The evolution of root hairs and rhizoids

Abstract

INTRODUCTION

The emergence of the first land plants sometime before 470 million years ago was a pivotal event in Earth history, which had far-reaching effects on the atmosphere and climate, and made possible subsequent invasions of the land by animals and the establishment of complex terrestrial ecosystems (Berner, 1997; Bateman et al., 1998). Land plants evolved from streptophyte algae, and a number of innovations were involved in their adaptation to terrestrial life. Among these was the evolution of rooting systems for anchorage, water uptake and nutrient acquisition, which was a key step that allowed the spread of plants on dry land (Bateman et al., 1998; Raven and Edwards, 2001). Others included the elaboration of meristems and complex tissue systems leading to the formation of complex land plant body plans; the evolution of desiccation tolerance, which allowed plants to survive life in the air; and ultimately the evolution of water transport systems, which allowed plants to move water from soil stores to the tops of tall trees. The appearance of land plants also had major impacts on geochemical cycles. Through the physical action of roots and the secretion of acids into the rhizosphere, plants greatly accelerated the weathering of silicate minerals. This increase in the rate of the reaction of atmospheric CO₂ with calcium and magnesium silicates caused a shift in the equilibrium of the long-term carbon cycle and drastically reduced atmospheric CO₂ levels, causing global climate cooling (Berner, 1997; Raven and Edwards, 2001; Lenton et al., 2012).

Filamentous cells develop at the interface between almost all land plants and the substrate in which they grow at some stage during the plant's life cycle. Root hairs form on the root surface of sporophytes (the multicellular diploid phase of the life cycle) in vascular plants. Rhizoids develop on the free-living gametophytes (the multicellular haploid phase of the life cycle) of extant vascular and non-vascular plants and on both the gametophytes and rootless sporophytes of extinct rhyniophytes. Extant lycophytes and monilophytes develop both free-living gametophytes and free-living sporophytes. These gametophytes and sporophytes grow in close contact with the substrate and develop rhizoids and root hairs, respectively. Both rhizoids and root hairs elongate by tip growth (Carol and Dolan, 2002; Pressel et al., 2008). During tip growth, growth is restricted to a small area at the apex of the elongating cell. This contrasts with the diffuse growth typical of most cell types in plants, in which growth occurs over more extensive areas of the cell surface.

This review outlines the diversity of filamentous, tip-growing cells that interface with the substrate of land plants and their aquatic ancestors. The literature describing their function and the genetic mechanisms that control their development is reviewed.

PHYLOGENETIC RELATIONSHIPS AMONG LAND PLANTS AND RELATED STREPTOPHYTE ALGAE

Land plants and the streptophyte algae (Charales, Coleochaetales, Zygnematales, Klebsormidiales, Chlorokybales and Mesostigma) together constitute a monophyletic group called the streptophytes. Streptophyte algae are a paraphyletic group, in that they do not include all descendants of a single common ancestor, while the land plants are monophyletic (include all descendants of a single common ancestor). It is unclear which algal group is most closely related to the land plants. It is likely that one of three groups, Coleochaetales, Charales or Zygnematales, is sister to the land plants. Further phylogenetic analyses are needed to define unequivocally the closest algal relative of the land plants (see, for example, Karol et al., 2001; Qiu et al., 2006; Finet et al., 2010; Wodniok et al., 2011; Leliaert et al., 2012) (Fig. 1). Three early diverging clades of land plants, liverworts, mosses and hornworts, are generally held to constitute a paraphyletic grade known as the bryophytes, though some recent molecular phylogenies do resolve the bryophytes as a monophyletic group (for example Finet et al., 2010). There is evidence that the earliest land plants had affinities to extant liverworts, and that liverworts are the earliest diverging land plant lineage (references in Kenrick and Crane, 1997a; Karol et al., 2001; Wellman et al., 2003; Qiu et al., 2006, 2007; Rubinstein et al., 2010). The vascular plants, which develop water-conducting tissues made up of cells with thickened lignified walls (xylem), are a monophyletic group that includes the lycophytes, the monilophytes (ferns and horsetails) and the seed plants (Kenrick and Crane, 1997b; Qiu et al., 2006). These phylogenetic relationships provide an evolutionary framework for understanding the distribution of filamentous cells at the interface between plant and substrate (Fig. 1).

RHIZOIDS AND ROOT HAIRS DEVELOP AT THE PLANT–SUBSTRATE INTERFACE IN STREPTOPHYTE ALGAE AND LAND PLANTS

Rhizoids develop in the haploid phase of some of the streptophyte algae, such as Chara (Charophytales) and Spirogyra (Zygnematales), but not in others such as the Coleochaetales (Lewis and McCourt, 2004). Rhizoids are unicellular in the Zygnematales and multicellular in the Charales. Rhizoids do not form in the diploid phase of the life cycle of streptophyte algae, which is unicellular and consists only of a zygote that undergoes meiosis. In contrast, the land plant life cycle consists of two distinct multicellular phases, comprising the diploid sporophyte and the haploid gametophyte. The gametophyte produces gametes that fuse to form a zygote that undergoes mitosis to form the multicellular diploid sporophyte. In turn, cells of the sporophyte undergo meiosis to form haploid spores that divide to form multicellular haploid gametophytes. Life cycles with multicellular haploid and diploid phases are said to display alternation of generations (Hofmeister, 1851; Strasburger, 1894; (Kenrick and Crane, 1997a).

In the earliest diverging land plant lineages, the liverworts, mosses, and hornworts, the gametophyte is the only free-living stage of the life cycle. As this phase of the life cycle is in direct contact with the substrate, the gametophyte develops a system of rhizoids. In contrast, the relatively simple sporophyte is either entirely (liverworts and mosses) or mostly (hornworts) nutritionally dependent upon the haploid phase, and does not make contact with the substrate or develop rhizoids (McManus and Qiu, 2008). Liverwort and hornwort rhizoids are unicellular, but those of mosses are multicellular (Crandall-Stotler and Stotler, 2008; Goffinet et al., 2008; Renzaglia et al., 2008).

In contrast to the bryophytes, vascular plants evolved hair-bearing axial organs that anchor the sporophyte and are involved in the absorption of nutrients and water. These organs, roots, possess unique defining characteristics including the formation of a protective root cap at the distal, growing end of the axes, and endogenous branching, in which lateral roots are derived from cells in the centre of the root (the pericycle). This contrasts with shoots where there is no cap and shoot cells at and near the surface of axes develop into branches (Raven and Edwards, 2001). Almost all roots develop filamentous cells (root hairs) along their surface at the plant–soil interface. These hairs have been shown to be important for nutrient uptake (discussed in detail below). In some vascular plants, namely monilophytes and lycophytes, the gametophyte is still present as a free-living but ephemeral organism and develops rhizoids (Banks, 1999, 2009) (Fig. 1). In contrast, the gametophyte is retained by and parasitic upon the sporophyte in all seed plants and reduced to a few cells (the pollen grains and embryo sac) in the angiosperms, where no rhizoids develop.

ROOTS EVOLVED AT LEAST TWICE IN THE LAND PLANTS

Fossil evidence indicates that roots had evolved among the lycophytes by the Early Devonian. For example, Asteroxylon mackei, found in the 411 million-year-old Rhynie chert, has simple root-like structures that contrast with the leafy shoot (Kidston and Lang, 1920; Kenrick and Crane, 1997b). Roots most probably evolved independently in other vascular plants from rootless ancestors (Kenrick and Crane, 1997b; Gensel et al., 2001; Raven and Edwards, 2001); although it has been argued that roots evolved only once in the vascular plants (Schneider et al., 2002), this is rebutted by phylogenetic analyses that consider fossil taxa (Friedman et al., 2004). It has also been suggested that roots evolved twice within the euphyllophyte clade because the orientation of the root axis relative to the shoot axis is different in the embryos of seed plants and monilophytes (Gensel and Berry, 2001; Raven and Edwards, 2001). This hypothesis remains to be tested phylogenetically, and difficulties may arise because of the poor preservation of this stage of the life cycle in the fossil record. Despite the probable independent evolution of root axes in different groups of plants, root hairs are found on the roots of the sporophytes of all major vascular plant lineages (Dittmer, 1949; Pearson, 1969; Banks, 2009) (Fig. 2).

ROOT HAIRS ARE IMPORTANT FOR NUTRIENT UPTAKE

Most angiosperms form root hairs at some stage during the development of the root system. They may be very short in some species such as onion (Alium cepa) or much longer in other species such as members of the Brassicales (reviewed in Jungk, 2001). Root hairs play a crucial role in the uptake of essential inorganic nutrients from the soil (Nye, 1966; Föhse et al., 1991; Gahoonia and Nielsen, 1997, 1998). These essential nutrients are taken up in ionic form from the soil water at the root surface (Marschner, 2012). As the nutrient is transported into the plant it is replaced at the root surface by diffusion if it is present in sufficient concentrations in the soil water. Nitrate and ammonium are soluble and diffuse through the soil water, thereby replenishing the supply of these ions at the root surface. Phosphate on the other hand is not mobile in the soil water because of its tendency to bind to clay particles and form insoluble precipitates in the soil (Brady and Weil, 2008). As a result, there is little diffusion of phosphate through the soil water to the root surface, where its concentration remains low after uptake into the root. Consequently, the concentration of phosphate of soil water in the vicinity of the root remains low. This region surrounding the root where nutrients are present in very low concentrations is known as a depletion zone. The length of root hairs determines the size of this zone. Plants with long root hairs develop large depletion zones (the zone has a relatively large diameter), while plants with short root hairs develop smaller depletion zones. Long root hairs enable the plant to extract nutrients from a greater volume of soil compared with plants with short root hairs (in the absence of mycorrhizae) (Nye, 1966; Gahoonia and Nielsen, 1996, 1998; Gahoonia et al., 1997). This explains why root hair length is positively correlated with the ability to absorb phosphate from the soil. Cultivars of barley with short root hairs take up less phosphate than cultivars with long root hairs when grown in field conditions with low available phosphate (Gahoonia and Nielsen, 1998, 2004). Furthermore, root hairless barley mutants take up less phosphate from the soil and yield much less grain than wild-type (long hair) cultivars in low phosphate conditions (Gahoonia et al., 2001). The role of root hairs in the uptake of other nutrients with limited mobility in the soil has also been demonstrated. For example, the rate of K⁺ uptake from the soil is also positively correlated with root hair length (Jungk, 2001). Together these data indicate that root hairs are important for the uptake of relatively immobile ions such as phosphate from the soil.

Mycorrhizae are symbioses between fungi and plants in which fungi provide inorganic nutrients to the plant in exchange for reduced carbon compounds (Parniske, 2008). Approximately 80 % of land plants develop symbioses with glomeromycotan fungal partners called vesicular arbuscular mycorrhizae (AM). Glomeromycota are obligate symbionts, have been found in 411 million-year-old fossil plants from the Rhynie Chert and are likely to have coevolved with the land plants (Remy et al., 1994; Wang et al., 2010). They form extensive hyphal networks in the soil and interface with the plant at branched intracellular structures called arbuscules where nutrients and carbon compounds are exchanged. There is a general trend among angiosperms that those species that develop mycorrhizae tend to form relatively short root hairs, while those that do not form mycorrhizae develop relatively long root hairs (Baylis, 1975; St John, 1980). Given that the ability to form mycorrhizae is an ancestral state among land plants, and that many unrelated groups of plants do not form mycorrhizae, it can be inferred that the ability to form mycorrhizae was lost independently among different lineages of plants (such as the Brassicales) (reviewed in Parniske, 2008). These non-mycorrhizal plants have evolved long root hairs that facilitate nutrient uptake in the absence of the fungal symbiont. Extreme examples of this are found among the Cyperaceae, where some species that lack mycorrhizae growing in nutrient-poor soils develop ‘dauciform roots’ where very long root hairs develop in patches along the root system. The formation of cluster roots with very long root hairs among species of the Proteaceae that lack mycorrhizae is another example where long root hairs provide the ability to extract limiting nutrients from the soil in the absence of mycorrhizae (reviewed in Lambers et al., 2010).

RHIZOIDS HAVE ROLES IN ANCHORAGE AND THE UPTAKE OF WATER AND NUTRIENTS

To date there have been few studies on the function of rhizoids, and what knowledge there is has been gleaned from diverse species. Nevertheless there are indications that rhizoids are important in anchorage and the uptake of water and nutrients. The examples presented here are from species that are distantly related phylogenetically, and it should be borne in mind that rhizoids may not carry out all of the above functions in every species.

It is often asserted in the literature that the primary role of rhizoids is in attachment to the substrate (Duckett et al., 1998; Goffinet et al., 2008; Crandall-Stotler et al., 2009). The rhizoids of many liverworts form discs or ramify at their tips when they contact solid particles and adhere strongly (Haberlandt, 1914; Odu and Richards, 1976; Pocock and Duckett, 1985; Duckett et al., 1991). Similar branching has also been observed at the tips of moss rhizoids in contact with hard substrates (Duckett, 1994a; Pressel and Duckett, 2009), as well as in the rhizoids of filmy fern gametophytes (Hymenophyllaceae) (Duckett et al., 1996), while moss rhizoids can also display thigmotropic responses, coiling around objects in the substrate (Duckett, 1994b; Duckett and Matcham, 1995). A role in attachment is also suggested by the observation that, at least among the highly branched pleurocarpous mosses, rhizoids are more abundant and highly branched in plants growing on bare, hard substrates such as rocks than those growing on soil (Odu, 1978). Attachment to the substrate may be facilitated by the production of adhesive sulfated non-cellulose polysaccharides by rhizoid tips (Odu, 1989).

Rhizoids have also been shown to be involved in the uptake and transport of water. Many bryophytes are ectohydric, i.e. they lack thick cuticles and absorb water over their whole surface (Proctor, 2000). Though rhizoids are not required for direct uptake of water in these species, many mosses produce a tomentum, a thick covering of rhizoids growing from the stem, and the spaces that form between the hairs aid water transport by capillary action (Proctor, 1984). In contrast, some bryophytes are endohydric, with internal water transport. The rhizoids of the endohydric moss Polytrichum have been shown to take up water from the substrate, though the importance of this route of water uptake is probably minor compared with uptake across aerial surfaces of the plant (Mägdefrau, 1938; Trachtenberg and Zamski, 1979). In the complex thalloid liverworts of the Marchantiales, rhizoids are involved in the uptake and transport of water from the substrate. These liverworts possess two kinds of rhizoids: smooth-walled rhizoids and tuberculate rhizoids, the latter having peg-shaped thickenings that project into the lumen of the rhizoid. These thickened rhizoids form bundles (like the moss tomenta) that run along the ventral surface of the thallus. Aqueous dyes are rapidly transported by capillarity between these rhizoids (Bowen, 1935; Czaja, 1936; McConaha, 1939, 1941; Kobiyama and Crandall-Stotler, 2011). In addition to this external movement along bundles of rhizoids, it has been shown that water can travel inside both smooth-walled and tuberculate rhizoids, into the cells of the thallus surrounding the rhizoid base (Kamerling, 1897; Clee, 1943). In Conocephalum conicum and C. japonicum, the movement of water from the rhizoids into the thallus is promoted by specialized pitted cells on the ventral surface (Kobiyama and Crandall-Stotler, 2011). Together, these observations indicate that rhizoids are important for water transport in the Marchantiales.

Rhizoids may be active in inorganic nutrient uptake in diverse species. The rhizoids of Chara species grow into the substrate where they play an important role in anchoring the plant. In addition, these rhizoids contains a higher concentration of mineral nutrients than the open water (Barko et al., 1991) and take up nitrate, ammonium and phosphates from sediments (Box, 1986; Vermeer et al., 2003). There are, to our knowledge, no reports demonstrating the role of liverwort rhizoids in nutrient uptake. However, they often form mycorrhiza-like associations with fungi (Read et al., 2000; Russell and Bulman, 2005), which can substantially increase the uptake of nutrients from the soil (Humphreys et al., 2010). Nutrient acquisition by mosses is also not well understood, but it is generally thought that the majority of mosses get most of their nutrients from precipitation and the deposition of dust (Bates, 1992). Soil-growing mosses have been shown to be able to obtain nutrients from the substrate (Chapin et al., 1987; Bates and Farmer, 1990; Van Tooren et al., 1990), though it has not been shown whether this is a result of direct absorption by the rhizoids or external transport of nutrient-bearing soil water over the plant surface to leafy parts.

Although the evidence is fragmentary, it seems that root hairs and rhizoids carry out similar functions, albeit probably to different extents in different species.

DO THE SAME GENES CONTROL THE DEVELOPMENT OF ROOT HAIRS AND RHIZOIDS?

Much of our knowledge about the molecular and cellular events during the development of root hairs comes from studies in the model angiosperm Arabidopsis thaliana, where the mechanisms of tip growth have been characterized in some detail (reviewed, for instance, in Libault et al., 2010). Less is known about the molecular and cellular events that control the development of rhizoids.

Moss rhizoids develop on the gametophyte and, unlike root hairs and the rhizoids of liverworts and hornworts, they are multicellular. Root hairs and rhizoids probably have similar functions as well as similar modes of growth and development. However, it might be argued that they are analogous because they are not structurally correspondent, since they are produced by different phases of the life cycle (Scotland, 2010). It is possible that the resemblance between rhizoids and root hairs is a result of convergent evolution because these cell types have similar functions. Since the developmental mechanism for producing rhizoids already existed in early land plants, it is possible that it was co-opted during the rise to dominance of the sporophyte, and deployed in a new context to produce root hairs. Do rhizoids and root hairs share an ‘ancient toolkit’ of developmental genes?

There are preliminary indications that rhizoid and root hair development may indeed share a common genetic mechanism. Two related basic helix–loop–helix transcription factors, ROOT HAIR DEFECTIVE 6 (AtRHD6) and ROOT HAIR DEFECTIVE 6-LIKE 1 (AtRSL1), control root hair development in Arabidopsis. These proteins accumulate in cells that will go on to develop root hairs, where they promote the transcription of genes necessary for tip growth such as ROOT HAIR DEFECTIVE 6-LIKE 4 (RSL4) (Menand et al., 2007; Yi et al., 2010). Mutants that lack AtRHD6 and AtRSL1 function do not develop root hairs (Menand et al., 2007; Yi et al., 2010). Two similar genes were identified in the genome of the model moss Physcomitrella patens and named PpRSL1 and PpRSL2. Few rhizoids develop in double mutants that lack both PpRSL1 and PpRSL2 function (Menand et al., 2007). Furthermore, constitutive co-expression PpRSL1 and PpRSL2 transforms the gametophore into a mass of rhizoids (Jang et al., 2011). These data indicate that PpRSL1 and PpRSL2 together are necessary and sufficient for rhizoid development. This necessity and sufficiency indicate that these PpRSL genes are key regulators of rhizoid development in mosses.

Remarkably, Arabidopsis mutants lacking RHD6 function develop root hairs if transformed with the PpRSL1 gene from Physcomitrella. This indicates that the function of RSL proteins has been conserved for >420 million years since mosses and vascular plants diverged from a common ancestor (Menand et al., 2007). We propose that the RSL network was co-opted to promote the formation of root hairs, when the free-living vascular plant sporophyte increased in size and came into contact with the soil. It is likely that at least some components of the regulatory network downstream of the RSL genes have also been conserved during the intervening period, although this hypothesis remains to be tested.

Comparative genetic analyses have shown that changes in the expression of regulatory genes have been important in the evolution of morphological novelties in both animals and plants, and that these changes often involve modifications to the cis-regulatory regions of the genes (Carroll, 2008; De Robertis, 2008; Shubin et al., 2009; Pires and Dolan, 2012; Wittkopp and Kalay, 2012). Changes in the cis-regulatory regions of RSL genes are likely to have played a role in altering their expression during land plant evolution. RSL genes are expressed in the gametophytes of non-vascular plants but not in sporophytes. Changes in cis-regulatory elements could have promoted the expression of RSL genes in the sporophyte and repressed transcription in the gametophyte in vascular plants. According to this model, expression of RSL genes in sporophytes would then have initiated developmental programmes for rhizoid-like filamentous tip-growing cells (root hairs) in this phase of the life cycle. Such changes in the cis-regulatory regions of key genes may have played an important role in the elaboration of the sporophyte as large multicellular diploid plants rose to dominance during the Palaeozoic.

AUXIN POSITIVELY REGULATES THE DEVELOPMENT OF RHIZOIDS AND ROOT HAIRS

Auxin signalling controls the rhizoid developmental programme that was probably redeployed in the sporophyte to control the development of root hairs. Auxin has a stimulatory effect on rhizoid development in the alga Chara (Klämbt et al., 1992), liverworts (Kaul et al., 1962; Maravolo and Voth, 1966) and ferns (Hickok and Kiriluk, 1984). Physcomitrella mutants defective in auxin perception develop few rhizoids; conversely, the application of auxin to wild-type plants promotes the development of supernumerary rhizoids (Ashton et al., 1979; Sakakibara et al., 2003; Prigge et al., 2010). Auxin controls the development of Physcomitrella rhizoids by regulating the expression of PpRSL1 and PpRSL2 genes (Jang et al., 2011). This indicates that RSL genes and auxin signalling interact during the development of rhizoids.

Auxin positively regulates root hair development in angiosperms. In Arabidopsis, auxin gradients generated by auxin influx carriers are responsible for limiting the position of root hair initiation to the region of the cell closest to the root tip, since root hairs develop in more shootward parts of the cell in mutants with defective auxin signalling (Pitts et al., 1998; Grebe et al., 2002; Ikeda et al., 2009). Furthermore, auxin positively regulates root hair elongation; the root hairs of auxin signalling mutants are shorter than in the wild type (Pitts et al., 1998; Knox et al., 2003). Auxin exerts this effect at least partly by influencing the expression of the gene encoding the RSL4 basic helix–loop–helix transcription factor (Yi et al., 2010). Treatment with auxin increases the transcription of RSL4, which is required for root hair growth; these data indicate that auxin stimulates root hair elongation by increasing the transcription of RSL4. In Physcomitrella, auxin promotes the expression of different genes – PpRSL1 and PpRSL2 and not orthologs of RSL4 (Jang and Dolan, 2011). Despite this difference, these data suggest that auxin and RSL genes are part of an ancient genetic network controlling the development of rooting cells that was present in the last common ancestor of mosses and vascular plants (Jang and Dolan, 2011).

CONCLUSIONS

Character mapping on current phylogenies implies that the earliest land plant life cycles resembled that of extant liverworts in which the gametophyte was free living and the sporophyte was embedded in the gametophyte. If true, this suggests that the regulatory module controlling rhizoid and root hair development first operated in the gametophyte of early land plants. We propose that this regulatory module became expressed in sporophytes as the diploid phase became free living, developing filamentous cells at the interface of the plant and soil. If such transference of regulatory modules between phases of the plant life cycle is widespread, it could, at least in part, explain the genetic basis for the increase in the morphological diversity of sporophytes that occurred during the radiation of land plants in the Devonian Period.

OUTLOOK AND OUTSTANDING QUESTIONS

An ancient mechanism controls the development of filamentous cells at the plant–soil interface in mosses and angiosperms. How ancient is this mechanism? Was the development of filamentous rooting cells at the plant–soil interface an evolutionary novelty acquired by the ancestor of all land plants during the conquest of the land? Or was the mechanism already present in plants' algal ancestors, a pre-existing developmental network that facilitated the transition to the terrestrial environment? Characterization of the genetic networks controlling the development of rhizoids in liverworts and streptophyte algae will allow us to answer this question.

Often a trait that has evolved multiple times in independent lineages is the result of wiring the same ancestral genes in new ways, as seen, for example, in the reduction of pelvic structures in separate stickleback populations (Shapiro et al., 2004; Chan et al., 2010), the convergent evolution of butterfly wing patterns (Reed et al., 2011) and the independent evolution of wing spots in different Drosophila species (Gompel et al., 2005; Prud'homme et al., 2006). Roots bearing root hairs probably evolved independently on at least two separate occasions (among the lycophytes and euphyllophytes). Was the same network co-opted on each occasion to regulate the development of root hairs? This question can be addressed by characterizing the genetic control of root hair growth and the roles of RSL genes in the lycophyte Selaginella moellendorffii, the genome sequence of which has recently been published (Banks et al., 2011).

The possible role of changes in cis-regulatory elements in the co-option of genes controlling rhizoid development has not been examined. cis-regulatory elements specific to root hairs have been identified in angiosperms (Cho and Cosgrove, 2002; Kim et al., 2006; Won et al., 2009). The characterization of further rhizoid-specific genes and their regulatory elements in basally diverging plants will allow us to assess the importance of changes in cis-regulatory elements in re-wiring a rhizoid developmental programme to give root hairs on the sporophyte.

ACKNOWLEDGEMENTS

We thank John Baker for photographic assistance; Clemence Bonnot, Andrew Plackett, Heather Sanders, Thomas Tam and Eftychios Frangedakis for plant material; and James Doyle for helpful comments on the manuscript. This work is supported by a Newton-Abraham Studentship at the University of Oxford to V.A.S.J., and grants from the European Union–Marie Curie–Integrated Training Network (PLANTORIGINS), European Research Council (EVO500) and the Biotechnology and Biological Science Research Council of the UK to L.D.

LITERATURE CITED