Unravelling rootstock×scion interactions to improve food security

Highlight

Understanding the physiological and (epi)genetic mechanisms underlying the phenotypic variability resulting from rootstock×scion×environment interactions will contribute to developing and exploiting rootstocks for food security.

Abstract

While much recent science has focused on understanding and exploiting root traits as new opportunities for crop improvement, the use of rootstocks has enhanced productivity of woody perennial crops for centuries. Grafting of vegetable crops has developed very quickly in the last 50 years, mainly to induce shoot vigour and to overcome soil-borne diseases in solanaceous and cucurbitaceous crops. In most cases, such progress has largely been due to empirical interactions between farmers, gardeners, and botanists, with limited insights into the underlying physiological mechanisms. Only during the last 20 years has science realized the potential of this old activity and studied the physiological and molecular mechanisms involved in rootstock×scion interactions, thereby not only explaining old phenomena but also developing new tools for crop improvement. Rootstocks can contribute to food security by: (i) increasing the yield potential of elite varieties; (ii) closing the yield gap under suboptimal growing conditions; (iii) decreasing the amount of chemical (pesticides and fertilizers) contaminants in the soil; (iv) increasing the efficiency of use of natural (water and soil) resources; (v) generating new useful genotypic variability (via epigenetics); and (vi) creating new products with improved quality. The potential of grafting is as broad as the genetic variability able to cross a potential incompatibility barrier between the rootstock and the scion. Therefore, understanding the mechanisms underlying the phenotypic variability resulting from rootstock×scion×environment interactions will certainly contribute to developing and exploiting rootstocks for food security.

Introduction

Securing food production for the growing population will require the breeding of new high-yielding varieties and closing of the gap between yield potential under optimal conditions and the yield actually achieved by farmers (Godfray et al., 2010; Albacete et al., 2014a ). Grafting has been used in agriculture for over 2000 years in Asia to improve plant production, reduce disease susceptibility, and increase agricultural sustainability by reducing inputs (Kubota et al., 2008; Haroldsen et al., 2012b ). This surgical alternative to breeding couples two independent genotypes to combine desired traits in the scion and rootstock. It has been used extensively to improve crop quality and productivity and also to induce variability and to propagate woody perennial crops like fruits, nuts, and ornamental plants. Indeed, the first contribution of rootstocks to food security (at the beginning of the first millennium bc) was by domesticating and propagating woody species difficult to root from cuttings, such as apples, pears, and plums (Mudge et al., 2009). Nevertheless, the best example of rootstock contributions to food security was the rescue of the European grape and wine industry from the devastating effects of the soil-borne insect phylloxera in the 19th century, and it continues to be the most efficient way to manage this pest (Pouget, 1990).

However, commercial grafting in vegetables did not become common practice until the 20th century (Lee and Oda, 2010). After the publication of the first scientific paper showing that grafting watermelon (Citrullus lanatus) onto pumpkin (Cucurbita moschata Duch.) increased pathogen (Fusarium and leaf beetle larvae) resistance and fruit yield (Tateishi, 1927), its use has been rapidly implemented to enhance productivity and disease resistance of intensive high-value solanaceous and cucurbitaceous crops (King et al., 2010). Indeed, the major impact of rootstocks in food security will probably occur in those vegetables and other annual crops including other families such as Fabaceae (soybean, peanut), Asteraceae (artichoke), and Euphorbiaceae (cassava), on which this review is mostly focused. More than 600 and 1000 million t of fruits and vegetables are produced annually in the world, respectively, with around 100 million ha cultivated (FAOSTAT, 2010). While fruit trees are almost exclusively grafted, accurate statistics do not exist in vegetables, with percentages depending on the species and country. Around 5% of the total vegetable production in countries like Korea and Japan is produced with grafted plants but with percentages close to 100% in some species like watermelon and melon (Lee et al., 2010). The situation is similar in many other horticultural areas around the world, but it is changing very quickly, mostly driven by the phasing out of soil chemical fumigants. Grafting provides opportunities to exploit natural genetic variation for specific root traits to influence the phenotype of the commercial aerial part (Albacete et al., 2009; Asins et al., 2010; Tsaballa et al., 2013). By selecting a suitable rootstock, grafting can manipulate scion morphology and manage other biotic stresses (soil-borne and air-borne pathogens) including viral, bacterial, and fungal diseases and nematodes, as well as abiotic stresses such as extreme temperatures, drought/waterlogging, and soil alkalinity/acidity (Mudge et al., 2009; Lee et al., 2010).

Comprehensive reviews on grafting have compiled existing information about historical aspects and the current state of the art in trees (Mudge et al., 2009), and in herbaceous vegetable and ornamental plants, including practical and agronomical aspects such as implementation, rootstocks, species, and crop performance (Lee and Oda, 2010) (Table 1). Sometimes, these studies are partially focused on very specific aspects of root-to-shoot communication (hydraulics and hormones) in trees (Gregory et al., 2013) and vegetables (Pérez-Alfocea et al., 2010, 2011; Ghanem et al., 2011b ), genetics and breeding (King et al., 2010), the response of grafted plants to salinity (Colla et al., 2010), nutrient and heavy metals stresses (Savvas et al., 2010; Schwarz et al., 2013), thermal stress and organic pollutants (Schwarz et al., 2010), pathogens (Louws et al., 2010), and fruit quality (Rouphael et al., 2010) (Table 1). However, few have compiled existing information about the mechanisms underlying rootstock-mediated effects on the scion and how these traits could be manipulated to improve crops.

Table 1.

List of reviews addressing different aspects of grafting

Review	Main aspects addressed
Lee (1994)	Relevant information on grafting and cultivating grafted seedlings
Lee (2003)	Avoiding crop loss due to various physiological and pathological disorders
Liu (2006)	Graft hybridization: sexual hybridization in which heritable changes may be induced by grafting
Kubota et al. (2008)	Grafting in North America to increase resistance to soil-borne diseases and nematodes and to increase yield
Mudge et al. (2009)	Information about historical aspects and the current state of the art mainly in trees
Lee and Oda (2010)	Grafting in herbaceous vegetable and ornamental plants, including practical and agronomical aspects such as implementation, rootstocks, species and crop performance
Louws et al. (2010)	Management of soil-borne pathogens and foliar soil-borne pests
Lee et al. (2010)	Reducing labour cost for grafting to improve disease resistance and crop yield and current status in Solanaceae and Cucurbitaceae
Aloni et al. (2010)	Understanding rootstock–scion communication processes and the involvement of phytohormones
King et al. (2010)	Advances in genetics and breeding of rootstocks
Martínez-Ballesta et al. (2010)	Physiological and biochemical aspects of rootstock×scion interaction, considering the mechanisms involved in graft compatibility
Colla et al. (2010)	Mechanisms of salt tolerance in grafted plants in relation to morphological root characteristics and physiological and biochemical processes
Schwarz et al. (2010)	Alleviation of adverse effects of environmental stresses on crop performance at agronomical, physiological and biochemical levels
Rouphael et al. (2010)	Effects on fruit quality including physical properties, flavour, and health-related compounds
Savvas et al. (2010)	Minimizing the negative effects of heavy metals, excessive nutrient availability, nutrient deficiency, and alkalinity stress
Guan et al. (2012)	Defence mechanisms for disease resistance of grafted vegetables
Gregory et al. (2013)	Linkage map locations for QTLs for disease resistance and other traits in rootstock breeding

Phenotypic variability in arable crops normally reflects the genotype×environment interaction (G×E). In grafted plants, the phenotype is more complex since it combines two different genotypes, causing R×S×E interactions that are driven by the communication between the rootstock (R) and the scion (S). Additionally, the effect of the graft union itself partially mediates the R×S interaction. These interactions determine the positive or negative influence of the rootstocks on plant performance and fruit quality of the scion (Fig. 1). Micrografting techniques in difficult-to-graft model species (e.g. Arabidopsis thaliana) offer opportunities to establish the influence of the root genotype on the phenotype of the aerial part and vice versa (Turnbull et al., 2002; Chen et al., 2011). Knowing the complete genome sequence of some species and the availability of many well-characterized mutants can assist in determining how root-localized gene expression can influence shoot physiology, as in tomato (Asins et al., 2010; http://www.rootopower.eu) and grapevine (Marguerit et al., 2012). Obviously, this information can be directly and immediately translated to other agronomically important species.

Fig. 1.

Phenotypic variability in grafted plants. The phenotype is more complex than in arable crops since it combines two different genotypes, causing R×S×E interactions that are driven by communication between the rootstock and the scion. The effect of the graft union itself partially mediates the R×S interaction. These interactions determine the positive or negative influence of the rootstocks on plant performance and fruit quality of the scion. PVPR, plant-growth-promoting rhizobacteria.

Communication between the rootstock and scion is bidirectional through the xylem and phloem and includes water, nutrients, hormones, metabolites, peptides, small organic molecules, and nucleic acids. Understanding the interactions and communication between scions and rootstocks will certainly allow the exploitation of new useful germplasm and new agronomic applications, especially in vegetable crops. This review explores some lesser-known mechanisms such as phytohormonal communication, gene expression, epigenetics, RNA silencing, breeding, and transgrafting as a strategy to exploit new genes in the rootstock with potentially powerful effects in the scion.

Rootstocks for invigorating or dwarfing crops

Although the use of invigorating and dwarfing rootstocks has been a classic contribution of grafting to agriculture, the regulation of developmental processes and vigour by the rootstock–scion interaction are still open questions. One hypothesis is that rootstock-induced vigour may be caused by signals such as water, nutrients, and especially hormones and nucleic acids that move through the graft union to affect scion growth. In support of this hypothesis, a significant number of mobile macromolecules can move through the vascular system (Kim et al., 2001; Ding et al., 2003; Mallory et al., 2003; Lucas and Lee, 2004), and more than 800 small organic compounds from different biochemical groups have been detected in the xylem sap of grafted tomato plants (A. Albacete, unpublished data; Table 2). Larger signalling molecules such as proteins or RNAs can also move through the graft union, and small differences in signal concentration or in their receptors/targets can alter gene expression and shoot growth (Kim et al., 2001; Mallory et al., 2003). Ultimately, the genetic background of the rootstock influences the scion, but the response of a scion grafted onto different rootstocks may be very different (Fig. 2), and vice versa.

Table 2.

Biochemical groups of compounds in the primary and secondary metabolism, and phytohormones detected in the m/z range 85–900, susceptible to being transported in the xylem sapHigh-resolution mass spectrometry with Orbitrap technology was used to identify the metabolites by exact mass.

Metabolic pathway	Biochemical group
Primary metabolites	Amino acids
	Sugars
	Sugar Phosphates
	Organic acids
	Fatty acids
	Polyols
Secondary metabolites	Alkaloids
	Flavonoids
	Glucosinolates
	Isoprenes
	Oxylipins
	Phenylpropanoids
	Pigments
	Saponins
Phytohormones	Abscisic acid and derivatives
	Gibberellins (GA1, GA3, GA4)
	Cytokinins (tZ, ZR, iP)
	Auxins (IAA and other indoles)
	Jasmonic acid and methyl jasmonate
	Salicylic acid
	Ethylene precursor (ACC)
	Brassinosteroids
	Strigolactones

Fig. 2.

(A–F) Shoot fresh weight frequency distribution of tomato plants (Solanum lycopersicum cv. Boludo F1) grafted onto a population of recombinant inbred lines (RILs) from a cross between Solanum lycopersicum×Solanum pimpinellifolium (Villalta et al., 2008; Asins et al., 2010) phenotyped within the framework of the EU Project ROOTOPOWER (#289365) under control (A) and different abiotic stress conditions: low (1mM) potassium experiment (B), performed at the CEBAS-CSIC by Dr Francisco Pérez-Alfocea and his research team; low (0.1mM) phosphorus experiment (C), performed at the EEZ-CSIC by Dr Juan Manuel Ruiz-Lozano and his research team; low (2mM) nitrogen experiment (D), performed at the Université catholique de Louvain by Dr. Xavier Draye and his research team; drought stress experiment (E), performed at the Lancaster University by Dr Ian C. Dodd and his research team; and high mechanical impedance (F), performed at the Cranfield University by Dr Andrew Thompson and his research team. (G, H) Leaf fresh weight (G) and fruit yield (H) frequency distribution of a population of tomato plants (Solanum lycopersicum cv. Boludo F1) grafted onto a RILs population from a cross between Solanum lycopersicum×Solanum cheesmaniae grown under moderate salinity (75mM NaCl, replotted from Albacete et al., 2009). ∆ indicates fold change in the measured parameter between the highest and lowest vigour-inducing lines.

Rootstock-altered leaf area was correlated with a 20% greater tomato fruit yield compared with the self-grafted scion variety under moderate salinity (75mM NaCl) but not under control (non-salinized) conditions (Albacete et al., 2009; Estañ et al., 2009), suggesting that canopy development only limited yield under salinity. Thus, rootstock modulation of leaf growth can increase yield and/or reduce inputs (water and nutrients) depending on the environmental conditions. Indeed, low-vigour rootstocks reduce tomato water use and leaf area by 40% without a negative effect on yield under optimal (no imposed stress) conditions (Cantero-Navarro, 2014). Using recombinant inbred lines (RILs) and mutants as rootstocks creates phenotypic variability depending on the rootstock and specific environment (Figs 2 and 3). Modifying only the root genotype through grafting allows the discovery of rootstock-mediated quantitative trait loci (QTLs) and the responsible candidate genes, and allows the development of linkage genetic maps for root-targeted breeding (Asins et al., 2010; Gur et al., 2010). Nutrient and water uptake, hormonal communication, and changes in gene expression are addressed in relation to invigorating-dwarfing effects.

Fig. 3.

Relative change in stomatal aperture ratio (SAR), stomatal conductance (gs) and whole-plant transpiration (ET) of self- and reciprocally grafted wild-type (WT) and abscisic acid (ABA)-deficient mutants (aba), where 100% represents the value for self-grafted (aba/aba=scion/rootstock) ABA-deficient mutants. Specific graft combinations were WT Ailsa Craig (Jones et al., 1987; Dodd et al., 2009) and Rheinlands Ruhm (Holbrook et al. 2002) tomatoes, and the sitiens (Jones et al., 1987; Holbrook et al., 2002) and flacca (Dodd et al., 2009) mutants grown in well-watered (WW) and drying (Dry) soil and at high (92%) and low (75%) relative humidity (RH), and WT Columbia A. thaliana and aba2 mutant grown under control (MS) and osmotic (Osm) stress where Ψ=–1.0MPa (Christmann et al. 2007).

Nutrient and water uptake

Water and/or nutrient capture through more efficient root system architecture and/or uptake and transport mechanisms are major rootstock traits that can regulate plant growth and crop yield (Ghanem et al., 2011b ; Gregory et al., 2013). Although the effects of different rootstocks (mostly commercial) on nutrient and water uptake in relation to plant performance are known, little information is provided about the underlying physiological mechanisms.

Root hydraulic conductivity can regulate plant growth by altering water supply to the shoot, but its role in rootstock-mediated vigour is still controversial (Gregory et al., 2013). Indeed, manipulating this trait by modulating root abscisic acid (ABA) production, thereby regulating aquaporin expression, has been suggested (Thompson et al., 2007; Parent et al., 2009; Marguerit et al., 2012).

Turning to nutrients, increased leaf nitrogen (N) concentration was positively correlated with fruit yield (10–15%) in melon and watermelon varieties grafted onto different Cucurbita maxima×C. moschata rootstocks (Colla et al., 2010). Increased N-use efficiency was due to rootstock-mediated efficiency in uptake, translocation, and assimilation (higher nitrate reductase activity and free amino acids, soluble proteins, and total-N concentrations). Similarly, grafting tomato onto commercial hybrid rootstocks (Solanum lycopersicum×Solanum habrochaites) improved irrigation water and N-use efficiency, and significantly increased tomato yields by 30% compared with non-grafted plants (Djidonou et al., 2013). Increased phosphorus (P)-use efficiency was correlated with improved yield (11%) in open-field grafted tomatoes (Khah et al., 2006). Rootstock-mediated improvement for other nutrients such as K⁺, Ca²⁺, and Mg²⁺ uptake has also been reported in different species such as eggplant, melon, watermelon, and tomato (Savvas et al., 2010). Although nutrient uptake and assimilation seem important in determining rootstock-mediated vigour, seldom have increased nutrient concentrations been correlated with yield benefits, as reported with K⁺ in salinized grafted tomatoes (Albacete et al., 2009). Rather, rootstock nutrient uptake capacity was explained by root system vigour (i.e. 71% more root hairs) compared with the self-grafted control plants (Oztekin et al., 2009). Similarly, the rootstock can also reduce the uptake and/or translocation of toxic nutrients to the scion (reviewed by Savvas et al., 2010).

Most studies suggest the need to identify highly nutrient-efficient rootstock and/or R×S combinations, and to develop irrigation and fertilization recommendations for grafted plants. However, more research is needed to identify the physiological and genetic determinants of such improvements.

Phytohormones

Regulation of plant hormones has been proposed as mechanisms by which rootstocks control scion growth (Ghanem et al., 2011b ; Pérez-Alfocea et al., 2011; Gregory et al., 2013). For example, leaf growth and crop productivity of a salinized (75mM NaCl) commercial tomato variety grafted onto different rootstocks (RILs derived from a S. lycopersicum L.×S. cheesmaniae L. Riley cross) was positively related to leaf xylem trans-zeatin (t-Z) concentration (and ratios of t-Z with other hormones) and with ABA and indole-3-acetic acid (IAA) (shoot and leaf growth; Albacete et al., 2008, 2009, 2010). Moreover, grafting wild-type (WT) plants onto a constitutively (35S cauliflower mosaic virus promoter) expressing isopentenyl adenosine transferase (IPT) gene [responsible for de novo cytokinin (CK) synthesis] rootstock increased fruit yield by 30% compared with salinized WT self-grafts. This effect was probably due to increased shoot development and/or reduced flower abortion (as suggested by 25% more fruits), where increased t-Z concentrations (from 1.5- to 2-fold) in the actively growing fruits promoted cell division and expansion, thus slightly increasing (by 5%) fruit weight (Ghanem et al., 2011a ) by promoting sink metabolism, namely the apoplastic (cell-wall invertase) and cytoplasmic (sucrose synthase and neutral invertase) sucrolytic activities (Albacete et al., 2014b ).

Grafting can also delay natural (Dong et al., 2008) and stress-induced (Albacete et al., 2009; Ghanem et al., 2011b ; Wang et al., 2012) leaf senescence, thereby improving productivity. In all cases, CKs seem important (Albacete et al., 2010). Thus, grafting early-senescent cotton genotypes onto late-senescent lines delayed leaf senescence compared with the self-grafted early-senescent lines, and higher photosynthetic rates (1.5- to 1.8-fold), chlorophyll, and soluble protein contents were correlated with enhanced accumulation of the CKs Z+ZR, and iP+iPA rather than DHZ+DHZR (Dong et al., 2008). Similar results were reported with reciprocal grafts of two cotton varieties with contrasting sensitivity to leaf senescence induction by K⁺ deficit (0.01–0.03mM; Wang et al., 2012). However, xylem sap and foliar phytohormone concentrations were scion regulated, suggesting that a feedback mechanism exists in the hypocotyl around the graft union of cotton plants affected by K⁺ deficiency to control xylem hormone concentration.

In tomato, delayed salt-induced senescence was related to increased root CKs (t-Z and IP type) export together with higher K⁺ (20%) concentration in both leaf tissue (Ghanem et al., 2011a ) and leaf xylem sap (Albacete et al., 2009), and attenuated stomatal closure and decreases in leaf xylem ABA concentration and accumulation in mature transpiring leaves. In these studies, root(stock) effect on shoot phytohormone physiology through root-to-shoot communication was clearly demonstrated by using natural genetic variability with RILs from S. lycopersicum×S. cheesmaniae (Albacete et al., 2009), transgrafting (constitutive 35S::IPT-overexpressing rootstock), and root-specific IPT expression (heat-shock-induced HSP70::IPT) (Ghanem et al., 2011a ). Increased CK delivery from root to shoot seemed to improve yield under salt stress by promoting source (photosynthesis) and sink (lower flower abortion and increased fruit growth) relations (Albacete et al., 2014a, b ).

Rootstocks can also affect growth by regulating transpiration and water stress acclimation. Hybrid rootstocks (Vitis vinifera×Vitis riparia) demonstrated QTL markers for these characters in genomic regions of the root that co-localized with genes related to ABA, water regulation (aquaporins), and root architecture (Marguerit et al., 2012). Natural constitutively elevated ABA production in the autotetraploid rootstock clone from Rangpur lime (Citrus limonia) increased leaf ABA concentrations and tolerance to water stress in sweet orange (Citrus sinensis) scions, by limiting both shoot gas exchange and growth under control conditions, and by inducing expression of drought-responsive genes, compared with the diploid rootstocks (Allario et al., 2013). These results highlight the importance of both constitutive (non-induced) gene expression and long-distance ABA signalling in growth and stress adaptation. Nevertheless, longer-term studies are needed to determine whether enhanced root ABA production affects yield under optimal or suboptimal conditions.

However, whether shoot physiological changes following stress are due to changes in root hormone export, or local changes in the shoot or roots, is uncertain. Reciprocal grafting of ABA-deficient mutants (Holbrook et al., 2002; Chen et al., 2003; Christmann et al., 2007; Dodd et al., 2009) or ABA-overproducing transgenics (Thompson et al., 2007) and WT plants evaluated the influence of root ABA biosynthesis on the shoot phenotype. Although scion genotype always had a stronger effect than rootstock genotype in reciprocal grafts, WT rootstocks partially restored growth and shoot ABA content of ABA-deficient scions (Chen et al., 2003; Dodd et al., 2009), indicating that root-synthesized ABA can influence shoot physiology. With WT scions, an ABA-deficient scion usually had limited effects on shoot water relations and stomatal conductance (Holbrook et al., 2002; Christmann et al., 2007; Dodd et al., 2009; see Ghanem et al., 2011b , for a review), suggesting that R×S×E interactions are critical in determining the physiological outcome of a given rootstock (Fig. 3).

Indeed, rootstock effects seem to be related to the environment (i.e. rootzone), and diverse mechanisms may be operating. For example, rootstocks overproducing ABA [sp12 and sp5 genotypes overexpressing the cis-epoxycarotenoid dioxygenase (NCED) gene, a key enzyme in ABA biosynthesis; Thompson et al., 2007) and CK (IPT gene; Smigocki et al., 2000) provoked similar invigorating effects on a cherry tomato variety grown in soil with unconstrained root systems in the greenhouse and saline irrigation water (3 dS m^–1, equivalent to 30mM NaCl) (Fig. 4). In the case of IPT, rootstock effects were related to increased CK delivery to the scion that stimulated photosynthesis and vegetative and reproductive growth (as described by Ghanem et al., 2011b , and Albacete et al., 2014b ). Although ABA-overproducing rootstocks had no effect on stomatal conductance of WT scions in well-watered pot-grown plants (Thompson et al., 2007), these rootstocks promoted vegetative vigour and fruit yield of salinized plants (Fig. 4), apparently by promoting deeper roots and better water and nutritional status (A. Martínez-Pérez et al., unpublished). Similarly other solanaceous crops with enhanced root system development buffered crop yield against the effects of limited water availability (Puértolas et al., 2014).

Fig. 4.

Pictures showing a mature leaf (A, B), and the 7th (C, D) and 2nd (E, F) fruit trusses of two ABA (sp12 and sp5, left panel) and CK (IPT_F and IPT_G, right panel) overproducing tomato lines and their respective WTs (Ailsa Craig: AC and UC-82B) used as rootstocks of a commercial tomato variety (Sugar Drop: SD) grown under a low salinization regime (3 dS m^–1, equivalent to 30mM NaCl) in a commercial greenhouse for 100 d.

Interestingly, rootstock-induced vigour provokes a transcriptional response in the scion similar to that of hybrid vigour, suggesting: (i) a common genetic/physiological basis of plant-growth promotion (Cookson and Ollat, 2013); and/or (ii) that the invigorating signal can be synthesized in the roots and transported to the shoot when the rootstock is a hybrid (as with most commercial vegetable rootstocks such as Maxifort, King Kong, He Man, and Bigpower in tomato). Phytohormones like CKs (Albacete et al., 2009, 2014b; Lee et al., 2010; Ghanem et al., 2011a ) may invigorate scions by mediating source–sink relationships, metabolism, and growth (Albacete et al., 2014a ), promoting transmethylation reactions including DNA methylation (Li et al., 2008), and regulating auxin transport and signalling (Marhavý et al., 2014).

Dwarfing rootstocks can increase crop sustainability without affecting yield by reducing resource allocation to the vegetative parts of the plant, thereby increasing the profitability of fruit production by reducing costs and chemical use and by increasing planting density (Prassinos et al., 2009). Hypotheses aiming at explaining rootstock-induced dwarfing should also be related to the control of vigour. The auxin/CK hypothesis states that dwarfing rootstocks decrease the basipetal transport of IAA within the phloem and cambial cells of the rootstock stem to the root, thereby limiting root growth, and the subsequent CK biosynthesis and transport to the scion, thereby limiting shoot growth (Kamboj et al., 1999; Lochard and Schneider, 2011). A second hypothesis postulates that rootstock-induced dwarfing is caused by reduced water and solute transport across the graft union (Atkinson et al., 2003; Basile et al., 2003). A third hypothesis assumes that phenols accumulating at the graft union reduce tissue viability and perhaps the rate of auxin breakdown and the subsequent CK response in the root (Errea et al., 1994; Lochard and Schneider, 2011), but no causal link has been demonstrated between phenol accumulation and growth (Prassinos et al., 2009, and references therein).

Although local auxin accumulation at the graft union and signalling seem to play a key role in graft establishment through the differentiation and connection of vascular tissues (Yin et al., 2012), it is unlikely that auxin promotes CK production in the root and transport to the shoot but rather that it inhibits it (Nordström et al., 2004; Albacete et al., 2008; Ghanem et al., 2008; Dun et al., 2009), suggesting a more direct role of the decreased rootstock export of CKs in the dwarfing process. Alternatively, using ramosus (rms) branching mutants in pea (Pisum sativum), a second messenger (shoot multiplication signal) was proposed as a root–shoot–root signal that could function in a feedback loop that regulates both CK production in the root and transport to the shoot and strigolactone biosynthesis (Dun et al., 2009). This shoot multiplication signal is related to carotenoid metabolism and responds to rhizosphere conditions (Dun et al., 2009), but is unlikely to be a strigolactone (Young et al., 2014). Similarly, the Arabidopsis bypass1 (bps1) mutant over-produces a root-to-shoot signal that arrests growth of the WT scion by interfering with auxin signalling (Sieburth and Lee, 2010). This signal also requires carotenoid biosynthesis and responds to environmental conditions like soil drying but is neither ABA nor strigolactone. Identifying these molecules and their signalling components will contribute to optimize plant growth and crop productivity to changing environments by manipulating root-to-shoot signalling.

Gene expression

Transcriptomics is an important tool to study both R×S interactions, and their responses to environmental conditions (R×S×E), but it is almost unexplored in vegetables. Such studies will help to explain any resulting positive phenotype and the underlying genetic mechanisms, thus opening new opportunities to exploit rootstocks, as suggested in some woody species (Cookson and Ollat, 2013).

In spite of apparently harmonious interactions between different genotypes, gene expression at the graft interface revealed the coordinated upregulation of genes from numerous functional categories related to stress responses in heterografts compared with autografts in Vitis spp. (Cookson et al., 2014). Grafting V. vinifera cv. Cabernet Sauvignon N with rootstocks originating from American Vitis spp. profoundly affected scion gene expression in the shoot apex, although rootstocks inducing high (Vitis berlandieri×Vitis rupestris hybrid cv. 1103 Paulsen) or low (V. riparia cv. Riparia Gloire de Montpellier) vigour had similar effects on gene expression (Cookson and Ollat, 2013). In scions of heterografted plants, upregulated genes included those related to DNA, chromatin structure, histones, flavonoids, and leucine-rich-repeat-containing receptor kinases. In rootstocks, upregulated genes comprised those involved in chromatin regulation, cell organization, and hormone signalling, suggesting some degree of self- and non-self root recognition. Many genes that were differentially expressed in the shoot apex between hetero- and autografted plants are also involved in defence responses, suggesting that the scion can detect the presence of a non-self rootstock.

Grafting grapevines with invigorating rootstocks upregulated genes in the scion apical meristem involved in carbohydrate metabolism and sugar transport (Cookson and Ollat, 2013). Similarly, sorbitol dehydrogenase (SDH) was the best-correlated transcript with plant size in grafted apple trees, suggesting increased sink strength in the shoot apex for the primary transported carbon form (Jensen et al., 2010). SDH activity, modulated by phytohormonal changes, may regulate shoot growth, as it is reduced along with growth in water-stressed apple plants, and is negatively (ABA) and positively (CKs) correlated with drought-induced hormonal changes (Li and Li, 2007). Indeed, xylem CK concentration has been positively related to rootstock-induced vigour in apple (Kamboj et al., 1999) and tomato (Albacete et al., 2009). Thus, hormonal signals may regulate rootstock-mediated vigour by modulating gene expression in the scion, including sink activity (Albacete et al., 2014a, b ).

Similar transcriptional responses in the shoot apex to invigorating non-self rootstocks and in vigorous hybrids suggest that common regulatory mechanisms and/or signalling molecules are involved (Cookson and Ollat, 2013). Shen et al. (2012) reported that the genome of Arabidopsis hybrids had high DNA methylation level, especially in the transposable elements. Methylome remodelling in these hybrids included the downregulation of genes involved in flavonoid biosynthesis and the upregulation of genes involved in auxin signalling and transport, similar to the rootstock-invigorated scion apex in grapevine (Cookson and Ollat, 2013). The upregulation of genes associated with DNA methylation and histone modification is at the basis of epigenetic mechanisms and can be affected by small interfering RNAs (siRNAs) (He et al., 2011). Since small RNAs are graft transmissible in plants (Molnar et al., 2010; Brosnan and Voinnet, 2011) and CKs seem to be involved in promoting transmethylation reactions (Li et al., 2008), these molecules and their interactions could be good candidates for the rootstock-induced alteration of gene expression in heterografted versus autografted plants and for the resulting vigour.

Transcriptomic analyses and the timing of the differential gene expression at the scion and graft union are also providing new insights to explain rootstock-mediated dwarfing. For example, 99 transcripts were differentially expressed in sweet cherry cv. Bing grafted onto dwarfing (Gisela 5) and semi-vigorous (Gisela 6) rootstocks, including genes involved in transcription regulation, brassinosteroid (BR) signalling, flavonoid metabolism, and cell-wall biosynthesis or modification (Prassinos et al., 2009). An interesting gene, encoding a protein similar to the Arabidopsis BR insensitive 1-associated receptor kinase 1 (BAK1/SERK3, RGUSS1166), was upregulated in the stem region of dwarfing apple grafts (Jensen et al., 2003), suggesting a potential role of BR signalling in the dwarfing process that should be investigated.

Reciprocal grafting experiments using the dwarf tomato cultivar Micro-Tom (MT) as a rootstock revealed that it could dwarf a normal variety, and the normal variety could invigorate an MT scion (Fig. 5). The mechanistic interpretation of this response is not straightforward, as MT harbours multiple dwarfing genes that cause BR deficiency (d mutation in BR-6 oxidase) and reduced sensitivity to gibberellins (mnt), but with normal DELLA proteins (Marti et al., 2006). Moreover, grafting WT plants onto BR-deficient rootstocks (including the d mutant) has no phenotypic effects (Symons and Reid, 2008), and thus alternative mechanisms are required to account for the dwarfing effect of the MT rootstock. However, if scion growth is due to a mobile signal from the rootstock, it should be differentially found in the xylem and/or the phloem of those grafts combinations. Thus, ectopic expression of the gain-of-function mutant Mhgai1 encoding a DELLA protein in tomato reduced plant height, and a grafting experiment demonstrated the long-distance movement of Mhgai1 mRNAs and their effect on dwarfing of the scion genotype (Wang et al., 2012). Thus, root-targeted strategies to manipulate giberellin signalling and, consequently, shoot growth and stress responses (Albacete et al., 2014a ) may be profitable.

Fig. 5.

Total fresh weight (FW; A) and appearance (B) of reciprocally grafted tomato plants, between the commercial F1 variety TT115 and the dwarf cultivar Micro-Tom (MT).

Rootstocks for long-distance gene silencing

Grafting is also useful to study the communicative signals between sensitive and responsive organs. A classic example is the control of flowering (activation and inhibition) by florigen in day-neutral tobacco plants by grafting on plants exposed to long and short days (Lang et al., 1977). Recent studies revealed that some specific RNA molecules are also transported through the phloem to coordinate organ development. Therefore, a precisely regulated RNA transport system could be applied to improve vegetable cultivars (Harada, 2010).

The phloem’s role for distributing photoassimilates and nutrients among different plant organs has long been recognized, but its role in R×S interactions has barely been investigated (Omid et al., 2007). Comparing mRNA libraries from melon phloem sap, leaves, and fruits indicated that the transcript profile of phloem sap is unique. Functional analyses revealed that over 40% of the transcripts are related to stress and defence responses, while over 15% of them are related to signal transduction (Omid et al., 2007). Experiments established that six out of 43 examined transcripts could be transported from melon rootstocks to pumpkin scions, and three were associated with auxin-signal transduction and were identified as Aux/IAA (F-308 and F-571) and small auxin-up RNA (Un-131) (Omid et al., 2007). The role of these needs further clarification before potential exploitation in grafting.

RNA interference is likely to be applied to produce virus-resistant transgenic plants (Lemgo et al., 2013), if silenced rootstocks can efficiently transmit the silencing signal to non-transformed scions, as in herbaceous plants (Kasai et al., 2011). Emerging evidence suggests that endogenous siRNA molecules can spread between cells and systemically over long distances (Dunoyer et al., 2010; Molnar et al., 2010; Haroldsen et al., 2012b ; Zhao and Song, 2014). Furthermore, microRNAs and trans-acting siRNAs have been implicated in the transmission of silencing signals from cell to cell via the plasmodesmata, and systemically via the phloem (Melnyk et al., 2011; Nazim Uddin and Kim, 2013). Silencing across a graft junction appears to require the presence of a ‘receiver’ at the other end, in the nature of the transgene target or a highly expressed endogenous target gene (Hohn et al., 2007). Thus, by grafting infected or mildly inoculated non-transformed scions containing the target viral sequence onto a silenced rootstock (expressing a double strain dsRNA homologous to a region of the viral genome), the silencing signal can be transmitted to initiate systemic silencing and thus virus resistance (Fig. 6).

Fig. 6.

Schematic representation of rootstock-induced virus resistance in plants. A silenced rootstock expressing a small RNA homologous to a region of the viral genome produces a double strain dsRNA that moves to the scion and initiates the siRNA silencing signal that interferes with replication of the viral genetic machinery, inducing systemic silencing and virus resistance.

However, the spread of RNA gene-silencing signals through graft unions has only been demonstrated in vitro, and not in vivo, for apple rootstocks transformed with a hairpin construct for the gusA reporter gene grafted onto a scion overexpressing the gusA gene (Flachowsky et al., 2012). Systemic silencing mechanisms have also been proposed in walnut, where the RNA signal produced by the rootstock was only detected in the kernels of the WT scion and in very low concentrations (Haroldsen et al., 2012a ). These limitations could explain why the systemic spread of the resistance to the WT scion from the only virus-resistant transgenic grapevine rootstock has not been further investigated (Vigne et al., 2004; Lusser et al., 2012). More recently, it has been demonstrated that siRNA can be transported long distances (1.2 m above the graft union) in woody plants, thereby alleviating viral disease by gene silencing (Zhao and Song, 2014). The success of this technology depends on optimizing expression levels in the rootstocks and the mobility and efficiency of the siRNA interaction with the target gene in the scion (Lemgo et al., 2013).

Rootstocks for plant breeding and improvement

Graft hybridization and epigenetics

Grafting has also contributed to horticultural improvement and development by creating new genetic variability and as a tool for vegetative propagation. The creation of new varieties by asexual hybridization inducing heritable changes by grafting is known as ‘graft hybridization’. Although Darwin first proposed this concept (Darwin, 1868) and the existence of graft hybrids produced from the cellular tissue of two different plants has been extensively documented, its acceptance is still controversial, since the phenomenon is commonly perceived as ‘simple’ chimaeras (Liu, 2006; Wu et al., 2013). Graft hybridization provides important evidence in favour of Darwin’s notions about Pangenesis, a developmental theory of heredity that helps to explain non‐Mendelian inheritance in grafted fruit trees. Nowadays, graft hybrids are divided into two categories—chimaera graft hybrids (so‐called graft chimaeras) and non-chimaera graft hybrids (so‐called vegetative hybrids) (Fig. 7). These differ with respect to grafting methods, characteristics, and potential mechanisms (Liu, 2006).

Fig. 7.

Schematic figure explaining the methods for obtaining graft chimaeras (upper panel) and graft hybrids (lower panel) (adapted from Liu, 2006. Historical and modern genetics of plant graft hybridization. Advances in Genetics 56, 101–129. Copyright 2006, with permission from Elsevier). A branch of the scion is grafted onto the rootstock (A), the scion is cut at the graft union (B), to allow the growth of a sectorial graft chimaera from the junction (C). The ‘mentor graft’ method consists of grafting a young seedling (cotyledonary stage) onto an old rootstock (D) and eliminating all the leaves except the two to three upper ones from the scion during the growing period (E).

Graft hybridization plays an important role in guiding parental selection in situations involving sexual hybridization associated with fruit‐tree breeding. As reviewed by Liu (2006), Michurin was one of the first investigators to use not only interspecific but also intergeneric hybridization between taxonomically remote species, thus obtaining dozens of valuable fruit and ornamental varieties. To overcome the resistance to crossing in distantly related species, he developed the method of ‘preliminary vegetative approximation’, which combines graft hybridization and sexual hybridization and has been widely applied not only to fruit trees but also to annuals. Using this method, Michurin and others (Hall, 1954; Nirk, 1959; Evans and Jones, 1964; Rao and Ram, 1971) obtained sexual hybrids from different species and genera in fruit trees (apricot×plum, sweet and sour cherry) and annuals (wheat×rye; S. lycopersicum×Solanum peruvianum). Moreover, several independent groups of scientists repeatedly showed that mentor grafting (old scion grafted onto young seedling stock or vice versa) was a simple and effective means of plant breeding, and that graft-induced variant characteristics were stable and inheritable (Taller et al., 1999; Liu, 2006).

By applying such methods, workers in the Soviet Union (Michurin, 1949; Morton, 1951; Glushcenko, 1963; cited in Liu, 2006) and China (Zhang et al., 2002; Liu et al., 2004) created many new fruit and vegetable varieties with excellent quality and high yields. Moreover, new characters induced by grafting are stable traits and can be used as novel genetic source material in Solanaceae and other species (Hirata et al., 1994; Taller et al., 1999; Liu, 2006). Genetic studies revealed the absence of Mendelian inheritance in the first generation of the selfed graft-induced variants but detected rootstock-specific randomly amplified polymorphic DNA markers in the scion, suggesting DNA transfer from the rootstock to the scions (Hirata et al., 1994). Exchanges of both RNA and DNA molecules between both partners through long‐distance mRNA translocation in the phloem (Lucas et al., 2001) or through horizontal bidirectional gene transfer via either large DNA pieces or an entire plastid genome (Stegemann and Bock, 2009) may help to explain the molecular basis for grafting-induced genetic variation. Indeed, traffic of macromolecules (including proteins and nucleic acids) between plant cells is most promiscuous in young, undifferentiated tissues with larger plasmodesmata to ease molecular transport (Ueki and Citovsky, 2005).

Importantly, graft hybridization may affect multigenic quantitative characters either alone or in combination with qualitative characters and may provide simpler alternatives to in vitro transformation, mutagenesis, somatic cell hybridization, and regeneration. Moreover, this technique not only may reduce the time needed to produce new varieties compared with conventional breeding but also, by breaking down linkage relationships, may allow selective transfer of desirable genes alone from relatively distantly related species (Liu, 2000, 2006; Liu et al., 2004). The new graft hybrids can be either clonally propagated or can be used as a source of new genetic variability in seed-propagated species (Taller et al., 1999). Some graft hybrids can boost food production, as in cassava, the principal food for more than 800 million people in the tropics and subtropics (FAO, 2013). For example, clones from a periclinal graft chimaera obtained from the epidermal tissue of a cultivar (Manihot esculenta) and the subepidermal tissue from the wild species Manihot fortalezensis produce 5-fold more edible roots than common varieties (Bomfim and Nassar, 2014). Similar results were obtained by grafting a cassava rootstock with its inedible relative the Ceara rubber tree (Manihot glaziovii) as scion (De Bruijn and Dharmaputra, 1974). Epigenetic interactions might explain the root phenotype.

Epigenetics is defined as the interactions of genes with their environment to ‘bring the phenotype into being’ (Waddington, 1942) and is considered a semi‐independent non‐Mendelian inheritance system, which enables environmentally induced phenotypes to be transmitted to subsequent generations (Jablonka and Lamb, 1998; Liu, 2006). Since DNA methylation responds to internal and external perturbations such as grafting and environmental conditions (Wu et al., 2013), it will be interesting to know how this epigenetic alteration influences the creation of new genetic variability in grafted plants and/or its adaptive capacity to environmental stresses.

Recently, global DNA methylation levels and locus-specific methylation patterns were analysed by a methylation-sensitive amplified polymorphism marker and locus-specific bisulfite sequencing in seed plants (WT controls), self- and heterografted scions/rootstocks, selfed progenies of scions, and their seed-plant controls, involving three Solanaceae species (Wu et al., 2013). Heterografting extensively altered DNA methylation patterns in a locus-specific manner (especially in scions) and the steady-state transcript abundance of genes encoding a set of enzymes functionally relevant to DNA methylation. Furthermore, altered methylation patterns in heterografted scions could be inherited by sexual progenies (Wu et al., 2013). The authors concluded that interspecific grafting in plants produced extensive and heritable epigenetic alterations in DNA, providing new insights into the molecular mechanisms for the still contentious concept of graft hybridization that could open up new commercial opportunities.

Flowering and seed production

Tree grafting is widely used to avoid the juvenile phase of seedlings and to induce flower precocity (Mudge et al., 2009). Similarly, grafting has been used to maintain germplasm collections in herbaceous species R×S interactions and to breed Solanum spp. by helping to overcome flowering-related problems and tolerance to soil pathogens and environmental conditions (Chetelat and Petersen, 2003). For example, Solanum juglandifolium (which typically refuses to flower under greenhouse conditions) and its close relative Solanum ochranthum (too tall at flowering) are difficult to handle, while the xerophyte Solanum sitiens is hypersensitive to soil-borne fungal pathogens, usually exacerbated by overwatering or transplant stress. Rootstocks of S. lycopersicum cv. VF36 promoted flowering in S. juglandifolium (Rick, TGC 37:62), but scions lost vigour as the rootstock was sensitive to root diseases. However, the interspecific hybrid F1 resulting from S. lycopersicum cv. VF36× Solanum pennellii LA0716 resulted in a very promising rootstock with resistance to several races of Fusarium wilt and high vigour that allowed growth under high soil-borne pathogen pressure. This F1 rootstock allowed the maintenance of the above-described species, with a daylength-insensitive flowering and a wide graft compatibility with other solanaceous species, including eggplant (Solanum melongena) and pepper (Capsicum spp.). Similarly, grafting onto flowering-inducing rootstocks is utilized in soybean breeding programmes in particular geographical areas and environmental conditions affecting flowering and fruit set (Kiihl et al., 1977) and to increase flower and seed production in sweet potato (Lardizabal and Thompson, 1990).

In agreement with Michurin’s observations (Liu, 2006), nowadays, hybrid seed production on grafted plants is a common commercial practice to improve seed production and quality, although a putative genetic imprint of the rootstock on the seed offspring cannot be ruled out. Indeed, experiments of this kind have revealed that grafting induces a low frequency of hereditary changes in annual plants (such as 1–22% in peppers, 20% in eggplants, and 10–15% tomatoes; Liu, 2006, and references therein). While the frequency and impact of grafting on genetic purity of commercial hybrids still requires proper industrial and scientific evaluation, using selected rootstocks not only increases seed production and quality (for agriculture and human/livestock nutrition) but also generates new genetic variability that can be incorporated into breeding programmes.

Rootstocks for transgrafting

Transgrafting can be defined as the use of a genetically engineered rootstock to support a WT scion, or vice versa, and has been considered a promising biotechnological tool to address food security ((Lusser et al., 2012; Lemgo et al., 2013). Applications include management of fan leaf virus in grapes (Gambino et al., 2010), fungal diseases in citrus (Mitani et al., 2006), and pathogen damage in tomato (Haroldsen et al., 2012a ). Creating new transgenic rootstocks to resist biotic and abiotic stresses, especially those for which resistance genes are rare in the plant genome, is a promising development (Lusser et al., 2012). Another advantage of transgrafting is the potential to minimize consumer concerns of transgene flow, since the non-transgenic scion would only produce non-transgenic pollen, provided that axillary stems from the rootstocks are eliminated (Lev-Yadun and Sederoff, 2001; COGEM, 2006). Secondly, obtaining a virus-resistant transgenic rootstock might reduce regulatory costs, as a single transgenic rootstock can be used to propagate several cultivars (Kalaitzandonakes et al., 2007).

The use of RNA signalling technology in modified roots represents an excellent opportunity to induce virus resistance(s) but maintain production of non-transgenic fruits, although the presence of the trans-genetic material in the shoot tissues may be a concern, considering that both mRNAs and siRNAs can cross a graft junction, as stated above. In Nicotiana benthamiana, a silencing signal can move across graft junctions, from a silenced rootstock to a non-silenced scion (Kasai et al., 2011). Initially, this movement was reported to be unidirectional from rootstock to scion in both tobacco and sunflower (Palauqui et al., 1997; Hewezi et al., 2005), and preferentially from scion to rootstock in Arabidopsis (Dunoyer et al., 2010; Molnar et al., 2010). Subsequently, it was demonstrated in N. benthamiana that silencing signals can be transmitted bidirectionally from silenced rootstock to non-silenced scion and vice versa (Tournier et al., 2006). Although this species is not agronomically interesting, its silencing capacity could be transferred, through intergeneric grafting, to other economically important crops, such as open-field and greenhouse tomatoes, where viruses such as tomato yellow leaf curl virus are a serious problem in many countries.

Researchers have also successfully employed strategies that utilize the expression of siRNAs to protect plant roots and tubers from pests and pathogens (Escobar et al., 2002; Klink and Matthews, 2009). Systemic protection has been achieved by grafting WT scions onto genetically modified rootstocks that target cyst nematode genes, including those associated with stimulating root growth in infected soybean (Huang et al., 2006; Steeves et al., 2006; Klink and Matthews, 2009). The systemic protection may be achieved in a manner similar to the virus resistance reported in tobacco (Smirnov et al., 1997) and more recently in cassava (Yadav et al., 2011). Aside from pathogen resistance, downregulation and/or epigenetic modification of transcripts and genetic networks in the scion or the rootstock could influence traits such as flowering time and fruit production or quality, or root characteristics such as potato tuberization (Martin et al., 2009). Therefore, this technology could quickly alleviate the devastating social effects of some pathogens such as cassava brown streak virus, which destroys the edible root in some African countries, where small family farms grow more than 90% of crops.

Research on protein production by transgenic rootstocks for delivery to scions arguably is more advanced than analogous work with the use of nucleic acids. For example, grape rootstocks can deliver hybrid lytic peptides to the scion to control bacterial and fungal diseases (Dutt et al., 2007; Gray et al., 2007). Similarly, bidirectional transport of phytochelatins (PC), peptides that function in heavy-metal chelation and detoxification in plants and fungi by sequestering metals into the vacuole, has been demonstrated using PC-defective mutants, transgenic plants over expressing PC synthase, and reciprocal grafting in Arabidopsis (Gong et al., 2003; Chen et al., 2011). These studies suggest important practical implications for increasing crop resistance to heavy-metal stress (i.e. Cd, Hg, Pb, and As) and soil nutrient concentrations (i.e. Fe, Cu), and may limit heavy-metal consumption in the human diet. For example, some rootstocks like Solanum torvum efficiently restrict Cd loading into the xylem (by 22%) and accumulation (by 70%) by S. melongena scions (see Savvas et al., 2010, for a review). Identifying, creating, and exploiting natural or biotechnological rootstocks with altered capacity to transport or exude PC could help to regulate the uptake, transport and accumulation of toxic ions in the leaves and edible parts of the plant.

Another interesting example of transgrafting is rootstock transmission of herbicide (glyphosate) resistance to a non-resistant scion in soybean (Jiang et al., 2013). Grafting conventional (CN; non-transgenic and glyphosate sensitive) soybean scions onto Roundup Ready (RR; glyphosate resistant) soybean rootstocks transferred glyphosate resistance (to 0.84 and 1.68kg ha^–1 applications), but this was achieved not in the opposite (RR scion-to-CN rootstock) direction. The glyphosate resistance trait, conferred by the shikimate biosynthetic enzyme CP4 5-enolpyruvylshikimate-3-phosphate synthase, was detected in the scion of CN/RR plants but at only 0.001% of that detected in RR leaf. Since this concentration seems insufficient to explain the glyphosate resistance observed in CN/RR plants, other factors such as aromatic amino acid and/or other herbicide-affected substances (i.e. auxins, folic acid) systemic trafficking and/or tissue specific glyphosate resistance (i.e. root detoxification of glyphosate transported from the leaves) are more likely explanations (Jiang et al., 2013). Although soybean belongs to a not (yet) conventionally grafted family (Fabaceae), these results suggest the potential of the approach and its possible implementation in crop species with much more impact in securing food production for the increasing population in the near future.

In designing and utilizing transgrafting strategies, it will be important to consider the mechanisms that regulate long-distance translocation of DNA, RNA, small RNAs, hormones, and metabolites to assess the durability and efficacy of alternative strategies. While advances to date have focused on delivery of single gene products with specific functions to scions, future advances may target transport of transcription factors that influence expression of multiple genes, which could coordinate concerted scion responses to complex challenges such as pathogens, pests, or abiotic stresses (Haroldsen et al., 2012b ). Key questions regarding regulatory consideration also must be assessed as this technology matures and research projects approach commercial situations and real social problems affecting food security, as mentioned above. Alternatively, cis-grafting, where an elite variety can be grafted on its own genetically manipulated root system (without involving foreign genes) may be more acceptable to the public.

Conclusions

Inosculation, the naturally joining of vascular tissues, occurs in tree branches and more often roots of the same (and sometimes different) species. Natural root grafts have been reported in about 200 species (Graham and Bormann, 1966) and probably exist in thousands of others (reviewed by Lev-Yadun, 2011). The grafted trees form functional clusters transferring water, minerals, organic materials and developmental signals but also diseases. There are several probable benefits of being connected to the root systems of other individuals, and these seem to have been the evolutionary agents that selected for enhanced formation of root grafts. In James Cameron’s film Avatar (2009), the mother tree, Eywa (as the Michurinian’s ‘mentor’), is the guiding force and deity of the world (Pandora), and its inhabitants (the Na’vi) believe that Eywa acts to keep the ecosystem of Pandora in perfect equilibrium. Although root grafting is probably a much simpler and real technology than the neuroconductive system connecting all living things on Pandora to Eywa, the ecologically and economically important phenomenon of cooperation between individual plants has not received the attention that it deserves (Lev-Yadun, 2011). Besides inosculation and its ecological role, man-driven grafting has solved important problems in agriculture, first in fruit trees and more recently in vegetable crops. The development of plant physiology, genetics, and molecular biology provides new tools not only to explain old phenomena but also to create a scientific basis for further development of plant grafting in general and of vegetable grafting in particular. Much more work is needed to understand the R×S and the R×S×E interactions in terms of mobile signals, gene expression, and the genetic and epigenetic control of favourable changes in the plant above and below ground. Gaining insights about the functioning and role of these natural and artificial communication processes between and within plants could bring new knowledge and agricultural applications through grafting, by identifying, creating, and exploiting new genetic biodiversity, and thereby improving the vigour, quality, sustainability, and pathogen/pest and environmental stress resistances of the food-producing crops.

Glossary

Abbreviations:

ABA

abscisic acid

BR

brassinosteroid

CK

cytokinin

IAA

indole-3-acetic acid

MT

Micro-Tom

PC

phytochelatins

QTL

quantitative trait locus

RIL

recombinant inbred line

siRNA

small interfering RNA

t-Z

trans-zeatin

WT

wild type.