Can the Cambrian explosion be inferred through molecular phylogeny?

Abstract

ABSTRACT Most of the major invertebrate phyla appear in the fossil record during a relatively short time interval, not exceeding 20 million years (Myr), 540-520 Myr ago. This rapid diversification is known as the ‘Cambrian explosion’. In the present paper, we ask whether molecular phyloge netic reconstruction provides confirmation for such an evo lutionary burst. The expectation is that the molecular phy logenetic trees should take the form of a large unresolved multifurcation of the various animal lineages. Complete 18S rRNA sequences of 69 extant representatives of 15 animal phyla were obtained from data banks. After elimi nating a major source of artefact leading to lack of resolu tion in phylogenetic trees (mutational saturation of sequences), we indeed observe that the major lines of triploblast coelomates (arthropods, molluscs, echinoderms, chordates…) are very poorly resolved i.e. the nodes defining the various clades are not supported by high bootstrap values. Using a previously developed procedure consisting of calculating bootstrap proportions of each node of the tree as a function of increasing amount of nucleotides (Lecointre, G., Philippe, H. Le, H. L. V. and Le Guyader, H. (1994) Mol. Phyl. Evol., in press) we obtain a more infor mative indication of the robustness of each node. In addition, this procedure allows us to estimate the number of additional nucleotides that would be required to resolve confidently the currently uncertain nodes; this number turns out to be extremely high and experimentally unfea sible. We then take this approach one step further: using parameters derived from the above analysis, assuming a molecular clock and using palaeontological dates for cali bration, we establish a relationship between the number of sites contained in a given data set and the time interval that this data set can confidently resolve (with 95% bootstrap support). Under these assumptions, the presently available 18S rRNA database cannot confidently resolve cladogenetic events separated by less than about 40 Myr. Thus, at the present time, the potential resolution by the palaeontolog ical approach is higher than that by the molecular one.