Broad taxon and gene sampling indicate that chaetognaths are protostomes

Despite advances in phylogenetic methods, there are still a number of enigmatic phyla whose affinities remain poorly resolved. One of the most recalcitrant of these is a group of small predatory marine invertebrates, the chaetognaths (arrow worms). Resolution of the phylogenetic position of the chaetognaths is key for reconstructing the evolutionary history of some of the most fundamental features of animals, including those that have been used to delineate two major clades of animals — the protostomes and deuterostomes. An affiliation of chaetognaths to deuterostomes was inferred from a number of shared embryological and morphological features including radial, indeterminate cleavage, a posterior position of the blastopore (deuterostomy), enterocoelous coelom formation, and a tripartite adult body plan with a post-anal tail [1,2]. Various other features (e.g., a ventrally positioned nervous system, ecdysozoan-like spines, and a lack of circular body wall muscle) have been interpreted as evidence of a protostome affinity [3]. Here, we analyse the phylogenetic position of the chaetognaths using existing datasets, i.e. small and large subunit nuclear ribosomal RNAs (SSU and LSU), and complete mitochondrial genomes. We also report a new dataset from the tropomyosin gene of many Metazoa. Each of these three datasets is broadly sampled across the Metazoa. We have also sequenced ~ 5000 ESTs from the pelagic chaetognath Flaccisagitta enflata, recovering 72 of the genes used by Philippe et al. [4]. Preliminary analyses of SSU and LSU ribosomal RNAs showed chaetognath sequences clustering with taxa that also show long phylogenetic branches and a higher than average GC content (see Supplemental data published with this article online). In order to address the possibility that this grouping was an artifact of GC content, we recoded all nucleotides as purines (A,G = R) and pyrimidines (C,T/U = Y). Maximum likelihood and Bayesian analyses of this recoded dataset grouped the chaetognaths with the priapulids within the Ecdysozoa (Supplemental data). However, with priapulids excluded, chaetognaths grouped with the Lophotrochozoa (100% posterior probability), without resolving their position within the group (Supplemental data). The association with the Ecdysozoa, therefore, depends on a single taxon and is not likely to reflect a general affinity for the group. Furthermore, we analyzed two complete chaetognath mitochondrial genomes in combination that previously were analyzed independently with differing results [5,6]. Bayesian analysis of the nucleotide matrix places chaetognaths within the Lophotrochozoa allied to molluscs, and Bayesian analysis of the amino acid matrix places them as sister to the …