Keywords

1 Relationships of Setaria to Other Monocots

The genus Setaria is a member of subtribe Cenchrinae, tribe Paniceae, subfamily Panicoideae, family Poaceae, order Poales, in the commelinid clade of monocots (Soreng et al. 2015; Kellogg 2015). Because Setaria inherits the characters of each of these larger clades, we will consider each in turn, progressing from the most to the least inclusive.

1.1 Characteristics of Setaria Shared with the Commelinid Clade, Poales, and Poaceae

1.1.1 Commelinids

The commelinid clade includes Poales, Zingiberales (gingers and bananas), and Commelinales (spiderworts and their relatives). All members of this clade have endosperm that is well developed and persists in the seed (Stevens 2012). In addition, the commelinid monocots have unique cell walls. The hemicelluloses in the primary walls are largely arabinoxylans, specifically glucuronoarabinoxylans (Carpita 1996; Withers et al. 2012); the latter appear to be unique to the commelinids. The amount of pectin and protein is fairly low in comparison to eudicots (Carpita 1996). In the secondary walls, major lignin subunits are p-coumaric and ferulic acid (Harris and Hartley 1980; Harris and Tretheway 2009). Much of the information on the biochemistry of the commelinid secondary wall comes from grasses (Withers et al. 2012; Petrik et al. 2014; Molinari et al. 2013) but is presumed to apply to all commelinids. All commelinids accumulate silica in their leaves (Stevens 2012). Thus, Setaria could be a useful model for studies of endosperm, cell walls, and silica accumulation, with the results of such studies applying not only to grasses, but to other members of the commelinid clade.

1.1.2 Poales

The Poales in its current broad sense includes 16 families (Angiosperm Phylogeny Group 2009; Givnish et al. 2010), all of which accumulate silica specifically in the leaf epidermis (Stevens 2012). Silica accumulation protects the plant from pathogenic bacteria and fungi (Isa et al. 2010; Ma and Yamaji 2006), and also appears to reduce insect herbivory (Massey and Hartley 2009; Garbuzov et al. 2011). In addition, deposition of silica provides structural support, reduces the uptake of toxic metals, and regulates water loss (Isa et al. 2010; Ma and Yamaji 2006). One popular theory suggested that production of silica was selected as a defense against mammalian grazers because it would wear down their teeth (Simpson 1951; Baker et al. 1959). However, this idea is not supported by data (Sanson et al. 2007; Strömberg 2006).

Endosperm development in all Poales is unique among angiosperms, with multiple rounds of nuclear division before cell walls form (nuclear endosperm) (Stevens 2012). Few other morphological characters are shared by members of Poales, although many are wind pollinated, often occur in nutrient-poor habitats, and are often fire adapted (Linder and Rudall 2005). Flowers generally occur in tiny clusters called “spikelets” in both Cyperaceae and Poaceae, but the structure of these is quite different and nonhomologous in the two families.

Within Poales, Poaceae fall into the graminid clade, a well-supported group that also includes Flagellariaceae, Restionaceae (which now includes members of the former Centrolepidaceae), Anarthriaceae, Joinvilleaceae, and Ecdeiocoleaceae. Members of the graminid clade have monoporate pollen, with a raised ring or annulus around the pore (Stevens 2012), a character that is retained in Setaria. The functional significance of this pollen form is unknown. Nearly all graminids have two-ranked (distichous) sheathing leaves. The endothecium of the anther has girdle-like thickenings. Stigmas are generally plumose, with receptive cells on multicellular branches. The graminids also all share the ability to produce flavones.

The immediate sister group of Poaceae is uncertain. Possible candidates are Joinvillea, the sole genus in Joinvilleaceae, or Ecdeiocoleaceae, a family with two genera, Ecdeiocolea and Georgeantha. Current data suggest that the two families are sisters, and that the clade is then sister to Poaceae (McKain et al. 2016). Members of both families have conventional monocot flowers with two whorls of perianth parts, and thus their structure sheds little light on the homologies of the grass spikelet (but see (Preston et al. 2009; Kellogg 2015)). Like Setaria and other grasses, both Joinvilleaceae and Ecdeiocoleaceae have dumbbell-shaped stomatal guard cells. This guard cell shape is thought to enhance the speed of pore opening (Haworth et al. 2011; Franks and Farquhar 2007). Also Joinvilleaceae shares with the grasses the pattern of alternating long and short cells in the epidermis (Campbell and Kellogg 1987). As with many such morphological characteristics, the genetic controls and functional significance of this character are unknown.

1.1.3 Poaceae

Poaceae , or Gramineae (both names are correct), is the most speciose of the families in Poales. It includes ca. 12,000 species (Clayton et al. 2006 onwards; Kellogg 2015) and is clearly monophyletic (Kellogg and Campbell 1987; Kellogg and Linder 1995; Vicentini et al. 2008; GPWG 2001; GPWG II 2012).

The grasses all share a distinctive embryo and fruit (GPWG 2001; Kellogg 2015). The seed coat is generally fused to the inner epidermis of the pericarp, forming a single seeded fruit or caryopsis (Fig. 1.1a). Unlike all other commelinid monocots (and indeed most monocots), the grass embryo is highly differentiated (Kellogg 2000; Campbell and Kellogg 1987; Rudall et al. 2005), with a well-developed shoot apical meristem surrounded by a sheath-like structure, the coleoptile, and bearing two or more leaves (Sylvester et al. 2001) (Fig. 1.1a). The root apical meristem is also differentiated and surrounded by a coleorhiza. Attached to the embryo is a large shield-shaped haustorial organ, the scutellum. Together the coleoptile and scutellum appear to represent the sheath and blade, respectively, of a highly modified cotyledon (Takacs et al. 2012).

Fig. 1.1
figure 1

(a) Embryo of Setaria viridis. As in all other grasses, the Setaria embryo has well-developed shoot and root meristems and a clear scutellum. The deep cleft between the base of the scutellum and the coleorhiza is common in most grasses except for the Pooideae. en, endosperm; sam, shoot apical meristem; ram, root apical meristem; s, scutellum. (b) Apex of young bristle showing collapsed cells. Bristles appear to lose their meristems early in development. dm, degenerating meristem. Photos by John G. Hodge

Full size image

Also characteristic of Poaceae is the formation of tiny trichomes (microhairs) from the short cells of the leaf epidermis. Microhairs are two-celled, with an elongate apical cell (Johnston and Watson 1976). The functional significance of microhairs is unknown, although it is possible that they could be secretory in some instances. While the ability to produce microhairs is clearly ancestral in the grasses (GPWG 2001) and occurs in Setaria as well as all other panicoid species, the cool-season grasses in subfamily Pooideae do not produce them so they are not universal in grasses.

Poaceae genomes have been studied extensively because genomic information is so essential for breeding efforts. A whole genome duplication occurred in the common ancestor of all grasses, so that many loci are retained in duplicate (Goff et al. 2002; Yu et al. 2002; Paterson et al. 2004, 2009; McKain et al. 2016).

Poaceae is divided into 12 subfamilies (Fig. 1.2) (Kellogg 2015; Soreng et al. 2015). The ones that diverged early in the evolution of the family include only a handful of species (Anomochlooideae, four species, Pharoideae, 12, and Puelioideae, 11) (Kellogg 2015). The vast majority of species fall in the BOP and PACMAD clades, the names of which are acronyms for the included subfamilies. BOP includes Bambusoideae, Oryzoideae, and Pooideae, whereas PACMAD includes Panicoideae, Arundinoideae, Chloridoideae, Micrairoideae, Aristidoideae, and Danthonioideae.

Fig. 1.2
figure 2

Phylogeny of the grasses, based largely on GPWG II (2012) and redrawn from Kellogg (2015). Arrow points to subfamily Panicoideae, tribe Paniceae, which includes Setaria

Full size image

The grass spikelet, which is a tiny spike delimited by two bracts (glumes) and with one or more flowers, characterizes all grasses except Anomochlooideae. The precise timing of origin of the spikelet, however, is unclear. Either the spikelet was present in the common ancestor of all grasses and was then highly modified in Anomochlooideae, or the spikelet originated in the common ancestor of Pharoideae and all remaining grasses (GPWG 2001; Preston et al. 2009; Kellogg 2015) (node 1, Fig. 1.2). In either case, Setaria is like all but about four species of grasses in having flowers borne in spikelets.

Other widespread aspects of grasses characterize Puelioideae plus the BOP + PACMAD clades (i.e., descendants of node 2, Fig. 1.2), and not Anomochlooideae or Pharoideae. For example, style branches and stigmas are reduced to two in this group (although the character reverses in some taxa), and the stigmas have two orders of branching (GPWG 2001). Spikelets each have multiple flowers (another character that reverses frequently). Anther walls have a middle layer that breaks down during development, and the inner walls of the endothecial cells become fibrous at maturity.

The female gametophyte in the grasses is fairly conventional in early development, with an egg, two synergids, a binucleate central cell, and antipodal cells. However, in all investigated species other than the early diverging genera Streptochaeta and Pharus (Sajo et al. 2007, 2008), the antipodals continue to divide (Anton and Cocucci 1984; Evans and Grossniklaus 2009; Shadowsky 1926). The function of these extra divisions is unknown.

The BOP + PACMAD clade (descendants of node 3, Fig. 1.2) has no obvious morphological synapomorphy. The Grass Phylogeny Working Group (GPWG 2001) suggested that lack of a pseudopetiole in the leaves, reduction of lodicule number to 2 and stamen number to 3 might be synapomorphic. Although these characters reverse in a number of lineages, they all characterize Setaria.

Members of the PACMAD clade (node 4, Fig. 1.2) have nothing obvious in common. They are thought to share an elongated mesocotyl internode in the embryo, but relatively few species have actually been investigated for this character and it is unclear how reliable or consistent it is (GPWG 2001). The clade also includes all 24 origins of the C4 photosynthetic pathway in grasses (GPWG II 2012).

1.2 Characteristics of Setaria Shared with Subfamily Panicoideae, Tribe Paniceae, and Subtribe Cenchrinae

1.2.1 Panicoideae

Nearly 1/3 of the species of Poaceae are in subfamily Panicoideae . Panicoideae s.s. (node 6, Fig. 1.2) was one of the earliest subfamilies to be recognized as distinct. In 1810, Robert Brown noted that the group (which he called Paniceae) mostly has spikelets with exactly two flowers, with the upper one bisexual and the lower one staminate or sterile (Brown 1810, 1814). Recent phylogenetic work has shown that Panicoideae s.s. is part of a larger clade which now bears the name Panicoideae (node 5, Fig. 1.2), in which the spikelet morphology is more variable (Sánchez-Ken and Clark 2007, 2010).

Spikelets in Panicoideae s.s. are dorsiventrally compressed. The glumes and lemmas are generally not folded and are borne ab- and adaxially in relation to the spikelet-bearing axis. This pattern of compression contrasts with that of most other grasses such as rice, tef, and Brachypodium, in which the glumes and lemma are both folded along the midrib, a pattern known as lateral compression. In these taxa, the glumes and lemmas initiate at right angles to the spikelet-bearing axis. As with many such morphological characters, the significance of this highly consistent difference is unknown.

Spikelet development in the panicoids is basipetal, with the distal flower maturing before the proximal one (Bess et al. 2005; Doust and Kellogg 2002; Malcomber and Kellogg 2004). This pattern is similar to that in rice, but distinct from what is found in Pooideae, Chloridoideae, and other major groups.

Silica bodies in the leaf epidermis of panicoid grasses are generally bilobed in surface view and symmetrical in cross section (Piperno 2006; Piperno and Pearsall 1998). Nothing is known about deposition of silica in the grass epidermis and the mechanism by which silica body shape is defined.

Early phylogenetic work in subfamily Panicoideae found that the phylogeny reflected chromosome numbers rather than photosynthetic pathway as had been thought previously (Gómez-Martinez and Culham 2000; Aliscioni et al. 2003; Giussani et al. 2001). The ancestral base chromosome number of the subfamily is unknown but most likely to be 11 or 12, and the number was then reduced in the common ancestor of Panicoideae s.s. One descendant of this ancestor acquired a base number of x = 9, a number that now characterizes the tribe Paniceae, whereas the other descendant acquired a number of x = 10, which is shared by the tribes Andropogoneae and Paspaleae.

1.2.2 Paniceae

Within Paniceae, major clades are strongly supported by both nuclear, chloroplast, and mitochondrial sequences (Vicentini et al. 2008; GPWG II 2012; Washburn et al. 2015). All analyses to date have identified clades corresponding to subtribes Cenchrinae, Melinidinae, Panicinae, Boivinellinae, Neurachninae, and Anthephorinae (Fig. 1.3). The genus Dichanthelium is monophyletic and could be placed in its own subtribe (Dichantheliinae (Soreng et al. 2015)). In addition, there is an unnamed clade made up of the genera Sacciolepis, Trichanthecium, and Kellochoa plus a number of species formerly placed in Panicum (node 4, Fig. 1.3) (Morrone et al. 2012; GPWG II 2012; Zuloaga et al. 2011; Nicola et al. 2015).

Fig. 1.3
figure 3

Phylogeny of Paniceae based on chloroplast, mitochondrial, and nuclear rDNA gene sequences (Washburn et al. 2015; Nicola et al. 2015). The clade made up of Cenchrinae, Melinidinae, and Panicinae is found in all molecular phylogenies. Other relationships, particularly those surrounding node 1, are contradicted by other gene trees

Full size image

Cenchrinae, Melinidinae, and Panicinae form a robust group (the MPC clade) (Morrone et al. 2012; GPWG II 2012) (node 2, Fig. 1.3), with Cenchrinae and Melinidinae sisters (node 3, Fig. 1.3 (Washburn et al. 2015)). The clade was first identified as the “C4 three subtypes” clade by Giussani et al. (2001) because all members are C4, but each subtribe exhibits a different subtype of the C4 pathway. Cenchrinae includes species that are NADP-ME subtype, Melinidinae members are PCK, and Panicinae are NAD-ME. Each C4 subtype has characteristic leaf anatomy (Hattersley 1987; Hattersley and Watson 1992; Prendergast and Hattersley 1987; Prendergast et al. 1987). In Panicinae and Melinidinae, each vein is surrounded by an inner sheath of thick walled cells, the mestome sheath, and an outer sheath of parenchymatous cells. Carbon reduction occurs in the latter. In Cenchrinae, and thus in Setaria, as in most NADP-ME grasses, veins are surrounded by a single sheath, a derived condition in the subtribe.

Relationships among the other clades of Paniceae are unclear. Chloroplast data place Dichanthelium as sister to the MPC clade, whereas nuclear data place it sister to all other Paniceae (Vicentini et al. 2008) and combined chloroplast, mitochondrial, and nuclear RNA data place it sister to Neurachninae (Washburn et al. 2015). Conversely, Anthephorinae is placed sister to all Paniceae by chloroplast data (node 1, Fig. 1.3), but sister to the MPC clade by nuclear genes.

1.2.3 Cenchrinae

The Cenchrinae is also known as the “bristle clade ,” because almost all members of the clade form sterile branches (“bristles”) in the inflorescence. These sterile branches originate as ordinary branches, but instead of forming spikelets, they grow out and terminate blindly (Doust and Kellogg 2002). The bristles may form a meristem at their apex, but this often simply aborts, leaving a small collapsed set of cells (Fig. 1.1b). Bristles may be restricted to the ends of branches, or a bristle may be paired with each spikelet, or individual spikelets may be surrounded by an involucre of bristles. In the latter case, the bristles may be terete or flattened.

Two species, Zuloagaea bulbosa and “Panicum” antidotale, lack bristles. (The latter species is unrelated to true Panicum, which is in Panicinae, but has not yet been transferred to another genus.) Developmental studies in Zuloagaea show that early development in that species is strikingly similar to that of Panicum miliaceum (Panicum s.s.) and there is no evidence of bristle formation at any point in development (Bess et al. 2005, 2006).

The phylogeny of the group is poorly resolved largely because no one has yet investigated it using a sufficient number of markers. Nonetheless, a few strong clades can be identified. The Cenchrus clade (Cenchrus sensu lato, Fig. 1.4) includes both Cenchrus and the former genus Pennisetum, plus the monotypic Odontelytrum (Chemisquy et al. 2010; Donadio et al. 2009; Morrone et al. 2012). All species of Cenchrus s.l. form an abscission zone at the base of the primary branch such that the spikelets fall from the plant surrounded by an involucre of bristles. Developmentally, the species are also distinct because the primary branch enlarges isodiametrically, rather than growing primarily in a proximo-distal direction (Doust & Kellogg 2002). While many species of Cenchrus s.s. are easily identified by their flattened bristles forming an involucre around the spikelets, others intergrade morphologically with the former Pennisetum. Thus, the boundary between the genera is not sharp, consistent with the pattern found in phylogenetic studies (Donadio et al. 2009; Chemisquy et al. 2010).

Fig. 1.4
figure 4

Phylogeny of Cenchrinae based on the chloroplast gene ndhF, focusing on species in Setaria. Redrawn from Kellogg et al. (2009). Arrows show the approximate placement of additional taxa included in the study of GPWG II (2012). All branches shown have either parsimony or maximum likelihood bootstrap values >80, or Bayesian posterior probability >0.95, or both. Brown box indicates taxa native to Africa or tropical Asia, yellow is Australia and Australasia, blue is temperate Asia and green is the Americas, mostly South and Central America

Full size image

Primary branches of the inflorescence are spirally arranged in most species, whereas the secondaries and higher order branches are distichous (Bess et al. 2005; Doust and Kellogg 2002; Kellogg et al. 2004, 2013).

2 Relationships Within the Genus Setaria

2.1 Phylogeny and Characteristics of the Genus

Species of Setaria are described in detail in two monographs, which together cover 99 species (Morrone et al. 2014; Pensiero 1999), although Clayton et al. (2006 onward) lists 103 names. The genus has no unique character and as currently defined is likely to be para- or polyphyletic. The current phylogeny shows a number of well-supported clades corresponding largely to geography, but relationships among them are unresolved, making generic circumscription impossible at the moment (Fig. 1.4). As currently circumscribed, Setaria includes the members of Cenchrinae that do not fall in the Cenchrus clade, have bisexual spikelets (i.e., are not Spinifex or Zygochloa), and lack the distinctive characters of the various oligotypic genera such as Dissochondrus, Paractaenum, or Plagiosetum. Almost certainly, Setaria will need to be expanded to include some of these elements, but without a solid phylogeny it is hard to find a good rationale for doing so.

Species of Setaria occur in warm regions throughout the world, and in diverse habitats (Morrone et al. 2014). Some species, such as S. sulcata and S. palmifolia, occur in disturbed areas in moist forest shade. Others, such as S. nigrirostris and S. sphacelata, are found in damp grasslands, and still others, such as S. rara and S. reflexa, in dry open habitats. A handful of species, notably S. viridis and S. pumila, are weedy and have followed human activity to spread far beyond their original distribution.

The number and position of bristles in the inflorescence varies considerably among Setaria species (Morrone et al. 2014). In species such as Setaria palmifolia, each spikelet is accompanied by a single bristle. In other species, such as S. parviflora and S. viridis, each mature spikelet is surrounded by multiple bristles. The relationship between the number of spikelets and the number of bristles is developmentally complex however (Doust and Kellogg 2002). Bristle and spikelet identity are specified early in inflorescence development. In some cases, all spikelets develop to maturity so that the number of bristles per spikelet reflects meristem identity decisions. In other species, however, late forming spikelets fail to develop so that high numbers of bristles per spikelet reflect a process of spikelet abortion rather than branch identity specification.

In some species of Setaria the primary branches of the inflorescence are themselves unbranched (i.e., the spikelets are borne directly on the primary branches) and the branches end in a sharp bristle-like tip (Morrone et al. 2014). The inflorescence thus looks superficially similar to that in Paspalum, but the presence of the terminal bristle is diagnostic; species with this inflorescence morphology have often been placed in a genus Paspalidium. However, species occur in which some spikelets are associated with bristles in addition to the one at the branch tip and thus the morphology intergrades with that of Setaria sensu stricto. Recognizing this morphological intermediacy, all Paspalidium species have been transferred to Setaria (Veldkamp 1994; Webster 1993, 1995), and the transfer has been supported by phylogenetic data (Morrone et al. 2012; Kellogg et al. 2009; GPWG II 2012).

The species of Setaria vary widely in inflorescence architecture and leaf form (Morrone et al. 2014). Inflorescences may be narrow with short stiff lateral branches (the inflorescence thus shaped like a bottle brush, e.g., S. pumila, S. sphacelata, S. nigrirostris), broad and lax with spreading branches (shaped like a Christmas tree, e.g., S. grandis, S. sulcata, S. lindenbergiana), or sparse with few spreading primary branches (like an antenna, e.g., S. jubiflora, S. flavida). Each primary branch may produce spikelets directly (e.g., S. jubiflora, S. flavida, S. rara) or may rebranch up to six times (e.g., S. parviflora, S. pumila). Plants may be annual (e.g., S. faberi, S. acromelaena, S. sagittifolia, S. viridis) or perennial (most species), and may be a few cm (e.g., some specimens of S. clementii, S. ustilata) to over 1 m (e.g., S. grandis) tall. Spikelets are generally ovate but sometimes may be elongate or orbicular. Leaves are generally flat, but some species (e.g., S. sulcata, S. palmifolia) have striking folded leaves. The latter were once placed in their own section because the leaf morphology is so distinctive, but they do not form a clade in molecular phylogenies. Sagittate leaves are found in S. sagittifolia and S. appendiculata.

As in all groups of grasses, polyploids are common (Table 1.1). Except for polyploids involving Setaria viridis (see next section) the history of few of these has been disentangled, although it would be straightforward to do so using low-copy nuclear genes. Sequences of the nuclear gene Knotted1 have shown that S. flavida and S. jubiflora are the products of a single polyploidization event (Doust et al. 2007). One of the genomes that produced the tetraploid ancestor of the two species also appears to be shared with S. grisebachii (diploid) and also with Stenotaphrum secundatum (a presumed diploid), Ixophorus unisetus (tetraploid), and Zuloagaea bulbosa (tetraploid). The origin of the other genome is less clear.

Table 1.1 Published chromosome numbers for species of Setaria
Full size table

A handful of published chromosome counts appear to have a base number other than x = 9 (Table 1.1). One accession of S. homonyma is reported to have n = 10 (Singh and Gupta 1977), one accession of S. sphacelata is apparently n = 18 + 1B (Dujardin 1978); chromosome spreads are illustrated in both papers, and the counts appear accurate, although Singh and Gupta acknowledge that the S. homonyma count might be 9 + 1B. An accession of S. sulcata may have n = 16 and 2n = 32 (Oliveira Freitas-Sacchet et al. 1984; Oliveira Freitas-Sacchet 1980), but the count is not documented photographically.

2.2 The Temperate Asian Clade

The annual species Setaria italica, S. viridis, S. faberi, and S. verticillata form a clade in chloroplast phylogenies (Kellogg et al. 2009; Layton and Kellogg 2014) and in phylogenies using the nuclear genes encoding knotted1 and 5S rDNA (Zhao et al. 2013; Layton and Kellogg 2014; Doust et al. 2007) (Fig. 1.4). S. viridis is the wild ancestor of the cultivated species S. italica, as documented by data from many sources, and the two remain interfertile (Le Thierry d’Ennequin et al. 2000; Hunt et al. 2008; Hirano et al. 2011; Darmency et al. 1987; Shi et al. 2008; Huang et al. 2014). This relationship is discussed more extensively in the chapters by Jia (Chap. 2), Huang and Feldman (Chap. 3), and Diao and Jia (Chap. 4) in this book.

The diploid genome of S. italica was designated as the A genome by Li et al. (1945); diploid S. viridis shares the A genome with S. italica, as verified by hybrid fertility, and cytogenomic, enzymatic, and molecular markers (Li et al. 1945; Wang et al. 1998; Benabdelmouna et al. 2001a, b; Darmency and Pernes 1987). The diploid genome of S. adhaerens is distinct from that of S. viridis and S. italica and has been designated as the B genome by Benabdelmouna et al. (2001b); this designation has been confirmed by Wang et al. (2009) and Zhao et al. (2013). In addition, sequence data show that S. adhaerens and S. viridis are not closely related (Fig. 1.4) (Layton and Kellogg 2014). The diploid genome of S. grisebachii from America was identified as genome C due to its poor hybridization signals with both the A genome of S. viridis and the B genome of S. adhaerens (Wang et al. 2009).

A combination of molecular phylogenetics and cytogenetics has identified the progenitors of several polyploid taxa. Chromosomes of the tetraploid species S. pumila (often erroneously known as S. glauca (Morrone et al. 2014)) and S. parviflora strongly cross-hybridized but no hybridization signal was detected when the chromosomes of these two species were hybridized with probes derived from the known A, B, and C genomes; thus S. pumila and S. parviflora were designated as having genome D (Zhao et al. 2013). Similarly, the lack of hybridization signals with A, B, C, and D donor genomes led to the recognition of the E genome from S. palmifolia and the F genome from S. arenaria (Zhao et al. 2013). (Note that S. arenaria is a dubious name according to Morrone et al. (2014) and thus it is unclear what material was used by Zhao et al. (2013).) GISH also identified an apparent A genome autotetraploid, S. apiculata (= queenslandica). Analysis of kn1 and 5S rDNA sequences was consistent with the GISH results (Zhao et al. 2013).

The tetraploid S. faberi formed from an A genome (S. viridis) plus another genome from an unknown source closely related to S. viridis (Layton and Kellogg 2014). S. faberi is morphologically similar to S. viridis, but in the former species the upper glume is slightly shorter, so that the upper 1/4–1/3 of the upper lemma is visible (Layton and Kellogg 2014). In contrast, the upper glume of S. viridis completely covers the upper lemma and often slightly overlaps the lower lemma at the apex of the spikelet. S. faberi also has macrohairs on the adaxial surface of the leaves, whereas the leaves of S. viridis are glabrous. In addition, S. faberi is less tolerant of drought than S. viridis is and often grows in more mesic habitats such as the margins of cultivated fields, whereas S. viridis occurs more frequently in poor soil and cracks in pavement (Layton and Kellogg 2014).

Tetraploids classified as S. verticillata and S. verticilliformis each have one genome from the diploid S. adhaerens and one from S. viridis (Benabdelmouna et al. 2001b). Phylogenetic data using the low-copy nuclear marker knotted1 on the same plant accessions confirmed the cytogenetic results, showing that S. verticillata and S. verticilliformis each have two loci, consistent with their ploidy level (Layton and Kellogg 2014). One kn1 locus is related to that of the diploid S. adhaerens and the other to that of the diploid S. viridis.

The phylogenetic and cytogenetic data settle confusion over the taxonomy of S. verticillata, S. adhaerens, and S. verticilliformis. Some authors have considered the three species as synonymous (Rominger 1962; Doust et al. 2007; Kellogg et al. 2009), others have distinguished S. verticillata and S. verticilliformis but synonymized S. adhaerens with S. verticillata (Morrone et al. 2014), and still others maintain all three species as distinct (Rominger 2003). Both S. verticillata s.s. and S. adhaerens have retrorse prickles on the bristles, a character that is distinctive within the genus and easily observed in the field because it makes the inflorescence adhere to clothing. Because this morphological character appears in the diploid and its derived polyploid, we can infer that it must be controlled by the B genome. S. adhaerens is distinctive in being a small plant with glabrous sheath margins and strigose hairs with papillose bases on the abaxial leaf surfaces. S. verticillata, in contrast, has sheath margins that are ciliate distally and abaxial leaf surfaces that are scabrous (Layton and Kellogg 2014). Some diploid chromosome numbers are reported for S. verticillata (Table 1.1) (Rominger 2003), but these are likely to be from misidentified specimens of S. adhaerens (Layton and Kellogg 2014).

S. verticilliformis, in contrast, has antrorse prickles but otherwise is morphologically indistinguishable from S. verticillata. Because of the genomic and morphological similarity of the two species, Layton and Kellogg (2014) recommend that the former species be placed in synonymy with the latter. They also recommend that S. adhaerens be recognized as a separate entity.

3 Summary

All organisms contain evidence of their evolutionary history and Setaria is no exception. Like all commelinid monocots, species of Setaria have extensive endosperm in their seeds, cell walls with glucuronoarabinoxylans, lignin with p-coumaric and ferulic acid, and leaves that accumulate silica. Like all Poales, the leaf silica accumulates in the epidermis, and like members of the graminid clade, the pollen is monoporate with a raised ring around the pore. Setaria has dumbbell shaped stomatal guard cells, and alternating long and short cells in the epidermis.

Like all other grasses, Setaria has a highly differentiated embryo, including a large haustorial derivative of the cotyledon, the scutellum. The fruit is a caryopsis in which the seed coat is fused to the inner epidermis of the pericarp. The antipodals proliferate in the female gametophyte. Short cells of the leaf epidermis produce bicellular microhairs. The genome has undergone a whole genome duplication dating to the origin of the family in the late Cretaceous. Like most but not quite all grasses, Setaria has flowers in spikelets with more than one flower per spikelet and two style branches and stigmas.

Like most other panicoid grasses (i.e., those in tribes Paniceae, Paspaleae, and Andropogoneae), Setaria has spikelets with exactly two flowers, the upper one bisexual and the lower one staminate or sterile. Spikelets are dorsiventrally compressed and develop basipetally. Leaf silica bodies are bilobed in surface view and symmetrical in cross section. Along with other members of tribe Paniceae, Setaria has a chromosome base number of x = 9. Members of subtribe Cenchrinae share C4 photosynthesis of the NADP-ME subtype and a single sheath surrounding the vascular bundle. The inflorescence bears sterile branches or bristles.

The genus Setaria itself has no diagnostic characters and is probably para- or polyphyletic as currently defined. Phylogenetic data are insufficient and not conclusive. Major clades correspond to geography and do not correlate well with morphology. The Asian clade includes the model species S. viridis, along with its domesticated derivative S. italica, and its polyploid descendants S. faberi, S. verticillata, and S. verticilliformis.