Animal phylogeny

Abstract

In the Origin of Species, Darwin used the metaphor of a tree to describe the relationship of life; indeed a tree is famously the only illustration in the book. The origin of species didn't include any explicit phylogeny of animal life, yet Darwin's book is said to have prompted the German biologist Ernst Haeckel to abandon his medical practice and turn to the study of evolution. In 1866, Haeckel produced a tree depicting the evolutionary relationships of living taxa including the animals (Figure 1). Looking at Haeckel's tree we can observe two contradictory things. First, we can see that much of the detail is very familiar in today's phylogenies — we recognise groups such as echinoderms, arthropods, molluscs and vertebrates; these groups represent the animal phyla. Second, many of the relationships in Haeckel's tree between these phyla (molluscs with the seasquirt (Himatega), echinoderms with annelids and arthropods) do not accord at all well with a more modern concept of animal phylogeny (Figure 2). This dichotomy seems to derive in large part from the definition of a phylum. Members of a phylum typically show a well-defined body plan that is not obviously related to that of any other group. What this means practically is that it is easy to recognise an echinoderm, arthropod, mollusc or annelid but much harder to see how they might be related to one another. This apparent uniqueness of phyla probably stems in part from their definition as distinct groups and in part from the manner of phylogenesis in the Precambrian during which they originated. Whatever the explanation, it is this dichotomy that has made understanding phylum level animal relationships so difficult and so controversial. Inevitably, the use of molecular genetic data has had an enormous impact on unraveling animal phylogeny. Far and away the most important contribution has come from the study of the small and large subunit ribosomal RNA genes (SSU and LSU). Our current view of animal phylogeny, which is often referred to as the ‘new animal phylogeny’, is in large part derived from studies of SSU but various conclusions regarding this tree have been strengthened by a diversity of additional sources of molecular evidence — in particular from the ‘phylogenomic approach’, which entails the use of whole genome sequences and large EST datasets to assemble datasets of many tens or hundreds of concatenated genes for phylogenetic analysis. Further support for individual groups has come from the discovery of ‘rare genomic changes’, which are complex genetic novelties such as a change in mitochondrial genetic code or gene order; their complexity and rarity should, in principle, exclude the possibility of convergent evolution in unrelated groups. The most ancient division within the animals is that between diploblasts and triploblasts. Diploblasts have two distinct germ layers — endoderm/gut and ectoderm/skin — triploblasts have an additional layer, the mesoderm, between these two. The relationships of the principal diploblast phyla — Cnidaria (jellyfish and sea anemones), Ctenophora (sea gooseberries/comb jellies) and Porifera (sponges) — are uncertain; most importantly, they do not seem to form a natural or monophyletic grouping. The cnidarians and ctenophores are closer to the triploblasts than to the sponges, which may themselves be paraphyletic. Triploblastica is synonymous with Bilateria and the group is classically defined both by their three tissue layers and by their bilateral symmetry. This said, recent analyses of cnidarians show evidence both of a largely hidden bilaterality and for some form of mesoderm. The anthozoan Nematostella vectensis has an oral/aboral axis patterned by the same genes that pattern the bilaterian antero-posterior axis and a second axis genetically similar to the bilaterian dorso-ventral axis; this means it is, in essence, bilaterally symmetrical. The presence in Nematostella of orthologs of a number of triploblast mesoderm patterning genes indicates, at the very least, a form of mesodermal precursor if not secondary loss of a tissue homologous to the mesoderm of the triploblasts. While the defining characteristics of the triploblasts/bilaterians may have arisen earlier than previously thought, there can be little doubt as to the monophyly of triploblastic bilaterians. This clade is strongly supported in numerous molecular phylogenetic analyses including phylogenomic datasets of over 140 genes and also by three novel changes in mitochondrial genetic code (with rare exceptions due to reversions). Within the triploblasts, there are two major branches, the Protostomia and Deuterostomia. Their names reflect the fundamentally different fates of the primary embryonic opening, the blastopore. The blastoporal opening is formed when a region of the spherical blastula embryo invaginates to form an inpocketing of tissue, called the archenteron, which will later form the gut; in the protostomes, the opening of the archenteron archetypically forms the mouth (protostome = primary mouth). In the deuterostomes, it forms the anus while the mouth will be derived from a new opening (deuterostome = secondary mouth). Both of these monophyletic groups are well supported by molecular evidence. The monophyly of the deuterostomes (Chordata, Hemichordata and Echinodermata) was recognised at the beginning of the 20th century on the basis of similarities of the very earliest embryonic stages including the fate of the blastopore. Deuterostomes share a radial pattern of early cell division in the embryo and the mesoderm and coelomic cavities originate from outpocketings of the archenteron (enterocoely). Confusingly, deuterostomy and radial cleavage are seen in many protostomes (see below) and seem likely to be a primitive characteristic of the Bilateria, rather than a defining trait of the deuterostomes. Enterocoely too is present in some but not all deuterostomes and also in various protostomes. One characteristic that does seem to be restricted to the deuterostomes is the presence of gill slits, which are found in chordates, hemichordates and certain fossil echinoderms. The homology of hemichordate and chordate gill slits, at least, is supported by conserved expression of developmental genes. Within the deuterostomes, the monophyletic chordates — Vertebrata, Urochordata (sea squirts) and Cephalochordata (amphiouxus/lancelet) — share a notochord, dorsal nerve chord, somites and hypophysis. The dorsal position of the nervous system may be the best synapomorphy (shared derived character) for the chordates as it contrasts with the ventral nerve chord seen in almost all other animals and seems likely to have originated through inversion of the dorso-ventral axis of a chordate ancestor. Within the chordates, the urochordates are the sister group of the vertebrates; this recently discovered relationship is contrary to the prevailing wisdom, which viewed the fish-like cephalochordates as more likely close relatives of the vertebrates. The other group within the deuterostomes is the Ambulacraria, which consists of Hemichordata (e.g. acorn worms) and Echinodermata (e.g. starfish and sea urchins). This close affinity of these two groups whose adults are far from similar was also long ago recognised on the basis of strikingly similar larvae (Tornaria) in members of both phyla. The chordate-like body and gill slits of the adult hemichordates, along with presumed homologs of the notochord and dorsal nerve chord, had led many to consider them as proto-chordates. However, the emphasis on hemichordate and echinoderm larval similarities has been resoundingly supported by molecular data. The sister group of the Ambulacraria is the worm Xenoturbella, formerly linked to the flatworms and, due to dietary contamination rather than to any similarity, to the bivalve molluscs. Xenoturbella is likely to be secondarily simplified, as it has no organs or brain; however, it shares with the Ambulacaria a diffuse, non-centralised basiepithelial nervous system. Initial molecular support for a monophyletic protostome clade contradicted the widely held belief that the Protostomia were a paraphyletic group with some of the ‘simpler’ forms (e.g. flatworms and nematodes) being early branches on the tree diverging before the split separating coelomate protostomes (e.g. molluscs, annelids, arthropods) from the deuterostomes. Although phylogenomic analyses have by no means reached all of the 20 or so protostomian phyla, those that are represented do clearly form a monophyletic group. Additional evidence for the monophyly of Protostomia comes from numerous conserved amino acid signatures within their mitochondrial NADH dehydrogenase subunit 5 (nad5) genes. This robust nad5 character extends to many protostomian phyla not covered by phylogenomic sampling, including some formerly considered to be deuterostomes. Two taxa in particular were long associated with the deuterostomes due, among other characters, to the fate of their blastopore: the Chaetognatha or arrow worms and the lophophorates including Brachiopoda (lamp shells) and Phoronida. All three of these phyla have been shown by various means to be bona fide protostomes. The presence within the protostomes of deuterostomous phyla leads to the paradoxical observation that by no means all protostomes actually form their mouth by protostomy. In many protostomes the blastoporal opening constricts laterally leaving openings at either end, one of which forms the mouth and the other the anus — a situation known as amphistomy. Of greater concern for the definition of the protostomes are numerous protostome taxa — not just chaetognaths and lophophorates — that form just the anus from the blastopore. Clearly another name is desirable for the Protostomia but a good morphological synapomorphy on which to base it is lacking. Within the protostomes two principal clades have been recognised primarily on the basis of analysis of SSU and LSU rRNA: the Lophotrochozoa and the Ecdysozoa. The first part of the name Lophotrochozoa derives from members of the group — the brachiopods and phoronids — that have a ciliated feeding organ called a lophophore. A very similar, but presumably convergently evolved, structure exists in the pterobranch hemichordates which are deuterostomes. In addition to deuterostomy, this character was an important reason for thinking that lophophorates were deuterostomes. The second part of the name Lophotrochozoa comes from other phyla — Mollusca, Annelida, Platyhelminthes (flatworms) and the less familiar Echiura, Sipunculida and Nemertea — which have a characteristic larval stage called a trochophore. These trochozoan phyla also show spiral cleavage in early embryos and are known alternatively as the Spiralia. In spiral cleavage, cell divisions are angled alternatively to the left and right of the primary axis of the embryo. None of the lophotrochozoan phyla (with the possible exception of the Entoprocta) has both defining characters meaning that there is no single morphological characteristic uniting the Lophotrochozoa; there are even several lophotrochozoan phyla with neither lophophore nor trochophore larva. The monophyly of the Lophotrochozoa is, however, well supported by molecular data although several of the constituent phyla are poorly sampled (SSU and LSU and mitochondrial genomes only). Within the Lophotrochozoa, molecular analyses have not yet proved capable of resolving the relationships between most phyla; certain relationships are, nevertheless, well supported. Two phyla, Echiura (common name ‘spoon worms’ or, more quaintly, ‘fat inn-keeper worms’) and Sipuncula (‘peanut worms’) have been grouped within the annelid radiation by molecular phylogenetic analyses and both have been shown to have mitochondrial genomes with an arrangement of genes that is strikingly similar to that of certain annelids. The annelidan affinities of the Echiura are further strengthened by recent work demonstrating a metameric or segmented arrangement of their nervous system as is typical of the annelids. A metameric nervous system has also been demonstrated for the Myzostomida — tiny ten-legged parasites found on various echinoderms which also posses a typically annelidan larva. The morphologically diverse mollusc classes (compare an octopus and a mussel to grasp how diverse they are) form a monophyletic group as do the lophophorate brachiopods and phoronids; phoronids can probably best be considered as shell-less brachiopods. The trochophore type larva and associated spiral cleavage are common within the Lophotrochozoa and so there is a question as to whether those phyla that possess it form a monophyletic group which might be termed the Spiralia or Trochozoa. Molecular studies using SSU and LSU do not support this idea; Mollusca, Annelida, Echiura and Sipunculida seem closely related, yet there are three other groups, the Platyhelminthes, Entoprocta and Nemertea that have spiralian cleavage and a trochophore larva, whose close relationship to the principle spiralian phyla is in doubt. The position of the Nemertea within the Lophoptrochozoa is uncertain, but the Platyhelminthes (flatworms) and Entoprocta have been specifically grouped outside the Spiralia in a clade termed the ‘Platyzoa’ which might be thought of as including the odds and ends of the Lophotrochozoa — the Rotifera/Acanthocephala, Gnathostomulida, Cycliophora, Entoprocta, Ectoprocta and, according to some, the Myzostomida. The preponderance of taxa showing great phylogenetic distances in the Platyzoa is of some concern, suggesting that support for this group may result from systematic errors in phylogeny reconstruction. A final intriguing question of lophotrochozoan relationships is the position of the lophophorates. One suggestion comes from a group of Cambrian fossils called halkieriids which have been interpreted either as containing a mixture of brachiopod and annelid characteristics or as being related to the molluscs. The former interpretation suggests that the halkieriids are in the stem lineage of a monophyletic group consisting of annelids and brachiopods. This would indicate that these phyla are grouped to the exclusion of the molluscs, raising the interesting proposition that spiral cleavage and trochophore larva ware secondarily lost or modified in the brachiopods. Overall, due to limited sampling, possibly compounded by a rapid radiation, the relationships between most lophotrochozoan phyla remain uncertain. The Ecdysozoa are named after the periodic moulting or ecdysis found in all phyla within the group. By far the largest and best-known ecdysozoan phylum is the Euarthropoda comprising insects, crustaceans, myriapods (millipedes and centipedes) and chelicerates (horeseshoe crabs and arachnids). The Euarthropoda can almost certainly be further grouped with the Onychophora (velvet worms) and Tardigrada in the Arthropoda, sharing segmentation, appendages with claws and a haemocoel. The remaining ecdysozoans have been termed the Introverta and consist of 5 phyla of introvert or proboscis bearing worms; Priapulida, Kinorhyncha, Loricifera, Nematoda and Nematomorpha. The priapulids, kinorhynchs and loriciferans may be closely related (Scalidophora) as may be the nematodes and nematomorphs (Nematoidea), but, apart from the nematodes and a priapulid, these phyla are very poorly sampled in terms of molecular data and their relationships are uncertain. The ecdysozoan phyla that have been sampled are well supported as a monophyletic group in phylogenomic analyses once the problem of rapid evolution of the nematode sequences has been addressed. The observation that the order of genes in the mitochondrial genome of the priapulid Priapulus caudatus is almost identical to that of most arthropods is another powerful indication of the monophyly of the Ecdysozoa, which is also supported by the shared secondary absence of large numbers of genes in euarthropods and nematodes. The relationships of the euarthropod classes have been controversial for many decades. One broadly accepted aspect of their relationships, however, was the close relationship between insects and myriapods: both groups lack a second antenna and both possess malpighian tubules, tracheal breathing and unbranched (uniramous) legs. It is now clear that even this relationship is wrong and it is now widely accepted that the closest relatives of the insects are the crustaceans and not the myriapods. It is likely that the insects are actually nested within the Crustacea and as such should be regarded as terrestrial crustaceans although there is no agreement as to where they sit. The position of Pancrustacea or Tetraconata (insects plus crustaceans) relative to the myriapods and the chelicerates is unclear. The shared characteristic of a mandible suggests that myriapods group with the Tetraconata in a clade named the Mandibulata, yet some molecular analyses find limited support for a surprising alliance of myriapods and chelicerates in a grouping that has been named the Paradoxopoda or the Myriochelata. There remain a few animal taxa that we cannot confidently place into any of the major groups mentioned above. Two of these rather obscure groups are most likely to be located amongst the diploblasts: the placozoan Trichoplax adhaerens, which lacks any axis of symmetry and has just four distinguishable cell types; and the myxozoans which were once thought to be unicellular protists but are now recognised as highly degenerate animals. Characteristics of the Trichoplax mitochondrial genome sequence shared with non-metazoans suggest that Trichoplax is very basal amongst the diploblasts and phylogenetic analysis of the mitochondrial protein coding gene sequences similarly suggests it branches before the Porifera. However, Trichoplax has epithelial cells connected by belt desmosomes and septate junctions, characters missing in the sponges and perhaps calling into question this early divergence from the metazoan lineage. The myxozoans have been shown by molecular phylogenetic evidence to have animal affinities. This conclusion has been strengthened by the discovery that one species of myxozoan, Tetracapsula bryozoides, has a multicellular worm-like stage previously classified as a separate species called Buddenbrockia plumatellae. The existence of cells within the myxozoans closely resembling cnidarian nematocysts (stinging cells) in terms of morphology and ontogeny, suggests that they may well be highly degenerate cnidarians. Also generally placed as an early branch within the Metazoa, the dicyemid and orthonectid mesozoans may have been grouped simply on the common basis of their morphological simplicity. Their mitochondrial genetic code shows that the dicyemids at least are triploblasts. They share features of one of their Hox genes with the lophotrochozoans and so seem likely to be very divergent members of this group. The position of the orthonectids is unknown. The Acoelomorpha comprises two related groups of worms, the acoels and the nemertodermatids, previously grouped with the flatworms. Several lines of molecular evidence have suggested that this association is incorrect and that they diverged early from all other triploblasts. However, each one of these analyses may have been affected by the rapid evolution of the acoelomorph sequences which would tend systematically to attract them artefactually towards the base of the tree. Finally, two independent phylogenomic analyses have shown that the chaetognaths are certainly protostomes and not deuterostomes although there is no consensus as to whether they are lophotrochozoans or diverged before the lophotrochozoan/ecdysozoan split. Two analyses of their complete mitochondrial genomes were similarly in disagreement but their protostomian nad5 signature at least is unambiguous. A complete understanding of the relationships of the animals is a very laudable goal in itself but the animal phylogeny also functions as a framework to further our understanding of the evolution of the animals. At least three active fields of research depend upon the new animal phylogeny; evo-devo which attempts to find the basis of changes in morphology plotted over the phylogeny; molecular clocks and the timing of divergences within the animals; and the reconstruction of features of long extinct common ancestors through the identification of homologous characters in its descendants. The most prominent ‘Ur-animal’ in this last category is Urbilateria, the last common ancestor of the protostomes and the deuterostomes. The logic is simply that any homologous character found in both protostomes and deuterostomes must have been inherited from Urbilateria. The problem then is to identify homologous characters and this involves the search for complex similarities, usually the involvement of orthologous patterning genes in the ontogeny of the character of interest. Likely characteristics of Urbilateria identified to date are an antero-posterior axis patterned by Hox genes, photoreceptors patterned by a pax6 ortholog and, more controversially, a pumping heart and a segmented body. Finally, one thing that is obvious from the above discussion is that many nodes of the tree are unresolved. The relationships of the diploblasts, and those within the Ecdysozoa and Lophotrochozoa are very poorly understood. One important reason for this is that most of these phyla are very poorly sampled in terms of molecular characters and so the use of phylogenomic scale datasets is a very encouraging trend in this respect. As a note of caution, however, the continued failure to resolve the position of the chaetognaths with two respectably sized EST data sets shows that this approach is not a guarantee of success.