Grass Phylogeny and Classification: Conflict of Morphology and Molecules

The grass family (Poaceae) has been recognized as a unique lineage for more than 200 years, based initially on characters of inflorescence and spikelet morphology and corroborated by anatomical and other features in the latter half of the twentieth century. There have been a number of attempts to derive a phylogeny of sections of the family from structural (morphological and anatomical) characters. However, more recently the Grass Phylogeny Working Group (2001) applied cladistic principles to the whole family using molecular and structural characters. This study produced a single most-parsimonious tree from six molecular sequence data sets, chloroplast restriction site data, and structural data from a representative set of 62 grasses and four outgroup taxa. However, the cladogram reveals a weak correlation with structural characters. Only three out of 12 subfamilies have unequivocal anatomical synapomorphies and there is no unequivocal morphological synapomorphy in any subfamily. Homoplasious character states outnumber synapomorphic states to a large degree. Cladistic analyses of grass genera in recent studies, based on both micromorphological, anatomical, and molecular data, have often produced clades that are difficult or impossible to define by readily observable features. Examples are shown in this paper for some genera of Danthonioideae, the panicoid tribe Andropogoneae, and the chloridoid tribe Triodieae. A recommendation is made that formal taxonomic and nomenclatural changes should only be applied, particularly at the species level, when the lineages within a phylogeny correlate with marker morphological characters.