OENO One Influence of the three main genetic backgrounds of grapevine rootstocks on petiolar nutrient concentrations of the scion, with a focus on phosphorus 1 UMR EGFV, Bordeaux Sciences Agro, INRA, University of Bordeaux, ISVV, 210 Chemin de Leysotte, 33882 Villenave d’Ornon, France. Gautier AntoineT. 1 Cookson SarahJ. 1 Lagalle Loïc 1 Marguerit NathalieOllat1andElisa * 1 * nutrient, petiole, scion, Vitis berlandieri, Vitis riparia, Vitis rupestris, Vitis spp. Introduction

Grafting is a horticultural technique that has been used for millennia (Mudge et al., 2009) and is still practised today in the cultivation of perennial fruit crops (e.g. citrus fruit, apple, peach, cherry and grape) and annual fruits and vegetables (e.g. watermelon and tomatoes). It enables the desirable traits of two different genotypes to be combined in the same plant. Use of rootstocks affects plant vigour, phenology, resistance to pests, fruit quality, yield, and tolerance to deleterious environmental conditions such as water deficit and nutrient limitations (Warschefsky et al., 2016).

The accidental introduction of the American aphid pest Phylloxera to Europe in the nineteenth century caused the near destruction of the European vineyard, because of the lack of phylloxera tolerance of roots of the Eurasian grapevine species Vitis vinifera. However, scientists of the time quickly realized that naturally resistant Vitis spp. of North American origin were the solution to the crisis. To maintain wine production in Europe and cater to the taste expectations of consumers, European Vitis vinifera varieties were grafted onto American Vitis spp. to combine fruit quality with phylloxera tolerance.

There are about 30 different American Vitis spp.; however, only a limited number have been used to breed rootstocks. In fact, most rootstocks used in viticulture are the result of breeding between just three different American Vitis spp. originating from different geographical areas: V. riparia, V. rupestris and V. berlandieri (Supplementary figure 1). According to the Vitis International Variety Catalogue database (http://www.vivc.de), 47 % of the 83 rootstocks used in Europe are the result of interspecies crosses of Vitis berlandieri, V. riparia and V. rupestris. Additionally, all rootstocks used in Europe have at least one of these three Vitis spp. in their genetic background (47 % with V. berlandieri, 52 % with V. riparia, and 30 % with V. rupestris). Furthermore, the potentially high genetic variability of American Vitis spp.is poorly exploited, because it is estimated that 90 % of Vitis vinifera plants in the world are grafted onto rootstocks of fewer than 10 different genotypes (Galet, 1988; Huglin and Schneider, 1998; Keller, 2015; Ollat et al., 2016). Therefore, there is considerable scope to improve current grapevine rootstock breeding programmes, to select not only for tolerance to phylloxera but also for various other agronomical traits, such as adaptability to soil properties or environmental conditions, resistance to other soil pathogens, and ability to control scion vigour (as defined by extent of shoot growth) (May, 1994; Keller, 2015).

Roots acquire mineral nutrients and water from the soil, anchor plants in the ground, and interact with soil organisms. Plants require at least 14 mineral elements for adequate development, most of which are mainly taken up by the roots from the soil solution (Marschner, 2011). They include the macronutrients nitrogen (N), phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg) and sulphur (S), and the micronutrients chlorine (Cl), boron (B), iron (Fe), manganese (Mn), copper (Cu), zinc (Zn), nickel (Ni) and molybdenum (Mo). These are required in different proportions, and deficiency or excess (toxicity) of any one of them reduces plant growth and crop yields. Numerous reviews have highlighted the facts that mineral homeostasis is highly regulated in plants and that the metabolism of different elements is interconnected (Amtmann and Armengaud, 2009; Williams and Salt, 2009). Additionally, in perennial species the root system is a storage organ for carbohydrates and other nutrients needed for the resumption of growth in the spring (Richards, 1983).

Differences in the mineral composition of plants may be due to genotypic factors responsible for differences in the nutrient- and water-uptake capacities of the roots, mobilization of stored nutrient reserves, allocation of nutrients and water within the plant, efficiency of nutrient use, and growth (all of which alter concentrations of mineral elements via dilution). The ability of a root system to take up nutrients or water is also related to development of the root system, its distribution in the soil (i.e. shallow or deep exploration), and its interactions with the rhizosphere (Dakora and Phillips, 2002; Lambers et al., 2006). Plant mineral composition is also affected by external factors, mainly soil structure and composition, nutrient availability and climatic conditions. Plants modify the pH of the rhizosphere and release various compounds (e.g. hormones, carboxylates and enzymes) that affect microbial activity or nutrient availability in the soil. Consequently, plant mineral composition is related to interactions between genotype and environment as well as the relative plasticity of plant development in response to nutrient availability.

In many grafted crops, vegetative growth and the mineral composition of the scion are strongly affected by the rootstock (as reviewed recently by Nawaz et al., 2016). In grapevine, rootstocks have long been known to modify the mineral element profile of the scion (Cordeau, 1998; BavarescoandPezzutto, 2003). Ibacache and Sierra (2009) investigated N, P and K content in the petioles of four scion cultivars (Vitis vinifera) grafted onto 10 different rootstocks; they showed rootstock effects on concentrations of N, P and K in the scion. Furthermore, several studies have shown differences in the response of grapevine rootstocks to the supply of nutrients such as N (Lecourt et al., 2015; Cochetel et al., 2017), P (Grant and Matthews, 1996a; Grant and Matthews, 1996b; Gautier et al., 2018), K (Ruhl, 1989; Ruhl, 1991) and Fe (Covarrubias and Rombolà, 2015; Covarrubias et al., 2016). In these studies, grapevine rootstocks with different genetic backgrounds showed different responses to low nutrient supply.

Despite the considerable agronomical knowledge regarding the influence of rootstock on the mineral composition of the scion, few studies have shown a strong relation between rootstock genetic background and scion mineral composition in grapevine (Wolpert et al., 2005). In the present study, we characterized the effects of rootstock genetic background on the elemental composition of scion petioles by means of an experiment in which grapevines were grafted onto rootstocks of many different genotypes and grown under non–nutrient and water-limited field conditions.

 Materials and methods 1. Vineyard site and experimental design

In 2014, Vitis vinifera cv. Cabernet-Sauvignon clone 169 was omega-grafted onto rootstocks of 13 different genotypes and then grown in a nursery for 1 year. The rootstocks were RGM, 3309C, 101-14MGt, 420A, SO4, 44-53M, Gravesac, Freedom, Dog ridge, Rupestris du Lot, 1103P, 110R and 41B. This panel was sufficiently diverse to include representatives from the three main American Vitis spp. used in rootstock crosses (i.e. V. riparia, V. rupestris and V. berlandieri). The rootstocks and their genetic backgrounds are summarized in Supplementary table 1. The experimental vineyard was planted in 2015 with three blocks of five grapevines per block for each rootstock.

The experiment was carried out during the 2017 season in an experimental vineyard near Bordeaux, France (44°47′N, 0°34′W; elevation, 22 m) with a row spacing of 1.5 × 1 m (6666 vines/ha). Vines were winter-pruned to 3–5 buds per vine on Guyot simple and trained to a vertical trellis system.

Canopy management practices included shoot hedging. Mean canopy height and width were about 1.6 and 0.4 m, respectively. The soil at the site was a gravelly sandy soil. Mean cation exchange capacity was 3.2 cmol+/kg, and available water capacity was 0.85 mm/cm. Phosphorus content was 0.015 %, which is about average for the type of soil. The soil was not tilled during the experiment; weeds appearing between rows and in the vine row were controlled by mowing several times a year (additional information on soil profile is given in Supplementary table 2).

 2. Growth measurements

Annual growth of each vine was evaluated by quantifying winter cane pruning weight at the end of 2017. For each different rootstock, the cane pruning weights of five grapevines in each of the three blocks were measured (i.e. 15 grapevines per rootstock genotype).

 3. Mineral analysis

For each different rootstock, four petioles were collected at veraison (berry softening, 14 August 2017) from two plants in each block (i.e. eight petioles per rootstock genotype), with two replicates (i.e. two samples per block; n = 6). Petioles were harvested near to a bunch, from nodes 2 to 4, and were dried in an oven at 60°C until a constant mass was achieved.

The mineral composition of the samples was analysed by Waypoint Analytical (Richmond, VA, USA). Nitrogen concentration was determined using a Leco FP-528 instrument (Leco, St Joseph, MI, USA). Concentrations of other elements (P, K, S, Mg, Ca, Na, B, Zn, Mn, Fe, Cu and Al) were determined by ICP-OES MS 730-ES (Varian, Palo Alto, CA, USA), using plant samples that had first been digested with nitric acid and hydrochloric acid in a CEM Mars5 microwave digester (CEM, Matthews, NC, USA). Concentrations are expressed in terms of percentage (weight/weight) or ppm.

4. Statistical analysis

All statistical analyses were done using R Statistics Environment (R Development Core Team, 2005). Rootstock effects on shoot growth and petiolar nutrient concentrations were determined by one-way analysis of variance (anova; significance level P < 0.05 with Tukey’s honest significant difference test) if the normality of residuals and the homogeneity of variances were respected (i.e. in the cases of cane pruning weight and petiolar N, K, S, Ca, Cu and Al concentrations). When the conditions of an anova were not met (i.e. in the cases of petiolar P, Mg, Na, B, Zn, Mn and Fe concentrations), the data were analysed using a Kruskal–Wallis test (P < 0.05). The effects of Vitis riparia, V. rupestris and V. berlandieri genetic backgrounds on mean petiolar mineral concentrations were tested, using t-tests (significance level, P < 0.05) with a Bonferroni multiple-testing correction. In the first comparison, data for rootstocks with a Vitis riparia genetic background (i.e. RGM, 3309C, 101-14MGt, 420A, S04, 44-53M and Gravesac) were compared with those for rootstocks without it. In the second comparison, data for rootstocks with a Vitis rupestris genetic background (i.e. 3309C, 101-14MGt, Dog ridge, Rupestris, 1103P and 110R) were compared with those for rootstocks without it. A third comparison was done by comparing data for rootstocks with a Vitis berlandieri genetic background (i.e. 420A, S04, 1103P, 110R and 41B) with those for rootstocks without it.

Principal component analysis was carried out on Pearson correlations, using the FactoMineR package.

Results 1. Scion vigour was significantly affected by rootstock genotype

Petiolar elemental composition was analysed in Cabernet-Sauvignongrapevines planted in a 4-year old experimental vineyard. At this developmental stage, there were various significant differences in vigour (P = 0.01) between plants with different rootstocks (Figure 1). However, only 3309C conferred a significantly lower vigour than Dog ridge.

Two-way anova was carried out to determine the potential effect of the different blocks of the experimental vineyard. The results showed that rootstock genotype explained 31.4 % of phenotypic variance, whereas block explained only 0.5 %.

Figure 1. Winter cane pruning weight in 2017 for Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The full name of each rootstock is given in Supplementary table 1. Bar colour indicates the genetic background of the rootstock: brown, Vitis riparia (Rip); dark orange, hybrids of V. riparia and V. rupestris (Rup); pale orange, hybrids of V. riparia and V. berlandieri (Ber); sand beige, other hybrids containing V. riparia; grey, hybrids of other Vitis spp.; light blue, other hybrids containing V. rupestris; turquoise, V. rupestris; green, hybrids of V. rupestris and V. berlandieri; and dark green, other hybrids containing V. berlandieri. Means and standard deviations are shown (n = 6). Different letters indicate significant differences at P < 0.05, tested by one-way anova using rootstock genotype as a factor.

Figure 1. Winter cane pruning weight in 2017 for Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The full name of each rootstock is given in Supplementary table 1. Bar colour indicates the genetic background of the rootstock: brown, Vitis riparia (Rip); dark orange, hybrids of V. riparia and V. rupestris (Rup); pale orange, hybrids of V. riparia and V. berlandieri (Ber); sand beige, other hybrids containing V. riparia; grey, hybrids of other Vitis spp.; light blue, other hybrids containing V. rupestris; turquoise, V. rupestris; green, hybrids of V. rupestris and V. berlandieri; and dark green, other hybrids containing V. berlandieri. Means and standard deviations are shown (n = 6). Different letters indicate significant differences at P < 0.05, tested by one-way anova using rootstock genotype as a factor.

2. Differences in petiolar mineral element composition were related to rootstock genotype

The petiolar concentration of each nutrient analysed, except Na, was significantly affected by rootstock genotype. The concentration of each of the three major mineral elements (i.e. N, P and K) varied significantly depending on the rootstock (Figure 2). Petiolar N concentration was significantly affected by rootstock genotype (P = 0.0005); 101-14MGt conferred the highest N concentration (0.69 %), whereas 1103P, 44-53M and RGM conferred the lowest (0.49 %) (Figure 2a). Petiolar P concentration was also significantly affected by rootstock genotype (P = 0.0002), varying 4-fold between the lowest concentration (for RGM) and the highest concentrations (for 110 R and Dog ridge) (Figure 2b). It appeared to be related to the genetic background of the rootstock, being lower when the rootstock had a V. riparia parent and highest in hybrids without a V. riparia parent (Figure 2b). Petiolar K concentration was also affected by rootstock genotype (P = 0.035), varying more than 2-fold between the lowest concentrations (for 1103P and 101-14MGt) and the highest (for Dog ridge) (Figure 2c).

Figure 2. Petiolar concentration of (a) nitrogen, (b) phosphorus and (c) potassium in Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The full name of each rootstock is given in Supplementary table 1. See the legend to Figure 1 for an explanation of the bar colour code. Means and standard deviations are shown (n = 6). Different letters indicate significant differences at P < 0.05, tested by one-way anova (for N and K) or Kruskal–Wallis test (for P), using rootstock genotype as a factor.

Figure 2. Petiolar concentration of (a) nitrogen, (b) phosphorus and (c) potassium in Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The full name of each rootstock is given in Supplementary table 1. See the legend to Figure 1 for an explanation of the bar colour code. Means and standard deviations are shown (n = 6). Different letters indicate significant differences at P < 0.05, tested by one-way anova (for N and K) or Kruskal–Wallis test (for P), using rootstock genotype as a factor.

Petiolar concentrations of other macronutrients also varied depending on the rootstock (Figure 3). The petiolar concentration of both S and Mg was significantly different (P = 6.3 × 10–12 and P = 1.2 × 10–6, respectively), varying 3-fold between the highest concentration and the lowest (Figure 3a and 3b, respectively). The petiolar S concentration of grapevines grafted onto 101-14MGt was 0.17 %, versus about 0.06 % in those grafted onto 110R or 3309C (Figure 3a). Regarding petiolar Mg concentration, 44-53M conferred a lower concentration (0.23 %) than 1103P, 101-14MGt, 41B and Gravesac (about 0.75 %) (Figure 3b). In contrast, although petiolar Ca concentration was also affected by rootstock genotype (P = 0.007), it varied by less than 2-fold between the lowest concentration (2.03 % for 3309C) and the highest (3.45 % for Freedom) (Figure 3c). Petiolar Na concentration was unaffected by rootstock genotype (P = 0.21) (Figure 3d).

Figure 3. Petiolar concentration of (a) sulphur, (b) magnesium, (c) calcium and (d) sodium in Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The full name of each rootstock is given in Supplementary table 1. See the legend to Figure 1 for an explanation of the bar colour code. Means and standard deviations are shown (n = 6). Different letters indicate significant differences at P < 0.05, tested by one-way anova (for S and Ca) or Kruskal–Wallis test (for Mg and Na), using rootstock genotype as a factor.

Figure 3. Petiolar concentration of (a) sulphur, (b) magnesium, (c) calcium and (d) sodium in Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The full name of each rootstock is given in Supplementary table 1. See the legend to Figure 1 for an explanation of the bar colour code. Means and standard deviations are shown (n = 6). Different letters indicate significant differences at P < 0.05, tested by one-way anova (for S and Ca) or Kruskal–Wallis test (for Mg and Na), using rootstock genotype as a factor.

Petiolar concentrations of microelements were significantly affected by rootstock genotype (Figure 4). The effect on petiolar B concentration was significant (P = 0.001); RGM and 420A conferred lower concentrations (about 23 ppm), whereas Dog ridge and 110R conferred higher concentrations (about 30 ppm) (Figure 4a). The petiolar concentration of both Zn and Mn was also significantly affected by rootstock genotype (P = 0.038 and P = 0.023, respectively). For Zn, the difference was about 2-fold: grapevines grafted onto 110R had a petiolar concentration of 56 ppm, whereas those grafted onto Gravesac had a petiolar concentration of 102 ppm (Figure 4b). Petiolar Mn concentration showed the greatest variation, with a 20-fold difference between the lowest and highest values (45 ppm for 44–53 M versus 865 ppm for SO4). However, for both Zn and Mn, the Kruskal multiple comparison did not allow separation of different statistical groups (Figure 4b and 4c). Petiolar Fe concentration was significantly affected by rootstock genotype (P = 2.1 × 10–6), showing 3-fold variation between SO4 (17 ppm) and 101-14MGt (60 ppm) (Figure 4d). Regarding petiolar Cu concentration, several rootstocks (RGM, 110R, 44-53M, Rupestris du Lot and 420A) conferred lower concentrations (< 15 ppm), whereas 1103P and 41B conferred higher concentrations (> 20 ppm) (P = 1.71 × 10–6) (Figure 4e). Petiolar Al concentration also showed significant variation depending on rootstock genotype (P = 2.18 × 10–7). Dog ridge and Freedom conferred high Al concentrations (> 20 ppm), whereas 420A, 101-14MGt, 1103P, 44-53M and 3309C conferred low Al concentrations (< 12 ppm) (Figure 4f).

Figure 4. Petiolar concentration of the microelements (a) boron, (b) zinc, (c) manganese, (d) iron, (e) copper and (f) aluminium in Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The full name of each rootstock is given in Supplementary table 1. See the legend to Figure 1 for an explanation of the bar colour code. Means and standard deviations are shown (n = 6). Different letters indicate significant differences at P < 0.05, tested by one-way anova (for Cu and Al) or Kruskal–Wallis test (for B, Zn, Mn and Fe), using rootstock genotype as a factor.

Figure 4. Petiolar concentration of the microelements (a) boron, (b) zinc, (c) manganese, (d) iron, (e) copper and (f) aluminium in Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The full name of each rootstock is given in Supplementary table 1. See the legend to Figure 1 for an explanation of the bar colour code. Means and standard deviations are shown (n = 6). Different letters indicate significant differences at P < 0.05, tested by one-way anova (for Cu and Al) or Kruskal–Wallis test (for B, Zn, Mn and Fe), using rootstock genotype as a factor.

3. Rootstock genetic background influenced petiolar mineral concentration profile

Mineral concentration data were subjected to principal component analysis. The first two principal components, PC1 and PC2, explained 19.9 % and 16.1 % of total variability, respectively (Figure 5). The positive values of PC1 are related to higher petiolar concentrations of P, Ca, Cu and Mg, whereas the negative values are related to higher petiolar concentrations of Mn, Zn, S and N. Furthermore, PC1 separated rootstocks with a V. riparia genetic background (negative values) from those without it (positive values). The positive values of PC2 are related to higher petiolar concentrations of K, Al, Na and B, whereas the negative values of PC2 are related to higher petiolar concentrations of Fe and Mg.

Figure 5. Principal component (PC) analysis of the petiolar concentrations of minerals in Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The distribution of variables (mineral concentrations indicated by arrows) and individual observations (symbols) on PC1 and PC2 are shown. Ellipses of confidence at the 95 % level are given for each rootstock parentage. Ber, Vitis berlandieri; Rip, V. riparia; Rup, V. rupestris. Asterisks indicate significant differences between rootstock genetic background, using Student’s t-test with the Bonferroni correction: *, P < 0.5; **, P < 0.01; ***, P < 0.001. Standard critical concentrations were added to confirm the absence of any deficiency (Delas, 2011).

Figure 5. Principal component (PC) analysis of the petiolar concentrations of minerals in Vitis vinifera cv. Cabernet-Sauvignon grafted onto 13 different rootstocks.

The distribution of variables (mineral concentrations indicated by arrows) and individual observations (symbols) on PC1 and PC2 are shown. Ellipses of confidence at the 95 % level are given for each rootstock parentage. Ber, Vitis berlandieri; Rip, V. riparia; Rup, V. rupestris. Asterisks indicate significant differences between rootstock genetic background, using Student’s t-test with the Bonferroni correction: *, P < 0.5; **, P < 0.01; ***, P < 0.001. Standard critical concentrations were added to confirm the absence of any deficiency (Delas, 2011).

4. Pair-wise comparisons of the effect of Vitis riparia, V. rupestris and V. berlandieri parents on petiolar nutrient concentrations

The effects of the species most commonly used in rootstock creation (i.e. Vitis riparia, V. rupestris and V. berlandieri) on the mineral composition of scion petioles were investigated by comparing data for the concentration of each element in the petioles of grapevines grafted

onto rootstocks containing or not containing at least 50 % of each one of these parental species in their genetic background (Table 1). In our field experiment, mineral status was non-limiting for any of the nutrients (Table 1; Delas, 2011). Regarding the macronutrients, rootstocks with a Vitis riparia parent conferred a lower petiolar concentration of P, Mg, B and Al but a higher petiolar concentration of S (Table 1). Rootstocks with a Vitis rupestris genetic background conferred a higher petiolar concentration of P, B and Fe. Rootstocks with a Vitis berlandieri genetic background conferred a higher petiolar concentration of P and Cu but a lower petiolar concentration of S and Al.

Table 1. The effect of Vitis riparia, V. rupestris and V. berlandieri parents on mean cane pruning weight and petiolar mineral concentrations in the scion, tested using Student’s t-test with the Bonferroni correction
titre du tableau
  Shoot weight (g) Petiolar concentrationa
N P K S Mg Ca Na B Zn Mn Fe Cu Al
With riparia (n = 42) 776 0.548 0.083 3.086 0.112 0.48 2.833 0.059 25.54 80.04 421.6 37.67 15.81 13.38
Without riparia (n = 36) 791 0.586 0.185 3.353 0.08 0.621 3.077 0.063 27.47 69.75 230.5 39.42 17.11 16.61
t-test     ***   *** **     *         **
With rupestris (n = 36) 807 0.557 0.169 3.219 0.093 0.596 2.888 0.059 27.53 71.17 301.6 44.08 16.61 14.47
Without rupestris (n = 42) 793 0.532 0.097 3.201 0.101 0.502 2.984 0.063 25.50 78.83 365.2 33.93 16.23 15.21
t-test     **           *     *    
With berlandieri (n = 30) 773 0.553 0.169 2.929 0.085 0.559 3.105 0.063 26.40 74.50 332.5 43.63 18.23 13.13
Without berlandieri (n = 48) 816 0.549 0.106 3.384 0.105 0.537 2.846 0.060 26.58 75.79 338.1 35.48 15.27 15.96
t-test     *   *               *** *
Standard critical concentration (of dry biomass) (Delas, 2011) 0.4–0.6 % 0.1–0.18 % 1.5–2.5 % 0.4–0.6 % 2–4% > 15 ppm > 20 ppm > 20 ppm

aExpressed as percentage (weight/weight) for N, P, K, S, Mg, Ca and Na, and as ppm for B, Zn, Mn, Fe, Cu and Al.

Discussion 1. Rootstocks influence petiolar mineral concentration

It is well known that in many grafted cultivated species, mineral nutrition is affected by rootstock genotype (Nawaz et al., 2016). In the present study, carried out under non-limited mineral conditions, the mineral composition of Cabernet-Sauvignon petioles was affected by the rootstock; this result is consistent with the literature (Cordeau, 1998; Ibacache and Sierra, 2009). The mineral composition of scion petioles was related to rootstock genotype, with groupings based on the presence of absence of a Vitis riparia, V. rupestris or V. berlandieri genetic background. This is the first time that, because of the use of such a large panel of rootstocks, the genetic background of rootstocks has been related to the mineral profile of the scion, although similar studies have been done for only K content in the petiole (Wolpert et al., 2005), in ungrafted plants for variables associated with drought tolerance (Rossdeutsch et al., 2016), and conferred vigour (Jones et al., 2009).

 2. The genetic background of rootstocks did not confer differences in scion vigour

In viticulture, vigour is often defined by the rate of shoot growth (Keller, 2015), and rootstocks affect the vigour of the scion in many species, including grapevine (Zhang et al., 2016). In grapevine, vigour is typically quantified by cane pruning weight at the end of the growing season, and rootstock-induced differences in scion development are often striking in mature plants growing in a vineyard (Jones et al., 2009; Wooldridge et al., 2010). In the present study, rootstock genotype affected cane pruning weight and explained 31.4 % of phenotypic variance; however, we were unable to show any significant differences between the results for rootstocks with different genetic backgrounds. The results of several studies and agronomical data from vineyards have shown that rootstocks with a Vitis rupestris or V. berlandieri genetic background confer a higher of scion vigour than rootstocks with a V. riparia genetic background (Galet, 1988; Cordeau, 1998; Galet and Smith, 1998; Bettiga, 2003). However, such differences were not found in the present study. Given that the plants were only 4 years old, differences will probably appear over time. Woody perennial species such as grapevine generally require a number of years between grafting and reproductive (and fully productive) maturity.

 3. The genetic background of rootstocks conferred significant differences in petiolar P concentration

In the present study, the presence of a Vitis riparia parent in the genetic background of a rootstock reduced the concentration of P in the petioles of the scion, whereas that of a V. rupestris or V. berlandieri parent increased it. The results of agronomical studies have previously shown that Vitis riparia cv. RGM reduces petiolar P concentration (Cordeau, 1998). Experiments in pots have also shown that RGM reduces shoot P concentration (except under very high N supply) in comparison with the Vitis rupestris × V. berlandieri hybrid cv. 1103P when grafted to Cabernet-Sauvignon (Lecourt et al., 2015). Using ungrafted cuttings, we have previously shown that RGM is less efficient than 1103P in terms of P uptake and P remobilization from perennial woody parts during the initial stages of grapevine development (Gautier et al., 2018). In the present study, we have shown that the presence of only one Vitis riparia parent also confers this low P concentration phenotype.

Differences in P acquisition and use between Vitis riparia and other Vitis spp. could be related to their geographical origin. Vitis riparia has a large geographical range, being found across most of North America. In comparison, other Vitis spp. used for rootstock breeding, such as V. rupestris, V. berlandieri, V.  cordifolia, V. longii and V. candicans have a more limited geographical range (Galet, 1988). Vitis rupestris and V. berlandieri are native to the south of the USA (Galet, 1988), particularly Texas, where soil is calcareous and often deficient in P due to precipitation of calcium phosphate (McLean, 1973). These species may have acquired efficient mechanisms to increase P acquisition or use in response to the limited-P environment.

 4. Petiolar P concentration correlates with known vigour conferred by the rootstock

Phosphorus is an essential element for plant growth (Bieleski, 1973). It is involved in many fundamental processes, including photosynthesis, biosynthesis and respiration, because of its role in energy generation via adenosine triphosphate. Additionally, P has a structural role in phospholipids and phosphate esters (Marschner, 2011). Reduction in growth of the shoot is a typical response to P starvation in plants (Vance et al., 2003; Hermans et al., 2006). The results of agronomical studies carried out in a Bordeaux vineyard have previously shown that grapevine rootstocks that confer high vigour to the scion also confer high petiolar P concentration in the scion at veraison (Cordeau, 1998). However, in these studies, the influence of the age of plants, soil characteristics and agricultural practices were not considered. In the present study, no differences in rootstock-conferred vigour were found, for the reasons described above. However, rootstocks (with a Vitis riparia genetic background) that confer low vigour to the scion in a mature vineyard also confer low P concentrations, when compared with V. rupestris and V. berlandieri hybrids.

 5. Rootstocks with a Vitis riparia genetic background have increased petiolar S concentration

In addition to reducing petiolar P concentration, use of rootstocks with at least one Vitis riparia parent also increases petiolar S concentration. The interaction between P and S nutrition is well known; under conditions of P deficiency, phospholipids are replaced by sulpholipids, and conversely, under conditions of S deficiency, sulpholipids are replaced by phospholipids (Essigmann et al., 1998; Härtel et al., 1998; ; Yu et al., 2002; Sugimoto et al., 2007). We have previously shown that the P acquisition efficiency of cuttings of RGM is lower than that of 1103P (Gautier et al., 2018). This may suggest that Vitis riparia rootstocks are less able to take up P from the soil, which could induce a compensatory increase in S uptake and transport to the scion.

 6. Rootstocks with a Vitis riparia genetic background confer low petiolar Mg concentration

Petiolar Mg concentration of Cabernet-Sauvignon was affected by rootstock genotype. In some cases, grapevines show symptoms of Mg deficiency, and the rootstock 44-53M has previously been described as being associated with low Mg acquisition capacity (Cordeau, 1998). In agreement with the literature, 44-53M conferred the lowest petiolar Mg concentration in the present study. Furthermore, rootstocks with a Vitis riparia genetic background conferred lower petiolar Mg concentration than rootstocks without it. In the present study, petiolar P and Mg concentrations were positively correlated (Pearson correlation coefficient, 0.34), suggesting an interaction between P and Mg assimilation in grapevine.

In California, grapevines with symptoms of Mg deficiency have been found on low-P soils (Skinner et al., 1988). Skinner et al. (1988) showed that grapevines growing under conditions of low P supply showed symptoms of Mg deficiency that were relieved by application of P fertilizer. In the present study, compared with other rootstocks, those with a Vitis riparia genetic background conferred lower petiolar P concentration. This is presumably linked to a lower capacity to take up P from the soil (Gautier et al., 2018). This lack of P-uptake capacity could also affect uptake of Mg by rootstocks with a Vitis riparia genetic background and its subsequent translocation to the shoots.

 Conclusion

Grapevine rootstocks have long been known to modify the mineral profile of the scion. However, the mechanisms underlying this phenomenon are poorly understood. For the first time, the capacity of different rootstocks to influence petiolar P concentration has been shown to be associated with the genetic background of the rootstock.

Phosphorus was the major mineral element differentially accumulated in the petiole in response to rootstock genotype. Petiolar P concentration was reduced when rootstocks with a Vitis riparia genetic background were used. Conversely, it was increased when rootstocks with a Vitis rupestris or V. berlandieri genetic background were used. Rootstocks with a Vitis riparia genetic background generally confer a lower level of vigour to the scion, whereas those with a V. rupestris or V. berlandieri background confer higher vigour. Therefore, the results may suggest that P nutrition corresponds to rootstock-conferred vigour in grapevine. Additionally, the poor efficiency of P uptake from the soil and of P remobilization from the perennial woody tissues of Vitis riparia (Gautier et al., 2018) may indirectly influence the concentration of other nutrients, such S and Mg.

 Acknowledgements

We would like to acknowledge the assistance of all the technical and scientific staff involved in the setting of the GreffAdapt experimental vineyard in INRA from Bordeaux ; particularly the Unité Expérimentale Viticole de Bordeaux 1442, INRA, F-33883 Villenave d'Ornon. This experimental vineyard was established due to the financial support from Council Interprofessional of Wine from Bordeaux (CIVB), and from Burgundy (BIVB) and from France Agrimer.

 Funding

This study was carried out with financial support from the French National Research Agency (Agence Nationale de la Recherche, ANR) in the context of the Investments for the Future programme, within the Cluster of Excellence COTE (ANR-10-LABX-45).

 Amtmann A. and Armengaud P. (2009) . Effects of N, P, K and S on metabolism: new knowledge gained from multi-level analysis. Curr Opin Plant Biol, 275-283. doi: 10.1016/j.pbi.2009.04.014, 12(3). Bavaresco L.E.G. and Pezzutto S. (2003) . Grapevine Rootstock Effects on Lime-induced Chlorosis, Nutrient Uptake, and Source–Sink Relationships (Vol. 26). Philadelphia, ETATS-UNIS: Taylor & Francis. doi:10.1081/PLN-120021054, NJ. Bettiga L.J. (2003) . Wine Grape Varieties in California. . Bieleski R.l. (1973) . Phosphate pools, phosphate transport, and phosphate avalability. Annual Review of Plant Physiology, 225-252. doi: 10.1146/annurev.pp.24.060173.001301, 24. Cochetel N. , Escudie F. , Cookson S.J. , Dai Z. , Vivin P. , Bert P.F. , Muñoz M.S , Delrot S. , Klopp C. , Ollat N. and Lauvergeat V. (2017) . Root transcriptomic responses of grafted grapevines to heterogeneous nitrogen availability depend on rootstock genotype. J Exp Bot, 68(15)-4355. doi: 10.1093/jxb/erx224. Cordeau J. (1998) . Creation of a Vineyard. Grafting the Vine and Rootstocks. Virus Disease Elimination. Éditions Féret. Covarrubias J.I. and Rombolà A.D. (2015) . Organic acids metabolism in roots of grapevine rootstocks under severe iron deficiency. Plant and Soil, 165-175. doi: 10.1007/s11104-015-2530-5, 394(1-2). Covarrubias J.I. , Retamales C. , Donnini S. , Rombolà A.D. and Pastenes C. (2016) . Contrasting physiological responses to iron deficiency in Cabernet-Sauvignon grapevines grafted on two rootstocks. Scientia Horticulturae, 1-8. doi: 10.1016/j.scienta.2015.12.013, 199. Dakora F.D. and Phillips D.A. (2002) . Root exudates as mediators of mineral acquisition in low-nutrient environments Food Security in Nutrient-Stressed Environments: exploiting plants’ genetic capabilities (pp. 201-213): Springer. doi:10.1007/978-94-017-1570-6_23. Delas J. (2011) . Fertilisation de la Vigne: contribution à une viticulture durable. Éditions Féret. Essigmann B. , Güler S. , Narang R.A. , Linke D. and Benning C. (1998) . Phosphate availability affects the thylakoid lipid composition and the expression of SQD1, a gene required for sulfolipid biosynthesis in Arabidopsis thaliana. Proc. Natl. Acad. Sci. USA, 95-1955. doi:10.1073/pnas.95.4.1950. Galet P. (1988) . Cépages et Vignobles de France, Tome 1. Les Vignes Américaines. (Second edition). Montpellier: Charles DEHAN. Galet P. and Smith J. (1998) . Grape Varieties and Rootstock Varieties: Oenoplurimédia. . Gautier A. , Cookson S.J. , Hevin C. , Vivin P. , Lauvergeat V. and Mollier A. (2018) . Phosphorus acquisition efficiency and phosphorus remobilization mediate genotype-specific differences in shoot phosphorus content in grapevine. Tree Physiol. doi: 10.1093/treephys/tpy074. Grant R.S. and Matthews M.A. (1996) . The influence of phosphorus availability, scion, and rootstock on grapevine shoot growth, leaf area, and petiole phosphorus concentration. Am. J. Enol. Vitic., 47(2), 217-224. Grant R.S. and Matthews M.A. (1996) . The influence of phosphorus availability and rootstock on root system characteristics, phosphorus uptake, phosphorus partitioning; and growth efficiency. Am. J. Enol. Vitic. , 47(4), 403-409. Härtel H. , Lokstein H. , Dörmann P. and Benning C. (1998) . The use of lipid headgroup mutants to explore the function of thylakoid lipids in photosynthesis. Photosynthesis: Mechanisms and Effects (pp. 1787-1792): Springer. doi:10.1007/978-94-011-3953-3_418. Hermans C. , Hammond J.P. , White P.J. and Verbruggen N. (2006) . How do plants respond to nutrient shortage by biomass allocation? Trends Plant Sci, 11(12), 610-617. doi: 10.1016/j.tplants.2006.10.007. Huglin P. and Schneider C. (1998) . Biology and ecology of the grapevine. Biology and ecology of the grapevine. (Ed. 2). Ibacache G.A. and Sierra B.C. (2009) . Influence of rootstocks on nitrogen, phosphorus and potassium content in petioles of four table grape varieties. Chilean Journal of Agricultural Research, 503-508. doi:10.4067/S0718-58392009000400004, 69(4). Jones T.H. , Cullis B.R. , Clingeleffer P.R. and Rühl E.H. (2009) . Effects of novel hybrid and traditional rootstocks on vigour and yield components of Shiraz grapevines. Australian Journal of Grape and Wine Research, 284-292. doi: 10.1111/j.1755-0238.2009.00061.x, 15(3). Keller M. (2015) . The Science of Grapevines: anatomy and physiology. Academic Press. Lambers H. , Shane M.W. , Cramer M.D. , Pearse S.J. and Veneklaas E.J. (2006) . Root structure and functioning for efficient acquisition of phosphorus: matching morphological and physiological traits. Ann Bot, 693-713. doi: 10.1093/aob/mcl114, 98(4). Lecourt J. , Lauvergeat V. , Ollat N. , Vivin P. and Cookson S.J. (2015) . Shoot and root ionome responses to nitrate supply in grafted grapevines are rootstock genotype dependent. Australian Journal of Grape and Wine Research, 311-318. doi: 10.1111/ajgw.12136, 21(2). Marschner H. (2011) . Marschner’s Mineral Nutrition of Higher Plants. Academic Press. May P. (1994) . Using Grapevine Rootstocks: the Australian perspective. Winetitles. McLean E. (1973) . Testing soils for pH and lime requirement. Soil Testing and Plant Analysis (revised edition), 77-95. Mudge K. , Janick J. , Scofield S. and Goldschmidt E.E. (2009) . 9 A History of Grafting. Horticultural reviews, 437. doi:10.1002/9780470593776.ch9, 35. Nawaz M.A. , Imtiaz M. , Kong Q. , Cheng F. , Ahmed W. , Huang Y. and Bie Z. (2016) . Grafting: a technique to modify ion accumulation in horticultural crops. Front Plant Sci, 7. doi: 10.3389/fpls.2016.01457. Ollat N. , Bordenave L. , Tandonnet J.P. , Boursiquot J.M. and Marguerit E. (2016) . Grapevine rootstocks: origins and perspectives. Acta Horticulturae(1136), 11-22. doi: 10.17660/ActaHortic.2016.1136.2. R Development Core Team (2005) . R: A language and environment for statistical computing. R Foundation for Statistical Computing, Austria. ISBN 3-900051-07-0, Vienna, URL: http://www.R-project.org. Richards D. (1983) . The grape root system. Horticultural Reviews, 127-168. doi:10.1002/9781118060728.ch3, 5. Rossdeutsch L. , Edwards E. , Cookson S. J. , Barrieu F. , Gambetta G. A. , Delrot S. and Ollat N. (2016) . ABA-mediated responses to water deficit separate grapevine genotypes by their genetic background. BMC Plant Biol, 91. doi: 10.1186/s12870-016-0778-4, 16. Ruhl E. (1989) . Uptake and distribution of potassium by grapevine rootstocks and its implication for grape juice pH of scion varieties. Australian Journal of Experimental Agriculture, 707-712. doi:10.1071/EA9890707, 29(5). Ruhl E. (1991) . Effect of potassium supply on cation uptake and distribution in grafted Vitis champinii and Vitis berlandieri × Vitis rupestris rootstocks. Australian Journal of Experimental Agriculture, 687-691. doi:10.1071/EA9910687, 31(5). Skinner P.W. , Cook J.A. and Matthews M.A. (1988) . Responses of grapevine cvs Chenin blanc and Chardonnay to phosphorus fertilizer applications under phosphorus-limited soil conditions. Vitis, 27, 95-109. Sugimoto K. , Sato N. and Tsuzuki M. (2007) . Utilization of a chloroplast membrane sulfolipid as a major internal sulfur source for protein synthesis in the early phase of sulfur starvation in Chlamydomonas reinhardtii. FEBS Lett, 581(23)-4522. doi: 10.1016/j.febslet.2007.08.035. Vance C.P. , Uhde-Stone C. and Allan D.L. (2003) . Phosphorus acquisition and use: critical adaptations by plants for securing a nonrenewable resource. New Phytologist, 423-447. doi:10.1046/j.1469-8137.2003.00695.x, 157. Warschefsky E.J. , Klein L.L. , Frank M.H. , Chitwood D.H. , Londo J.P. , von Wettberg E.J. and Miller A.J. (2016) . Rootstocks: diversity, domestication, and impacts on shoot phenotypes. Trends Plant Sci, 418-437. doi: 10.1016/j.tplants.2015.11.008, 21(5). Williams L. and Salt D.E. (2009) . The plant ionome coming into focus. Curr Opin Plant Biol, 247-249. doi: 10.1016/j.pbi.2009.05.009, 12(3). Wolpert J.A. , Smart D.R. and Anderson M. (2005) . Lower petiole potassium concentration at bloom in rootstocks with Vitis berlandieri genetic backgrounds. American Journal of Enology and Viticulture, 56(2), 163-169. Wooldridge J. , Louw P.J.E. and Conradie W.J. (2010) . Effects of rootstock on grapevine performance, petiole and must composition, and overall wine score of Vitis vinifera cv. Chardonnay and Pinot noir. South African Journal of Enology and Viticulture, 45-48. doi:10.21548/31-1-1399, 31(1). Yu B. , Xu C. and Benning C. (2002) . Arabidopsis disrupted in SQD2 encoding sulfolipid synthase is impaired in phosphate-limited growth. Proc Natl Acad Sci U S A, 99(8)-5737. doi: 10.1073/pnas.082696499. Zhang L. , Marguerit E. , Rossdeutsch L. , Ollat N. and Gambetta G.A. (2016) . The influence of grapevine rootstocks on scion growth and drought resistance. Theoretical and Experimental Plant Physiology, 143-157. doi: 10.1007/s40626-016-0070-x, 28(2).