Current Biology

Review

Phylogenomic Insights into Animal Evolution

Maximilian J. Telford1,*, Graham E. Budd2, and Hervé Philippe3,4
1Department of Genetics, Evolution and Environment, University College London, WC1E 6BT, UK
2Department of Earth Sciences, Palaeobiology, Uppsala University, Villavägen 16, 75236 Uppsala, Sweden
3Centre de Théorisation et de Modélisation de la Biodiversité, Station d’Ecologie Expérimentale du CNRS, USR CNRS 2936 Moulis,

09200, France
4Département de Biochimie, Centre Robert-Cedergren, Université de Montréal, Montréal, Québec, Canada

*Correspondence: m.telford@ucl.ac.uk
http://dx.doi.org/10.1016/j.cub.2015.07.060

Animals make up only a small fraction of the eukaryotic tree of life, yet, from our vantage point as members of
the animal kingdom, the evolution of the bewildering diversity of animal forms is endlessly fascinating. In the
century following the publication of Darwin’s Origin of Species, hypotheses regarding the evolution of the ma-
jor branches of the animal kingdom — their relationships to each other and the evolution of their body plans —
was based on a consideration of the morphological and developmental characteristics of the different animal
groups. This morphology-based approach had many successes but important aspects of the evolutionary
tree remained disputed. In the past three decades, molecular data, most obviously primary sequences of
DNA and proteins, have provided an estimate of animal phylogeny largely independent of the morphological
evolution we would ultimately like to understand. The molecular tree that has evolved over the past three de-
cades has drastically altered our view of animal phylogeny and many aspects of the tree are no longer conten-
tious. The focus of molecular studies on relationships between animal groups means, however, that the disci-
pline has become somewhat divorced from the underlying biology and from the morphological
characteristics whose evolution we aim to understand. Here, we consider what we currently know of animal
phylogeny; what aspects we are still uncertain about and what our improved understanding of animal phy-
logeny can tell us about the evolution of the great diversity of animal life.

Introduction those characters. The intellectual basis for this process has
Our understanding of the evolutionary relationships between been summarised as ‘‘the criterion for a reliable scenario [being]
major animal (metazoan) groups has matured to an extraordinary its inner coherence’’ [3].
degree in the past quarter of a century [1]. This is very largely due There were many different morphology-based schemes for
to the widespread use of computational analysis of molecular relating the animals (and for explaining the origins of their
data. Genomes contain a vast quantity of informative data, and morphology and embryology) but a few significant themes can
molecules have various desirable qualities as phylogenetic be recognised [4]. One major theme was the generally sensible
markers. Not least of these qualities is that most of the informa- idea of a progression from simple to more complex, with a spe-
tive heritable change molecules contain is independent of the as- cial focus on the evolution of the number of developmental tissue
pects of phenotype whose evolutionary paths might interest us. layers (diploblast to triploblast), planes of body symmetry (radial
Perhaps as a result of this relative independence from pheno- to bilateral), organ systems (organized as tissues or as more
type, research publications using molecular data tend to be specialized organs) and guts (blind-ended guts to through guts
rather removed from morphology — the most common finished with mouth and anus). A second major theme concerned the
product of such a study ideally being a more accurate evolu- likely characteristics of the common ancestor of the bilaterally
tionary tree. Here, we aim to begin to bridge this gap by consid- symmetrical animals (‘Urbilateria’): solid like a sponge planula
ering what our improved knowledge of animal phylogeny might larva or an acoelomate platyhelminth (acoeloid/planuloid),
tell us about the patterns and processes of animal evolution. schizocoelomate (forming coelomic body spaces by splitting
the mesoderm, supposed to be typical of protostomes) or enter-
Post-Darwinian Phylogenies ocoelomate (coelom formation by budding from the gut,
The separate treatment of morphology and phylogeny is in strik- supposed to be typical of deuterostomes). Equally important
ing contrast to the standards prevailing for most of the first hun- were different interpretations of primitive and derived character-
dred or so years of animal phylogenetics, starting with Ernst istics of other aspects of embryology, most notably the pattern
Haeckel [2], in which phylogeny generally emerged directly of cleavage (primitively radial or spiral) and the fate of the
from a consideration of morphological evolution — of the likely blastopore becoming either mouth (protostome) or anus
homology and transformations of characters and the scientist’s (deuterostome). Each of these aspects of hypothetical ancestors
conception of how that morphology evolved. This process was has a strong effect on the topology of the tree and hence on
one of reciprocal illumination in which the evolution of pheno- the implied direction of character evolution, because each
typic characters suggested the topology of the tree at the manoeuvres a different modern group, with its particular
same time as the tree implied a certain evolutionary history of body plan and mode of development, towards the base of the

R876 Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved
Current Biology

Review
Figure 1. Current consensus phylogeny.
Gastrotricha Our current best estimate of the phylogenetic re-
Platyhelminthes lationships of major animal phyla. Major clade
names are indicated. Alternative possible posi-
Ectoprocta/Bryozoa tions for groups with the most contentious posi-
Ph
Brachiopoda/Phoronida a tions are indicated by dashed lines. In the case of
Ctenophora and Xenacoelomorpha systematic
Nemertea
biases in evolutionary patterns, in particular rapid
Mollusca evolution, seem a likely cause of the difficulty in
Lophotrochozoa Annelida (sensu lato) positioning the clades. The lack of resolution within
the Lophotrochozoa seems to be plausibly ex-
Entoprocta
plained by a rapid radiation. While Deuterostomia
Cycliophora is a very long accepted clade, support for this
Mesozoa-Orthonectida grouping is weak compared to the support for
Protostomia, suggesting the branch leading to
Mesozoa-Dicyemida Deuterostomia is (at best) short. A short branch
Protostomia Rotifera/Acanthocephala leading to Deuterostomia has the corollary that
Gnathostomulida there is a shorter evolutionary path from Urbilateria
to Urdeuterostomia than from Urbilateria to
Micrognathozoa Urprotostomia and that Urbilateria might be pre-
Chaetognatha dicted to be more deuterostome-like than proto-
stome-like. In any case, the overall phylogenetic
Kinorhyncha
distribution of the characters ‘deuterostomy’ and
Priapulida ‘radial cleavage’ implies Urbilateria shared both
Bilateria Loricifera these canonical deuterostomian characters. Ani-
mal silhouettes from Phylopic (www.phylopic.org),
Nematoda
credits from top: Michelle Site, Mali’o Kodis, Mali’o
Nematomorpha Kodis, uncredited, Scott Hartman, Scott Hartman,
Tardigrada Scott Hartman, Mali’o Kodis, Michelle Site,
Ecdysozoa (Rotifera: Diego Fontaneto, Elisabeth A. Herniou,
Onychophora Chiara Boschetti, Manuela Caprioli, Giulio Melone,
Eumetazoa Arthropoda Claudia Ricci, and Timothy G. Barraclough, vec-
Deuterostomia
Chordata Vertebrata torized by T. Michael Keesey), Matthew Hooge
(vectorized by T. Michael Keesey), Bam:ejnap
Urochordata (vectorized by T. Michael Keesey), Scott Hartman,
Ambulacraria Cephalochordata Michelle Site, Michelle Site, Mali’o Kodis, Frank
Hemichordata Förster, Eduard Solà Vázquez (vectorized by Yan
Wong), Dinah Challen, Michelle Site, Nicolas
Echinodermata Gompel, T. Michael Keesey, Yan Wong, Mali’o
Xenacoelomorpha Kodis, Mali’o Kodis, (drawing by Manvir Singh),
Hans Hillewaert (photo, T. Michael Keesey, vec-
Cnidaria
torization), Noah Schlottman, uncredited, Noah
Ctenophora Schlottman, Michelle Site, uncredited.
Placozoa
Porifera
Current Biology

bilaterian animals and derives other phyla from this early body 19th century onwards reveal various super-phyletic groups also
plan and ontogeny. seen in the most credible of today’s molecular phylogenies
While this multitude of ideas shows a profound lack of (Figure 1). The most obvious grouping of phyla is the bilaterians
consensus, some aspects of phylogeny and character evolution (all bilaterally symmetrical animals); other currently accepted
seem, nevertheless, to have been accepted. Most significant is groups, such as Ambulacraria (echinoderms and hemichordate
the accurate and, for the most part, unchallenged grouping of worms), Deuterostomia (Ambulacraria and chordates) and Pro-
species into phyla, such as chordates, molluscs, annelid worms tostomia (all bilaterally symmetrical animals excluding deutero-
or arthropods, whatever the subsequent discussion of between- stomes) have also all been initially proposed over 100 years
phylum and within-phylum relationships. A phylum is the most ago. However, other groups that in the past provoked little
inclusive classificatory subdivision within the animal kingdom. dispute we now know are not monophyletic: Articulata
Phyla (like all clades) are characterised by a set of diagnostic incorrectly linked annelids and arthropods and implied that the
characters unique to the group — notochord and dorsal nerve latter is derived from something resembling the former, perhaps
chord for chordates, shell, radula and muscular foot in molluscs. by way of an onychophoran worm [5–7]. And many authors
To an extent, the designation of a phylum is an admission of linked both the chaetognaths and the lophophorates (brachio-
ignorance regarding these higher-level relationships: the body pods and phoronids) to the deuterostomes (chordates,
plan defining a phylum is well defined but relationships to groups echinoderms and hemichordates) because of their shared
with other body plans are less clear. embryological characters such as enterocoely and a pattern of
In truth, of course, some phyla are more closely related to each ‘radial’ cleavage in the early embryo [8] as well as the ciliated
other than others, so a higher level of classification linking sub- feeding structure (lophophore) common to the lophophorates
sets of phyla should be possible. Indeed, further to the largely ac- and the deuterostome hemichordates [9]. It is important to
curate assigning of species to phyla, phylogenies from the late emphasise the success of morphology-based estimates of

Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved R877
 Current Biology

Review

animal relationships — almost all species are assigned to the by stochastic error but turns out to be the main limitation when
correct phylum based purely on morphology — but establishing larger datasets are available. The most famous (and frequently
the relationships between the phyla using morphology and encountered) systematic error in phylogenetics is the long-
embryology has had a more mixed success. branch attraction artefact, whereby fast-evolving branches in
In comparison to morphological and embryological features, the tree are artefactually clustered. A fast evolving lineage
characters encoded in the genome, in addition to being largely (long branch) may either be grouped with a second fast-evolving
independent from phenotypic evolution, have three major ad- lineage or by the long branch (not necessarily fast-evolving) lead-
vantages for reconstructing accurate phylogenetic trees. First ing to a distant outgroup [23]. The problem derives from the fact
is the ease with which one can detect primary homology — the that, while long-branch taxa evolve away from their true sibling
likelihood of convergent evolution of specific sequences of hun- species, the large number of changes along independent long
dreds or thousands of nucleotides or amino acids is negligible branches results in occasional convergent changes in non-sib-
[10]. Second, genes are 100 times more numerous than ling taxa which artificially link them. The artefact is not in fact
codable morphological characters, and each gene contains hun- due to the accelerated rate per se, since branch length is always
dreds or thousands of nucleotide characters. Third, compared to modelled in probabilistic methods, but to the fact that fast line-
the enormous diversity of morphological characters, it is rela- ages often evolve differently, violating the homogeneity hypoth-
tively straightforward to mathematically model the evolution of eses of the model. A second well-known systematic error results
the large sample sizes of genetic characters (4 nucleotides, 20 from cases of heterogeneity of nucleotide or amino acid compo-
amino acids, 64 codons) [11]. These enormous advantages allow sition across lineages and leads to the incorrect grouping of taxa
us to address the two limitations of any statistical inference very sharing the same bias.
efficiently: stochastic error arising from a limited amount of data, Despite many years of progress in improving tree reconstruc-
and systematic errors resulting from incorrect modelling of the tion methods (Box 1), artefacts remain and, once stochastic error
underlying process of change [12]. All these advantages of mo- has been eliminated, violations of the assumptions underlying
lecular data explain why it is preferable to map morphological the models used to infer the tree explain much of the incongru-
characters onto a molecular tree rather than vice versa (see re- ence observed between (and within) studies. Models generally
view by Lee and Palci in this issue). Reconstructing a tree from assume, for example, that all characters in the data set evolve
molecular data is not quite as straightforward as this discussion according to the same rules and that the process of evolution
implies, however. is homogenous across different branches of the tree; these as-
sumptions are likely to be frequently violated. The rest of the
Stochastic and Systematic Errors in Phylogenetic incongruence between studies comes from data errors (e.g. un-
Reconstruction detected contaminations, use of non-orthologous genes or
The first comprehensive efforts to reconstruct the relationships frameshifts affecting predicted amino acids) and high amounts
between animal phyla using molecular data used 5S ribosomal of missing data (Figure 2). In theory, systematic error should be
RNA sequences, but the topologies these analyses produced solvable by improving the model of evolution. However, the
based on just 120 nucleotides [13,14] suffered from major sto- evolutionary process is highly complex, both at the level of the
chastic or sampling errors, ameliorated somewhat by the subse- gene (duplication, horizontal transfer, incomplete lineage sorting
quent use of the 1800 nucleotides of 18S rRNA [1,15–17]. or conversion, all of which conspire to make gene trees different
Stochastic error derives from the use of small samples of molec- from the species tree), and at the level of primary sequences (e.g.
ular data that contain randomly distributed homoplasy (the pres- heterogeneity across positions and over time, mutation/selec-
ence of the same character in distantly related animals due to tion balance, interdependence of sites within and between
convergent evolution). With few nucleotides or amino acids in a genes).
data set, the likelihood of random homoplasies predominating The gene-level problems (apart from the incomplete lineage
over informative sites at certain branches of the tree is high. sorting) can be dealt with through the careful identification and
Larger samples will allow the informative sites to predominate. use of orthologous genes (i.e. genes whose relationships reflect
Sequencing hundreds of nucleotides from many thousands of speciation), a feasible task for animals. For the characters within
genes in a transcriptome or even an entire genome is now sequences, the problem of correctly modelling character evolu-
straightforward and, except for very ancient comparisons, tion depends on the handling of the heterogeneity of the
there are numerous conserved regions that can be unambigu- substitution process across sites and over time. In this context,
ously aligned, typically yielding >100,000 homologous positions CAT-like models which, in addition to the well known site-spe-
[18,19]. Such large data sets drastically reduce most effects cific rates, have site-specific equilibrium frequency profiles (cat-
of stochastic error and contrast with morphological data egories hence ‘CAT’). The CAT model allows different characters
sets, where the biggest matrices contain a few hundred within an alignment to be modelled using different parameters,
characters. meaning, for example, that an amino acid found buried in a cell
While there is a strong psychological bias towards preferring membrane (typically hydrophobic) will not be assumed to evolve
phylogenetic trees with the greatest number of characters according to the same rules as one sticking out into the cyto-
used, the impact of careful taxon sampling makes it clear that plasm (hydrophilic). Such non-homogenous models constitute
the remaining problems of tree reconstruction cannot neces- the most significant recent trend towards improving the accu-
sarily be solved simply by using more characters [20–22]. Sys- racy of phylogenomics [24]. An alternative, simpler approach to
tematic error, where the mode of sequence evolution differs avoiding such errors is the careful selection of taxa, genes and
from our simplified models, was previously generally masked positions, ultimately identifying and discarding the data that

R878 Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved
Current Biology

Review

Box 1. Sources of phylogenetic error.

Phylogenetic reconstruction is susceptible to systematic error whatever the type of character used, but the cause of error has been
best characterised in the use of classical methods based on primary sequences of orthologous genes, typically on a concatenation
of multiple gene alignments (i.e. a super matrix). The evolutionary process is extremely complex, meaning that dealing with sys-
tematic error via model improvement is challenging. The heterogeneity of mutational and selective pressures across time and
across genomes, for example, due to epistasis and heterogeneity of environment, makes the evolution of a given nucleotide or
amino acid position highly site-specific. This makes perfect modelling of sequence evolution an impossible task and forces re-
searchers to focus on modelling heterogeneity that is the most mathematically and computationally accessible. The handling of
the heterogeneity of the substitution process across sites through the CAT-like models, which drastically improve the fit of the
models to data and hence the accuracy of phylogenomics, is the most significant area of recent progress. In contrast, heteroge-
neity of processes over time has been addressed for rate (heterotachy) and global amino acid composition [103] but improvements
have been less marked. The heterogeneity of the substitution process over time (heteropecily) has been shown to bias animal phy-
logeny based on mitochondrial genomes [104] but has not yet been modelled. Each of these diverse model improvements have
generally, for mathematical and computational tractability, been made separately but in the same way that the joint estimate of
alignment and phylogeny is desirable, joint modelling of these factors is ultimately required [105].
Alternative approaches to avoiding tree reconstruction errors that are less technical but perhaps easier to implement are the
careful selection of taxa, genes and positions with genes. The principle here is that, by preferentially discarding the data that
most significantly violate model assumptions (e.g. a fast evolving species with an aberrant nucleotide composition), one can mini-
mise errors [25].
An important question that deserves further research is how to find the best compromise between increasing the number of spe-
cies and increasing model complexity; both these approaches are known to improve accuracy at the cost of increased computa-
tional time. Finally, it should be noted that the incompleteness of numerous phylogenomic matrices (often >50% missing data)
decreases accuracy by reducing the effective number of species [106]. Such a consideration may explain some previously
observed incongruence between studies (Figure 2).

violate model assumptions most significantly, such as a fast in different taxa, which are then tested by their distribution in
evolving species with an aberrant nucleotide composition [25]. the most parsimonious tree(s). Synapomorphies — shared novel
characters that define monophyletic groups — are in this view
Molecular Phylogeny of Animals and Implications for hypotheses of ‘secondary’ homology posterior to the tree anal-
Character Evolution ysis. However, with accurate molecular trees, another approach
Despite problems due to incomplete taxon sampling for some is to map morphological states onto the morphology-indepen-
elusive animals, systematic biases and some nodes remaining dent tree (Box 2). The distribution on such a tree of putatively
difficult to resolve, molecular phylogenetics of animals has (i.e. primarily) homologous morphological characters can also
made a great deal of progress [26]. Our current best estimate establish (secondary) homology, and provide an answer as to
of the tree relating the animal phyla is shown in Figure 1; see whether primary estimates of morphological homologies are
[26] for a recent alternative view. As discussed, some important supported by their distribution on a molecular tree. In both cases,
aspects of the tree, such as the grouping of Bilateria, Deuteros- the relationship between trees and an understanding of pheno-
tomia, Ambulacraria and Protostomia, support conclusions from typic evolution, intimately linked in the pre-molecular era, re-
the study of morphology. A number of other features, however, mains essential if we are interested in understanding how the
diverge from previous ideas of animal relationships. The old diversity of the animals arose. In parallel with the increasing
consensus linked annelids and arthropods (Articulata); in confidence in phylogenetic trees, there is a renaissance in
contrast, molecular trees show that these phyla are found on morphology, with many new techniques becoming available
opposite sides of the most basic division of Protostomia, arthro- that have led to new insights from the morphology side of the
pods are members of the Ecdysozoa alongside nematodes and equation [34–36]. With this in mind, it is interesting to consider
priapulids, while annelids are in the Lophotrochozoa with a number of important conclusions derived from the modern
molluscs and others [20]. The acoelomate Platyhelminthes — phylogenetic consensus regarding character evolution deep
once thought to be one of the earliest branches amongst the within animal evolution and the likely make up of metazoan
Bilateria — are also in the Lophotrochozoa [27]. The lopho- ancestors.
phore-bearing taxa and chaetognaths, previously linked by A Moulting Clade of Arthropods, Nematodes and
many to the deuterostomes, are in fact closer to these protosto- Priapulids
mian taxa [8,28–33]. Each of these phylogenetic discoveries has The existence of a clade of moulting animals, the Ecdysozoa,
had a significant influence on our interpretation of the evolution comprising arthropods, priapulids, nematodes and relatives,
of morphological characters and on the reconstruction of animal and that moulting (ecdysis) is thus a homologous character
ancestors and these new interpretations will be returned to later defining this group are by now well established [20,37,38].
in this discussion The homology of this character in animals as morphologically
In morphology-based analyses, trees are based on hypothe- distinct as nematodes and arthropods was by no means
ses of so-called ‘primary’ homology between character states obvious. Previously, pseudocoelomate nematodes and related

Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved R879
 Current Biology

Review

Porifera Ctenophora
Ctenophora Porifera

Cnidaria Cnidaria

Acoela
Chordata
Chordata Xenoturbellida
Hemichordata
a
Xenoturbellida
Hemichordata Echinodermata
Echinodermata Chaetognatha
Gnathostomulida
Bryozoa
Chaetognatha
Rotifera
Rotifera Gnathostomulida
Acoela
Platyhelminthes Myzostomida
Gastrotricha
Entoprocta
Bryozoa Platyhelminthes
Phoronida
Gastrotricha Entoprocta
Brachiopoda Phoronida
Nemerta
Nemerta Brachiopoda
Chaetopterus
Myzostomida
Annelida
Annelida

Mollusca Mollusca

Priapulida Priapulida
Kinorhyncha Kinorhyncha
Nematoda Nematoda
Nematomorpha Nematomorpha
Tardigrada Tardigrada
Onychophora Onychophora

Arthropoda Arthropoda

Current Biology

Figure 2. Effects of data quality and optimal models on the prevalence of systematic error.
The original dataset is Figure 1 from Dunn et al. [65]. These authors used relatively relaxed parameters for selection of reliably aligned positions from their super
matrix of concatenated genes, which have 21,152 positions and 55.5% of missing data. The resulting data set was analysed using the WAG +G model. Updated
data set: The same data set updated by Philippe et al. [21]. These authors identified some instances of frameshift and some contaminating sequences. They used
more stringent parameters when discarding unreliably aligned positions (producing a super-matrix of 18,463 positions) and used the site heterogenous CAT
model to reconstruct the tree on the data set, which had fewer instances of missing data (35.6%). The effects of these procedures aimed at minimizing the
systematic bias of long-branch attraction can be seen in the different positions for three fast evolving (long branched) taxa highlighted in red. Ctenophores move
from the earliest branching position within Metazoa to a position closer to Cnidaria and Bilateria. Acoela and Myzostomida move from being grouped with other
long-branch taxa to positions with Deuterostomia (Acoela) or Annelida (Myzostomida).

introvert-bearing worms were considered an early branch within reports of spiral cleavage in the lophophorate phoronids [42]
the Bilateria, distant from the coelomate arthropods. Ecdysis of and in entoprocts [43] make sense in the light of their inclusion
an external cuticle and lack of locomotory cilia were thought of in the Lophotrochozoa alongside the canonical spiral cleavers.
as convergently evolved. Moulting, however, turns out to be a This single origin of spiral cleavage in a sub-group of proto-
better indication of relationships than a seemingly complex char- stomes is in contrast to the interpretation of this character in
acter such as segmentation. That this result is not obvious even phylogenetic schemes placing the platyhelminths and nemer-
when knowing the corresponding genetic basis of the character teans at the base of the Bilateria. These acoeloid/planuloid the-
illustrates the limitations of phylogeny based only on morpholog- ories must interpret spiral cleavage either as a primitive bilaterian
ical characters; it is worth noting here that some aspects of the character lost in taxa with radial cleavage or as convergently
new phylogeny, such as a Lophotrochozoa-like clade, were pro- evolved. Of related significance is the recognition that the arthro-
posed as long ago as 1899, and that palaeontologists also sug- pods are not close relatives of the spirally cleaving annelids. The
gested this by the end of the 1980s [5,39]. previously widespread acceptance of the Articulata (annelids
Evolution of Early Embryonic Cleavage Patterns plus arthropods) had led to the interpretation of arthropod early
The spiral arrangement of blastomeres seen when an early cleavage stages as being a derived form of spiral cleavage
cleavage stage embryo is viewed from the animal pole is found [6,7]. While taxa without spiral cleavage show various modes
in several phyla, most notably annelids, molluscs, nemerteans of development, the most common alternative to spiral cleavage
and platyhelminths [40,41]. This pattern is strongly associated is radial cleavage, most firmly associated with the deutero-
with a conserved set of defined blastomere fates. The recogni- stomes. According to its distribution on the new molecular
tion that the Lophotrochozoa are monophyletic — a clade that tree, radial cleavage seems certain to be the primitive character
includes all the phyla with classic spiral cleavage — strongly sup- state of metazoans; in addition to deuterostomes radial cleavage
ports the homology of this mode of early development. Recent is found in diploblasts (poriferans, cnidarians and ctenophores),

R880 Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved
Current Biology

Review

Box 2. Mapping morphological data onto a phylogeny.

condition in which the blastopore is slit-like and closes in the
middle to form both mouth and anus. The recognition that
Mapping morphological characters onto a phylogenetic tree is some phyla once classed as deuterostomes, specifically the
a complex task. Unfortunately, by far the mot common chaetognaths and at least some lophophorates, group with the
method used is maximum parsimony. While parsimony consti- protostomes may suggest that deuterostomy was the primitive
tuted a major advance in evolutionary biology by formalizing condition. The recent discovery of deuterostomy in the ecdyso-
hypothesis comparisons, it is now well established that prob- zoan priapulids [50] and possibly in onychophorans [51] rein-
abilistic methods, even using the simple Mk or threshold forces this view. In reality, there is no clear pattern as in the
models [107,108], outperform parsimony [109]. The obvious protostomes everything from protostomy, through amphistomy
advantages of probabilistic methods are that they take into ac- to deuterostomy can be found in different taxa. It thus seems
count rate heterogeneity across sites and branches (e.g. they that at least the Ecdysozoa are likely to be primitively deuterost-
can account for different branch lengths) and allow multiple omous, radial cleavers, and that this may extend to all of the Pro-
changes on a single branch. Few efforts have been made in tostomia (and thus Bilateria).
developing an accurate model of morphological character
evolution (but see [110]), but this is a crucial step to take full Major Areas of Phylogenetic Controversy
advantage of the now well-established animal phylogeny. In Xenacoelomorpha
the long run one can envision uniting trait evolution models While the precise position of Platyhelminthes (‘flatworms’) within
with more mechanistic insights/knowledge from develop- the Lophotrochozoa is still uncertain, one aspect of platyhel-
mental genetics and in the short run phenomenological minth evolution that has been resolved is the removal from this
models using the existing mathematical tools, allowing great phylum of three groups of marine worms: the single species Xen-
flexibility [111]. oturbella bocki and the two related acoelomorph flatworm
groups the Acoela and the Nemertodermatida. That none of
these three groups is a flatworm is where the current consensus
ecdysozoans (priapulids, as well as in some arthropods with ends. After having been briefly assigned to the molluscs due to
primitive non-yolky eggs and holoblastic cleavage) and chaeto- DNA contamination with their principal food (bivalve molluscs
gnaths [9,44–46]. [52]), Xenoturbella has since been linked to the Ambulacraria
Ancestry of Through Guts (echinoderms and hemichordates) [53]. Several studies, by
Cnidarians and ctenophores both have blind-ended guts, in contrast, place the acoelomorphs as the earliest branch of
which the single opening functions both as mouth and anus. It bilaterians, branching off from the main lineage of before the
is reasonable to suppose (but see [47]) that this is the primitive protostome–deuterostome split [54–56]. More recent work
condition, although if coelenterates (Cnidaria and Ctenophora) groups Xenoturbella with the acoelomorphs (Xenacoelomorpha)
turn out to be monophyletic [48], this character would be unpo- [57,58]; the remaining tension is between those who put xena-
larised — i.e. coelenterates could as easily have lost a through coelomorphs as a branch outside the Protostomia and Deuter-
gut as bilaterians gained one. The widespread presence of a ostomia and those who place them as the sister group of the
through gut with separate mouth and anus in deuterostomes Ambulacraria [49].
and in both major clades of protostomes suggests that the If xenacoelomorphs are an early branching clade intermediate
through gut is an innovation that was present in the common between diploblasts (sponges, cnidarians and ctenophores) and
ancestor of both groups — an ancestor that is synonymous all other bilaterians, their morphological simplicity [47] as well as
with ‘Urbilateria’ unless Xenacoelomorpha branch before this the lack in Xenacoelomorphs of some Hox genes [54,59] and
point (see below). Platyhelminthes, which have a blind gut, are microRNAs [60] present in other Bilateria would result from their
nested within the Lophotrochozoa and so seem most likely to having branched off from the main bilateral lineage before the
have lost a through gut (all their closest relatives have one); the evolution of these characters [57]. A deuterostome affinity, on
same can be said of the Xenacoelomorpha if they are indeed the other hand, would imply that the Xenacoelomorpha are
members of the Deuterostomia rather than an early branch of Bi- simple and lack the characters mentioned not because they
lateria [49]. This inference, along with the observation that acoe- retain primitive features, but because they have lost these char-
lomate, pseudocoelomate and coelomate taxa are intermingled acters, as well as characters typical of deuterostomes such as
in the Deuterostomia, Ecdysozoa and Lophotrochozoa shows gill slits and a through gut through a process of simplification
that the traditional picture of a steady increase in complexity in and loss [18,61].
these characters (from blind gut to through gut or from acoelo- Chaetognatha, Ecdysozoa and Lophotrochozoa
mate via pseudocoelomate to coelomate) is untenable. While lophophorates and chaetognaths, previously believed to
Primitive Deuterostomy? be associated with deuterostomes, have been shown unambig-
The primitive gut or archenteron archetypically forms by an uously to belong to the protostomes, only the lophophorate
ingression of cells when the spherical blastula undergoes gastru- phyla have been positioned more precisely, within the Lophotro-
lation; the opening of the tube that forms during gastrulation is chozoa. The chaetognaths, or ‘arrow worms’, in contrast have
called the blastopore. The fate of the blastopore has a clear sig- been linked by different studies to Ecdysozoa, Lophotrochozoa
nificance in the name of the two major clades of Bilateria — the or as a sister group to both [29–33]. The likely early branching po-
protostomes, in which the blastopore forms the mouth and sition of chaetognaths relative to other protostomes means their
the deuterostomes in which the blastopore forms the anus and deuterostome-like early developmental features (radial cleav-
the mouth is secondary. Amphistomy refers to an intermediate age, deuterostomy and enterocoely) may indicate that such

Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved R881
 Current Biology

Review

traits are primitive within the protostomes and hence primitive metazoan group more distant than sponges to the Eumetazoa
within the Bilateria as a whole. The composition of the Ecdyso- [65,71–73]. This early divergence of ctenophores has been sup-
zoa and Lophotrochozoa is now well established but relation- ported by analyses of the content of ctenophore genomes that
ships between their constituent phyla are much less clear lack many characters present in sponges, cnidarians and
[18,62]. The lack of resolution amongst ecdysozoan phyla is bilaterians [74,75]. Such an early branching placement of cteno-
simpler as there are fewer clades and the Priapulida (and phores would suggest that the characters they have in common
possibly their likely sister taxa the Kinorhyncha and Loricifera) with cnidarians and bilaterians appeared convergently in both
seem likely to be the earliest branch [58]. This would mean the clades. The alternative explanation of independent losses of
‘cycloneuralian’ worm-like phyla of the Ecdysozoa (Priapulida, these morphological characters in both sponges and Trichoplax
Kinorhyncha. Loricifera, Nematoda and Nematomorpha) do not is less parsimonious [76].
form a clade and would imply that the ecdysozoan ancestor While an early branching position of ctenophores has repeat-
possessed the features (most notably a worm-like body and pro- edly been recovered in molecular phylogenetic studies [77],
boscis) shared by these phyla. The phylum-rich Lophotrochozoa the fast evolutionary rate and the low extant taxonomic diversity
are more complex: certain former phyla (Sipunculida, Echiura, of this clade make long-branch attraction artefacts possible
the ectoparasitic Myzostomida and the beard worms Pogono- (Figure 2). Phylogenetic studies addressing the perceived
phora/Vestimentifera) have been subsumed into the phylum problem of long branches leading to the ctenophores have sup-
Annelida, and the Sipunculida and Echiura, unsegmented as ported instead an unresolved multifurcation of cnidarians, cteno-
adults, have been shown to go through a segmented phase as phores, placozoans and bilaterians excluding the poriferans
embryos [63,64]. [21,48]. Further studies are required, and the question of the
The Lophotrochozoa uncontroversially contain the phyla characteristics of the common ancestor of all animals remains
Annelida (sensu lato), Mollusca, Nemertea, Platyhelminthes, open [76,77].
Brachiopoda and Phoronida, as well as several phyla of small an-
imals, such as Gastrotricha, Rotifera, Entoprocta, Cycliophora, Interpreting Patchy Characters
Gnathostomulida and Bryozoa/Ectoprocta. However, there is Characters formerly widely considered homologous and used to
no obvious consensus for the relationships between them. Bra- define clades include the body segments of the (widely
chiopods and phoronids are very likely linked — phoronids may accepted) Articulata and the coelomic cavities of the (less
even be shell-less brachiopods — and molecular phylogenies broadly accepted) Coelomata. In the modern molecular tree,
and recent reports of spiral cleavage link these two phyla to these characters show a patchy phylogenetic distribution, i.e.
the spirally cleaving annelids, molluscs and nemerteans [42]. they are shared by various distantly related taxa. This patchy dis-
The spiral cleavage and trochophora-like Müller’s larva of tribution immediately calls into question the idea that these
some platyhelminths also suggests a link to these phyla. A num- characters are homologous. If segmentation in annelids and
ber of molecular studies, however, group platyhelminths with arthropods is homologous and thus derived from their common
the diminutive phyla mentioned above in a cluster that has ancestor, then it has been lost in multiple other protostome phyla
been called Platyzoa [58,62,65,66]. Most studied members of that also descend from the common ancestor of both ecdysozo-
these phyla, including platyhelminths, are fast evolving (long- ans and lophotrochozoans.
branched), however, and a systematic error is the most plausible A second character with a patchy distribution is the biphasic
explanation for this grouping (Figure 2) [21,67,68]. One inter- life cycle involving a larval stage followed by a metamorphosis
esting interpretation of the difficulty in resolving the relationships into a very different looking adult. Restricting ourselves to the cili-
between the phyla of the Lophotrochozoa is that the lack of res- ated larvae of aquatic taxa (the larva of holometabolous insects
olution is caused by the phyla having diverged from one another is clearly derived, for example, as are life stages of parasites
in a rapid radiation of new body plans. Lack of resolution here such as digenean Platyhelminthes), these seem at least superfi-
argues for the use of the adaptive radiation framework, well cially similar — from the dipleurula of echinoderms and hemi-
developed for recent cases (e.g. cichlid fishes [69]). chordates in the deuterostomes to the trochophore of annelids
Non-Bilaterian Animals and molluscs, the pilidium of nemerteans and the Müller’s larva
Outside the bilaterally symmetric animals, the ctenophores (sea of polyclad Platyhelminthes.
gooseberries/comb jellies) share characteristics with the cnidar- Whether all or even some of these larvae are indeed homolo-
ians and bilaterians including nervous system, true muscle cells gous and so derived from a common ancestor is, however, far
and aspects of their early development [9,70]. These characters from clear [78]. On the one hand, the morphological similarities
are absent in sponges (Porifera) and the single placozoan spe- are obvious, with all the larval types mentioned using bands of
cies Trichoplax adhaerens. Ctenophores were linked in many cilia to swim and having a similar ciliated apical organ [79]. On
morphological phylogenies to the cnidarians in a group called the other hand, many phyla have no ciliated larva and the poten-
the ‘Coelenterata’ (‘sack-guts’, reflecting their blind guts). This tial for convergent evolution of such a useful life-history char-
phylogenetic position of ctenophores would suggest that the acter has been emphasized [80]. The potential to re-use adult
shared characters just mentioned appeared after the branching developmental circuits for patterning larval equivalents might
off of the sponges. The placement of the ctenophores using mo- even produce similar expression patterns of homologous genes
lecular phylogenetic data is another simmering controversy in non-homologous larval organs.
involving another fast-evolving clade: some recent phyloge- There are two related ways in which to think about patchy
nomic studies place ctenophores not with bilaterians and characters. First, the more complex the similarity (and more
cnidarians (collectively Eumetazoa) but as the earliest diverging importantly, the more complex the evolutionary path), the more

R882 Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved
Current Biology

Review

reasonable it becomes to accept homology, despite multiple los- primitive within the Bilateria. Perhaps the adult form of Urbilateria
ses in taxa that lack the character [81]. Second, likelihood of was most similar to a deuterostome as well. Evolution from this
convergence is higher where there is an obvious external reason complex, possibly deuterostome-like ancestor by simplification
for the similarities we find, other than shared ancestry. For provides a straightforward explanation to the conundrum of the
example, the fact that the same adult anterior patterning genes multitude of patchy characters discussed above. However, a
(e.g. Six3, rx, FoxQ2) might pattern (anterior) apical organs in complex Urbilateria does not mean that simpler ancestors did
ambulacrarian and trochophora larvae can be explained as a not exist and that complexity appeared instantaneously. Instead,
convergent re-use of a conserved adult patterning mechanism complexity appeared more or less gradually and, at a given stage,
[82]. In contrast, detailed similarities in the patterning of some- it would have provided a sufficient selective advantage to
thing that cannot have come from an adult (for example, the outcompete the closely related lineages that, had they survived,
same neuropeptide provoking the onset of larval metamor- could have revealed all the intermediate steps. The only access
phosis) are less likely to have evolved through convergence [83]. we now have to these intermediates is through fossils.
Given that all evolution leading to the animal phyla took place
A Complex ‘Urbilateria’ — Integrating Molecular Trees in now extinct stem groups [94], the fossil record might provide
and Fossils help to interpret the trees generated by molecular phylogeny
A major difference between pre-molecular and molecular trees is (and vice versa). Animals with hard parts start to appear in the
that numerous simple organisms had been misplaced and, owing fossil record just before the beginning of the Cambrian period
to their lack of ‘advanced’ characters, incorrectly located deeper some 542 million years ago (see primer by Briggs in this issue).
in the tree, as also observed for unicellular eukaryotes [84]. For While the Cambrian yields a continuous and diverse fossil re-
instance, acoelomates or pseudocoelomates were thought to cord, the fossil record before is discontinuous (i.e. fossil finds
branch closer to the root than more sophisticated bilaterians, rely on rare instances of exceptional preservation) and few
and the simple urochordates (e.g. sea squirts) rather than taxa can be seen to cross into the Cambrian. The fossils from
more fish-like cephalochordates (amphioxus) were incorrectly the Precambrian are generally problematic and no clear
assumed to be the earliest diverging chordate lineage [85]. Mo- consensus exists about their status, even though some have
lecular phylogenies have corrected these misconceptions and been claimed to be animals or even bilaterians and members
show that many seemingly simple animals are deeply nested of crown-group phyla. These problems of interpreting the Pre-
within clades of complex ones (e.g. Myzostomida within annelids cambrian record have led to wide variations in charts that
[86,87], Myxozoa within Cnidaria [88] and Platyhelminthes within attempt to plot animal phylogeny against the fossil record [95].
Lophotrochozoa). An analogous process happens with incom- One of the principal problems of interpreting the early fossil re-
plete or poorly preserved fossils which effectively preferentially cord has been that our search image for ancestors of various
‘lose’ advanced characters during fossilization, causing them to clades has been so poorly constrained. If the stem eumetazoans
be placed lower down in a tree. This systematic error of pre-mo- and stem bilaterians evolved from either benthic or planktonic
lecular phylogenetics is likely due to the strong prejudice, in- larvae, either from adult ancestors actually resembling modern
herited from Aristotle and his ‘great chain of being’, that simple larvae (a classic Haeckelian view) or via heterochronic modes
organisms are ancestral to more complex forms. The number of of evolution such as progenesis (the ‘planula hypothesis’), it is
simple organisms incorrectly placed as early branches in the mo- obvious that predicting the forms of stem lineage fossils
lecular trees may still be underestimated [89]. (or even finding them) will be challenging. This problem is
One implication of this observation is that simplification is a confounded by the fact that fossils on the stem leading to mod-
major driving force in evolution and that the ancestors of extant ern clades will lack a subset of the characters that define that
phyla are likely to have been more complex than previously clade as these characters accumulated gradually along this
thought. The eukaryotic ancestor, for example, appears to stem. Some control exists though. For example, despite the gen-
have had a genome rich in genes and introns to an extent that eral patchiness of the fossil record, some remarkable discov-
would have been unthinkable 20 years ago [90]. Numerous com- eries of microscopic fossils have been made in the Precambrian
plex characters are likely to have been present in Urmetazoa and (e.g. Doushantou Formation from South China), yet none of these
Urbilateria, the ancestors of metazoans and bilaterians, respec- can be confidently assigned to bilaterians [96]. In contrast, the
tively. In particular, numerous genetic networks involved in cell– emerging field of studies of microscopic organic fragments in
cell interactions and in multicellular development are shared by the Cambrian reveals many such fragments that are clearly
all animals (even the simple Trichoplax [91]) and even by the bilaterian in origin [97]. Even if bilaterians were tiny in the Pre-
close relatives of animals (choanoflagellates [92] or ichthyospor- cambrian, they would be capable of being preserved in the
eans). Given the prevalence of simplification during evolution microfossil record, suggesting that their absence is real.
and given its clearly rich gene repertoire, it is reasonable to Uncertainties about the timing of animal origins are exacer-
assume a morphologically complex Urbilateria. bated by the continuing problems in both molecular clock meth-
Another finding of molecular phylogenetics — the shorter odology and how to calibrate them [98,99]. For example, dating
branch leading from Urbilateria to the deuterostomes compared of the split of crown-group bilaterians to about 675 Ma [95] im-
to protostomes, or even the possibility of paraphyletic deutero- plies a gap of something like 140 million years between their
stomes [93] — leads to the intriguing possibility that Urbilateria appearance and their first definitive traces in the fossil record.
was more deuterostome-like than protostome-like. We have These early dates for Urbilateria seem, however, to be incom-
seen already that deuterostomian developmental characters, patible with the view that Urbilateria was complex. One argu-
such as radial cleavage and deuterostomy, are likely to be ment along these lines is that if Urbilateria were complex, then

Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved R883
 Current Biology

Review

Complex Bilateria Figure 3. A simplified order of appearance

trace fossils of major taxa in the fossil record.
(total group Green bars represent (relatively) reliable occur-
bilateria) Protostomia Deuterostomia rences of the clade, thin lines represent implied
‘ghost lineages (no distinction between stem and
Porifera Ctenophora Cnidaria Chaetognatha Arthropoda Chordata Echinodermata
crown-group members). Note that the fossil record
in itself does not necessitate an appearance of
crown group metazoans before about 565 million
years ago (Ma). Sponges: hexactinellid spicules
520 from China. Ctenophores: we accept Eoan-
Cambrian

dromeda as a probable stem-group ctenophore,

otherwise early Cambrian examples from China.
Cnidarians: probably Corumbella and allied forms
from the latest Ediacaran of Brazil. Total group
bilaterians: complex trace fossils from later than
540 Z
560 Ma. Protostomia: total-group chaetognaths
(e.g. from China) from close to the base of the
Cambrian and arthropod trace fossils from a
similar time. Deuterostomia: total-group echino-
derms from around 521 Ma (data from [112]);
W chordates from the Chengjiang biota (e.g. [113]; a
Ediacaran

560 Y tunicate is also described [114]). Several earlier

X fossils imply earlier occurrences of some groups
but their affinities are unclear. A recently published
Doushantuo fossil that may date as early as
600 Ma was described as a sponge [115]. (Key to
580 possible earlier fossils: W, various forms from the
Ediacaran assemblages from c. 565–550 Ma have
been suggested to be sponges; X, trace fossils
Current Biology from the about 565 Ma Mistaken Point assem-
blage may reflect a ‘‘coelenterate’’ grade; Y,
Haootia from Mistaken Point has been suggested
to be a cnidarian [116]; Z, a clear protostome
diversification takes place as reflected in the
‘‘small shelly fossil’’ record [112].

it is likely to have been relatively large as well [94] and thus to of the animals in the fossil record. From this discussion, the pos-
have been capable of leaving either body fossils or, at the very sibility of reciprocal illumination between the fossil record and
least, trace fossils. Despite various controversial earlier findings, molecular phylogenies becomes clear: phylogenies can help
the oldest widely accepted trace fossils that are complex enough direct us to what sort of organism we should be looking for,
necessarily to have been made by bilaterians (either stem or whereas the fossil record can help test such theories.
crown group) are younger than about 560 million years [100].
This is an important time datum that allows us to state quite Conclusion
clearly that by this time at least stem-group sponges, cteno- Many aspects of the animal phylogeny have been established for
phores, cnidarians, placozoans and bilaterians must have diver- some time and backed up with corroborating evidence from
sified. A further corollary of the view that Urbilateria was sources such as unique genomic features. Nevertheless, we
complex, however, is that this time interval also represents the have highlighted a number of issues that remain unresolved. Im-
period when bilaterians were diversifying — if complex bilater- provements in the accuracy of animal phylogeny will, we sug-
ians existed much earlier, they should surely have been able to gest, require a ‘more and less’ approach. More genomic data
leave some sort of evidence, at least as trace fossils. From are required from a broader diversity of species. The aim for
very low diversity in the late Ediacaran (from around 560 million now, however, should not be to build complete trees of tens of
years ago onwards) to the very high diversity of exceptional pres- thousands of species using thousands of genes, but rather to
ervation in the later Cambrian, such as the Sirius Passet and concentrate on ensuring the accuracy of the principal features
Chengjiang biota, it is clear that a very rapid expansion of bilat- of the tree. To achieve this, we should set out to select the least
erian, and incidentally, cnidarian and poriferan, clades took problematic taxa (i.e. slow evolving with minimal systematic
place. This sense of rapidity is reinforced by emphasizing that biases in substitution patterns); to sample taxa with the aim of
there are no even vaguely plausible candidates for early animal breaking long branches; and to select amongst all genes the
fossils before 600 million years ago. A straightforward reading most relevant data (e.g. genes for which we are best able to
of the early fossil record combined with the view above that model their evolutionary properties). In parallel, it is essential
the Urbilateria was complex suggests that the earliest stages that more sophisticated evolutionary models be developed to
of animal evolution took place perhaps around 590–580 million approximate more closely the inherently complex reality of
years ago, and that it was not until around 560 million years genomic evolution.
ago that bilaterians began to emerge (Figure 3). If the ‘mainline’ Our central theme emphasises the use of the increasingly sta-
of animal evolution did in fact take place in large complex adult ble phylogenetic framework of the animal kingdom as the basis
benthic forms [94,101,102], then this means that there is a for understanding the pattern and process of morphological evo-
much better chance of tracing the earliest phylogenetic stages lution. Trees contain information concerning both relationships

R884 Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved
Current Biology

Review

between species and, in the lengths of internal and terminal 15. Field, K.G., Olsen, G.J., Lane, D.J., Giovannoni, S.J., Ghiselin, M.T., Raff,
E.C., Pace, N.R., and Raff, R.A. (1988). Molecular phylogeny of the animal
branches, on the timing of divergences as well as the non-unifor- kingdom. Science 239, 748–753.
mity of genetic evolution. Mapping characters onto accurate
trees is the next step in this process. It should be clear from 16. Field, K.G., Olsen, G.J., Giaovannoni, S.J., Raff, E.C., Pace, N.R., and
Raff, R.A. (1989). Technical comment: Phylogeny and molecular data.
our discussion that the character-mapping step is liable to be Res. Sci. 243, 550–551.
as fraught with problems as has been the construction of
17. Telford, M.J. (2013). The Animal Tree of Life. Science 339, 764–766.
the underlying framework. While determining character state
homology may be helped to an extent by mapping onto an accu- 18. Telford, M.J., and Copley, R.R. (2011). Improving animal phylogenies
rate tree, this is still a far from trivial problem in practice. More with genomic data. Trends Genet. 27, 186–195.
sophisticated methods for mapping characters are necessary 19. Philippe, H., and Telford, M.J. (2006). Large-scale sequencing and the
(Box 2) but the potential for convergent evolution on the one new animal phylogeny. Trends Ecol. Evol. 21, 614–620.
hand and for character loss or character state reversion on the 20. Aguinaldo, A.M., Turbeville, J.M., Linford, L.S., Rivera, M.C., Garey, J.R.,
other is not easily overcome (especially given that, as we have Raff, R.A., and Lake, J.A. (1997). Evidence for a clade of nematodes, ar-
underlined, simplification is an evolutionary driving force). The thropods and other moulting animals. Nature 387, 489–493.
next step of mapping homologous characters onto trees will 21. Philippe, H., Brinkmann, H., Lavrov, D.V., Littlewood, D.T.J., Manuel, M.,
require the collaboration between morphologists, develop- WOrheide, G., and Baurain, D. (2011). Resolving difficult phylogenetic
questions: why more sequences are not enough. PLoS Biol. 9, e1000602.
mental biologists, comparative genomicists, palaeontologists
and phylogeneticists. 22. Jeffroy, O., Brinkmann, H., Delsuc, F., and Philippe, H. (2006). Phyloge-
nomics: the beginning of incongruence? Trends Genet. 22, 225–231.

ACKNOWLEDGEMENTS 23. Felsenstein, J. (1978). Cases in which parsimony or compatibility

methods will be positively misleading. Syst. Zool. 27, 401–410.
The manuscript has been improved by the helpful comments of two referees.
24. Lartillot, N., Brinkmann, H., and Philippe, H. (2007). Suppression of long-
M.J.T. is supported by the European Research Council (ERC-2012-AdG
branch attraction artefacts in the animal phylogeny using a site-heteroge-
322790) and a Royal Society Wolfson Research Merit Award. neous model. BMC Evol. Biol. 7, S4.

25. Philippe, H., and Roure, B. (2011). Difficult phylogenetic questions: more
REFERENCES
data, maybe; better methods, certainly. BMC Biol. 9, 91.

1. Telford, M.J. (2013). Field et al. Redux. EvoDevo 4, 5. 26. Dunn, C.W., Giribet, G., Edgecombe, G.D., and Hejnol, A. (2014). Animal
Phylogeny and Its Evolutionary Implications. Annu. Rev. Ecol. Evol. S 45,
2. Haeckel, E. (1866). Generelle morphologie der organismen (Berlin: Georg 371–395.
Reimer).
27. Balavoine, G. (1997). The early emergence of platyhelminths is contra-
3. Reif, W.E. (2002). Evolution of organ systems: phylogeny, function and dicted by the agreement between 18S rRNA and Hox gene data. C.R.
reciprocal illumination (Senckenbergiana lethaea). Acad. Sci., Paris, Sciences de la Vie 320, 83–94.

4. Willmer, P. (1990). Invertebrate relationships. Patterns in animal evolution 28. Halanych, K.M., Bacheller, J.D., Aguinaldo, A.M., Liva, S.M., Hillis, D.M.,
(Cambridge: Cambridge university press). and Lake, J.A. (1995). Evidence from 18S ribosomal DNA that the lopho-
phorates are protostome animals. Science 267, 1641–1643.
5. Eernisse, D.J., Albert, J.S., and Anderson, F.E. (1992). Annelida and
Arthropoda are not sister taxa: a phylogenetic analysis of spiralian meta- 29. Papillon, D., Perez, Y., Caubit, X., and Le Parco, Y. (2004). Identification
zoan morphology. Syst. Biol. 41, 305–330. of chaetognaths as protostomes is supported by the analysis of their
mitochondrial genome. Mol. Biol. Evol. 21, 2122–2129.
6. Anderson, D.T. (1979). Embryos, fate maps, and the phylogeny of arthro-
pods. In Arthropod Phylogeny, A.P. Gupta, ed. (New York: Van Nostrand 30. Matus, D.Q., Copley, R.R., Dunn, C.W., Hejnol, A., Eccleston, H., Hala-
Reinhold Co.). nych, K.M., Martindale, M.Q., and Telford, M.J. (2006). Broad taxon
and gene sampling indicate that chaetognaths are protostomes. Curr.
7. Anderson, D.T. (1973). Embryology and phylogeny in annelids and Biol. 16, R575–R576.
arthropods (Oxford: Pergammon Press).
31. Marletaz, F., Martin, E., Perez, Y., Papillon, D., Caubit, X., Lowe, C.J.,
8. Telford, M.J., and Holland, P.W. (1993). The phylogenetic affinities of the Freeman, R., Fasano, L., Dossat, C., Wincker, P., et al. (2006). Chaeto-
chaetognaths: a molecular analysis. Mol. Biol. Evol. 10, 660–676. gnath phylogenomics: a protostome with deuterostome-like develop-
ment. Curr. Biol. 16, R577–R578.
9. Nielsen, C. (2011). Animal Evolution. Interrelationships of the living phyla
(Oxford: OUP). 32. Helmkampf, M., Bruchhaus, I., and Hausdorf, B. (2008). Multigene anal-
ysis of lophophorate and chaetognath phylogenetic relationships. Mol.
10. Altschul, S.F., Gish, W., Miller, W., Myers, E.W., and Lipman, D.J. (1990). Phylogenet. Evol. 46, 206–214.
Basic local alignment search tool. J. Mol. Biol. 215, 403–410.
33. Paps, J., Baguna, J., and Riutort, M. (2009). Bilaterian phylogeny: a broad
11. Yang, Z. (2014). Molecular evolution: A statistical approach (Oxford: sampling of 13 nuclear genes provides a new Lophotrochozoa phylogeny
Oxford University Press). and supports a paraphyletic basal acoelomorpha. Mol. Biol. Evol. 26,
2397–2406.
12. Felsenstein, J. (2004). Inferring Phylogenies (Sunderland MA: Sinauer
Associates Inc). 34. Budd, G.E., and Olsson, L. (2007). Editorial: a renaissance for evolu-
tionary morphology. Acta Zool. 88, 1.
13. Ohama, T., Kumazaki, T., Hori, H., and Osawa, S. (1984). Evolution of
multicellular animals as deduced from 5S rRNA sequences: a possible 35. Ziegler, A., Ogurreck, M., Steinke, T., Beckmann, F., Prohaska, S., and
early emergence of the Mesozoa. Nucleic Acids Res. 12, 5101–5108. Ziegler, A. (2010). Opportunities and challenges for digital morphology.
Biol. Direct 5, 45.
14. Hendriks, L., Huysmans, E., Vandenberghe, A., and De Wachter, R.
(1986). Primary structures of the 5S ribosomal RNAs of 11 arthropods 36. Harzsch, S. (2006). Neurophylogeny: Architecture of the nervous system
and applicability of 5S RNA to the study of metazoan evolution. J. Mol. and a fresh view on arthropod phyologeny. Integr. Comp. Biol. 46,
Evol. 24, 103–109. 162–194.

Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved R885
 Current Biology

Review
37. Valentine, J.W., and Collins, A.G. (2000). The significance of moulting in 58. Hejnol, A., Obst, M., Stamatakis, A., Ott, M., Rouse, G.W., Edgecombe,
Ecdysozoan evolution. Evol. Dev. 2, 152–156. G.D., Martinez, P., Baguna, J., Bailly, X., Jondelius, U., et al. (2009). As-
sessing the root of bilaterian animals with scalable phylogenomic
38. Telford, M.J., Bourlat, S.J., Economou, A., Papillon, D., and Rota-Sta- methods. Proc. Biol. Sci. 276, 4261–4270.
belli, O. (2008). The evolution of the Ecdysozoa. Philos. Trans. R. Soc.
Lond. B. Biol. Sci. 363, 1529–1537. 59. Hejnol, A., and Martindale, M.Q. (2009). Coordinated spatial and tempo-
ral expression of Hox genes during embryogenesis in the acoel Convolu-
39. Dzik, J., and Rumbiegel, G. (1989). The oldest ‘‘onychophoran’’ Xenu- triloba longifissura. BMC Biol. 7, 65.
sion: a link connecting phyla? Lethaia 22, 169–181.
60. Sempere, L.F., Martinez, P., Cole, C., Baguñà, J., and Peterson, K.J.
40. Nielsen, C. (2005). Trochophora larvae: cell-lineages, ciliary bands and (2007). Phylogenetic distribution of microRNAs supports the basal posi-
body regions. 2. Other groups and general discussion. J. Exp. Zool. B. tion of acoel flatworms and the polyphyly of Platyhelminthes. Evol. Dev.
Mol. Dev. Evol. 304, 401–447. 9, 409–415.

41. Nielsen, C. (2004). Trochophora larvae: cell-lineages, ciliary bands, and 61. Telford, M.J. (2008). Xenoturbellida: the fourth deuterostome phylum and
body regions. 1. Annelida and Mollusca. J. Exp. Zool. B. Mol. Dev. the diet of worms. Genesis 46, 580–586.
Evol. 302, 35–68.
62. Edgecombe, G.D., Giribet, G., Dunn, C.W., Hejnol, A., Kristensen, R.M.,
42. Pennerstorfer, M., and Scholtz, G. (2012). Early cleavage in Phoronis Neves, R.C., Rouse, G.W., Worsaae, K., and Sørensen, M.V. (2011).
muelleri (Phoronida) displays spiral features. Evol. Dev. 484–500. Higher-level metazoan relationships: recent progress and remaining
questions. Org. Divers. Evol. 11, 151–172.
43. Merkel, J., Wollesen, T., Lieb, B., and Wanninger, A. (2012). Spiral cleav-
age and early embryology of a loxosomatid entoproct and the usefulness 63. Brinkmann, N., and Wanninger, A. (2010). Capitellid connections: contri-
of spiralian apical cross patterns for phylogenetic inferences. BMC Dev. butions from neuromuscular development of the maldanid polychaete
Biol. 12, 11. Axiothella rubrocincta (Annelida). BMC Evol. Biol. 10, 168.

44. Wennberg, S.A., Janssen, R., and Budd, G.E. (2008). Early embryonic 64. Kristof, A., Wollesen, T., and Wanninger, A. (2008). Segmental mode of
development of the priapulid worm Priapulus caudatus. Evol. Dev. 10, neural patterning in Sipuncula. Curr. Biol. 18, 1129–1132.
326–338.
65. Dunn, C.W., Hejnol, A., Matus, D.Q., Pang, K., Browne, W.E., Smith, S.A.,
45. Nielsen, C. (2001). Animal Evolution. Interrelationships of the living phyla, Seaver, E., Rouse, G.W., Obst, M., Edgecombe, G.D., et al. (2008). Broad
2nd Edition (Oxford: O.U.P.). phylogenomic sampling improves resolution of the animal tree of life.
Nature 452, 745–749.
46. Hertzler, P.L., and Clark, W.H. (1992). Cleavage and gastrulation in the
66. Giribet, G., Distel, D.L., Polz, M., Sterrer, W., and Wheeler, W.C. (2000).
shrimp Sicyonia ingentis: invagination accompanied by oriented cell divi-
Triploblastic relationships with emphasis on the acoelomates and the po-
sion. Development 116, 127–140.
sition of Gnathostomulida, Cycliophora, Plathelminthes, and Chaetogna-
tha: a combined approach of 18S rDNA sequences and morphology.
47. Hejnol, A., and Martindale, M.Q. (2008). Acoel development indicates the
independent evolution of the bilaterian mouth and anus. Nature 456, Syst. Biol. 49, 539–562.
382–386.
67. Struck, T.H., Wey-Fabrizius, A.R., Golombek, A., Hering, L., Weigert, A.,
Bleidorn, C., Klebow, S., Iakovenko, N., Hausdorf, B., Petersen, M., et al.
48. Pick, K.S., Philippe, H., Schreiber, F., Erpenbeck, D., Jackson, D.J.,
(2014). Platyzoan paraphyly based on phylogenomic data supports a
Wrede, P., Wiens, M., Alie, A., Morgenstern, B., Manuel, M., et al.
noncoelomate ancestry of Spiralia. Mol. Biol. Evol. 31, 1833–1849.
(2010). Improved phylogenomic taxon sampling noticeably affects non-
bilaterian relationships. Mol. Biol. Evol. 27, 1983–1987. 68. Egger, B., Lapraz, F., Tomiczek, B., Muller, S., Dessimoz, C., Girstmair,
J., Skunca, N., Rawlinson, K.A., Cameron, C.B., Beli, E., et al. (2015).
49. Philippe, H., Brinkmann, H., Copley, R.R., Moroz, L.L., Nakano, H., A transcriptomic-phylogenomic analysis of the evolutionary relationships
Poustka, A.J., Wallberg, A., Peterson, K.J., and Telford, M.J. (2011). of flatworms. Curr. Biol. 25, 1347–1353.
Acoelomorph flatworms are deuterostomes related to Xenoturbella.
Nature 470, 255–258. 69. Fan, S., Elmer, K.R., and Meyer, A. (2012). Genomics of adaptation and
speciation in cichlid fishes: recent advances and analyses in African
50. Martin-Duran, J.M., Janssen, R., Wennberg, S., Budd, G.E., and Hejnol, and Neotropical lineages. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 367,
A. (2012). Deuterostomic development in the protostome Priapulus cau- 385–394.
datus. Curr. Biol. 22, 2161–2166.
70. Scholtz, G. (2004). Coelenterata versus Acrosomata - zur Position der
51. Janssen, R., Jorgensen, M., Lagebro, L., and Budd, G.E. (2015). Fate and Rippenquallen (Ctenophora) im phylogenetischen System der Metazoa.
nature of the onychophoran mouth-anus furrow and its contribution to In Kontroversen in der Phylogenetischen Systematik, Volume 43, S.
the blastopore. Proc. Biol. Sci. 282, 20142628. Richter and W. Sudhaus, eds. (Berlin: Sber. Ges. Naturf. Freunde),
pp. 15–33.
52. Bourlat, S.J., Nielsen, C., Lockyer, A.E., Littlewood, D.T., and Telford,
M.J. (2003). Xenoturbella is a deuterostome that eats molluscs. Nature 71. Ryan, J.F., Pang, K., Schnitzler, C.E., Nguyen, A.D., Moreland, R.T., Sim-
424, 925–928. mons, D.K., Koch, B.J., Francis, W.R., Havlak, P., Program, N.C.S., et al.
(2013). The genome of the ctenophore Mnemiopsis leidyi and its implica-
53. Bourlat, S.J., Juliusdottir, T., Lowe, C.J., Freeman, R., Aronowicz, J., tions for cell type evolution. Science 342, 1242592.
Kirschner, M., Lander, E.S., Thorndyke, M., Nakano, H., Kohn, A.B.,
et al. (2006). Deuterostome phylogeny reveals monophyletic chordates 72. Moroz, L.L., Kocot, K.M., Citarella, M.R., Dosung, S., Norekian, T.P., Po-
and the new phylum Xenoturbellida. Nature 444, 85–88. volotskaya, I.S., Grigorenko, A.P., Dailey, C., Berezikov, E., Buckley,
K.M., et al. (2014). The ctenophore genome and the evolutionary origins
54. Cook, C.E., Jimenez, E., and Akam, M.E. (2004). The Hox gene comple- of neural systems. Nature 510, 109–114.
ment of acoel flatworms, a basal bilaterian clade. Evol. Dev. 6, 154–163.
73. Whelan, N.V., Kocot, K.M., Moroz, L.L., and Halanych, K.M. (2015). Error,
55. Telford, M.J., Lockyer, A.E., Finch, C.C., and Littlewood, D.T.J. (2003). signal, and the placement of Ctenophora sister to all other animals. Proc.
Combined large and small subunit ribosomal RNA phylogenies support Natl. Acad. Sci. USA 112, 5773–5778.
a basal position of the acoelomorph flatworms. Proc. R. Soc. B. Biol.
Sci. 270, 1077–1083. 74. Ryan, J.F., Pang, K., Mullikin, J.C., Martindale, M.Q., Baxevanis, A.D.,
and Progra, N.C.S. (2010). The homeodomain complement of the cteno-
56. Ruiz Trillo, I., Riutort, M., Littlewood, D.T.J., Herniou, E.A., and Baguñà, J. phore Mnemiopsis leidyi suggests that Ctenophora and Porifera diverged
(1999). Acoel flatworms: earliest extant bilaterian metazoans, not mem- prior to the ParaHoxozoa. Evodevo 1, 10.
bers of Platyhelminthes. Science 283, 1919–1923.
75. Pang, K., Ryan, J.F., Baxevanis, A.D., and Martindale, M.Q. (2011). Evo-
57. Nielsen, C. (2010). After all: Xenoturbella is an acoelomorph! Evol. Dev. lution of the TGF-beta signaling pathway and its potential role in the
12, 241–243. ctenophore, Mnemiopsis leidyi. PLoS One 6, e24152.

R886 Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved
Current Biology

Review
76. Jekely, G., Paps, J., and Nielsen, C. (2015). The phylogenetic position of 96. Bengtson, S., and Budd, G.E. (2004). Comment on ‘‘Small Bilaterian Fos-
ctenophores and the origin(s) of nervous systems. EvoDevo 6, 1. sils from 40 to 55 Million Years Before the Cambrian’’. Science 306, 1291.

77. Marlow, H., and Arendt, D. (2014). Evolution: ctenophore genomes and 97. Harvey, T.H., Velez, M.I., and Butterfield, N.J. (2012). Exceptionally pre-
the origin of neurons. Curr. Biol. 24, R757–R761. served crustaceans from western Canada reveal a cryptic Cambrian ra-
diation. Proc. Natl. Acad. Sci. USA 109, 1589–1594.
78. Telford, M.J., Lowe, C.J., Cameron, C.B., Ortega-Martinez, O., Arono-
wicz, J., Oliveri, P., and Copley, R.R. (2014). Phylogenomic analysis of 98. Sperling, E.A., Robinson, J.M., Pisani, D., and Peterson, K.J. (2010).
echinoderm class relationships supports Asterozoa. Proc. Biol. Sci. R. Where’s the glass? Biomarkers, molecular clocks, and microRNAs sug-
Soc. 281, 20140479–20140479. gest a 200-Myr missing Precambrian fossil record of siliceous sponge
spicules. Geobiology 8, 24–36.
79. Nielsen, C. (2009). How did indirect development with planktotrophic
larvae evolve? Biol. Bull. 216, 203–215. 99. Love, G.D., Grosjean, E., Stalvies, C., Fike, D.A., Grotzinger, J.P., Brad-
ley, A.S., Kelly, A.E., Bhatia, M., Meredith, W., Snape, C.E., et al. (2009).
80. Raff, R.A. (2008). Origins of the other metazoan body plans: the evolution Fossil steroids record the appearance of Demospongiae during the
of larval forms. Phil. Trans. R. Soc. Lond. B. 363, 1473–1479. Cryogenian period. Nature 457, 718–721.
81. Hessling, R., and Westhiede, W. (2002). Are Echiura derived from a 100. Martin, M.W., Grazhdankin, D.V., Bowring, S.A., Evans, D.A., Fedonkin,
segmented ancestor? Immunohistochemical analysis of the nervous sys- M.A., and Kirschvink, J.L. (2000). Age of Neoproterozoic bilatarian
tem in developmental stages of Bonellia viridis. J. Morphol. 252, body and trace fossils, White Sea, Russia: implications for metazoan
100–113. evolution. Science 288, 841–845.
82. Hunnekuhl, V.S., and Akam, M. (2014). An anterior medial cell population 101. Dewel, R.A. (2000). Colonial origin for Emetazoa: major morphological
with an apical-organ-like transcriptional profile that pioneers the central transitions and the origin of bilaterian complexity. J. Morphol. 243, 35–74.
nervous system in the centipede Strigamia maritima. Dev. Biol. 396,
136–149. 102. Budd, G.E., and Jensen, L.J. (2015). The origin of the animals and a
‘‘Savannah’’ hypothesis for early bilaterian evolution, in press.
83. Conzelmann, M., Williams, E.A., Tunaru, S., Randel, N., Shahidi, R., Asa-
dulina, A., Berger, J., Offermanns, S., and Jékely, G. (2013). Conserved 103. Groussin, M., Boussau, B., and Gouy, M. (2013). A branch-heteroge-
MIP receptor-ligand pair regulates Platynereis larval settlement. Proc. neous model of protein evolution for efficient inference of ancestral se-
Natl. Acad. Sci. USA 110, 8224–8229. quences. Syst. Biol. 62, 523–538.
84. Philippe, H., Germot, A., and Moreira, D. (2000). The new phylogeny of
104. Roure, B., and Philippe, H. (2011). Site-specific time heterogeneity of the
eukaryotes. Curr. Opin. Genet. Dev. 10, 596–601.
substitution process and its impact on phylogenetic inference. BMC
85. Delsuc, F., Brinkmann, H., Chourrout, D., and Philippe, H. (2006). Tuni- Evol. Biol. 11, 17.
cates and not cephalochordates are the closest living relatives of verte-
105. Blanquart, S., and Lartillot, N. (2008). A site- and time-heterogeneous
brates. Nature 439, 965–968.
model of amino acid replacement. Mol. Biol. Evol. 25, 842–858.
86. Weigert, A., Helm, C., Meyer, M., Nickel, B., Arendt, D., Hausdorf, B.,
106. Roure, B., Baurain, D., and Philippe, H. (2013). Impact of missing data on
Santos, S.R., Halanych, K.M., Purschke, G., Bleidorn, C., et al. (2014).
Illuminating the base of the annelid tree using transcriptomics. Mol. phylogenies inferred from empirical phylogenomic data sets. Mol. Biol.
Evol. 30, 197–214.
Biol. Evol. 31, 1391–1401.

87. Bleidorn, C., Eeckhaut, I., Podsiadlowski, L., Schult, N., McHugh, D., Ha- 107. Lewis, P.O. (2001). A likelihood approach to estimating phylogeny from
lanych, K.M., Milinkovitch, M.C., and Tiedemann, R. (2007). Mitochon- discrete morphological character data. Syst. Biol. 50, 913–925.
drial genome and nuclear sequence data support myzostomida as part
108. Felsenstein, J. (2012). A comparative method for both discrete and
of the annelid radiation. Mol. Biol. Evol. 24, 1690–1701.
continuous characters using the threshold model. Am. Nat. 179,
88. Jiménez-Guri, E., Philippe, H., Okamura, B., and Holland, P. (2007). Bud- 145–156.
denbrockia is a cnidarian worm. Science 317, 116–118.
109. Wright, A.M., and Hillis, D.M. (2014). Bayesian analysis using a simple
89. Philippe, H., and Laurent, J. (1998). How good are deep phylogenetic likelihood model outperforms parsimony for estimation of phylogeny
trees? Curr. Opin. Genet. Dev. 8, 616–623. from discrete morphological data. PLoS One 9, e109210.

90. Fritz-Laylin, L.K., Prochnik, S.E., Ginger, M.L., Dacks, J.B., Carpenter, 110. Revell, L.J. (2013). Two new graphical methods for mapping trait evolu-
M.L., Field, M.C., Kuo, A., Paredez, A., Chapman, J., Pham, J., et al. tion on phylogenies. Methods Ecol. Evol. 4, 754–759.
(2010). The genome of Naegleria gruberi illuminates early eukaryotic
versatility. Cell 140, 631–642. 111. Rodrigue, N., and Philippe, H. (2010). Mechanistic revisions of phenom-
enological modelling strategies in molecular evolution. Trends Genet. 26,
91. Srivastava, M., Begovic, E., Chapman, J., Putnam, N.H., Hellsten, U., Ka- 248–252.
washima, T., Kuo, A., Mitros, T., Salamov, A., Carpenter, M.L., et al.
(2008). The Trichoplax genome and the nature of placozoans. Nature 112. Kouchinsky, A., Bengtson, S., Runnegar, B., Skovsted, C., Steiner, M.,
454, 955–960. and Vendrasco, M. (2012). Chronology of early Cambrian bio-
mineralisation. Geol. Mag. 149, 221–251.
92. King, N., Westbrook, M.J., Young, S.L., Kuo, A., Abedin, M., Chapman,
J., Fairclough, S., Hellsten, U., Isogai, Y., Letunic, I., et al. (2008). The 113. Shu, D.-G., Luo, H.-L., Conway Morris, S., Zhang, X.-L., Hu, S.-X., Chen,
genome of the choanoflagellate Monosiga brevicollis and the origin of L., Han, J., Zhu, M., Li, Y., and Chen, L.-Z. (1999). Lower Cambrian ver-
metazoans. Nature 451, 783–788. tebrates from south China. Nature 402, 42–46.

93. Lartillot, N., and Philippe, H. (2008). Improvement of molecular phyloge- 114. Shu, D.-G., Chen, L., Han, J., and Zhang, X.-L. (2001). An early Cambrian
netic inference and the phylogeny of Bilateria. Philos. Trans. R. Soc. tunicate from China. Nature 411, 472–473.
Lond. B. Biol. Sci. 363, 1463–1472.
115. Yin, Z., ZHhu, M., Davidson, E.H., Bottjer, D.J., Zhao, F., and Tafforeau,
94. Budd, G., and Jensen, S. (2000). A critical reappraisal of the fossil record P. (2015). Sponge grade body fossil with cellular resolution dating 60 Myr
of the bilaterian phyla. Biol. Rev. Camb. Philos. Soc. 75, 253–295. before the Cambrian. Proc. Natl. Acad. Sci. USA 112, E1453–E1460.

95. Erwin, D.H., Laflamme, M., Tweedt, S.M., Sperling, E.A., Pisani, D., and 116. Liu, A.G., Matthews, J.J., Menon, L.R., McIlroy, D., and Brasier, M.D.
Peterson, K.J. (2011). The Cambrian conundrum: early divergence and (2014). Haootia quadriformis n. gen., n. sp., interpreted as a muscular
later ecological success in the early history of animals. Science 334, cnidarian impression from the Late Ediacaran period (approx. 560 Ma).
1091–1097. Proc. R. Soc. Lond. B. 281, 20141202.

Current Biology 25, R876–R887, October 5, 2015 ª2015 Elsevier Ltd All rights reserved R887