Crinoid phylogeny: a preliminary analysis (Echinodermata: Crinoidea)

We describe the first molecular and morphological analysis of extant crinoid high-level inter-relationships. Nuclear and mitochondrial gene sequences and a cladistically coded matrix of 30 morphological characters are presented, and analysed by phylogenetic methods. The molecular data were compiled from concatenated nuclear-encoded 18S rDNA, internal transcribed spacer 1, 5.8S rDNA, and internal transcribed spacer 2, together with part of mitochondrial 16S rDNA, and comprised 3,593 sites, of which 313 were parsimony-informative. The molecular and morphological analyses include data from the bourgueticrinid Bathycrinus; the antedonid comatulids Dorometra and Florometra; the cyrtocrinids Cyathidium, Gymnocrinus, and Holopus; the isocrinids Endoxocrinus, and two species of Metacrinus; as well as from Guillecrinus and Caledonicrinus, whose ordinal relationships are uncertain, together with morphological data from Proisocrinus. Because the molecular data include indel-rich regions, special attention was given to alignment procedure, and it was found that relatively low, gene-specific, gap penalties gave alignments from which congruent phylogenetic information was obtained from both well-aligned, indel-poor and potentially misaligned, indel-rich regions. The different sequence data partitions also gave essentially congruent results. The overall direction of evolution in the gene trees remains uncertain: an asteroid outgroup places the root on the branch adjacent to the slowly evolving isocrinids (consistent with palaeontological order of first appearances), but maximum likelihood analysis with a molecular clock places it elsewhere. Despite lineage-specific rate differences, the clock model was not excluded by a likelihood ratio test. Morphological analyses were unrooted. All analyses identified three clades, two of them generally well-supported. One well-supported clade (BCG) unites Bathycrinus and Guillecrinus with the representative (chimaeric) comatulid in a derived position, suggesting that comatulids originated from a sessile, stalked ancestor. In this connection it is noted that because the comatulid centrodorsal ossicle originates ontogenetically from the column, it is not strictly correct to describe comatulids as unstalked crinoids. A second, uniformly well-supported clade contains members of the Isocrinida, while the third clade contains Gymnocrinus, a well-established member of the Cyrtocrinida, together with the problematic taxon Caledonicrinus, currently classified as a bourgueticrinid. Another cyrtocrinid, Holopus, joins this clade with only weak molecular, but strong morphological support. In one morphological analysis Proisocrinus is weakly attached to the isocrinid clade. Only an unusual, divergent 18S rDNA sequence was obtained from the morphologically strange cyrtocrinid Cyathidium. Although not analysed in detail, features of this sequence suggested that it may be a PCR artefact, so that the apparently basal position of this taxon requires confirmation. If not an artefact, Cyathidium either diverged from the crinoid stem much earlier than has been recognised hitherto (i.e., it may be a Palaeozoic relic), or it has an atypically high rate of molecular evolution