Phylogeny and Evolution of Glass Sponges (Porifera, Hexactinellida)

Abstract

Reconstructing the phylogeny of sponges (Porifera) is one of the remaining challenges to resolve the metazoan Tree of Life and is a prerequisite for understanding early animal evolution. Molecular phylogenetic analyses for two of the three extant classes of the phylum, Demospongiae and Calcarea, are largely incongruent with traditional classifications, most likely because of a paucity of informative morphological characters and high levels of homoplasy. For the third class, Hexactinellida (glass sponges)--predominantly deep-sea inhabitants with unusual morphology and biology--we present the first molecular phylogeny, along with a cladistic analysis of morphological characters. We collected 18S, 28S, and mitochondrial 16S ribosomal DNA sequences of 34 glass sponge species from 27 genera, 9 families, and 3 orders and conducted partitioned Bayesian analyses using RNA secondary structure-specific substitution models (paired-sites models) for stem regions. Bayes factor comparisons of different paired-sites models against each other and conventional (independent-sites) models revealed a significantly better fit of the former but, contrary to previous predictions, the least parameter-rich of the tested paired-sites models provided the best fit to our data. In contrast to Demospongiae and Calcarea, our rDNA phylogeny agrees well with the traditional classification and a previously proposed phylogenetic system, which we ascribe to a more informative morphology in Hexactinellida. We find high support for a close relationship of glass sponges and Demospongiae sensu stricto, though the latter may be paraphyletic with respect to Hexactinellida. Homoscleromorpha appears to be the sister group of Calcarea. Contrary to most previous findings from rDNA, we recover Porifera as monophyletic, although support for this clade is low under paired-sites models.

New insights into the phylogeny of glass sponges (Porifera, Hexactinellida): Monophyly of Lyssacinosida and Euplectellinae, and the phylogenetic position of Euretidae

“…Full-length 18S, and partial 28S and mitochondrial 16S rDNA sequences were obtained as described, and added to the previously reported alignments (see Dohrmann et al, 2008). A few additional sites (9 bp overall) had to be excluded due to further ambiguities regarding positional homology introduced by the new sequences.…”

“…A few additional sites (9 bp overall) had to be excluded due to further ambiguities regarding positional homology introduced by the new sequences. The final concatenated alignment [TreeBase (www.treebase.org) matrix accession number M4266] consisted of 3426 bp for 60 taxa, including the 17 outgroup species (from Choanoflagellata, Demospongiae, Homoscleromorpha, Calcarea, Placozoa, and Cnidaria) used in our previous study (see Dohrmann et al, 2008). The total amount of missing data decreased from 18.75% to 16.82% because of the more complete sequencing of Farrea and one Hyalonema species (see Table 1).…”

“…We chose the RNA16C model because it is equivalent to models 7D and 6B (Savill et al, 2001); more general variants (16A, 16D) would have probably resulted in over-parameterization as this was already the case with the most general 6-and 7-state models 6A and 7A (see Results and Dohrmann et al, 2008). Phylogenetic analyses were performed with PHASE v2.0 (http://www.bioinf.manchester.ac.uk/resources/phase/) essentially as in Dohrmann et al (2008), except that chains were run for 5 Â 10 7 post-burnin generations and sampled every 10 3 generations. Because currently available Bayesian Markov Chain Monte Carlo (MCMC) implementations might not be able to accurately infer branch lengths under certain conditions (e.g., Kolaczkowski and Thornton, 2007;Yang and Rannala, 2005;Yang, 2007), we re-estimated branch lengths of the consensus phylograms with the maximum likelihood (ML) program optimizer of the PHASE package, using exactly the same models as in the Bayesian analyses.…”

“…Implications of these results for evolution of morphological characters are discussed. We also elaborate further on comparison of RNA (paired-sites) substitution models (see Savill et al, 2001;Dohrmann et al, 2008) and highlight the non-trivial relationship of taxon sampling and model choice in deep metazoan phylogenies inferred from rDNA data.…”

“…Here, we have addressed these issues by including 10 additional species that were previously unavailable to us (i.e., Dohrmann et al, 2008). Among them was a Farrea species that yielded high-quality sequences of all three markers, and a representative of the most species-rich family of Sceptrulophora, the Euretidae.…”