ABSTRACTTo investigate the evolutionary diversification and morphological evolution of grass supertribe Poodae (subfam. Pooideae, Poaceae) we conducted a comprehensive molecular phylogenetic analysis including representatives from most of their accepted genera. We focused on generating a DNA sequence dataset of plastid matK gene–3’trnK exon and trnL– trnF regions and nuclear ribosomal ITS1–5.8S gene–ITS2 and ETS that was taxonomically overlapping as completely as possible (altogether 257 species). The idea was to infer whether phylogenetic trees or certain clades based on plastid and nuclear DNA data correspond with each other or discord, revealing signatures of past hybridization. The datasets were analysed using maximum parsimony, maximum likelihood and Bayesian approaches. Instances of severe conflicts between the phylogenetic trees derived from both datasets, some of which have been noted earlier, unambiguously point to hybrid origin of several lineages (subtribes, groups of genera, sometimes genera) such as Phalaridinae, Scolochloinae, Sesleriinae, Torreyochloinae; Arctopoa, Castellia, Graphephorum, Hyalopodium, Lagurus, Macrobriza, Puccinellia plus Sclerochloa, Sesleria, Tricholemma, American Trisetum, etc. and presumably Airinae, Holcinae and Phleinae. ‘Calamagrostis’ flavens appears to be an intergeneric hybrid between Agrostis and Calamagrostis. Most frequently there is good agreement of other regions of the trees, apart from intrinsic different phylogenetic resolution of the respective DNA markers. To explore the to date rather unclear morphological evolution of our study group a data matrix encompassing finally 188 characters was analysed for ancestral state reconstructions (ASR) using the tree from the combined molecular dataset as presumably best approximation to the species phylogeny. For 74 characters ASRs were feasible and yielded partly surprising results for the study group as a whole but also for some of their subdivisions. Considering taxonomy and classification it became evident, that many morphological characters show a very high degree of homoplasy and are seemingly able to change within comparatively short timespans in the evolution of our grasses. Most of the taxonomic units distinguished within our study group, e.g. as subtribes, are defined less by consistent morphological characters or character combinations and should be rather understood as clades revealed by molecular phylogenetic analysis. One reason for this extreme homoplasy concerning traditionally highly rated characters of inflorescences or spikelets and their components might be that they have little to do with pollination (always wind) or adaptation to pollinators as in other angiosperms but rather with dispersal and diaspores. Easily changing structure of spikelet disarticulation, of glume, lemma or awn characters might be advantageous in the rapid adaptation to different habitats and micro-habitats, which was evidently most successfully accomplished by these grasses. A partly revised classification of Poodae is presented, including a re-instatement of tribes Aveneae and Poeae s.str. Following a comparatively narrow delineation of preferably monophyletic subtribes, Antinoriinae, Avenulinae, Brizochloinae, Helictochloinae, Hypseochloinae are described as new. New genera are Arctohyalopoa and Hyalopodium. New combinations are Arctohyalopoa lanatiflora, A. lanatiflora subsp. ivanoviae, A. lanatiflora subsp. momica, Colpodium biebersteinianum, C. kochii, C. trichopodum, C. verticillatum, Deschampsia micrathera, Dupontia fulva, Festuca masafuerana, Hyalopodium araraticum, Paracolpodium baltistanicum, Parapholis cylindrica, P. ×pauneroi. Festuca masatierrae is a new name.Supporting Information may be found online in the Supporting Information section at the end of the article.
A systematist’s guide to estimating Bayesian phylogenies from morphological data
