Introduction

Roots are fundamental organs of vascular plants. The root performs essential functions necessary for the survival and development of plants, such as anchoring the plant body to a substrate, providing mechanical support, and absorbing water and nutrients. Despite functional similarities among all roots, phylogenetic analyses integrating key fossil taxa and extant lineages have suggested that roots evolved at least twice independently among vascular plants, once in lycophytes and once in euphyllophytes (monilophytes and seed plants) (Friedman et al. 2004; Kenrick and Crane 1997; Raven and Edwards 2001). Recent researches on Devonian fossil plants, especially lycophytes, have suggested that roots evolved several times independently from different source organs; underground shoots, rhizomorphs, basal swelling or protocorms (Doyle 2013; Hao et al. 2010; Hetherington and Dolan 2017; Matsunaga and Tomescu 2016). This hypothesis is consistent with the molecular clock-based estimate that three families of extant lycophytes (Selaginellaceae, Isoetaceae, and Lycopodiaceae) diverged in the Early Devonian (Kenrick and Strullu-Derrien 2014; Wikstrom and Kenrick 2001) before lycophyte ancestors acquired roots. However, it is not clear which organs gave rise to the roots of extant lycophytes; further neo- and palaeobotanical studies on lycophyte roots should therefore be conducted (Harrison 2017; Hetherington and Dolan 2017).

Root tissues consist of the epidermis, cortex, vascular cylinder, and root cap (Esau 1965; Evert 2006; Zhu et al. 1998). These tissues are entirely produced by the activity of the root apical meristem (RAM), a self-perpetuating tissue. RAM exhibits diversified structure and organization among the major taxa (Clowes 1961; Esau 1965), although research on these diversified structures has mainly been performed for euphyllophytes, especially for angiosperms. In contrast, information on lycophyte RAM structures is very limited. Very recently, we reported diverse root structures and cell division activities in extant lycophytes and provided evidence from these extant plants to support the multiple origin hypotheses. In the present article, we review recent discussions on the evolution and origin of roots in lycophytes, monilophytes, and seed plants, incorporating our previous results on lycophyte roots.

Comparative morphology of RAM of extant euphyllophytes and lycophytes

Euphyllophyte RAM organizations have been described separately for each major plant group. The RAM is similar across monilophytes in that it contains a single pyramidal initial cell, the apical cell (Gifford 1991, 1993; Ogura 1972). Gymnosperm RAMs have one to three initiating zones, depending on the taxon (von Guttenberg 1966; Pillai 1964, 1966). Angiosperm RAMs have two basic types of organization, open and closed, originally defined by von Guttenberg (1960). In the open RAM, all root tissues appear to arise from a common group of initial cells, whereas in the closed RAM, the epidermis, cortex, vascular tissue, and root cap are separately traceable to discrete tiers of initial cells, i.e., protoderm-, ground meristem-, procambium-, and root cap-initials (Esau 1965; Evert 2006). In addition, with respect to the origin of the epidermis, angiosperm RAMs are classified into monocot and dicot types; the epidermis has a common origin with the cortex in most monocots, but with the root cap in most dicots (Clowes 1994, 2000). Angiosperm RAMs were later further classified into 15 types (Heimsch and Seago 2008) based on differences in initial cell behavior and epidermis deviation.

Previous studies distinguished only two types of RAM organization in lycophytes. In Selaginellaceae, the RAM has an apical cell as the ultimate source of all tissues constituting roots (Imaichi and Kato 1989); Isoetaceae and Lycopodiaceae have a layered RAM structure, with multiple tiers of initial cells from which different tissues are derived (von Guttenberg 1966; Imaichi 2008; Ogura 1972; Yi and Kato 2001). Thus, RAM structural diversity has been underestimated in lycophytes, mainly due to a lack of studies on Lycopodiaceae (Bruchmann 1874; Campbell 1918; Stokey 1907). However, our recent work based on tissue organization has resulted in the recognition of three RAM types in addition to the apical cell type in Selaginella (Fig. 1, Fujinami et al. 2017). Type I RAM, which is found in some Lycopodium and Diphasiastrum species of Lycopodiaceae, has a non-layered structure with a mass of common initials, which are surrounded by procambium-, ground meristem-, protoderm-, and root cap-initials (Fig. 2a). This RAM type is reminiscent of open angiosperm RAM. Types II and III RAM similarly comprise multiple tiers of initial cells, but they are distinguished by the origin of the root epidermis. In type II RAM of Huperzia and Lycopodiella (Lycopodiaceae), initials of epidermis and of cortex, i.e. protoderm- and ground meristem-initials, form discrete cell layers that separate the root proper from root cap cells (Fig. 2b), whereas in type III RAM of Isoetaceae species, a common initial cell layer produces both the root cap and epidermis. The structure of types II and III RAM resembles that of some closed angiosperm RAM (Fujinami et al. 2017); however, these similarities evolved independently in lycophytes and euphyllophytes, as discussed below.

Fig. 1
figure 1

RAM structures mapped onto a phylogenetic tree of vascular plants. Character evolutions are based on Fujinami et al. (2017) and Hetherington and Dolan (2017, 2018, 2019). For seed plants, four RAM types are selected from Clowes (1981) and Heimsch and Seago (2008), which have similarities to those of lycophytes. Asteroxylon rooting axis is hypothesized as an ancestor of roots with type II RAM

Full size image
Fig. 2
figure 2

Schematic illustrations and EdU staining images of type I and type II RAM. a, b Schematic illustrations of type I (a) and type II (b). The figures are based on Fujinami et al. (2017). c, d EdU image of five sections and a phase-contrast of a median longitudinal section. c EdU image of Lycopodium clavatum RAM (type I). EdU signals (green) are scarce in the center of RAM. Traced lines indicate the layers of the initial cells (the center of RAM) and the boundary of root tissues. d EdU image of Huperzia serrata RAM (type II). Scale bars, 100 µm. GMI ground meristem initials, PCI procambium initials, PDI protoderm initials, RCI root cap initials

Full size image

Lycophyte RAM exhibits different cell division activity depending on RAM type

A quiescent center (QC) is generally observed at the heart of angiosperm and gymnosperm RAM (Clowes 1954, 1956, 1994; Peterson and Vermeer 1980). QC cells divide infrequently and serve as an organizing center to maintain the stem cell niche (van den Berg et al. 1997; Heyman et al. 2013). The presence of a QC in non-seed plants had been controversial. Some studies have argued that the root apical cell in monilophytes is quiescent, whereas others have considered the apical cell as mitotically active (Avanzi and D’Amato 1967; D’Amato and Avanzi 1965; Gifford 1991, 1993; Gifford and Kurth 1982). We analyzed the frequency and pattern of mitotic division in the four lycophyte and monilophyte RAM types using a thymidine analog, 5-ethuniyl-2′-deoxyuridine (EdU), as a proxy to identify actively dividing cells (Fujinami et al. 2017). We showed that type I RAM possesses a region with very low cell division frequency, reminiscent of the QC in angiosperm roots (Figs. 1, 2a, c). In contrast, type II and III layered RAM and apical cell-type RAM in Selaginellaceae do not show QC-like activity (Figs. 1, 2b, d, Fujinami et al. 2017). In the monilophyte Hypolepis punctata, the apical cell divided more frequently than the immediate derivative cells (Fig. 4d in Fujinami et al. 2017). Thus, a QC-like area and a QC likely evolved independently in lycophytes and seed plants. Nevertheless, it is interesting that RAM with a mass of initial cells shows quiescence in both lycophytes and seed plants. This commonality may indicate that quiescence facilitates maintenance of stem cell niche also in the type I RAM of lycophytes, although the function of the QC-like area in lycophytes remains unclear at present.

Root organization is superficially similar in lycophytes and euphyllophytes; however, the cell division dynamics observed in lycophyte RAM suggests a different mechanism of root growth from euphyllophyte RAM, supporting the independent evolution of roots in lycophytes and euphyllophytes (Doyle 2013; Friedman et al. 2004; Hetherington and Dolan 2019; Kenrick 2013; Raven and Edwards 2001).

Genes involved in establishing a QC-like area in lycophytes

In angiosperms, the proliferation rate of QC cells is controlled by the WUSCHEL-RELATED HOMEOBOX5 (WOX5) transcription factor, whose transcripts are exclusively found in QC cells (Haecker et al. 2004; Sarkar et al. 2007). Other key regulators of the QC include SCARECROW (SCR), which is required by QC for the maintenance of stem cell activity (Sabatini et al. 2003), and the ETHYLENE RESPONSEFACTOR family, a pacemaker of QC cell division (Heyman et al. 2013).

Transcriptomes for some lycophyte roots have recently been made available (Mello et al. 2019); however, very limited tissue-scale information is available on gene expression. WOX family genes have been isolated in Selaginella kraussiana (Selaginellaceae), but phylogenetic analyses showed that this species does not have a WOX5/WUS ortholog (Ge et al. 2016; Nardmann and Werr 2012, 2013; Nardmann et al. 2009). Ge et al. (2016) reported that SkWOX11C is expressed in microphylls, rhizophores, shoots, and stems, but not in roots. In our preliminary gene expression analyses, no WOX genes were expressed in the QC-like area of Lycopodium clavatum (R. Fujinami, unpublished observations). This finding may indicate that different molecular mechanisms may be responsible for the QC of seed plants and the QC-like area of lycophytes; further analyses are required to confirm this possibility.

Evolution of RAM organization and the origin of roots in vascular plants

Voronin (1969, as cited in Barlow 1995) proposed that the propensity to form centrally placed tetrahedral apical initials is primitive by assuming the apical initials of Selaginellaceae RAM as the archetype. This hypothesis further predicts that polyhedral apical initials with increased cutting faces were multiplied to form plural initials in seed plants. Such polyhedral apical initials are found in Osmunda and Angiopteris (Barlow 1995; Freeberg and Gifford 1984), which are phylogenetically basal monilophyte groups (Schuettpelz and Pryer 2008). However, the hypothesized evolutionary course of apical initials is unlikely because roots were acquired independently in lycophytes, monilophytes and seed plants (Friedman et al. 2004; Fujinami et al. 2017).

Fossil evidence rather suggests that, if the rooting axis of Early Devonian lycophyte Asteroxylon mackiei represents a primitive root without a root cap or root hairs, the ancestral organ of roots had RAM with multiple initials (Hetherington and Dolan 2018). These rooting axes form an epidermis over the surface of rooting axis meristems. This feature is similar to type II RAM in Lycopodiaceae in that epidermal (protoderm) initials and cortical (ground meristem) initials form discrete cell layers. Type II RAM also has discrete initials for the root cap. Therefore, the trajectory to type II RAM from rooting axis meristem can be explained by the addition of root cap initials outside the rooting axis proper (Fig. 1).

Many Devonian fossils suggest that roots originated at least twice within a relatively short period (Freidman et al. 2004; Kenrick and Crane 1997; Raven and Edwards 2001). A large land mass was present at high latitudes in the Southern Hemisphere prior to the Silurian; however, after it began to split, land pieces moved to mid-latitude and equatorial areas around the onset of the Devonian (Golonka 2000). This change in continent distribution triggered rapid rock weathering due to high precipitation in these areas (Le Hir et al. 2011). Consequently, rooting organs are likely to have become adaptive on the Early Devonian land. However, the primitive soil was still nutritionally poor, especially in phosphates and nitrates. Therefore, the functions of the earliest roots may have been limited to anchoring plants to pebbly substrates rather than absorbing nutrients, as in the rooting axes of Asteroxylon, which had no root hairs (Hetherington and Dolan 2018). The cultivation of land by pioneer plants would have led to accumulation of nutrients in the soil, offering an advantage to plants that could utilize these nutrients effectively. Interestingly, the earliest Devonian roots have mainly been reported in the lycophyte lineage (e.g., Hao et al. 2010; Hetherington and Dolan 2018, 2019; Matsunaga and Tomescu 2016), suggesting that euphyllophyte roots evolved later than lycophyte roots, allowing the exploitation of nutrients made available through lycophyte efforts. Such differences in the timing of root origins, as well as in their functional requirements, are likely related to multiple root origins in the evolution of vascular plants.

Conclusion and future perspectives

The diversity of RAM among extant lycophytes suggests that root evolution and diversification occurred through different evolutionary pathways, supporting paleobotanical hypothesis of the multiple origins of roots. The presence of a QC-like region in non-seed plants allowed us to compare the molecular mechanisms regulating root structure between lycophytes and seed plants, and it will contribute to the elucidation of root evolution. More importantly, lycophytes and euphyllophytes differ in the branching pattern of their roots. Lycophyte roots branch dichotomously from their apex, whereas euphyllophyte roots typically form lateral roots endogenously (Gifford and Foster 1989). This difference in root branching patterns could be the key to understanding how roots and shoots evolved from a common vascular plant ancestor that had no shoots and roots, but rather axial organs. The origin of roots will be further elucidated as we improve our understanding of branching patterns through molecular genetic analysis and further exploration of fossil evidence.