 18). Variable Monot Noris ibmp Fi Vagp Monot Noris ibmp Fi Vagp A W T min. Tmax. T aver. Intercept T min, T max, T aver. A W ± ± ± ± ± ± ± ± = Total free monoterpenes (relative concentration) = Total bound monoterpenes and Cj3-norisoprenoids (relative concentration) = 2-Methoxy-3-isobutylpyrazine concentration (ng/l) = Fruitiness intensity (%) = Vegetative/asparagus/green pepper intensity (%) = Visible light radiation above the canopy (MJ) = Visible light radiation within the canopy (MJ) = Minimum temperature ( C) = Maximum temperature ( C) = Average temperature { Q = Not selected AxW A 2 W z R % sl< % ± sl= % sl=0.046 _ % ± sko.ol * ^ ft Y = A W + 0.5Q92AxW R = 43.4% si < 0.01 Y = T max A R 2 = 33.3% si < 0.01 rm '«Won( Mj, " ^ -*" FIGURE 1 Effect of above-canopy (A) and within-canopy (W) radiation (between veraison and ripeness) on 2-methoxy-3-isobutylpyrazine concentration in Sauvignon blanc grapes. FIGURE 2 Effect of above-canopy (A) radiation and average maximum temperature (T max.) (between veraison and ripeness) on the fruitiness intensity of Sauvignon blanc wine. tion. The higher the average above-canopy radiation, the lower the vegetative/asparagus/green pepper intensity of the wine (Fig. 3). This is in agreement with data presented by Marais et al. (1999). Although natural shading plays an extremely important role in Sauvignon blanc quality development (Marais et al., 1999), models developed separately for the shaded and control treatment data were more or less the same as for the full data set and are therefore not reported. Models arising from the data of ripening week four (harvest time) only are presented in Table 2. As much as 70.5% of the variation in monoterpene levels was explained by the average maximum temperature for the week in which the grape samples were taken. Monoterpenes may contribute to the flower-like nuances of Sauvignon blanc wine, but are probably less important than the cultivar-typical methoxypyrazines. The negative coefficient indicated that the higher the maximum temperature, the lower the monoterpene concentrations in the grapes, which is in agreement with results reported by Marais et al. (1999), There was not enough evidence, i.e. no variables were selected, to fit a model on the norisoprenoids in the grapes. 2-Methoxy-3-isobutylpyrazine, S. Afr. J. Enol. Vitic., Vol. 22, No. 1, 2061

12 A Model for Predicting Sauvignon blanc Wine Quality 25 Y = 265, A A - 8.3B95W W = 68.6% si < g 30 Is 25 Q.D) IT 20 «s ««A 15 Y = A A R = 40.8% si = Above-canopy radiation (MJ) FIGURES Effect of above-canopy (A) and within-canopy (W) radiation (between veraison and ripeness) on the vegetative/asparagus/ green pepper intensity of Sauvignon blanc wine. FIGURE 4 Effect of above-canopy (A) radiation at harvest (week 4) on 2-methoxy-3-isobutylpyrazine concentration in Sauvignon blanc grapes * Y = A A R = 65.0% si < S." is5 40.* Y = A R = 22.7% si = Above-canopy radiation (MJ) FIGURES Effect of above-canopy (A) radiation at harvest (week 4) on fruitiness intensity of Sauvignon blanc wine Above-canopy radiation (MJ) FIGURE 6 Effect of above-canopy (A) radiation at harvest (week 4) on vegetative/asparagus/green pepper intensity of Sauvignon blanc wine. fruitiness intensity and vegetative/asparagus/green pepper intensity showed a significant response to the above-canopy radiation only (Figs. 4, 5 and 6, respectively). Again the vegetative/asparagus/green pepper intensity variation was best explained by a quadratic response of above-canopy radiation with 65.0%, and a minimum turning point of 24.9% for the above-canopy radiation. In spite of the fact that only 18 data points were applied for ripening week four, results obtained for vegetative/asparagus/green pepper intensity (65.0%, Table 2) were close to those of the full data set (68.8%, Table 1). CONCLUSIONS There is a significant relationship between gas chromatographically-analysed grape aroma components, sensorially-evaluated wine quality parameters and microclimatic data. Models are presented, one of which can predict Sauvignon blanc wine quality (cultivar-typical vegetative/asparagus/green pepper intensity ^ from above- and within-canopy radiation data, collected ovei grape ripening, with a fairly high degree of accuracy. It is therefore possible to predict wine quality from easily-measurabh microclimatic data. The results of this investigation wert S. Afr. J. Enol. Vitic., Vol. 22, No. 1,2001

13 26 A Model for Predicting Sauvignon blanc Wine Quality obtained under specific South African conditions and it is not claimed that they are valid under all conditions. Nevertheless, they can be considered as a contribution towards wine quality prediction. Future research is necessary to test and refine this model for application under different environmental conditions. LITERATURE CITED Archer, E. & Strauss, H.C., Effect of shading on the performance of Vitis vinifera L. cv. Cabernet Sauvignon. S. Afr. J. Enol. Vitic. 10, Blackburn, W.J. & Proctor, J.T.A., Technical note. Estimating photosynthetically active radiation from measured solar irradiance. Solar Energy 31, Du Plessis, C.S. & Van Rooyen, P.C., Grape maturity and wine quality. S. Afr. J. Enol. Vrtic. 3, Francis, I.L., Hand, P.G., Cynkar, W.U., Kwiatkowski, M., Williams, P.J., Armstrong, H., Bolting, D.G., Gawd, R. & Ryan, C., Assessing wine quality with the G-G assay. In: Blair, R.J., Sas, A.N., Hayes, P.p. & Hoj, P.B., (eds.). Proc. 10 th Aust. Wine Ind. Tech. Conf., 2-5 August 1998, Sydney, Australia, pp Goudriaan, J., A simple and fast numerical method for the computation of daily totals of crop photosynthesis. Agric. For. Meteorol. 38, Haasbroek, P.D., Myburgh, J. & Hunter, J.J., Modelling sunlight interception and photosynthetic activity - correlation with field data. Poster. 21 SI Annual SASEV Congress, November 1997, Cape Town, South Africa. Holgate, A., Berry colour index assay - application and interprelation of the assay as an objective measure of red wine grape quality in a commercial vineyard. Proc. 5th Int. Symp. Cool Climate Vitic. & Enol., January 2000, Melbourne, Australia. In press. Le Roux, E.G., 'n Klimaatsindeling van die Suidwes-Kaaplandse wynbougebiede. M.Sc. Thesis, University of Stellenbosch, StelSenbosch, South Africa. Marais, J., Hunter, J.J. & Haasbroek, P.D., Effect of canopy microclimate, season and region on Sauvignon blanc grape composition and wine quality. S. Afr. J. Enol. Vitic. 20, Marais, J., Hunter, J.J., Haasbroek, P.D. & Augustyn, O.P.H., Effect of canopy microclimate on Sauvignon blanc grape composition. In: Slockley, C.S., Sas, A.N., Johnstone, R.S. & Lee, T.H. (eds.). Proc. 9th Aust. Wine Ind. Tech. Conf., July 1995, Adelaide, Australia, pp Noble, A.C., Elliot-Fisk, D.L., & Alien, M.S., Vegetative flavor and methoxypyrazines in Cabernet Sauvignon. In: Roussef, R.L. & Leahy, M.M. (eds.). Fruit flavors. Biogenesis, Characterization and Authentication. ACS Symposium. Series 596,22-27 August 1993, Chicago, Illinois, USA. pp SAS SAS/Stat User's Guide. Version 6, 4 lh Edition, Vol. 2, SAS Institute Inc., SAS Campus Drive, Cary, NC Spitters, C.J.T., Separating the diffuse and direct component of global radiation and its implications for modelling canopy photosynthesis. Part II, Calculation of canopy photosynthesis. Agric. For. Meteorol. 38,