Glume absence in the Orcuttieae (Gramineae: Chloridoideae) and a hypothisis of intratribal relationships

This study addresses glume absence in tribe Orcuttieae. In Orcuttia califomica, 0. inaequalis, and 0. viscida, all spikelets possess two glumes except for the terminal spikelet of the inflorescence, which lacks both glumes. In 0. pilosa and Tuctoria greenei the terminal spikelet lacks only the first (proximal) glume, whereas in 0. tenuis, T. fragilis, and T. mucronata both glumes are developed on all spikelets. This is the first report of glume absence in species of Orcuttieae other than Neostapfia colusana, which has been long reported to lack both glumes on all spikelets. A hypothesis of phylogenetic relationships in the tribe is presented and characters involved are discussed. We hypothesize Neostapfia to be sister to a Tuctoria/Orcuttia clade, and Tuctoria to be a grade leading to a monophyletic Orcuttia.

The grass tribe Orcuttieae are noted for their morphological and ecological specialization (Crampton 1959;Reeder 1965Reeder , 1982Griggs 1976;Keeley 1998a) and rarity (Reeder and Reeder 1980;Skinner and Pavlik 1994 ). This unusual group of annuals is distinct from all other grasses: plants are beset with glands that produce a viscid, aromatic exudate, and the leaves are eligulate, without distinction into sheath and blade, and bear small, sunken, mushroom-button-shaped bicellular microhairs (Reeder 1965). Although the tribe is confidently placed in subfamily Chloridoideae (Stebbins and Crampton 1961;Reeder 1965), its closest relatives have yet to be identified. The manynerved (7-17) lemmas led Stebbins and Crampton (1961) to place the species in tribe Pappophoreae, a relationship that is unlikely in light of other data (Reeder 1965). Distich/is Rafin. and Eragrostis N. M. Wolf have been suggested as potential close relatives (Columbus pers. comm. in Keeley 1998a), but discovery of the elusive sister group awaits molecular phylogenetic study.
While exarmmng plants of Orcuttia californica grown up from germination trials at Rancho Santa Ana Botanic Garden, we discovered that both glumes were lacking from the terminal spikelet in all inflorescences. All other spikelets, however, were observed to possess the normal pair of glumes. Herbarium specimens at RSA were then studied to ascertain the extent of this dimorphism. From this survey we report the following findings. In Orcuttia californica, 0. inaequalis, and 0. viscida, all spikelets possess two g1umes except for the terminal spikelet of the inflorescence, which lacks both glumes. In 0. pilosa and Tuctoria greenei the terminal spikelet lacks only the first (proximal) glume, whereas in 0. tenuis, T. fragilis, and T. mucronata both glumes are developed on all spikelets. This is the first report of glume absence in species of Orcuttieae other than Neostapfia colusana, which has been long reported to lack both glumes on all spikelets. Interestingly, Neostapfia can have glumes present in the spikelet, as discussed in the original description by Davy (1898). He notes, "The 2 or 3 uppermost spikelets are subtended by linear or lanceolate linear empty glumes." When specimens are critically studied, this is indeed the case, with the addition that glumes may also be present on the lowermost spikelets of the inflorescence. It is clear that these bracts are indeed glumes and not foliacious bracts of the inflorescence, as they are positioned on the lateral spikelet axis, not the main axis.
It was then discovered that glumes are absent in those species with more congested inflorescences. The terminal spikelet in the dense, capitate inflorescence of Orcuttia inaequalis lacks both glumes, whereas the widely spaced spikelets of 0. tenuis all have two Table I. Characters mapped on the phylogenetic tree. These characters are from Metcalfe (1960), Reeder (1965Reeder ( , 1982, and Keeley (1998a, b Gain of glands 4 Gain of malate deposition in leaf glands 5 Gain of small, sunken, mushroom-button-shaped bicellular microhairs 6 Loss of embryonic epiblast 7 Gain of long, ribbon-like floating juvenile leaves 8 Loss of stomata on submerged juvenile leaves 9 Gain Of lacunae in submerged juvenile leaves 10 Loss of kranz anatomy in submerged juvenile leaves II Increase in number of cells (5-7 vs. 1-3) between vascular bundles in submerged juvenile leaves 12 Gain of NADP-malic enzyme catalyzed decarboxylation glumes. These species represent the extremes in inflorescence form. Griggs (1976) considered the inflorescence morphology of seven Orcuttia and Tuctoria species in context of pollination. His quantification of inflorescence density, as a percentage of the inflorescence axis bearing spikelets, serves to illustrate the relationship between inflorescence density and presence/absence of glumes in the terminal spikelet: 0. inaequalis, 22% (0 glumes); 0. viscida, 42% (0); 0. californica, 48% (0); 0. pilosa, 55% (1); T. greenei, 64% (1); T. mucronata, 70% (2); and 0. tenuis, 81% (2). Tuctoria fragilis, not included in Griggs (1976) study also has two glumes on the terminal spikelet. Employing characters extracted from Metcalfe (1960), Reeder (1965Reeder ( , 1982, and Keeley (1998a, b) (Table 1), we estimated the phylogeny of Orcuttieae intuitively (Fig. 1). It should be noted that only Neostapfia, 0. californica, 0. greenei, and T_ greenei have been examined for characters 4, 8, 9, 11, and 12 (all from Keeley 1998a), but an assumption was made that all congeners are the same with respect to these characters. Although the sister group of the Orcuttieae has yet to be identified, characters were confidently polarized based on the rare suite of characters states found in the tribe with respect to the other members of Chloridoideae. These data suggest that Neostapfia represents the sister lineage to the Tuctoria/Orcuttia lineage and Orcuttia, by virtue of its numerous synapomorphies, is monophyletic. The conclusion regarding the position of Neostapfia was the same reached by- Keeley (1998a), who employed the genera as terminal taxa in a cladistic analysis. While we have only coded one character (12) that separates Neostapfia from the rest of the Orcuttieae, there is other evidence for this position, though less convincing. Tuctoria is known to have intermediate states between Neostapfia and Or-cuttia. The most notable are the number of juvenile submerged leaves, the RUBISCO:PEP carboxylase ratio, and germination time. These characters help to intuitively place Tuctoria between Neostapfia and Orcuttia. Unlike Orcuttia, monophylesis of Tuctoria is doubtful. Evidence that Tuctoria may represent a grade (i.e., paraphyletic genus) instead of a monophyletic group primarily involves chromosome number. Reeder (1982) reported a diploid chromosome number of 40 for Neostapfia, T. fragilis, and T. mucronata, whereas T. greenei and all Orcuttia species have fewer chromosomes, ranging from 24 to 30 (Fig. 1 ). There are no known synapomorphies for Tuctoria.
Characteristics of the lodicules in the Orcuttieae have become confused in recent literature. One of the characters used in the recent phylogenetic analysis of the Orcuttieae (Keeley 1998a) is lodicule reduction. Keeley characterizes the lodicules of Tuctoria as being reduced relative to Neostapfia. In the revision of the tribe, Reeder (1982) suggests that lodicule size varies among the Tuctoria species, and there is no clear reduction in size from Neostapfia to Tuctoria. In addition, the character "lodicules fused" used by Keeley (1998a) for Tuctoria is problematic. This is referring to fusion of the lodicules to the palea (not each other), and is also suggested to be variable for the genus (Reeder 1982). Tuctoria mucronata has lodicules fused to the palea, but in T. fragilis the lodicules are only "slightly" fused to the palea, and fusion of lodicules and palea is not addressed for T. greenei in Reeder's revision (1982), implying no fusion at all. Given these contradicting reports, more detailed study of lodicule variation is necessary.
In light of the foregoing phylogenetic hypotheses, the loss or gain of glumes appears to have occurred several times independently. Given that spikelets of all but a few other members of Chloridoideae possess both glumes, it is more likely than not that the common ancestor of the Orcuttieae bore glumes, and that glume development was subsequently arrested to varying degrees in several lineages. Although the tribe is predisposed to glume suppression, the complete absence of glumes in Neostapfia is a situation considerably different from that in the five other species lacking glumes, wherein only the terminal spikelet of the inflorescence is involved.
As can be seen from Fig. 1, the topology of the cladogram was not resolved solely from the characters listed in Table 1, but by considering inflorescence congestion (Griggs 1976), glume absence, and chromosome numbers (Reeder 1982) as well. We placed Tuctoria greenei as the sister species of Orcuttia because of its nontetraploid chromosome number. Arrangement of the Orcuttia species, however, was based on glume absence and inflorescence congestion, which do not correlate with chromosome number. 12 loss of glome I 4,6,7,8,9,10,11 Reeder (1982), and percentage of axis bearing spikelets values are from Griggs (1976). This hypothesis of interspecific relationships serves as the starting point for future phylogenetic work in the Orcuttieae. The study by Keeley (1998a), which provided a number of useful anatomicaVphysiological characters, should be expanded to include all species. Also, our understanding of the evolution of the tribe would likely be improved by employing molecular approaches.