Evolution of centralized nervous systems: Two schools of evolutionary thought

Abstract

Understanding the evolution of centralized nervous systems requires an understanding of metazoan phylogenetic interrelationships, their fossil record, the variation in their cephalic neural characters, and the development of these characters. Each of these topics involves comparative approaches, and both cladistic and phenetic methodologies have been applied. Our understanding of metazoan phylogeny has increased greatly with the cladistic analysis of molecular data, and relaxed molecular clocks generally date the origin of bilaterians at 600–700 Mya (during the Ediacaran). Although the taxonomic affinities of the Ediacaran biota remain uncertain, a conservative interpretation suggests that a number of these taxa form clades that are closely related, if not stem clades of bilaterian crown clades. Analysis of brain–body complexity among extant bilaterians indicates that diffuse nerve nets and possibly, ganglionated cephalic neural systems existed in Ediacaran organisms. An outgroup analysis of cephalic neural characters among extant metazoans also indicates that the last common bilaterian ancestor possessed a diffuse nerve plexus and that brains evolved independently at least four times. In contrast, the hypothesis of a tripartite brain, based primarily on phenetic analysis of developmental genetic data, indicates that the brain arose in the last common bilaterian ancestor. Hopefully, this debate will be resolved by cladistic analysis of the genomes of additional taxa and an increased understanding of character identity genetic networks.

The fact that some of these building stones are universal does not, of course, mean that the organs to which they contribute are as old as these molecules or their precursors.

von Salvini-Plawen and Mayr (1)

Any consideration of the evolution of centralized nervous systems is inextricably linked to an understanding of the phylogeny of living metazoans, their fossil history, the vast range of complexity in their nervous systems, and the development of these nervous systems. For this reason, any attempt to reconstruct the phylogeny of metazoan CNSs must be based on all lines of evidence available. The molecular phylogenetic studies of the last 20 y are particularly important in understanding metazoan interrelationships as well as the time frame in which these animals arose and radiated, and we now have increased insights into the genetics underlying the development of CNSs.

First, I will review the fossil history of the earliest putative metazoans, and then, I will discuss different comparative approaches to analyzing both molecular and morphological data: the molecular clock hypothesis, which has yielded a range of possible dates for the origin and divergence of metazoans; developmental genetics and its contribution to our understanding of the patterning of metazoan bodies, particularly patterning of the CNS; and conclusions based on the first outgroup analysis of metazoan central neural characters. Finally, I will review two hypotheses concerning the morphological complexity of the last common bilaterian ancestor.

Fossil Record

The fossil record is notoriously incomplete. Fossils essentially exist as snapshots in time, and these snapshots are of varying quality. Some are grainy, providing only a glimpse of organisms and their ecology; others are fine-grained photographs of individual taxa and their ecology (Lagerstätten). Regardless, each snapshot provides unique and critical insights into the minimal age of a radiation. Each snapshot helps calibrate molecular clocks, establish ecological settings of evolutionary events, and reveal unsuspected morphological characters that challenge current conclusions regarding character transformation (2).

Ediacaran Biota.

The earliest reported fossils of possible metazoan embryos and adults are in the Ediacaran Doushantuo Formation (∼570 Mya) in southern China (3–5). Small globular fossils, ∼200 μm in diameter, show remarkable cellular details and have been interpreted as cnidarian gastrulae and planulae as well as bilaterian gastrulae comparable with living molluscans and echinoderms (4). However, the interpretation of these fossils as bilaterian metazoans has been questioned, and they have been reinterpreted as encysted holozoan protists (6). Similar problems plague the earliest reported adult bilaterian, Vernanimalcula, which is also from the Doushantuo Formation of southern China (7, 8). Fossils of Vernanimalcula (∼200 μm in diameter) have been described as broadly oval and triploblastic with a mouth, a differentiated gut surrounded by paired coeloms, and an anus. The rostral end of these “small spring animals” is also reported to have three pairs of external pits that have been interpreted as sensory organs (7). This interpretation has been questioned, however, and these fossils have been claimed to be taphonomic artifacts in which phosphates were deposited within a spherical object, such as the cysts of algal acritarchs (9).

The earliest fossils of macroscopic organisms interpreted as metazoans, including bilaterians, are in the Ediacaran strata above the Doushantuo formation (10). They average 10 cm but reach an extreme of 1 m in length, and they include forms that are frond-, disk-, and worm-like (Fig. 1A); their interpretation has had a tumultuous history. Many of these fossils were discovered in the late 1940s and were interpreted as representatives of living metazoan phyla. Forms like Eoporpita (Fig. 1A, 1) were interpreted as cnidarian pelagic medusa (11), and frond-like forms, such as Charniodiscus (Fig. 1A, 2), were interpreted as possible cnidarian sea pins (12). Still other forms of these fossils were interpreted as stem bilaterians. For example, Dickinsonia (Fig. 1A, 3) was interpreted as a flatworm (13), Arkarua (Fig. 1A, 4) was interpreted as an echinoderm (14), Spriggina (Fig. 1A, 5) was interpreted as an annelid capable of active swimming (15), and Praecambridium (Fig. 1A, 6) and a soft-bodied “trilobite” not formally described (Fig. 1A, 7) were interpreted as stem arthropods (16, 17). After this burst of descriptions, Ediacaran anatomy was reevaluated; claims were made that all Ediacarans were organized on a quilt-like pattern and represented an independent experiment of nonmetazoan animals, termed the Vendobionta, that failed with the evolution of macrophagous bilaterian metazoans (18–20). The concept of the Ediacaran biota as Vendobionta was generally abandoned, because paleontologists came to realize that the Ediacaran biota represents a wide range of morphological forms (10, 21).

Fig. 1.

Reconstruction of the Ediacaran (A) and Burgess Shale (B) biotas. The Ediacaran biota is reconstructed to convey maximal morphological complexity. (A) 1, Eoporpita; 2, Charniodiscus; 3, Dickinsonia; 4, Arkarua; 5, Spriggina; 6, Praecambridium; 7, soft-bodied “trilobite”; 8, Kimberella. (B) 1, Burgessochaeta; 2, Lingulella; 3, Ottoia; 4, Marrella; 5, Olenoides; 6, Naraoia; 7, Canadaspis; 8, Sidneyia, 9, Opabinia; 10, Anomalocaris; 11, Gogia; 12, Eldonia; 13, Pikaia; 14, Aysheaia; 15, Hallucigenia; 16, Odontogriphus; 17, Dinomischus.

To date, the Ediacaran biota includes some 160 taxa (10) found in 40 separate locations representing all parts of the globe except Antarctica. These biotas are dispersed among three stratigraphic zones in named assemblages based on a cladistic analysis of their spatial and temporal distribution (22): an Avalon assemblage (579 to ∼560 Mya), a White Sea assemblage (∼560 to ∼550 Mya), and a Nama assemblage (∼550–541 Mya).

The Avalon assemblage is dominated by simple, frond-like taxa, such as Charnia, grouped into a clade termed the Rangeomorpha (21). Macroscopic bilaterians are absent from the Avalon assemblage as well as trace fossils, such as surface tracks or shallow horizontal burrows, that would indicate the presence of small bilaterians that had developed muscles and coeloms to hydraulically locomote. One could infer from the Avalon assemblage that small bilaterians had not yet evolved, but this assemblage is the only Ediacaran assemblage from deep water; therefore, it is possible that small locomotory bilaterians existed at this time but were restricted to shallow ecological zones (23).

The White Sea assemblage represents the peak diversity of Ediacaran biota, including all of the taxa in Fig. 1A, which are grouped into nine clades (21). The clade Kimberellomorpha is of particular interest, because it includes Kimberella (Fig. 1A, 8), a small, oval-shaped animal that seems to have glided on a muscular foot and have an anterior end that houses a retractable arrow-shaped organ that was used to scratch the upper surface of the microbial mats on which it moved (24, 25). Trace fossils in the White Sea assemblage are diverse and suggest the presence of small bilaterians (21).

The Nama assemblage has less diversity than either the Avalon or White Sea assemblages, and it is dominated by frond-like taxa, called arboreomorphs, and simple cylindrical, sessile taxa, called erniettomorphs (21, 22). Bilaterian body fossils are absent, but small calcified shells of Cloudina and Namacalathus and the earliest evidence of predation in the form of holes bored into these calcified shells do occur (26).

Our understanding of body organization and phylogeny of Ediacarans is incomplete, but a conservative interpretation of the paleontological data indicates that most animals existed primarily on microbial mats; it was likely a 2D world, with sessile frond-like forms and vagile, small organisms that trophically were suspension feeders and grazers. There is little to no evidence that pelagic medusae existed (Fig. 1A, 1), but there is considerable evidence that sponges and sessile cnidarians were scattered across the microbial mats as were a number of bag-, frond-, and spindle-shaped taxa, forming clades that may be unrelated to any living metazoans. One or more of the three radiations of the small vagile organisms may be close relatives, or even stem members, of three clades of extant metazoans.

Cambrian Explosion.

The close of the Ediacaran was marked by a massive reduction in the Ediacaran biota, with only a small number of Ediacaran taxa continuing into the Early Cambrian (27). The small, calcified shells first seen in the late Ediacaran continue to, however, diversify in the early Cambrian to include a wide variety of plates, spines, and small shelly fauna, which seem to be the skeletal elements of bilaterians that ranged from a few millimeters to several centimeters in length (28). Clearly, bilaterians became armored in the Ediacaran–Cambrian transition. This finding suggests that the development of hard mouth parts may have been a key innovation to allow for additional expansion of macrophagous predators, giving rise to the first arms race (26). The small shelly taxa diversified over the next 14 Myr, which culminated in the Cambrian explosion of bilaterian diversity; this explosion seems to have occurred over a relatively short 10 Myr (29). Despite the rapidity of the Cambrian explosion, we have two fine-grained snapshots of the event captured in the exceptionally well-preserved, soft-bodied Lagerstätten of the Chengjiang biota of the early Cambrian (∼525 Mya) from the Yunan Province in South China and the Burgess Shale biota of the Middle Cambrian (∼505 Mya) from British Columbia. Each of these biotas has been extensively described (30–34) and analyzed for community composition and structure (35, 36).

Despite the Chengjiang and Burgess Shale biotas existing in distinctly different environments—the Chengjiang community is thought to have existed in a relatively shallow marine environment, possibly a partially enclosed embayment subject to periodic, storm-generated turbidity (30, 35), whereas the Burgess Shale biota is thought to have existed as a deep-water community on muddy sediments banked against the front of a stromatolite reef (Fig. 1B), where it was, thus, unstable and subject to periodic slumps, carrying parts of the community into deeper, anaerobic waters (32, 33)—both communities share numerous similarities. Both were dominated by arthropods, brachiopods, and priapulid worms (35, 36). Approximately 20% of each fauna consisted of sessile or burrowing infaunal species, and each fauna was dominated by epifaunal species, only 4% of whom were pelagic. Feeding strategies included suspension feeding and hunting/scavenging, forming complex food chains comparable with those food chains in many modern benthic marine ecosystems (37). Members of stem and/or crown groups at the bilaterian phylum level are in these Cambrian communities and occupy niches similar to those niches in modern benthic marine ecosystems, suggesting that competition among taxa was as high then as it is now. For example, in the Burgess Shale biota, infaunal species included polychaete annelids such as Burgessochaeta (Fig. 1B, 1), brachiopods such as Lingulella (Fig. 1B, 2), and priapulid worms such as Ottoia (Fig. 1B, 3), which seems to have been an aggressive predator (33). Epifaunal taxa included suspension- and detritus-feeding arthropods such as Marrella (Fig. 1B, 4), the trilobites Olenoides (Fig. 1B, 5) and Naraoia (Fig. 1B, 6), and a possible crustacean, Canadaspis (Fig. 1B, 7), as well as predatory arthropods such as Sidneyia (Fig. 1B, 8), Opabinia (Fig. 1B, 9), and Anomalocaris (Fig. 1B, 10), which grew to over 1 m in length and were clearly apex predators. Deuterostomes are also represented in the Cambrian biota of the Burgess Shale: the eocrinoid echinoderm Gogia (Fig. 1B, 11) and a possible pelagic holothurian, Eldonia (Fig. 1B, 12), as well as a possible cephalochordate, Pikaia (Fig. 1B, 13).

Unlike the Burgess Shale biota, the Chengjiang biota contains a rich variety of chordates. Two sessile, putative urochordates, Cheungkongella and Shankouclava, have been described (38, 39) as well as another Pikaia-like chordate, Yunnanozoon, which was initially described as a possible cephalochordate with a notochord, segmented trunk muscles, and an expanded pharynx with an endostyle (40). This taxon was subsequently reinterpreted as an early vertebrate (41), and it was then reinterpreted again as the earliest known enteropneust hemichordate (42). Subsequently, a third Pikaia-like taxon, Cathaymyrus, was described in the Chengjiang deposits and interpreted as a cephalochordate (43).

Early craniates (hagfishes and vertebrates) may also occur in the Chenjiang biota. Haikouella has been interpreted as a craniate-like chordate with a well-developed brain, lateral eyes, a pharynx with gills, and a ventral heart (44, 45). A subsequent interpretation of the Haikouella material suggests that the head consisted of separate dorsal and ventral movable units connected by external gills (46) and that Yunnanozoon and Haikouella are stem group deuterostomes that are allied to vetulicolians, another problematic group in the Chengjiang biota (46, 47). Thus, the yunnanozoans (Yunnanozoon, Cathaymyrus, and Haikouella) may be stem cephalochordates, or they may be closely allied to vetulicolians and may possibly be stem deuterostomes.

The situation is somewhat clearer regarding the first vertebrates from the Chengjiang Lagerstätte. Two genera, Haikouichthys and Myllokunmingia, have been described as agnathan vertebrates, with Haikouichthys said to be closely allied to living lampreys and Myllokunmingia said to be closely allied to living hagfishes (48). However, it has been claimed that this interpretation is based on tenuous characters and that both taxa may form a clade (myllokumingids) that is basal to living craniates (49, 50). Subsequently, another described genus, Zhongjianichthys, seems to be a myllokumingid (51). In any case, these taxa seem to have had paired nasal capsules, large lateral eyes, and possibly, paired otic capsules, all of which suggest that that they may have possessed brains comparable with living agnathan vertebrates (51).

Similar taxonomic problems plague a number of the Burgess Shale taxa. Aysheaia (Fig. 1B, 14) and Hallucigenia (Fig. 1B, 15) have been considered to be primitive onychophoran worms (33) but are more probably an extinct clade (the lobopods, which were possibly a stem group of arthropods) (52). This finding is also true of Opabinia (Fig. 1B, 9) and Anomalocaris (Fig. 1B, 10), which may be members of a clade of stem arthropods, although their exact relationship to other arthropods is still unclear (52). Odontogriphus (Fig. 1B, 16), a pelagic, flattened, 12-cm-long animal with tooth-like elements surrounding the mouth, remains an enigma but may be a basal lophotrochozoan related to annelids, brachiopods, or molluscs (33). Dinomischus (Fig. 1B, 17) poses similar problems. The bodies of these 10-cm-long animals consisted of a calyx, which housed the mouth and anus opening onto the upper surface of the calyx, and a stem, which was anchored in the sediment (33). These animals have been compared with both echinoderms and entoprocts, but their taxonomic affinities are presently unclear. Continued study and the discovery of new fossils will likely resolve their positions.

Comparative Approaches

Comparative biologists use two very different approaches in formulating evolutionary statements: cladistics (or phylogenetics) (53, 54) and phenetics (55, 56). Both involve comparing traits or characters (any definable attribute of an organism) among multiple species, but each treats similar characters differently. Cladists, following Hennig (53), divide similar characters among organisms into three categories: shared primitive characters, shared derived characters, and uniquely derived characters. Furthermore, cladists hold that only shared derived characters, which they define as phylogenetic or taxic homologs (57), can form the basis for establishing genealogical relationships. Such relationships are usually illustrated as a dendrogram or a sequence of branches (a cladogram). In contrast, phenetists say that overall similarities define homologies, which can be recognized by structural and compositional correspondence and are said to be phenetic or operational homologies (56). Phenetists also believe that genealogical relatedness depends on the degree of similarity (i.e., the number of operational homologies) shared by a group of organisms.

In most groups of organisms, multiple hypotheses of genealogical relationships can be proposed. Hennig (53) was the first to discover that these hypotheses can be evaluated objectively by grouping organisms based on shared derived characters. For example, given three taxa—A, B, and C—there are only three possible hypotheses regarding their relatedness: A is more closely related to B; A is more closely related to C; or A, B, and C are equally related. Because phylogeny is a historical process that has occurred only one time, only one of these hypotheses can be valid. The distribution of postulated shared derived characters will indicate the valid hypothesis. That is, the genealogical hypothesis that reveals the largest number of shared derived characters and thus, requires the fewest independent origins is the one that is supported. (This conclusion was initially based on simple parsimony but is recently supported by sophisticated algorithms, such as Bayesian inference.) In rejecting the alternate genealogical hypotheses, their “shared derived characters” are revealed to be shared primitive characters or homoplasies (54). This finding does not mean that such characters are of no phylogenetic interest. Many shared primitive characters are, in fact, shared derived characters at some lower level of the tree of life and thus, linked as transformational homologs to shared derived characters at a higher level. In addition, analysis of homoplasous characters can reveal structural and functional constraints in phylogeny. Although transformational homologies do not specifically define taxonomic groups, they become critical in evaluating the phylogenetic history of characters across clades. Hennig (53) also discovered that character polarity (i.e., primitive or derived) could be determined by an outgroup rule, which proposes that, when two or more homologous characters occur within a group, the character outside the group is the primitive character, whereas the character found only within the group is the derived character. Realization of the predictive power of the outgroup rule in the work by Hennig (53) has given rise to a wide range of evolutionary studies that have attempted to reconstruct the phylogenetic history of molecular characters (58), morphological characters (59–61), behavioral and ecological characters (62), and biogeographical events (63). Such studies are usually called cladistic studies, because they rely on the outgroup rule in the work by Hennig (53), although they deal with the phylogeny of a character rather than reconstructing the phylogenetic history of taxa. It has been claimed that studies involving the outgroup rule in the work by Hennig (53) in this way are not truly cladistic analyses (64), presumably because they rely on cladograms generated in other studies; however, there is a general consensus that they do fall under the rubric of cladistic methodology (65, 66).

Because of its logic and methodological transparency, cladistics has largely replaced phenetics in zoological systematics, except at the species level, and much of its methodology is widely used to analyze the phylogenetic history of both genotypic and phenotypic characters. The phenetic approach is still widely used, however, in developmental genetic studies, in which evolutionary statements are based on a two-taxon approach, possibly because until recently, it has been difficult to explore the genetic basis of phenotypic characters widely among different taxa. The roles of both cladistic and phenetic approaches are examined in the next three sections dealing with the molecular clock hypothesis, the genetic basis of bilaterian body plans, and an outgroup analysis of metazoan neural characters.

Molecular Clock Hypothesis.

Evolutionary biologists seek to date the origin of metazoan clades and determine the rate at which they evolve. Initially, clade origins were based solely on the earliest occurrence in the fossil record of that clade, and the temporal rate was established by current rates of sedimentation and subsequently, radiometric dating (67). Given the incompleteness of the fossil record, however, the accuracy of these estimates of origin and tempo were open to question. In the early 1960s, it was discovered that differences between lineages in the number of amino acids in several proteins seemed to be roughly linear in time and that evolutionary changes in these proteins, as well as in genes, could be used to infer the separation in time of different lineages (68–70). This discovery led to the neutral theory of molecular evolution, which claimed that most changes in proteins and genes would be neutral and that fixation of these molecules would accumulate at a clock-like rate (71). It has become clear, however, that the rate of change in different molecules in different clades varies tremendously (72–74). This problem has been addressed by modeling relaxed molecular clocks, in which mean rates of sequence divergence for each molecule have been calculated (75) or the molecular clock is calibrated by one or more points based on fossil dates (67). Relaxed molecular clocks are frequently generated in a two-step process: the most supported cladogram is generated by cladistic analysis of the molecules of interest, and then, a clock is calibrated by using the time of origin of several clades based on the fossil record.

Conclusions based on mean rates of sequence divergence differ greatly from those conclusions based on multiple fossil calibration points in regard to the time of origin for metazoan clades. Using mean rates of sequence divergence, the origin and divergence of bilaterians has been placed at ∼1.0–1.2 billion y ago (75). In contrast, fossil-calibrated molecular clocks place the origin and divergence of bilaterians at ∼700–600 Mya (21, 76–79). These later dates suggest that, although animals arose during the Cryogenian Period (∼850–635 Mya), bilaterians arose and began to radiate during the Ediacaran Period.

Genetic Basis of Bilaterian Body Plan.

In the last 20 y, developmental biologists have made spectacular strides in revealing the genetic basis of the regulatory networks that underlie anterior–posterior and dorsoventral patterning of body organization in many bilaterian metazoans (80–85). Anterior–posterior patterning involves a homeobox gene superfamily in which orthodenticle and its paralogue (Otx) are expressed in the rostral head, followed more caudally by the expression of paired -box (Pax) genes and most caudally by Hox genes, which continue to be expressed in the trunk. Much early work on body patterning was based on two-taxon comparisons involving fruit flies and mice, but this research has now included extensive outgroup analyses (86–89).

One consequence of the discovery of the genetic basis of anterior–posterior body patterning in bilaterian metazoans was the realization that Otx, Pax, and Hox genes are also expressed in a rostral to caudal sequence in those bilaterians that possess brains (90–94). This finding gave rise to the tripartite brain hypothesis, which proposes that there is a monophyletic origin of the brain in bilaterians. The original hypothesis (91) was based on a two-taxon comparison (between an arthropod and chordates), but more recently, this hypothesis was extended to a rudimentary three-taxon outgroup analysis (93) involving an annelid (a spiralian protostome), an arthropod (an ecdysozoan protostome), and a mammal (a deuterostome). To date, however, there has been no attempt to polarize expression of these homeobox genes in ecdysozoan or spiralian protostomes, which possess less complex CNSs, although a tripartite brain in deuterostomes has traditionally been interpreted as a derived character (65, 66). Hopefully, continued study of the genetic regulatory networks underlying anterior–posterior patterning in ecdysozoan and spiralian protostomes will provide additional insights into the phylogeny of these networks and brain evolution in bilaterians.

The discovery of the genetic processes involved in the dorsoventral patterning of the body in several bilaterian metazoans (85, 95, 96) may also provide support for the tripartite brain hypothesis. It has been shown that the gene short gastrulation (sog) in Drosophila is functionally homologous to the gene chordin (chd) in Xenopus; both promote dorsal development, whereas the gene decapentapelgic (dpp) in Drosophila and its homolog bmp-4 in vertebrates promote ventral development (85). Both sog and chd, in conjunction with other genes, also promote formation of neurogenic ectoderm. However, the expression of sog and chd are inverted in the blastula of Xenopus relative to their expression in stage 12 embryos of Drosophila. This finding leads to the suggestion that vertebrates evolved from protostomes by a dorsoventral inversion (95, 96), resurrecting an earlier inversion hypothesis proposed in the work by Geoffroy Saint-Hilaire (97). If a dorsoventral inversion of the body axis occurred with the origin of chordates, then their brains could be considered homologous (90, 95, 96). Once again, this claim is based on a two-taxon phenetic comparison and not a cladistic one. Work on body patterning in an enteropneust hemichordate Saccoglossus, which has a diffuse nerve net, reveals the same expression of homeobox genes in the anterior–posterior body axis as in other bilaterians, but the antagonistic actions of sog and dpp do not restrict neural development to the dorsal body surface of this bilaterian (98, 99). This finding suggests that, although the genetic signaling network is homologous between protostomes and deuterostomes, this network can be deployed to regulate the development of fundamentally different nervous systems. It is possible that the ancestral roles of the regulatory networks involved in anterior–posterior as well as dorsoventral patterning did not extend to patterning CNSs and that elements of these networks were subsequently co-opted in neural development.

Outgroup Analysis of Metazoan Central Neural Characters.

Clearly, it is difficult to discern the connection between genetic networks and phenotypic characters (100, 101). This discernment will become easier as the phylogeny of particular genetic networks is mapped cladistically, with particular attention paid to taxa that occupy critical positions in the metazoan cladogram. Meanwhile, it is useful to conduct an outgroup analysis of the distribution of central neural characters in extant bilaterians. Although we recognize ∼40 metazoan phyla, comprising some 1.3 million described species (102), only 8 of these phyla (cnidarians, platyhelminthes, annelids, molluscs, nematodes, arthropods, echinoderms, and chordates) comprise 99% of extant metazoan species, and 4 of these phyla (annelids, some molluscs, arthropods, and chordates) have brains; thus, bilaterian metazoans with brains comprise 90% of the extant metazoan species. Clearly, the evolution of a brain as part of an adaptive suite has been under heavy selective pressure. If an outgroup analysis of central neural characters reveals that a brain is a shared primitive character for bilaterians, then the tripartite brain hypothesis might be supported. This finding would be the case, however, only if a brain divided into three parts is a shared primitive character. If tripartite brains are revealed to be a derived neural character, then this finding would be evidence again for the tripartite brain hypothesis. If brains are revealed to be a derived character, then brains in bilaterians must have evolved a number of times independently, which suggests that elements of the genetic network underlying anterior–posterior patterning have also been co-opted for brain patterning a number of times independently.

An outgroup analysis of central neural characters in metazoans is complicated by the lack of a consensus regarding a single metazoan cladogram (21, 102–105) and difficulties in defining distinct central neural characters. Despite these difficulties, comparative molecular studies have clarified much of the phylogeny. All molecular studies recognize bilaterians as a monophyletic taxon divided into two major clades: the protostomes and the deuterostomes (Fig. 2). Furthermore, the protostomes can be divided into two superphyla or clades termed the ecdysozoans and the spiralians (lophotrochozoans). Conflicts regarding metazoan phylogeny currently center on the contentious relationships of acoelomorph flatworms (Acoela and Nemertodermatida), the genus Xenoturbella (a small ciliated marine worm), and the basal metazoan clades (cnidarians, ctenophores, placozoans, and poriferans). The cladogram generated in the work by Glenner et al. (104) was chosen for the present outgroup analysis (with some modifications), because it is the only Bayesian phylogenetic analysis that includes both molecular and morphological data. The Acoela group has been interpreted as the sister group to all other bilaterians (104) or a deuterostome clade, the xenocoelmorphs (105, 106), which includes acoelomorph flatworms, nemertodermatids, and Xenoturbella. When xenacoelmorphs are interpreted as deuterostomes, they are considered either as the sister group to Ambulacraria (Echinodermata and Hemichordata) or the sister group to all other deuterostomes (107). The latter interpretation is accepted in the present outgroup analysis. The following characters—or levels of increasing morphological and functional complexity in the cephalic CNSs of extant metazoans—are recognized in the present outgroup analysis: (i) diffuse nerve nets or subepidermal nerve plexuses; (ii) simple cerebral ganglia; and (iii) brains, defined as a central collection of neural centers, with distributed and hierarchical functions. A considerable range of morphological complexity occurs within each of these cephalic neural characters (108). Diffuse nerve nets range from those nets of cnidarians and ctenophores to those nets in enteropneust hemichordates, in which neural cell bodies occupy a subepidermal nerve plexus with centralized bundles of fast-conducting axons forming dorsal and ventral nerve cords. Cerebral ganglia range from simple, bilobed ganglia in polyclade flatworms to more complex multiple cephalic ganglia in many gastropod molluscs (108). In most annelids, arthropods, and some cephalopod molluscs, brains form by elaboration of one or more cephalic ganglia, whereas in vertebrates, they form by elaboration of their dorsal hollow neural tube. Categorizing cephalic nervous systems in protostomal bilaterians as simple cerebral ganglia or brains is somewhat arbitrary, because the criteria are based on the relative size and functional complexity of the cephalic structure in question. Simple cerebral ganglia and brains in protostome bilaterians, thus, represent grades of increasing morphological and functional complexity. There is a similar problem in defining a “brain” among chordates: cephalochordates possess a brain that is only slightly more complex than their spinal cord; they do not seem to have a homolog of the cerebrum in vertebrates; and separation of a thalamus and midbrain does not appear to exist, nor does a cerebellum (65, 109). It is possible that future analyses will reveal additional morphological categories among cephalic neural characteristics in metazoans; if so, this information may help to resolve the definition of a brain and thus, contribute to our understanding of the evolution of centralized nervous systems.

Fig. 2.

Outgroup analysis of cephalic neural characters across extant metazoans. The cladogram is modified from the work by Glenner et al. (104), with the inclusion of xenacoelmorphs as the sister taxon to other deuterostomes (106). The analysis indicates that the last bilaterian common ancestor possessed a diffuse nerve plexus and that brains independently evolved at least four times among bilaterians.

Distribution of the three cephalic neural characters is plotted on the cladogram in Fig. 2. Polarization of these characters among deuterostomes suggests that a diffuse nerve plexus is the primitive character, and a brain is the derived character. In both ecdysozoan and spiralian clades, simple cerebral ganglia seem to be the primitive character, whereas a brain is the derived character. If so, brains have evolved three times independently among protostome bilaterians (Fig. 2). The condition in the last bilaterian common ancestor could be either a diffuse nerve plexus or a simple cerebral ganglion, but examination of the metazoan outgroups suggests that the last bilaterian common ancestor possessed a diffuse nerve plexus like the last common ancestor of all metazoans.

Different conclusions might be reached, however, if xenocoelmorphs were determined not to be deuterostomes (106) but the sister group to all other bilaterians (103). If this finding was the case, it would be impossible to polarize the characters diffuse nerve plexus and brain in deuterostomes. Because the third neural character, simple cerebral ganglion, is the primitive condition in protostome bilaterians, additional examination of the outgroups, xenocoelmorphs and the basal metazoan clades, would still indicate that the last common bilaterian ancestor and the last common metazoan ancestor possessed a diffuse nerve plexus.

Outgroup analysis of intracladal variation in central cephalic neural characters also indicates that brains have evolved numerous times independently. For example, analysis of cephalic neural characters within the molluscan clade reveals that basal molluscans (monoplacophorians and polyplacophorians) have simple cerebral, pleural, and pedal ganglia interconnected by ventral and lateral medullar cords (108), which are retained in some basal gastropods. Hypertrophy of the various ganglia seems to have occurred independently in several gastropod groups as well as independently in octopod cephalopods (110), which have evolved the most complex brains among invertebrates.

It could be said that the present outgroup analysis may be flawed by mistaking secondary character reductions (degenerative events) for primitive characters (90, 94, 111). If this flaw was the case, 23 of ∼30 phyla would have to possess secondarily degenerated cephalic neural characters, which in the context of this cladogram, would have to have occurred at least 11 times independently. Needless to say, this interpretation would not be parsimonious, but it does raise the question of how secondarily simplified characters can be recognized from primitive simple characters.

In addition to the outgroup rule, there are at least four auxiliary criteria that suggest to zoologists that primitive characters are actually secondarily simplified characters: (i) when the characters are in sessile taxa, (ii) when the characters are in parasitic taxa, (iii) when the characters are in paedomorphic taxa, and (iv) when the characters are in taxa with secondary loss of microRNAs. Three of these criteria have been recognized by zoologists for almost 50 y (108). Secondarily simplified characters have long been suspected in sessile tunicate urochordates, bryozoans, phoronids, entoprocts, and parasitic cestode and trematode flatworms and rhomboaoans, which are obligate symbiots in the nephridia of cephalopod molluscs. Similar secondary simplification also frequently occurs when ancestral ontogenies are truncated (paedomorphosis), leading to reduction in body size and morphological complexity, which is widely documented in salamanders (112). The fourth criterion, taxa with secondary loss of microRNAs, may offer a molecular explanation for the first three criteria, and it may also identify additional taxa characterized by multiple character reductions or losses. Acoela, Platyhelminthes, and Xenoturbella each seem to have secondarily lost microRNAs (21), which suggests that many of their morphological characters are secondarily simplified. It should be noted that, in most if not all of the taxa said to possess secondarily simplified characters, these characters are widespread throughout most organ systems in that taxon rather than confined to a single system. Furthermore, the fossil record can be of immense value in polarizing life histories when the earliest members of a clade are vagile and shift to a sessile existence or when there are clear trends in body size.

Last Common Bilaterian Ancestor

There are two very different reconstructions of the morphology of the last common bilaterian ancestor (LCBA) or urbilaterian (95). Many developmental biologists, relying on the roles of numerous genes and gene networks in the development of arthropods and vertebrates (summarized in ref. 113), suggest that the LCBA was a morphologically complex organism with anterior–posterior differentiation of a head that possessed paired eyes, a tripartite brain, and a segmented trunk with a differentiated gut, heart, and appendages. In contrast, many paleontologists and zoologists would suggest that the LCBA was far more simple morphologically, perhaps a small vernanimalcular-like organism that was patterned in both the anterior–posterior and dorsoventral axes but not segmented. This ancestor would have possessed a mouth and anus connecting a differentiated gut surrounded by coelomic cavities. The nervous system would have been a diffuse nerve plexus, and the apical pole of the organism would have had simple ocelli composed of both ciliary and rhapdomeric photoreceptors. The trunk may have contained contractile muscle cells but no heart, segmented muscles, or appendages, and locomotion would primarily have involved ciliary gliding.

Both molecular clock and paleontological data indicate that bilaterian metazoans arose ∼600–700 Mya during the Ediacaran, and they radiated rapidly into most bilaterian crown clades by the end of the Cambrian (21). It is also clear that most genes involved in developmental genetic networks determining anterior–posterior and dorsoventral patterning must already have been in place in the LCBA (21, 114). If the fossils of the Doushantuo Vernanimalcula and some of the macroscopic fossils of the Ediacaran biota, such as Dickinsonia and Kimberella, are interpreted as stem bilaterians, then the body plans of the earliest bilaterians must have been relatively simple and comparable with the body plans of living placozoans, platyhelminthines, and aplacophoran molluscs. Although neural structures are rarely fossilized, it is possible to relate neural complexity to specific grades of body complexity (108). A conservative interpretation of body complexity of the macroscopic Ediacaran biota suggests that these organisms were characterized by diffuse nerve plexuses. A more heterodox interpretation of organisms, such as Spriggina, Praecambridium, and so-called soft-bodied “trilobites”, is that they are members of clades closely related to annelids and arthropods, which would suggest that some Ediacaran organisms may have already evolved cerebral ganglia sufficiently complex to be termed brains. Given the body complexity of Cambrian annelids, arthropods, and chordates, it is reasonable to assume that the CNSs in these clades were characterized by brains. Interestingly, this level of neural complexity may not have been reached by cephalopod molluscans until the Devonian some 70 Myr later, with the origin of octopod cephalopods (115).

Outgroup analysis of inter- and intraclade variations in cephalic neural characters (Fig. 2) supports an LCBA model with a diffuse nerve plexus, which subsequently coalesced into a number of cephalic ganglia and nerve cords or a dorsal hollow neural tube. Hypertrophy and increase in cellular differentiation of cephalic ganglionated and dorsal neural tube systems independently reached levels of neural complexity that are defined as brains in arthropods, annelids, and some molluscs and chordates.

Conservation of genetic regulatory networks, which has been termed deep homology (116, 117), has been invoked to claim that all bilaterian brains are homologous (a shared derived character of all bilaterian metazoans) and consist of three anterior–posterior divisions (tripartite brain hypothesis). A basic assumption of this claim is that conserved genetic regulatory networks also have a conserved role in the development of phenotypes. As developmental biologists dissect the genetic mechanisms that control processes underlying the development of phenotypic characters, it seems that some genetic networks determine character identity, whereas others determine character state (101). Only as the genomes of additional taxa are probed and analyzed cladistically will it be possible to determine if homologous character identity networks underlie phenotypically recognized brain divisions across all bilaterian metazoans. Meanwhile, metazoan interrelationships and the evolution of their nervous systems will continue to be debated, hopefully, with the reminder by the late renowned invertebrate zoologist, Donald P. Abbott, to “Cultivate a suspicious attitude toward people who do phylogeny” (118).