Reproductive performance of Cabernet Sauvignon and Merlot (Vitis vinifera L.) is affected when grafted to rootstocks

Kidman et al. Reproductive performance of grafted grapevines 1 Reproductive performance of Cabernet Sauvignon and Merlot (Vitis vinifera L.) is affected when grafted to rootstocks C.M. KIDMAN 1,2, P.R. DRY 1,3, M.G. MCCARTHY 1,4 and C. COLLINS 1 1 School of Agriculture, Food and Wine, The University of Adelaide, Waite Research Institute, PMB 1, Glen Osmond, SA 564, Australia 2 Wynns Coonawarra Estate, Memorial Drive, Coonawarra, SA 5263, Australia 3 The Australian Wine Research Institute, Wine Innovation Cluster, Glen Osmond, SA 564, Australia 4 South Australian Research and Development Institute, Research Road, Nuriootpa, SA 5355, Australia Corresponding author: Dr Cassandra Collins, email cassandra.collins@adelaide.edu.au Abstract Background and Aims: Cabernet Sauvignon and Merlot are Vitis vinifera cultivars known to be susceptible to poor fruitset in cool climates (MJT 19 C 2.9 C). The importance of rootstocks in viticulture is well documented, particularly in relation to yield, salinity and water relations; little is known, however, about how rootstocks affect reproductive performance. This study assessed the reproductive performance of Cabernet Sauvignon and Merlot grafted to rootstocks. Methods and Results: Cabernet Sauvignon and Merlot grafted to rootstocks, Ramsey, 5C Teleki, Schwarzmann and 113 Paulsen, were assessed for reproductive performance over three consecutive growing seasons. This was measured by assessing the following: bud fruitfulness, flower number per inflorescence, fruitset (%), berry number per bunch, coulure index (CI), and millerandage index (MI). Fruitset was higher when grafted to rootstocks compared to that of ungrafted vines which corresponded to a decrease in MI and CI. For Cabernet Sauvignon, there were no observed differences in fruitset, however, fruitfulness and bunch number were higher when grafted to rootstocks compared to ungrafted vines. Conclusion: Rootstocks affect fruitfulness and fruitset in Cabernet Sauvignon and Merlot, however, reproductive performance differs between cultivars when grafted to the same rootstock. Significance of the Study: Rootstocks may be used as a management tool to manipulate the reproductive performance of Cabernet Sauvignon and Merlot in cool climates. Keywords: flower number, fruitfulness, fruitset, reproductive performance, rootstock Introduction The process of reproductive development in grapevines can be divided into several sequential stages that occur over two successive growing seasons. As a consequence, the growth and success of the crop is dependent on bunch initiation in the previous season, inflorescence development, flowering and fruitset and the development of seeds and flesh within a grape berry in the current season. Factors affecting these processes may include the cultivar (Longbottom 27, Dry et al. 21), climatic conditions (Buttrose 1969, Dunn and Martin 2, Sommer et al. 2, 21), excessive vigour or shading (Dry 2, Collins et al. 26) and choice of rootstock through an effect on scion vigour (Candolfi-Vasconcelos and Castagnoli 1995, Cirami 1999, Whiting 23, Dry 27, Candolfi- Vasconcelos et al. 29, Keller et al. 211). There are several ways that reproductive performance can be measured in grapevines. For example, bud fruitfulness can be estimated prior to budburst by a measure of the number of inflorescence primordia present in a compound bud (Buttrose 1969). Other parameters used to measure reproductive development include fruitset, millerandage and coulure (Dry et al. 21). Fruitset is a measure of the number of flowers that successfully develop into berries. In some instances, berries on an inflorescence may not develop seeds or only develop seed traces (May 24). These seedless berries are smaller in size than seeded berries yet still undergo veraison and ripen normally (May 24). Seedless berries generally account for a low proportion of total berry mass approximately.9% in Cabernet Sauvignon and 2% in Chardonnay (Collins and Dry 29), but in Merlot, seedless berries may account for approximately 1% of all berries (Longbottom 27). In contrast, live green ovaries (LGOs), which are formed after pollination, but without fertilisation, are seedless or may contain only seed traces, remain small, green and hard and make up less than 1% of total bunch mass (Friend et al. 23, Longbottom 27, Collins and Dry 29). Therefore, the relative proportion of seeded and seedless berries and LGOs on a bunch is indicative of grapevine reproductive performance. Millerandage and coulure are important reproductive phenomena of fruitset as they can have a negative impact on final yield (Dry et al. 21). Millerandage occurs when flowers develop abnormally into either seedless berries or LGOs (May 24, Collins and Dry 29). Coulure results when flowers fail to develop into a berry or LGO, also defined as excessive shedding of ovaries or young berries (May 24, Collins and Dry 29). Coulure can result from a deficiency in the concentration of soluble and insoluble sugars and may be caused by a doi: 1.1111/ajgw.1232

2 Reproductive performance of grafted grapevines Australian Journal of Grape and Wine Research 213 disturbance in the concentration of growth regulators (Lebon et al. 28). To measure the expression of coulure and millerandage, two indices have been developed: millerandage index (MI) and coulure index (CI) (Collins and Dry 29). For both indices, the higher the numerical value, the greater the incidence of the condition. Poor fruitset can result from variation in temperature, nutrition, growth regulators, carbohydrate reserves, cultural practices and water stress (Alexander 1965, Ebadi et al. 1995, May 24, Longbottom 27, Lebon et al. 28, Collins and Dry 29, Dry et al. 21). The classification of wine grapes based on reproductive parameters revealed that certain cultivars were more susceptible to poor fruitset than others (Dry et al. 21). Cabernet Sauvignon and Merlot have been grouped together as they are both susceptible to poor fruitset due to a high incidence of both millerandage and coulure (May 24, Longbottom 27, Dry et al. 21). Propagation of a scion with a rootstock results in a grafted vine, and ungrafted vines differ from grafted vines due to the development of the graft union (Tandonnet et al. 21). The formation of callus tissue and vascular tissue from the grafting process house the newly formed xylem and phloem vessels that maintain the flow of solutes between the scion and the rootstock (Nicholas 1992, May 1994). Tandonnet et al. (21) suggested that the interaction between the combined scion and rootstock may have a greater effect than the rootstock effect alone. In some combinations, rootstock genotype has been shown to influence biomass allocation between roots and shoots, while in other combinations the scion genotype had an influence on shoot development of grafted vines. Tandonnet et al. (21) also concluded that root development including root length and root system structure of rootstock genotypes is strongly influenced by scion genotype. Reproductive development of grapevines may potentially be managed through the use of rootstocks (Candolfi-Vasconcelos and Castagnoli 1995, Cirami 1999, Whiting 23, May 24, Dry 27). For example, fruitfulness of the scion has been found to increase or decrease depending on the rootstock to which the scion was grafted (Hedberg et al. 1986, Sommer et al. 2, 21, Keller et al. 21, 211, Stevens et al. 28, Candolfi- Vasconcelos et al. 29). An attempt to classify rootstocks according to their fruitset potential has been previously reported by several authors (Candolfi-Vasconcelos and Castagnoli 1995, Cirami 1999, Whiting 23, May 24, Dry 27, Keller et al. 211). Candolfi-Vasconcelos and Castagnoli (1995) reported that rootstocks significantly improved fruitset of Pinot Noir vines at two of five sites investigated in the Oregon Wine Region, USA. Wrattonbully has been previously described as having poor fruitset for red cultivars with lower berry number per bunch as a consequence (Phylloxera and Grape Industry Board of South Australia 212). More recent studies contradict previous reports inferring that rootstock may have no effect on fruitset and that the interaction between the scion and the rootstock is greater than the rootstock effect alone (Tandonnet et al. 21, Keller et al. 211). The aim of this study was to assess reproductive performance of Cabernet Sauvignon and Merlot grafted to four rootstocks. In addition, a V. vinifera control was included for the two cultivars to further assess the differences between grafted and non-grafted vines. Materials and methods Experimental site The experimental site was in Wrattonbully, South Australia, Australia (37 2 27.62 S, 14 52 9.87 E). The mean January temperature (MJT) for Wrattonbully is 19.6 C and the degree days (DD) (October April) is 1421 (Longbottom et al. 211). This is comparable to other cool climate regions, such as the Yarra Valley (MJT 19.3 C, DD 1489) and Coonawarra (19.6 C, DD 1396) (Coombe and Dry 1988). The vineyard was planted in 22 at 1818 vines per hectare, with vine spacing and row spacing at 2 m 2.75 m, respectively, and was trained to a bilateral cordon with vertical shoot positioned canopy. All grafted and ungrafted vines were sourced from Yalumba Nursery, South Australia, Australia. All certified clones (CW44 and D3V14) and rootstocks were tested for viruses and other diseases and have corresponding class and source identification as well as batch number for traceability. Prior to grafting using the Omega graft technique, rootstocks were hot water treated for a period of 3 min at 5 C. All vines were planted on the same day as potted one-year-old vines, selected from the nursery specifically for vine uniformity and size. The vineyard had an elevation of 83 m. Vines were spur pruned by hand to approximately 4 nodes per vine to match the commercial pruning level of the vineyard. Vines were drip irrigated, using an underground water source (bore). Scheduling of irrigation was based on Gbug (gypsum block) sensor assessments, and irrigation was approximately 1.4 ml/ha (14 mm) each season. Meteorological conditions were monitored using daily temperature and rainfall data sourced from the Bureau of Meteorology weather station at Naracoorte Airport, located approximately 6 km to the west of the trial site. Longterm average temperature, growing degree days (GDD) and rainfall data were calculated from weather data archived on the Bureau of Meteorology website (http://www.bom.gov.au/ climate/dwo/idcjdw544.latest.shtml). The long-term average rainfall for the region is 561 mm (http://www.bom.gov.au/climate/dwo/idcjdw544. latest.shtml). The site is located within a phylloxera-free region that allows for the planting of ungrafted V. vinifera vines. The soil is a mixture of loamy sand over red clay on calcrete; medium thickness loamy sand over a well structured red clay on calcrete, with areas of exposed red clay on calcrete (Longbottom et al. 211). Experimental design Vitis vinifera L. cultivars Merlot (clone D3V14) and Cabernet Sauvignon (clone CW44) were grafted to four American Vitis rootstocks: Ramsey (Vitis champinii), 5C Teleki (Vitis berlandieri X Vitis rupestris), Schwarzmann (Vitis riparia X Vitis rupestris) and 113 Paulsen (Vitis berlandieri X Vitis rupestris) and compared with ungrafted Merlot and Cabernet Sauvignon. A completely randomised block design of four rootstock treatments and one ungrafted control with five replicates of one vine per plot was used at the vineyard. Measurements were taken over three consecutive seasons starting in the 28/9 growing season. Vegetative and reproductive measurements During winter dormancy, cane number and pruning mass were recorded, and average cane mass determined from these measures. All variables are presented on a per metre of cordon basis. Twenty canes from each plot were also collected at this time to assess bud fertility. Compound buds at node positions one to four on each cane were dissected and scored for the number of inflorescence primordia (IP) and the presence of primary bud necrosis (PBN) using a binocular microscope (Leica model MS5, Leica Microsystems, Wetzlar, Germany) at 1 4 magnification. The number of IP per compound bud was recorded in the primary (N+2) bud; however, when the primary bud was

Kidman et al. Reproductive performance of grafted grapevines 3 necrotic, then the largest secondary (N+3) bud was scored for the number of IP. An average of the number of IP at nodes one to four was determined to give an indication of the potential fruitfulness per node. Primary bud necrosis was also assessed at each node, and the incidence expressed as a proportion. Actual fruitfulness per shoot was determined by the number of inflorescences per metre of cordon divided by the total number of shoots (count plus non count) per metre of cordon at pruning time to give a mean number of inflorescences per shoot. No bud dissection results were obtained for the first season of the trial (28/9) due to a contamination of samples which rendered the analysis inconclusive. To assess fruitset, three inflorescences per vine from five vines per treatment were randomly selected and enclosed in a fine mesh bag prior to flowering. After flowering, bags were removed and the dehisced caps of the flowers in each bag were counted to determine flower number per inflorescence. Corresponding bunches were collected to determine berry number per bunch on the same day as harvest. The bunch number and mass of reference bunches were included in the final vine yield measurement. Berries within each reference bunch were assessed for the proportion of seeded berries, seedless berries and live green ovaries (LGOs) while average mass was determined as the mass of the bunch (g) divided by the sum of seeded berries and seedless berries on the bunch. At harvest, all bunches were counted, and the total number of bunches per vine were weighed and recorded. Results are reported per metre of cordon. Yield components and fruitset indices were calculated according to formulae in Collins and Dry (29): total berry number per bunch, fruitset (%), coulure index (CI), millerandage index (MI), berry mass (g), fruit yield per metre cordon. Statistical analyses Each cultivar was subjected to a one-way factorial with repeated measures analysis of variance (ANOVA), using Genstat Version 1.2 (Lawes Agriculture Trust 27, Rothamsted, England). The interaction between cultivar x season x rootstock treatment was assessed by ANOVA and principal component analysis (PCA) in Microsoft Excel 27 and XLSTAT Version 212 1.1 (Addinsoft SARL, Paris, France). Details of individual analyses are provided in the text or captions. Results Climatic conditions over three seasons Over the three seasons, temperature at budburst, flowering date and length of flowering period differed between cultivars and seasons. The flowering period corresponded to the modified Eichhorn and Lorenz (E L) stages 19 25 (Coombe 1995). Budburst and flowering was earlier for Merlot than for Cabernet Sauvignon in every season (Figure 1). The interval between budburst and flowering was 57 to 67 days for Merlot and 54 to 59 days for Cabernet Sauvignon. The flowering period for Merlot and Cabernet Sauvignon ranged from 7 to 1 days and 6 to 11 days, respectively, across the three seasons. In 29, the average temperature at flowering (E-L 19 25) was 18.2 C compared to 16.1 C for Cabernet Sauvignon, and flowering took 7 days for Merlot and 6 days for Cabernet Sauvignon. For the 21 flowering period, average temperature was 22.8 C for Merlot and 22.2 C for Cabernet Sauvignon and took 11 days for Merlot and 8 days for Cabernet Sauvignon. In 211, flowering lasted for 9 days for Merlot at an average temperature of 18 C and 11 days for Cabernet Sauvignon at an average temperature of 2.3 C. In addition, Cabernet Sauvignon had 4 mm of rainfall throughout this period. (http://www.bom.gov.au/climate/dwo/idcjdw544. latest.shtml). The summation of growing degree days (GDD) (base 1 C) from budburst to flowering was assessed for both cultivars: GDD was higher in 29 than in either 21 or 211. For Merlot, 21 had the lowest GDD of 176 calculated from budburst to flowering than the other seasons 247 in 29 and 223 in 211 whereas for Cabernet Sauvignon, 211 had the lowest GDD summation of 261 from budburst to flowering than 297 in 29 and 265 in 21. Annual rainfall for the growing seasons 29, 21 and 211 (seasons were calculated from 12 September to 31 March) was 182, 242, 439 mm, respectively. The summation of rainfall from budburst to flowering was 129, 155 and 25 mm, respectively (Figure 1). Vine growth Pruning mass for Cabernet Sauvignon ranged from.46 kg/m cordon to 1.26 kg/m cordon with a mean mass of.78 kg/m cordon. For Merlot, pruning mass ranged from.27 kg/m cordon to.75 kg/m cordon with a mean mass of.42 kg/m of cordon (Tables 1 and 2). Pruning mass is a product of cane mass and cane number; there was, however, a stronger relationship observed for pruning mass and average cane mass (R 2 =.781) than for pruning mass and cane number (R 2 =.545). Pruning mass was lowest in 29,.57 kg/m and.35 kg/m of cordon for Cabernet Sauvignon and Merlot, respectively, and highest in 211, 1. kg/m and.51 kg/m of cordon for Cabernet Sauvignon and Merlot, respectively. Cabernet Sauvignon vines grafted to Ramsey had a pruning mass higher than that of ungrafted vines (Table 1). Cane mass was also higher for vines grafted to 5C Teleki, Ramsey and Schwarzmann than for ungrafted vines. In contrast, Merlot vines grafted to 113 Paulsen, Ramsey (and in the final year, Schwarzmann) had a pruning mass higher than that of ungrafted and 5C Teleki. Cane mass was higher for 113 Paulsen and Ramsey than for ungrafted vines, and cane number was also higher for 113 Paulsen in 21 and 211 than for ungrafted vines (Table 3). A significant seasonal effect on yield was found for both cultivars. Yield was significantly lower in 29 than 21 (Tables 1 and 2). Cabernet Sauvignon grafted to Schwarzmann had consistently higher yield in each season than that of the other rootstocks (Table 1). For Merlot, 113 Paulsen and Ramsey had a yield significantly higher than that of ungrafted vines and 5C Teleki (Table 2). There was no significant effect of rootstock or season on the ratio of fruit mass/pruning mass (FM/PM) of Cabernet Sauvignon (Table 1). Merlot vines grafted to Schwarzmann in 211, however, had a significantly higher FM/PM than all other treatments (Table 2). Reproductive performance Bud fertility. Cabernet Sauvignon vines grafted to 5C Teleki had lower potential fruitfulness than for all other rootstocks. For actual fruitfulness, 5C Teleki was lower only than Schwarzmann. Schwarzmann had an actual fruitfulness higher than that of the other rootstocks (Table 3). Ungrafted Cabernet Sauvignon had an actual fruitfulness significantly lower than that of 113 Paulsen, Ramsey and Schwarzmann (Table 3). In Merlot, potential fruitfulness was higher in 21 than in 211. Potential fruitfulness was lower for 5C Teleki and 113

4 Reproductive performance of grafted grapevines Australian Journal of Grape and Wine Research 213 Temperature ( C) 2 18 16 14 12 1 8 6 4 2 a) Budburst Merlot Budburst Cabernet Sauvignon Flowering Merlot Flowering Cabernet Sauvignon 8 7 6 5 4 3 2 1 Rainfall (mm) Temperature ( C) 2 18 16 14 12 1 8 6 4 2 b) Budburst Merlot Budburst Cabernet Sauvignon Flowering Merlot Flowering Cabernet Sauvignon 8 7 6 5 4 3 2 1 Rainfall (mm) 2 18 c) Flowering Cabernet Sauvignon 8 7 16 14 Budburst Merlot Budburst Cabernet Sauvignon Flowering Merlot 6 Temperature ( C) 12 1 8 5 4 3 Rainfall (mm) 6 4 2 2 1 Figure 1. Daily rainfall ( ) and average temperature (maximum + minimum temperature/2) ( ) at 8% flowering for the (a) 28, (b) 29 and (c) 21 seasons for Cabernet Sauvignon and Merlot at Wrattonbully, SA, Australia. Paulsen than Ramsey in 211. Actual fruitfulness was significantly lower in 21 than in the other seasons, and 29 was significantly higher than preceding seasons (Table 4). Rootstock type had no effect on the incidence of PBN for Cabernet Sauvignon. There was a significant influence of season on PBN: higher in 211 than in 21 (Table 3). In contrast, a significant rootstock x season interaction was observed for PBN with Merlot. In 21, ungrafted vines, and Ramsey had a lower incidence of PBN than 5C Teleki and 113 Paulsen; but in 211, there was no effect of rootstock (Table 4).

Kidman et al. Reproductive performance of grafted grapevines 5 Table 1. Effect of rootstock and season on vine growth measures for Cabernet Sauvignon grown in Wrattonbully in the 29, 21 and 211 growing seasons. Variable Season Treatment P-value LSD (5%) Control (CAS) 113 Paulsen 5C Teleki Ramsey Schwarzmann Season mean Pruning mass (kg/metre cordon) 29.6.5.6.7.6.6 a.18 (R).13 (R) 21.6.7.8.8.8.8 b <.1 (S).7 (S) 211.9 1. 1. 1.2 1. 1. c NS (R*S) NS (R*S) Treatment mean.7 a.7 a.8 ab.9 b.8 ab Cane mass (g) 29 31 3 45 35 3 34 a <.1 (R) 6.15 (R) 21 35 39 58 55 45 47 b <.1 (S) 4.48 (S) 211 39 45 59 63 51 51 b NS (R*S) NS (R*S) Treatment mean 35 a 38 ab 54 c 51 c 42 b Cane no. 29 19 def 16 bc 13 a 19 def 2 ef 17 a <.1 (R) 1.49 (R) 21 17 bcd 16 bc 15 ab 15 ab 19 def 17 a <.1 (S) 1.6 (S) 211 24 h 22 gh 18 cde 2 ef 21 fg 21 b.4 (R*S) 2.42 (R*S) Treatment mean 2 18 15 17 19 Yield (kg/metre cordon) 29.9 a 1.3 abc 1. ab 1.3 abcd 1.7 cd 1.3 a.6 (R).39 (R) 21 2. de 1.6 bcd 2.6 e 1.7 cd 2.6 e 2.1 b <.1 (S).4 (S) 211 2.5 e 2.4 e 1.5 abcd 1.7 bcd 2.8 e 2.2 b.4 (R*S).65 (R*S) Treatment mean 1.8 1.7 1.7 1.6 2.4 FM/PM 29 2.8 3. 1.4 1.7 3. 2.4 NS (R) NS (R) 21 3.7 2.8 3.3 2.7 3.2 3.1 NS (S) NS (S) 211 1.8 3. 2.6 2.2 3.5 2.6 NS (R*S) NS (R*S) Treatment mean 2.7 2.9 2.4 2.1 3.2 Statistical significance of the effects of rootstocks on Cabernet Sauvignon is given by P <.5(*), P <.1(**), P <.1(***) and not significant (NS). For all treatments and seasons, each value represents the mean of five replicate samples for each group. The 5% LSD values listed are for comparison treatments (R) and for comparison seasons (S). Where there were no significant (R x S) interactions, the treatment means were compared using the (R) 5% LSD, and the season means were compared using the (S) 5% LSD. Letters account for significant differences among treatments. Fruit mass/pruning mass is the yield divided by the pruning mass per metre of cordon. CAS, Cabernet Sauvignon; LSD, least significant difference. Fruitset. For both cultivars, flower number per inflorescence was highest in 211 (Tables 3 and 5). Flower number for Cabernet Sauvignon was significantly higher for 5C Teleki than for ungrafted vines (Table 3). For Merlot, a significant season x rootstock interaction for flower number was observed although there were no distinct differences between rootstocks over each of the three seasons. In 29, 113 Paulsen and 5C Teleki had significantly higher flower number than Ramsey and ungrafted vines. In 21, there was no significant difference in flower number and in 211, 113 Paulsen had significantly lower flower number than ungrafted, Ramsey and Schwarzmann vines (Table 5). Fruitset was significantly affected by rootstock treatment for Merlot (Table 5). Fruitset of ungrafted Merlot was significantly lower than that for all other rootstocks: 41% higher for Ramsey, 57% for Schwarzmann, 63% for 113 Paulsen and 75% for 5C Teleki. For Cabernet Sauvignon, with the exception of flower number, no significant difference between rootstock treatments was observed for seeded berry number, total berry number, LGOs, MI, CI, bunch mass or berry mass (Table S1). For Merlot, seeded berry number was consistently lower for the ungrafted vines than rootstocks (Table 5). A significant rootstock x season interaction was also observed for total berry number and number of LGOs (Table 5). Total berry number was significantly higher for 113 Paulsen than the ungrafted vines in 29. In 21, 5C Teleki, Ramsey and Schwarzmann had a higher berry number than the ungrafted, whereas in 211, Ramsey and 5C Teleki had higher berry numbers than ungrafted. Large seasonal variation in the number of LGOs was observed, but no clear pattern between rootstocks was apparent. In 21 and 211, the number of LGOs was lower than 29. Although Cabernet Sauvignon showed no rootstock effect for fruitset, CI or MI, the opposite was true for Merlot CI and MI values for ungrafted vines were 19 33% and 11 33% higher, respectively, than the rootstock treatments. Merlot grafted to Ramsey and 5C Teleki had significantly higher berry mass than both ungrafted and 113 Paulsen. In addition, Merlot bunch mass was significantly lower for ungrafted than 5C Teleki, Ramsey and Schwarzmann (Table 5). Season, cultivar and rootstock interaction No significant interaction was found between cultivar x rootstock x season (Table 6). However, there were significant interactions for rootstock x cultivar and cultivar x season for the following variables: actual fruitfulness, PBN, seeded and seedless berries, yield and fruit mass/pruning mass (FM/PM). The number of seeded berries was significantly higher for Merlot than Cabernet Sauvignon. In all three seasons, Merlot had significantly higher yield, FM/PM and lower MI than Cabernet

6 Reproductive performance of grafted grapevines Australian Journal of Grape and Wine Research 213 Table 2. Effect of rootstock and season on vine growth measures for Merlot grown in Wrattonbully in the 29, 21 and 211 growing seasons. Variable Season Treatment P-value LSD (5%) Control (MER) 113 Paulsen 5C Teleki Ramsey Schwarzmann Season mean Pruning mass (kg/metre cordon) 29.28 a.48 de.28 a.4 cd.33 ab.4 a <.1 (R).81 (R) 21.27 a.55 e.35 ab.45 d.35 ab.4 a <.1 (S).33 (S) 211.36 bc.75 f.4 cd.53 e.53 e.5 b.24 (R*S).99 (R*S) Treatment mean.3.6.3.5.4 Cane mass (g) 29 2 35 29 27 24 26 a <.1 (R) 2.24 (R) 21 21 34 26 3 28 28 a <.1 (S) 4.52 (S) 211 3 47 36 42 4 39 b NS (R*S) NS (R*S) Treatment mean 24 a 38 d 3 b 33 c 31 b Cane no. 29 14 bcd 15 cd 1 a 15 cd 14 bc 14.1 (R) 1.96 (R) 21 13 bc 16 d 14 bc 15 cd 13 bc 14 NS (S) NS (S) 211 13 ab 16 d 12 ab 13 abc 13 bc 13.3 (R*S) 2.35 (R*S) Treatment mean 13 15 12 14 13 Yield (kg/metre cordon) 29 2.1 4.4 2.7 3.3 3.2 3.1 a.4 (R) 1. (R) 21 1.5 3.1 2.6 3.9 2.7 2.8 a <.1 (S).4 (S) 211 5. 7.1 4.9 6.5 5.2 5.7 b NS (R*S) NS (R*S) Treatment mean 2.9 a 4.9 c 3.4 a 4.6 bc 3.7 ab FM/PM 29 14.2 c 9.7 abc 13. bc 12. abc 1. abc 11.9 b NS (R) NS (R) 21 5.4 a 6.2 ab 7.7 abc 9.1 abc 7.4 abc 7.2 a.1 (S) 3.11 (S) 211 8.8 abc 9.1 abc 9.6 abc 8.5 abc 22. d 11.6 b.28 (R*S) 7.14 (R*S) Treatment mean 9.5 8.3 1.1 1. 13.2 Statistical significance of the effects of rootstocks on Merlot is given by P <.5(*), P <.1(**), P <.1(***) and not significant (NS). For all treatments and seasons, each value represents the mean of five replicate samples for each group. The 5% LSD values listed are for comparison treatments (R) and for comparison seasons (S). Where there were no significant (R x S) interactions, the treatment means were compared using the (R) 5% LSD and the season means were compared using the (S) 5% LSD. Letters account for significant differences among treatments. Fruit mass/pruning mass is the yield divided by the pruning weight per metre of cordon. LSD, least significant difference; MER, Merlot. Sauvignon, and Cabernet Sauvignon had higher flower numbers than Merlot (Table 6). Principal component analysis was used to assess the interaction between cultivar and rootstock for variables that were found to be significantly different (Table 6). Principal component analysis was performed on the aggregated data and PC1 and PC2 accounted for 9.4% of the variation (Figure 2). In PC1, yield was highly correlated with bunch mass, fruitset and seeded berry number and negatively correlated with MI and CI. The scores for the two cultivars and five rootstock treatments were also projected onto the vector biplot. Cabernet Sauvignon had higher pruning mass, cane mass, cane number, MI and CI than Merlot. Cabernet Sauvignon grafted to Ramsey, Schwarzmann and 5C Teleki had higher pruning mass, cane mass and cane number than ungrafted and 113 Paulsen vines. Merlot grafted to rootstocks had higher yield, seeded berries, bunch mass and fruitset than Cabernet Sauvignon and the ungrafted Merlot. Discussion This study has confirmed the influence of rootstock type on reproductive development for Merlot and Cabernet Sauvignon. This is one of the few studies where reproductive performance has been assessed on scions grafted onto non-vinifera rootstock vs ungrafted V. vinifera. For the measures of reproductive performance, rootstock effect differed between the two scion cultivars. The influence of rootstock on reproductive performance and yield in Cabernet Sauvignon was mainly due to an effect on fruitfulness. In contrast, a combination of fruitfulness, fruitset and bunch mass differences influenced yield in Merlot. Effect of climate Site temperature at both budburst and flowering differed between seasons and cultivars. Previously, a lower temperature at budburst has been reported to increase the number of flowers on a grapevine inflorescence (Dunn and Martin 2). Furthermore, floral induction in response to environmental conditions such as low temperature has been reported in other species (Kinet et al. 1993). Mean temperature at budburst for Merlot and Cabernet Sauvignon was lower in 211 than in previous seasons, and for both cultivars, budburst was delayed compared to 29 and 21. This delayed budburst was due to a cooler and wetter start than the previous seasons. Environmental conditions at budburst have been shown to affect the degree of branching of the inflorescence primordium, and this can affect flower development (May 2). The variation in flower numbers can be explained by the number of branches on the inflorescence (Dunn and Martin 2). Previous studies have shown that a cooler temperature at budburst of 12 C has been found to promote flower formation, whereas a warmer

Kidman et al. Reproductive performance of grafted grapevines 7 Table 3. Effect of rootstocks and season on fruitfulness, flower number and incidence of Primary Bud Necrosis (PBN) for Cabernet Sauvignon grown in Wrattonbully. Variable Season Treatment P-value 5% LSD Control (CAS) 113 Paulsen 5C Teleki Ramsey Schwarz-mann Season mean Potential fruitfulness 29.2 (R).18 (R) 21 1.72 1.82 1.36 1.81 1.76 1.69 NS (S) NS (S) 211 1.79 1.79 1.65 1.7 1.64 1.71 NS (R*S) NS (R*S) Treatment mean 1.75 b 1.8 b 1.51 a 1.75 b 1.7 b % PBN 29 NS (R) NS (R) 21 2 9 8 1 1 8 a.2 (S).5 (S) 211 16 15 3 11 1 17 b NS (R*S) NS (R*S) Treatment mean 9 12 19 11 1 Actual fruitfulness 29 2.7 2.37 1.77 2.25 2.73 2.24 c.1 (R).28 (R) 21 1.33 1.95 1.67 1.87 2.6 1.78 b <.1 (S).2 (S) 211 1.7 1.38 1.63 1.38 1.78 1.45 a NS (R*S).46 (R*S) Treatment mean 1.49 a 1.9 c 1.69 abc 1.83 bc 2.19 d One-way ANOVA was performed using repeated measures analysis to assess the interaction between rootstock and season. The comparison of both cultivars on treatments over three seasons is given by P <.5(*), P <.1(**), P <.1(***) and not significant (NS). Letters account for significant differences among treatments. For all cultivars, treatments and seasons, each value represents the mean of five replicate samples for each group. Calculations of fruitfulness and PBN are determined on a per node basis. Actual fruitfulness was calculated by number of inflorescences per metre of cordon divided by canes per metre of cordon. CAS, Cabernet Sauvignon; LSD, least significant difference. Table 4. Effect of rootstock and season on fruitfulness and incidence of Primary Bud Necrosis (PBN) (%) for Merlot grown in Wrattonbully. Variable Season Treatment P-value 5% LSD Control (MER) 113 Paulsen 5C Teleki Ramsey Schwarz Season mean Potential 29 NS (R) NS (R) fruitfulness 21 1.79 ab 1.91 b 1.74 ab 1.75 ab 1.91 b 1.82 b.1 (S).9 (S) 211 1.74 ab 1.54 a 1.52 a 1.87 b 1.75 ab 1.69 a.49 (R*S).23 (R*S) Treatment mean 1.77 1.74 1.63 1.81 1.83 % PBN 29 NS (R) NS (R) 21 5 a 18 bc 2 c 4 a 1 abc 11 NS (S) NS (S) 211 9 abc 5 a 11 abc 15 abc 6.2 ab 9.1 (R*S).113 (R*S) Treatment mean 7 11 16 9 8 Actual 29 2.62 2.28 2.13 2.51 2.21 2.35 c NS (R) NS (R) fruitfulness 21 1.41 1.54 1.48 1.95 1.76 1.63 a <.1 (S).6 (S) 211 1.78 2. 2.5 1.88 1.82 1.91 b NS (R*S) NS (R*S) Treatment mean 1.94 1.94 1.89 2.11 1.93 One-way ANOVA was performed using repeated measures analysis to assess the interaction between rootstock and season. The comparison of both cultivars on treatments over three seasons is given by P <.5(*), P <.1(**), P <.1(***) and not significant (NS). Letters account for significant differences among treatments. For all cultivars, treatments and seasons, each value represents the mean of five replicate samples for each group. Calculations of fruitfulness and PBN are determined on a per node basis. Actual fruitfulness was calculated by number of inflorescences per metre of cordon divided by canes per metre of cordon. LSD, least significant difference; MER, Merlot. temperature 25 C reduces flower formation. Inflorescences on later bursting shoots are known to have fewer flowers than on earlier bursting shoots due to the progressive increase in soil and air temperature over time (Dunn and Martin 2). The cooler budburst weather resulted in significantly higher flower numbers for both cultivars in the 211 season. Flowering of individual clusters has been reported to extend for between 4 and 8 days under optimal conditions but may be delayed due to cool or wet conditions (May 24, Candolfi-Vasconcelos et al. 29). In 211, flowering occurred at an average temperature of 18 C for Merlot and 2 C for Cabernet Sauvignon, and although temperature was optimal at

8 Reproductive performance of grafted grapevines Australian Journal of Grape and Wine Research 213 Table 5. Effect of rootstock and season on reproductive performance for Merlot grown in Wrattonbully in 29, 21 and 211 seasons. Variable Season Treatment P-value 5% LSD Control (MER) 113 Paulsen 5C Teleki Ramsey Schwarzmann Season mean Flower no. 29 224 abc 37 def 262 cde 18 ab 253 bcd 245 b NS (R) NS (R) 21 196 abc 164 a 185 ab 195 abc 218 abc 192 a <.1 (S) 33.3 (S) 211 418 h 314 def 362 efgh 369 fgh 392 gh 371 c.19 (R*S) 79.7 (R*S) Treatment mean 279 262 27 248 288 Seeded berry 29 66 abc 122 fg 88 cde 45 a 12 efg 85 <.1 (R) 18.3 (R) no. 21 53 ab 68 abcd 87 cde 91 cdef 9 cde 78 NS (S) NS (S) 211 61 abc 82 bcde 97 defg 19 efg 123 g 95 <.1 (R*S) 31.3 (R*S) Treatment mean 6 92 91 82 15 Seedless berry 29 19 16 17 8 2 16 a NS (R) NS (R) no. 21 23 35 3 19 19 25 b.2 (S) 5.47 (S) 211 15 14 18 2 15 16 a NS (R*S) NS (R*S) Treatment mean 19 22 22 16 18 LGO no. 29 13 cd 24 e 16 d 8 bc 25 e 17 b.1 (R) 3.75 (R) 21 1 a 2 a 1 a 1 a 1 a 1 a <.1 (S) 3.44 (S) 211 a 1 a 1 a 1 a 1 a 1 a.25 (R*S) 7.49 (R*S) Treatment mean 5 9 6 3 9 Total berry 29 85 abc 139 e 15 bcde 53 a 122 de 11 <.1 (R) 19.33 (R) no. 21 76 ab 13 bcd 117 cde 11 cde 18 bcde 13 NS (S) NS (S) 211 76 ab 97 bcd 115 cde 129 de 138 e 111 <.1 (R*S) 33.3 (R*S) Treatment mean 79 113 113 98 123 % fruitset 29 3 49 51 32 5 42 b <.1 (R) 7.98 (R) 21 41 64 71 59 53 57 c <.1 (S) 6.45 (S) 211 2 34 36 37 39 33 a NS (R*S) NS (R*S) Treatment mean 3 a 49 bc 53 c 42 b 47 bc CI 29 6.6 4. 4.4 6.2 3.7 5. b <.1 (R).84 (R) 21 5.8 3.4 2.9 4. 4.6 4.2 a <.1 (S).65 (S) 211 8. 6.5 6.3 6.3 6.1 6.6 c NS (R*S) NS (R*S) Treatment mean 6.8 c 4.7 ab 4.6 a 5.5 b 4.8 ab MI 29 3.2 cde 2.5 bcd 2.6 bcde 2.4 bcd 3.4 de 2.8 b.12 (R).56 (R) 21 3.4 de 3.6 e 2.6 bcde 1.8 ab 1.8 ab 2.6 b <.1 (S).45 (S) 211 2.1 abc 1.6 ab 1.9 ab 1.6 ab 1.2 a 1.7 a.21 (R*S) 1. (R*S) Treatment mean 2.9 c 2.5 bc 2.3 ab 1.9 a 2.1 ab Bunch mass (g) Berry mass (g) 29 8 147 176 135 128 133 a.3 (R) 45.1 (R) 21 79 13 164 131 124 12 a.6 (S) 22.7 (S) 211 98 131 181 17 214 159 b NS (R*S) NS (R*S) Treatment mean 86 a 127 ab 174 c 145 bc 155 bc 29 1.12 1.2 1.2 1.68 1.45 1.31 ab.15 (R).22 (R) 21.99.99 1.4 1.27 1.17 1.18 a.44 (S).16 (S) 211 1.32 1.34 1.5 1.31 1.43 1.34 b NS (R*S).48 (R*S) Treatment mean 1.1 a 1.1 a 1.4 b 1.5 b 1.3 ab Statistical significance of the effects of rootstocks on Merlot is given by P <.5(*), P <.1(**), P <.1(***) and not significant (ns). For all treatments and seasons, each value represents the mean of five replicate samples for each group. The 5% LSD values listed are for comparison treatments (R) and for comparison seasons (S). Where there were no significant (R x S) interactions, the treatment means were compared using the (R) 5% LSD and the season means were compared using the (S) 5% LSD. Letters account for significant differences among treatments. Per bunch. CI, Coulure Index; LGO, live green ovary; LSD, least significant difference; MER, Merlot; MI, Millerandage Index; NS, not statistically significant difference among means at a.5 level.

Kidman et al. Reproductive performance of grafted grapevines 9 Table 6. Analysis of variance (ANOVA) for the interactions between season x cultivar x rootstock for all parameters measured at Wrattonbully in 29, 21 and 211 seasons. Variable Cultivar Rootstock Season Cultivar x rootstock Cultivar x season Rootstock x season Cultivar x rootstock x season Pruning mass (kg/m canopy) P-value <.1.4 <.1.4..911.953 Significance *** ** *** ** NS NS NS Cane no. P-value <.1.1.19.29..198.857 Significance *** *** * * NS NS NS Cane mass (g) P-value <.1 <.1 <.1.4.86.661.982 Significance *** *** *** ** NS NS NS Pred. fruitfulness P-value.953.1.232.216.79.136.456 Significance NS *** NS NS NS NS NS Actual fruitfulness P-value <.1.544.3.141.1.143.881 Significance *** NS ** NS *** NS NS PBN (%) P-value.66.17.43.919.4.546.328 Significance NS * * NS ** NS NS Flower no. P-value <.1.244 <.1.58.33.139.787 Significance *** NS *** NS * NS NS Seeded berries P-value <.1.11.25.26.293.19.24 Significance *** * * * NS * NS Seedless berries P-value.4.28 <.1.33 <.1.283.423 Significance ** NS *** NS *** NS NS LGOs P-value.68.348 <.1.117.4.711.113 Significance NS NS *** NS * NS NS Fruitset (%) P-value <.1.56 <.1.1 <.1.744.428 Significance *** NS *** *** *** NS NS Total berry no. P-value <.1.8.795.181.148.8.334 Significance *** ** NS NS NS NS NS CI P-value <.1.42 <.1...639.253 Significance *** * *** NS NS NS NS MI P-value <.1.189 <.1.149 <.1.19.342 Significance *** NS *** NS *** * NS Berry mass (g) P-value..344.272.59.583.539.82 Significance NS NS NS NS NS NS NS Bunch mass (g) P-value <.1.2.17.39.12.87.322 Significance *** ** * * NS NS NS Yield (kg/m canopy) P-value <.1.26 <.1.3 <.1.392.99 Significance *** * *** ** *** NS NS FM/PM ratio P-value <.1.681.6.137 <.1.52.314 Significance *** NS ** NS *** NS NS ANOVA of cultivar, rootstock and season effects and cultivar x rootstock, season x cultivar, season x rootstock and season x cultivar x rootstock interactions. The comparison of both cultivars on treatments over three seasons is given by P <.5(*), P <.1(**), P <.1(***) and not significant (ns). For all cultivars, treatments and seasons, each value represents the mean of five replicate samples for each group. (FM: PM), Fruit mass: Pruning mass ratio and is the yield divided by the pruning mass per metre of cordon. CI, Coulure Index; LGOs, live green ovaries; MI, Millerandage; PBN, primary bud necrosis. EL 19 25, 4 mm of rainfall was experienced for Cabernet Sauvignon at this time which may have contributed to the delay in flowering for Cabernet Sauvignon and contributed to higher CI and lower fruitset. Rain can be detrimental to fruitset at flowering as it prevents the dehiscence of the cap on flowers, and this delays ovule fertilisation (May 24). The increase in flower number for both cultivars corresponded to a high proportion of flowers that were unable to set fruit or develop into berries (seeded or seedless) or LGOs and was observed to correlate with a strong inverse relationship (R 2 =.982) between CI and fruitset for 211. Vasconcelos and Castagnoli (2) also found an inverse relationship between flower number and fruitset. Mϋller Thurgau grafted to 5C Teleki has been reported to have a high flower number, but the functionality of these flowers was lower and coulure higher than that for other rootstocks in the trial (Keller et al. 21). Cabernet Sauvignon grafted to 5C Teleki also had a higher flower number than that of ungrafted vines, but this had no effect on the incidence of CI.

1 Reproductive performance of grafted grapevines Australian Journal of Grape and Wine Research 213 4 Biplot (axes PC1 and PC2: 9.38 %) 3 2 113 Paulsen PC2 (1.78 %) 1-1 Ramsey Cane mass Pruning mass Schwarzmann Cane number 5C Teleki MI 113 Paulsen CI Ungra ed Yield Seeded Bunch mass Fruit set Ramsey 5C Teleki Schwarzmann -2 Ungra ed -3-4 -4-3 -2-1 1 2 3 4 PC1 (79.59 %) Figure 2. Principal component analysis of reproductive performance and vegetative growth variables for Cabernet Sauvignon ( ( ) on ungrafted (control) and grafted to Ramsey, 5C Teleki, Schwarzmann and 113 Paulsen rootstocks. ) and Merlot Therefore, it is more probable that the prolonged flowering period for both cultivars coupled with climatic conditions contributed to the decrease in fruitset during the flowering period of 211, although, further studies on flower number and functionality are warranted. Effect of rootstocks on fruitfulness Rootstocks can affect fruitfulness through changes in scion vigour (Candolfi-Vasconcelos et al. 29). Ramsey, 113 Paulsen and Schwarzmann have previously been shown to increase yield of the grafted scion through an increase in fruitfulness (Hedberg et al. 1986, Sommer et al. 2, Keller et al. 21, Stevens et al. 28). Fruitfulness for Ramsey has also been described as poor when compared to that of Sultana on own roots due to a higher vegetative growth of Sultana grafted to Ramsey (Sommer et al. 21). In the present study, both Merlot and Cabernet Sauvignon on Ramsey and 113 Paulsen had a higher fruitfulness than that of ungrafted vines, while Cabernet Sauvignon grafted to Schwarzmann had a significantly higher fruitfulness than this cultivar grafted to all other rootstocks, and this resulted in a significantly higher yield. The difference in fruitfulness for scions on Ramsey between the study of Sommer et al. (21) and the present study may be partly due to the different vigour potentials of the scion high for Sultana, moderate for Cabernet Sauvignon and low for Merlot (Dry 27). Although vegetative growth (as measured by pruning mass and cane number) was higher on Ramsey than the ungrafted for both scion cultivars, this was not to the extent whereby higher shoot vigour or denser canopies were observed as described in Sommer et al. (21). Conversely, a reduced fruitfulness was observed when both cultivars were grafted to 5C Teleki. In particular, both potential and actual fruitfulness were lower when 5C Teleki was used as the rootstock for Cabernet Sauvignon. Pruning mass was not significantly higher for 5C Teleki than for the other rootstocks, but a higher average cane mass may have contributed to an increased vegetative potential that consequently lowered fruitfulness. No significant difference in PBN between treatments was observed for Cabernet Sauvignon. Dry et al. (23) similarly found no rootstock effect on PBN for Shiraz. Other studies have shown site and climate affect the responses of rootstocks to PBN: no effect at one site and a significant effect at the other when rootstocks were compared at two separate locations (Cox et al. 212). In the present study, the rootstock significantly affected the incidence of PBN for Merlot only. The highest incidence of PBN was observed for rootstocks 5C Teleki and 113 Paulsen in 21. Incidence of PBN greater than 2% in a vineyard is considered to have a significant impact on fruitfulness and therefore final yield (Pool 2). Both cultivars had, in general, a low level of PBN. For example, the incidence of PBN in Cabernet Sauvignon was between 9 and 19% and 7 and 16% for Merlot across the analysis. The incidence of PBN has previously been associated with high shoot vigour (Lavee et al. 1981, Dry and Coombe 1994) and canopy shading (May 1965, Perez and Kliewer 199, Wolf and Warren 1995). An indication of vegetative growth may be deduced at pruning time through pruning mass and its components, cane mass and cane number (Smart and Robinson 1991). Ideally, pruning mass should be.3 to 1. kg/m cordon (Shaulis and Smart 1974, Kliewer and Dookoozlian 21, Smart 21) and cane mass 25 to 45 g (Reynolds 21, Smart 21). Based on these studies, it is

Kidman et al. Reproductive performance of grafted grapevines 11 apparent that neither Cabernet Sauvignon nor Merlot had excessive vegetative growth to a cause high incidence of PBN. Effect of rootstock on reproductive performance No significant changes to fruitset were observed for Cabernet Sauvignon in our study, however, the level of fruitset for Cabernet Sauvignon was low (average 24%), and the season had a greater influence on fruitset than rootstock type. In contrast, Merlot grafted to rootstocks had significantly higher fruitset and bunch parameters than that of ungrafted Merlot. Fruitset was 41 to 75% higher for Merlot grafted to rootstocks. Fruitset was highest for 5C Teleki, followed by 113 Paulsen, Schwarzmann and Ramsey. The increased fruitset of Merlot grafted to rootstocks resulted in a significantly higher number of berries per bunch and a greater number of seeded berries. A relationship between seed content and berry size has previously been reported (May 2, Friend et al. 29), and although seed number per berry was not measured in the present study, it was observed that rootstock treatments with a higher bunch mass had more seeded berries (R 2 =.633). Merlot ungrafted and grafted to 113 Paulsen had a higher measure of MI than grafted to Schwarzmann and Ramsey. Previous studies have identified that cultivar differences contribute to MI and CI (Dry et al. 21) and that management practices such as shoot topping can decrease the degree of these disorders (Collins and Dry 29). It is also apparent some rootstocks (Ramsey and Schwarzmann) may have a negative influence on the expression of MI (i.e. decrease the incidence) for Merlot but not for Cabernet Sauvignon. The cultivars Cabernet Sauvignon and Merlot were selected for this study as they are commonly regarded as susceptible to poor fruitset due to a high incidence of both millerandage and coulure, along with lower documented yields than cultivars such as Chardonnay or Sangiovese (May 24, Dry et al. 21). Furthermore, both scions have been shown to be responsive to rootstocks (Zapata et al. 21, 24, Tandonnet et al. 21). In the current study, reproductive performance differed between the two cultivars: Cabernet Sauvignon (24%) had a lower fruitset than Merlot (44%) over the 3-year period. Ungrafted Merlot is regarded as having poor fruitset at 31% (Dry et al. 21). An average fruitset of 3% was reported for ungrafted Merlot which supports the findings of Dry et al. (21). In this study, reproductive development for Merlot was able to respond more favourably to rootstocks than Cabernet Sauvignon. Previous studies have identified the reliance of Merlot on reserve carbohydrates for nutrient supply to the developing inflorescence at flowering, whereas other cultivars such as Pinot Noir sequester nutrients from photosynthesis (Zapata et al. 24). Sugars provide the main energy source for reproduction and a reduction in sugars has been shown to lead to disorders such as bud necrosis (Vasudevan et al. 1998) and coulure (Lebon et al. 28).The reliance of Merlot inflorescences on reserve carbohydrates has been previously suggested (Zapata et al. 24). These authors found a requirement on root reserves for Merlot until E L 31 (pea size berries), whereas autotrophy in Pinot Noir occurred prior to E L 27 (fruitset). These differences were shown to increase susceptibility of Merlot to coulure when the level of remobilisation of carbohydrates and nutrients was deficient (Zapata et al. 24). For Merlot, grafting to a rootstock increased both vegetative growth and yield. It is likely that root density and root distribution differed between the grafted and ungrafted vines, and this may have lead to the observed increase in reproductive development. Previously, rootstock genotype has been shown to affect root density, root distribution, nutrient status and starch concentration (Swanpoel and Southey 1989, Keller et al. 21, Dry 27, Cox et al. 212), and rootstocks with a higher root density have been shown to have higher vegetative mass and yield per vine (Swanpoel and Southey 1989). The observed reduction in coulure and increase in fruitset for Merlot vines grafted to rootstocks is likely to be due to an improved remobilisation of reserve carbohydrates attributed to the rootstock root system. While we did not measure reserve carbohydrates in the vines, evidence for a reliance on reserve carbohydrates for Merlot has previously been documented (Zapata et al. 24). This study observed that grafting Merlot to a rootstock increased fruit set and bunch mass and decreased CI. Further examination of grafted vine data showed that Merlot on Schwarzmann, 113 Paulsen and 5C Teleki had more seeded berries and total berries per bunch than that of ungrafted vines. In addition, Merlot on 113 Paulsen, Schwarzmann and Ramsey had higher actual fruitfulness and yield than that of Merlot ungrafted and on 5C Teleki. A high yield from Ramsey and 113 Paulsen has been reported previously for Chardonnay as a scion (Stevens et al. 28). Conclusion The reproductive performance of Cabernet Sauvignon and Merlot scions was affected when grafted to rootstocks. The comparison of the same rootstocks with the two cultivars at the same site, coupled with the indices for fruitset, enabled a thorough examination of rootstock effects on various components of reproduction. This study highlights the cultivar-specific interactions that occur for individual rootstocks and as a result, identified that cultivars can differ in their reproductive performance when grafted to the same rootstock. For Cabernet Sauvignon, reproductive performance was affected by rootstock treatments through increased fruitfulness. For Merlot, a combination of fruitfulness and fruitset effects by rootstock increased yield compared to that of ungrafted vines. In summary, fruitset was increased in Merlot for all rootstock treatments relative to that of ungrafted vines. This corresponded to an increase in berry number per bunch and proportion of seeded berries within the bunch. Further work to classify reproductive performance of rootstocks over a wider climatic and cultivar spectrum will benefit our knowledge of rootstocks in relation to fruitset. The use of rootstocks on cultivars considered to be susceptible to poor fruitset would be beneficial at Wrattonbully and other cool regions where poor fruitset occurs. Acknowledgements This project was funded by Australia s grapegrowers through their investment body, the Grape and Wine Research and Development Corporation along with The Phylloxera and Grape Industry Board of South Australia. Thanks to Teresa Fowles and to the staff at Waite Analytical Services, Glen Osmond, SA, Australia for their help with ICP-OES and also to Scholefield Robinson Horticultural Services, Parkside, SA, Australia for their assistance with bud dissections. Special thanks go to Yalumba Wine Company for the use of their vineyard at Wrattonbully and to James Freckleton, Daniel Newson, Wendy Smith and John Kenny for their valued assistance throughout the trial.