The Gnetales: fossils and phylogenies

Abstract: The phylogeny of seed plants has been debated for more than a hundred years and is still not fully understood. Morphological analyses have consistently resulted in a phylogeny in which cycads are the earliest diverging seed plants, and Gnetales and angiosperms are sisters. Molecular data has, however, rarely supported this result. Mitochondrial data resolves Gnetales as sister to the conifer family Pinaceae, but results from the nuclear and chloroplast genomes are ambiguous. My studies showed a conflict between transitions and transversions within the chloroplast genes rbcL and atpB, and consequently, different topologies are obtained from different substitution models and character codings. Transversion parsimony as well as maximum likelihood using GTR-model results in an association between Gnetales and conifers, whereas equally weighted parsimony and a simpler model (F81) resolves Gnetales as sister to all other seed plants. Taxon sampling error may be another problem. I could easily obtain seemingly well-supported, but conflicting, topologies simply by substituting a few terminals in bayesian and parsimony analyses of rbcL. These problems do not only concern the systematic position of the Gnetales, but other fundamental issues such as the monophyly of conifers and gymnosperms, the sister group of Ginkgo etc. The most recent articles on the subject argue that monophyletic gymnosperms and a sister relationship between Gnetales and Pinaceae is the most probable scenario because this topology is supported by mitochondrial data and by chloroplast data from which the fastest evolving sites (i.e. transitions or third codon positions) have been removed or down-weighted.The Gnetales have often been considered a small remnant of a former much greater diversity due to the pronounced morphological and ecological differences between the genera. But until recently, megafossils, which could support this were rare. We have described a seedling from the Early Cretaceous of Brazil that shares several synapomorphies with Welwitschia and dates the split between Gnetum and Welwitschia to before 110 Myr. Ephedra, the sister group of the Gnetum-Welwitschia clade, comprises a homogenous group of species; very similar in gross morphology and with few informative characters in the investigated gene regions. The similarities made it difficult to resolve relationships within the group and to assign fossils to subgroups within Ephedra. We have studied Early Cretaceous fossils from Portugal, Virginia and China and they share several unique features with modern Ephedra, e.g. the peculiar naked male gametophyte. They differ in detailed morphology, however, indicating that a diversity of Ephedra species was present in the Early Cretaceous flora. We have compared the fossils to extant species and to the extinct Erdtmanispermum, and discuss character evolution within the ephedroid lineage. The paucity of information in most gene regions has indicated a very young age (8-32 Myr) for modern species in molecular dating analyses. The fossils document that the characteristic Ephedra features were present and widespread in the Early Cretaceous and they indicate that the crown group may be of Mesozoic origin (110-127 Myr). A recently published fossil from China, which possesses a character only present in a few extant species, further supports this idea. Ephedra might once again demonstrate the absence of a linear correlation between the amount of substitutions and time.