Comparing patterns of evolution: larval and adult life history stages and ribosomal RNA of post-Palaeozoic echinoids

Publisher Website
Google Scholar
Abstract
We have taken a total-evidence approach to the phylogeny of 29 extant echinoids, combining data from larval morphology, adult morphology, small subunit rRNA complete gene sequence and large subunit rRNA partial gene sequence: a total of 176 morphological and 121 molecular phylogenetically informative characters. Also included are 13 extinct taxa for which we know only adult morphology. Parsimony analysis of the combined data generated 28 equally parsimonious solutions, differing primarily in the positioning of a few fossil taxa. We reduced these to a single working hypothesis of echinoid relationships by pruning fossil taxa and one extant species. Patterns of morphological evolution of larval and adult stages were compared by optimizing character sets onto the total evidence tree and assigning each character transformation to a branch. Branch nodes were dated by reference to the first appearance of one or other sister taxon in the fossil record. From this we demonstrate that larval and adult morphological evolution has proceeded in a mosaic-like fashion over the last 250 Ma. A similar comparison between morphological and molecular data finds equally weak correlation between rates of ribosomal RNA evolution and rates of morphological evolution, implying that morphology and ribosomal genes have also evolved largely independently. Larval characters appear to be more prone to homoplasy than adult characters, even when comparison is restricted to adult organs of similar size and structural complexity as the larvae. As morphological and molecular apomorphies accrue over time, there is a general correspondence between the duration of a particular branch and the number of apomorphies assigned to that branch. However, we found no evidence that overall molecular rates of evolution were any more strictly clock-like than morphological character transformations, although mapping transversions only improved the fit to a clock-like model for molecular data.