Poaceae subfam. Arundinoideae
Plants usually perennial; cespitose or not, sometimes rhizomatous, sometimes stoloniferous. Culms 15-1000 cm, annual, herbaceous to somewhat woody, internodes usually hollow. Leaves usually mostly cauline, often conspicuously distichous; sheaths usually open; auricles usually absent; abaxial ligules usually absent (of hairs in Hakonechloa); adaxial ligules membranous or of hairs, if membranous, often ciliate; blades without pseudopetioles, sometimes deciduous at maturity; mesophyll usually non-radiate (radiate in Amnio); adaxial palisade layer absent; fusoid cells absent; arm cells usually absent (present in Phragmites); Kranz anatomy absent; midribs simple; adaxial bulliform cells present; stomatal subsidiary cells low dome-shaped or triangular; bicellular microhairs usually present, usually with long, narrow terminal cells; papillae usually absent. Inflorescences usually terminal, ebracteate, usually paniculate, occasionally spicate or racemose. Spikelets laterally compressed, with 1-several bisexual florets or all florets unisexual and the species dioecious; florets 1-several, terete or laterally compressed, distal florets often reduced; disarticulation above the glumes. Glumes 2, from shorter than the adjacent lemmas to exceeding the distal florets; lemmas (3)5-7-veined, lanceolate to elliptic, acute to acuminate, sometimes awned; awns 1 or 3, if 3 not fused into a single basal column; paleas subequal to the lemmas; lodicules 2, usually free, occasionally joined at the base, fleshy, usually glabrous, not, scarcely, or heavily vascularized; anthers (1)2-3; ovaries glabrous; styles 2, usually free, bases close together. Caryopses usually punctate (long-linear in Molinia); endosperm hard, without lipid; starch grains compound; haustorial synergids absent; embryos usually large compared to the caryopses, waisted or not; epiblasts absent; scutellar cleft present; mesocotyl internode elongate; embryonic leaf margins usually meeting (overlapping in Hakonechloa). x = 6,9, 10, 12.
Discussion
The Arundinoideae are interpreted here as including only one tribe, the Arundineae. The tribe used to be interpreted more broadly (e.g., Watson et al. 1985; Clayton and Renvoize 1986; Kellogg and Campbell 1987), but the broader interpretation was generally acknowledged to be somewhat artificial. Hsiao et al. (1998) showed support for inclusion of the Danthonieae, Aristideae, and Amndineae in a more broadly interpreted Arundinoideae, but other studies (e.g., Hilu et al. 1990; Barker et al. 1995, 1998; Grass Phylogeny Working Group 2001) have failed to support such a treatment.