Rootstocks were first widely used in viticulture in the 1800s to address the phylloxera outbreak occurring at that time in Europe and in California. Phylloxera-resistant rootstocks, involving various combinations of Vitis species from northern America, the source of phylloxera, were developed to provide protection against the aggressive pest. In modern viticulture, rootstocks are now used to address a broad spectrum of production issues. For example, in Australia, the major drivers for rootstock adoption are tolerance of soil-borne pests such as phylloxera and root knot nematodes, appropriate vigour conferred to scions, enhanced fruit quality traits (in particular reduced potassium uptake to reduce grape juice pH and acid adjustment during winemaking), tolerance of abiotic stresses including drought and salinity, and sustainability involving improved productivity and enhanced water-use efficiency (Walker and Clingeleffer, 2009). It is unlikely that a single rootstock cultivar can be developed that is suited to all environments and with the range of adaptive traits to address these issues. Indeed, Morton (1979) considered that the development of a ‘universal rootstock’ was a ‘myth’ because different rootstocks would be required to meet the challenges of different growing conditions, and that varying combinations of different Vitis species would be required to breed rootstocks for varying soil characteristics, including resistance to lime-induced chlorosis, and suitability for wet or dry conditions, heavy clay and acid soils.

Morton (1979), citing the studies of Bioletti et al. (1921) and Husmann et al. (1939), suggested that different scion varieties may require different rootstocks to optimise performance. Rives (1971) analysed growth and yield data from a number of rootstock trials with different scions, and in all cases was able to demonstrate significant scion × rootstock interactions (i.e. non-additive effects) and that concluded rootstocks could be selected to modulate overall vine vigour and growth of different scions. In contrast, Ferree et al. (1996), found no scion × rootstock interactions for cluster and fruit quality data in studies involving White Riesling and Cabernet Franc. Despite the issues raised above, there are few studies that adequately address the issues involving rootstock × environment or scion × rootstock interactions, or the studies are limited to a small number of rootstocks, for example Habran et al. (2016), Ough et al. (1969) and Wooldridge et al. (2010). Walker et al. (2010) showed that the performance of eight common rootstocks grafted with either Chardonnay or Shiraz, across four Australian regions with varying climatic conditions, soil characteristics, irrigation practices and salinity of irrigation water was not consistent with respect to yield, pruning weight (conferred vigour), and chloride and sodium concentrations in leaves and grape juice, an indication of the importance of genotype (rootstock and scion) × environment and scion × rootstock interactions. Such interactions are poorly understood, not only in regard to growth and productivity but also fruit composition and final wine quality. Gautier et al. (2019) propose various mechanisms by which the scion × rootstock interaction may influence phenotype, including the capture and transport of soil resources (water and nutrients) and root to shoot or shoot to root signalling involving metabolites, hormones, peptides and micro RNAs.

The present study, undertaken in a hot irrigated vineyard, aimed to assess the consistency of rootstock effects on growth and fruit composition of major scion varieties grown in Australia (Chardonnay, Cabernet-Sauvignon and Shiraz). It included the commonly used commercial rootstocks (Ramsey, 1103 Paulsen and 140 Ruggeri) and four promising rootstock selections from the CSIRO breeding programme (Clingeleffer et al., 2011; Clingeleffer et al., 2017).

Rootstock evaluation was undertaken over five seasons (2012/13 to 2016/18) in a replant situation with well-established grafted vines in a hot irrigated vineyard located near Mildura, Victoria, Australia. The grafted vines, produced by chip budding of potted rootstock vines in a shade house, were planted in a sandy loam soil in spring 2005. The vines, trained on a two-wire vertical trellis as a quadrilateral cordon system with wires at 1.3 m and 1.7 m, were spur-pruned by hand until conversion, to simulate light mechanical hedging (Clingeleffer, 2013) typical of that used in the region in winter 2015. A 3.0 m row × 1.8 vine spacing was utilised. The first crops were produced in 2009. The vines were drip-irrigated with approximately 5.5–6.0 ML/ha of water each season.

When established, the trial included three commercial rootstocks (1103 Paulsen, 140 Ruggeri, and Ramsey), and 26 new rootstock genotypes from the CSIRO breeding programme grafted with three scion varieties: Chardonnay (clone I10V5), Cabernet-Sauvignon (clone 22-4) and Shiraz (clone PT23). The trial was a fully randomised replicated block design with single vine plots with five replicates. The new rootstock genotypes had been selected for low potassium uptake of ungrafted vines, high rate of root strike and grafting success, and moderate-to-high vigour of ungrafted vines. This study focussed on four of the 26 rootstock genotypes, selected for their performance and resistance to root knot nematode (Smith et al., 2016) and phylloxera (Korosi et al., 2007) using rapid screening techniques. The new rootstock genotypes were C112 and C113 (V. champinii × V. cinerea), C114 (V. champinii × V. berlandieri) and C20 (V. champinii × (V. rupestris × V. riparia)).

Each season, prior to harvest a total of 100 berries were collected from each vine by sampling five berries from each of five bunches selected randomly from each of the cordons. Sampling dates ranged from 27 January to 10 February for Chardonnay; 15 February to 10 March for Cabernet-Sauvignon and 4 February to 10 March for Shiraz. The berry samples were used to determine berry weight, fruit maturity and were retained by freezing for fruit composition analyses. At harvest, yield and number of bunches were recorded for each vine. This allowed all yield components to be calculated, including bunch weight and berries per bunch, ignoring the rachis weight. Pruning weight was collected in winter and used to calculate the Ravaz Index (yield/pruning weight, Ravaz 1903) as a measure of vine balance. Analysis of fruit composition included juice total soluble sugars (TSS) determined by refractometer, pH and titratable acidity by titration (Metrohm Auto-titrator) and organic acids (malate and tartrate) and yeast assimilable nitrogen (YAN) by FTIR (Oenphos) and total berry anthocyanin and phenolic substances using spectrophotometry (Thermo Scientific).

The data was subjected to analysis of variance using the Systat statistical package, version 5, removing the effects of block and season (data not shown). For each scion variety, correlation analyses were conducted using rootstock and means generated across seasons to identify key factors contributing to rootstock effects on crop development and fruit composition.

The study was conducted over five seasons with mature grafted vines of Chardonnay, Cabernet-Sauvignon and Shiraz grown in a hot irrigated vineyard. It was shown that scion performance and fruit composition of the major varieties was not consistent across seven rootstock genotypes, as significant scion × rootstock interactions were determined. These findings indicate that specific rootstocks will be required to optimise scion variety performance. The underpinning cause for the interactions were associated with the large, three-fold difference in the inherent vigour of the scion varieties (Table 3). Rives (1971) found that that both the inherent vigour of the scion (own vigour) and that conferred by the rootstock were contributing factors to yield performance. This study has extended the approach to include not only yield but also yield components and fruit composition. It has demonstrated significant scion × rootstock interactions not only for yield, but for number of bunches, bunch weight, berry weight, berries per bunch, pruning weight, Ravaz Index, pH and YAN.

Currently, the Australian industry is reliant on rootstocks bred and selected overseas for conditions that may not be the same as those in Australia (Walker and Clingeleffer, 2009). Breeding and selecting new locally adapted rootstocks offers a potential positive impact on vine performance and wine quality while addressing the issues of sustainability and risk management. This study has shown that the phylloxera- and nematode-resistant CSIRO selections offer alternatives to existing commercial rootstock varieties. In the context of this study, the seven rootstocks could be classified into three conferred vigour categories based on mean pruning weight: C114, Ramsey and C112 with high vigour; C113 and 1103 Paulsen with moderate vigour; and 140 Ruggeri and C20 with low vigour. However, the significant scion by rootstock interaction shows that care must be taken in using general vigour classifications, a view supported by Lefort and Legisle (1977) based on their studies with young grafted vines. For example, the highest vigour was recorded in C113 with Chardonnay, in Ramsey and C112 with Cabernet-Sauvignon, and in C112 and C114 with Shiraz. C20 rootstock consistently produced low vigour. The vigour of 140 Ruggeri was also low with both Cabernet-Sauvignon and Shiraz. Santarosa et al. (2016) found that the scion × rootstock interaction for vegetative growth of Cabernet-Sauvignon and Merlot was associated with differences in the vascular systems, and specifically with xylem vessel size. The results of this study are consistent with those of Tandonnet et al. (2010), who demonstrated in pot studies that the scion had a significant effect on root growth and biomass allocation. It is possible that the relative low vigour of 140 Ruggeri, and to a lesser degree 1103 Paulsen, may be due to a breakdown in resistance to root knot nematodes in the replant situation (Smith et al. 2017).

The significant scion × rootstock interaction for yield provides evidence that different rootstocks may be required for each scion to maximise productivity. For example, with the low vigour Chardonnay, Ramsey, 1103 Paulsen and C113 were most productive whereas C20 was least productive. In contrast, with the very high vigour Cabernet-Sauvignon, Ramsey, C113, C114 and C20 produced the highest yield while 1103 Paulsen, 140 Ruggeri and C112 were the least productive. The most productive rootstocks with Shiraz were Ramsey and C114, and 1103 Paulsen and C113 were least productive. These responses can be attributed to differences in crop development across the rootstocks for the different scions. In the case of the low vigour Chardonnay, the main driver of yield was number of bunches, which was strongly linked with conferred vigour. In contrast, with the higher vigour scion varieties Cabernet-Sauvignon and Shiraz, conferred vigour was not correlated with yield. With Cabernet-Sauvignon, number of bunches and bunch weight both contributed to yield variability across the rootstocks. With Shiraz, number of bunches and to a lesser degree berry weight (r=0.61) were significant contributors to yield variability across the rootstock genotypes. Further detailed studies of crop development, including the assessment of retained nodes, budburst, shoot fruitfulness, inflorescence flower number and % fruit set, are required to further elucidate the different varietal responses. It is likely that shading, associated with the high vigour of Cabernet-Sauvignon and Shiraz, may have contributed to reduced shoot fruitfulness as shown for Sultana (May, 1965; May and Antcliff, 1963) and that the use of larger, more open trellis may have produced a higher number of bunches and potentially a higher yield for these varieties (Clingeleffer, 2009; Kliewer and Dookoozlian, 2005; May et al.,1976).

The Ravaz Index (yield/pruning weight), often referred to as vine balance, provides surrogate estimates for carbon assimilation efficiency and water-use efficiency based on total assimilation and transpiration by the canopy with pruning weight used as an indicator of canopy size (Clingeleffer et al.,2011). Kliewer and Dookloozian (2005) found that the Ravaz Index generally ranged from 5 to10 across varieties. Wooldridge et al. (2010) showed that wine quality of Chardonnay and Pinot Noir was inversely proportional to pruning weight but positively correlated with the Ravaz Index. In this study, the vigour of the scion variety had a stronger impact than rootstock on the Ravaz Index. The very low values (mean 3.3) for Cabernet-Sauvignon were well below the desirable values of between 8 and 10 determined by Dookloozlian et al. (2011) to optimise wine-quality attributes for that variety. In this study, the low yield and high vigour of Cabernet-Sauvignon contributed to the very low Ravaz Index, indicative of reduced carbon assimilation efficiency and low water-use efficiency. In contrast, Chardonnay with a high Ravaz Index (mean 12.1) associated with its high yield and low inherent vigour, had increased carbon assimilation and water-use efficiency and was capable of easily maturing the large crop. The Ravaz Index value of Shiraz (6.3), which fell between that of Cabernet-Sauvignon and Chardonnay, was close to the optimum described by Dookloozlian et al. (2011). While the results for Ravaz Index were dominated by the inherent scion vigour, the results also provide evidence of potential to select rootstocks with enhanced carbon assimilation and water-use efficiency. In this study, C20 rootstock consistently had a high Ravaz Index with all scions. The weak scion × rootstock interaction for the Ravaz Index indicates inconsistencies in the combined yield and vigour responses for the different scions.

For example, Ravaz Index values were lowest for Chardonnay with C113 and C114, for Cabernet-Sauvignon with 1103 Paulsen, and C112 and for Shiraz, with 140 Ruggeri and C20. These results suggest that individual scion/rootstock combinations may require specific irrigation, pruning or canopy management to optimise the relationship between yield and pruning weight. In this study, mechanical hedging was imposed in the last two seasons. Comparing the results from different seasons, the preliminary results indicate that the lighter pruning treatment has enhanced vine balance, with shifts in the Ravaz Index for Cabernet-Sauvignon from 2.81 to 4.21 and for Shiraz from 4.87 to 8.21.

Over the five seasons, the individual scion varieties were harvested with similar levels of maturity (i.e. 25° Brix) across the rootstocks, indicating that maturity can be eliminated as a source of variability in assessing the effects of scion × rootstock interactions on fruit composition. The significant effect of rootstock on TSS (24.2–25.2° Brix) indicates some potential to select rootstocks for early maturity (140 Ruggeri and C113) or to delay ripening (1103 Paulsen), although the difference of 1.0° Brix is unlikely to be of practical significance to industry (i.e. less than one week).

Commercial experiences and research results have shown that the widely adopted high vigour, nematode tolerant rootstock varieties contribute to negative impacts on wine quality associated with high potassium uptake and high pH and malate levels, which require tartaric acid supplements in winemaking for pH adjustment, and reduced colour in berries and poor spectral properties in red wine (Walker and Clingeleffer, 2009, 2016). Rootstocks with low potassium uptake offer a solution to problems of high juice and wine pH and associated negative impacts on wine quality (Clingeleffer 1996; Walker and Clingeleffer, 2009, 2016). While juice potassium was not measured in this study, the results indicate the potential to manipulate fruit composition using rootstocks with significant effects on pH (4.17–4.31), malate (4.45–3.95 g/L) and YAN (189–251 mg/L). However, effects of rootstock were not significant for titratable acidity, tartrate and for the red varieties, berry anthocyanin or phenolic substances. This result was unexpected as other studies in the same region have reported rootstock effects on these parameters (Ruhl et al., 1988; Clingeleffer, 1996; Walker et al., 1998; Walker and Clingeleffer, 2009).

Effects of rootstock on pH have been previously reported in similar environments with high pH linked to high vigour and high potassium uptake (Hale and Brien, 1978; Ruhl et al.,1988; Clingeleffer, 1996; Walker, et al. 1998; Walker and Clingeleffer, 2009). The overall low juice pH and low vigour of C20 rootstock is consistent with such results. However, the low pH of the high vigour rootstock C114 was unexpected, suggesting it could be a useful high vigour rootstock to enhance fruit composition compared to high vigour rootstocks such as Ramsey. However, other factors such as the weak, but significant scion × rootstock interaction, and the confounding significant effect of rootstock on TSS and vine conferred vigour and growth characteristics must also be considered in assessing the rootstock effects on pH. In particular, it should be noted that C20 and Ramsey had the highest and lowest pH, respectively, with the high vigour Cabernet-Sauvignon. The effects of growth characteristics on pH varied with the scion variety. With the low vigour Chardonnay, pH was positively linked to bunch weight and berries per bunch. With the very high vigour Cabernet-Sauvignon, growth characteristics were not associated with pH, suggesting that expected rootstock effects were masked by the effect of excessive shade (Hale and Brien, 1978; Smart et al.,1985). In contrast, with the moderately high vigour Shiraz, pH was negatively linked to pruning weight and berry weight. This unexpected result can be explained at least in part by the confounding effect of rootstock on maturity as there was a high highly significant correlation between TSS and pH (p=0.82), a relationship not found with the other scion varieties. The confounding effect of TSS, although significant only with Shiraz, indicates that detailed studies of changes in fruit composition during ripening, as undertaken by Walker and Read (2000), are required to fully understand rootstock effects on composition.

The effects of rootstock on malate were consistent across scions despite large differences in inherent vigour, with 140 Ruggeri, 1103 Paulsen and C20 having the highest malate concentrations and C114 and Ramsey the lowest concentrations. This result was unexpected, based on previous studies linking high vigour and excessive shade with high malate concentrations (Hale and Brien, 1978; Walker and Clingeleffer, 2009). Unlike pH, the effects of rootstock on malate concentration were not linked to fruit maturity for any variety. While the effects of rootstock on malate concentration were consistent across rootstocks, the relationship between pH and malate appeared to be dependent on inherent scion vigour. With Chardonnay there was no relationship between pH and malate whereas with Cabernet-Sauvignon (r=0.93) and Shiraz (r=0.74) the relationships were strong. Measurement of juice potassium and its links to pH and malate concentrations (Ruhl, 1989; Clingeleffer, 1996; Walker and Clingeleffer, 2016) would provide further insights into these responses.

All rootstocks produced acceptable concentrations of YAN required for successful fermentation (i.e. >150 mg/L). However, the significant scion × rootstock interaction indicates that YAN juice concentrations of C114 may be an issue with Shiraz, and to a lesser degree with Cabernet-Sauvignon. For all varieties YAN was strongly linked to rootstock conferred vigour and for Chardonnay it was linked with yield and number of bunches, again highlighting the importance of scion variety in assessment of rootstock effects on fruit composition. In this case, Chardonnay and Cabernet-Sauvignon had the highest and lowest YAN, respectively. For all varieties YAN was strongly correlated with juice malate concentration and for both red varieties, with pH. Further study of the impacts of both scion and rootstocks on plant nitrogen status and linkages with fruit composition is required. For example, with Shiraz grafted on 60 different rootstocks, Clingeleffer (2000) demonstrated strong correlations between juice N and pH (r=0.78), TA (0.80) and K (0.82). Ough et al. (1968) showed that nitrogen and free amino content of wines could be influenced by rootstock. Habran et al. (2016) found in studies with grafted Cabernet-Sauvignon and Pinot Noir that YAN, primary metabolites, particularly malate and amino acid content, and secondary metabolites were impacted by scion, rootstock and nitrogen supply.

In conclusion this study has shown that scion performance and fruit composition of the major varieties was not consistent across seven rootstock genotypes, because significant scion × rootstock interactions were determined. These findings indicate that specific rootstocks will be required to optimise scion variety performance and fruit composition. The study has also shown that the new CSIRO rootstock selections, covering a range of conferred vigour classifications, may be useful alternatives to those currently grown. These results suggest that individual scion/rootstock combinations may require specific irrigation, pruning or canopy management to optimise the relationship between yield and pruning weight and optimise fruit and potentially wine composition.

The authors wish to acknowledge the input of Arryn Clarke, CSIRO Irymple farm manager for routine maintenance of the trial and coordination of berry sampling and harvest. Wine Australia and CSIRO provided funding for this study.