Systematics of Trigonochloa (Poaceae, Chloridoideae, Chlorideae)

Abstract A systematic treatment including descriptions and a key for identification is provided for the two species of Trigonochloa, a genus recently segregated from the polyphyletic Leptochloa s.l. Trigonochloa ranges from southern Africa east to India and Sri Lanka, reflecting the widely ranging Trigonochloa uniflora. Trigonochloa rupestris has a more limited distribution from East Africa to Yemen. Trigonochloa is diagnosable from other chloridoid grasses based on its unusually flaccid and membranous leaves that have uniquely enlarged lateral cells in the parenchyma sheath surrounding the vascular bundles in Trigonochloa uniflora (unconfirmed for Trigonochloa rupestris given limited material), primary and secondary vascular bundles that do not project above or below in fresh material, XyMS+ leaf anatomy, narrow spicate primary inflorescence branches, spikelets with one (or rarely two) florets, thinly membranous to hyaline lemmas, and a trigonous caryopses that bear a narrow but deep sulcus on the hilar side. Lectotypes are designated for Agrostis montana and Cynodon gracilis. The synonym Leptochloa laurentii De Wild. is confirmed for Trigonochloa uniflora.


Introduction
The generic boundaries of Leptochloa P. Beauv. have been contentious more or less continuously since the genus was first described in 1812 (Valls 1978;McNeill 1979;Phillips 1982;Snow 1997). Many previous discussions of generic boundaries focused on whether Diplachne P. Beauv., often included in Leptochloa, could be recognized as a distinct genus (Phillips 1974;Jacobs 1987). A global monograph of Leptochloa (Snow 1997) based on morphology, lemmatal micromorphology (Snow 1996), anatomy of stems and leaves, and caryopsis morphology (Snow 1998) tested the monophyly of the genus in the context of a few putatively related genera. Preliminary results from cladistic studies using morphology were far from conclusive and at best suggested that Diplachne was not easily segregated from Leptochloa (Snow 1997). One noteworthy aspect of the cladistic studies (Snow 1997) was the consistent sister status of the L. uniflora Hochst. ex A. Rich. and L. rupestris C.E. Hubb. to other members of Leptochloa s.l., which was predictable since these two species share many morphological and anatomical characteristics absent in other members of Leptochloa s.l.
Leptochloa s.l. has been considered a diverse assemblage of C 4 (nicotinamide adenine dinucleotide co-factor malic enzyme [NAD-ME] and phosphoenolpyruvate carboxykinase [PCK]) grasses in the tribe Chlorideae (Clayton and Renvoize 1986;Watson and Dallwitz 1992;Soreng et al. 2012) with approximately 32 annual or perennial species (Snow 1997(Snow , 2003. The range of morphological variation within many species of Leptochloa is significant (Snow 1997. However, Leptochloa s.l. is demonstrably polyphyletic when tested with molecular DNA markers (Columbus et al. 2007;Peterson et al. 2010Peterson et al. , 2012. In a large phylogenetic study of the Chloridoideae based on seven DNA sequence markers the species of Leptochloa s.l. were found to form three separate lineages (Peterson et al. 2010), and more recently using six DNA markers these species were found to form five separate lineages, each treated as a separate genus: Dinebra Jacq., Diplachne, Disakisperma Steud., Leptochloa sensu strictu, and Trigonochloa P.M. Peterson & N. Snow (Peterson et al. 2012).  found that Trigonochloa uniflora (Hochst. ex A. Rich.) P.M. Peterson & N. Snow and T. rupestris (C.E. Hubb.) P.M. Peterson & N. Snow consistently resolved as a strongly supported clade ) outside of subtribe Elusininae.
Leptochloa uniflora was first described by Richard (1851). Chippindale (1955) later transferred this species to Craspedorhachis Benth., although her concept of Craspedorhachis was broader than that of other treatments (Clayton and Renvoize 1986;Gibbs-Russell et al. 1991;Watson and Dallwitz 1992). One current concept of Craspedorhachis includes three species that collectively occupy parts of sub-Saharan Africa and Madagascar (e.g., Simon et al. 2011).
Bentham (1882: 108) described two sections in Leptochloa, which he maintained distinct from Diplachne. In L. sect. Pseudocynodon Benth., characterized by spikelets having only one or two florets, Bentham ascribed L. uniflora and L. neesii (the latter of which included his concept of L. polystachya Benth.). The latter species, which is also characterized by one floret per spikelet (or in rare cases two), differs significantly from Trigonochloa by characters of leaf anatomy and has been transferred to Dinebra .
The purpose of this paper is to present the systematics of Trigonochloa as the first step in revising Leptochloa s.l. into monophyletic genera.
Fresh leaf samples of Trigonochloa uniflora and Leptochloa s.l. were studied for, but not summarized, in Snow (1997). The first author also viewed black and white anatomical photographs of T. uniflora taken by R. Ellis housed at PRE (Davidse & Ellis 5925, Ellis s.n., Ellis 3635, but see in particular Ellis 4534) and was able to confirm observations of T. uniflora made previously by Metcalfe (1960). Stems (culms) were hand-sectioned with fresh or rehydrated material.  (Snow 1996), and caryopses were studied using simple light microscopy (Snow 1998).
Leaf anatomy. The transverse anatomical features of the leaves of these two species differ in several significant ways from the rest of Leptochloa s.l. The leaf blades of both species are quite thin (and flaccid) when fresh in T. uniflora and somewhat translucent. They also can be relatively broad basally and relatively short, thus appearing narrowly ovate.
Epidermal preparations of T. uniflora made separately by the first author (Davidse & Ellis 5925, MO; unpublished) and by Roger Ellis (Ellis 1928; photos on herbarium specimen; PRE) show the adaxial (more so) and abaxial (less so) surfaces (apart from areas above bulliform cells) to be covered with narrow rows of relatively small cells, virtually all of which are capped by a centrally located and prominent papilla (Snow, unpubl.). In addition, the cells of the leaf blade epidermis and lemma surface are not always clearly differentiated into short and long cells (Snow 1996).
Keels (areas of parenchyma in the middle of the leaf blade lacking vascular bundles) are absent, or if present then small, and if present then lacunae within the parenchyma are absent. Primary and secondary vascular bundles differed only slightly in size and in fresh material do not project adaxially or abaxially. Bulliform cells were noted between adjacent vascular bundles. Colorless cells were not observed between vascular bundles, but they do occur adaxially to the primary and secondary bundles, and may be the only cell layer between the epidermis and the secondary vascular bundles. As many as five successive colorless cells (in cross section) were observed adaxially to a secondary vascular bundle on Ellis 4534. Metcalfe (1960: 285) reported only a single bundle sheath in Leptochloa uniflora and first noted the significantly enlarged parenchyma sheath cells at 3 and 9 o'clock, which he termed lateral cells, which we confirmed for this species (see in particular images with Ellis 4534 at PRE). The lateral cells often penetrate deeply into laterally adjacent mesophyll and are nearly completely filled by a large chloroplast. Enlarged lateral cells were not clearly evident in the single specimen of T. rupestris examined from limited rehyrdated material so we cannot yet confidently confirm nor reject their presence in this species. Hattersley and Watson (1976: 303) reported the species as being XyMS-, based on the implied lack of intervening cells between the metaxylem and PCR sheath given the results of Metcalfe (1960). Valls (1978: 73-74) reported the presence of a double sheath for L. uniflora and reported that walls of the inner sheath were "exceptionally thin-walled". Valls (1978), however, may have used specimens incorrectly identified that in actuality were Leptochloa neesii. Specimens studied by Roger Ellis, and seen for this study, confirm the observations of Metcalfe (1960) of the enlarged lateral cells and the presence of a double sheath, supporting the XyMS+ condition. However, the "outer sheath" cells only occur on the distal edges of the enlarged lateral cells, such that colorless cells typically occur adaxially and abaxially to the promixal part of the lateral cell (i.e., that part adjacent to the vascular bundle), with smaller chloroplast-bearing cells bearing smaller and more diffuse chloroplasts adjacent to the part of the lateral cell that occurs closest to the intervascular region. With permission while on site at PRE in 1996, the first author took 35 mm SLR photographs of the original black and white images of Ellis 4534, which show the leaf anatomy clearly. However, photos of the original images did not reproduce at a high enough quality to include in this paper. Other images of the leaf anatomy of Trigonochloa are unknown to the present authors.
Stem anatomy. Both species have a solid culm. Trigonochloa uniflora has inner and outer sclerenchymatous rings, although an inner ring was absent for T. rupestris.
Lemmatal micromorphology. The two species share a unique combination of lemmatal micromorphological characters compared to other species of Leptochloa s.l. (Snow 1996). Cork cells and bicellular microhairs were present; macrohairs were terete (not crispate) and obtuse apically (not clavicorniculate; Snow 1996). Lemmatal characters in many species of Leptochloa s.l. but lacking in Trigonochloa included silica cells, and papillate long and papillate short cells (Snow 1996).

Key to species of Trigonochloa
Phenology. Flowering June through January. Distribution. This species is found in Yemen and Eritrea south to Kenya in woodlands, hillsides, bushland and on damp rocks along streams; 900-1800 m. Conservation status. Since many parts of its range are presently inaccessible to botanists or still remain inadequately surveyed this species is data deficient. Additional observations and collections are highly desirable.
Etymology. The epithet rupestis is Latin for "of rocks", presumably in reference to observations of the habitat of the type collection.
Comments. This species closely resembles T. uniflora, with which it is morphologically similar. The best character to recognize T. rupestris from T. uniflora is leaf blade width. However, its sprawling, branching, and perennial growth form with narrow culms typically distinguish it from T. uniflora. The holotype and isotype are aberrant in their lack of ciliate sheath margins, but otherwise accord with the diagnostic characters. The observation of Phillips (1974) that T. rupestris has more widely divergent leaf blades than T. uniflora cannot be reliably applied to dried material. None of the specimens we have seen confirmed her observation (Phillips 1974) that T. rupestris is rhizomatous (i.e., culms within the soil), but the stems can be somewhat sprawling and stoloniferous.
Phenology. Flowering throughout the year when adequate moisture is available.
Distribtution. This species is scattered through the eastern and southern portions of sub-Saharan Africa, rarely in India, most common in Sri Lanka in forests and shady areas on hillsides, well-drained and often sandy soils in disturbed and riparian areas; 0-1200 m. Conservation status. Since the species is widespread it is of least concern (IUCN 2010) given its widespread occurrence. However, the typical size of populations is undocumented. The thinly membranous leaves likely are sought after by grazers, although the relative nutrition content of the leaves is unstudied.
Etymology. The specific epithet is from the Latin uniflora, with reference to the single floret per spikelet. Vernacular names. Common triangle-seed grass. Kenya: Mkuse -Digo (Magogo and Glover 477, W).
Comments. The species description and distribution differ from Snow (1997) because some specimens therein were incorrectly identified as Leptochloa neesii.
Trigonochloa uniflora individuals vary significantly in growth habit. Most specimens are relatively delicate, sprawling annuals, but some specimens have more erect, relatively stout culms that appear to be weakly perennial. The more erect forms typically occur in somewhat more open areas and have thicker leaves, whereas the more slender forms that frequently root at the lower nodes, typically occur in shade and have thinner leaves.
The glumes of T. uniflora typically are longer than the single floret and may be mucronate. The caryopsis is sometimes dispersed with a tightly adnate lemma and palea,