Phylogenetic analyses of seed plant evolution have demonstrated the importance of the Gnetales. While these distinctive gymnosperms were initially considered a sister group to the angiosperms, all recent molecular analyses have refuted the anthophyte hypothesis, with some even suggesting a close affinity to the pines. However, a sister relationship of Gnetales to the Pinaceae is not easily reconciled with morphology. This problem is further compounded by our incomplete understanding of extant diversity in Ephedra, the basalmost member of the Gnetales. To test Stapf's (1889) longstanding hypothesis of relationships within Ephedra (Alatae, Asarca, Pseudobaccatae) both morphological and nucleotide sequence data sets were analyzed. These included data from the chloroplast intergenic spacer regions (atpB-rbcL, trnF-trnL) and the plastid coding region and spacer of the rps4 gene. Our preliminary cladistic analysis suggests that New World ephedras are paraphyletic, monophyly of this group requires 32 extra steps. While the analysis provides robust support for monophyly of the Alatae clade, the monophyly of the Asarca and Pseudobaccatae clades are not supported. The traditionally accepted Old and New World clades are artificial. The distribution of fleshy bracts, once considered taxonomically important, requires two separate origins and a loss in the North American "nevadensis-viridis alliance". Hence it appears a relatively poor indicator of affinity. The topology also supports a New World origin of Ephedra, with a single migration event to the Old World. This scenario appears congruent with the currently known fossil evidence, and may be further supported with additional sampling. The apparent contrast between the low rate of molecular divergence and remarkably rapid rates of morphological differentiation may be tied to adaptive radiation of Ephedra in extreme arid environments.

Key words: atpB-rbcL, Ephedra, morphology, phylogeny, rps4, systematics