The most robust previously published phylogeny for the overall structure of the grass family (Poaceae) shows three early diverging lineages and two major derived clades, the BEP clade and the PACCAD clade (Grass Phylogeny Working Group 2001). A few key taxa were incompletely sampled, however, and support for the BEP clade was moderate at best and relationships among the major lineages within the PACCAD clade remained unresolved. In addition, recent studies indicated that the sister group to Poaceae may be Joinvilleaceae and/or Ecdeiocoleaceae, the latter of which were not previously sampled. In this study, missing structural data were determined and analyzed as well as sequence data for ndhF and rbcL, the two most complete plastid sequence data sets. Sampling was increased with a particular focus on key taxa such as Danthoniopsis, Eriachne, Micraira, and Streptogyna and a representative of the outgroup, Ecdeiocoleaceae. A total of 61 ingroup and two outgroup taxa were analyzed using maximum parsimony for total data, and maximum parsimony, Bayesian inference, and neighbor joining for the molecular data. A strongly supported clade of ((Eriachneae, Isachne) Micraira) was recovered as a sister subfamily to Arundinoideae and excluded from Panicoideae. Arundinaria was strongly united with Bambusoideae. The position of Streptogyna was weakly supported among Ehrhartoideae, and is still unresolved. An outgroup effect on ingroup topology was observed demonstrating that highly divergent outgroups may unpredictably alter ingroup relationships.

Phylogenetic relationships among subfamilies of the well supported PACCAD clade of Poaceae remain uncertain. Several genera such as Micraira and Eriachne were considered incertae sedis in the most recent subfamilial classification of the grasses, but these two genera formed a well-supported clade in an analysis based on chloroplast and structural data. Another genus, Isachne, traditionally classified in the Panicoideae, also formed part of this well-supported clade. Despite strong molecular support for the clade, thus far no morphological synapomorphy has been found. Nevertheless, the strongly supported monophyly of this clade allowed us to suggest the recognition of a new subfamily within the PACCAD clade. Since there was already a name available, in this paper we propose the reinstatement and emendation of the circumscription of Micrairoideae. The reinstatement of Micrairoideae changes the acronym PACCAD to PACCMAD for this large clade of grasses.

Subfamily Centothecoideae is formally submerged into the Panicoideae, and a new tribal classification for the expanded Panicoideae is proposed based explicitly on the phylogeny. This classification includes 12 tribes of which Chasmanthieae and Zeugiteae are segretated from the Centotheceae; Tristachyideae is segregated from Arundinelleae, and a new tribe, Cyperochloeae, is validated to accommodate two isolated genera. A key to the tribes is provided.

Bambuseae (woody bamboos), one of two tribes recognized within Bambusoideae (true bamboos), comprise over 90% of the diversity of the subfamily, yet monophyly of the tribe is generally only moderately supported, and phylogenetic relationships within the tribe are poorly understood. In addition, there appears to be some level of conflict between morphological and molecular data within the tribe. We conducted a parsimony analysis of 43 species of Bambuseae, three of Olyreae (herbaceous bamboos), and two outgroup taxa using morphological and plastid rpl16 intron sequence data to (1) further test the monophyly of Bambuseae, (2) test the monophyly of Chusqueinae and Hickelinae (the two one-flowered, determinate subtribes), and (3) examine the apparent conflict between molecular and morphological data sets in the determinate, one-flowered genera of Bambuseae. We recovered a monophyletic Bambusoideae, Bambuseae, Olyreae, and Chusqueinae, although support for Bambuseae remained moderate. Our results suggest that the morphological similarities between Chusqueinae and Hickelinae are homoplasious, but robust resolution of relationships among the major lineages of woody bamboos is still wanting.

Question Niche differentiation may promote species co‐existence because it precludes competitive exclusion. However, in stressful environments, facilitation may predominate, and large niche overlaps may result from niche widening and positive spatial associations. We analysed the niche structure of a species‐rich system over a soil depth gradient (a niche axis positively correlated with water availability) and tested for various patterns expected from competition or facilitation. Location Semi‐arid grassland, Oaxaca, Mexico. Methods We recorded species abundance and soil depth in 1200 0.1 × 0.1 m quadrats from four sites. Community structure was described through correspondence analysis and correlated with soil depth. We measured niche breadth (Levine's index), position (mean depth) and overlap (Hurlbert's index). We used a null model to estimate the expected overlap if interactions play no role in determining community structure. We tested whether overlap changed along the depth gradient. We also analysed if niches become narrow with higher competition. Results Data supported three predictions derived from competition: most species’ niches were relatively narrow and differed in their position along the gradient, a pattern reflected in a changing community structure; realized niches became narrower where competition was stronger; and niche overlap was smaller than expected by chance. Facilitation had no support: large niche overlaps expected from stress amelioration were as frequent as predicted by chance. Overlap increased in the most stressful (shallow) soils, as expected if facilitation predominates there, but also in the deepest ones where evidence for competition was strong, suggesting an artifact at the extremes of the gradient. Conclusions In this grassland, competition seems to restrict species distribution, abating richness in the deepest soils and relegating poor competitors (succulent and poikilohydric species) to the shallow soils. There, drought tolerance rather than facilitation determined community composition and richness. Our results support the idea that hydrological niches play an important role in maintaining plant diversity because (1) differentiation in hydrological requirements (small niche overlaps over the water‐use axis) reduces competition, and (2) spatial heterogeneity and stress create competition‐free patches. Our results also support the hypothesis that facilitation is rare when stress is caused by resource (in this case, water) scarcity.