Conflicting phylogenetic signals at the base of the metazoan tree Antonis Rokas, a,1 Nicole King, a,1 John Finnerty, b and Sean B. Carroll a, * a Howard Hughes Medical Institute and Laboratory of Molecular Biology, University of Wisconsin-Madison, 1525 Linden Drive, Madison, WI 53706-1596, USA b Department of Biology, Boston University, 5 Cummington St., Boston, MA 02215, USA *Author for correspondence (e-mail: sbcarrol@facstaff.wisc.edu) SUMMARY A phylogenetic framework is essential for under- standing the origin and evolution of metazoan develop- ment. Despite a number of recent molecular studies and a rich fossil record of sponges and cnidarians, the evolutionary relationships of the early branching metazoan groups to each other and to a putative outgroup, the choanoflagellates, remain uncertain. This situation may be the result of the limited amount of phylogenetic information found in single genes and the small number of relevant taxa surveyed. To alleviate the effect of these analytical factors in the phylogenetic recons- truction of early branching metazoan lineages, we cloned multiple protein-coding genes from two choanoflagellates and diverse sponges, cnidarians, and a ctenophore. Comparisons of sequences for a-tubulin, b-tubulin, elongation factor 2, HSP90, and HSP70 robustly support the hypothesis that choanoflagellates are closely affiliated with animals. However, analyses of single and concatenated amino acid sequences fail to resolve the relationships either between early branch- ing metazoan groups or between Metazoa and choano- flagellates. We demonstrate that variable rates of evolution among lineages, sensitivity of the analyses to taxon selection, and conflicts in the phylogenetic signal contained in different amino acid sequences obscure the phylogenetic associations among the early branching Metazoa. These factors raise concerns about the ability to resolve the phylogenetic history of animals with molecular sequences. A consensus view of animal evolution may require investigations of genome-scale characters. INTRODUCTION The origin of multicellular Metazoa from a unicellular ancestor, and the subsequent metazoan radiation, involved the evolution of new modes of development. Comparing modern representatives of the earliest branching metazoan lineages (e.g., sponges, cnidarians, ctenophores) with bilater- ians may reveal the role of novel developmental mechanisms in the early evolution of Metazoa. Inferences regarding the polarity of changes in development along evolutionary lineages (i.e., distinguishing between what is ancestral and what is derived) require a robust phylogenetic framework. Therefore, a well-resolved phylogeny of early branching meta- zoans and candidate outgroups, such as the choanoflagellates, is essential for understanding the beginnings of animal evolution. The phylogenetic relationships of sponges, cnidarians, and ctenophores have recently received increased attention. From a histological perspective, sponges are perhaps the simplest of all animals, lacking tissue level organization. Higher level sponge taxonomy is based on the composition of small skeletal elements known as spicules (Brusca and Brusca 1990). Members of the sponge classes Demospon- giae and Hexactinellida are often treated as sister-groups because both synthesize siliceous spicules, whereas a third class, Calcarea, synthesizes spicules from calcium carbonate. However, despite the differences in the composition of their spicules, Calcarea and Demospongiae are sometimes linked to the exclusion of Hexactinellida because their cellulariza- tion and lack of symmetry differ dramatically from the syncytia and partial radial symmetry found in hexactinellids (Brusca and Brusca 1990). Uncertainty also exists about whether the sponges are mono- phyletic or paraphyletic. For instance, one study found that 18S rDNA analyses cannot differentiate between paraphyly and monophyly of the sponges (Medina et al. 2001), whereas another analysis supported sponge paraphyly but failed to resolve the position of Hexactinellida (Borchiellini et al. 2001). Furthermore, analyses of single protein-coding genes either fail to reveal the interrelationships between Calcarea, Demospongiae, and Hexactinellida (Borchiellini et al. 1998) or support a hypothesis (the grouping of Calcarea and Demospongiae; Kruse et al. 1998) that is in conflict with 18S rDNA analyses (Medina et al. 2001). Whether sponges are monophyletic or paraphyletic bears directly on interpreta- tions of the early stages in the evolution of metazoan development. 1 These authors contributed equally to this work. EVOLUTION & DEVELOPMENT 5:4, 346–359 (2003) & BLACKWELL PUBLISHING, INC. 346