The genome sequences of Arabidopsis 1 and rice 23 have stimulated great advances throughout the plant sciences. Comparisons of these eudicot and monocot genomes have provided many insights into the genome characteristics and evolutionary histories of both lineages [e.g. 456], and comparisons involving additional species are generating a more global picture of angiosperm genome evolution 789.

These multispecies comparisons, and comparative plant sciences more generally, have been aided by the well-supported understanding of evolutionary relationships among flowering plants that has emerged over the last decade [e.g. 10111213]. Among the most noteworthy phylogenetic results is the well-supported inference that whereas monocots form a clade, the dicots as traditionally circumscribed do not. Rather, monocots are derived from within the "primitive" dicot grade, now collectively referred to as basal angiosperms (Fig. 1). The "eudicots" (or "tricolpates" 14; Fig. 1) do form a clade that comprises ca. 75% of all angiosperm species 15, and most of this diversity is found among the "core" eudicots, which include the rosids, asterids, and Caryophyllales (Fig. 1). Model systems such as Arabidopsis thaliana, tomato (Lycopersicon esculentum), cotton (Gossypium), poplar (Populus), barrel medic (Medicago truncatula) and ice-plant (Mesembryanthemum crystallinum), are all representatives of the core eudicot clade (Fig. 1). Rice (Oryza savita), maize (Zea mays), wheat (Triticum aestivum), barley (Hordeum vulgare), sorghum (Sorghum bicolor), and sugarcane (Saccharum officinale) are all members of the grass family (Poaceae), a phylogenetically derived lineage within the monocots (Fig. 1). Although comparisons of the rice and Arabidopsis genomes will undoubtedly identify many features of the ancestral angiosperm genome, this pair-wise comparison alone will not be able to distinguish Arabidopsis-specific attributes from those specifically absent in rice or visa versa. The recent posting of high coverage genome sequence for Populus trichocarpa 16 is a major advance for comparative plant genomics, but even Populus-Arabidopsis-rice comparisons cannot distinguish features common to all angiosperms from those that arose in the most recent common ancestor of eudicots and monocots, which existed at least 125 million years ago 17 and perhaps more than 140 million years ago [e.g. 181920]. In general, the resolving power of comparative plant genomics will increase with additional taxa representing key lineages in plant phylogeny (Fig. 1). Increased genomic resources for phylogenetically diverse plant species will lead to a better understanding of plant genome evolution, the diversification of gene families, and the origins of reproductive characteristics common to all flowering plants.

Basal angiosperms and basal eudicots (e.g., Ranunculales), while comprising a small percentage of the total number of extant angiosperm species, nonetheless encompass an astonishing spectrum of developmental patterns and floral forms 2122. In turn, this diversity provides a clear opportunity to reconstruct to the basal condition of angiosperms, and thereby bridge the evolutionary gap between model eudicot and monocot genomes. Understanding the evolution of angiosperm genes and genomes, including the floral transcriptome, requires three-way and higher-order comparisons that extend beyond Arabidopsis and rice. This point is widely appreciated, and comparative plant genomics is being fueled by the availability of genomic resources for a growing number of plant species. The addition of species representing basal angiosperms, basal eudicot, and non-grass monocot lineages will be especially valuable, not only for flowering research, but also for more general "reconstructomic" studies of housekeeping and transcription factor functions.

A primary objective of the Floral Genome Project (FGP; 23) is to uncover patterns of conservation and divergence of the floral transcriptome among angiosperms, particularly to elucidate the role of gene duplications and shifting expression patterns in the origin and diversification of angiosperms. The FGP has constructed a large collection of non-normalized nor tissue subtracted cDNA libraries and 5' EST sets from developing reproductive tissues for selected species of basal eudicots, basal angiosperms and gymnosperms (Table 1). These species represent not only key nodes in the angiosperm phylogenetic tree and its sister group (gymnosperms), but also a diversity of reproductive structures and developmental patterns. While multi-species comparisons of large sequence data sets are already possible for Poaceae 924 Solanaceae 25, and Brassicaceae 72627, the addition of large EST sets for basal angiosperms opens the door to fundamental comparative genomics investigations of the origin and diversification of flowering plants.

Random 5' sequencing of cDNAs from basal flowering plants has so far (as of 12/01/04) generated 70,514 ESTs assembled into 48,170 unique gene sequences (Table 1). These materials should provide essential resources for comparative genomic research because they represent previously poorly sampled genomes placed at crucial points in angiosperm phylogeny. Gene sequences from the gymnosperm Welwitschia, the basalmost angiosperms Amborella and Nuphar, the basal monocot Acorus, the magnoliids Persea, Liriodendron and Saruma, and the basal eudicot Eschscholzia will aid in placing boundary dates on the origins of florally-expressed gene families, help resolve patterns of gene and genome evolution within the flowering plants, and bridge critical gaps in comparative analyses involving monocot and eudicot model systems. Identification of cDNA clones for finished sequencing has been aided by efficient sorting of EST sequences into putative gene families based on whole Arabidopsis/rice proteome comparison. Phylogenetic analyses of ESTs are providing new insights into the process of gene family evolution in relation to the origin and diversification of the angiosperms. Here we introduce our EST database and provide some examples of broad utility of these data in comparative analyses.