Crinoid phylogeny: new interpretation of the main Permo-Triassic divergence, comparisons with echinoids and brachiopods, and EvoDevo interpretations of major morphological variations

An alignment of nuclear and mitochondrial gene sequences (1529 data columns from three genes: nuclear‐encoded 28S ribosomal DNA, mitochondrial 16S rDNA, and cytochrome oxidase subunit 1) from 120 specimens representing a wide selection of extant crinoids is used in Bayesian relaxed‐clock analyses to construct the maximum clade credibility tree and to locate the root. This tree, and its root position, largely agree with those previously published on the basis of mid‐point and outgroup rooting, and, using a likelihood ratio test, we find that our independently aligned data do not depart significantly from a strict molecular clock. Our tree, similar to those published previously, shows a basal split into two main clades but, unlike previous reports, we identify a phenotypic basis for the split and name the clades accordingly, ‘Stalked and quasi‐stalked’ (SqS) and ‘Stalked but facultatively unstalked’ (SfU), with the latter corresponding to the ‘comatulids’. There is some correspondence between clades and morphologically‐defined Linnean categories in the SqS clade (e.g. the family Hyocrinidae corresponds to a clade). However, there is almost no correspondence between them in the SfU clade, within which we abandon Linnean names and instead use the letters A to E, suggesting, as previously noted, that considerable morphology‐based taxonomic oversplitting has occurred, such that critical revision that takes account of molecular results is overdue. The distinctive, living crinoid Caledonicrinus is placed as sister to all other members of the SqS clade (i.e. its detailed relationships remain obscure). The results of the present study do not exclude the Articulata hypothesis under which all extant crinoids belong to a single radiation but, using the first appearance of fossil cyrtocrinid ossicles as a calibration datum, the root node of our time‐tree dates to before 258 Mya [i.e. in the Permian (Kungurian), rather than in the Triassic]. Simple comparisons are made between crinoids, echinoids, and selected articulate brachiopods using p‐distance networks and maximum likelihood pairwise patristic distances as proxies for molecular evolution, and the numbers of extant families as (the only available but potentially unreliable) proxy for morphological evolution. These analyses suggest that, over approximately the same period (Triassic to present), echinoids show approximately five‐fold lower rates of molecular evolution than crinoids or brachiopods, whereas both crinoids and echinoids show higher rates of morphological (taxonomic, family‐level) evolution than brachiopods. These differences are discussed briefly in terms of developmental constraint and character plasticity. The evolutionary history of extant crinoids is given a speculative interpretation in terms of evolutionary developmental gene expression (EvoDevo) on the basis of four hypothetical gene regulatory networks or cassettes (Stalk, Cirrus, Abscission, and Multiply‐or‐Divide) whose presence/absence and activity/inactivity could combine to account for the phenotypes of the main crinoid clades. The conventional division of living crinoids into ‘stalked’ and ‘unstalked’ categories is strongly deprecated