Systematic Studies On The Tribe Agrostideae And Related Genera From China

The grass family includes approximately 10000 species classified into 600-700 genera. There are about 200 genera and 1500 species distributed in China. The grass family is of particular interest to humans. However, subfamily and tribal rank has been in dispute.The Agrostideae was first erected by Kunth (1815) to include genera that have a paniculate inflorescence with one-flowered spikelets. Agrostideae Kunth included Muhlenberginae, Agrostidinae, Stipinae. Subsequent studies based on microcharacters showed that the group was rather artificial, and consequently its genera were referred to the tribes Poeae, Aveneae, Brachyelytreae, Stipeae, Eragrosteae and Aristideae. Agrostideae was again circumscribed. Agrostideae retained in Flora Reipublicae Popularis Sinicae in a narrower sense, belonging to Pooideae. Currently, Agrostideae was merged with a broad Aveneae by many botanists. Thus, it is clear that the systematics of Agrostideae would benefit from a further study. In this paper, a holistic approach of the leaf abaxial epidermis, pollen grains, caryopsis morphology, plus ITS and trnL-F sequences were utilized to infer the relationship among Agrostideae and to assess the phylogenetic position of some genera of Agrostideae. The results are summarized as follows:Leaf morphology: Leaf abaxial morphology of 73 taxa allows the recognition of four types and four sub-types according to shape, rows and walls of intercostal long-cells, presence and absence of short-cells, micro-hairs or papillae between costal zone, presence or absence of prickles in costal zone. The results indicated that leaf morphology of Agrostideae, Phalarideae, Aveneae and Poeae were highly consistent, which belong to the same sub-type. The view that Agrostideae, Phalarideae, Aveneae were merged with Poeae was proposed. In addition, on the basis of leaf morphology, Brachyelytrum was promoted as Brachyelytreae. The genus Beckmannia belong to typeⅠ, while, the genera of Chlorideae belong to typeⅣ. The removal of Beckmannia from Chlorideae was also well supported. Meanwhile, Beckmannia was identical in leaf abaxial morphology with the genera Phleum, Alopecurus. Based on leaf morphology, Cinna was unnecessary to treat as Cinnae. Stephanachne was ascribed to Poeae s.l other than Stipeae. Moreover, our results didn't support that A. shandongensis F. Z. Li was treated as a synonym of A. infirma Büse. Species of A. shandongensis F. Z. Li and A. transmorrisonensis Hayata var. kunyushanensis F. Z. Li were maintained. A. transmorrisonensis Hayata var. kunyushanensis F. Z. Li was new named as A. sozanensis Hayata var. Kunyushanensis (F. Z. Li) F. Z. Li & C. M. Xu. Palynology: Pollen grains of Agrostideae, Aveneae, Poeae and Triticeae were investigated using light and scanning electron microscopy (including cited Ma et al., 2004). Pollen morphology was highly identical, pollen grains were spherical or ellipsoidal with a single operculate aperture surrounded by a raised rim. The surface of exine was interspersed with granules. group.Except for some differences in operculm, exine ornamentation, there was limited systematic value in recognizing taxa at generic level or above.Carpology: Four types and five subtypes were distinguished based on length of caryopsis, vental view, colour of caryopsis, and hilium type. Caryopsis of Agrostideae, Phalarideae, Aveneae, and Poeae were unanimous , belong to typeⅠ. The genus Beckmannia shared the same subtype with the genera Poa and Puccinellia. Although Alopecurus and Phleum were ascribed to typeⅡ, spermoderm sculpture were similar to Beckmannia which was reticulate. On the other hand, their spermoderm sculptures were different from other species which were striate sampled in this study. Based on caryopsis morphology, Cinna was unnecessary to treat as Cinnae. Moreover, spermoderm sculpture of Brachyelytrum was distinct. This would provide new evidence for the erection of Brachyelytreae.Molecular phylogeny: Phylogenetic analyses were conducted on sequences of nuclear ITS and chloroplast trnL-F sequences for species of Agrostideae and allies. The analyses were performed using parsimony and Bayesian reconstruction methods, and the species Phyllostachys glauca as outgroup. Better phylogenetic resolution was obtained from the trnL-F dataset compared to the ITS dataset. In combined tree, Agrostis, Polypogon, Calamagrostis (Agrostideae) and Avena (Aveneae), Phalaris (Phalarideae) gathered together with the bootstrap value of 98%. Molecular evidence indicated that Agrostideae, Aveneae, and Phalarideae should be merged with a broad Aveneae. Furthermore, our results suggested that the lineages Aveneae s.l. and Poeae have a close relationship. The two tribes were no distinct boundaries, and consequently they should be merged. Based on molecular analysis, it was appropriate to treat the two as subtribes, Poinae and Aveninae. Based on molecular sequences sampled in this study, Beckmannia, Alopecurus, and Phleum was more related to Poa, Sclerochloa and Puccinellia than to Agrostis, Deyeuxia, Calamagrostis and Polygonon. We were inclined to put Beckmannia, Alopecurus, and Phleum in Poinae. The treatment of some problematical species of Agrostis: One new species and one new variety were described e.g. Agrostis tashanensis F. Z. Li & C. M. Xu and Agrostis clavata var. ludongensis F. Z. Li & C. M. Xu.In conclusion, micro-morphology including leaf abaxial epidermis, pollen grains, caryopsis and molecular evidence all supported Phalarideae and Agrostideae were merged with Aveneae. This view is congruent with currently most botanists. Moreover, this treatment is also concordant with Flora of China (2006). Meanwhile, our study also displayed that Aveneae and Poeae have a close relationship to each other. Microcharacters and ITS and trnL-F data all indicated Aveneae s.l. and Poeae have arbitrary boundaries. We are in agreement with Hilu and Wright, who suggested Aveneae s.l. should be merged with Poeae s.l.. We are inclined to divide Poeae s.l. into two subtribes, Poinae and Aveninae. According to our study, we further make sure that the genus Beckmannia should not be placed in Chlorideae. On the basis of molecular sequences and caryopsis morphology sampled in this study, Beckmannia was more related to Poa, Puccinellia than to Agrostis, Deyeuxia, Calamagrostis and Polygonon. We concluded that the genus Beckmannia was more suitably to place in Poinae. In addition, our results indicated Alopecurus, Beckmannia and Phleum had close relative, but Phalaris was distant from them. Brachyelytrum was promoted as Brachyelytreae. We also consider it unnecessary to treat Cinna as Cinneae.