At high temperatures and relatively low CO 2 concentrations, plants can most efficiently fix carbon to form carbohydrates through C 4 photosynthesis rather than through the ancestral and more widespread C 3 pathway. Because most C 4 plants are grasses, studies of the origin of C 4 are intimately tied to studies of the origin of the grasses. We present here a phylogeny of the grass family, based on nuclear and chloroplast genes, and calibrated with six fossils. We find that the earliest origins of C 4 likely occurred about 32 million years ago (Ma) in the Oligocene, coinciding with a reduction in global CO 2 levels. After the initial appearance of C 4 species, photosynthetic pathway changed at least 15 more times; we estimate nine total origins of C 4 from C 3 ancestors, at least two changes of C 4 subtype, and five reversals to C 3 . We find a cluster of C 4 to C 3 reversals in the Early Miocene correlating with a drop in global temperatures, and a subsequent cluster of C 4 origins in the Mid-Miocene, correlating with the rise in temperature at the Mid-Miocene climatic optimum. In the process of dating the origins of C 4 , we were also able to provide estimated times for other major events in grass evolution. We find that the common ancestor of the grasses (the crown node) originated in the upper Cretaceous. The common ancestor of maize and rice lived at 52 AE 8 Ma.

Panicum L. is a cosmopolitan genus with approximately 450 species. Although the genus has been considerably reduced in species number with the segregation of many taxa to independent genera in the last two centuries, Panicum remains a heterogeneous assemblage, as has been demonstrated in recent years. The genus is remarkably uniform in its floral characters but exhibits considerable variation in anatomical, physiological, and cytological features. As a result, several classifications, and criteria of what the genus should really include, have been postulated in modern literature. The purpose of this research, based on molecular data of the chloroplast ndhF gene, is to test the monophyly of Panicum, to evaluate infrageneric classifications, and to propose a robust phylogenetic hypothesis. Based on the present results, previous morphological and molecular phylogenetic studies, and inferred diagnostic morphological characters, we restrict Panicum sensu stricto (s.s.) to the former subgenus Panicum and support recognition of Dichanthelium, Phanopyrum, and Steinchisma as distinct genera. We have transfered other species of Panicum to other genera of the Paniceae. Most of the necessary combinations have been made previously, so few nomenclatural changes have been required. The remaining species of Panicum sensu lato (s.l.) are included within Panicum incertae sedis representing isolated species or species grouped within monophyletic clades. Additionally, we explore the performance of the three codon position characters in producing the supported phylogeny.

The genus Setaria is the largest genus in the so-called bristle clade, a monophyletic group of panicoid grasses distinguished by the presence of sterile branches, or bristles, in their inflorescences. The clade includes both foxtail millet and pearl millet, the latter an important cereal crop in dry parts of the world. Other members of the clade are weeds that are widespread agricultural pests. Previous molecular phylogenetic studies have suggested that Setaria might not be monophyletic but did not have a large enough sample of species to test this rigorously. In addition, taxonomic studies have suggested a close relationship between Setaria and Paspalidium, with some authors combining them into a single genus, but molecular studies included too few Paspalidium accessions for a meaningful conclusion. Accordingly, we have produced 77 new sequences of the chloroplast gene ndhF for 52 species not in previous analyses. These were added to available sequences for 35 species in 10 genera of the bristle clade and four outgroup taxa. We find that Setaria species fall into several moderately to strongly supported clades that correlate with geography but not with the existing subgeneric classification. Relationships among these clades and among other genera within the bristle clade are unclear. Constraint experiments using the approximately unbiased test reject the monophyly of Pennisetum, Setaria, and Setaria plus Paspalidium, as well as several other groupings, although the test may be overly sensitive and prone to Type I error. The more conservative Shimodaira-Hasegawa test fails to reject monophyly of any of the tested clades.