Long-term Dynamics of Nitrogen and Carbohydrate Reserves in Woody Parts of Minimally and Severely Pruned Riesling Vines in a Cool Climate

Experimental design.

Field experiments were conducted with Vitis vinifera L. cv. Riesling (clone: Gm 198/rootstock: 5C, planted in spring 1977) from 1999 to 2003 in Geisenheim, Germany (50°N, 8°E). Compared to other cool-climate locations where MP systems have been tested, such as Coonawarra, Australia, this site has about 300 heat units (calculated as average monthly temperature above 10°C with a 19°C upper cut-off multiplied by the number of days per month) (Gladstones 1992) less over the growing season (1045 as compared to 1337) with an annual precipitation of 534 mm (as compared to 638 mm for Coonawarra). The experimental vineyard had a south to southwest exposure with a slope of 15 to 20% and a loam to clay-loam soil. The minimally pruned (MP) plots were converted from a Sylvoz trellis during the winter 1995 to 1996 and compared to the standard vertical shoot-positioned system (VSP). Both systems were arranged in three replicated randomized blocks of two to five rows each, with a row by vine spacing of 2.8 m x 0.85 m for MP and 2.0 m x 1.2 m for VSP, respectively. The replicates were blocked down the slope and each block contained 48 vines (total of 144 vines per pruning system). The VSP was cane-pruned to 8 buds/m² (19 buds/vine).

Plant material.

Wood samples were collected from three plants per field replicate during six phenological stages following the developmental scale (BBCH) of Eichhorn and Lorenz (1977). Sampling times in all years were during dormancy (prebudbreak, January to February; D) (BBCH 00), at budbreak (end of April to beginning of May; BB) (BBCH 11), bloom (June; B) (BBCH 65), bunch closure (July; BC) (BBCH 77), veraison (mid-August; V) (BBCH 81), and end of leaf fall (end of October to beginning of November; LF) (BBCH 97), except for 2003 when no samples were collected during dormancy and bloom. Wood samples were cut from the vines with a pair of pruning shears and separated into three fractions: one-year-old wood, two-year-old wood, and wood older than two years (>2 years contained parts of trunks and cordons). Obviously, the older wood fractions also contained annual rings of younger wood. About 30 to 75 g was sampled per replicate (three vines) and wood fraction. The same vines were not re-used the next year. Root samples were also collected during the 1997 season, for which roots were excavated with a back hoe at a distance of 50 cm from the trunk to a depth of ~70 cm. Samples were taken from a total of 12 vines across the three blocks. The root material taken from three successive vines was pooled into a replicate (~75 to 125 g). No distinction was made between different root categories. Vines were repeatedly sampled over the season. Both wood and root samples were weighed to determine fresh weight, dried for 1 hr at 105°C to deactivate enzymes, again dried 48 hr at 60°C, and then reweighed to determine dry weight. The dried samples were ground in two steps: first with a coarse grinder (type SM1, Retsch GmbH, 42781, Haan, Germany), then with a fine grinder to pass mesh size <0.2 mm (type cyclotec 1093, tecator, 26383, Höganös, Sweden) in preparation for carbohydrate and nitrogen analyses.

Carbohydrate and nitrogen analyses.

Analyses of total nonstructural carbohydrates (TNC: the carbohydrate not bound into structural components such as cellulose) and its components starch, sucrose, fructose, and glucose followed the enzymatic assays of Boehringer (1989). For sugar determination, 500 mg of dry homogenized sample were extracted in 25 mL distilled water at 60°C in a swirling water bath for 1 hr. The pH was measured and if necessary regulated to 4 to 5 by adding some drops of sodium hydroxide (NaOH, 5 molar sol.). The entire solution was then transferred into a 100-mL volumetric flask, brought to volume with distilled water, mixed, and 20 mL was filtered (N 33 wet filter, Ederol) into a test tube for the assay. For starch determination, 500 mg dry homogenized sample material was extracted in 20 mL dimethyl sulfoxide with 5 mL HCL (8 molar sol.) for 1 hr at 60°C. All remaining steps were as with the sugars. Enzymatic sample preparation was according to Boehringer (1989) and quantification was performed using a photometer (Specord 200, analytikjena AG, D-07745, Jena, Germany).

Nitrogen (N) was analyzed on samples collected in 1999 and 2000 by the Kjeldahl (Kj) method using 500 mg of dried, homogenized material. From 2001 forward, N was analyzed using an automated high-temperature combustion (HTC) method coupled to a nitrogen gas analyzer (Vario MAX CNS, Elementar Analysensysteme GmbH, 63452 Hanau, Germany) using 200 mg of dried homogenized samples. All laboratory analyses were performed in duplicate.

A population of samples (40) from the two training systems, all three wood fractions, and all six phenological stages from 1999, 2000, and 2001 were analyzed for nitrogen with both methods to examine possible differences between the two techniques. Results were linearly correlated, but the Kj-N consistently gave higher N-values (Figure 1⇓). These results were independent of canopy system, phenological sampling stage, year, or wood fraction. As a goal was to present long-term data in this study, the regression result was used to recalculate the N-values from 1999 and 2000 so that they were comparable to the results of the HTC method, which is a more recent and advanced technique.

• Download figure
• Open in new tab

Figure 1

Relationship between values of nitrogen content determined with the Kjeldahl method (Kj-N) and values obtained with the high-temperature combustion method (HTC-N) on a randomized population (n = 40) comprising samples from all wood ages of both pruning systems from 1999 to 2001. Data points represent the mean of a double laboratory analysis.

Leaf area determination.

Leaf area development was determined on four dates (bloom, bunch closure, veraison, and at the beginning of leaf fall [mid-October]) in 2002 and 2003 for both systems. For VSP, four plants per date were chosen at random with a minimum of one plant per block, while for MP, four strips of a canopy length of 0.85 m each (= planting distance) were sampled because it was impossible to separate individual vines in this treatment. In 2002, all leaves were stripped off the plants. Individual leaf length (midrib) was measured and correlated to individual leaf area (Schultz 1992). This method was extremely time-consuming, so measurements were simplified in 2003. For each plant or planting distance sampled, the leaf length of 10 leaves was measured for the VSP treatment (total n = 40 sampling date⁻¹) and 60 for the MP treatment (total n = 240 sampling date⁻¹). Leaves were then dried at 65°C (24 hr) and individual leaf dry weight determined. Individual leaf area was expressed as a function of leaf weight, which gave linear regression coefficients, R², between 0.69 and 0.86 depending on the sampling date. The remaining populations of leaves on these plants were harvested, split into four groups per plant or planting distance, and dried. Established regressions were extrapolated to these dry weight values to calculate leaf area. This method somewhat overestimates total leaf area at the beginning of the season when growing leaves are present, but in general the results agreed well with leaf areas determined during the previous year for the same phenological stages.

Measurements of plant water status.

Leaf water potential (Ψ_pd) was determined predawn with a pressure chamber (Soilmoisture Corp., Santa Barbara, CA) on eight fully expanded leaves per treatment (each from a different vine and sampled across all blocks) from the central part of the canopy. Measurements were conducted between 8 (2000) and 16 (2003) times per season.

Statistical analysis.

Linear regressions and analyses of variance (ANOVA, Holm–Sidak method) were calculated with SigmaStat 3.1 (Systat Software, Point Richmond, CA).

Nitrogen reserves: seasonal dynamics.

Levels of reserve-N were highest in all woody fractions of both pruning systems during dormancy (BBCH 0) (January to February), budbreak (BBCH 11), and leaf fall (BBCH 97) (Figure 2⇓). Highest values during dormancy were found in one-year-old wood and ranged from 0.50 to 0.68% (on a dry weight basis) depending on the year and the training system. Reserve pools were lower between bloom and veraison, with minimum values often attained at bunch closure. N-concentrations were lower in older wood fractions irrespective of the time during the season and the year in question, and the range of N-concentrations over the six phenological stages was smaller with increasing wood age (Figure 2⇓).

• Download figure
• Open in new tab

Figure 2

Nitrogen dynamics of different wood fractions for VSP and MP vines as a function of phenological stage over the course of five seasons (1999 to 2003). Phenological picture code is according to the BBCH system. Values are means ± SE with n = 3. *, **, and *** indicate significance at p ≤ 0.05, 0.01, and 0.001, respectively.

These general results confirm those obtained for young (three years) (Schaller et al. 1989) and mature (>18 years) grapevines in the field (Löhnertz 1988) and those obtained on either potted vines or cuttings grown in sand culture and trenches (Conradie 1990, Cheng et al. 2004, Zapata et al. 2004). However, while Cheng et al. (2004) and Zapata et al. (2004) reported a 20 to 40% decrease in N-concentration of perennial plant parts (with or without roots) from budbreak to early bloom for small potted or greenhouse grown plants, our data suggest greater remobilization during the same period with a 55 to 70% reduction of total reserve-N. These values corroborate those of Schaller et al. (1989), where the main N-storage form mobilized was arginine, but contrast those of Bates et al. (2002), which indicated little change in N-concentration of aerial woody parts throughout the growing cycle for three-year-old field-grown Concord grapevines.

Nitrogen reserves: differences related to training system.

One-year-old wood of MP vines had higher N than VSP vines at dormancy during the four years it was measured. These differences between MP and VSP were less consistent at budbreak and leaf fall and for older wood (Figure 2⇑). Rühl and Clingeleffer (1993) also found slightly, albeit not significantly, higher values for MP vines than for spur-pruned vines during dormancy with 0.68 (MP) to 0.62% (spur-pruned) for one-year-old wood and 0.30 to 0.29% for old wood, respectively.

N-concentration tended to be higher at budbreak than at dormancy in most samples (Figure 2⇑). This increase is thought to be related to N-transport from the roots to the aerial parts of the vine (Löhnertz 1988, Schaller et al. 1989), since most N-reserves are located in the roots (Conradie 1990) and substantial amounts of N-compounds are transported in the xylem during the prebudbreak period (Campbell and Strother 1996). We have only limited information on the dynamics of N in roots for our experiment, but N-concentration at budbreak was similar in MP and VSP vines in the 1997 season (Figure 3⇓), and other data from the same field also show no clear differences.

• Download figure
• Open in new tab

Figure 3

Nitrogen and total nonstructural carbohydrate (TNC) dynamics of roots from VSP and MP vines as a function of phenological stage during the 1997 season. Values are means ± SE with n = 4.

The higher N-concentration observed in the aerial woody parts of MP as compared to the VSP vines at dormancy and budbreak (in most cases) may be due to differences in root size of the two pruning systems and therefore differences in the reserve-N pool. However, both the smaller (Rühl and Clingeleffer 1993) and the larger root systems (Sommer 1995) have been reported for MP compared with VSP vines. A fraction of nitrogenous compounds may also have been lost through bleeding sap (Glad et al. 1992) in the pruned as compared to the unpruned vines.

It is generally assumed that nitrogenous reserves for new shoot and leaf growth are more readily available from one-year-old wood than from the older wood fractions, and least available from roots (Schaller et al. 1989, Conradie 1990), and our data would support this (Figures 2⇑ and 3⇑). Depletion of N-reserves between budbreak and bunch closure and refilling of N-reserves between veraison and leaf fall was greater for MP than for VSP vines irrespective of year and wood fraction (Figure 2⇑). In grapevines and other woody perennials, N-reserves play a crucial role in supplying early season growth (Löhnertz 1988, Chen et al. 2004), but the importance of N as compared to carbohydrate reserves has recently been questioned (Zapata et al. 2004). We correlated changes in N in different wood fractions occurring during different phenological phases with the corresponding changes in leaf area for each training system in 2002 and 2003 (Figure 4⇓). For both years and all wood fractions, good correlations (r² = 0.65 to 0.99) were obtained between change in leaf area, Δ-LA (m² vine⁻¹), and change in %N, Δ-N (%), of the wood. The slope of the linear regression between Δ-LA and Δ-N was substantially higher for MP (−56 to −97) than for VSP (−28) because of greater changes in leaf area of MP versus VSP vines (Figure 4⇓). The reported contribution of N-reserves from the wood to leaf development until bloom in the field varies widely. For young pruned vines, it was between 14 and 26% in California (Araujo and Williams 1988) and between 25 and 87% in Germany (Löhnertz 1988, Schaller et al. 1989), with the latter value being the amount mobilized from one-year-old wood. In total this amounts to 1.1 to 2.3 g N/vine for a VSP system such as used in the present study in a cool-climate situation (Schaller et al. 1989). At bloom, leaf area for MP was 12 m²/vine on average over the years as compared with 3.6 m²/vine for VSP, which would indicate that roughly 3.6 to 7.6g N/vine would be needed from the wood fraction to support growth for MP vines. The portion of wood from which these N-reserves could be drawn in the MP system was estimated to be at least 10 times as large as the VSP system. Additionally, MP vines may have mobilized more N from the roots between bunch closure and leaf fall (Figure 3⇑), which may have been necessary to sustain the higher number of shoots and greater leaf area during summer and the higher amount of fruit formed by the MP system (Table 1⇓). Both have been shown to enhance N-depletion in storage organs of grapevines (Balasubrahmanyam et al. 1978).

• Download figure
• Open in new tab

Figure 4

Change in leaf area (Δ-LA) from budbreak to bloom, bloom to bunch closure, bunch closure to veraison, and veraison to leaf fall as a function of change in nitrogen concentration (Δ-N) for wood fractions of different age over two seasons (2002a–c and 2003d–f) for VSP and MP vines. For 2003, differences were calculated for budbreak to bunch closure because no bloom-time measurements were made (d–f). Linear regression lines are shown to underline tendencies.

N-retranslocation into MP wood and roots seemed to have started earlier or was more rapid than that into VSP wood because concentrations at leaf fall were higher for the wood fraction of MP vines (Figure 2⇑), which may be related to earlier leaf senescence frequently observed for these systems (Sommer 1995). During the senescence process, proteins are hydrolyzed and organelles are broken down to form nitrogenous compounds which are translocated back to the storage organs (Yang et al. 2001). Although leaves contain a large proportion of N, Williams (1987) found only small amounts that were retranslocated to the trunk and roots of Thompson Seedless vines, whereas other studies have shown that up to 40% of leaf N is recycled (Conradie 1990). Data from Schaller et al. (1989) suggest that substantial amounts of arginine from the leaves were contributing to the increase in N-reserves in one-year-old wood.

Over the years, a trend was observed for veraison N-concentrations to decrease for both systems and all wood fractions (Figure 2⇑). This decrease was greater for VSP than for MP vines in one-year-old wood. In general, the decrease in N-concentration may be related to plant water status (Figure 5⇓). Predawn water potential around veraison slowly decreased over the five-year study period and specifically from 2001 to 2003 (see arrows, Figure 5⇓). Since veraison has been identified as the second peak period for N-uptake from roots with the majority incorporated into the fruit and woody reserves (Löhnertz 1988, Bates et al. 2002), low water availability may have hampered this uptake. While this development may be explicable by reduced root uptake, it does not explain why MP vines had higher N-values despite lower water potentials and why these higher N-values were generally found up to leaf fall, the period with the lowest yearly water potential values in most years (Figure 5⇓). It could be argued that MP vines, because of their much larger leaf area (about three times that of VSP), transpired more during the early parts of the season when soil moisture was still adequate (Schmid and Schultz 2000), thereby also increasing whole-vine N-uptake. This may have increased leaf N-content (Falcetti et al. 1995) before soil water became depleted and leaf water potential decreased later during the season (Figure 5⇓). Low water potential may have accelerated leaf aging and senescence in older leaves and thus nutrient retrans-location into the wood (Yang et al. 2001). Additionally, we observed an average leaf area to fruit ratio (2001 to 2003) of about 20 cm²/g fruit for MP vines as compared to about 16 cm²/g fruit for VSP vines at maximum canopy development, which may have also influenced nutrient demand and partitioning into reserves (Murisier 1996).

• Download figure
• Open in new tab

Figure 5

Predawn leaf water potential during the growing seasons of 1999 to 2003 for VSP and MP vines of the cultivar Riesling. Arrows indicate date of veraison. Data are the mean ± SE (n = 8).

Carbohydrate reserves: seasonal dynamics.

Dry matter, total nonstructural carbohydrates (TNC), and the component sugars glucose, fructose, and sucrose decreased between dormancy and budbreak (Figures 6⇓, 7⇓, 8⇓, 9⇓, and 10⇓), whereas starch concentration tended to increase during this period in most of the studied wood fractions (Figure 11⇓). The increase in starch was probably related to assimilation from sugars associated with an increase in ambient temperature before budbreak (Eifert et al. 1961, Korkas et al. 1994), which replenished starch depleted for maintenance respiration in winter (Mooney and Gartner 1991). The magnitude of seasonal change in dry matter and TNC was greater in younger woody parts (Figures 6⇓ and 7⇓), which is possibly related to a decrease in availability of stored carbohydrates from older wood (Winkler and Williams 1945, Mooney and Gartner 1991). Dry weight was slightly higher and TNC was lower in older wood fractions during the dormant season, and both reached a minimum at bunch closure (Figures 6⇓ and 7⇓). The starch fraction had a similar minimum at bunch closure in one-and two-year-old wood in most years, but did not decrease in wood older than two years. This lack of change in starch in wood older than two years supports current belief that older wood has less remobilization capacity, possibly because of reduced activity of enzymes such as amylase, phosphorylase or acid phosphatase, and their isoenzymes (Loescher et al. 1990).

• Download figure
• Open in new tab

Figure 6

Evolution of dry matter (%) as a function of phenological stage (BBCH) of different wood fractions for VSP and MP vines over the course of three seasons (2001 to 2003) (D: dormancy; BB: budbreak; B: bloom; BC: bunch closure; V: veraison; LF: end of leaf fall). Values are means ± SE with n = 3. *, **, and *** indicate significance at p ≤ 0.05, 0.01, and 0.001, respectively.

• Download figure
• Open in new tab

Figure 7

Dynamics in total nonstructural carbohydrates (TNC) in different wood fractions as a function of phenological stage (BBCH) for VSP and MP vines over the course of three seasons (2001 to 2003) (abbreviations as in Figure 6⇑). Values are means ± SE with n = 3. *, **, and *** indicate significance at p ≤ 0.05, 0.01, and 0.001, respectively.

• Download figure
• Open in new tab

Figure 8

Changes in glucose concentration in different wood fractions as a function of phenological stage (BBCH) for VSP and MP vines over the course of three seasons (2001 to 2003) (abbreviations as in Figure 6⇑). Values are means ± SE with n = 3. *, **, and *** indicate significance at p ≤ 0.05, 0.01, and 0.001, respectively.

• Download figure
• Open in new tab

Figure 9

Changes in of fructose concentration in different wood fractions as a function of phenological stage (BBCH) for VSP and MP vines over the course of three seasons (2001 to 2003) (abbreviations as in Figure 6⇑). Values are means ± SE with n = 3. *, **, and *** indicate significance at p ≤ 0.05, 0.01, and 0.001, respectively.

• Download figure
• Open in new tab

Figure 10

Changes in of sucrose concentration in different wood fractions as a function of phenological stage (BBCH) for VSP and MP vines over the course of three seasons (2001 to 2003) (abbreviations as in Figure 6⇑). Values are means ± SE with n = 3. *, **, and *** indicate significance at p ≤ 0.05, 0.01, and 0.001, respectively.

• Download figure
• Open in new tab

Figure 11

Dynamics of starch concentration in different wood fractions as a function of phenological stage (BBCH) for VSP and MP vines over the course of three seasons (2001 to 2003) (abbreviations as in Figure 6⇑). Values are means ± SE with n = 3. *, **, and *** indicate significance at p ≤ 0.05, 0.01, and 0.001, respectively.

This seasonal pattern in carbohydrate dynamics with starch being the main component of TNC seems a general feature for grapevines and has been observed before (Winkler and Williams 1945, Eifert et al. 1961, Scholefield et al. 1978, Williams 1996, Hamman et al. 1996, Bates et al. 2002, Zapata et al. 2004). However, amounts in TNC and starch may differ substantially depending on climate or cultivation practices.

In this and other studies conducted under cool-climate conditions, maximum TNC values of 13 to 16% (during the dormant phase) and a minimum value of 3 to 6% at bunch closure were found (Eifert et al. 1961, Korkas et al. 1994). However, Winkler and Williams (1945) in California (cv. Grenache) and Scholefield et al. (1978) in Australia (cv. Sultana), reported much higher TNC values throughout the year in warmer climate conditions, suggesting that climatically favorable conditions may lead to a carbohydrate “surplus” (Mooney and Gartner 1991).

We found maximum starch concentrations of ~5.0% to 8.0% at budbreak and veraison and minimum concentrations of <1.0% to 3.0% at bunch closure for one-and two-year-old wood (Figure 11⇑). While TNC levels increased until leaf fall (Figure 7⇑), starch did not (Figure 11⇑), possibly because of conversion into glucose, fructose, and sucrose (Figures 8⇑, 9⇑, 10⇑). Eifert et al. (1961) also reported two peaks in starch concentration of perennial tissue of grapevines just before budbreak and during midripening when shoots lignified.

Concentrations in glucose, fructose, and sucrose decreased to almost the lowest values during the season at budbreak and increased again just before leaf fall (Figures 8⇑, 9⇑, 10⇑). This pattern confirms the bulk of data found in the literature (Winkler and Williams 1945, Mooney and Gartner 1991, Hamman et al. 1996), although actual absolute concentrations vary from study to study. The increase in sugars before leaf fall, usually attributed to frost acclimation (Hamman et al. 1996), was not in response to frost exposure in this study. In 2001 and 2002, the last samples were collected before the first frost (date of first frost: 9 Nov 2001; 8 Dec 2002; 24 Oct 2003). The ratio of [(Glucose + Fructose)/Sucrose], an indicator of frost hardiness, remained >2 throughout the seasons in every year, which is usually not reported for Vitis vinifera (Hamman et al. 1996).

Sucrose, the major transport form of carbohydrates in higher plants and grapevines (Swanson and El-Shishiny 1958, Williams 1996), reached a maximum at dormancy before budbreak and leaf fall in all years and in wood fractions of both training systems. The high concentrations at these phenological stages suggest that the periods before budbreak and leaf fall are active for transport out and into reserve tissue, respectively. Changes for glucose and fructose may be more related to conversion into and out of starch, respectively, at these times (Eifert et al. 1961).

Carbohydrate reserves: differences related to training system.

Dry weight was higher in general at bloom and bunch closure in one-year-old wood of MP compared with VSP vines. The opposite trend was observed in wood older than two years (Figure 6⇑). There was a combined effect of the year and pruning system noticeable for TNC, sugar, and starch concentrations. While glucose, fructose, and sucrose concentrations were similar in all wood fractions for both pruning systems in 2001 and 2002, they tended to be lower from budbreak through veraison in one-year-old wood of MP in 2003 (Figures 8⇑, 9⇑, 10⇑). Starch and TNC concentrations were also similar, with a trend toward lower values during leaf fall of MP (Figures 7⇑ and 11⇑). Old wood had higher TNC and starch concentrations for MP vines for most of the season in 2001 (Figures 7c⇑ and 11c⇑), but this trend seemed to gradually reverse during midsummer through the 2002 to 2003 seasons (Figures 7f,i⇑ and 11f,i⇑).

Rühl and Clingeleffer (1993) found no differences in sugars and starch on a percent dry weight basis between spur-pruned and minimally pruned Cabernet franc vines during the dormant season in one-year-old canes, but slightly higher sugar concentrations for MP vines in both old wood and roots. However, on a g/vine basis, spur-pruned vines had higher values for all carbohydrate fractions in one-year-old wood, lower starch and sugars in old wood, but higher starch in the roots as compared to MP vines of similar size. If vine size differs, as in the current study, even if concentration differences are small or absent, then the total amount of carbohydrates stored in roots and old wood is much higher for the MP vines (Sommer 1995).

Lower budbreak and leaf-fall values for TNC and starch in one-year-old wood of MP vines could have three explanations. One, carbohydrates mobilize to support spring leaf growth, since swelling buds are a sink for carbohydrates (Glad et al. 1992). Two, reproductive organs compete for reserve TNC in perennial storage tissue (Loescher et al. 1990), where minimal pruning because of higher yield (Table 1⇑) exerts a higher demand for carbohydrates. Three, above-ground carbohydrate dynamics are coupled to below-ground growth, and roots can be a large sink for carbohydrates (Bates et al. 2002) for storage and respiration (Williams 1996, Zapata et al. 2004). Assuming larger root systems for MP vines (Sommer 1995), increased demands for root growth may have caused this temporary depletion, since active root growth coincides with the phases of depletion reported here (Freeman and Smart 1976).

The decrease in dry weight with increasing wood age for the MP as compared with the VSP system may indicate a reduction in absolute carbohydrate storage capacity even at constant carbohydrate levels on a percentage basis. It is possible that shoot trimming of the VSP vines during the season in our study redirected partitioning of carbohydrates into permanent parts of the vine (Loescher et al. 1990).

TNC and starch concentrations in summer decreased slightly more for MP vines, especially in wood older than two years (Figures 7⇑ and 11⇑). This decrease may have been related to the level of water deficit present in 2002 and 2003 (Figure 5⇑). Predawn water potential between bunch closure and harvest ranged between −0.3 and −0.4 MPa in 2002 and between −0.4 and −0.6 MPa in 2003, with lower values for MP vines. This water stress can cause substantial inhibition of photosynthesis and may have necessitated an increase in carbohydrate mobilization from storage tissues such as wood and roots (Candolfi-Vasconcelos et al. 1994). However, in a pot study, Rühl and Alleweldt (1990) found an increase rather than a decrease in carbohydrate levels (on a % dry weight basis) in wood and root tissue under prolonged water deficit for most varieties because of the decreased demand of aboveground growth.

Contrary to other data showing a decrease in leaf area to fruit ratio for MP vines as compared with standard pruning (Clingeleffer 1984, Downton and Grant 1992, Kliewer and Dokoozlian 2005), we observed an increase (20 cm²/g fruit compared with ~16 cm²/g fruit). Whereas higher leaf area to fruit ratio has been positively correlated with the amount of carbohydrate stored in perennial tissues (Murisier 1996), our data did not show such an effect. The absence of such a correlation for MP vines is probably related to the larger proportion of shaded leaf area in MP canopies as compared with VSP canopies, since shade leaves contribute little to whole-plant carbon gain (Intrieri et al. 2001).

Despite the slightly lower reserve carbohydrate concentrations in perennial organs of MP vines, there was no limiting effect on N-accumulation by the roots (Loescher et al. 1990), since N was higher in MP wood even in the driest year (2003). From the available data, it can be inferred that there are no obvious limitations in terms of carbohydrate reserves for MP vines even in very dry years under cool-climate conditions.