Grasses as a single genetic system: reassessment 2001.

Grasses are a large, diverse, and successful family of monocotyledonous flowering plants characterized most dramatically by their spikelet style of floral structure (Fig. 1a) and by the number and pattern of these spikelets in the inflorescence. There are about 10,000 species that are categorized as grasses by taxonomists who traditionally follow embryonic, architectural, and anatomical characters (e.g. Clayton and Renvoize, 1986). More recent comparative DNA sequence information confirms that all grasses examined to date did indeed diverge from one common ancestral population of “grass alleles” and are distinct from the nearest non-grass family, the Joinvilleaceae. In general, the taxonomic distinction within the grasses continuously divides these species into subfamilies, supertribes, tribes, subtribes, genera, species, and subspecies. As additional species have been added to the sequence databases, a phylogenetic trend has emerged. Where Clayton and Renvoize (1986) recognized six subfamilies, Kellogg (1998 and refs. therein) recognized 13, and the “in progress” data of the Grass Phylogenetics Working Group (http://www.ftg.fiu.edu/grass/gpwg) could be used to justify recognition of at least 16 subfamilies, often composed of one or a few species. At present the phylogenetic relationship—the exact branch relationships—among most of these 16 subfamilies is not known, as if they originated in a single grass adaptive radiation, estimated to be about 70 million years ago (Kellogg, 1998 and refs. therein). In fact, there is likely one specific order by which the several grass subfamilies are related, one to another, but sequence-based trees are necessarily quantitative and often cannot resolve deep branches. Perhaps phylogenetic trees based on chromosomal breakpoints, not being time-dependent, will fare better. Until about 8 years ago, one grass species maps and mutant collections, however interesting, did not directly affect research on another grass species, even though grass genomes were known to be related. This state of being isolated by commodity changed in 1992–1993. Of particular notice was the research of Dr. Steven Tanksley and coworkers on gene order comparisons of maize and rice (Ahn and Tanksley, 1993) and Dr. Mike Gale and coworkers and their many international collaborators who moved from mapping wheat relatives to more distant grasses (summarized by Moore et al., 1995). The graphic summary of these mapping data, greatly simplified, has become popularly known as “The Circle Diagram” because of a method used to draw the expressed gene sequence (EST) maps of several different grass species on one radial axis. A recent Circle Diagram (Gale and Devos, 1998) includes crop grass species from four subfamilies: Pooids (wheat and oat), Panicoids (maize, sorghum, sugarcane, and foxtail millet), Oryzoids (rice), and the Chlorinoids (finger millet). In general, gene probes (ESTs) were chosen to span entire genomes at intervals of 10 or 20 map units. Comparative mapping in the grasses has been reviewed recently (Devos and Gale, 2000). The general conclusion is that all of the grasses in the four subfamilies examined have their genes in about the same order so that one can conclude with confidence that one ancestral genome remains recognizable in its descendents. The huge differences in DNA content/ haploid genome and the differences in chromosome number seem to have little or nothing to do with gene number or order. Recent polyploids, like bread wheat, for example, certainly have multiples of the ancestral gene number, but even this simple expectation about polyploids proves to be false for descendents of ancient duplication events, as will be discussed.