Keywords: grape, H +, hue, K +, ph, wine

Walker and Blackmore K + concentrations and ph in grape juice and wine 183 Potassium concentration and ph inter-relationships in grape juice and wine of Chardonnay and Shiraz from a range of rootstocks in different environments_189 183..193 R.R. WALKER and D.H. BLACKMORE CSIRO Plant Industry, Waite Campus, PO Box 350, Glen Osmond SA 5064, Australia Corresponding author: Dr Rob Walker, fax +61 8 8303 8601, email rob.walker@csiro.au. Abstract Background and Aims: ph adjustment during winemaking is a significant cost to the Australian wine industry. This study addresses potassium (K + ) concentration and ph inter-relationships in grape juice and wine of Chardonnay and Shiraz. Methods and Results: Chardonnay and Shiraz on own roots, and on Ramsey, 1103 Paulsen, 140 Ruggeri, K51-40, Schwarzmann, 101-14, Rupestris St. George and 1202 Couderc were compared at Koorlong and Merbein (Victoria), and Padthaway, Nuriootpa and Rowland Flat (South Australia). Petiole K + concentrations at flowering were a poor indicator of grape juice and wine K + concentrations. The concentration of H + ions in grape juice and wine decreased as K + concentrations increased resulting in increased ph. The relationship between H + and K + concentrations was linear for Chardonnay but exponential for Shiraz, where K + concentrations were higher. Wine K + and grape juice K + concentrations exhibited a positive linear relationship, with slope for Chardonnay about half that for Shiraz, indicating a net loss of K + between grape juice and wine of 58% for Chardonnay and 13% for Shiraz. Conclusions: The study has linked higher wine ph to both higher juice soluble solids and K +, and to poorer wine colour hue. Loss of K + during fermentation and cold stabilisation appeared higher for Chardonnay than for Shiraz. Significance of the Study: Rootstocks that lead to lower K + concentrations and ph in grape juice and wine are identified. Differences in the K + concentration dynamics between grape juice and wine of Chardonnay and Shiraz are described and quantified. Keywords: grape, H +, hue, K +, ph, wine Introduction High soil potassium (K + ) contents are found in a number of Australia s major wine regions (Dundon et al. 1984) that can lead to relatively high grape juice K + concentrations. High K + concentrations in grape juice are linked to high grape juice ph (Somers 1975). In the 1970s, it was observed that Australian dry red wines had a higher ph range (3.4 4.3) compared with French wines (3.0 3.9) (Rankine 1977, Rankine et al. 1977), and this was linked with higher juice K + concentrations (Somers 1975). However, Godden and Gishen (2005) noted that mean ph of Australian dry red wines has fallen substantially from 1970s values because of adoption of better winemaking practice. Winemaking practice generally involves ph adjustment with tartaric acid to bring grape juice ph to within the range 3.0 3.3 for white wines and 3.3 3.5 for red wines (Iland et al. 2000), which adds significant additional cost to the winemaking process. Rootstocks differ in ability to accumulate K + (Cirami et al. 1984, Hedberg et al. 1986, Rühl et al. 1988, Rühl 1989, Kodur et al. 2010a,b), and low K + accumulating rootstocks provide an option to mitigate against the need for ph adjustment. The scion (Kodur et al. 2010b) and vineyard soil K + concentration (Rühl et al. 1992) are additional factors that may affect K + concentration in grape juice. Previous studies on rootstock effects on Shiraz wine composition, including K + and ph (Hale and Brien 1978, Cirami et al. 1984, Hedberg et al. 1986), have involved the use of low-salinity water for irrigation. To our knowledge, there have been few comprehensive investigations of rootstock effects on grape juice and/or wine K + concentrations and ph across a range of irrigation water and soil salinities. Irrigation-water electrical conductivities in the range 0.40 3.5 ds/m applied to Sultana grapevines on own roots and on seven different rootstocks at one site over a 5-year period had no effect on Sultana grape juice K + concentration (Walker et al. 2004) and ph (Walker et al. 2007). Just two exceptions were noted: rootstock 1103 Paulsen, where Sultana grape juice ph was reduced by 10% at 3.5 ds/m (Walker et al. 2004), and rootstock R3 (a Commonwealth Scientific and Industrial Research Organisation (CSIRO) hybrid), where Sultana grape juice ph increased by 2% at 3.5 ds/m (Walker et al. 2007). This relative consistency in grape juice K + concentration and ph with increasing salinity occurs despite reductions in grape berry size (Prior et al. 1992, Stevens et al. 1999, Walker et al. 2002) and significant increases in grape berry chloride (Cl - ) and sodium (Na + ) concentrations (Prior et al. 1992, Stevens et al. 1999, Walker et al. 2004). This suggests that grape juice K + and ph are more likely to be responsive to other site factors such as soil type and soil K + status, although rootstock type (Rühl 1991), variety type (Schaller et al. 1992) and canopy management, including influences on light and shade (Jackson and Lombard 1993), may interact with soil conditions to affect the outcome. In a companion study, we reported on yield performance and concentrations of Cl - and Na + in petioles, grape juice and doi: 10.1111/j.1755-0238.2012.00189.x

184 K + concentrations and ph in grape juice and wine Australian Journal of Grape and Wine Research 18, 183 193, 2012 wine of Chardonnay and Shiraz grown on a range of rootstocks in coastal and inland environments (Walker et al. 2010). In the study reported here, involving the same scions and rootstocks at the same four sites, that also differed in soil K + status, we investigated the inter-relationships between grape juice soluble solids, K + concentration and ph. Titratable acidities were also measured because ph is reported to be a function of acid concentrations, K + and Na + concentrations (Boulton 1980a). We also used Chardonnay and Shiraz wines from vines on own roots and from vines grafted to selected rootstocks to compare changes in K + concentration between grape juice and wine and to compare inter-relationships between wine K + concentration, ph, soluble solids and colour hue. Materials and methods Plant material and experimental design Of the sites used in this study, two were located in the Sunraysia region in northwest Victoria at Merbein (34 13 41 S, 142 2 38 E) and Koorlong (34 15 29 S, 142 8 1 E), and three were located in commercial vineyards in South Australia at Padthaway (36 38 45 S, 140 31 0 E Shiraz; 36 38 44 S, 140 30 42 E Chardonnay), Nuriootpa (34 29 26 S, 139 0 20 E) and Rowland Flat (34 35 47 S, 138 55 43 E) (Walker et al. 2010). The trials were planted in 1992 as complete randomised blocks with 1-year-old vines. Main treatments were scion varieties, i.e. Shiraz and Chardonnay (Vitis vinifera) on their own roots or grafted to rootstocks Ramsey (V. champinii), 1103 Paulsen (V. berlandieri V. rupestris), 140 Ruggeri (V. berlandieri V. rupestris), K51-40 (V. champinii V. riparia Gloire ), Schwarzmann (V. riparia V. rupestris), 101-14 Millardet et de Grasset, referred to in the text as 101-14 (V. riparia V. rupestris), Rupestris St. George (V. rupestris) and 1202 Couderc (V. vinifera V. rupestris), referred to in the text as 1202C. Chardonnay clone I10V1 was used at all sites. For Shiraz, clone AC72-8189 was used at Merbein and Koorlong, and clone BGVSS Cl.30 was used at the South Australian sites. There were six blocks at Merbein and Koorlong, and ten blocks at Padthaway, Nuriootpa and Rowland Flat. Each main treatment was replicated once and randomised within each block. Trellis type and spacings between rows and vines Trellis type and row and vine spacings at each site were detailed previously (Table 1, Walker et al. 2010). All vines were spurpruned to about 40 spurs where possible, leaving two buds per spur or approximately 80 buds per vine during winter dormancy. Soil Soil at Merbein was described as gradational yellow, calcareous clay; at Koorlong as gradational reddish, calcareous, mottled, sandy clay loam; at Padthaway as uniform, shallow, brownish, gravelly, calcareous mottled clay; at Nuriootpa as duplex, yellowish, mottled clay; and at Rowland Flat as duplex, reddish clay (Alf Cass and Cliff Hignett, unpubl. data, 1999). Soil samples were taken during the post-harvest periods at each site in 1997. Samples were taken at four depths at all sites with a hand-held auger, including a surface sample (0 10 cm), and at 30 40 cm, 60 70 cm and 90 100 cm. Electrical conductivity (EC) and sodium adsorption ratio (SAR) of 1:5 (w/v) soil : water extracts were determined and reported previously (Table 2, Walker et al. 2010). K + concentrations were determined using soil saturation extracts. At Merbein and Koorlong, there were four sampling sites in each of the Shiraz and Chardonnay. There were five sampling sites in each of the Shiraz and Chardonnay at Padthaway, and in the Chardonnay at Nuriootpa and Shiraz at Rowland Flat. Each sample was stored in a sealed plastic bag until processing and analysis. Weather data Weather data for the different sites (Table 1, Walker et al. 2010) were obtained from the closest Australian Bureau of Meteorology station, which was approximately 5 km away from the Merbein, Koorlong, Nuriootpa and Padthaway sites, and approximately 15 km from that of Rowland Flat. Table 1. Rootstock effects on K + concentration (g/100 g dw) in petioles of Chardonnay and Shiraz at flowering at Merbein (M) and Koorlong (K), Padthaway (P) and at the Barossa Valley sites, Nuriootpa (N) (Chardonnay) and Rowland Flat (RF) (Shiraz). Rootstock Chardonnay Shiraz M N P K M RF P K Own roots 3.82 a 3.47 b 3.35 d 3.66 bcd 2.92 c 2.57 e 3.51 d 3.22 c Ramsey 3.55 ab 3.12 c 3.62 cd 4.07 b 3.79 b 3.70 d 4.38 c 4.55 ab 1103 Paulsen 2.82 c 2.69 d 3.56 cd 3.26 de 3.00 c 3.86 cd 4.10 c 3.93 b 140 Ruggeri 3.68 ab 2.72 d 3.59 cd 4.80 a 4.45 a 3.86 cd 4.44 bc 4.68 a K51-40 3.58 ab 3.08 c 3.34 d 3.16 e 4.50 a 4.13 c 4.44 bc 4.33 ab Schwarzmann 3.52 ab 3.53 b 3.74 c 3.82 b 4.10 ab 4.54 b 4.92 ab 4.04 ab 101-14 3.70 ab 4.02 a 4.71 b 3.72 bc 4.04 ab 4.68 b 5.21 a 4.31 ab Rupestris St. George 3.16 abc 4.06 a 5.12 a 3.78 b 3.82 b 5.13 a 5.40 a 4.42 ab 1202 C 3.10 bc 3.06 c 2.76 e 3.33 cde 4.17 ab 3.84 cd 4.26 c 3.94 b LSD R 0.47* 0.27* 0.24* 0.31 0.49* 0.21* 0.36* 0.48* LSC R Y 0.67** ns 0.34*** 0.44*** ns 0.30*** 0.51** 0.68*** *P < 0.05; **P < 0.01; ***P < 0.001. Values are means of two seasons, 1996 and 1997. Different superscript letters indicate significant differences between rootstock means. LSD R Y is used where the R Y interaction is significant. Bold type highlights higher values. LSD, least significant difference; ns, not significant; R, between rootstock means; R Y, rootstock year interaction.

Walker and Blackmore K + concentrations and ph in grape juice and wine 185 Table 2. Rootstock effects on total soluble solids (TSS) ( Brix), K + concentration (mmol/l) and ph of grape juice at harvest of Chardonnay and Shiraz at Merbein, Barossa Valley (Chardonnay Nuriootpa, and Shiraz Rowland Flat), Padthaway and Koorlong. Parameter Chardonnay Shiraz TSS K + ph TSS K + ph Merbein (2.1 ds/m) Own roots 23.3 c 44.4 ab 3.57 c 26.7 bc 57.5 b 3.67 b Ramsey 23.8 abc 47.4 ab 3.65 ab 26.2 bc 58.5 ab 3.77 ab 1103 Paulsen 24.0 abc 47.3 ab 3.59 abc 28.6 a 62.2 ab 3.81 a 140 Ruggeri 23.8 abc 40.8 b 3.58 bc 27.0 b 56.8 b 3.72 ab K51-40 23.7 bc 50.9 a 3.58 bc 26.7 bc 66.5 a 3.74 ab Schwarzmann 24.2 abc 47.4 ab 3.66 a 26.1 bc 57.1 b 3.76 ab 101-14 24.7 a 47.5 ab 3.64 abc 26.9 bc 56.0 bc 3.74 ab Rupestris St. George 24.4 ab 46.0 ab 3.63 abc 27.4 ab 55.7 bc 3.73 ab 1202C 23.4 c 45.8 ab 3.49 d 25.4 c 48.8 c 3.49 c LSD R 0.6* 5.9* 0.07* 1.5* 7.9* 0.11* LSD R Y 0.9*** 8.3* ns ns ns ns Barossa Valley Nuriootpa (1.8 ds/m) Rowland Flat (3.3 ds/m) Own roots 24.8 cd 55.1 d 3.64 bcd 26.4 d 70.5 c 3.89 d Ramsey 24.7 cd 60.6 bc 3.70 ab 27.1 bcd 75.6 bc 4.02 c 1103 Paulsen 24.4 d 58.0 cd 3.64 bcd 28.0 ab 79.7 b 4.04 bc 140 Ruggeri 25.0 c 55.6 d 3.63 cd 28.3 a 80.1 b 4.05 bc K51-40 25.6 ab 65.2 a 3.74 a 28.5 a 92.2 a 4.11 ab Schwarzmann 25.0 c 61.2 bc 3.69 abc 26.8 cd 76.5 b 4.02 c 101-14 25.8 a 63.5 ab 3.70 ab 27.7 abc 74.8 bc 4.01 c Rupestris St. George 25.2 bc 66.2 a 3.74 a 27.6 abc 87.4 a 4.15 a 1202C 25.2 bc 55.4 d 3.61 d 27.0 bcd 75.0 bc 3.91 d LSD R 0.5* 3.2* 0.06* 1.0* 5.5* 0.07* LSD R Y ns ns ns ns ns ns Padthaway (2.5 ds/m) Own roots 23.6 c 54.8 bc 3.55 d 23.6 a 52.4 ab 3.48 b Ramsey 24.1 bc 54.6 bc 3.59 bcd 23.1 a 51.5 ab 3.52 ab 1103 Paulsen 23.6 c 58.8 b 3.62 b 22.8 a 51.5 ab 3.49 b 140 Ruggeri 23.9 bc 54.4 bc 3.56 cd 23.5 a 53.4 ab 3.53 ab K51-40 24.1 bc 58.2 bc 3.61 bc 23.2 a 52.4 ab 3.48 b Schwarzmann 24.3 ab 55.5 bc 3.59 bcd 23.6 a 52.1 ab 3.51 ab 101-14 24.8 a 64.6 a 3.68 a 23.6 a 54.5 a 3.52 ab Rupestris St. George 24.2 b 56.9 bc 3.69 a 23.4 a 54.7 a 3.57 a 1202C 24.1 bc 53.9 c 3.55 d 23.4 a 49.6 b 3.46 b LSD R 0.4 3.4* 0.05* 1.0* 2.7* 0.05* LSD R Y 0.5** 4.7*** ns 1.5* 3.8* 0.07** Koorlong (0.4 ds/m) Own roots 23.6 cd 45.6 b 3.49 bc 24.3 bc 48.8 cd 3.52 cd Ramsey 23.2 d 48.8 a 3.56 a 24.5 bc 56.5 a 3.68 a 1103 Paulsen 24.2 abc 50.3 a 3.56 a 24.8 abc 52.0 abcd 3.56 bcd 140 Ruggeri 24.8 a 48.1 ab 3.56 a 24.8 abc 53.5 abc 3.63 ab K51-40 23.5 cd 51.2 a 3.53 ab 25.8 a 56.1 a 3.68 a Schwarzmann 23.8 bcd 49.6 a 3.52 ab 24.2 bc 49.2 bcd 3.57 bc 101-14 24.4 ab 50.1 a 3.56 a 23.9 c 55.3 ab 3.62 ab Rupestris St. George 24.8 a 48.2 ab 3.55 a 25.3 ab 47.0 d 3.47 d 1202C 23.8 bcd 40.5 c 3.44 c 24.6 abc 46.5 d 3.47 d LSD R 0.5* 3.1* 0.05* 1.2* 6.1* 0.09* LSD R Y 0.7** ns ns ns ns ns *P < 0.05; **P < 0.01; ***P < 0.001. EC iw at each site is in brackets. Values are means of two seasons, 1996 and 1997. Different superscript letters indicate significant differences between rootstock means. LSD R Y is used where the R Y interaction is significant. Bold type highlights higher K + values. EC, electrical conductivity. LSD, least significant difference; ns, not significant; R, between rootstock means; R Y, rootstock year interaction.

186 K + concentrations and ph in grape juice and wine Australian Journal of Grape and Wine Research 18, 183 193, 2012 Irrigation practices Irrigation was applied at all sites via drippers, with spacings, dripper flow rates and total amounts of water applied detailed previously (Table 1, Walker et al. 2010). At Merbein, the desired salinity of the irrigation water was achieved by the addition of thoroughly mixed concentrated brine via a fertiliser injector pump to the main irrigation pipe containing River Murray water (Walker et al. 2010). Soluble nutrients, principally calcium nitrate (Ca(NO 3) 2) at 31 mmol/l and potassium nitrate (KNO 3) at 139 mmol/l were also added to the River Murray water using fertiliser injector pumps. The final salinity of the irrigation water was 2.1 ds/m with a SAR of approximately 10 (Walker et al. 2010). At Koorlong, vines were irrigated with water from the River Murray. Soluble nutrients (Ca(NO 3) 2 and KNO 3) were added via fertiliser injector pumps. There were no additions of NaCl. Final EC iw was 0.4 ds/m (Walker et al. 2010). Soil matric potential at Merbein and Koorlong was assessed with tensiometers (Irrometer Company, Inc., Riverside, California, USA) installed in the drip line at 0.3, 0.6 and 0.9 m below the surface. Irrigations were scheduled to ensure that soil matric potential between 0.3 and 0.9 m was kept above -30 kpa. A leaching fraction was applied at each irrigation (Stevens and Walker 2002). At Padthaway, soil water monitoring to assist irrigation scheduling for Chardonnay and Shiraz vines was carried out using a capacitance probe (EnviroSCAN RT-5, Sentek Sensor Technologies, Stepney, South Australia, Australia), with the lowest sensor at 90 cm depth (Chardonnay) and 100 cm depth (Shiraz) within the drip line. Neutron moisture probes (ICT International, Armidale, New South Wales, Australia) were also used, with access tubes to 100 cm depth. For Chardonnay vines, irrigation was not normally applied until approximately the middle of November and thereafter once the readily available water (RAW) in the soil profile had been used (Table 2, Tregeagle et al. 2006). For Shiraz vines, irrigation was not normally applied until the middle of December, and with the exception of a period of regulated deficit irrigation between fruitset and veraison, irrigations were applied thereafter once the RAW had been used. Winter rainfall was relied on at Padthaway to provide adequate leaching of salts from the soil profile. The Padthaway site was irrigated with bore water, which had a mean EC across both seasons (1996 and 1997) of 2.5 ds/m and a SAR of 4.7. At Nuriootpa and Rowland Flat, vines were also irrigated with bore water. A cap of 1 mega L/ha restricted the total amount of irrigation water that could be applied per season. Different bores were used for each site, delivering irrigation water with a mean EC across both seasons (1996 and 1997) of 1.8 ds/m at Nuriootpa and 3.3 ds/m at Rowland Flat. At both sites, irrigation was not normally applied until the end of December. Thereafter, it was applied mainly in the period post-veraison to harvest and in the post-harvest period taking care to ensure that total allocations remained within the cap. Composition of irrigation water Mean concentrations of Cl -,Na +,K +, magnesium (Mg 2+ ) and calcium (Ca 2+ ) in the irrigation water at the various sites for 1996 and 1997 have been presented previously (Walker et al. 2010). Petiole sampling Leaves were sampled at flowering at each site in each year. Twenty leaves were sampled per grapevine, ten each from opposite sides of the vine. Of these ten, five were sampled from each side of the trunk. Leaves were randomly selected opposite bunches from approximately nodes 2 5. Each fresh leaf was separated into lamina and petiole. Samples of 20 petioles per vine were washed and rinsed then processed, as described by Walker et al. (2010). Samples were then stored until analysis. Harvest scheduling Trials were harvested when juice soluble solid concentrations were as close as possible to 22 Brix for Chardonnay and as close as possible to 24 Brix for Shiraz. Harvest dates at the various sites have been listed previously (Walker et al. 2010). Samples of 80 berries per vine were taken at all sites on the day before harvest. Each 80 berry sample was obtained by taking 20 berries from each quarter of the vine (left and right of the main trunk, and on each side); five berries were sampled from four bunches in each quarter, making a total of 16 bunches sampled per vine; for each bunch, two berries were sampled from the shoulders of the bunch, two from the middle and one from the tail. All berry samples were placed in plastic bags and held in an ice-cooled insulated container for transport to the laboratory. Berry samples were gently crushed with mortar and pestle to extract juice. The juice samples were then centrifuged (Beckman J25i, Beckman Coulter, Inc., Fullerton, California, USA) for 15 min at 4 C, at 5930 g. The clear juice was decanted and used immediately for measurement of total soluble solids, ph and titratable acidity. The samples were then frozen for later analysis of K +. Total soluble solids, ph and titratable acidity Fresh grape juice samples were analysed for total soluble solids ( Brix) with a digital refractometer (PR-1, Atago Co., Ltd, Tokyo, Japan). Measurement of ph and titratable acidity was made using an autotitrator (Radiometer Analytical A/S, Copenhagen, Denmark). Fermentation Small-scale wines were made from Chardonnay and Shiraz on own roots and on Ramsey, 1103 Paulsen, 140 Ruggeri, K51-40 and 1202C rootstocks at Merbein, and the Barossa Valley sites in 1996 and at all sites in 1997. The must fermentation procedure was used for Shiraz and the juice procedure for Chardonnay. Duplicate fermentations of 15 L of juice or must were made. For each winemaking replicate, grapes were crushed on day 0 in a crusher/destemmer (Amos, Heilbronn, Germany). Pressing was done with a Willmes airbag press (Joseph Willmes, Hessen, Germany). Dry yeast, strain SIHA 4 (Getränkeschutz GmbH, Nahe, Germany) was used at the rate of 0.2 g/l, and sulfur dioxide (SO 2) (as sodium metabisulfite, 10% solution, 1 ml/l) and diammonium hydrogen orthophosphate (DAP, 0.5 g/l) were added at times indicated later. This rate of DAP is slightly higher than the usual rate (0.1 0.2 g/l) but has, in our experience, no observable effects on ph, ionised anthocyanins or colour hue. Post-fermentation treatment of the wine was as described by Kerridge (1983). Must fermentations. The crush was placed into wide-mouth 20-L polyethylene containers. SO 2 was added soon after crushing, and 4 5 h later after warming to room temperature, dry yeast and DAP were added. The must, placed at 25 C, was stirred twice daily. On day 3, the must was pressed, and the liquid was placed in 9 L glass containers to continue fermentation at 25 C.

Walker and Blackmore K + concentrations and ph in grape juice and wine 187 Juice fermentations. The crushed grapes were pressed on day 0; SO 2, Polyclar (polyvinylpolypyrrolidone, 0.2 g/l) and Clariphase GL (liquid pectolytic enzyme, 0.05 ml/l) were added. The juice was left to settle for 3 days at 4 C, was then racked and was placed in 9-L glass containers. DAP and Celite (diatomaceous earth, 0.2 g/l), and, after warming to room temperature for 4 5 h, dry yeast were added. The containers were held overnight at room temperature, then placed in a 18 C fermentation room. K + analysis Grape juice and wine samples. Concentrations of K + were measured with an atomic absorption (AA) spectrophotometer (Model 1200, Varian Techtron Pty Ltd, Melbourne, Victoria, Australia). Directly prior to analysis, the frozen grape juice samples were defrosted in a water bath at 34 C for 10 min. Aliquots (1 ml) of grape juice were then diluted to 10 ml by addition of deionised water. The AA was optimised (flame and burner height and orientation) and then calibrated with a standard solution of KCl to establish a standard curve in an appropriate range for the samples. The top standard and blank were checked after every ten samples to account for drift. If the value had changed by greater than 10%, a new calibration curve was run. Dried, powdered petiole samples. Nitric acid extracts (3.5%) of the samples were analysed for concentrations of K + by inductively coupled plasma optical emission spectrometry (ICP- OES) (Spectroflame ICP, Spectro Analytical Instruments, Kleve, Germany). Calibration of the ICP was carried out at the commencement of each analysis run using a 3.5% HNO 3 acid blank solution and a multi-element standard containing 1000 ppm K +. After every ten samples, calibration of the instrument was verified with a multi-element standard containing 50 ppm K +. When the instrument drift exceeded high or low limits, recalibration was undertaken with the blank and multi-element standard. Soil saturation extracts. Saturation extracts were obtained from saturation pastes using an automatic extractor procedure described by Rayment and Higginson (1992), Method 14A2. Saturation extracts were clarified by passing through a 0.450-mm Millipore filter prior to analysis. The clarified sample was analysed by ICP-OES, as outlined earlier. Wine colour hue Colour hue of the red wines was evaluated 2 3 months after bottling according to the method of Somers and Evans (1977) using a spectrophotometer (DMS 90, Varian Australia Pty Ltd, Melbourne, Victoria, Australia). Statistical methods One-way analysis of variance was applied to the grape juice and wine K +, ph and titratable acidity data (mean values for the two seasons) for each scion variety, with rootstocks as the main source of variation at each site using Genstat5 Release 3.1 (VSN International, Hemel Hempstead, UK) (Payne et al. 1993). Where significant (P < 0.05) within site rootstock effects were found, comparison between means at that site were made using the Fisher s protected least significant difference test with 5% level of significance or better. Regression techniques were applied to examine relationships between grape juice and wine H + ion concentrations and K + concentrations, wine K + and grape juice K + concentrations, grape juice K + and grape juice soluble solids, wine ph and grape juice soluble solids, and between wine colour hue and wine ph. Results Soil K + K + concentration (mean standard error of the sampling sites and depths at each sampling site; mmol(+)/l) in soil saturation paste extracts for the various site locations were Nuriootpa (0.15 0.03), Padthaway Shiraz site (0.24 0.04), Padthaway Chardonnay site (0.27 0.02), Merbein (0.30 0.05), Koorlong (0.33 0.02) and Rowland Flat (0.65 0.22). Petiole K + concentrations at flowering Highest petiole K + concentrations for Chardonnay were recorded on own roots at Merbein, on Rupestris St. George and 101-14 at Nuriootpa, on Rupestris St. George at Padthaway, and on 140 Ruggeri at Koorlong. Highest concentrations for Shiraz were recorded on K51-40 and 140 Ruggeri at Merbein, on Rupestris St. George at Rowland Flat, on Rupestris St. George and 101-14 at Padthaway, and on 140 Ruggeri at Koorlong (Table 1). Lowest concentrations for Chardonnay were recorded on 1103 Paulsen at Merbein, on 1103 Paulsen and 140 Ruggeri at Nuriootpa, on 1202C at Padthaway, and on K51-40 at Koorlong. Lowest concentrations for Shiraz were recorded for own roots at all sites (Table 1). Grape juice total soluble solids and titratable acidity at harvest For Chardonnay, average soluble solid content at harvest across the sites was in the range 23.2 25.8 Brix, with highest values recorded at Nuriootpa (24.4 25.8 Brix) (Table 2). Titratable acidity was in the range 4.5 6.1 g/l. In this case, highest values were recorded at the Merbein and Koorlong sites (5.0 6.1 g/l), and lowest values were recorded at the Nuriootpa and Padthaway sites (4.5 5.2 g/l) (detailed rootstock data not shown). For Shiraz, average total soluble solid content at harvest across the sites was in the range 22.8 28.6 Brix, with highest values recorded at the Merbein and Rowland Flat sites (25.4 28.6 Brix) and lowest values recorded at Padthaway (22.8 23.6 Brix) (Table 2). Titratable acidity was in the range 3.9 5.3 g/l, in this case with highest values recorded at Padthaway (4.9 5.3 g/l), lowest values at Koorlong and Merbein (3.9 4.5 g/l) and Rowland Flat intermediate (4.3 4.9 g/l) (detailed rootstock data not shown). Grape juice K + concentrations and ph at harvest For Chardonnay, mean harvest time grape juice K + concentrations for seasons 1996 and 1997 were in the range 40.5 66.2 mmol/l (Table 2). Highest concentrations were recorded at Padthaway and Nuriootpa (range 53.9 66.2 mmol/ L), and lowest values at Koorlong and Merbein (range 40.5 51.2 mmol/l) (Table 2). Chardonnay on K51-40 rootstock had higher juice K + concentrations than Chardonnay on 140 Ruggeri at Merbein and Nuriootpa (Table 2). Mean harvest time grape juice ph for Chardonnay for the two seasons was in the range 3.44 3.74, with highest values mostly recorded at the Nuriootpa site (range 3.61 3.74) (Table 2). For Shiraz, mean harvest time grape juice K + concentrations for seasons 1996 and 1997 were in the range 46.5 92.2 mmol/l (Table 2). Highest concentrations were recorded at Rowland Flat (range 70.5 92.2 mmol/l) (Table 2). Shiraz on K51-40 rootstock had higher juice K + concentrations than Shiraz on 140 Ruggeri at Merbein and Rowland Flat. Mean harvest time grape juice ph for Shiraz for the two seasons was also highest at the

188 K + concentrations and ph in grape juice and wine Australian Journal of Grape and Wine Research 18, 183 193, 2012 Rowland Flat site (range 3.89 4.15). Grape juice ph for Shiraz on own roots and 1202C at Rowland Flat (range 3.89 3.91) was lower than that of Shiraz on the other rootstocks at that site (range 4.01 4.15). At the other sites, Shiraz grape juice ph was generally in the range 3.46 3.81, with the ph of Shiraz juice at Padthaway having the lowest values (range 3.46 3.57) (Table 2). Regressions of grape juice K + concentration on petiole K + concentration at flowering at each site for each scion (eight regressions in total) showed a poor relationship (data not shown). Coefficients of determination (R 2 ) ranged from 0.003 to 0.37. None were significant at P < 0.05. Grape juice H + concentration decreased with increasing grape juice K + concentration for both Chardonnay and Shiraz (P < 0.001), with R 2 of 0.51 for Chardonnay (Figure 1a) and 0.89 for Shiraz (Figure 1b). The relationship for Chardonnay was linear (slope 0.005) (Figure 1a), whereas for Shiraz, the relationship was exponential (Figure 1b). Wine K +, ph and titratable acidity K + concentrations in wine of Chardonnay and Shiraz on K51-40 rootstock were higher than in wine from vines on own roots and from all other rootstocks at the Merbein and Barossa Valley sites (Table 3). However, there was no significant difference in ph of Chardonnay wine from own roots and grafted vines at Nuriootpa and Merbein. Furthermore, there was no difference between ph of Shiraz wine from K51-40, 140 Ruggeri and 1103 Paulsen rootstocks at Rowland Flat, or at Merbein (Table 3). Wine H + concentration decreased with increasing wine K + concentration for both Chardonnay and Shiraz (P < 0.001), with R 2 of 0.56 for Chardonnay (Figure 1c) and 0.81 for Shiraz (Figure 1d). The relationship for Chardonnay was linear (slope 0.0014) (Figure 1c), whereas for Shiraz, the relationship was exponential (Figure 1d). Wine titratable acidity was higher at Merbein (range 5.88 6.92 g/l) than at the Barossa Valley site (range 5.20 5.51 g/l) Figure 1. Regression of grape juice H + (mmol/l) on grape juice K + (a,b) and of wine H + on wine K + (c,d) for Chardonnay (left, open symbols) and Shiraz (right, closed symbols). Grape juice data are means for seasons 1996 and 1997. Wine data are for season 1997. Values for grape juice H + and K + are means of 6 10 replicate vines (depending on site) for own roots and all rootstocks, and wine H + and K + values are means for two winemaking replicates for own roots and five rootstocks. Left side y-axis shows H + and right side y-axis shows equivalent ph values. Sites are distinguished by different symbols as follows: Merbein (circle), Barossa Valley sites (squares), Padthaway (upright triangle) and Koorlong (inverted triangle). Chardonnay grape juice K + : y = 0.005x - 4.35 10-6 (R 2 = 0.51); Shiraz grape juice K + : f = 0.0027exp(-0.437x) (R 2 = 0.89); Chardonnay wine K + : y = 0.0014x - 3.64 10-5 (R 2 = 0.56); Shiraz wine K + : f = 0.0035exp(-0.0554x) (R 2 = 0.81); (P < 0.001 for all four regressions).

Walker and Blackmore K + concentrations and ph in grape juice and wine 189 Table 3. Rootstock effects on K + concentration (mmol/l), ph and titratable acidity (TA) (g/l) of wine of Chardonnay and Shiraz at Merbein and at the Barossa Valley sites (Nuriootpa Chardonnay, and Rowland Flat Shiraz). Parameter Chardonnay Shiraz K + ph TA K + ph TA Merbein (2.1 ds/m) Own roots 20.3 bc 3.24 ab 6.19 c 46.7 bc 3.58 b 5.87 b Ramsey 20.7 bc 3.33 a 6.33 c 43.1 d 3.63 b 5.67 c 1103 Paulsen 21.5 b 3.30 a 6.27 c 47.4 b 3.80 a 5.90 b 140 Ruggeri 17.1 d 3.29 ab 5.88 d 44.3 cd 3.73 a 5.77 bc K51-40 24.1 a 3.31 a 6.92 a 62.1 a 3.74 a 5.76 bc 1202C 19.3 c 3.20 b 6.62 b 31.7 e 3.30 c 6.59 a LSD R 1.0* 0.06* 0.14* 2.1* 0.05* 0.11* LSD R Y 1.4** 0.09** 0.19* 3.0*** 0.07* 0.16** Barossa Valley Nuriootpa (1.8 ds/m) Rowland Flat (3.3 ds/m) Own roots 22.4 d 3.47 5.27 cd 50.1 c 3.76 d 6.15 b Ramsey 26.0 b 3.58 5.51 a 54.5 bc 3.99 b 6.19 b 1103 Paulsen 23.8 cd 3.59 5.47 ab 58.7 b 4.03 ab 6.24 b 140 Ruggeri 22.6 d 3.53 5.42 abc 56.1 b 4.06 a 6.12 b K51-40 30.7 a 3.66 5.20 d 70.8 a 4.05 a 6.70 a 1202C 25.4 bc 3.49 5.31 bcd 56.2 b 3.89 c 6.21 b LSD R 1.1* ns 0.19* 3.4* 0.03* 0.20* LSD R Y 1.6*** ns ns 4.8* 0.04*** 0.28** *P < 0.05; **P < 0.01; ***P < 0.001. EC iw at each site is in brackets. Values are means of two seasons, 1996 and 1997. Different superscript letters indicate significant differences between rootstock means. LSD R Y is used where the R Y interaction is significant. Bold type highlights higher K + values. EC, electrical conductivity; LSD, least significant difference; ns, not significant; R, between rootstock means; R Y, rootstock year interaction. for Chardonnay but vice versa for Shiraz (range 6.12 6.70 g/l at Barossa Valley and 5.67 6.59 g/l at Merbein) (Table 3). Wine K + in relation to grape juice K + There was a positive linear relationship between wine K + and grape juice K + for both Shiraz and Chardonnay, with R 2 values of 0.80 (Chardonnay) and 0.86 (Shiraz) (Figure 2). However, it was not a 1:1 relationship for either variety. The slope of the relationship for Chardonnay (0.40) was about half that for Shiraz (0.89) (Figure 2). Relationship between grape juice K + and wine ph with grape juice soluble solids Positive linear relationships existed between grape juice K + and grape juice soluble solids (Figure 3a,b), with R 2 values of 0.65 (Chardonnay) and 0.71 (Shiraz), and between wine ph and grape juice soluble solids (Figure 3c,d), with R 2 values of 0.56 (Chardonnay) and 0.68 (Shiraz). The slope of the relationship in each case was similar between scion varieties (Figure 3). Wine colour hue and ph inter-relationship For Shiraz wine made from grapes on own roots and on Ramsey, 1103 Paulsen, 140 Ruggeri, K51-40 and 1202C rootstocks at each site in 1997, wine colour hue (based on means from vines on own roots and all rootstocks) was positively correlated with wine ph in all cases, with R 2 values of 0.82 (Merbein, Figure 4a), 0.77 (Koorlong, Figure 4b), 0.90 (Rowland Flat, Figure 4c) and 0.71 (Padthaway, Figure 4d). Additions of SO 2 to musts were considered sufficient to prevent oxidative reactions that may have contributed to wine colour hue, with no correlation observed between residual-free SO 2 in the wine at bottling and either wine ph of Chardonnay and Shiraz or with Shiraz wine colour hue (data not shown). Discussion Grape juice K + concentrations were highest at Nuriootpa and Padthaway for Chardonnay and at Rowland Flat for Shiraz. For Shiraz, this reflected the higher soil K + at Rowland Flat and the reported increase in juice K + with higher levels of K + fertilisation (Morris et al. 1980, Rühl et al. 1992). However, for Chardonnay, K + concentrations in soil saturation extracts post-harvest and overall mean K + concentration in petioles at flowering (3.31, 3.44, 3.73 and 3.75 g/100 g dw for Nuriootpa, Merbein, Koorlong and Padthaway, respectively) were lowest at Nuriootpa than at the other sites, which does not reconcile with the higher grape juice K + concentrations and ph at this site relative to that at Merbein and Koorlong. Lack of a consistent relationship between soil K + and grape K + has also been observed by Dundon et al. (1984). Soil samples were taken post-harvest, and it cannot be ruled out that soil K + concentrations were different at other times in the season. Furthermore, remobilisation of K + from vegetative organs to the berries after veraison is known to occur (Smart et al. 1985, Williams and Biscay 1991), and it is possible that varietal differences in the extent of remobilisation may be a factor in the observed differences. K + concentration in petioles at flowering was a poor indicator of K + concentrations in grape juice and wine. This was in

190 K + concentrations and ph in grape juice and wine Australian Journal of Grape and Wine Research 18, 183 193, 2012 Figure 2. Regression for wine K + concentration on grape juice K + concentration for Shiraz (closed symbols) and Chardonnay (open symbols) for season 1997. Values for grape juice K + are means of 6 10 (depending on site) replicate vines for own roots and five rootstocks, and values for wine K + are means of two winemaking replicates also for own roots and the same five rootstocks. Sites are distinguished by different symbols, as described for Figure 1. Merbein (circle), Barossa Valley sites (squares), Padthaway (upright triangle) and Koorlong (inverted triangle). Chardonnay: y = 0.40x + 1.36 (R 2 = 0.80); Shiraz: y = 0.89x - 1.54 (R 2 = 0.86); (P < 0.001 for both regressions). contrast with the observed correlation between K + concentration in petioles (opposite inflorescences at flowering) of rootstocks growing as ungrafted vines in the CSIRO germplasm collection at Merbein (Victoria), with grape juice ph of Chardonnay and Ruby Cabernet from rootstock/scion trials at Nuriootpa and Loxton (Rühl 1993). This may be related to wider range in petiole K + concentrations (1 5 g/100 g dw) (Rühl 1993) compared with 2.7 5.1 g/100 dw in this study (Table 1). Rootstock K51-40 led to highest K + concentration in grape juice and wine at the Merbein and Barossa Valley sites. These sites also resulted in highest Cl - concentrations in grape juice and wine (Tregeagle et al. 2006, Walker et al. 2010), with Chardonnay and Shiraz on K51-40 in particular having the highest or equal highest Cl - concentrations (Tregeagle et al. 2006, Walker et al. 2010). Regressions of grape juice K + on grape juice Cl - and of wine K + on wine Cl - were not significant (data not shown), supporting observations made in an earlier study on Sultana by Walker et al. (2004) where grape juice K + concentrations were unaffected by soil salinity. At Padthaway, K + concentrations in grape juice were highest on 101-14 rootstock. Rootstock 101-14 was shown to result in higher K + accumulation in vegetative tissues than 140 Ruggeri in glasshouse studies (Kodur et al. 2010a,b). While partially supported here by highest (Padthaway) or equal highest (Merbein, Nuriootpa, Koorlong) K + concentrations in grape juice of Chardonnay on 101-14 rootstock, it was not supported for Shiraz. The relative rankings of rootstocks in terms of K + accumulation in grape juice revealed few consistencies with other studies. For example, Hale and Brien (1978) found that grape berry K + was higher for Shiraz on Ramsey rootstock than for Shiraz on own roots. Rühl et al. (1988) found no difference between Shiraz on Ramsey and Shiraz on own roots in grape juice K + concentrations, and in our study, a significant difference was found only at the Koorlong site. Similarly, for Chardonnay, Rühl et al. (1988) found that own-rooted vines had the highest grape juice K + concentrations, whereas in our study, this was not the case. The reduction in H + concentrations in grape juice and wine as K + concentrations increased, with resultant increase in grape juice and wine ph, agrees with the established relationship between ph and K + concentration in grape juice and wine, as reported by Somers (1975). Wine ph has also been reported to be a function of acid concentrations, K + and Na + concentrations (Boulton 1980a), but we found no correlation between H + and Na + concentrations for Chardonnay grape juice and wine (R 2 = 0.05 and 0.09, respectively), no correlation for Shiraz wine, and only a weak correlation for Shiraz grape juice (R 2 = 0.36). The exponential relationship between H + and K + concentrations in Shiraz grape juice and wine suggests a more moderated reduction in H + concentration, as grape juice and wine K + concentrations exceed approximately 55 70 mmol/l. Hale (1977) also reported that grapes with higher K + concentration had a higher titratable acidity and ph. It was further noted that berries from Sultana vines grafted to Dog Ridge and Ramsey rootstocks had significantly higher K + concentrations and malic acid concentrations at harvest than Sultana on own roots. It was speculated that the higher ph was probably related to the higher K + concentration and to the greater proportion of malic acid, which is a weaker acid than tartaric (Hale 1977). Instances of higher wine K + concentration but no increase in ph, for example, with Chardonnay on K51-40 rootstock at Merbein and Nuriootpa, are also probably due to the fact that in addition to the effect of K +,ph also depends on the total acidity and the relative proportions of the major organic acids (Boulton 1980b). In situations of high grape juice K + and therefore high ph, this highlights the potential cost of acid adjustment required to return juice ph to acceptable levels. For example, for a 1000-L juice sample with a starting ph of 3.8, approximately 1.5 kg of tartaric acid would be required to bring juice ph to 3.5 (P. Rogers, pers. comm., 2011). Industry wide in Australia, at current costs of tartaric acid, the total cost of ph adjustment is likely to be in the millions of dollars per annum. The relationship between wine K + concentration and grape juice K + concentration for Shiraz was slightly less than 1:1 (0.87), while for Chardonnay, it was approximately 50% lower (0.42). Using grape juice to wine extraction rates for Cl - during the same winemaking process for these same samples in the same years (Walker et al. 2010), there was a 1:1 relationship between the grape juice Cl - concentration and wine Cl - concentration for Chardonnay (Walker et al. 2010), indicating that virtually all Cl - in the grape juice was recovered in the wine. However, the much lower recovery (approximately 42%) of K + in Chardonnay wine from the initial grape juice K + concentration indicates a 58% loss of K + between grape juice and finished wine. This level of loss is supported by the observation of Somers (1975) that juice K + levels decrease by more than 50% during the making of white wines. For Shiraz, the ratio of Cl - in wine to Cl - in grape juice was 1.7:1 (Walker et al. 2010), indicating that during fermentation on skins, 70% more Cl - is extracted from skins into the fermenting must. In this case, the difference between Chardonnay and Shiraz in recovery of Cl - in wine relative to that in juice results from fermentation for a period on skins for Shiraz relative to no skin contact for Chardonnay (Walker et al. 2010). Like Cl -, it is possible that some K + is similarly extracted from skins of Shiraz into the fermenting must, particularly during the

Walker and Blackmore K + concentrations and ph in grape juice and wine 191 Figure 3. Regressions of grape juice K + concentration (a,b) and wine ph (c,d) on grape juice soluble solids for Chardonnay (left, open symbols) and Shiraz (right, closed symbols) in season 1997. Grape juice values are means for 6 10 replicate vines (depending on site) for own roots and five rootstocks, and wine values are means for two winemaking replicates also for own roots and the same five rootstocks. Sites are distinguished by different symbols, as described for Figure 1. Merbein (circle), Barossa Valley sites (squares), Padthaway (upright triangle) and Koorlong (inverted triangle). Chardonnay juice K + : y = 10.1x - 183.8 (R 2 = 0.65); Shiraz juice K + : y = 9.66x - 187.1 (R 2 = 0.71); Chardonnay wine ph: y = 0.16x - 0.49 (R 2 = 0.56); Shiraz wine ph: y = 0.17x - 0.69 (R 2 = 0.68); (P < 0.001 for all four regressions). early stages of fermentation (Walker et al. 1998). However, K + differs from Cl - in that it forms soluble and insoluble salts with tartaric and malic acids (Boulton 1980b), which suggests that its mobility during fermentation is more complex. Moreover, the study of Harbertson and Harwood (2009) has shown that the K + lost during fermentation can be accounted for by the amount present in the skin pomace, indicating that grape skins can adsorb K +, most likely involving a carboxylic acid on the cell wall polysaccharides (Harbertson and Harwood 2009). The ratio of Shiraz wine K + to juice K + (0.87) was significantly less than the 1.7:1 ratio for Cl -, indicating a 13% loss of K + from the Shiraz must during fermentation. This level of loss is supported by the observation of Somers (1975) that juice K + levels decrease by less than 20% during vinification of reds. It would seem likely that with Shiraz, the approximate 87% recovery of K + relative to the 42% recovery of K + for Chardonnay between the initial grape juice concentrations and final wine concentrations reflects a combination of initial extraction of K + from Shiraz skins (Walker et al. 1998), potentially some skin adsorption of K + (Harbertson and Harwood 2009) and some precipitation of potassium bitartrate during fermentation and cold stabilisation (Boulton 1980b). With Chardonnay, however, with no skin contact, the 58% loss of K + most likely reflects precipitation as potassium bitartrate during fermentation and cold stabilisation. Given there was no artificial ph adjustment during the winemaking process, and the strong positive association between ph and K + in grape juice and finished wine, it is interesting to speculate on what may be happening during fermentation with respect to K + and organic acid complexes and resultant effects on wine ph. Regressions of wine ph on grape juice ph showed Chardonnay wine ph to be 94% of grape juice ph and Shiraz wine ph to be similar to grape juice ph at the higher ph values (3.7 3.8) but slightly lower (96% of grape juice ph) at lower ph values (data not shown). This indicates a slight reduction in ph for both varieties in progressing from grape juice to finished wine. The reduction in ph is consistent with the precipitation of potassium bitartrate either during fermentation or stabilisation (Boulton 1980b). If this was the case, a reduction in titratable acidity would also be expected (Boulton 1980b). However, regressions of wine titratable acidity on grape juice titratable acidity showed Chardonnay and Shiraz wine titratable acidity to be approximately 14 and 39% higher, respectively, than in grape juice (data not shown). An increase in titratable acidity from grape juice to wine can be due to the formation of other acids, for example, succinic and acetic acids from neutral substrates by yeast and bacteria activity during fermentation (Boulton 1980b, P. Iland, pers. comm., 2011). For Shiraz, this increase could also be due to a relatively high pro-

192 K + concentrations and ph in grape juice and wine Australian Journal of Grape and Wine Research 18, 183 193, 2012 Figure 4. Regressions of wine colour hue on wine ph for Shiraz wine made from grapes harvested at Merbein, Koorlong, Rowland Flat and Padthaway in season 1997. Wine colour hue and ph values are means for two winemaking replicates. Merbein: y = 0.331x - 0.435 (R 2 = 0.82) (P < 0.01); Koorlong: y = 0.314x - 0.315 (R 2 = 0.77) (P < 0.01); Rowland Flat: y = 0.251x - 0.187 (R 2 = 0.90) (P < 0.01); Padthaway: y = 0.329x - 0.483 (R 2 = 0.71) (P < 0.05). Absorbance units (A) for colour hue are based on the ratio of A 450/A 520. portion of shrivelled berries in the ferment and the leaching of tartaric acid, hydrogen tartrate and malic acid from skins, as the berries soak in the ferment (P. Iland, pers. comm., 2011). The skin of Shiraz berries also contains higher concentrations of tartaric acid relative to the weaker malic acid (Gong et al. 2010). High colour hue indicates a browner, less bright colour (Somers 1975), and the positive relationship between wine colour hue and wine ph agrees with previous observations that high wine ph is linked to lower wine stability, poorer colour and overall, poorer quality (Somers 1975, Boulton 1980b). Shiraz grape ripeness levels at harvest were in the range 22.8 28.6 Brix, and there was no evidence to suggest that higher irrigation water salinity levels, for example at Rowland Flat, had a negative impact on the vine s capacity to achieve desired grape berry sugar concentrations. This agrees with the observations of Walker et al. (2007), although there is evidence that sugar accumulation is reduced in severely salt-stressed vines (Prior et al. 1992). Positive linear relationships between wine ph and grape juice soluble solids (Figure 3c,d) reflect positive linear relationships between grape juice K + and grape juice soluble solids (Figure 3a,b). In turn, this reflects the previously reported relationship between sugar (as measured by soluble solids, Brix) and K + accumulation in developing grape berries in the period between veraison and ripeness (Walker et al. 2000, Mpelasoka et al. 2003). This is potentially linked to the role of K + in phloem loading and unloading (Lang 1983). The potential link between sucrose and K + transport in the phloem during this period of berry development has been noted by Mpelasoka et al. (2003). Overall, this study has demonstrated varietal differences in the relationship between wine K + and juice K +. Extraction of additional K + into the must of Shiraz during fermentation on skins appears to be countered by loss of K + during fermentation potentially because of adsorption of K + on skin pomace and potentially also to formation and precipitation of potassium bitartrate. Similarly, for Chardonnay, while there is no contact with skins during fermentation, little additional K + is extracted into the must, but significant loss of K + also appears to occur potentially through formation and precipitation as potassium bitartrate. The study has linked higher wine ph to both higher juice soluble solids and K +, and to poorer wine colour hue. Acknowledgements We thank the Grape and Wine Research and Development Corporation for financial support in setting-up and maintaining the trial sites. The input of Peter Clingeleffer to trial design is acknowledged. The advice of Peter Rogers on matters relating to winemaking is also gratefully appreciated. We also thank Russell Johnstone for assistance in locating and establishing the South Australian field sites while formerly working with the Australian Wine Research Institute. Karin Lorenzen and Dr Alf Cass (both formerly of the Cooperative Research Centre for Soil and Land Management) are acknowledged for assistance in analysis of soil samples. We also thank growers in the Barossa Valley for