Inspired by the emerging new view of animal phylogeny, recent comparative morphological studies have applied the range of available new and established microscopical techniques to shed new light on many received wisdoms about animal evolution. Here I review some illustrative recent work of the participants in the symposium “The New Microscopy: Toward a Phylogenetic Synthesis” held at the annual Society for Integrative and Comparative Biology meeting in San Diego in January 2005. I will discuss 6 case studies. Of these, 3 have generated new morphological insights that are significantly at odds with received wisdom about high-level animal phylogeny but congruent with emerging molecular phylogenies. These examples examine the value of characteristics of arthropod neurogenesis in supporting a controversial new hypothesis of high-level arthropod phylogeny, the insights into annelid phylogeny provided by studies on central nervous system development in clitellates, and how new data on the source of mesoderm in phoronids corroborates their placement in the Lophotrochozoa. The remaining 3 case studies shed new light particularly on the evolution of important innovations that are at the base of successful clades. The examples deal with the evolution of torsion as a major synapomorphy of the gastropods, illustrate how the origin of gastrulation could be studied in sponges, and show that trochophore larvae may characterize a larger clade of spiralians than hitherto suspected.
Visualizing the unseen
In phylogenetics, traditions are among the most important characters, and all too often they are traditions masquerading as facts
Ghiselin 1997, p 295
The study of morphology needs no excuse. It is the uncontested and irreplaceable documentation of life's diversity in all its manifestations, across life cycles from egg to adult, and across the somatic hierarchy from molecules to colonies of organisms. As such, the value of morphological study is unassailable. However, over the past decade and a half, debate has erupted about morphology's dominion in a particular domain: phylogenetics.
Ever since phylogenetics was born, in the decades immediately following 1859, morphology, including adult anatomy and embryology, has held pride of place as the central source for insights into the pattern of evolution. However, for at least the past decade and a half, there has been a new kid on the block, which has managed to successfully challenge morphology's monopoly as the only arbiter in matters of genealogy. As Balter (1997, p 1032) aptly expressed the matter, morphologists have had to “learn to live with molecular upstarts,” ushering in an era of exciting dialogue between molecules and morphology in the quest to reconstruct the tree of life. However, in a recent paper in the prime journal of our trade, Scotland and colleagues (2003, p 543) attempted to mute this dialogue by claiming that “we disagree that morphology offers any hope for the future to resolve phylogeny at lower or higher taxonomic levels.” Although their critique of the role of morphology in phylogenetics was couched in strong language, upon closer inspection, their arguments lose much of their force (Jenner 2004a; Wiens 2004; Smith and Turner 2005). The importance and future promise of morphology for phylogenetics are left fully intact.
In this article I demonstrate the continuing value of morphology in contributing valuable data for understanding metazoan phylogeny and body plan evolution. In particular, I select examples from the most recent work of various participants in the symposium “The New Microscopy: Toward Phylogenetic Synthesis,” held at the annual Society for Integrative and Comparative Biology meeting in San Diego in January 2005. Strikingly, their important research challenges many widespread or traditional ideas about metazoan evolution.
The research of the symposium participants provides exemplary illustrations of the application of the diversity of available microscopical techniques to all stages of animals' life cycles, from early events in embryogenesis (Maslakova and others 2004a, 2004b; Leys and others 2006; Stollewerk and Chipman 2006) to larval and juvenile development (Page 2006; Schmidt-Rhaesa and Rothe 2006) and adult structures (Hooge and Tyler 2006; Klaus and Schawaroch 2006; Müller 2006). The techniques used range from scanning electron microscopy, transmission electron microscopy, and histological sectioning to confocal laser scanning microscopy (CLSM) and immunocytochemical staining. In particular, the recent use of CLSM in systematic studies has started to generate arresting insights into the intricacies of animal morphology and development in hitherto unattainable detail. The work by Klaus and colleagues (2003; Klaus and Schawaroch 2006) and Hooge and Tyler (2006) provides good examples of the great value of the unparalleled sensitivity of CLSM in facilitating the discovery of many new morphological characteristics that are useful in the systematics and phylogenetics of diverse animal taxa.
I have organized the remainder of this article under 2 headings. First, under “Toward a new animal phylogeny: Congruence between molecules and the new morphology,” I discuss studies that have generated new morphological insights that are significantly at odds with received wisdom about high-level animal phylogeny but are congruent with emerging molecular phylogenies. Second, under “Toward a new understanding of evolutionary novelties,” I discuss several studies that shed new light on the evolution, in particular, of important innovations that are at the base of successful clades.
The reader should note that this article is not intended as a comprehensive review of the relationship between molecules and morphology in metazoan phylogenetics. It merely aims to highlight important new work of the symposium participants as an introduction to their own, more detailed contributions.
Toward a new animal phylogeny: Congruence between molecules and the new morphology
Over the past decade and a half, the study of high-level animal phylogeny and the evolution of animal body plans has reached an unprecedented level of interest and activity. The emergence of a new molecular view of metazoan relationships (Halanych 2004) has necessitated the reconsideration of many established wisdoms. The following examples illustrate the contributions of the new morphology to understanding a surprising new hypothesis about high-level arthropod phylogeny, the divergence of annelid body plans, and the phylogenetic placement of the phoronids.
Arthropod phylogeny: Tentative morphological support for the monophyly of the Paradoxopoda?
Our understanding of high-level arthropod phylogeny has undergone dramatic changes over the past decade. Although unanimous agreement in phylogenetics is typically illusory, until very recently a “traditional” (Nielsen 2001, p 209) or “classical” (Mallatt and others 2004, p 179) view of arthropod phylogeny was widely accepted that grouped Hexapoda and Myriapoda together as Tracheata or Atelocerata. This clade was considered the sister group of the Crustacea to form a clade Mandibulata. The Mandibulata were considered to be the sister group of the Chelicerata. The combination of molecular sequence data and new morphological evidence has changed this view considerably.
The change that has received by far the most attention is the recovery of a close relationship between Hexapoda and Crustacea in a clade Pancrustacea or Tetraconata, either as sister taxa or with the hexapods evolving from within the crustaceans. In addition, Mallatt and colleagues (2004) recently christened a clade Paradoxopoda; Pisani and colleagues (2004) independently named this clade Myriochelata. The name Paradoxopoda reflects the entirely unexpected union of Chelicerata and Myriapoda as sister groups (or with myriapods paraphyletic with respect to Chelicerata; Negrisolo and others 2004). This clade has been found in independent phylogenetic analyses of a variety of mitochondrial and nuclear genes (Mallatt and others 2004; Pisani and others 2004) and was not expected on the basis of morphological evidence.
Inspired by the new molecular phylogenies, Stollewerk and colleagues (2001) performed a remarkable set of studies that compared the mechanisms of neurogenesis in the major arthropod groups, leading to some surprising results. In her studies of neurogenesis in a spider, Stollewerk (2002; Stollewerk and others 2001) found some similarities to neurogenesis in insects and crustaceans but also some striking differences. In insects and crustaceans individual neural stem cells known as neuroblasts can be recognized that will adopt a neural fate during development of the nervous system. In contrast, in spiders no reliable accounts of cells similar to neuroblasts are known, and rather than individual stem cells, it is groups of cells that will adopt the neural fate at a given time.
In extending these studies to myriapods, Dove and Stollewerk (2003) and Kadner and Stollewerk (2004) examined neurogenesis in a diplopod (millipede) and a chilopod (centipede), respectively. Interestingly, the mechanism of neurogenesis in these myriapods is much more similar to that of the spider than to that of insects and crustaceans. This conclusion is strengthened by preliminary data on neurogenesis in another chilopod (Stollewerk and Chipman 2006). The similarities of neurogenesis in myriapods and chelicerates extend from the invagination of groups of neural precursors to the expression patterns of proneural and neurogenic genes. Although the mechanisms of neurogenesis in the nearest relatives of the arthropods need to be studied to rule out the possibility that the shared similarities in myriapod and chelicerate neurogenesis are plesiomorphies (Harzsch and others 2005; Stollewerk and Simpson 2005), these unanticipated similarities in neurogenesis in chelicerates and myriapods provide the first potential morphological support for the surprising new clade Paradoxopoda.
It should be noted that arthropod phylogeny remains far from settled, even with the compilation and analysis of huge combined morphological and molecular data sets. A recent monster analysis by Wheeler and colleagues (2004), for example, included 240 extant taxa sampled from all major arthropod groups that were scored for 808 morphological characters and included more than 2 kb of 18S and 28S rDNA sequence data. This analysis supported monophyly of Pancrustacea and Mandibulata. However, when just 7 fossil taxa were included, support shifted radically to the traditional Tracheata hypothesis, with crustaceans as the sister group to this clade.
Annelid nervous systems and monophyly of Polychaeta
Traditionally, the structure of nervous systems has been considered to be both highly conservative and imbued with great phylogenetic value, and the overall configuration of nervous systems is still considered a useful tool in high-level animal phylogenetics (Breidbach and Kutsch 1995). The annelids provide a good example of the phylogenetic value of nervous system architecture (Orrhage 1995).
Nervous system organization provides some of the most consistent characteristics distinguishing the polychaetes from the clitellates, in particular the position of the cerebral ganglion and the structure of the circumesophageal connectives (Orrhage 1995; Purschke 1999, 2002; Purschke and others 2000; Müller 2004, 2006; Müller and Henning 2004). In polychaetes the cerebral ganglion is typically located in the prostomium, and the circumesophageal connectives are split into a ventral and a dorsal root. In clitellates (both oligochaetes and leeches) the cerebral ganglion is displaced posteriorly into the peristomium or anterior trunk segments, and the circumesophageal connectives are not split into separate ventral and dorsal roots. These findings seem to nicely support recent morphological cladistic analyses that consistently resolve Clitellata as the sister group of the Polychaeta, irrespective of whether the study is performed with pen and paper (Ax 1999) or on the basis of a comprehensive cladistic data matrix (Rouse and Fauchald 1997). However, as Müller's research clearly shows, such a typological interpretation of comparative morphology may be seriously misleading.
On the basis of regeneration studies in several species of clitellates, Müller (2004, 2006) shows that the possession of undivided circumesophageal connectives in clitellates is the result of the fusion of a separate ventral and dorsal root, a situation identical to that in most polychaetes. This suggests that the adult nervous system of clitellates may have been derived from the polychaete configuration, a conclusion supported by Hessling and Westheide (1999), who studied the normal embryology of another clitellate species and showed that the cerebral ganglion is initially developed in an anterior position, similar to the situation in polychaetes, followed by a posterior displacement. These studies show the great value of comparing morphology across entire life cycles, rather than restricting comparative study to presumed comparable semaphoronts (Jenner, 2001, 2004b).
Interestingly, these findings are in agreement with recent molecular phylogenetic evidence that suggests that clitellates have evolved within a paraphyletic “Polychaeta” (Jördens and others 2004; Siddall and others 2004), which conflicts with morphological cladistic analyses. This molecular hypothesis is in agreement with a recently proposed evolutionary scenario based on a detailed study of comparative morphology in the context of ecological niches and potential selection pressures, which explicitly derived the clitellate body plan from an ancestral polychaete body plan (Purschke 1999, 2002; Westheide and others 1999; Purschke and others 2000). As discussed in Jenner (2004c), this congruence between molecular phylogenetics and careful morphological adaptive reasoning, and their conflict with standard morphological cladistic analyses, is an interesting result that can be extended to several other taxa and calls out for further investigation.
Dissolving dichotomies: Phoronid mesoderm formation and the protostome/deuterostome dichotomy
The Protostomia/Deuterostomia dichotomy is widely seen as among the best-defined subdivisions in the animal kingdom (Nielsen 2001). According to this view, a principal distinguishing feature between protostomes and deuterostomes is the source of the mesoderm. In protostomes mesoderm is typically thought to be derived from cells at the edge of the blastopore lip or the 4d cell, including ectomesoderm, whereas in deuterostomes the mesoderm derives from the invaginated archenteron, sometimes in the form of mesoderm (enterocoelic) pouches. The presumed derivation of mesoderm from the archenteron in brachiopods and phoronids has therefore been nominated as an important synapomorphy uniting them with the Deuterostomia (Lüter 2000; Nielsen 2001, 2002). This echoes the close phylogenetic affinities of phoronids and brachiopods with deuterostomes, rather than protostomes, as has been accepted by many workers (for example, Siewing 1976; Salvini-Plawen 1982; Kozloff 1990; Meglitsch and Schram 1991; Ax 2001; Nielsen 2001; Brusca RC and Brusca GJ 2003).
However, until the publication of the important cell lineage study by Freeman and Martindale (2002), mesoderm formation was studied with relatively crude techniques, such as whole mounts of intact embryos or histological sections through gastrulating embryos. A rather different picture of mesoderm origins in phoronids emerges when instead early cleavage stage blastomeres are labeled and studied with CLSM (Freeman and Martindale 2002).
Freeman and Martindale (2002) show that phoronid mesoderm derives from the boundary between ectoderm and endoderm and that the majority of larval muscle cells, which are the chief mesodermal product, are derived from ectomesoderm. Furthermore they note that in contrast to deuterostomes, ectomesoderm is an important source of mesodermal derivatives in spiralian protostomes, where endomesoderm derives from the 4d cell and ectomesoderm derives from the first, second, or third quartet micromeres in a species-specific fashion. This discovery of ectomesoderm as a significant source of larval muscles in phoronids seems to dovetail nicely with insights from molecular phylogenetics that place phoronids (and brachiopods) in close association with the spiralians within the Lophotrochozoa (Halanych 2004).
The phoronids and brachiopods may provide further ammunition for breaching the traditional protostome/deuterostome dichotomy when we properly distinguish the source of mesoderm from mode of coelom formation. As argued in detail in Jenner (2004b), in his important contribution to coelom formation in brachiopods, and using standard scanning electron microscopy and transmission electron microscopy, Lüter (2000) argues that even when real outpouchings from the archenteron may be absent, we could still conclude that enterocoely is present if the archenteral epithelium is the source of the mesoderm (see also Lüter 2004). However, it should be appreciated that source of mesoderm and mode of coelomogenesis are distinct phenomena. Thus, on the basis of the archenteron being the source of mesoderm, Lüter (2000) concluded that the 2 species of brachiopods he studied exhibit enterocoely, which may be indicative of their deuterostome affinities, even though the coeloms develop by schizocoely, that is, the hollowing out of an initially solid mesodermal mass. This mode of coelomogenesis is characteristic of coelomate protostomes, such as annelids, mollusks, and nemerteans.
The hypothesis of the monophyly of the traditionally defined Deuterostomia, including the phoronids and brachiopods, received a final blow with several recent studies into coelomogenesis in brachiopods and phoronids. So-called trimery or archimery, which is the division of the body into 3 regions arranged along the anteroposterior axis, each with its own coelomic compartment(s) (protocoel, mesocoel, metacoel) has been considered to be “perhaps the most obvious characteristic of the deuterostomes” (Nielsen 2001, p 373; 2002). However, Bartolomaeus (2001), Lüter (2000), and Gruhl et al. (2005) have shown that only 2 coelomic cavities are present in phoronids and brachiopods.
Finally, as extensively discussed in Jenner (2004b), the traditional protostome/deuterostome dichotomy is artificially maintained by the widespread application of different criteria for diagnosing the origin of mesoderm in the 2 groups. By focusing on mesoderm specification in protostomes, such as establishing the 4d cell as the source of mesoderm in spiralians, and on morphological differentiation of mesoderm derivatives in deuterostomes, such as the outpouching of coeloms from the archenteron, falsely dichotomous characters and character states have been introduced into phylogenetic analyses that greatly exaggerate the distinctness of protostome and deuterostome mesoderm origins. A more realistic and less biased perspective, and a greater unity of mesoderm origin in protostomes and deuterostomes, is achieved when mesoderm is recognized as a product of inductive processes between ectoderm and endoderm, and by taking heterochrony in the timing of mesoderm origin into account.
In conclusion, the embryological and morphological studies of Martindale and Freeman, Lüter, and Bartolomaeus combine to provide convincing evidence against the traditional hypothesis that separates the protostomes from the deuterostomes, with the phoronids and brachiopods included in the latter. Significantly, these new morphological findings are congruent with the monophyly of the Lophotrochozoa, which groups phoronids and brachiopods together with the other nonmolting protostomes.
Toward a new understanding of evolutionary novelties
Understanding the origin of evolutionary novelties is at the same time among the most compelling and most frustrating topics in evolutionary biology. Recent developments in evo devo and comparative genomics have provided new approaches for studying the origin of novelties, notably through the comparative study of the expression patterns of conserved developmental genes and the inventory of genome content in early branching taxa. The increasing availability of such information recently inspired Stone and Hall (2004) to breathe new life into the concept of “latent homology” (Hall 1999, p 340) and led Conway Morris (2000, 2003a, 2003b) to propose the concept of “inherency.” According to these concepts, comparative information about genome organization and gene expression patterns can be used as a guide in the search for the origin of morphological novelties in early branching taxa that possess the requisite genomic components but lack the fully developed morphological character present in more derived taxa. However, careful comparative morphological studies can shed as much light on the evolutionary origins of important novelties.
The 3 examples discussed here illustrate how detailed comparative studies of animal development can provide new insights into an entrenched hypothesis of body plan evolution (gastropod torsion) and push back in time the evolutionary origin of 2 important novelties at the base of the Metazoa (gastrulation is sponges) and the Trochozoa (a “hidden” trochophore in Nemertea).
Reconnecting ontogeny and phylogeny: Disentangling gastropod torsion
Torsion is a paradoxical character. It has long been considered a key synapomorphy of gastropods (Ponder and Lindberg 1997), many zoologists have discussed the phenomenon from the late 19th century to the present, and all textbooks on invertebrate zoology prominently describe it. Torsion is defined as the 180° counterclockwise rotation of the gastropod visceropallium (shell and secreting mantle epithelium, mantle or pallial cavity with anus, gills, nephridiopores, and so on, and visceral organs) with respect to the cephalopodium (head and foot). Torsion occurs during gastropod development, and it brings the posterior mantle cavity from an initially posterior position to an anterior position located over the head. It is often described as a 2-step process (Brusca RC and Brusca GJ 2003), with the first step quickly rotating the visceropallium about 90°, probably with the aid of muscle contraction, and with the second step occurring over a more protracted period and driven by differential tissue growth.
Importantly, ontogenetic torsion is widely regarded as recapitulating evolutionary torsion, according to which a hypothetical ancestral mollusc (HAM) with a posterior mantle cavity evolved into a protogastropod with a fully rotated visceropallium and anterior mantle cavity. Surely, the many discussions of torsion in the literature and the continued showcasing of torsion in textbooks must reflect our triumphant understanding of this key feature? As the recent work of Louise Page (2002, 2003, 2006) shows, the manifest importance of torsion for gastropod evolution and development is matched only by our failure to properly understand it, both ontogenetically and evolutionarily.
As Page (2003, p 11) pointedly remarked: “speculative literature about ontogenetic torsion and its evolutionary significance has far outstripped empirical observations.” Most literature dedicated to gastropod torsion deals with the adaptive significance of the process, and very few studies have focused on the actual developmental details of torsion. With the notable exceptions of the studies by Wanninger and colleagues (1999, 2000) and Hickman and Hadfield (2001), only Page and her coworkers have attempted to attain a better understanding of ontogenetic torsion. They have now studied torsion in species from various major gastropod clades, including patellogastropods, vetigastropods, heterobranchs, and caenogastropods. Strikingly, the results appear to be significantly at odds with textbook views.
Rather than being a rigidly conserved process, the pattern, mechanisms, and duration of torsion are very variable. Most important, contraction of larval retractor muscles may or may not be involved in torsion. The visceropallium also need not undergo torsion as a single unit. For example, the shell may undergo 180° torsion in a single step, accompanied by just 90° of rotation of the mantle cavity and anus. During the second step of torsion the mantle cavity may expand to extend to the left side, covering the entire dorsal side of the head, with the anus ending up mid-dorsally over the head. Or the shell and anus may rotate only 90° in total, whereas the osphradial sense organ continues to rotate counterclockwise. And in some opisthobranchs, the shell is secreted only in its post-torsional orientation. Furthermore, the mantle cavity may initially develop either on the right side of the larva or at the posterior end.
Amidst all this variation there seems little support for the widespread depiction of torsion as a unitary 180° rotation of all components of the visceropallium. But faced with this diversity of process, Page (2003, 2006) has refocused on what she perceives to be a conserved developmental pattern that was previously unrecognized and that forms the basis for her alternative “asymmetry hypothesis” (2006) for the origin of the gastropods. On the basis of her work on vetigastropods, heterobranchs, and caenogastropods, Page concludes that the conserved stage during gastropod development is the stage in which the shell is fully rotated, and the mantle cavity and anus are located on the right side of the body. Development before and after this stage may differ between taxa.
But what about the HAM with its posterior mantle cavity that must surely serve as the starting point in evolution? One has only to open a random textbook to find this beast, which has remained strikingly similar since E. Ray Lankester originally conceived of it in the late 19th century as a superficially limpetlike gastropod (for example, compare Bowler 1996, fig. 2.6, with modern renditions of the HAM in Brusca RC and Brusca GJ 1990, fig. 54, p 761, “the basic body plan of mollusks”; Campbell 1993, fig. 29.20, “the generalized mollusc;” Ruppert and others 2004, fig. 12.1). However, as Page (2006) points out, evidence for the existence of such a HAM scarcely transcends the realm of fantasy. Instead, our current understanding of molluscan phylogeny suggests that a monoplacophoran-like ancestor, with lateral mantle cavities that joined posteriorly, seems a much more plausible model for the pregastropod ancestor. Consequently, by restricting its mantle cavity to the right side, an organization that recalls the conserved stage in gastropod development may be achieved, without the necessity of posing a largely conjectural ancestor with a posterior mantle cavity.
More studies of gastropod development are evidently needed to test Page's elegant new hypothesis, but her research has sown a fertile seed for revision of the received wisdom of our textbooks.
Recognizing an evolutionary precursor: Gastrulation in sponges
Sponges do not gastrulate, or so textbooks commonly proclaim. For example, in the chapter on sponge development in the popular embryology textbook by Gilbert and Raunio (1997), no mention is made of gastrulation at all (Fell 1997). The same is true for Brusca RC and Brusca GJ (2003) and Ruppert and colleagues (2004), who instead consider gastrulation a synapomorphy of the Eumetazoa, that is, all metazoans excluding Porifera and Placozoa. Similarly, Valentine (2004, p 55) states that only in metazoans other than sponges is the blastula stage followed by gastrulation. Accordingly, Nielsen (2001, p 23, 44) concludes that sponges are at the blastula or blastaea level of organization, which introduces an interesting irony.
This textbook knowledge would perhaps not lead one to suspect that gastrulation was in fact originally defined on the basis of embryological studies in sponges. In 1872, biology's first full-time phylogeneticist and inveterate coiner of new biological jargon defined the gastrula as “a globular or spheroidal, egg-shaped or somewhat elongated round body, which contains an inner cavity with an external opening” (Ernst Haeckel, quoted in Brauckmann and Gilbert 2004, p 6). This definition stipulates that a gastrula has an archenteron, which Haeckel claimed was primitively formed by invagination. In the same paper Haeckel introduced his famous “biogenetic law” and the Gastraea theory, in which he used the invaginated gastrula found in the ontogenies of various animals as the sole evidence for postulating that the Gastraea, or free living gastrula, represented the most recent common ancestor of the Metazoa, including sponges (Ghiselin and Groeben 1997; Hossfeld and Olsson 2003).
As Haeckel conceived of the Gastraea as a free-living organism, it needed an invaginated gut, with a blastoporal mouth for feeding, and as a result Haeckel vehemently refused to accept any alternative theories for the origin of the Metazoa that proposed gastrulation primitively took place by some means other than invagination, such as delamination or ingression. For this reason, Haeckel categorically dismissed the theories for metazoan origins proposed by E. Ray Lankester, Francis Maitland Balfour, and especially Elias Metschnikoff (Ghiselin and Groeben 1997; Brauckmann and Gilbert 2004). Ever since, Haeckel's insistence on invagination as the primitive mode of gastrulation has pressed a heavy stamp on thinking about metazoan evolution.
Haeckel based his initial description of invagination gastrulation in calcarean sponges on what modern zoologists now refer to as the process of metamorphosis that occurs at settlement. During metamorphosis calcarean amphiblastulae settle on the ciliated cells of the anterior pole, which may subsequently invaginate and contribute to the choanocyte chambers of the adult sponge. This process of invagination has been likened to gastrulation, but the homology of this invagination process with eumetazoan gastrulation has not been widely accepted (Leys 2004; Nielsen 2001). Two important reasons for this are that the sponge larva is already a differentiated diploblastic or multilayered organism with distinct cell types and that the homology of the choanocyte chambers with eumetazoan guts is not generally accepted. Thus, because most other sponges do not show any sign of invagination processes, gastrulation is generally considered to be lacking in the Porifera.
However, in contrast to Haeckel's insistence that gastrulation equals invagination, it has long been known that invagination is simply one possible gastrulation mechanism. It is agreed that the key characteristic of gastrulation is the production of a multilayered organism. Invagination is special because it links this with the development of a gut, but this link is by no means universal, and solid, multilayered gastrulas are commonly formed through processes of cell ingression and delamination. Researchers from Metschnikoff to Hyman have emphasized that this is the case, especially in lower metazoans such as cnidarians. These workers, together with Lankester and Balfour, have considered invagination gastrulation to be derived rather than primitive.
Yet, Haeckel's “invaginative dogma” (Brauckmann and Gilbert 2004, p 6) is still influential today, and the central tenets of the Gastraea theory are still accepted by various modern workers (Nielsen 2001; Arendt 2004). Unfortunately, according to modern sponge workers, the myopic focus on invagination as the sole mechanism of gastrulation has prevented the identification of a diversity of possible gastrulation mechanisms in sponges. These mechanisms may produce multilayered sponge larvae via ingression and delamination processes and a recently described and unique mode of hexactinellid gastrulation (Leys and others 2006). These mechanisms may be appreciated as bona fide instances of gastrulation (Leys 2004; Maldonado 2004). Accordingly, Maldonado (2004, p 4, 8) contends that “poriferan gastrulation is not essentially different from that in other invertebrates,” and “gastrulation in Porifera appears to follow recognizable models … taking place immediately after cleavage and before larval differentiation in most cases.” Leys (2004, p 23) states that in sponges “the various mechanisms of reorganizing the cells of the blastula are comparable to what is readily recognized as gastrulation in many cnidarians, and so the formation of the multilayered embryo during embryogenesis in sponges must be construed as gastrulation, and the cell layers must be therefore equated with ectoderm and endoderm that arise during development of other metazoans.”
Clearly, sponges present a diversity of developmental processes that can tentatively be labeled as gastrulation and that produce a bilayered larva. And with the recent description of invagination gastrulation in the demosponge Halisarca dujardini (Maldonado and Bergquist 2002; Leys 2004), even the application of Haeckel's restrictive criterion of invagination leaves no doubt that we have been systematically biasing our search for the origins of gastrulation by focusing solely on eumetazoans. Although the comparison of potential sponge gastrulation mechanisms with those of eumetazoans is not without its difficulties, it does appear to be clear that sponges can no longer be ignored in the study of the evolution of gastrulation. This is in line with recent molecular phylogenetic results that strongly suggest a paraphyletic grade of sponges at the base of the metazoan phylogeny (Halanych 2004).
Unveiling an evolutionary vestige: The nemertean trochophore and the paradox of modified basal taxa
In tracing the evolutionary origin of a synapomorphy, we focus especially on conditions observed in basal or early branching lineages. However, because evolutionary change is not restricted to more derived clades, basal branches are far from infallible guides to ancestral character states. The Nemertea provide a good illustration of this problem.
Nemerteans, together with platyhelminths, gnathostomulids, entoprocts, annelids, mollusks, sipunculans, and echiurans, exhibit spiral cleavage, which has long been regarded as a very conserved cleavage mode. With the exception of the gnathostomulids, whose development after the initial cleavage stages remains unknown, these taxa develop ciliated larval forms. These are unambiguously recognized as trochophore larvae in annelids, mollusks, entoprocts, sipunculans, and echiurans, and these 5 taxa are sometimes united into a clade Trochozoa on the basis of their possessing trochophore larvae (Rouse 1999). In contrast, the interpretation of the ciliated larvae of platyhelminths and nemerteans is problematic (Rouse 1999; Nielsen 2001; Jenner 2004b), and these larvae are variously interpreted as modified trochophores or as nontrochophores.
If platyhelminths and nemerteans were nested deeply inside the trochozoan clade, an argument could be made that their ciliated larvae are indeed modified trochophores. However, as reviewed in Jenner (2004b), the phylogenetic position of Platyhelminthes remains unclear, but is most likely outside of Trochozoa, while published analyses indicate that Nemertea is the most likely sister group of Neotrochozoa (Sipuncula, Echiura, Mollusca, Annelida), a clade possessing trochophore larvae. Thus one might expect Nemertea to shed light on the origin of trochophore larvae, in analogy to searching for latent homologues of neural crest cells in the sister group of craniates (Stone and Hall 2004), and mesoderm and bilateral symmetry in the sister group of the Bilateria (Finnerty and others 2004; Martindale and others 2004).
Maslakova and colleagues (2004a, 2004b) studied development of the palaeonemertean Carinoma tremaphoros. The Palaeonemertea is a paraphyletic group of nemerteans at the base of the Nemertea, from which the other major nemertean groups have evolved. Palaeonemerteans develop a ciliated “planuliform” larva that is distinct from a trochophore larva in not having a differentially ciliated prototroch. However, Maslakova and colleagues (2004a, 2004b) discovered a vestigial and hidden prototroch in C. tremaphoros. Although the prototroch of C. tremaphoros is not differentially ciliated, it does show the developmental characteristics of prototrochs in trochozoans, including an approximately equatorial and preoral position, similar cell lineage source, early ciliation with respect to other regions of the larval ectoderm, cleavage arrest of the prototroch cells, and ultimate degeneration of the prototroch cells. Unlike many “typical” prototrochs that comprise narrow and distinct bands, the prototroch of C. tremaphoros forms a broad uniformly ciliated belt that has also been described for other modified trochophores in several mollusks, annelids, and a sipunculan species (Maslakova and others 2004a). Similar to other prototrochs, the prototroch of C. tremaphoros is probably the first major locomotory organ of the larva.
Maslakova's discovery of this “hidden” trochophore larva in the life cycle of a palaeonemertean may fill “the “gap” in the distribution of trochophore larvae among the Trochozoa … and allows a meaningful comparison between larval development of nemerteans and other trochozoans” (Maslakova and others 2004a, p 225). Despite the fact that the palaeonemertean larva is probably a highly modified trochophore, its presence in a taxon basal to the Trochozoa nevertheless allowed it to be informative about the evolutionary origin of this important larval type. These findings provide the first important empirical support for a direct comparison of what were previously labeled the “direct development” of nemerteans and the “indirect development” of the trochozoans.
Conclusion: The evolutionary plasticity of animal body plans and finding unity among diversity
The new animal phylogeny forcefully conveys the extensive plasticity of animal body plans. The work of the participants in the symposium “The New Microscopy: Toward a Phylogenetic Synthesis” provides clear illustrations. Nervous systems are often considered to be particularly refractory to evolutionary change, but the work of Stollewerk and colleagues on arthropod neurogenesis and of Müller on annelid nervous system development reveals a diversity of developmental processes underlying apparently conserved adult morphology. The work of Leys shows that almost as soon as animal multicellularity was established, the hexactinellids started to experiment with syncytial organization. Moreover, the great diversity of mechanisms in sponges to produce a multilayered larva has caused many to fail to see the forest for the trees, so that the probable homology of these processes to gastrulation mechanisms observed in eumetazoans has remained unexplored. Similar diversity in developmental processes is characteristic of torsion in different gastropods and prototroch development in different spiralians.
It is the challenge of the new microscopy to find the unity hidden underneath this diversity. The works reviewed in this article show how this has been successfully achieved, thereby breaking through entrenched received wisdoms about the evolution of animal body plans. There is, therefore, no doubt that the study of morphology, inspired by the new animal phylogeny, will continue to yield fundamental insights into animal evolution across all taxonomic levels.
I thank Dr. Ruth Ann Dewel for her kind invitation to participate in the symposium and Dr. Rick Grosberg for his hospitality in providing me with space in his laboratory at University of California, Davis. I am grateful to Drs. Matthew Hooge, Louise Page, Andreas Schmidt-Rhaesa, Sally Leys, and Angelika Stollewerk for the kind sharing of their unpublished manuscripts. I thank 3 anonymous referees for their comments on the manuscript.